Show
Your pulse rate — which is also your heart rate — can tell a lot about your heart health and fitness. When you're resting or reclining, your heart is pumping the lowest amount of oxygen your body needs. That is your resting heart rate, says Harvard Health Publishing. Effect of Body Position on Heart Rate"A normal resting heart rate can be between 60 and 100 beats per minute," says Peter Santucci, MD, professor of cardiology at Maywood, Illinois-based Loyola University Medical Center. "This rate can vary slightly with body position changes. Most notably, when you go from reclining to standing. Your heart rate may go up by 10 to 15 beats per minute." According to the American Heart Association (AHA), after you go from a reclining or sitting position to a standing position, the increase in your pulse should settle back down after about 15 to 20 seconds. The AHA notes that body position is not the only thing that affects your resting heart rate. Other factors include:
Even though your reclining or resting heart rate can be normal between 60 and 100 beats per minute, very active people can have a resting heart rate as low as 40 and still be considered normal, the AHA says. Also, as explained by Harvard Health, some studies suggest that although a resting heart rate of up to 100 may be normal, a resting heart rate above 80 may indicate a higher risk for heart disease. What About Heart Rate During Exercise?When you're active, your body position is changing constantly, so your heart has to beat faster to keep all your muscles supplied with oxygen. This creates the opposite of your resting heart rate, which is your maximum heart rate. If your average maximum heart rate is too high, it means that your heart is working too hard during exercise, Harvard Health points out. Vigorous exercise is a good way to lower both your resting heart rate and your maximum heart rate. "Although your resting heart rate is not much affected by your age, the normal range for your maximum heart rate does change with age," Dr. Santucci says. "To find out what your average rate should be during exercise, subtract your age from 220. A target rate for moderate exercise is 50 to 70 percent of maximum, and for vigorous exercise, 70 to 85 percent of maximum. This may vary for some individuals." The AHA says to measure the effects of exercise, take your pulse before, during and after exercise. The harder you exercise, the quicker you will reach your maximum heart rate. When you rest after exercise, your heart rate will start to come down. The quicker your heart rate returns to your resting rate, the better your physical fitness. As your body moves during exercise, your blood vessels open up to get more blood to your muscles. Some people may worry that exercising to reach maximum heart rate may dangerously increase blood pressure. According to AHA, this is not the case. A rising heart rate does not increase blood pressure at the same rate. In fact, you may double your heart rate during exercise while only slightly increasing your blood pressure. According to the AHA, you should let your doctor know if your resting heart rate is very low or very high, especially if you have symptoms like weakness, dizziness or a fainting spell. "If you have an occasional resting heart rate outside the normal range, and you don't have symptoms, it is probably not significant," Dr. Santucci says. "However, if your heart rate is frequently below 50 or above 110, best to let your doctor know." This website uses cookies. By continuing to use this website you are giving consent to cookies being used. For information on cookies and how you can disable them visit our Privacy and Cookie Policy. Got it, thanks!orthostatic intolerance disorders, which are common in every age, are difficult to diagnose and treat. Typically, these disorders, with clinical manifestations including dizziness, syncope, orthostatic hypotension, falls, and cognitive decline, are a result of several biological mechanisms. To develop better strategies to treat and diagnose orthostatic intolerance, it is important to understand the underlying mechanisms leading to these disorders. One of the main mechanisms involved is the short-term cardiovascular regulation of blood flow to the brain, which includes autonomic regulation and cerebral autoregulation. The overall goal of this work is to develop a mathematical model that can predict dynamics in observed cerebral blood flow and peripheral blood pressure data and propose mechanisms that can explain the interaction between autonomic regulation and cerebral autoregulation. To this end, we have developed a mathematical model that can predict these two regulatory mechanisms. To validate the model, we compare model predictions with measurements of arterial finger blood pressure and middle cerebral artery blood flow velocity of a young subject. On the transition from sitting in a chair to standing, blood is pooled in the lower extremities as a result of gravitational forces. Venous return is reduced, which leads to a decrease in cardiac stroke volume, a decline in arterial blood pressure, and an immediate decrease in blood flow to the brain. The reduction in arterial blood pressure unloads the baroreceptors located in the carotid and aortic walls, which leads to parasympathetic withdrawal and sympathetic activation through baroreflex-mediated autonomic regulation. Parasympathetic withdrawal induces fast (within 1–2 cardiac cycles) increases in heart rate, whereas sympathetic activation yields a slower (within 6–8 cardiac cycles) increase in vascular resistance, vascular tone, and cardiac contractility and a further increase in heart rate (4, 7, 37). Simultaneously, cerebral autoregulation, mediated by changes in CO2, myogenic tone, and metabolic demand, leads to vasodilation of the cerebral arterioles (2, 18, 34, 38). Our mathematical model includes two submodels: 1) a cardiovascular model that can predict blood pressure and blood flow velocity during sitting and 2) a control model that can predict autonomic and cerebral regulatory mechanisms during the postural change from sitting to standing. Both submodels are based on the same closed-loop model with 11 compartments that represent the heart and systemic circulation. Our previous work (27, 29) also used compartmental models to describe the dynamics of the cardiovascular system. One (27) used an open-loop (3-element windkessel) model to analyze dynamics of cardiovascular control. This model used arterial blood pressure measured in the finger as an input to predict model parameters that describe dynamics of cerebral vascular regulation for young subjects. These parameters were obtained by minimizing the error between computed and measured middle cerebral artery blood flow velocity. Consequently, no equations were used to describe possible mechanisms of the underlying regulation. To further advance this study, we recently developed a seven-compartment closed-loop model (29) that can predict the dynamics observed in the data. This model did not rely on an external input; rather, it included a submodel that describes the pumping of the left ventricle. In addition, the seven-compartment model included simple equations that describe the short-term regulation. This model was able to accurately predict dynamics of cerebral blood flow velocity and arterial blood pressure during sitting (t < 60 s) and standing (t > 80 s), as well as the mean values during the transition from sitting to standing (60 < t < 80 s), but it was not able to predict detailed dynamics during the transition from sitting to standing. Furthermore, we were not able to achieve adequate filling of the left ventricle. To obtain a more accurate model, we developed the 11-compartment model, which overcomes limitations of the 7-compartment model by 1) predicting resistances as nonlinear functions of pressure, 2) adding essential compartments, 3) devising an empirical model of autoregulation, and 4) including a new physiological model describing pooling of blood in the lower extremities due to effects of gravity. A large body of work that describes cardiovascular control modeling (9–11, 30, 44) is based on predictions of mean values for arterial blood pressure and cerebral blood flow velocity. Consequently, these models cannot predict the pulsatile dynamics of the cardiovascular system. These models use optimal control to minimize the deviation between some observed quantity (e.g., arterial blood pressure) and a given set point. Although this strategy can provide good parameter estimates, optimal control models do not describe the underlying physiological mechanisms. Other modeling strategies have been proposed by Melchior et al. (19, 20) and Heldt et al. (8), who devised pulsatile models that include pulsatility, autonomic regulation, and effects of gravity. The latter was done by changing the reference pressure outside the compartments. However, these models do not include effects of autoregulation. One way to model the effect of autoregulation is to let the cerebrovascular resistance be a function of time, as suggested by Ursino and Lodi (39). However, this work does not include the effects of autonomic regulation. A second group of models described parts of the control system without validation against experimental data (5, 19–21, 31, 32, 35, 40–43). These models used a closed-loop compartmental description of the cardiovascular system combined with physiological descriptions of the control. Although these models can provide qualitative analysis of the system, they cannot be used for quantitative comparisons with data. Furthermore, most of the models in the second group describe the effects of autonomic regulation without including the effects of cerebral autoregulation. In contrast, our model includes autonomic and cerebrovascular regulations and provides quantitative comparisons with physiological data. Glossary
MODELING BLOOD PRESSURE AND BLOOD FLOW VELOCITYOur cardiovascular model is based on an 11-compartment closed-loop model. The model is designed to predict blood pressure and volumetric blood flow in the left atrium, left ventricle, aorta, vena cava, arteries, and veins in the upper body, lower body, and head, as well as arteries in the finger (Fig. 1). Each compartment represents all vessels in areas of similar pressure. Hence, in its simplest form, the systemic circuit could consist of one arterial (high-pressure) and one venous (low-pressure) compartment. In our model, we include five arterial compartments and four venous compartments. Fig. 1.Compartmental model of systemic circulation. The model contains 11 compartments: 5 represent systemic arteries (brain, upper body, lower body, aorta, and finger), 4 represent systemic veins (brain, upper body, lower body, and vena cava), and 2 represent left atrium and left ventricle. Because the pulmonary system is not included, systemic veins are directly attached to the left ventricle. Each compartment includes a capacitor to represent compliant volume of arteries or veins. All compartments are separated by resistors representing resistance of the vessels. Compartment representing the left ventricle has 2 valves (aortic and mitral). Following terminology from electrical circuit theory, flow between compartments is equivalent to electrical current, and pressure inside each compartment is analogous to voltage. Resistors (R, mmHg·s·cm−3) are marked with zigzag lines, capacitors (C, cm3/mmHg) with dashed parallel lines inside the compartments, and aortic and mitral valves with short lines inside the compartment that represents the left ventricle. Cer, cerebral; see Glossary for other abbreviations. The 11 compartments depicted in Fig. 1 are chosen to ensure that the level of detail in the model is adequate to describe the complex dynamics observed in the data and, at the same time, is not too complex to be solved computationally. Four compartments that represent the upper body and the legs are included to model venous pooling of blood and sympathetic contraction of the vascular bed. Two compartments that represent the brain are included to model effects of cerebral autoregulation and to enable model validation against cerebral arterial blood flow velocity measurements. One compartment that represents the finger is included to enable model validation against arterial blood pressure measured in the finger. To determine cardiac output and venous return, two compartments are included to represent the aorta and vena cava. Finally, to obtain a closed-loop model, it is necessary to include a source (i.e., the heart) that pumps blood through the system. Consequently, two compartments are included to represent the left atrium and left ventricle. Our previous work (29) included only the left ventricle; without an atrium, it is not possible to achieve adequate filling of the heart. The major system not included in our model is the pulmonary circulation. Addition of compartments that represent the pulmonary circulation would require more parameters, which would increase the computational complexity. Instead, the pulmonary circulation is represented as a resistance between the vena cava and the left atrium. To study dynamics of postural change from sitting to standing, it is not important to know how blood is distributed among various inner organs. Hence, the upper body is simply represented by an arterial and a venous compartment. Each compartment is represented by a compliance element (inverse elasticity) and is separated by resistance to flow. The design of the systemic circulation with arteries and veins separated by capillaries provides some resistance and inertia to the volumetric flow rate. In our model, we include effects of resistance between compartments but neglect effects due to inertia. The major resistance to flow is located in peripheral regions between compartments that represent arteries and veins. Compartments that represent large conduit vessels are also separated by resistances that represent the overall resistance of the compartment. Resistances between conduit vessels are very small compared with peripheral resistances. The description of blood pressure and volumetric flow in a system consisting of compliant compartments (capacitors) and resistors is equivalent to that of an electrical circuit (Fig. 1), where blood pressure plays the role of voltage and volumetric flow rate plays the role of current. To compare our model with data, we assume that the diameter of the middle cerebral artery remains constant, such that blood flow velocity can be obtained by scaling volumetric blood flow by a constant factor that represents the area of the vessel. Recent measurements of middle cerebral artery diameter by magnetic resonance imaging combined with transcranial Doppler assessment of cerebral blood flow velocity have demonstrated that the middle cerebral artery diameter does not change, despite large changes in cerebral blood flow velocity elicited by stimuli such as lower body negative pressure and CO2 changes (36). To predict blood pressure and blood flow within and between the compartments, we base our model on volume conservation laws (41). Blood pressure and volumetric blood flow can be found by computing the volume and change in volume for each compartment. The equations that represent the arterial and venous compartments are similar. For each of these compartments, the stressed volume V = Cp (cm3, volume pumped out during 1 cardiac cycle), where C (cm3/mmHg) is compliance and p (mmHg) is blood pressure. The cardiac output (CO) from the heart is given by CO = HVstroke (cm3/s), where H (beats/s) is heart rate and Vstroke (cm3/beat) is stroke volume. For each compartment, the net change of volume is given by (1) where q (cm3/s) is determined analogously to Kirchhoff's current law and R is the resistance to flow. Several compartments have more than one inflow or outflow. For example, the compartment that represents the aorta has three outflows (qout = qaf + qau + qac), whereas the compartment that represents the vena cava has three inflows (qin = qafp + qvu + qvc; Fig. 1).To model the left ventricle as a pump, the position of the mitral and aortic valves must be included. During diastole, the mitral valve is open, while the aortic valve is closed, allowing blood to enter the left ventricle. Then isometric contraction begins, increasing the ventricular pressure. Once the ventricular pressure exceeds the aortic pressure, the aortic valve opens, propelling the pulse wave through the vascular system. For healthy young people, both valves cannot be open simultaneously. To incorporate the state of the valves, we have modeled the resistances (Rav and Rmv; Fig. 1) as follows where v represents mitral and aortic valves. This equation results in a large resistance (and no flow) while the valve is closed and a small resistance (and normal flow) while the valve is open. The minimum (min) value is introduced to avoid numerical problems due to large numbers. A system of differential equations is obtained by differentiating the volume equation V = Cp and inserting Eq. 1 (2) The circuit in Fig. 1 gives rise to a total of nine differential equations in dp/dt, one for each of the arterial and venous compartments. For the two compartments that represent the atrium and the ventricle, differential equations are kept as dV/dt. For these two compartments, blood pressure is computed explicitly as a function of volume (see Ventricular and atrial contraction; for a complete list of equations, see the appendix).Atrial and ventricular contraction leads to an increase in blood pressure from the low values observed in the venous system to the high values observed in the arterial system. Our model is based on the work by Ottesen and coworkers (6, 33), which predicts atrial (pla) and ventricular (plv) pressure as a function volume and cardiac activation of the form (3) The parameter a (mmHg/cm3) is related to elastance during relaxation, b (cm3) represents volume at zero diastolic pressure, c(t) (mmHg/cm3) represents contractility, and d (mmHg) is related to the volume-dependent and volume-independent components of developed pressure.The activation function g(t), which is defined over the length of one cardiac cycle, is described by a polynomial of degree (n;m): g(t) = f(t)/f(tp) with (4) where T (s) is the duration of the cardiac cycle [t̃ = mod(t;T), s], β(H) (s) denotes the onset of relaxation, H = 1/T (1/s) is heart rate, n and m characterize the contraction and relaxation phases, and pp is the peak value of the activation. The ability to vary heart rate is included in the isovolumic pressure equation (Eq. 3) by scaling time and peak values of the activation function f. The time for peak value of the contraction [tp (s)] is scaled by introducing a sigmoidal function, which depends on the heart rate (H), of the form(5) where θ represents the median, ν represents steepness, and tm (s) and tM (s) denote the minimum and maximum values, respectively. The peak ventricular pressure [pp (mmHg)] is scaled similarly using a sigmoidal function of the form(6) where φ represents the median, η represents steepness, and pm (mmHg) and pM (mmHg) denote minimum and maximum values, respectively. Finally, the time for onset of relaxation is modeled by(7) which is obtained by recognizing that tp is related to the parameter β in the isovolumic pressure model (3). Initial values for all parameters were obtained from the work by Ottesen and Danielsen (33), in which parameters were based on data from dogs. To obtain human values for the young subject studied in this work, we identified the parameters in Table 1 during our model validation.
To our knowledge, previous modeling contributions (see the introduction) assume that, during steady state (i.e., sitting, for t ≤ 60 s), the small resistances between compartments that represent large conduit vessels are constant. Nevertheless, from the theory of fluid mechanics, it is well known that the resistance depends on the radii of the vessels and that the radii themselves depend on the corresponding transmural pressure. Our investigation has shown that such dependencies are important to include in regions that represent vessels with large diameters and high blood pressure (i.e., large arteries), whereas they are less important in regions of low blood pressure (i.e., the venous system). Furthermore, these “passive” changes in diameters are also negligible in regions with small vessels (i.e., small arteries and arterioles), where autonomic responses are active and dominate the change in vessel diameters. Our previous work (29) did not include nonlinear arterial resistances; therefore, we were not able to obtain a sufficiently wide pulse pressure immediately after postural change from sitting to standing. To model nonlinearities for these resistances, we base our derivation on Poiseuille's law. For flow in a cylinder with circular cross-sectional area, Poiseuille's law predicts the resistance to flow (14) as where R (mmHg·s·cm−3) is resistance, r (cm) is radius of the vessel, η (mmHg·s) is viscosity of blood, and l (cm) is length of the cylindrical vessel. If it is assumed that length of the vessel is constant (8) The first relation comes from Poiseuille's law, the second can be obtained by assuming a fixed length l, and the third can be obtained by assuming the validity of the pressure volume relation V = Cp. In the compartmental model discussed above each compartment, a number of vessels are lumped together; as a result, we have no specific information about r. The relation in Eq. 8 implies that the resistance is inversely proportional to pressure squared. For real arteries and veins, the resistance will have maximum and minimum values. Hence, we have chosen to model this nonlinear relation using a sigmoidally decreasing function of the form(9) where RM (mmHg·s·cm−3) and Rm (mmHg·s·cm−3) are the maximum and minimum values for resistance and p (mmHg) is the blood pressure in the compartment that precedes the resistance. [In our implementation, the actual blood pressure oscillates too much; therefore, for numerical stability, we base the prediction of R on the corresponding mean arterial blood pressure, p̄(t) (mmHg).] As shown in Fig. 2, the mean arterial blood pressure oscillates with the same frequency but with smaller amplitude than pa; k represents the steepness of the sigmoid, and the parameter α2 is calculated to ensure that R returns to the value of the controlled parameter found during steady state. For k = 2, the slope of the sigmoid approximates the relation in Eq. 8. However, the relation in Eq. 8 is valid only for a steady flow. Blood flow in arteries is unsteady, and the flow through a given vessel depends on the state of the vessel. Consequently, as shown in Table 2, we should not expect that k = 2.Fig. 2.Mean arterial pressure, p̄a(t) (pam), for 45 ≤ t ≤ 90 s, computed as a continuous function by solving differential Eq. 16. Similar results were obtained for p̄au(t).
In our model 3, resistances are computed as functions of pressure: Ral(p̄au), Rac(p̄a), and Raf(p̄a). The resistance of the aorta (Rau) could also be modeled using this method. Initial investigations showed that other mechanisms, e.g., autoregulation or autonomic regulation, may also affect Rau. As a consequence, we have used an empirical model to estimate Rau (see modeling autonomic regulation and cerebral autoregulation. Cerebral autoregulation). Gravitational effects are essential during postural change from sitting to standing. Consider a cylindrical vessel with length Δz (cm) and time-invariant cross-sectional area A (cm2), i.e., dA/dt = 0. Assume that there is no velocity across the vessel and that the blood pressure is only a function position along the vessel. Hence, dv/dr = 0, where v (cm/s) and r (cm) denote the velocity and radii, respectively, and the volumetric flow rate becomes q = Av (cm3/s). Finally, assume that the drag force due to viscous shear is proportional to q. Thus the drag force per cross-sectional area unit is proportional to q; i.e., the drag force can be written as −RAq, where R (mmHg·s·cm−3) may be interpreted as the resistance. In steady state, the resistance R is given by Poiseuille's law (23) To derive the mathematical model, we proceed by balancing inertial forces with the drag force, the pressure force, and the gravitational force. The inertial force is given by where ρ = 1.055 (g/cm3) is the density of the fluid and M (g) is the mass of the fluid contained in a piece of the vessel with length Δz (cm) and cross-sectional area A (cm2; Fig. 3). Thus Newton's second law, which describes balancing of forces, gives where g=981 (cm/s2) is the gravitational acceleration. From this, it follows that (10) where L = ρΔz/A (1/s2) is the inertance and Δh = Δzcos(ψ) = hin − hout (cm) is the vertical difference of the vessel inlet (at hin where pin and pout represent pressure at the inlet and outlet, respectively). During steady state, Eq. 10 reduces to(11) When modeling postural change from sitting to standing, we substitute Eq. 11 for Kirchhoff's current law. In the limit g → 0, Eq. 11 approaches the normal form of Kirchhoff's current law given in Eq. 1. In the case of energy conservation (R → 0), Bernoulli's law for steady flow is recovered; as a result, pin + ρghin = pout + ρghout. Thus Kirchhoff's current law is still valid if we interpret p as the hydrostatic pressure p + ρgh.Fig. 3.Vessel segment with cross-sectional area A (cm2) and length Δz (cm). At one end, pressure is pin (mmHg); at the other end, pressure is pout (mmHg). Vessel is at an angle θ with respect to gravity g (g/cm2) and at an angle ψ with respect to the horizontal axis. Difference in vertical latitude is Δh = Δzcos(ψ) (cm). To capture the transition from sitting to standing, h is defined for the lower body compartments as the exponentially increasing function (12) where Tup (s) is the time at which the subject stands up, hM (cm) is the maximum height needed for the mean arterial blood pressure in the finger to drop as indicated by the data, and δ (s) is the latency for the transition to standing. In our experiments, the subjects sit with their legs elevated and the hand, where the pressure is measured, held by a sling at the level of the heart. Therefore, compartments that represent the heart and the finger are not affected by gravity. Compartments that represent the brain and the upper body are exposed to constant hydrostatic conditions, which are neglected in the current formulation. However, compartments that represent the legs are affected by gravity. Consequently, equations for the flows qal and qvl will be modified as described in Eq. 11
In the first of these equations, hin = 0 and hout = h, where h is computed using Eq. 12. In the second of these equations, hin = h and hout = 0. MODELING AUTONOMIC REGULATION AND CEREBRAL AUTOREGULATIONTwo main control mechanisms play a role: autonomic regulation and cerebral autoregulation. Autonomic regulation is mediated via the autonomic nervous system and causes changes of resistances in the vascular bed, compliance, heart rate, and cardiac contractility. Autoregulation is a local control that maintains cerebral perfusion, despite changes in systemic pressure. Autoregulation is mediated via changes in myogenic tone, metabolic demands, and CO2 concentration. Autonomic regulation is modeled as a pressure regulation where heart rate (H, beats/s), cardiac contractility (ca and cv, mmHg/cm3), peripheral systemic resistance (Raup and Ralp, mmHg·s·cm−3), and systemic compliance (Ca, Cau, Cal, Cac, Caf, Cv, Cvu, Cvl, and Cvc, cm3/mmHg) are functions of mean arterial blood pressure (p̄a, mmHg). The change in the controlled parameters is modeled using a first-order differential equation with a set-point function dependent on p̄a (13) This simple model is able to predict the observed dynamics. The parameter x(t) is controlled, xctr(pa) is the set-point function, and τ (s) is a time constant that characterizes the time required for the controlled variable to obtain its full effect. Different values of τ were used for control of cardiac contractility, compliance, and resistance (Table 2). As described earlier, autonomic regulation yields increases in peripheral vascular resistance, heart rate, and cardiac contractility. Heart rate is directly obtained from data. Hence, it is not modeled using the set-point function (13). To obtain increases in peripheral resistances (Raup, Ralp, and Rafp) and cardiac contractility (cla and clv) in response to the decrease in arterial blood pressure, the following set-point function has been used(14) A sigmoidal function was used, because it displays saturation; i.e., the function has a maximum and a minimum value corresponding to maximum dilation and maximum constriction of the vessels. In addition, vascular tone is increased, leading to a decrease in compliance in response to a decrease in arterial blood pressure. Hence, for compliance, the set-point function has the form(15) Equation 14 gives rise to a decreasing sigmoidal curve (i.e., for a decreasing pressure, the value of xctr will increase), whereas Eq. 15 gives rise to an increasing sigmoidal curve (i.e., for a decreasing pressure, the value of xctr will decrease). The parameters xm and xM are minimum and maximum values for the controlled parameter x(t). The parameter α2 is calculated to ensure that x(t) returns the value of the controlled parameters found during steady state. Initial values of parameters for k, xm, and xM are from Danielsen (5) (Table 2).These control equations (Eqs. 13–15) are formulated as functions of mean arterial blood pressure. However, our model describes the instantaneous (pulsatile) pressure. Mean values are computed as weighted averages, where the present is weighted higher than the past (16) The normalization factor N is introduced to ensure that the correct mean arterial blood pressure is obtained for pa = 1, i.e.,(17) Because our mathematical model is described by differential equations, it is more efficient to implement a differential equation to compute the mean arterial blood pressure. Hence, we differentiate Eq. 16 to obtain(18) A similar equation is used to calculate pau.On the transition to standing, cerebral autoregulation mediates a decline in cerebrovascular resistance (Racp) in response to the decrease in arterial blood pressure. In addition, the autonomic system may also play a role, by decreasing the cerebrovascular resistance due to cholinergic vasodilation or by increasing the resistance due to release of norepinephrine (7). Consequently, it is not trivial to develop an accurate physiological model that describes cerebral autoregulation. Our strategy in this work has been to use a piecewise linear function with unknown coefficients to obtain a representative function that describes the time-varying response of the cerebrovascular resistance. Once such a function is obtained, we can interpret the result in terms of the underlying physiology. To obtain such a function, we have parameterized the cerebrovascular resistance using piecewise linear functions of the form (19) where Hi represents the standard “hat” functions given by(20) The unknown coefficients γi will be estimated together with the other control parameters in Table 2. As described above, we have used a similar method to estimate the resistance Rau, which may be affected by passive nonlinear resistances and autonomic regulation.PARAMETER ESTIMATIONEstimation of model parameters has been done in a number of steps. First, we used physiological properties of the system to determine initial values for all parameters and variables (Table 1). Then we solved the steady-state problem (without including effects of gravity and regulation); i.e., we solved 11 equations of the form of Eq. 2, one for each compartment. During steady state, all resistances and capacitors were kept constant; hence, terms that involve p(dC)/dt = 0. These equations are combined with Eqs. 3–7, which determine pressures in the left atrium and ventricle, and Eq. 18, which determines the mean arterial pressures p̄a and p̄au. Finally, we estimated a constant factor used to calculate cerebral blood flow velocity vacp = qacp/fact (cm/s). We have used a constant factor (fact), because we assume that the cross-sectional area of the middle cerebral artery does not change significantly (36). These equations involve a total of 53 parameters that were estimated using a nonlinear optimization method, the Nelder-Mead algorithm, which is based on function information computed on sequences of simplexes (13). Estimated parameter values are shown together with initial values in Table 1. To obtain the best possible parameter values, we used the following cost function to minimize the difference between measured and computed values of cerebral blood flow velocity and finger pressure where v = vacp and p = paf. The superscripts d and c refer to data and corresponding computed values, respectively. In the first two sums, i = [1:N], where N is the number of data points. To compare the computed values xc and the measured data values xd (x = v,p), interpolation is used to evaluate the computed value at the same points in time where the data are obtained. Each term is divided by the number of points and the mean value of the measured data. Our model is not able to predict second-order oscillations (see Fig. 7B). The error due to poor resolution of second-order oscillations is of the same order of magnitude as the error due to poor resolution of the maximum and minimum values. However, for our modeling purpose, it is important to resolve the maximum and minimum values, but it is not important to resolve second-order oscillations. To reward good resolution of the maximum and minimum values, we have added four additional sums predicting the error between systolic and diastolic (sys and dia, respectively) computed and measured values of vacp and paf. Because of the nature of the pulse wave, only one minimum and maximum value is obtained per period; hence, i = [1:M], where M is the number of periods for 45 ≤ t ≤ 90 s. After the steady-state parameters (constant values of all resistances and compliances) were obtained, we included all equations that describe the control and ran another optimization to fit parameters that describe the control functions. This second optimization included 27 ordinary differential equations: 11 of the form of Eq. 2, 2 of the form of Eq. 18, and 14 of the form of Eq. 13. These equations are solved together with the heart model described in Eqs. 3–7, equations for passive nonlinear resistances (Eq. 9), Eq. 12, which determines the height used to calculate gravitational pooling in the veins, and the piecewise linear functions used to parameterize Racp and Rau. This second optimization gave rise to a total of 111 parameters that were optimized: 59 parameters are shown in Table 2, and 52 parameters used to parameterize Racp and Rau are shown in Figs. 4 and 5. During this second optimization, all parameters found during steady-state (i.e., during sitting, for t < 60 s) optimization remained constant (at the optimized values). In general, the inverse problem for parameter estimation does not provide a unique solution. In addition, the optimized parameters depend on the initial guesses and on the optimization algorithm. Fig. 4.Cerebral vascular resistance [Racp(t)] for 45 ≤ t ≤ 90 s, computed using piecewise linear Eq. 19. *, 26 values used to estimate cerebrovascular resistance. Shortly after transition to standing (at t = 60 s), cerebral autoregulation leads to a decrease in cerebrovascular resistance followed by an increase to a new steady-state value slightly higher than the steady-state value during sitting (for t ≤ 60 s). Fig. 5.“Passive” resistances between compartments that represent large arteries. A: Rau(t) fitted, using Eq. 19, with 26 values (*). B: Rac(t) computed using Eq. 9. Rau and Rac are depicted for 45 ≤ t ≤ 90 s. Rau and Rac increase in response to decreasing pressure and then decrease to a new steady-state value. Models for Ral(t) and Raf(t) are similar to that for Rac(t) and show similar trends. The differential equations from our mathematical model, Eqs. 2, 13, and 18, are solved using MATLAB's (MathWorks, Natick, MA) differential equations solver “ode15s.” Initial values for the resistance and compliance parameters were found from the distribution of the total blood volume between compartments and steady-state estimates for the pressure values in the various compartments. The blood volume distribution is obtained using the quantities suggested by Beneken and DeWit (3). Initial values for the resistances and compliances were based on previously reported values for blood volumes and flow rates (3), whereas blood pressure values were obtained from standard physiology literature (4). Volumes for each compartment are given by where Vunstr is the unstressed volume, i.e., the part of the volume that is not pumped out during the cardiac cycle. Therefore, initial values for compliance and resistance are calculated by
These initial values are given in Table 1. Initial values for pressures and unstressed volumes are given in Table 3.
EXPERIMENTAL DATAOur model was validated against continuous physiological data from a young subject during the transition from sitting to standing. In particular, we used arterial blood pressure measurements from the finger and arterial blood flow velocity measurements from the middle cerebral artery (15). Each subject was instrumented with a three-lead ECG (Collins) to obtain heart rate and a photoplethysmographic cuff on the middle finger of the right hand supported at the level of the right atrium to obtain noninvasive beat-by-beat blood pressure (Finapres, Ohmeda). The middle cerebral artery was insonated by placement of a 2-MHz Doppler probe (Nicolet Companion) over the temporal window to obtain continuous measurements of blood flow velocity. The envelope of the velocity waveform was derived from the fast Fourier transform of the Doppler signal, as described by Aaslid et al. (1). All physiological signals were digitized at 500 Hz (Windaq, Dataq Instruments) and stored for offline analysis. Blood pressure reduction of ∼30 mmHg on the transition to standing was used as a challenge for cerebral autoregulation. Subjects sat in a straight-backed chair with their legs elevated at 90° in front of them. They were then asked to stand. Standing was defined as the moment both feet touched the floor. Subjects performed two 5-min trials in the sitting position followed by standing for 1 min and one 5-min trial in the sitting position followed by 6 min of standing. RESULTSWe were able to obtain excellent agreement between simulations and measured data. Figure 6 shows the characteristic features of the measured data. After the transition to standing at t = 60 s, blood pressure (systolic, diastolic, and mean values) dropped significantly. At the same time, mean blood flow velocity decreased during the transition from sitting to standing (dark line through pulsatile velocity data). However, although systolic and diastolic values of pressure decreased, only the diastolic value of the blood flow velocity was diminished. The systolic values remained at baseline or were even slightly increased. This yields a significant widening of the pulsatile flow, a feature typical for young people with normal regulatory responses (15). Fig. 6.Measured arterial blood pressure in the middle finger [paf(t)], cerebral blood flow velocity [vacp(t)], and heart rate [H(t)] for a young subject for 45 ≤ t ≤ 90 s. Gray traces, time-varying values; dark traces, corresponding beat-to-beat mean values. Heart rate is obtained as follows: H = 1/T, where T (s) is cardiac cycle duration. Immediately after transition to standing (at t = 60 s), pulsatile and mean blood pressure dropped significantly, mean blood flow velocity dropped, and pulsatile blood flow velocity widened (i.e., systolic value increased, and diastolic value decreased). Initially, heart rate increased and then reached a new steady state at a higher level than during sitting. First, we evaluated our model's ability to reproduce the dynamics during steady state (i.e., during sitting, for t ≤ 60 s). We applied initial parameter values from physiological considerations (see above). Then we fitted our model [without including equations that describe resistances of large arteries as nonlinear functions of pressure (Eq. 9) and those that describe active control (Eqs. 13 and 19)] to the data set. The duration of the cardiac cycles was obtained from the ECG (Fig. 6). Simulation results in Fig. 7 show that we obtained an excellent agreement between our model and the data during steady state. However, our model is not able to resolve details of the secondary oscillations observed within each cardiac cycle (Fig. 7B), a feature that is not included in our heart model. Fig. 7.A: middle cerebral blood flow velocity and arterial finger blood pressure during sitting, i.e., for 0 ≤ t ≤ 60 s. B: magnification of 29.4 ≤ t ≤ 34.2 s in A. During steady state, vacp(t) and paf(t) were obtained by solving differential equations of the form of Eq. 2 (see appendix for all equations). Dark traces, result of our computations; gray traces, corresponding data. Our model can accurately predict blood flow velocity and blood pressure profiles while the subject is sitting. As shown in B, our model is not able to capture secondary oscillations observed in the data. The second step in validating our model is to illustrate that we can model effects of venous pooling after the transition to standing. Venous pooling results in dramatic reductions of cerebral blood flow velocity and arterial pressure (Fig. 8): with the parameters listed in Tables 1 and 2, it is possible to decrease blood flow velocity and pressure. Two observations should be noted: 1) although we did not include effects of the control, we still see an increase in heart rate, because heart rate information is obtained from the data (Fig. 6), and 2) although blood flow velocity and pressure drop immediately after standing (at 60 s), the pulse amplitude for blood flow velocity and pressure remains very narrow. Fig. 8.Cerebral blood flow velocity and arterial finger blood pressure for 45 ≤ t ≤ 90 s. Effect of standing is shown without active control mechanisms. A: blood flow velocity and blood pressure (dark traces) decrease as a result of redistribution of volumes from changes in hydrostatic pressure. Results were obtained by solving equations of the form of Eq. 2, where gravity is included, as shown in Eq. 11. B: effect of including nonlinear functions of pressure for large arterial resistances as described in Eq. 9. Immediately after standing (from 60 ≤ t ≤ 65 s), pulse pressure is much wider. Dark traces, simulated model results; gray traces, data. Next, we demonstrated the impact of the nonlinear relation between pressure and the vascular resistance of the large arteries (see modeling blood pressure and blood flow velocity. Nonlinear resistance); i.e., we let Ral(pau), Rau(pa), Rac(pa), and Raf(pa) be functions of pressure. We used the same values for all remaining parameters, and the result of this simulation is shown in Fig. 8B. The pulse pressure amplitude is higher immediately after the transition to standing (from 60 ≤ t ≤ 65 s); thus the model better represents measured values (cf. dark lines in Fig. 8, A and B, in the transition region, for 60 ≤ t ≤ 65 s). The third step involved incorporation of all active control mechanisms. Results that include effects of autonomic regulation and autoregulation are shown in Fig. 9. Our model is able to predict the change in the overall profile during the transition from sitting to standing. The only minor difference is that the data include a slight overshoot in pressure in the transition to standing. Fig. 9.Autonomic regulation and cerebral autoregulation of arterial finger blood pressure and cerebral blood flow velocity for 45 ≤ t ≤ 60 s. Model is able to reproduce data well. Dark traces, model simulations; gray traces, data. Results were obtained by solving cardiovascular equations of the form of Eq. 2, including gravity, as described in Eq. 11, passive resistances (Eq. 9), and autonomic regulation and cerebral autoregulation (Eqs. 13 and 19). The main region, where the model does not capture the dynamics of the data, is just before return to steady state during standing, i.e., for t ∼ 60 s. Autonomic regulation was included using a model that predicts parameters as a function of pressure. Although this method does not incorporate effects of sympathetic vs. parasympathetic activation, it does include net effects of neurogenic regulation. Effects of cerebral autoregulation were modeled using the empirical model described in Eq. 19. We chose to include 26 points to represent the dynamics of cerebral vascular resistance, Racp (Fig. 4). Figure 4 shows that Racp decreases because of autoregulation in response to the decrease in pressure. From earlier work (27), we expected an initial increase before the decrease; however, the model used in our previous work was much simpler than the model used in the present work. In particular, the parameter that represents peripheral resistance (Rp) in our earlier work lumps the peripheral resistance from the entire body, i.e., it combines Racp, Raup, Rafp, and Ralp. Consequently, it can be difficult to use Rp to describe the dynamics of the cerebrovascular resistance Racp, as we attempted to do in our earlier work. The resistance of the upper body (Rau) was also modeled using a piecewise linear model with unknown parameters, as described elsewhere (19). We expected that Rau may depend on autonomic regulation and may be a nonlinear passive function of pressure. This resistance follows trends predicted by remaining resistances that represent the large arteries (Fig. 5). Finally, Fig. 10 depicts the dynamics of some of the controlled variables, e.g., arterial resistance (Raup), cardiac contractility of the left ventricle (clv), and venous compliance in the upper body (Cvu). These results display quite different dynamics of the three types of variables. In particular, the compliance and peripheral resistance do not reach a steady state during the 10 s after the transition from sitting to standing (from 80 ≤ t ≤ 90 s), perhaps because the dynamics that change the ventricular contractility occur over a much faster time scale than those that affect resistances and compliances. Finally, the dynamics of other resistances, capacitors, and atrial contractility are similar to the parameters shown in Fig. 10. Fig. 10.Dynamics of controlled variables for 45 ≤ t ≤ 90 s. A: peripheral resistance in the upper body [Raup(t)]. B: cardiac contractility of the left ventricle [clv(t)]. C: compliance of veins in the upper body [cvu(t)]. Results were obtained by solving Eq. 13 together with equations for the cardiovascular system (Eq. 2). Autonomic regulation yields increase in peripheral resistance, cardiac contractility, and vascular tone. The latter yields a decrease in compliance as shown. Timing of the different controls varies; especially, note that cardiac contractility changes faster than resistances and capacitances. Regulation of the remaining resistances, contractility, and compliances showed similar responses. CONCLUSIONIn summary, we have developed an 11-compartment model that can predict cerebral blood flow velocity and finger blood pressure. This model includes a physiological description of dynamics as a response to hydrostatic pressure changes during postural change from sitting to standing. Furthermore, our model includes nonlinear functions describing resistances of the large systemic arteries as functions of pressure. To regulate blood pressure and cerebral blood flow velocity after postural change from sitting to standing, our model includes autonomic regulation using first-order differential equations regulating cardiac contractility, peripheral resistance, and vascular tone (compliance). Furthermore, we have included an empirical model describing the dynamics of cerebral vascular resistance. Validation of our model against one data set showed that, by including the mechanisms described above, our model is able to reproduce the dynamics of blood flow velocity and blood pressure needed to compensate for hypotension observed during postural change from sitting to standing. Modeling of physiological responses to standing enables a better understanding of physiological mechanisms underlying disorders related to orthostatic tolerance, e.g., orthostatic hypotension and syncope. Our model predicts that, in the absence of regulatory mechanisms (Fig. 8), blood pressure and blood flow velocity declined on the transition to standing and did not recover to baseline in the upright position. This modeling result has not been validated against data. However, similar responses have been observed clinically. For example, sustained blood pressure reduction in the upright position is seen in clinical syndromes with orthostatic hypotension associated with autonomic failure (16, 17). Different etiologies and severity of autonomic failure may lead to differences in pathophysiological responses during the transition to standing. For example, severe peripheral autonomic failure, such as pure autonomic failure or diabetic neuropathy, may be associated with orthostatic hypotension with no heart rate increment. Cerebral autoregulation, which maintains cerebral perfusion over a wide range of pressure (25), may be preserved, expanded, or reduced in orthostatic hypotension. However, cerebral blood flow would decline with impairment of autoregulation and/or when blood pressure is diminished below the autoregulated range. A transient impairment of autonomic and cerebral blood flow control is common in young people with vasodepressor syncope. This is associated with a withdrawal of sympathetic tone followed by a decline of blood pressure and cerebral perfusion (12, 22, 24). Furthermore, our results show that, by including passive nonlinear responses of resistances in the large arteries, we can obtain sufficient widening of the pulse pressure amplitude observed immediately after the transition to standing. This response is immediate and, thus, not a regulatory response but, rather, a purely passive response that occurs because of the nature of the underlying fluid dynamics. We have described an elaborate model for predicting effects of hydrostatic changes, even though this model was only validated for the transition from sitting to standing, i.e., cos(ψ) = 1. The advantage of the model derived in the present work is that it may be applicable to prediction of hydrostatic effects observed during tilt-table experiments. The main accomplishment of this work is that our model describes how autonomic regulation and cerebral autoregulation play a synergistic role in the control of arterial blood pressure and cerebral blood flow velocity. In particular, the cerebral resistance first decreases and then increases during active standing. This result is different from previous findings (27), which suggested an initial increase followed by a decrease. However, the new result is not surprising, because the present study was performed with a more complex closed-loop model. The main advantage of the closed-loop 11-compartment model presented in this study is that the cerebrovascular resistance offers a more accurate representation of the brain. For example, in previous work (27), the measured pressure was an input and only one compartment was included. Hence, the peripheral resistance was not distinguished between resistance of the body and the brain. Furthermore, the curve for Racp displays hysteresis effects: Immediately after standing, the decrease of Racp is faster than the increase for t ≤ 70 s during the phase where blood flow velocity is returning to its normal value. Hysteresis in vascular resistance in response to decreasing and increasing pressures may reflect differences between cerebral and peripheral vasculature that account for time lags between central and peripheral responses. With normal autoregulation, blood flow velocity precedes changes in peripheral blood pressure, reflecting local adjustments to intracranial pressure (26). Finally, to obtain a blood flow velocity during standing that is equivalent to that during sitting, the resistance reaches a set point that is higher during standing than during sitting. Results for parameters representative of autonomic regulation show that these parameters react as expected: peripheral resistance and cardiac contractility increase, while compliance decreases (Fig. 10). As described in results, the contractility increases much faster than the peripheral resistance. This could be due to the more rapid effects of parasympathetic withdrawal acting on contractility than of sympathetic activation, which has a later effect on contractility, peripheral resistance, and compliance. Finally, the optimized parameters depend on the initial estimates and the optimization algorithm. In particular, some of the maximum values for the resistances and compliances have large values, which are physiologically unrealistic. APPENDIXThe complete system of differential equations needed to describe all flows and pressures shown in Fig. 1 consists of 11 ordinary differential equations. For each of the nine compartments that represent the arteries and veins, we obtain differential equations of the form of Eq. 2
where each of the flows is determined using Kirchhoff's current law. The flows are as follows
Finally, differential equations for the two compartments that represent the left atrium and ventricle are given by
For these compartments, pressures are computed using the heart model (see modeling blood pressure and blood flow velocity. Ventricular and atrial contraction). FOOTNOTESThis work was supported by US-Austria-Denmark Cooperative Research: Modeling and Control of the Cardiovascular-Respiratory System Grant 0437037 from the National Science Foundation. Work performed at Beth Israel Deaconess Medical Center General Clinical Research Center was supported by National Institutes of Health Grants M01 RR-01302, R01 NS-045745-01A2, and P60 AG-08812. L. Ellwein was supported by predoctoral National Research Service Award Training Grant TR32-AG-023480 and a Statistical and Applied Mathematical Sciences Institute graduate fellowship. Data collection and analysis were supported by a Joseph Paresky Men's Associates grant from the Hebrew Rehabilitation Center for Aged, National Institute on Aging Research Nursing Home Grant AG-04390, and Alzheimers Disease Research Center Grant AG-05134. H. Tran was supported in part by National Institute of General Medical Sciences Grant RO1 GM-067299-03. REFERENCES
Page 2iontophoresis allows for the delivery of vasoactive drugs through the human and animal skin by using anodal or cathodal galvanic current application. Drugs positively charged are iontophoretically delivered using an anodal current application, whereas drugs negatively charged are delivered with cathodal current application. Coupled with laser-Doppler flowmetry, iontophoretic delivery of drugs, such as acetylcholine or sodium nitroprusside, allows for endothelium-dependent and -independent vasodilation assessments, respectively. Because endothelial function is impaired in diseases such as diabetes (43), this noninvasive assessment of vascular response is of major interest. Unfortunately, several studies have demonstrated a so-called “nonspecific” vasodilation following iontophoresis of vehicle solutions devoid of vasoactive properties, such as deionized water, with either anodal or cathodal current (2, 5, 28). This current-induced vasodilation (CIV) interferes with the study of physiological cutaneous microcirculation response to the delivered drugs (13, 14). Despite this limitation, iontophoresis remains a promising tool to assess the interplay between pharmacology and physiology of the cutaneous microcirculation in vivo in healthy volunteers or in patients (3, 23). To limit this undesirable CIV and assess the vascular response induced by the delivered drug, the study of the underlying mechanisms of this CIV deserves further investigations. The vascular response to galvanic current application is suggested to rely on an axon reflex vasodilation with either anodal or cathodal current (9, 13, 24, 46), as observed following transcutaneous electrical nerve stimulation (46). Because it seems quite unlikely that the direction of the current would differently excite nerve fibers, CIV should be the same, whatever the current polarity. However, many differences have been observed between anodal CIV and cathodal CIV, such as their kinetics and their amplitude. Indeed, cathodal CIV appears 90 s following the start of current application, whereas anodal CIV only appears after the current is stopped, suggesting that anodal current interferes with the mechanism leading to vasodilation. Neither break stimulation (7) nor anodal block hypotheses (41) were the sole explanation for the delayed onset of anodal CIV. For the same charge, expressed in millicoulomb [product of current magnitude (mA) by duration (s)], the amplitude of the anodal CIV is weaker than the one observed at the cathode (2, 27), although the reason for this difference is unknown. The exact nature of the excitatory mechanism then remains unclear. The most obvious differences between the anode and the cathode would be the pH change observed, according to current polarity. Nevertheless, although CIV is suggested as being due to the accumulation of protons under the anode, the mechanism remains unknown at the cathodal level. Many studies have been conducted to investigate CIV at the anode, but, as a result of the differences observed between anodal and cathodal CIV, it is likely that the results previously observed at the anode cannot necessarily be extrapolated to the cathode. However, it is of major interest to study cathodal CIV, because cathodal current is used in various applications, such as sodium nitroprusside iontophoresis to test the integrity of vascular smooth muscle (2, 23, 27, 28) or insulin iontophoresis to investigate the local effect of insulin on skin blood flow (SkBF) (32, 38). In these particular cases, what is the role of cathodal CIV in the total vascular response? At the anode, our laboratory has shown that, when current of 0.1 mA was applied for at least 1 min, the amplitude of the CIV increases with the duration of current application (5). This was not true at the cathode (5). However, for sodium nitroprusside iontophoresis, cathodal current applications <1 min are widely used to study the integrity of smooth muscle cells (1, 2, 22, 23, 27). Then the relationship between the amplitude of cathodal CIV and the current duration <1 min deserves to be studied. In the aim to allow for large-molecule delivery or for dose-response studies, some authors have used segmented current application (13, 27), assuming that the vascular effect of repeated current applications, resulting in a defined cumulated charge, would be lower than the vascular effect of the all-at-once current application of the same total charge. However, at the anode, we have shown that segmented current application induced a greater vascular response than the one observed with all-at-once current application corresponding to the same total charge (6). This amplification observed following segmented anodal current application is assumed to occur via sensitization of afferent nerve endings. This sensitization is a long-lasting phenomenon (at least 60 min) and is aspirin sensitive, suggesting a participation of prostaglandins (PG) (6). Last, our laboratory has shown that a single oral high dose of aspirin resulted in a prolonged decrease of the amplified response induced by segmented anodal current application (8), but it had no effect on the slow vasodilation that occurred following the first application. The vascular response induced by segmented current application and its sensitivity to aspirin have not been studied at the cathode. The aim of the present work was to study the mechanisms involved in the cathodal CIV in human skin. 1) We investigated the effects of different durations of cathodal current (<1 min) on the amplitude of CIV. 2) We hypothesized that segmented cathodal current application would result in an amplified vascular response induced by a prolonged primary afferent fiber sensitization and that this eventual sensitization mechanism would be a long-lived phenomenon, as reported for the anode. For this purpose, we studied whether CIV observed following segmented cathodal application was higher than the CIV observed following the same total charge delivered all at once. We also analyzed the influence of the interstimulation interval on the CIV. 3) Last, we analyzed whether aspirin interferes with the cathodal CIV observed following all-at-once and segmented current applications and the duration of the aspirin effect. METHODSNonsmoking, healthy volunteers with no clinical sign of, or risk factor for, vascular diseases participated in this study. Anthropometric characteristics of studied subjects are summarized in Table 1.
In the 3 wk before each experiment, volunteers took no other drugs than those proposed in the protocols. Before their participation, all subjects were informed of the methods and procedures and gave their written consent to participate in this institutionally approved study, which was conducted according to the recommendations of the conference of Helsinki. We studied the variations in SkBF in response to 0.1-mA cathodal current application through deionized water on the volar aspect of the forearms, using laser-Doppler flowmetry. The laser-Doppler flowmetry technique has been shown to accurately monitor SkBF continuously (19, 34) and is not influenced by underlying muscle blood flow (35). The technique used has been extensively described elsewhere (5). In the aim to assess SkBF, laser-Doppler probes connected to laser-Doppler flowmeters (Periflux PF4001, Perimed) were positioned on the volar aspect of the forearm skin. Each probe, also called “active” probe, was fixed to the skin with an adhesive patch designed with a sponge. Before each experiment, the sponge was wetted with 0.2 ml of deionized water and connected to the cathodal terminal of a current intensity-regulated supplier (Periiont, Micropharmacology System, PF382 Perimed, or A395 R linear stimulus isolator, WPI Instruments) for iontophoresis. Each anodal terminal was connected to an Ag-AgCl disposable electrode (Care 610, Kendall, Neustadt, Germany) fixed 5 cm apart from its respective active probe. Finally, connected to the heating system (Peritemp PF4005 Perimed), the active probe allows for local heating of the skin up to 44°C, to attain maximal cutaneous vasodilation capacity (20, 33, 36, 42). The number of active probes was chosen according to the protocol. Connected to a laser-Doppler flowmeter (Periflux PF5000, Perimed), a reference laser-Doppler probe (PF408, Perimed) was systematically positioned on the volar aspect of the forearm skin in every experiment. This reference probe was used to control the stability of SkBF at an adjacent site without electrical or local heating application. Local cutaneous temperature was measured at another site of the forearm without electrical or local heating application, using a surface thermocouple probe connected to an electronic thermometer (BAT-12, Physitemp Instruments, Clifton, NJ). The surface thermocouple probe was positioned 5 cm from active probes. Systemic arterial blood pressure was monitored with the use of a Finapres 2350 (Ohmeda, Englewood, CO) positioned on the second or third finger of the hand. For each protocol, experiments were performed with the subjects placed supine in a quiet room with the ambient temperature set at 24 ± 1°C. They rested for 15 min before the start of the experiments. On the same subject, at least 1 day elapsed between two consecutive experiments, and at least 1 wk elapsed between two consecutive protocols. A reference period of 2 min was recorded in resting conditions. After the reference period, we performed the stimulation, according to each protocol. After completion of the electrical stimulations within each protocol, a recovery period of at least 20 min was recorded to study the long-lasting effects of the current applications on SkBF, except for protocol 1, where the recovery period was 10 min. At the end of the recovery period, local heating to 44°C was systematically applied on active probes for 24 min. The aim of this protocol was to study the effects of different durations of 0.1-mA cathodal application on the amplitude of CIV in 14 subjects. The durations tested were 5, 10, 20, 30, or 40 s in a random order, with one, two, or four active probes, until eight subjects were tested with each duration. Subjects underwent from one to three experiments, but the same duration was never tested twice in the same subject. Zero-current application was performed to test for the possibility of nonspecific effects of the patch and deionized water to the skin. This protocol was performed in two parts. On one of the two active probes used, cathodal current was applied for 20 s and on the other active probe for 10 s followed by a subsequent period of current application of the same duration. All-at-once current application and the first 10 s of the segmented current application were started simultaneously. The second 10-s cathodal current application was started 10 min following the end of the first current application period. Two consecutive periods of 10-s cathodal current application were performed on two active probes with 20- and 40-min interstimulation intervals. Fourteen subjects were included in protocol 2. Among these 14 subjects, 9 performed part 1, and 9 performed part 2. Only four subjects performed both parts. Each subject (n = 8) underwent two series of three consecutive experiments after both aspirin and placebo pretreatment in a random order, resulting in six experiments per subject. At least 3 wk elapsed between pretreatments. Aspirin (Aspegic adulte 1 g; Sanofi-Synthelabo) was dissolved in 125 ml of orange juice to disguise the taste and appearance of aspirin, whereas nothing was added to the orange juice in the placebo experiments. Two hours before the first experiment of each series, subjects drank the 125-ml orange juice, blinded from the presence or not of aspirin in the glass. The experiments were conducted following each pretreatment: at 2 h (H2), day 3 (D3), and day 10 (D10). Each experiment was performed according to protocol 2, part 1. SkBF was assessed with laser-Doppler flowmeter in arbitrary units and recorded on a computer via an analog-to-digital converter (Biopac System) with a sample frequency of 3 Hz. Due to instantaneous variability due to vasomotion, individual recordings were averaged over 1-min intervals throughout each experiment. Subsequently, SkBF was indexed as cutaneous vascular conductance (CVC), calculated as the ratio of SkBF to mean arterial blood pressure over the same 1-min intervals, to take into account possible changes in systemic hemodynamic conditions. Finally, normalization of CVC to the maximum achieved in response to local heating (last minute of the heating period) was performed to better reflect changes in SkBF (21, 31). Then results were expressed in percentage of heat-induced maximal CVC (%MVC). In protocol 1, the relationship between the duration of cathodal current application and the amplitude of the vascular responses was studied with a correlation test (Pearson) at 10 min following the start of current application. For data analysis and interpretation of protocols 2 and 3, we defined the following points: rest was the last minute of the resting period, and peak was the maximal value recorded in the recovery period. In protocol 2, comparisons between rest and its respective peak values were performed with a two-tailed paired t-test. Multiple comparisons were performed with one-way ANOVA followed by Dunnett's multiple-comparison tests. In protocol 3, comparisons between segmented and all-at-once current applications, as well as between placebo and aspirin within the all-at-once (20 s) and within the segmented (10 s + 10 s) current applications, were performed with a two-tailed unpaired t-test. Comparisons between rest and its respective peak values were performed with a two-tailed paired t-test. Statistical analyses were performed with Prism (Prism 2.01 Graphpad Software). SkBF was recorded in arbitrary units and expressed as means ± SE in %MVC. A P value < 0.05 was considered significant in all statistical analyses. RESULTSCompared with rest, no significant changes were observed for control SkBF at the “reference” probe, for mean arterial blood pressure, and for local skin temperature during each experiment. Consistent with our previous observation (5–8, 41), subjects did not report painful sensations during current application in any of the protocols. Although some sensations could be noted, they had no apparent relationship with the amplitude of the vascular response. There were no significant differences in rest values under the active probes between groups for all protocols. SkBF changes recorded following 5-, 10-, 20-, 30-, or 40-s cathodal current application are represented in Fig. 1. The effect of the patch and deionized water on SkBF (called control in Fig. 1) was also recorded. Fig. 1.Skin blood flow values observed before, during, and after a single cathodal current application of 40, 30, 20, 10, 5, or 0 (control) s. Values are expressed in percentage of heat-induced maximal cutaneous vascular conductance (%MVC, means ± SE). Skin blood flow values increase with the duration of cathodal current application. A significant relationship exists between duration of cathodal current application and the amplitude of the average skin blood flow increase. Under the active probe where no cathodal current was applied and following 5-s cathodal application, CIV was not observed, whereas a single 10-, 20-, 30-, or 40-s cathodal application was followed by a progressive CIV prolonged over the whole recovery period. Our results showed a significant relationship between duration of cathodal application and the amplitude of the average CIV at 10 min following the start of current application with r = 0.99. A significant increase in SkBF occurred in response to cathodal current application after both methods of current delivery (P < 0.05 vs. rest). However, the amplitude of the peak vascular response following segmented current application with 10-min interstimulation interval is higher (79.1 ± 8.6% MVC) than the one observed following all-at-once current application (39.5 ± 4.3% MVC, P < 0.05). Following segmented current application, the amplified response was not statistically different among interstimulation intervals (86.0 ± 14.27 and 83.1 ± 8.2% MVC with 20- and 40-min interstimulation intervals, respectively). Thus the difference of cathodal CIV amplitude between all-at-once and segmented applications was observed for all tested interstimulation intervals (Fig. 2). Fig. 2.Rest and peak skin blood flow values observed following 0.1-mA cathodal current application delivered all at once (20 s) or in two consecutive 10-s intervals separated by 10-, 20-, and 40-min interstimulation interval. Values are expressed in %MVC (means ± SE). For a comparable total charge of 2 mC, peak values following 0.1-mA cathodal current application is 1) increased significantly (*P < 0.05 vs. rest), or 2) amplified if the current is delivered segmented (#P < 0.05 vs. all at once), whatever the interstimulation interval (up to 40 min). No significant differences were observed between resting values recorded, with 14.3 ± 1.7% MVC for all-at-once current application, 13.0 ± 2.6% MVC with 10-min interstimulation interval, 15.6 ± 3.7 % MVC with 20-min interstimulation interval, and 11.3 ± 1.9% MVC with 40-min interstimulation interval for segmented current application. At H2, D3, and D10 following aspirin pretreatment, a significant increase from rest of SkBF occurred in response to both all-at-once and segmented current applications. However, this CIV in response to both methods of current delivery was significantly reduced at H2 compared with placebo (P < 0.05). The difference of cathodal CIV amplitude between all-at-once and segmented cathodal current applications was abolished at H2, still reduced at D3, and present at D10 (P < 0.05). Under placebo pretreatment, the vascular responses recorded following current application at H2, D3, and D10 mimicked the responses observed in protocol 2 for comparable interstimulation interval. In brief, we observed a significant increase of SkBF in response to cathodal current application after both methods of current delivery (P < 0.05 vs. rest) and a significant difference of cathodal CIV amplitude between all-at-once and segmented application (P < 0.05) (Fig. 3). Fig. 3.Rest and peak skin blood flow values observed following 0.1-mA cathodal current application (20 s or twice 10 s), 2 h (H2), 3 days (D3), and 10 days (D10) after 1-g aspirin pretreatment or placebo. Values are expressed in %MVC (means ± SE). The skin blood flow increase in response to cathodal current application is present for all experiments (*P < 0.05 vs. rest). Comparisons between placebo and aspirin, within all-at-once and within segmented current application, show significant difference at H2 ($P < 0.05) but not at D3 and D10. The difference of peak values amplitude observed between all-at-once and segmented applications is abolished at H2, reduced at D3, and present at D10 (#P < 0.05) following aspirin pretreatment. DISCUSSIONThe present study demonstrated 1) a close correlation between the amplitude of CIV and the duration of cathodal application, when the duration is <1 min; 2) that a segmented application induced an amplification of the cathodal CIV, which lasted for long interstimulation intervals (up to 40 min); and 3) that a single oral high dose of aspirin decreased the CIV observed following cathodal application and impaired the difference of cathodal CIV amplitudes observed between all-at-once and segmented applications at both H2 and D3. Cathodal current applications are widely used to perform iontophoresis of drugs such as sodium nitroprusside (2, 23, 27, 28) or insulin (32, 38). However, in humans, the vascular response recorded following cathodal iontophoresis of drugs does not result only from the delivered drug. Indeed, cutaneous CIV is observed following cathodal iontophoresis performed with deionized water. In a previous study, our laboratory reported that the cutaneous CIVs observed with 0.1-mA cathodal current applied for 1, 3, or 5 min were all close to 75% MVC (5). This study has shown that the amplitude of cathodal CIV was not correlated with the duration of current application when the duration was >1 min. In the present study, we observed that the amplitude of cathodal CIV was correlated with the duration of current application when duration is <1 min. In brief, when delivered all at once, 0.1-mA cathodal application of <1 min resulted in a CIV proportional to the duration of current application. This result is in contrast to the absence of correlation observed with longer duration. We then assume that the maximal cathodal CIV is obtained from 1-min current application. As a result, it is likely that, at the cathode, current application duration of <1 min would be preferred to limit CIV. However, cathodal CIV is observed with duration of current application as short as 10 s and thus remains a pitfall for noninvasive assessments of microvascular response to the drugs delivered. In an effort to decrease the CIV or test increased doses of drugs, some authors have used segmented cathodal current application (13, 27). However, segmented cathodal current application resulted in an amplified vascular response compared with all-at-once cathodal current application of comparable total charge. Because a pure additive effect should not explain this amplified CIV, this suggests an increased sensitivity of nerve fibers to the electrical current, induced by the first period of current application. Indeed, CIV is assumed to rely on an axon reflex following current-induced primary afferent fiber excitation, since it was abolished following local anesthesia with Emla cream (5). Furthermore, the main fibers involved are C fibers, as CIV was decreased following chronic capsaicin pretreatments (5). Capsaicin-sensitive primary afferent excitation is followed by the release of a large variety of neurotransmitters. Among these neurotransmitters, substance P and calcitonin gene-related peptide are powerful vasodilators involved in neurogenic inflammation (25). Although calcitonin gene-related peptide and/or substance P release could explain the CIV observed following single-current application, it cannot explain the amplification observed following segmented current application. A sensitization of nerve endings should be considered as the underlying mechanism of the amplified cathodal CIV to segmented application in our experiments. Substances such as PG are synthesized and released from small-diameter sensory neurons. In parallel to their direct vasodilator effects, PG can be involved in sensitization mechanisms by binding to specific receptors that are localized on sensory neurons where they lower the firing threshold of these neurons (37). Then, following a single cathodal current application, PG could play a key role, either as direct vasodilator and/or as a potent sensitizer, resulting in an amplified CIV, if a second cathodal current application is performed. Then, using segmented current application, resulting in the same total charge as all-at-once application, increases the CIV. This sensitization results in an amplified response, even with long interstimulation intervals (up to 40 min). Then, as hypothesized, this sensitization is a long-lasting phenomenon. It has been previously described that the cathodal CIV, recorded following long duration of current application, was sensitive to aspirin. Indeed, Berliner (4) demonstrated that the major vasodilation resulting from cathodal current application was decreased following aspirin treatment. Our laboratory also observed that the CIV resulting from 5-min cathodal application was reduced following aspirin treatment (5). The results of the present study showed the aspirin sensitivity of cathodal CIV recorded following a short duration of current application. Indeed, although there is some vasodilation following 20-s or 10-s + 10-s cathodal current application in the presence of aspirin at H2, Fig. 3 indicates that it was almost abolished. Furthermore, 2 h following aspirin treatment, the cathodal CIV recorded following either all-at-once or segmented applications was reduced compared with placebo. Then, we assumed that, following short duration of current application (all at once or segmented), cathodal CIV development is, in part, an aspirin-sensitive mechanism. Aspirin leads to a direct and irreversible blockade of PG synthesis through the acetylation of cyclooxygenase (COX), wherever these PGs could be synthesized. This principal effect of aspirin could be proposed to explain the inhibition of cathodal CIV induced by 20-s or 10-s + 10-s current application. In addition to its effect on PG synthesis, recent reports suggest that aspirin may exert an inhibition of vanilloid receptors (VR1) (40) and interfere with the function of acid-sensing ion channels (ASIC) (44). VR1 and ASIC can be activated by acidosis. However, as cathodal current application induces an alkalosis (26), proton accumulation could not occur, and VR1 or ASIC inhibition by aspirin pretreatment could not explain the decrease of the vascular response recorded following cathodal current application. Although cathodal CIV is aspirin sensitive, the blockade of PG synthesis decreased but could not totally abolish the CIV resulting from 20-s or 10-s + 10-s cathodal application, since various neuropeptides released at nerve endings may exert direct vasodilator effects, independent of the PG pathway (16, 18, 45). Then the exact mechanisms involved in this nonspecific vasodilation induced by cathodal current remain to be studied. In addition to its effect on cathodal CIV from either all-at-once or segmented applications, aspirin pretreatment interferes with sensitization mechanisms. In the present study, at H2 and D3 following the single oral high dose of aspirin intake, the difference of cathodal CIV amplitude between all-at-once and segmented current applications was abolished or reduced. The second effect of aspirin in the present study was an inhibition of the sensitization mechanism induced by the first application of current, limiting the CIV amplification and suggesting a major role of PG in this sensitization. PGs are synthesized in a large variety of cells, including endothelium, smooth muscle (39), nerves (11), and platelets (29). Because a few hours are sufficient for COX to be resynthesized in endothelial or smooth muscle cells (15, 17), PGs of these origins could be involved in the cathodal CIV observed following 20-s or 10-s + 10-s current application. However, although the response at D3 following aspirin pretreatment is not statistically different from placebo, either for all-at-once or segmented cathodal current applications, the normal response to both methods appeared only to be restored at D10. Consistently, a single oral dose of aspirin had a prolonged effect on the difference of cathodal CIV amplitude between all-at-once and segmented applications. Thus, although the results from this study cannot differentiate between the roles of PG as direct vasodilator and/or as a sensitizer, it is unlikely that PGs involved in cathodal CIV are only of endothelial or smooth muscle cell origin. In nerves, COX, as other molecules, is synthesized in the cytoplasm close to the nucleus. Thereafter, molecules are transported to the periphery through active nerve trafficking at a maximal rate of 40 cm/day (12). It would be hypothesized that the time required to supply nerve endings with unblocked resynthesized COX would result in prolonged inhibition of the cathodal CIV and sensitization mechanisms. This hypothesis had been tested without success for sensitization mechanisms at the anode and would require further experiments at the cathode (8). Contrary to nucleated cells, the effect of aspirin in nonnucleated cells, like platelets, which are unable to resynthesized COX, is reversed when cells are replaced (e.g., 10 days after aspirin administration). Although there is no in vivo evidence of platelet-mediated vasodilation in humans, there is in vitro evidence of platelet-mediated vasorelaxation in animal vessels (10, 30). Whether platelets participate in this in vivo human model of CIV is a fascinating but still unproven possibility and also deserves future studies. In conclusion, previous studies have shown differences between cathodal and anodal CIV, suggesting that anodal CIV mechanisms cannot be extrapolated at cathodal CIV. As observed at the anode, the present study showed that the amplitude of the cathodal CIV increased with the duration of the applied current. This relationship between duration and amplitude was observed for periods of <1 min, since cathodal CIV reached a plateau at 1 min. As reported for anodal CIV, an amplification of cathodal CIV is observed following segmented application, and the difference of amplitude between all-at-once and segmented applications is aspirin sensitive. Indeed, this difference is abolished at H2 and D3 following aspirin treatment, suggesting that sensitization likely relies on mediators other than PGs from endothelial or smooth muscle cells. In contrast to anodal CIV, our data showed that the CIV observed following a short duration of cathodal current application (all at once or segmented) is reduced at H2 or D3 and restored at D10 following aspirin pretreatment. This finding suggests that PG of nonendothelial or smooth muscle cell origin could also be involved in cathodal CIV as direct vasodilator. Because the durations of current application used in this study are usually reported in the literature for cathodal iontophoresis of sodium nitroprusside, aspirin pretreatment could be used to decrease the vasodilation resulting from single and repeated cathodal current applications and study the specific vascular effect induced by the delivered drug. GRANTSThe present project was supported in part by Région Pays de la Loire and Centre National de la Recherche Scientifique (UMR 6188). It was granted in part by the Direction Régionale et Départementale de la Jeunesse et des Sports and Programme Hospitalier de Recherche Clinique 2001 and promoted by the University Hospital in Angers. FOOTNOTESThe authors gratefully acknowledge A. Papin for technical assistance. REFERENCES
Page 3the link between chronic sodium intake and blood pressure (BP) has been well established in large-scale epidemiological studies (6, 9, 22). However, individual responses to dietary sodium vary, with some individuals exhibiting a sodium-sensitive phenotype and others a sodium-resistant phenotype (17, 28). Both neural (8) and hormonal (18) responses may contribute to the sodium-induced change in BP, with the kidney playing a prominent role in long-term pressure homeostasis (13). Presumably, a maladaptation of one of these mechanisms contributes to the sodium sensitivity exhibited in some individuals. It has been demonstrated that black hypertensive adults have a higher degree of sodium sensitivity than white hypertensive adults (30, 32) and that sodium sensitivity increases with age (31). These data are important, since sodium sensitivity has not only been associated with increased mortality in hypertensive and normotensive adults (29) but it may also predict future hypertension in normotensive adults (26, 31). Ultimately, however, an acute or chronic increase in BP must be due to alterations in cardiac output (Q̇c), peripheral vascular resistance (PVR), or both. Although many studies link a change in dietary sodium to a change in BP, few consider the effects of diet on sodium concentration in the blood or the effects of sodium concentration on BP. Very efficient regulatory mechanisms prevent large fluctuations in sodium concentration under normal physiological conditions. Nevertheless, He et al. (15) retrospectively examined studies conducted in their laboratory where the effects of small and large changes in sodium intake were related to changes in sodium concentration. They found that increases or decreases in salt intake do indeed cause changes (albeit modest) in sodium concentration, and, at least in the longer term studies in hypertensive individuals, they found a modest but significant correlation between the decline in sodium concentration and the decline in systolic BP (15). They speculated that changes in sodium concentration affect BP directly, independent of and additive to the expected sodium-induced change in extracellular volume (7). The purpose of this study was to investigate the acute hemodynamic effects of sodium chloride. We utilized a short-term infusion of 3% hypertonic saline as a robust sodium stimulus and recorded the BP responses to this stimulus. To elucidate the mechanisms underlying acute sodium-induced changes in BP, we assessed Q̇c and derived PVR throughout the infusion. The rationale for utilizing an infusion protocol was to examine these mechanisms under well-controlled laboratory conditions. We hypothesized that an acute sodium infusion would increase BP and that the mechanisms contributing to this increase could be quantified in a group of young, normotensive adults. Furthermore, similar to what is done in the baroreflex literature (24) (i.e., acute baroreflex sensitivity), we quantified these responses by regressing arterial pressure against serum sodium, with the slope of the relationship representing the sensitivity of the response. Finally, we examined the BP response to isotonic (0.9%) saline administration, where volume was modestly expanded with no change in serum sodium concentration or plasma osmolality. MATERIALS AND METHODSThirteen healthy adults (7 men and 6 women; all Caucasian) participated in the study. The subjects were 27 ± 2 yr old, were not obese (body mass index of 23.9 ± 0.7 kg/m2), and had normal resting, seated BP (systolic BP 113 ± 3 mmHg, diastolic BP 67 ± 2 mmHg). All subjects provided verbal and written consent before study participation. The study was approved by the Human Subjects Review Board at the University of Delaware. All subjects completed a medical history form. A baseline blood sample was obtained for a complete blood count, a lipid profile (i.e., total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and triglycerides), fasting glucose, liver function (i.e., aspartate transaminase and alanine transaminase), kidney function (i.e., creatinine and blood urea nitrogen), and electrolytes (i.e., sodium, potassium, and chloride). Height and weight were measured (Healthometer scale, Continental Scale, Bridgeview, IL), and body mass index was calculated. To ensure that all subjects were healthy and had normal cardiovascular function, resting and exercise 12-lead ECGs were performed (Schiller AT-10, Electra-Med, Flint, MI). The exercise ECG was performed on an electronically braked cycle ergometer (Corival V2 Ergometer, Lode Medical Technology, Groningen, The Netherlands). The initial workload was 75 W, with a 25-W increase every 2 min; all subjects achieved at least 85% of their age-predicted maximal heart rate with a rating of perceived exertion of >18 on the 6–20 category Borg scale (10). Resting and exercise BPs were manually assessed using a mercury sphygmomanometer and stethoscope. None of the subjects used tobacco products or were taking any medications. Subjects practiced the rebreathing maneuver (see Protocol) during the screening session. Subjects were instructed to drink 15 ml/kg of water in the 24 h before the experimental day. They were also instructed to avoid caffeine, alcohol, and exercise 12 h before the protocol. Women were tested during the early follicular phase of their menstrual cycle. On the morning of the study, subjects were instructed to drink an additional 5 ml/kg of water. On arriving at the laboratory (∼0700), subjects emptied their bladders (the start of a timed urine-collection period), and specific gravity of the urine was determined. Subjects assumed a semirecumbent position, ECG electrodes were placed on the chest, and an automated oscillometric upper arm BP cuff was placed on the right arm (Dinamap Dash 2000, GE Medical Systems, Milwaukee, WI). Respiratory bands were placed around the abdomen and chest (Inductotrace System, Ambulatory Monitoring, Ardsley, NY). A 20-gauge intravenous catheter was placed in a vein in the left and right antecubital area (the left intravenous catheter was used to infuse the hypertonic solution, the right for blood sampling). A cuff was placed on the left middle finger for beat-by-beat BP assessment (Finometer, Finapres Medical Systems). The manufacture's recommended calibrations were followed. BP values from the Finometer correlate very well with directly measured radial artery BP (20). Q̇c was assessed via the indirect Fick method of CO2 rebreathing (5). During steady-state breathing, CO2 production and end-tidal CO2 were measured (TrueOne 2400 Metabolic Measurement System, ParvoMedics, Sandy, UT). Standard volume and gas calibration procedures were followed before each trial. End-tidal CO2 provided an estimate of arterial CO2. To estimate venous CO2, subjects rebreathed a high CO2-O2 gas mixture from a rebreathing bag until the level of CO2 in the bag and lung reached equilibrium (this was confirmed visually by a plateau in CO2 concentration). This equilibrium value for CO2 provided an estimate of venous CO2. The following equation was then used to calculate Q̇c: Q̇c = V̇co2(Cv− Ca)−1, where V̇co2 is CO2 production, Cv is venous CO2 concentration, and Ca is arterial CO2 concentration. PVR was derived from mean arterial pressure (MAP) and Q̇c; stroke volume was derived from Q̇c and heart rate. Five minutes of baseline data were collected and included the assessment of respiration, heart rate, beat-to-beat BP, as well as a venous blood draw. To stabilize BP, a paced breathing protocol was utilized; all subjects were verbally cued from a recording to breath at 0.25 Hz (15 breaths/min) for the 5 min of baseline data. Compliance to the paced breathing protocol was visually confirmed. Whole blood was transferred into the appropriate vacutainer tubes and spun for 15 min at 3,500 rpm in a centrifuge (Allegra X-22R, Beckman Coulter, Fullerton, CA). The serum or plasma was pipetted off and used to determine serum sodium, potassium, and chloride (EasyElectrolyte Analyzer, Medica, Bedford, MA) and plasma osmolality (model 3D3 Osmometer, Advanced Instruments, Norwood, MA). Quality control standards were run. For the determination of hematocrit, whole blood was transferred into precalibrated capillary tubes and spun rapidly on a Readacrit Centrifuge (Clay Adams Brand, Becton Dickinson, Parsippany, NJ). All samples were run in either duplicate or triplicate. The total volume of blood drawn during the protocol was 150 ml. Baseline Q̇c was also determined at this time. After baseline data collection, a 60-min infusion of 3% sodium chloride (0.15 ml·kg−1·min−1) was started. Hemodynamic variables were assessed continuously. Venous blood samples were obtained at 15, 30, 45, and 60 min. Five minutes of paced breathing commenced at 18, 33, and 50 min. Q̇c was assessed at 23, 38, and 58 min. On completion of the infusion, the subjects emptied their bladders into a urine-collection container. Urine volume, sodium, osmolality, and specific gravity were determined. For the determination of plasma norepinephrine (NE) at baseline and at the end of the infusion, whole blood was transferred into specially prepared and chilled EDTA-sodium metabisulfite vacutainers. These samples were spun at 3,500 rpm in a refrigerated centrifuge. The plasma was then pipetted off and promptly frozen in a −70°C freezer for future analysis by high-performance liquid chromotography at the Mayo Medical Laboratories; interassay coefficient of variation for the NE control was 5.5%. Five additional subjects (all men) were recruited for a direct comparison of the BP change with 0.9 vs. 3% saline infusion, with assessment of serum sodium, plasma osmolality, and hematocrit. These 10 trials were done on separate days, with 1 mo separating the experiments for each individual subject. The same infusion volume and rate were used for this comparison. The respiratory, ECG, and BP signals were collected at 500 Hz using Windaq recording software (DATAQ Instruments, Akron, OH). The ECG was peak detected, and BP waveform peak and valley were detected (Windaq waveform browser, Advanced CODAS software). Each time point reported for HR and BP represents an average of 5 min of data. The percent change in plasma volume was calculated using the following formula (12): where HctB is the hematocrit at baseline and HctA is the hematocrit at a given time point during the infusion. Urine production (ml/min) and sodium excretion (mM/min) were assessed. Free water clearance was V̇ − (Uosm·V̇)·Posm−1, where Uosm and Posm are equal to urine and plasma osmolality, respectively (mosmol/kgH2O), and V̇ is equal to urine flow rate (ml/min). Data are expressed as means ± SE. A repeated-measures ANOVA was used to determine whether there was a change in a particular variable (e.g., MAP) during the infusion (SPSS 12.0). Post hoc pairwise comparisons were performed using Fishers least significant difference procedure (Statistica for windows, release 5.1). Linear regression analysis (least squares method, SigmaPlot 8.0) was used to examine the serum sodium-MAP relationship, the osmolality-MAP relationship, and the hematocrit-MAP relationship. A single line was applied to all of the data (e.g., see Fig. 3, top) and then applied to the data on an individual basis (Fig. 3, middle). The slopes of each individual were used as indexes of MAP responsiveness to acute changes in serum sodium (Fig. 3, bottom), plasma osmolality, and plasma volume, respectively. A stepwise linear regression was also applied to all of the data (SPSS 12.0; independent variables: serum sodium and hematocrit; dependent variable: MAP). For the 0.9 vs. 3% comparison, a two-way ANOVA was utilized (time and treatment), and regression analysis was used to examine the hematocrit-MAP relationship. A P value of ≤0.05 was considered significant for all statistical tests. RESULTSBaseline blood work was within clinically acceptable normal limits. Resting and exercise 12-lead ECGs and BPs were normal. None of the subjects was hypertensive (as defined in Ref. 4). Urine specific gravity the morning of the study was 1.013 ± 0.002. As shown in Fig. 1, the infusion was successful in acutely increasing serum sodium and plasma osmolality (P < 0.01 via ANOVA; post hoc comparisons). Hematocrit declined pre- to postinfusion (0.382 ± 0.013 vs. 0.344 ± 0.010; P < 0.01 via ANOVA), corresponding to a plasma volume expansion of 19.7 ± 2.0%. There was a significant increase in finometer-derived MAP over the course of the infusion (Fig. 2; P < 0.01 via ANOVA). This increase in MAP was confirmed with the automated oscillometric upper arm cuff (P < 0.05 via ANOVA). Fig. 1.Stimulus: serum sodium (top), plasma osmolality (middle), and hematocrit (bottom) during the 60-min infusion of 3% saline. There was a significant change in each variable during the infusion (*P < 0.05 vs. baseline value via post hoc pairwise comparisons). Fig. 2.Response: mean arterial pressure (MAP; top), cardiac output (middle), and peripheral vascular resistance (derived, bottom) during the 60-min infusion of 3% saline. There was a significant change in each variable during the infusion (*P < 0.05 vs. baseline value via post hoc pairwise comparisons). The mechanisms underlying this increase in MAP varied from the early phase of the infusion to the late phase of the infusion, with changes in Q̇c (Fig. 2, middle) contributing to the early rise in MAP and changes in PVR (Fig. 2, bottom) contributing to the late rise in MAP, as shown in Fig. 2, top (P < 0.05 for both via ANOVA). This late change in PVR corresponded with a 30% increase in plasma NE (n = 8, 175 ± 13.5 to 227 ± 19.9 pg/ml; P < 0.05). There was no significant change in heart rate during the infusion (P = 0.09 via ANOVA). The change in stroke volume paralleled the change in Q̇c (P < 0.05 via ANOVA). There was no significant correlation between baseline MAP and baseline serum sodium (r = 0.03, P = 0.92). However, there was a significant relationship between the change in serum sodium and the change in MAP across all subjects during the infusion (Fig. 3, top). MAP was regressed against serum sodium on an individual basis (Fig. 3, middle and bottom). All data are shown. Three of the 13 subjects had r values of <0.80 (arbitrarily defined). The change in serum sodium, plasma osmolality, and hematocrit for these 3 subjects was similar to the other 10 subjects, although the infusion-induced change in MAP varied considerably [MAP change: 2.1, −1.1, and 13.2 mmHg; see online data supplement table for additional information ( http://jap.physiology.org/cgi/content/full/00262.2005/DC1)]. Fig. 3.Stimulus-response relationship: responsiveness of MAP to acute changes in serum sodium. Top: change in MAP plotted against the change in serum sodium across all subjects (n = 13, r = 0.46, P = 0.003). Because this plot represents the change in each variable from baseline, there are 39 data points (13 subjects, 3 data points each; 37 points are clearly visible, 2 are obscured due to overlap). Middle: example of 1 subject, where the r value was 0.94 and the slope was 2.7 mmHg/mM. Bottom: lines of the 13 subjects. The mean r value for all subjects was 0.84 ± 0.05, and the slope (represented by the thick line) was 1.75 ± 0.34 mmHg/mM. MAP was also regressed against plasma osmolality. Across all subjects, there was a significant relationship between these variables (r = 0.56, P = 0.0002). When plotted on an individual basis, the mean slope was 1.06 ± 0.19 mmHg·mosmol·kgH2O−1 and the mean r value was 0.84 ± 0.04. See online data supplement for individual data. There was also a significant relationship between the decline in hematocrit and the increase in MAP during the infusion across all subjects (r = 0.42, P = 0.008). MAP was also regressed against hematocrit on an individual basis, and the mean slope was −2.2 ± 0.35 mmHg/%, and the mean r value was 0.84 ± 0.04. See online data supplement for individual data. In the stepwise linear regression where MAP was the dependent variable, the strong colinearity between serum sodium and hematocrit (i.e., serum sodium and hematocrit; r = −0.717) prevented the overall equation from being improved by including both independent variables in the model, and of the two independent variables, serum sodium was the slightly better predictor of MAP. Because the software excluded hematocrit from this model, the final result of the stepwise linear regression equation is identical to the single linear regression result presented in the Fig. 3 legend. Urine flow rate during the infusion was 2.6 ± 0.5 ml/min, and free water clearance was −0.49 ± 0.62 ml/min. Sodium excretion was 78 ± 12 mM/min. Sodium excretion as determined at the end of the test was not correlated with the change in MAP at the end of the test (r = 0.23, P = 0.46). Figure 4 depicts the change in MAP with a 0.9% saline infusion compared with a 3% saline infusion. During the 0.9% trial, from pre- to postinfusion, there was no significant change in serum sodium (137.4 ± 1.1 to 137.5 ± 0.6 mM; P > 0.40) or plasma osmolality (291.3 ± 1.1 to 290.9 ± 1.3 mosmol/kgH2O; P > 0.40); hematocrit declined significantly (0.412 ± 0.012 to 0.397 ± 0.012; P < 0.05). During the 3% trial, there was the typical robust increase in serum sodium (137.0 ± 1.4 to 140.9 ± 0.8 mM; P < 0.05) and plasma osmolality (289.0 ± 0.8 to 298.3 ± 0.8 mosmol/kgH2O; P < 0.05), and a decline in hematocrit (0.415 ± 0.018 to 0.373 ± 0.018; P < 0.05). There was a significantly greater increase in MAP (ANOVA time and treatment effect, P < 0.05 for both), and plasma volume was expanded more during the 3% saline infusion (19.6 ± 3.6 vs. 6.8 ± 1.7%; P < 0.05). The results of the regression analysis (hematocrit-MAP relationship) for these 5 subjects demonstrated that the slopes and fits were comparable during the hypertonic infusion (average slope −3.2 ± 0.69 mmHg/%, r = 0.92 ± 0.01) to the other 13 subjects, but the results of the isotonic infusion were much more variable (average slope −1.95 ± 1.20 mmHg/%, r = 0.61 ± 0.18). In particular, two of the five subjects had r values of <0.30, preventing any meaningful comparison of the isotonic vs. hypertonic slopes. Fig. 4.Change in MAP during a 0.9% saline infusion (○) compared with the change in MAP during a 3% saline infusion (•) (n = 5). *P < 0.05 for 0.9 vs. 3.0% at a given time point via post hoc pairwise comparison. DISCUSSIONThe major findings from the present investigation include 1) an acute intravenous hypertonic saline infusion was effective in increasing BP in a group of young, normotensive adults; 2) the mechanism underlying this increase varied from the early to late phase of the infusion, that is, the early rise in BP was mediated by an increase in Q̇c and the late rise due to an increase in PVR; and 3) the responsiveness of BP to sodium and volume can be estimated using this protocol. Although we hypothesized that an acute intravenous sodium load would increase BP, we anticipated a modest response due to the young age of the subjects. Nevertheless, the present protocol resulted in an ∼10-mmHg rise in MAP from pre- to postinfusion. Twelve of the 13 subjects demonstrated an increase, and 10 of the 13 subjects had an increase of >5 mmHg. This increase was noted with both a Finometer and an automated oscillometric device. Previously published data, using a variety of infusion protocols, have found mixed results. For example, Peskind et al. (discussed below; Ref. 21) reported a BP increase during a hypertonic saline infusion, whereas Stachenfeld et al. (3, 25) and Anderson et al. (1, 2) did not. Different infusion rates and volumes, as well as individual differences in the subjects recruited, may explain these disparate findings. We conclude that a robust sodium stimulus like that observed in the present study is effective in increasing BP. With regard to the mechanisms underlying salt-induced increases in BP, much of the literature focuses on the extracellular volume expansion associated with salt ingestion (indeed, this is often referred to as volume-loading hypertension) (14). In experiments that range from days to weeks, the initial rise in BP is attributed to a volume-induced increase in Q̇c, followed by a secondary increase in PVR [via an autoregulatory mechanism related to increased flow occurring at the local tissues (14)], which also contributes to the elevation in BP. This sequence has been emphasized by Guyton and Hall (14). However, although it is possible that these mechanisms are also operative in the present acute protocol (extracellular volume was not assessed, but plasma volume, a part of the extracellular space, was expanded ∼20% with this model), we speculate that sympathetically mediated vasoconstriction also was involved. Specifically, the late increase in PVR was associated with a 30% increase in plasma NE from the pre- to postinfusion period. This finding is consistent with that of Peskind et al. (16, 21). In the context of studying panic disorder, they found that hypertonic saline increased NE concentration (and MAP). This increase is impressive when one considers that the volume stimulus, acting through the baroreflex arc, would be expected to inhibit sympathetic outflow (19, 21). However, Scrogin et al. (23), using a water-deprived Sprague-Dawley rat model, have demonstrated that elevated baseline osmolality is associated with elevated lumbar sympathetic nerve activity. Furthermore, an infusion protocol that acutely lowered plasma osmolality in these rats also lowered lumber sympathetic nerve activity. Additional support for this osmotic-sympathetic link comes from other animal-based studies that have documented the sympathoexcitatory effects of an osmotic stimulus (11, 33). The central neural mechanisms that link changes in osmolality to alterations in sympathetic outflow have been reviewed by Toney et al. (27). Briefly, neurons within the brain stem that detect peripheral alterations in osmolality project to brain stem areas known to receive baroreceptor and cardiopulmonary input (27). Collectively, it appears that plasma osmolality is a regulator of sympathetic outflow in animal models, and the increase in plasma NE in the present study is consistent with this view. Furthermore, this increase in plasma NE suggests that the sodium stimulus is distinct from that of the volume stimulus. As part of an exploratory analysis, we regressed MAP against serum sodium and, separately, MAP against plasma osmolality in an attempt to assess MAP responsiveness to sodium and osmolality. We plotted all of the data in Fig. 3, top, but our emphasis is on the individual plots, as demonstrated in Fig. 3, middle. Although the utility of expressing sodium and BP this way clearly must be confirmed in a larger group of subjects, perhaps including those with a statistically higher incidence of salt sensitivity to BP, our data suggest that there is a relationship between these variables in a group of subjects predicted (based on age and race) (32) to have a fairly low incidence of sodium sensitivity. We speculate that those predicted to have a higher degree of salt sensitivity will demonstrate a steeper sodium-MAP slope. This relationship (observed in the acute setting) does not imply causation, but, in light of the abundance of information on the link between dietary salt and BP and the recent emphasis on circulating sodium and BP (15), this issue merits additional study. It is of physiological interest to tease out the possible separate effects of sodium and volume, but in normal daily living these stimuli are intertwined. Nevertheless, Fig. 4 represents a comparison of BP responses to an infusion of 0.9% saline (where volume was expanded but serum sodium did not change) and 3% saline (where volume and serum sodium increased, as observed in the 13 other subjects). It is clear with these five additional subjects that BP increases more during a 3% saline infusion than during a 0.9% saline infusion. It is tempting to conclude that the difference between the two represents the “sodium” component of the BP increase. However, the intravascular volume stimulus is not matched with this simple comparison (experimentally, it is difficult to match the plasma volume expansion). Although we used the same infusion rate and volume for the 0.9 and 3% trials in these five subjects, plasma volume was expanded more during the 3% infusion. Presumably, with the hypertonic saline infusion, there was a greater shift in fluid from the intracellular to extracellular space. This might lead one to conclude that mainly volume is causing BP to increase (consistent with the “Guyton” explanation offered above). Consistent with this view, there was a significant relationship between the decline in hematocrit (used as a way to track the plasma volume expansion) and the increase in MAP. And, similar to the serum sodium-MAP plots discussed above, it is possible, within the confines of the current experimental protocol, to create hematocrit-MAP plots on an individual basis. We examined the data on an individual basis to see whether those with poor sodium-MAP fits had better hematocrit-MAP fits as a way to determine whether BP was related more to sodium in some and volume in others. However, this was not the case, and it is difficult to separate these two stimuli since both are changing during the infusion. Indeed, the results of the stepwise linear regression indicated a high degree of colinearity between the change in serum sodium and hematocrit. Alternate experimental paradigms will be needed to truly tease out the possibly separate effects of sodium and volume on the BP increase in humans. But, consistent with the recent hypothesis put forth by He et al. (15) and de Wardener et al. (7), we believe the statistically significant relationship between circulating sodium and MAP provide preliminary support for the concept that circulating sodium contributes to the increase in BP. This view does not discount the volume contribution to the increase in BP but rather supports the conclusion that both serum sodium and volume may be involved in the BP increase. There are several limitations that should be mentioned. First, the present cohort contained both men and women, and with a total sample size of 13 subjects, it is not possible to fully explore possible sex differences in BP and hemodynamic responses to a sodium and volume load. There were no statistical differences or trends in any of the variables assessed, but we hesitate to make any firm conclusions regarding the lack of sex differences with the present cohort. Second, we are not able to link what occurs in the acute setting (using the infusion protocol) to what occurs in the chronic setting (using dietary manipulations of sodium). Therefore, any discussion on the similarities of these two distinct perturbations is speculative in nature. Third, we did not assess baseline extracellular volume or the change in the extracellular volume in response to the infusion. Including an assessment of the extracellular volume would have improved our ability to discern sodium- vs. volume-induced alterations in BP. We conclude that a short-term infusion of hypertonic saline in humans permits the assessment of stimulus-response characteristics of serum sodium and volume and BP. Both sodium and volume appear to be related to the increase in BP. The early hypertonic saline-induced increase in BP is mediated by an increase in Q̇c and the late increase via changes in PVR. The increase in plasma NE provides indirect evidence for the involvement of the sympathetic nervous system. The novelty of this model for studying BP regulation is that some of these mechanisms can be examined during a very short-term infusion protocol. Future studies will need to relate what occurs in the acute setting using intravenous fluids to what occurs in the chronic setting using dietary manipulations of salt. GRANTSThis research was supported by the University of Delaware Research Foundation and National Heart, Lung, and Blood Institute Grant 1 R15 HL-074851-01. FOOTNOTESThe authors thank the subjects for participation in this study. The assistance of Ryan T. Allen, Angela Disabatino, Cheryl Katz, and Steven Johnson is gratefully acknowledged. REFERENCES
Page 4sympathetic nervous activity in humans is closely related to the level of gravitational stress on the cardiovascular system. The baroreflexes adjust sympathetic activity to maintain a constant blood pressure despite variations in venous return of blood to the heart. Thus sympathetic activity is high when venous return is low such as in the upright position, whereas it is suppressed when venous return is increased such as during water immersion and antiorthostatic maneuvers. Head-down bed rest (HDBR) has been applied to simulate cardiovascular changes during microgravity. In 1995 our laboratory reported, however, that plasma norepinephrine values were elevated during microgravity and above values obtained in the seated position on the ground (10). Blood samples were collected from an antecubital vein on the fifth to sixth day of flight. Later Ertl et al. (4) reported that baseline sympathetic activity as measured by microneurography was increased moderately by spaceflight. Furthermore, in the same study it was demonstrated that the norepinephrine spillover rate was significantly increased in space. These studies therefore suggested, contrary to expectations, that sympathetic activity is increased by spaceflight of 1- to 2-wk duration. The aim of the present study was therefore to evaluate long-term changes in sympathoadrenal activity during HDBR and during microgravity by measuring platelet norepinephrine and epinephrine values. The HDBR study and ambulatory control study were done both with the subjects on a normocaloric diet and on a hypocaloric diet (−25%). The hypocaloric diet experiment was performed because cosmonauts tend to have a reduced calorie intake in space compared with subjects in ground-based observations, which might modulate sympathetic nervous activity. Platelets circulate through all parts of the body and take up catecholamines from plasma. Platelet norepinephrine and epinephrine values are unaffected by acute changes in sympathoadrenal activity such as exercise and reflect chronic, more long-term changes (1, 2, 15). Furthermore, platelet epinephrine measurements may be a more reliable technique for detection of changes in epinephrine release than comparable measurements of epinephrine in forearm venous blood, because the platelet epinephrine concentration does not depend on the extraction ratio in forearm tissues. Platelets take up catecholamines from plasma in vivo, both norepinephrine and epinephrine. This is an active process. The final concentration of norepinephrine and epinephrine in the platelets is primarily dependent on the average plasma catecholamine concentrations. This is known especially from patients with pheochromocytoma, who have high values of platelet catecholamines in proportion to the plasma level. The concentration in the platelets decreases after the tumor as been removed although more slowly than in plasma (19). Long-term changes in sympathoadrenal activity can also be recorded by multiple sampling procedures, but such procedures cannot easily be applied to spaceflight. We did not apply urine measurements of norepinephrine and epinephrine because renal function is influenced by microgravity and urinary excretion rate of water and sodium are decreased during spaceflight but not during HDBR (10, 11). SUBJECTS AND METHODSAll study protocols were reviewed and approved by Ethics Committees at the European Space Agency medical board and were in compliance with the Declaration of Helsinki II. All subjects provided informed consent to the procedures. Nine normal subjects participated in all four study phases (Table 1). One subject participated only in phase 1 and another subject participated in the following three study periods. The mean age was 23.8 yr (range 21–29 yr). The mean body mass index was 23.0 kg/m2 (range 19.2–27.8 kg/m2). All subjects were healthy and had a normal heart rate and blood pressure.
Study phases 1 and 2 started with an adaptation period of 9 days (−9 to −1) when the subjects were ambulatory. This was followed by an intervention period of 14 days (+1 to +14), when the subjects at random were either ambulatory or subjected to −6° HDBR and vice versa. Study phases 3 and 4 were performed in the same way except that all subjects were on a hypocaloric diet during the intervention period. Phases 2 and 4 were the HDBR study, and phase 1 and 3 the ambulatory study. The daily normocaloric diet consisted of protein (1 g per kg body wt per day), fat (30% of the energy; fatty acid composition was saturated and polyunsaturated fatty acids), and carbohydrate (remaining calories). In addition, the subjects received 50 ml of water per kg body wt, 2.5 mmol sodium/kg, 1,000 mg calcium, and 400 IU vitamin D per day. The hypocaloric diet had an energy intake of 75% of the respective normocaloric ambulatory diet. All subjects received the same amounts of water, protein, sodium, calcium, and vitamin D. Intake of alcohol and caffeine was not allowed. All other nutrients without experiment-specific requirements matched dietary recommended intake levels of the German Nutrition Society. In the adaptation period in the four phases all subjects received the normocaloric diet of identical nutrient composition. Total energy expenditure was calculated as basal metabolic rate multiplied by the physical workload plus the calculated thermic effect of feeding. The test subjects stayed in the laboratory all the time also during the ambulatory study phases. They were not allowed to do any exercise on a voluntary basis. However, in phases 1 and 3 they followed an exercise protocol that was two times 15 min of bicycle ergometry (∼125 W). Blood samples for plasma and platelet catecholamines were collected from an antecubital vein. The samples were always collected in the morning at 7 AM with subjects in the supine position. The blood samples were immediately brought to the laboratory and prepared as described below. Blood samples were obtained on days −4 and −2 in the adaptation period and again on days +5, +9, and +14 during the intervention period. No samples for cateholamine analysis were obtained in the recovery period. For all practical reasons, the preparation of blood samples differed in the HDBR study compared with the microgravity study. For this reason we have not compared absolute values in the two groups. Relative changes observed in relation to the corresponding basal values in the adaptation period and preflight may be compared. Ten milliliters of blood were collected in polycarbonate tubes that contained 50 μl of 0.2 M NaEDTA solution per milliliter of blood. The EDTA blood was centrifuged without brake for 15 min at 20°C at 350 g. The upper two-thirds of the plasma, ∼3 ml, was transferred to a new tube and mixed gently. The number of platelets was counted in a Coulter counter and expressed as number × 106/ml. Samples of 108 platelets were collected and pipetted into Eppendorf tubes. These tubes were centrifuged for 15 min at 4°C at 1,800 g with brake. Supernatants were decanted and platelets frozen and stored at −20C° until analysis. The loss of platelets with the second centrifugation procedure and the decantation procedure was ∼4%. It is important that no catecholamines are lost from the platelets during the preparation procedure. There was no agglutination of platelets before the final centrifugation and of course no coagulation (owing to the binding of Ca2+ to EDTA). In addition, we have analyzed platelet norepinephrine and epinephrine in 12 samples obtained from two subjects. Four samples were analyzed by the standard procedure; i.e., centrifugation was done without any delay. Four samples were allowed to stand for 25 min before centrifugation, and an additional four samples were centrifuged after a delay of 50 min. There was no decrease in platelet norepinephrine and epinephrine with time. The mean platelet norepinephrine concentration (±SE) was 62 ± 5, 62 ± 2, and 70 ± 2 pg/108 platelets with the standard procedure and with a delay of 25 and 50 min, respectively. The corresponding values for platelet epinephrine were 5 ± 1, 7 ± 2, and 8 ± 1 pg/108 platelets, respectively. None of these changes was significant. 2.5 ml of blood was collected in tubes as described above and centrifuged for 10 min at 1,800 g. The plasma was collected and stored at −20°C until analysis. Blood samples for platelet measurements were collected from an antecubital vein in five male cosmonauts. The mean age was 41 yr (range 37–45 yr). They participated in three Soyuz missions to the International Space Station. Samples were collected ∼14 days before launch, after 11–12 days in flight within 12 h upon landing, and finally at least 14 days thereafter. Samples for platelet norepinephrine and epinephrine measurements should preferably have been obtained inflight, but this was not possible because no centrifuge with an adjustable speed was available on the International Space Station. Because of the long half-life of platelet norepinephrine and epinephrine (see below), a sample taken after 11–12 days in flight and within 12 h after landing would still reflect the microgravity state. We tested the half-life of platelet norepinephrine in five normal subjects during the first 4 days of another HDBR study. The mean half-life for platelet norepinephrine was 54 ± 12.5 h (±SE). There was a tendency for an inverse relationship between the half-life and the basal platelet norepinephrine values. The half time is of the same magnitude as reported by Chamberlain et al. (2) (44 h). The platelets could not be counted at the sampling site, and therefore the preparation of the platelets had to be modified. After the initial centrifugation at 350 g, samples of 0.5 ml were added to Eppendorf tubes and centrifuged as described above (1,800 g with brake), decanted, and frozen at −20°C. In addition, at least two times 0.3 ml plasma samples were obtained and added to Eppendorf tubes. These samples were not centrifuged but were frozen and later applied for counting the number of platelets. A preliminary study indicated that the number of platelets remained the same before and after freezing. The mean platelet level in the cosmonauts preflight and within 12 h after landing averaged 243 ± 20 and 261 ± 56 × 109/l. These values were not significantly different and were well within normal range (140–340 × 109 platelets/l). Plasma and platelet norepinephrine and epinephrine concentrations were quantified by a sensitive and precise radioenzymatic assay described earlier (8). Results are presented as means ± SE. For the analysis of data obtained in the same group of subjects at different time intervals, we applied one-way repeated-measures ANOVA. Pairwise multiple comparison procedures were done with the Tukey's test. The paired t-test, the t-test, regression, and correlation analysis were applied in some tests. Statistical analysis was done with the SigmaStat version 2.0. A P value < 0.05 was considered significant. RESULTSPlatelet norepinephrine decreased significantly during HDBR (phase 2; Fig. 1; P < 0.001). Fig. 1.Platelet norepinephrine, pg/108 platelets, during adaptation and intervention periods in the normocaloric experiment. Results are presented as means ± SE. Solid bars, ambulatory, phase 1 (not significant); shaded bars, head-down bed rest (HDBR), phase 2. P < 0.001 for a decrease in platelet norepinephrine during HDBR. The tendency for platelet norepinephrine to decrease during the normocaloric ambulatory study was not significant. The hypocaloric diet had no effect on platelet norepinephrine levels, which decreased during the HDBR (phase 4; P < 0.001) but remained unchanged during the ambulatory study period (Fig. 2). Fig. 2.Platelet norepinephrine, pg/108 platelets, during adaptation and intervention periods in the hypocaloric experiment. Results are presented as means ± SE. Solid columns, ambulatory, phase 3 (not significant); shaded columns, HDBR, phase 4. P < 0.001 for a decrease in platelet norepinephrine during HDBR. The mean platelet norepinephrine level in the four experiments in the adaptation period before the intervention averaged 42.9 ± 9.8 (mean ± SE; phase 1), 41.2 ± 7.4 (phase 2), 34.4 ± 8.3 (phase 3), and 32.9 ± 8.3 (phase 4) pg/108 platelets. The values obtained in the adaptation period in phases 1 and 2 tended to be higher than in phases 3 and 4 (one-way repeated-measures ANOVA + Tukey's test; P < 0.05). Platelet norepinephrine levels varied between individual subjects, but values in the same subject in the two samples obtained in the four adaptation periods were correlated [R values ranging from 0.98 to 0.60, P < 0.05 to 0.000 in all but two comparisons (n = 28 comparisons)]. There was also a strong positive correlation between platelet norepinephrine values in the adaptation period and during the intervention (phase 2, r = 0.92, P < 0.001; phase 4, r = 0.95, P < 0.001), indicating that the relative decrease in platelet norepinephrine was approximately the same in all subjects. The corresponding values for platelet epinephrine in the adaptation period were 2.7 ± 0.7 (phase 1), 2.9 ± 0.9, 2.3 ± 0.8, and 2.6 ± 0.5 pg/108 platelets (not significant). Platelet epinephrine did not change significantly during HDBR and was not influenced by the hypocaloric diet. Table 2 shows plasma norepinephrine during the four phases. In the HDBR with normocaloric diet plasma norepinephrine decreased significantly, but the decrease occurred already between the first and second sample in the adaptation period. A similar response was seen in the adaptation period phase 4. During phase 3 (hypocaloric and ambulatory), plasma norepinephrine increased significantly at the end of the intervention period. Thus there was no change in plasma norepinephrine that could be related to HDBR. Plasma epinephrine values were low, with mean values ranging from 0.00 to 0.02 ng/ml. No significant differences were observed.
Figure 3 shows platelet norepinephrine values in the cosmonauts. Fig. 3.Mean platelet norepinephrine values ± SE in the 5 cosmonauts. One value was missing postflight. Preflight values were within normal range but in the lower end. Epinephrine values averaged preflight 1.5 ± 0.3, inflight 3.8 ± 1.1, and postflight 2.1 ± 0.2 pg/108 platelets. During microgravity, platelet norepinephrine and epinephrine increased in four of the five cosmonauts, but the change was not significant. Platelets from subjects participating in the HDBR study and from the cosmonauts were processed and stored in different ways as described earlier, and a comparison of the absolute values may not be relevant. The relative changes may, however, be compared. Platelet norepinephrine during microgravity and during HDBR expressed in percentage of basal values (preflight or pre-HDBR values, respectively) were significantly different (152.6 ± 28 vs. 59.8 ± 5.7%, P < 0.004) for a difference between inflight and HDBR (Fig. 4). Fig. 4.Platelet norepinephrine values, pg /108 platelets, inflight and during HDBR expressed in percentage of basal values [preflight values or pre-HDBR values (phase 2), respectively]. *P < 0.004 for a difference between inflight and HDBR. Comparison of inflight values with values from the phase 4 study was also significant (152.6 ± 28 vs. 57 ± 6.6%, P < 001). Comparing platelet epinephrine in the same way as norepinephrine indicated that platelet epinephrine was significantly different during microgravity compared with the HDBR experiment [293 ± 85 vs. 90 ± 12% (phase 2) and 89 ± 18% (phase 4), P < 0.01 and 0.02, respectively]. Thus there was a marked and highly significant difference in platelet norepinephrine and epinephrine responses during microgravity compared with HDBR. The lack of a decrease in platelet norepinephrine in cosmonauts compared with participants in the HDBR study cannot be explained by the relatively lower preflight values. In the phase 4 study four subjects had mean values in the adaptation period below 20 pg/108 platelets (range from 5 to 17.5), and all values decreased during HDBR. DISCUSSIONThe different sympathoadrenal response to HDBR compared with microgravity is most likely related to cardiovascular changes. Both microgravity and HDBR cause an increase in the end-diastolic volume at the beginning of exposure to these conditions. The underlying mechanism may, however, be different. During microgravity, the increase in central transmural venous pressure is due to an increase in venous return and to a more negative intrapleural pressure (17, 18), whereas the increase during HDBR is only due to increased venous return. Intravascular volumes decrease during both conditions and to approximately the same extent by 12% (9, 14). Later on there may also be remodeling of the heart muscle and some atrophy (13). Cardiac ouput is increased during weightlessness in parabolic flights and during the early phase of microgravity but gradually decreased during prolonged exposures (16; Norsk P, unpublished observations). We did not observe any changes in plasma norepinephrine during HDBR, but it must be noted that all samples were obtained with the subjects in the supine position. Forearm venous plasma norepinephrine is correlated to muscle sympathetic nerve activity (3), which also has been found to be unchanged during short-term HDBR (7, 12). Goldstein et al. (5) observed that urinary excretion rates of norepinephrine decreased during HDBR compared with the supine position. In the present experiment, platelet norepinephrine, but not epinephrine, decreased markedly during HDBR. The decrease in platelet norepinephrine during HDBR is most likely due to the fact that the subjects were not allowed to sit up or stand up during HDBR and they were not allowed to perform exercise. Several studies have now demonstrated, contrary to expectations, that sympathetic nervous activity is not decreased during microgravity, and it is most likely increased compared with ground-based values. In 1995 our laboratory reported that plasma norepinephrine values were elevated inflight and above values observed in the seated position in ground-based experiments (10). Ertl et al. (4) concluded that baseline sympathetic neural outflow was increased moderately inflight. Furthermore, in the same study it was demonstrated that the norepinephrine spillover rate was significantly increased in space. In this study, the steady-state concentration of the norepinephrine tracer was measured in venous blood and not in arterial blood, and the calculated clearance values are therefore too high and to some extent dependent on variations in the local uptake of the tracer in the forearm tissue. Results of the present study are in accordance with our previous study in which we showed that plasma norepinephrine concentrations were increased during microgravity. Thus results from all three studies, which applied different techniques to study sympathetic nervous activity, support the concept that sympathetic activity is moderately increased during microgravity. The platelet measurements showed high epinephrine values during microgravity that were not observed in our previous study (10). The reason may be that in the first study samples were obtained from a forearm vein and epinephrine in arterial blood is extracted by forearm tissues (3). Platelets circulate through all parts of the body and take up catecholamines from plasma. Platelet epinephrine may therefore be a more reliable index of epinephrine release in the body than epinephrine in forearm venous blood. The exact interrelationship during microgravity between the initial increase and gradual decrease thereafter in cardiac output and plasma volume and the increment in sympathetic nervous activity during spaceflight remains to be elucidated. Furthermore, the difference in the norepinephrine response to HDBR and microgravity should also be explained. Most likely the decrease in plasma volume inflight plays a major role for the increase in sympathetic nervous activity. There does not appear to be a pronounced early increase in urine output during weightlessness, but there may be a relative increase compared with the intake of fluid, because fluid and food intake decreases (14). The reduction in plasma volume during HDBR has little influence on basal sympathetic nervous activity as long as the subjects are supine. After prolonged bed rest the subjects have a tendency to develop orthostatic hypotension, which largely can be corrected for by fluid intake (6). The relationship between cardiac output and sympathetic nervous activity during spaceflight is more difficult to explain. The dilatation of the cardiopulmonary area during spaceflight should inhibit sympathetic nervous activity, but at the same time the distended central vasculature would induce a decrease in vascular compliance. This is probably also what occurs during the early period of spaceflight as observed during parabolic flights (17). During the subsequent decrease in cardiac output and stroke volume, arterial pulsation will decrease. This decrease in pulsation combined with a decrease in compliance of the central vascular wall may activate sympathetic nervous activity. We cannot exclude the possibility that changes in sympathetic activity during microgravity are due to a decrease in the sensitivity to catecholamines, but this suggestion can hardly explain the difference between bed rest and microgravity. Furthermore, the sympathetic response during spaceflight is also unlikely to be an arousal reaction due to mental stress, because neither blood pressure nor heart rate increased inflight. In conclusion, a relative high sympathoadrenal activity compared with preflight values seems to be an integrated part of the regulatory response to microgravity. Furthermore, HDBR cannot be applied to simulate changes in sympathoadrenal activity in humans during microgravity. Platelet norepinephrine values were measured in blood samples obtained early after landing (<12 h). Because of the long half-life of norepinephrine in platelets of ∼2 days (Ref. 2 and the present study), values obtained early postlanding will still reflect the microgravity state. Although there is likely to be an increase in sympathetic activity during the return of the crew to Earth, increments in plasma norepinephrine are not likely to be very high owing to vasoconstriction, which will retain norepinephrine in the extracellular fluid. Furthermore, it has been shown in several studies that platelet norepinephrine and epinephrine are unaffected by acute short-term increments in sympathoadrenal activity such as exercise (1, 2), in which there is a marked increase in plasma norepinephrine and epinephrine. In further studies of sympathoadrenal activity during microgravity, platelet catecholamines should be measured in samples obtained in space. This can be done provided there is an access to a centrifuge with an adjustable speed on the international space station. The preparation of platelet samples was different during HDBR compared with microgravity. The platelet norepinephrine levels in the preflight samples were also in the lower end of the normal range observed in the HDBR study. We cannot exclude that the loss of cateholamines from the samples may have been greater during the microgravity study compared with HDTB. The platelet number was, however, normal in the samples obtained from the cosmonauts and not different in the preflight and early postlanding sample. All samples in the microgravity experiments were prepared in the same way, and the relative changes in platelet catecholamines between the two groups can therefore be compared. It is unclear whether the relatively low platelet catecholamine values found in the cosmonauts in preflight samples is a characteristic finding in this group of subjects. If this is the case it may indicate that cosmonauts have a relatively large plasma volume on Earth, because in normal subjects sympathetic nervous activity and plasma volume are inversely related (3). GRANTSThis study was supported by the Danish Research Councils with grant no. 2006-01-0012. FOOTNOTESUlla Kjærulff-Hansen is thanked for excellent technical assistance. REFERENCES
Page 5in the world of sports most disciplines require some degree of both strength and motor skill for the athlete to be successful, and athletes often use a combination of resistance training and skill learning to optimize their performance. Whereas plastic changes in the central nervous system are well documented in relation to acquisition of new skills (59, 61), such neural adaptations are generally reported mainly to take place in the initial stages of strength training (21), and their significance for the increased strength compared with the well-documented muscular changes is still debated (8, 16, 18, 20, 36). There is now increasing evidence suggesting that plastic changes in the primary motor cortex play an important role in skill acquisition. Motor skill learning has thus been demonstrated to be associated with anatomical and physiological changes within the primary motor cortex in primates and nonprimate animals (31–33, 41, 53, 65). In humans, neuroimaging techniques and transcranial magnetic stimulation (TMS) have demonstrated that motor skill training induces changes in the organization of movement representations in the primary motor cortex in the form of expansion and increased excitability of the cortical representation of specific muscles (or movements) involved in the tasks (10, 11, 17, 27, 30, 35, 43–45, 47, 48). In contrast to skill acquisition, nonskill training or passive motor training is mainly reported to elicit no or only minor changes in excitability (35, 45). For instance, Plautz et al. (50) demonstrated in squirrel monkeys that movement repetition in the absence of motor skill acquisition was not sufficient to produce changes in cortical representational organization. On the basis of these findings, learning thus seems to be a prerequisite or an important factor in driving cortical representational plasticity related to motor experience. To fully acknowledge this hypothesis, however, it remains to be elucidated which aspects of motor experience relate to motor learning or how motor learning is to be defined. Little is known about the neuronal mechanisms involved in the increased neuronal drive in the early stages of strength training, although it has been suggested that increased cortical drive to the spinal motoneurons may be of importance (1). Strength increments arise as a consequence of numerous factors, but in many ways it would make sense to consider strength training as a kind of motor learning process. As reviewed by Carroll et al. (8), strength training relates to motor learning because of the fact that athletes learn to produce muscle recruitment patterns associated with optimal performance of the specific task. It is thus likely that strength training in parallel with motor learning can lead to improved muscular coordination. Considering this, it would be reasonable to assume that similar plastic changes in the primary motor cortex as reported for acquisition of new motor tasks are also involved in the improved ability to generate force in the early stages of strength training. Remple et al. (53) demonstrated in the rat that training of skilled reaching movements involving either a progressive increase of the maximal load (strength) or a control condition induced a similar degree of plastic changes in the motor cortical movement representations. This finding indicated that the observed plasticity related to the development of skilled movements rather than increased muscle strength per se (64). Electron microscopy of the ventral spinal cord showed that rats training power reaching had a significantly greater density of synapses onto spinal motor neurons than both a normal reaching group and a nonreaching control group, which led to the suggestion that strength training is supported by spinal cord synaptogenesis (64). The only study that has addressed the issue of supraspinal adaptations to strength training in human subjects observed a decrease rather than an increase in corticospinal excitability in relation to strength training index finger abduction (7). Isolated finger abduction may, however, have little relevance to normal strength training, which often involves complex exercises involving large proximal muscle groups in combination with distal muscle groups, as is seen in biceps curl. We thus speculate that “normal” strength training involving more complex muscle recruitment patterns and a more prominent role of muscular coordination may have the potential to induce learning-related phenomena in the central nervous system. In the present study, we used TMS to investigate whether 4 wk of strength training of the biceps brachii (BB) muscle is associated with increased excitability of corticospinal projections to the muscle. Similar measurements were also made in relation to the acquisition of a difficult motor task requiring precise control of elbow joint movements. We found that increased corticospinal excitability was only observed in relation to acquisition of the difficult motor task, whereas 4 wk of strength training that increased the dynamical strength of the biceps muscle by 31% caused a depression of corticospinal excitability at rest. A significant correlation between the changes in corticospinal excitability and motor performance was only observed for the skill-trained subjects. METHODSThe experiments were performed on 24 healthy volunteers (11 women, 13 men) with an average age of 25 ± 5 yr. All subjects gave their written, informed consent to the experimental procedures, which were approved by the local ethics committee. The study was performed in accordance with the Declaration of Helsinki. All subjects were right-handed according to the Edinburgh Handedness Inventory (42), and no volunteers had any history of neurological disease. The subjects were randomly allocated to three groups, two different motor training groups and a control group (n = 8, 5 men and 3 women) that did not train but participated in all testing procedures. Motor training consisted of either strength training (n = 8, 4 men and 4 women) or visuomotor skill learning (n = 8, 4 men and 4 women). There were no age differences between the three groups. Thirteen training sessions were performed by the participants over a 4-wk training period. At the beginning of the training period, after 2 wk, and at the end of the training period, each volunteer participated in a longer lasting experimental session. These experimental sessions involved 1) strength tests evaluating the maximal voluntary dynamic and isometric elbow flexor muscle strength of the subjects, 2) an electrophysiological testing procedure involving peripheral electrical nerve stimulation and TMS at rest and during tonic contraction, 3) one training session of either motor skill training or strength training, and 4) repeated measures of TMS and peripheral electrical nerve stimulation after training. This experimental design aimed at investigating short-term adaptations to training defined as the effect of a single training session as well as long-term adaptations to training defined as the effect of training 2–4 wk. For every testing and training procedure involved in the study, subjects were familiarized with the equipment and the measuring procedures on separate occasions before data sampling. Six months after completion of the training, it was possible to retest four of the subjects in the skill learning group to investigate reversibility of the training-induced phenomena. At this occasion, TMS and peripheral electrical nerve stimulation were applied. At the beginning of the testing procedure, the subjects' maximal dynamic muscle strength was determined as one-repetition maximum (1 RM) biceps curl and the maximal isometric muscle strength was determined as the peak torque of a maximal voluntary contraction (MVC). For the 1 RM test, the subjects were standing in a standardized position at a custom-made biceps curl bench. Before the test, subjects performed a warm-up procedure and received instructions in how to perform unilateral biceps curl. During the test, the subject was handed a submaximal weight and performed one extension-flexion cycle of the elbow joint with the forearm supinated. As this task was completed, the load increased progressively until failure of the biceps curl occurred. 1 RM was determined as the highest load at which the task was fulfilled. In all tests, the subject performed 5–8 trials with increasing load depending on maximal strength. Because of a large similarity between the 1 RM test and the strength training paradigm, it was hypothesized that the 1 RM test could be influenced by effects of learning (28, 58). Therefore, in addition to the 1 RM test, a MVC test was used as a control to validate any training-induced alterations in the maximal strength of the subjects. For the MVC test, subjects were seated in a custom-built rigid chair and firmly strapped to an upright backrest. The elbow was flexed to 90° and the forearm was supinated and rested on a table. A nonelastic strap around the wrist was connected to a strain-gauge transducer. Subjects were instructed to perform a maximal contraction of the right arm elbow flexors by increasing the torque to maximum within a few seconds and then to exert maximal torque for 2 s, while maintaining the standardized position. Verbal encouragement and visual feedback of the torque exerted were provided. Typically four or five successive trials were performed until the peak torque did not increase any further. The peak torque recorded in either of the trials was taken as the MVC. Strength measurements were only obtained during the first and the last of the three testing sessions. After completing the strength tests, subjects were seated in an armchair for the electrophysiological testing procedure involving TMS and peripheral electrical nerve stimulation. Subjects were positioned with the head supported and the examined right arm fixed on a cushioned arm support, the shoulder joint flexed 45° and the elbow joint almost fully extended. Surface electrodes were used for electrical nerve stimulation and recording of electromyographic activity (EMG). EMG activity was recorded from the BB and triceps brachii muscle by nonpolarizable bipolar Ag-AgCl electrodes (1 cm2, interelectrode distance 1 cm). The amplified EMG signals were filtered (band-pass, 25 Hz to 1 kHz), sampled at 2 kHz, and stored on a personal computer for offline analysis. Furthermore, the EMG was full-wave rectified, integrated, and displayed to the subject as visual feedback during tonic contraction. Motor evoked potentials (MEPs) were evoked by TMS of the left hemisphere (contralateral) motor cortical arm area at the hot spot for activation of BB by using a magnetic stimulator (Magstim 200, Magstim) with the capability to deliver a magnetic field of 2 T for 100 μs through the figure-of-eight coil (loop diameter, 9 cm; type no. 8106). The MEPs were recorded from BB and triceps brachii EMG. Before TMS stimulation, a cap with a coordinate system marked on it was placed on the subject's head and the hot spot for activation of BB was identified through a motor cortical mapping procedure. The hot spot was identified as the coordinates in which the lowest intensity of magnetic stimulation was required to evoke a MEP of 50 μV peak-to-peak amplitude in at least three of five consecutive trials (55). The coil was oriented and positioned with the handle of the coil pointing backward to induce posterior to anterior current flow across the primary motor cortex, and the coil was secured to ensure that the same area of the cortex was stimulated throughout the experiment. Single pulse stimuli were delivered at an interstimulus interval of 4 s. During the experiment, MEPs were displayed and averaged online for visual inspection as well as stored on a computer for offline analysis. At first, TMS was applied at rest. Magnetic stimuli were applied at 10–15 different stimulation intensities from 0.6–2.0 of the minimal stimulation intensity required to elicit MEPs (MEPthreshold) with 10 stimulations at each intensity. The sequence of intensities was randomly varied. Responses were measured as the peak-to-peak amplitude and expressed as a percentage of the corresponding maximal M-wave (Mmax). For each stimulation intensity, responses were averaged and the peak-to-peak amplitude was plotted until a stimulus-response curve with a well-defined MEPthreshold, slope, and maximal level (MEPmax) had been obtained. Figure 1 illustrates an example of the obtained BB MEPs, their increase with stimulus intensity, and the creation of a stimulus-response curve. Fig. 1.Transcranial magnetic stimulation (TMS) procedure and generation of stimulus-response curves. When transcranial magnetic stimulation is applied over the motor cortex, contraction of contralateral muscles may be elicited because of activation of corticospinal cells and spinal motoneurons (A). B: typical surface EMG recordings of motor evoked potentials (MEPs) of the biceps brachii (BB) obtained in a subject resting (left) and exerting a voluntary tonic contraction of 5% of maximum voluntary contraction integrated EMG (right). Responses to stimuli of increasing strength are aligned. For each stimulating intensity, a sequence of 10 stimuli (<0.25 Hz) was delivered and the peak-to-peak amplitudes of the elicited MEPs were averaged and normalized to the corresponding maximal M-wave (Mmax). When all mean MEP amplitudes are plotted against stimulation intensity, a stimulus-response curve is obtained (C). Each stimulus-response curve is characterized by a set of parameters including minimal stimulation intensity required to elicit MEPs (MEPthreshold), peak slope, maximum level of MEP amplitude (MEPmax), and stimulus intensity at which the MEP amplitude size is 50% of MEPmax (S50). Before training, 2 stimulus-response curves were obtained at rest and during a tonic contraction, and immediately after training 2 additional stimulus response curves were obtained during tonic contraction and at rest. For comparison, all stimulation intensities were normalized to the individual pretraining motor threshold of the baseline test. After a pretraining stimulus-response curve was obtained at rest, the maximal amplitude of the integrated BB EMG was determined and a TMS stimulus-response curve was obtained during tonic contraction of BB corresponding to 5% of maximal amplitude of the integrated EMG. This procedure was followed by a training session. Immediately after training, two additional stimulus-response curves were generated during tonic contraction and at rest. Before generation of a stimulus-response curve, maximal compound muscle action potentials of BB (maximal M-waves or Mmax) were elicited by bipolar surface electrical stimulation of the musculocutaneus nerve. A custom-built stimulator applied current to the nerve via ball-shaped electrodes fixed in the axilla with an interelectrode distance of 4 cm. The intensity of stimulation was increased from a subliminal level until there was no further increase in the peak-to-peak amplitude of the M-wave with increasing intensity. Mmax was determined by using this procedure before the generation of every stimulus-response curve (54) by application of TMS. The purpose of this procedure was to normalize the TMS data to the corresponding individual Mmax, thereby making it possible to compare the different test sessions. The strength training group performed heavy-load strength training of the dominant right arm elbow flexors three times per week for 4 wk. Training sessions never took place on consecutive days. The subjects performed standing unilateral biceps curl using a curl bench supporting both arms in a position of 20° shoulder flexion. Biceps curl was performed by doing flexion-extension movements of the elbow joint with the forearm supinated and the left hand placed on the right shoulder. After a warm-up procedure, the subjects performed five sets of 10–6 repetitions maximum. The sets were separated by a few minutes of rest, and the load was progressively adjusted throughout the training period to maximize the training response. For this purpose, all training sessions were also monitored by a supervisor. The subjects in the skill learning group also performed motor training of the right arm elbow flexors three times a week for 4 wk and training sessions never took place on consecutive days. During training, subjects were seated in an armchair with the right arm positioned on an arm support, the shoulder joint flexed, and the forearm supinated. This position was chosen to match the strength training paradigm anatomically and kinematically in the sense that only simple elbow flexion-extension movements were allowed. Because of the setup, flexion was caused by concentric contraction of the elbow flexors whereas extension primarily was caused by eccentric contraction of the same muscles. For the skill training, a purpose-build computer program was used. The position of the elbow joint was measured by a SG110 twin axis elbow goniometer (Biometrics) and displayed as a circular cursor on a computer screen in front of the subject. On the screen, a series of six figures were presented in a randomized order, each of them sketching a different series of combinations of flexion and extension movements. The cursor automatically moved from the left to the right at a velocity that was predetermined for each screen paradigm. Subjects were able to control the vertical movement of the cursor by varying the position of the elbow joint, thereby tracking the presented figures as precisely as possible. During extension of the elbow, the cursor moved to the bottom of the screen, whereas during flexion the cursor moved to the top of the screen. Figure 2 illustrates the different screen figures presented to the subjects. Time over the screen varied between 1.88 and 3.14 s, and the range of movement varied between the six different figures as well. After 2 wk of training, two additional screen figures were introduced to ensure maximal attention to the task. For this purpose, all training sessions were also monitored by a supervisor. Each training session consisted of four sets of 4-min continuous tracking with 1 min of rest in between the sets. Fig. 2.Motor skill training. A: general setup during skill training. B: training paradigm. During the first 2 wk of training, events 1–6 were presented to the subjects in a randomized sequence. During the last 2 wk of training, events 7 and 8 were added. The cursor speed varied between the events so time over screen was 1.88–3.14 s. The cursor trajectory (goniometer data) was sampled during training and motor performance (error) was calculated for each of the 4 sets of the 1st, 5th, 9th, and 13th training sessions. During the first, fifth, ninth and thirteenth training session, goniometer data were digitally sampled at 2,000 Hz with a QNX real-time analog-to-digital capturing system for calculation of performance. For each set of 4 min, training goniometer data were averaged for each screen figure and superimposed on the optimal track. Motor performance was quantified as the total distance between the performed and the optimal track in 10 different points. Measures of corticospinal excitability include, among other parameters, MEP amplitudes, MEPthreshold/motor threshold (55), and stimulus-response curves (14, 54). To characterize the stimulus-response function, the stimulus-response curve data were quantified through several procedures. The data of each curve were fitted with a three-parameter sigmoid function where S is stimulus intensity, MEPmax represents the maximum MEP defined by the function, and m is the slope parameter of the function. S50 is the stimulus intensity at which the MEP amplitude size is 50% of MEPmax. This equation has previously been referred to as an analog of the Boltzmann equation and has been used to fit data points by the Levenberg-Marquard algorithm (6, 7, 14, 29, 54). From this analysis, the maximal amplitude of the stimulus-response curve MEPmax was obtained. Furthermore, every curve was characterized by the slope parameter of the function and the MEPthreshold. The slope was calculated for the steepest part of the curve (i.e., at S50), indicating the maximal increase of MEP amplitude with increasing stimulator intensity. Because the MEPthreshold is not an explicit parameter of the equation and cannot be directly derived, it was calculated by using linear regression analysis. The data points on the steepest part of the curve were fitted by a straight-line regression formula (y = a + bx) and the baseline activity ± 1 SD were included in another linear regression. MEPthreshold was then calculated as the intercept between these two regression lines. For comparison and to be able to pool group data, all stimulus intensities were normalized to the resting or tonic MEPthreshold of the individual stimulus-response curves obtained pretraining on the day of the first training session. Normalization to MEPthresholds of the initial test was preferred because an analysis based on this procedure could detect whether stimulus-response curves were shifted left or right as a consequence of training. It has previously been demonstrated that the sigmoidal function parameters can be obtained reliably in testing sessions conducted on different days (9). Before statistical comparison, all data sets were tested for normal distribution by a Kolmogorov-Smirnoff test. Changes in maximal strength and motor skill performance were tested by using paired t-tests for each of the three groups. The stimulus-response curve parameters MEPthreshold, MEPmax, slope, and S50 were analyzed by comparing pre- and posttraining values in each of the three testing sessions. Resting and contraction values were analyzed separately for each of the three groups by paired t-tests, and a criterion of P < 0.05 was used. Significant P values are marked with an asterisk. Long-term adaptations to training were investigated by comparing pretraining data from the three testing sessions with repeated-measures ANOVA. For multiple-comparison analysis, Tukey's test was used for all pairwise comparisons between the group mean responses. Data are presented as means ± SE unless reported otherwise. Correlation between changes in the neurophysiological parameters and changes in motor performance capacity was tested using the Pearson's product-moment correlation test. RESULTSAt the end of the training period, the strength training group displayed a significant improvement in both maximal isomeric and dynamic muscle strength. After 4 wk of strength training, the group average maximal dynamic strength increased significantly by 31.2% from 10.5 ± 2 to 13.8 ± 1.8 kg (P < 0.001*). MVC also increased significantly by 12.5% from 21.9 ± 2.7 to 24.8 ± 2.3 N·m (P = 0.045*). The maximal dynamic as well as isometric muscle strength of the skill learning group and the control group remained unaltered. Skill learning group mean 1 RM was 12.9 ± 1.7 before and 12.9 ± 1.9 kg after the training period, whereas MVC was 25.1 ± 2.5 before and 24.7 ± 2.6 N·m after training. 1 RM in the control group decreased slightly from 12.4 ± 1.6 to 12.2 ± 1.6 kg. MVC in the control group was 21.6 ± 2.5 N·m before training and 21.7 ± 2.2 N·m after the training. These results imply that the strength training paradigm caused significant improvements of the subjects' maximal muscle strength. Neither the skill learning paradigm nor the experimental procedures induced changes in the maximal strength of the subjects. The level of performance was tested in the subjects in the skill learning group during the first, fifth, ninth, and thirteenth of the training sessions, and the performance was quantified as the mean deviation from the optimal track for each of the four training sets in the individual sessions (Fig. 3). The skill training group improved mean tracking performance significantly during the first training session from (mean ± SD) 162.8 ± 14.5 mm deviation to 142.6 ± 10.2 mm (P < 0.001*). During the fifth training session, the mean deviation decreased from 101 ± 13 to 91.7 ± 18.3 mm (P = 0.092). During the ninth training session, the mean deviation decreased significantly from 46.2 ± 11.3 to 37.8 ± 19.6 mm (P = 0.037*), and during the thirteenth (last) training session deviation decreased from 30.1 ± 16.6 to 27.3 ± 17.8 mm (P = 0.052). It follows from this marked improvement of motor performance during the individual training sessions that the long-term improvement of motor performance capacity over the 4 wk was highly significant (P < 0.001*), which is also evident from Fig. 4. Fig. 3.Strength tests. Measures of maximal dynamic muscle strength (1 RM) and maximal isometric muscle strength (MVC) before (shaded bars) and after (solid bars) the 4-wk motor training period (means ± SE). Left: MVC group mean peak torque (N·m) for the strength training group, the skill learning group, and the control group. Right: group mean 1 RM data (kg) for the 3 groups. *P < 0.05. Fig. 4.Skill learning motor performance. On the day of the 1st, 5th, 9th, and 13th (last) training sessions, the motor performance of the subjects in the skill learning group was quantified. Each session consisted of 4 times 4 min of training, and the individual motor performance was calculated as mean performance for each set of 4-min tracking. The performance (ordinate) is quantified as the deviation (error) from the optimal track to the actual cursor trajectory (group mean ± SE). The figure illustrates the improvement of motor performance in the skill learning group over the complete training period of 4 wk (52 sets). Statistical comparisons are based on the Wilcoxon's signed rank test. The TMS measurement aimed at investigating both short-term and long-term adaptations to the motor training paradigms. None of the control group measurements showed any significant changes during the whole period. The measurements from the control group will therefore not be considered further. Figure 5 illustrates measurement before and after one training session at rest and during tonic contraction on the day of the first, the seventh, and the final training session. Because only the skill learning group subjects exhibited any significant short-term changes in response to single training sessions, only the results of this group are illustrated. Fig. 5.TMS results: short-term effects of training. The figure illustrates pooled TMS data from the stimulus-response curves of the subjects of the skill learning group on the day of the 1st, 7th (after 2 wk), and 13th (after 4 wk) training sessions. Measurements were obtained at rest (A–C) and during tonic contraction (D–F) before (•) and after (○) each of the 3 training sessions. The abscissa of each graph represents intensity of stimulation normalized to the individual pretraining MEPthreshold of the baseline test and the ordinate represents MEP amplitudes (group mean ± SE) normalized to the corresponding individual Mmax. The characteristics of the stimulus-response curves reflect the short-term effects of the individual training session when comparing data obtained before training and data after training. In the skill learning group, there was a significant effect of the first training session. At rest, the MEPs were generally facilitated after training, and this was reflected in a significant increase of MEPmax (pretraining = 3.89 ± 0.8% of Mmax to posttraining = 6.03 ± 0.91% of Mmax; P = 0.02*). As an effect of training, the MEPthreshold seemed to decline and the slope seemed to increase; none of these alterations were, however, significant. S50 remained unchanged. The same pattern of changes was seen on the day of the seventh and the last training sessions (Fig. 5, B and C). However, none of these changes were significant. The short-term effect of training thus seemed to be largest in response to the first training session. The same tendencies as those seen at rest were evident during tonic contraction. As shown in Fig. 5, D, E, and F, MEPmax also tended to increase after training during tonic contraction. However, none of the changes after training were significant in any of the three testing sessions. Strength training did not induce any significant short-term changes in the TMS stimulus-response curves. The long-term adaptations to training are defined as the differences that occur when comparing the pretraining values obtained in the baseline test, the 2-wk test, and the 4-wk test. The adaptations that occurred after 2 and 4 wk of training are illustrated in Fig. 6. Fig. 6.TMS results: long-term effects of training. Pooled TMS data from the skill learning group, the strength training group, and the control group. The figure illustrates measurements obtained before motor training on the day of the first training session (•), after 2 wk (shaded circles), and after 4 wk of training (○). Measurements were obtained at rest (A, C, E) and during tonic contraction (B, D, F). The intensity of stimulation (abscissa) is normalized to the individual MEPthreshold of the day of the 1st training session. The MEP amplitude is normalized to the corresponding individual Mmax amplitude. For the skill learning group, MEPmax at rest increased from 3.9 ± 0.8 to 6.9 ± 1.5% of Mmax after 2 wk (P = 0.04*) and 6.8 ± 1.1% of Mmax after 4 wk of training (P = 0.046*). MEPthreshold decreased from 48.7 ± 4.8% of maximal stimulator output in the baseline test to 42.5 ± 5% after 2 wk (P = 0.07) and 41.2 ± 5.1% after 4 wk (P = 0.03*). No other parameters exhibited any significant changes. MEPmax also increased during tonic contraction in response to training [from 33.2 ± 4.3% of Mmax to 52.5 ± 10.6% (P = 0.04*)] after 2 wk and 50.28 ± 10% after 4 wk of training (P = 0.07). MEPthreshold decreased from 32.7 ± 1.4% of maximal stimulator output initially to 29.5 ± 2% after 2 wk (P = 0.16) of training and 28.1 ± 1.5% after 4 wk of training (P = 0.04*). No other parameters exhibited any significant changes during tonic contraction. For the strength training group, there was no change of MEPmax at rest after the first 2 wk of training. After 4 wk of training, however, MEPmax decreased significantly from the initial 6.5 ± 1.4 to 3.8 ± 1.5% of Mmax (P = 0.01*). The slope of the stimulus-response curves decreased from 0.24 ± 0.07 to 0.17 ± 0.06 after 2 wk of training (P = 0.11) and to 0.11 ± 0.04 after 4 wk of strength training (P < 0.01*). Similar changes were observed during tonic contraction; these changes did not, however, reach a statistically significant level. It was possible to test four of the eight subjects in the skill training group again 6 mo after they had completed the training period. The results from the four subjects are illustrated in Fig. 7. As can be seen, MEPmax, MEPthreshold, and the slope of the recruitment curve were almost similar to the measurements before the training. Because of the small number of subjects the data were not subjected to a statistical analysis. The material was also too limited to determine a difference in the performance of the task at the three occasions (before training, after training, and 6 mo after training). Fig. 7.Retest: effect of detraining. Retest after 6 mo of detraining (n = 4). Group mean responses to TMS normalized to the corresponding individual Mmax of the 4 subjects. Intensity of stimulation is normalized to the individual MEPthreshold of the baseline test. The correlation analysis using the Pearson product moment correlation test showed a significant correlation between the long-term changes in the skill learning group TMS parameters and the motor performance (skill) of the subjects. The correlation analysis of the measurements obtained at rest including MEPmax and skill performance (error) showed a correlation coefficient of R = 0.356 (R2 = 0.127) and P = 0.021* whereas the analysis with MEPthreshold and skill performance showed that R = 0.431 (R2 = 0.186) and P = 0.006*. For the measurements obtained during tonic contraction, the corresponding analysis showed that R = 0.369 (R2 = 0.136) and P = 0.026* for skill and MEPmax whereas the analysis of skill and MEPthreshold showed a correlation coefficient of R = 0.486 (R2 = 0.236) and P = 0.001*. In contrast to skill learning, there was no correlation between the neurophysiological changes and the increase of maximal strength observed in the strength training group. DISCUSSIONIn this study, we have demonstrated that acquisition of a visuomotor skill is associated with increased corticospinal excitability over several weeks. Strength training for a similar amount of time was in contrast associated with decreased corticospinal excitability. Increased corticospinal excitability and expansion of the cortical representation of hand and finger muscles in relation to the acquisition of motor tasks is well documented (45, 46, 48). The present data demonstrate that similar changes also take place for proximal arm muscles in relation to acquisition of a visuomotor tracking task requiring precise control of the elbow flexor muscles. The excitability changes are thus not restricted to distal finger muscles, which are generally believed to receive a more significant corticospinal control than proximal muscle groups (52). This is of importance in relation to many forms of sports in which large proximal muscle groups are more important for the performance than the smaller distal muscle groups. It is not possible from our study to determine the underlying physiological mechanisms responsible for the changes in corticospinal excitability. Changes in the spinal motoneurons, corticospinal neurons, subcortical neurons contacted by corticospinal tract fibers and projecting to the spinal motoneurons, as well as intracortical inhibitory and/or excitatory interneurons may be involved. However, previous studies have demonstrated that changes within the primary motor cortex are involved in the expansion of the cortical representation of the muscles as well as the increased corticospinal excitability changes demonstrated by recording of the input-output relation for the MEP as in the present study (4, 5, 9, 12–14, 23, 49, 52, 56, 57, 61). Experiments in primates and rats give strong support to this (15, 32, 40, 41, 60, 61). Pascual-Leone et al. (45) demonstrated changes in the representation of hand muscles in the course of a 5-day training program involving acquisition of a finger motor skill, and Pascual-Leone et al. (46) added information to this study by including measurements during 28 days of training. The results demonstrated that the main improvement in the performance occurred during the first week of (quite intense) training after which the subjects continued to perform the task at a high level in the rest of the training period. The expansion of the cortical representation of the tested muscle was also mainly seen in the first week of training after which it gradually declined. We similarly found that the corticospinal excitability mainly increased within the first 2 wk of training, which was also the period where the main improvement in the performance of the task was observed. Some improvement of the performance of the task was still observed between the second and fourth weeks, but this was not accompanied by any changes in corticospinal excitability. This is in line with the idea that the increased corticospinal excitability is involved in the early acquisition of the visuomotor skill is not necessary for the skilled performance of the task as such. Convincing evidence of a crucial role of the motor cortex in early acquisition of motor skills has also been provided by Muellbacher et al. (38), who observed that the retention of motor learning could be blocked by repetitive TMS over the primary motor cortex in the early stages of learning. The exact time course of the changes in corticospinal excitability probably reflects the complexity of the task and the intensity of the training. This likely explains why the corticospinal excitability declined in the study by Pascual-Leone et al. (46) in the third and fourth weeks of training, whereas we observed no change. Evidence for a continuous reorganization in the primary motor cortex over several weeks has also been obtained in relation to the acquisition of a finger sequence learning task using brain imaging (30, 63). Several studies have suggested that strength training is associated with increased neuronal drive to the muscles: 1) Significant increases in strength precede muscular hypertrophy in the course of a strength training program (25, 28, 34, 39, 51). 2) “Cross education,” whereby movements contralateral to the trained limb exhibit increased strength, has been observed in a number of studies (19, 26, 37, 39). 3) Subjects who train imaginary muscle contractions have been shown to exhibit significant MVC increases (Refs. 66, 67; see, however, Ref. 24). 4) Several studies have reported increased maximal EMG recorded from the trained muscle after a period of training and used this to infer an increased neuronal drive to the muscles (2, 24, 37, 39). A few studies have also argued against any changes in neuronal drive in relation to strength training. Using twitch interpolation to evaluate the voluntary drive to the muscle has thus generally shown that subjects are able to voluntarily activate the muscle almost to its maximal capacity before training and that no or only minor changes occur in relation to strength training (3, 22, 24, 62). A number of studies have reported various adaptations in the central nervous system in relation to strength training. Aagaard et al. (1) demonstrated significant increases of evoked H-reflex and V-wave responses during maximal contraction after 14 wk of strength training and suggested that this reflected increases in descending motor drive from higher centers leading to increased α-motoneuronal excitability. However, in the present study we found no evidence of increased corticospinal excitability either at rest or during voluntary contraction of the muscle. In line with a recent study by Carroll et al. (7), we actually observed a decrease of corticospinal excitability. In rats, Remple et al. (53) have also found that reorganization of the movement representation within the motor cortex is similar, whether the rats perform the movements against a low or a high load. There is thus no evidence that stronger muscles are related to a larger representation of the muscles in the primary motor cortex. From our study, we cannot decide whether the observed decrease in MEPmax and the slope of the input-output relation is explained by changes at a cortical or subcortical level. In the study by Carroll et al., MEPs evoked by transcranial electrical stimulation (TES) showed similar changes after strength training as MEPs evoked by TMS. Because TES is assumed to be only little influenced by cortical excitability changes, this observation favors a subcortical mechanism, although it should be pointed out that the low sensitivity of MEPs evoked by TES to cortical excitability changes mainly applies for small MEPs, whereas the changes in the MEP after strength training was mainly seen for large MEPs in the study by Carroll et al. (7) as well as in the present study. In the present study, a significant depression of the MEPs after strength training was only observed at rest. Carroll et al. (7) observed no changes in the MEPs at rest, but only during stronger contractions, whereas we only found significant changes at rest (a decrease in MEPmax was observed also during voluntary contraction, but it did not reach a significant level; this is in all likelihood explained by the higher variability of the recordings during voluntary contraction compared with rest). This discrepancy may be explained either by the different muscles being studied (Carroll et al. studied the first dorsal interosseus muscle), differences in the training design (the subjects in the study by Carroll et al. performed 4 times 6 abduction-adduction movements against external loads 3 times per week), or the MEP measurements (Carroll et al. investigated contraction levels up to 60% of MVC). Carroll et al. suggested that the decrease in the MEP size in their study was most easily explained by changes in the firing rate of the spinal motoneurons and/or their intrinsic firing properties. The findings in the present study do not exclude that such changes may occur in relation to strength training but suggest that other changes, such as changes in the excitability of cortical and/or spinal neurons and the transmission across synaptic connections between corticospinal fibers and spinal neurons, may also occur. We did not observe any correlation between the changes in MEPmax and the increased muscle strength in the subjects. Although this negative finding should not be overinterpreted, it does question whether changes in corticospinal excitability have any functional significance for the increased muscle strength. A 4-wk strength training program is usually considered to be too short for any structural muscular changes to take place, but we cannot fully exclude that some changes did take place in the muscles in our study and thus explain at least partly the increased muscle strength in the subjects. Nevertheless, most studies agree that increased neuronal drive is manifest very soon after the onset of strength training and is responsible for the main part of the initial gain in muscle strength. Then why did we not see similar changes in the MEPs as during the visuomotor skill training? Does this imply that the initial strength gain during strength training is not explained by the neuronal adaptations involved in learning and optimizing a skill? We do not think so. There are several factors that distinguish learning the visuomotor task and “learning” to generate maximal force: 1) novelty of the task, 2) visual feedback, 3) complexity of the task, and 4) pattern of somatosensory feedback related to the training. We find it likely that changes in the MEP and expansion of the cortical area might be involved when subjects are forced to use visual feedback to improve their ability of generating force as quickly and possibly also as precisely as possible. This would be of clear relevance to most sports activities, where it is not only important to be able to generate large force, but also to do it at a precise time and with maximal precision. In conclusion, these experiments have demonstrated that increased corticospinal excitability occurs over the course of several weeks of skill learning and that these changes seem to be closely related to the acquisition of new visuomotor skills. Such changes do not occur in relation to strength training. In contrast, strength training was associated with decreased corticospinal excitability at rest, which was not correlated to the increased muscle strength. The findings in the study thus emphasize the role of plastic changes in the corticospinal pathway in the acquisition of new motor skills but question whether similar changes play a role in possible neuronal adaptations to muscle strength training. GRANTSThis study was supported by The Danish Ministry of Culture (Sports Research Foundation). FOOTNOTESREFERENCES
Page 6aging is a complex biological process defined as a general decline in organ function associated with a state of decreased adaptiveness to changes and capacity to restore disrupted homeostasis. The mechanisms underlying aging are far from being utterly understood and are under continual scrutiny to improve the quality of life of the growing number of elderly people (2). No single theory is generally accepted to account for the aging process. On the contrary, nowadays, aging is recognized as a process involving the interplay between a network of damaging agents and a network of cellular defenses (20). The effects of aging on distinct organs and tissues have been widely demonstrated, as well as their relationship with the alteration of the organism oxidative stress (1, 19). Several are the cellular sources of reactive oxygen species (ROS), and, to limit the damage they inflict, organisms have evolved a complex antioxidant defense mechanism, composed of enzymatic and nonenzymatic substances, which scavenge free radicals, thus contributing to maintain an adequate redox status (for review, see Ref. 7). In physiological conditions the balance between prooxidant and antioxidant substances is kept slightly in favor of prooxidant products, thus favoring a mild oxidative stress state (9). According to the “aging free radical theory,” introduced by Harman (13), aging is caused by an accumulative enhancement of the oxidative stress on various organs (29), tissues, and cell components. In addition, Meydani et al. (25) proposed that the redox balance is gradually more delicately set as the aging process develops, causing deregulation of cellular functions. The most evident and well-understood age-related effects on skeletal muscle and heart are loss of muscle mass (or sarcopenia) and a significant hypertrophy, respectively (4, 28). On the other hand, there are few reports concerning the effects of aging on intestinal smooth muscle, despite recurring complaints of gastrointestinal motility disorders and chronic functional constipation by elderly people (14). Most studies focus mainly on the dysfunctions of intestinal mucosa absorption (10, 40). Deleterious effects of aging on human, rat, guinea pig, and mouse intestine have been associated with specific loss of cholinergic neurons from the myentheric plexus, leading to impaired motility (12) and muscular layer thickening (24). However, if aging is caused by increased tissue oxidative stress, it is quite possible that lipid peroxidation of plasma membrane, one of the most dramatic phenomena triggered by increased free radical production (36), might be contributing to changing membrane transport mechanisms, including ionic channels, carriers, and active pumps, thus disturbing signaling transduction mechanisms. Aerobic exercise programs were introduced as a tool to improve cardiovascular conditioning and physical performance, whereas resistance exercise training programs are assigned for the development of muscle hypertrophy and strength. Hence, a prospective strategy to minimize the deleterious effect of aging is to take advantage of the beneficial effects of an appropriate exercise training either on antioxidant mechanisms (26, 29, 31, 35) or on its capacity to induce hypertrophy (28). However, because exercise benefits are intimately dependent on intensity and volume, it was first crucial to establish the appropriate type of exercise the aged animals should be submitted to. Considering that elderly people usually have a dramatic reduction in daily physical activity, we investigated whether a lifelong physical activity pattern is a good strategy to counteract the age-related deleterious effects. In this study, we investigated the effects of aging on morphology, oxidative status, and reactivity of C57BL/6 murine ileum, heart, and gastrocnemius muscle and hypothesized that a lifelong moderate exercise program would have a protective role against aging deleterious effects. MATERIAL AND METHODSInbred male C57BL/6 mice were obtained from Centro de Desenvolvimento de Modelos Experimentais para Medicina e Biologia—Universidade Federal de São Paulo animal facilities, housed five animals/cage, with water and food ad libitum. The animals were kept on a 12:12-h light-dark cycle (0600 to 1800) and maintained at 23°C in our animal facility for at least 5 days before any experimental procedure. They were divided in three groups: sedentary young 3-mo-old (S 3), aged 18-mo-old (S 18), and continuously exercised animals from 3 to 18 mo old (E 3–18). Mice were killed by cervical dislocation, and the tissues of interest were isolated, frozen, and stored at −20°C for lipid peroxidation analysis. The animals’ handling was approved by our University Ethics Committee, in adherence to the International Guiding Principles for Biomedical Research involving Animals (Geneva, 1985). Mice were submitted to treadmill running, a kind of exercise in which intensity and duration can be easily manipulated and quantified, in opposition to voluntary wheel or swimming exercises (6). Animals from the E 3–18 group were initially acclimated to the treadmill environment by a 30-min running session at 13 m/min, 0% grade, for 5 successive days. Then they were submitted to an aerobic exercise-training program, which consisted of a daily session as follows: 1) 3-min warm-up at 5 m/min; 2) 60-min endurance run at speeds between 13 and 21 m/min, according to the tolerance of each animal; and 3) 3-min warm-down at 5 m/min. The intensity of the applied exercise program was considered moderate, because it corresponded to 55–65% of C57BL/6 maximal oxygen uptake, and the speed range was adjusted between 13 and 21 m/min at 0% grade (34). Mice were motivated to run with gentle hand prodding. Electrical shock was avoided as a negative reinforcement because this would add undue stress not typically associated with volitional exercise. Sedentary animals (S 18 group) were exposed to the same environment conditions (handling, treadmill motor noise, vibration, and deprivation of food and water) while the other animals performed their daily exercise session. Physical performance was assessed by determining the maximum treadmill speed reached by the animals during the incremental test. The schedule of this test consisted of 3 min of warm-up at 5 m/min, initial speed set at 10 m/min, followed by progressive increases of 1 m/min every min until animal exhaustion, and 3 min of cooling down at 5 m/min. Animals were killed 24 h after the last exercise session. Fresh tissue sections were appropriately isolated and stained with hematoxylin and eosin. In brief, tissue samples were fixed in 10% buffered formalin dehydrated by graded concentrations of alcohol (from 50 to 85% ethylic alcohol), cleared in four rinses of xylene, embedded in paraffin wax at 58 ± 2°C, and sectioned at 4 μm. Morphological features of the ileum (100-fold magnification) and gastrocnemius muscles (200-fold magnification) were evaluated by sorting out 10 measurements from three fields per tissue sample. Muscular layer thickness of the ileum and cross-sectional areas from gastrocnemius muscle fibers were quantified by computer software (Image Tool 3.00 for Windows, University of Texas Health Science Center in San Antonio, San Antonio, TX). Fresh ileum samples were fixed in phosphate-buffered 2% glutaraldehyde buffered at pH 7.2 with 0.2 M sodium phosphate for 4 h at 4°C. Then they were fixed in phosphate-buffered 1% osmium tetroxide for 1 h at 4°C, dehydrated in graded ethanol, treated with propylene oxide, and embedded in araldite epoxy resin. Ultrathin sections were cut with an ultramicrotome (Sorvall, Poter-Blum MT-1) and stained with uranyl acetate and lead citrate for transmission electron microscopy studies. Electron micrographs were taken with a digital camera, coupled to a 80-kV electron microscope (Carl Zeiss, Heidelberg, Germany). The ends of an ileum strip, ∼1.0 cm long, were tied up to a steel hook support, which was suspended in a 5-ml perfusion chamber containing Tyrode solution at 37°C, pH 7.4, and bubbled with air. Tissue strips were allowed to equilibrate under a 0.5-g basal tension for at least 30 min before any experimental procedure. During this rest period, the chamber solution was renewed every 10 min. Isometric tension was recorded by means of a force transducer (TRI 210, Letica, Barcelona, Spain) connected to an amplifier (model AECAD-0804, Solução Integrada, São Paulo, Brazil). Acquisition and analysis of the isometric contractions were done by means of the KitCad8 software (Software & Solutions, São Paulo, Brazil). Intestinal tissue responsiveness was evaluated by determining the potency and efficacy, obtained from noncumulative concentration-response curves to either carbachol (CCh) or KCl. The corresponding pharmacological parameters are EC50 (concentration of the stimulant that causes 50% of the maximum response, expressed in M) and maximum effect (expressed in g). Tissue strips were stimulated for 1.5 min at 5-min interval between two successive challenges. Just one concentration-contractile response curve was done per ileum segment. Peroxidative damage to membrane lipid constituents from heart, gastrocnemius muscle, and ileum were determined by measuring the chromogen reaction product of 2-thiobarbituric acid (TBA) with one of the products of membrane lipid peroxidation, malondialdehyde (MDA), according to the technique described by Winterbourn et al. (38) and modified by Fraga et al. (11). In brief, homogenate tissue pools were incubated for 30 min with the reaction mixture at 95°C. The chromogen reaction product was extracted in n-butanol, and its concentration was determined spectrophotometrically (N-200, Hitashi, Tokyo, Japan) at 532 nm. Results are expressed as nanomoles per milliliter per gram dry tissue. The following solutions were used for lipid peroxidation assays: phosphate buffer solution (in mM) consisted of 20 KH2PO4, 150 KCl, and 40 HEPES; the reaction mixture contained phosphate buffer, 11% acetic acid, 0.1% tungstophosphoric acid, 0.5% SDS, and 0.2% TBA. For contractile assays, the composition of the Tyrode solution was (in mM) 135 NaCl, 2.68 KCl, 1.36 CaCl2·2H2O, 0.49 MgCl2·6H2O, 12 NaHCO3, 0.36 NaH2PO4, and 5.5 d-glucose, pH 7.4. All chemicals were analytical grade. Salts, d-glucose, n-butanol, TBA, tungstophosphoric acid, SDS, ethylic alcohol, acetic acid, and xylene were purchased from Merck (Darmstadt, Germany); carbachol, osmium tetroxide, glutaraldehyde, araldite epoxy resin, and HEPES were from Sigma (St. Louis, MO); and hematoxylin and eosin were from Nuclear (Diadema, Brazil). Data are presented as means ± SE with n representing the number of experiments. Statistical significance was analyzed by one-way ANOVA followed by Tukey's test. P values <0.05 were considered statistically significant. RESULTSFigure 1 shows that the aged group (S 18) presented a 27% significant decrease of the maximum velocity compared with younger animals (S 3). On the other hand, animals submitted to a continuous moderated aerobic exercise during 15 mo (E 3–18) displayed not only a 100% enhancement in the maximum velocity compared with the 18-mo aged animals (S 18) but also a surprising 44% improvement of physical performance compared with younger ones. Fig. 1.Physical performance of young (S 3, n = 8), aged (S 18, n = 5), and continuously exercised (E 3–18, n = 5) C57BL/6 mice. The animal performance was evaluated by the maximum speed reached during a treadmill running incremental test. *Significant difference relative to S 3 animals, P < 0.05. #Significant difference relative to S 18 mice, P < 0.05. A common phenomenon associated with aging is the reduction of the skeletal muscle mass known as sarcopenia. Figure 2 illustrates that an 18-mo aging period caused a 47% reduction of the cross-sectional area of gastrocnemius fibers compared with the corresponding muscle from young animals. In contrast, the gastrocnemius muscle fibers from 18-mo-old animals submitted to a prolonged aerobic exercise program over 15 mo, with no interruptions (E 3–18), were well preserved with cross-sectional area ∼2,000 μm2, similar to those observed in the young animals (S 3) (Fig. 2). Fig. 2.Histograms of cross-sectional area of the gastrocnemius fiber muscles isolated from the S 3 (n = 5), S 18 (n = 6), and E 3–18 (n = 5) animal groups. *Significant difference relative to S 3 animals, P < 0.05. Cardiac hypertrophy, commonly evaluated through the heart wet weight-to-body weight ratio (HW/BW ratio), is considered a good marker of endurance conditioning. However, because aging also promotes cardiac hypertrophy, it was interesting to verify whether this marker also stands for aged exercised animals. Table 1 illustrates the heart wet weight, body weight, and HW/BW ratio for the three animal groups. There was a significant enhancement of the HW/BW ratio in both S 18 and E 3–18 animal groups compared with that ratio exhibited by S 3 group. This increased ratio was rather due to a higher increase in the heart weight, which was of 97% for the S 18 animals and 62% for E 3–18 animals, than in body weight, which remained unchanged in aged mice, exercised or not (Table 1).
We investigated one of the well-known damaging processes associated to oxidative stress, the membrane lipid peroxidation (36), of distinct tissues of the three animal groups. As shown in Fig. 3, the S 3 level of lipid peroxidation was quite variable according to the organ studied. The heart presented the highest level, ∼620 nmol·ml−1·g−1 dry tissue, and the gastrocnemius muscle and ileum intermediary levels, ∼350 nmol·min−1·g−1 dry tissue. Aging caused 81 and 48% significant increase in gastrocnemius muscle and ileum membrane lipid peroxidation, respectively. However, when a continuous moderate exercise program was introduced as a daily routine (E 3–18 group), a drastic reduction to levels down to or lower than 200 nmol·min−1·g−1 dry tissue was observed in the MDA concentrations in heart, gastrocnemius muscle, and ileum. Fig. 3.Comparison of the membrane lipid peroxidation of distinct tissues isolated from S 3 (n = 4), S 18 (n = 4), and E 3–18 (n = 4) animal groups. [MDA], malondialdehyde concentration. *Significant difference in relation to S 3, P < 0.05. #Significant difference relative to S 18 animals, P < 0.05. The enhanced aged oxidative stress on the ileum and the protective effect exerted by 15 mo of aerobic exercise led us to explore the ileum cellular and ultracellular structure alterations of the three animal groups. Figure 4 illustrates representative light micrographs of hematoxylin-eosin-stained ileum sections isolated from the three animal groups. Aging and aging associated with prolonged aerobic exercise caused opposite cellular effects on the intestine structure. Compared with the ileum histological structure pattern from young animals (Fig. 4, S 3), the ilei isolated from the S 18 animal group (Fig. 4, S 18) exhibited a moderate to high damage of the mucosa, with some epithelial lifting of the epithelial layer, the presence of few denuded villi, and a mild level of architectural distortion. These structural alterations of the mucosa layer can be classified as grade 2 or 3, according to Chiu et al. (5). Besides damage of the mucosa, the most impressive aging effect was the significant thickening of the muscular layer (Fig. 4). These structural damages of both intestinal mucosa and muscular layer were not observed in the exercised aged animal group (Fig. 4, E 3–18), which exhibited an organized ileum architecture, presence of a normal mucosa with very regular villi and submucosal layers, and a muscular layer slightly thinner than in young animals (Fig. 4, S 3). Quantitatively, this impressive distinction corresponded to 57% significant increase of muscular layer thickness in aged animals and a small, but significant, 24% reduction in exercised aged animals compared with young ones. In contrast, there was a drastic and significant reduction, ∼50%, of the muscular layer thickness in exercise aged animals compared with the aged ones (Fig. 4). Fig. 4.A: representative ileum thin section light micrographs from S 3, S 18, and E 3–18 animal groups. Magnification, 100-fold. B: histogram of the muscular layer thickness of the ileum from S 3 (n = 5), S 18 (n = 7), and E 3–18 (n = 5) animal groups. *Significant difference in relation to S 3 animals, P < 0.05. #Significant difference in relation to S 18 mice, P < 0.05. Considering these morphological results, it was interesting to perform ultrastructure analysis, mainly to evaluate the influence of aging and/or continuous aerobic exercise on cellular organelles. The representative electron micrograph of S 18 group ileum (Fig. 5) showed relaxed hypertrophied muscle fibers with a central elongated voluminous euchromatic nucleus, presence of interdigitated myofilaments, and elevated number of caveolae. Typically, some signals of organelle deterioration were observed in the aged animal micrographs, mainly the presence of mitochondria with distinct level of disorganization of the cristae, lamellar corpuscles, and lysosomes. In contrast, the electron micrographs of the muscular fibers of the ileum from E 3–18 group showed a more dense interdigitated myofilament content, accumulation of numerous mitochondria, with different forms and sizes, located nearby the nuclear poles, and a central euchromatic nucleus (Fig. 5). In both animal groups, the extracellular matrix elements contained mainly type I collagen fibers and amorphous elements. Fig. 5.Representative ileum electron micrographs isolated from S 18 (n = 3) and E 3–18 (n = 3) animal groups. F, muscular fiber; N, nucleus; ★, extracellular matrix; •, myofilaments; m, mitochondria; mc, degenerated mitochondria; l, lysosomes; c, caveolae; d, desmosomes; sr, granular sarcoplasmic reticulum. To verify the effects of these structural features on the intestinal reactivity, the ileum was stimulated by increasing concentrations of a muscarinic agonist, CCh, or to KCl, in an attempt to infer on the pharmacomechanical and electromechanical couplings, respectively. As illustrated in Fig. 6A, there were no differences in the potency of the muscarinic receptor activation with CCh, as the EC50 values were ∼0.6 μM for the three animal groups. Similar results were obtained for the contractile responses triggered by KCl depolarization. The EC50 values were similar and ∼16 mM in young (S 3), aging (S 18), and exercised aged animals (E 3–18) (Fig. 6A). Regarding efficacy, there were no differences in the ileum maximum contractile responses triggered by addition of either CCh (∼1.3 g) or KCl (∼1 g) in the three animal groups (Fig. 6B). These results suggest the presence of compensatory mechanisms in each group to obtain the same response despite the morphological damage induced by aging. Fig. 6.A: comparison of isolated murine ileum responsiveness to KCl and carbachol (CCh), assessed by concentration-response curves from S 3 (n = 6), S 18 (n = 12), and E 3–18 (n = 12) animal groups. Tension is expressed relative to the stimulant maximum response. Brackets denote concentration. B: comparison of the maximum contractile responses elicited by KCl and CCh on isolated ileum from S 3 (n = 6), S 18 (n = 12), and E 3–18 (n = 12) animals. DISCUSSIONWe provide evidence, for the first time, that lifelong physical activity pattern has beneficial effects on C57BL/6 mice intestinal tissue, mainly by reversing the aging deleterious effects on the structure and oxidative status of the ileum. Despite these alterations, its contractile response was unaffected by aging or exercise. In addition, we confirmed that this exercise program exerts a clear positive effect on some well-known deleterious responses triggered by aging, such as reduced physical activity, sarcopenia of the gastrocnemius, cardiac hypertrophy, and oxidative status of these tissues. An increasingly number of authors claim that lifestyle factors, including diet, regular aerobic exercise, vitamin supplementation, and nonsmoking habits, may favorably modulate the related adverse physiological and pathophysiological consequences of aging (3). Herein, we evaluated the effect of a continuous 15-mo period of moderate treadmill running program, starting at 3 mo old, on the murine physical performance, gastrocnemius muscle sarcopenia, HW/BW ratio, and lipid peroxidation level of distinct tissues or organs. We chose this period instead of the usual 8- to 16-wk period reported in most studies involving the beneficial effects of a prolonged exercise program (22, 39) because 8 wk cover only 15–30% of the animal life span and thus cannot be considered as a habitual practice. One consistent deleterious effect of aging described in animal or human models is the impairment of physical performance (28). The maximum velocity reached by aged mice in the incremental running test was significantly reduced to 50% of the value attained by young animals (Fig. 1). However, by exercising during 15 mo, the animals preserved their capacity of adaptive response to aerobic exercise. Indeed, there was a drastic physical performance improvement in aged exercised animals, which was not only higher than their aged-matched pairs but also 44% higher than the young animal values (Fig. 1). These results may be attributed to intrinsic aspects of the designed exercise program. Although not intended to improve performance, because the principles of overloading and overcompensation were not applied (15) and the exercise stimulus was maintained ∼60% of maximal oxygen uptake all over the animal life span, one might suppose that the mice were still submitted to a certain level of overloading considering they were growing older at the same time. Corroborating the beneficial effect of exercise, the expected age-induced sarcopenia of the gastrocnemius muscle was totally prevented in aged animals continuously exercised (Fig. 2). This shows once again the preservation of the adaptive response of the elder animals to exercise stimulus, as they are able to develop the specific musculature engaged in the movement. In contrast, the increased HW/BW ratio, widely accepted as a good marker of aerobic conditioning (8), must be interpreted with caution, because it is known that both aging and endurance exercise induce cardiac hypertrophy (4, 8). The same heart weight values observed in aged animals regardless of exercise routine suggest that the daily physical activity somehow protected the senescent heart against the pathological age-related hypertrophy (Table 1). Obviously, this is a mere speculation, albeit an attractive one, because we did not carry out further morphological studies of the heart to better clarify this point. Consequently, cardiac hypertrophy cannot be used as an aerobic conditioning marker in aging research, as usually done in exercise physiology studies (8). As far as we know, this is the first time that such a prolonged moderate exercise program was investigated and its remarkable beneficial effect against aging deleterious effect was so clearly demonstrated. Aging has been hypothesized to be caused by the cumulative and deleterious effects of ROS over the life span (13). On the other hand, exercise intensity is also known to interfere with the cell redox status, by altering the balance between oxidant production and antioxidant defense mechanisms (19). Whereas a moderate exercise program can improve the organism redox status by augmenting the defense mechanisms to a higher level, thus compensating any simultaneous increase of oxidant products, strenuous physical exercise, in turn, increases ROS production and oxidative stress (18). The better physical performance of aged exercised animals compared with the physical performance of aged animals led us to suppose that this could be due to a better overall oxidative status of the former animal group. So, to clarify this point, we indirectly evaluated the redox status by measuring membrane lipid peroxidation of distinct tissues and organs isolated from the young, aged, and exercised aged animals (Fig. 3). As expected, and already described by other authors (23, 37), the level of oxidative stress was quite variable according to the tissues or organs studied (Fig. 3). Nevertheless, the main point here is to remark that the gastrocnemius muscle, which is one of the muscles directly engaged in the treadmill running, underwent a drastic and significant prooxidative shift of its redox status with aging, because there was 81% increase in the MDA concentration, one of the end products of the membrane lipid peroxidation (Fig. 3). More interesting was the clear and strong improvement of the redox status of the gastrocnemius muscle when the regular moderate exercise program was associated with the aging process (Fig. 3). This improvement was also observed for the heart, which was the organ with the highest lipid peroxidation level among the organs and tissues studied, in both aged and young animals (Fig. 3). These data strongly suggest that the heart is under a high oxidative stress condition in this rodent species, which is likely due to their accelerated aerobic metabolism (6). In addition, these results also evidenced that aerobic heart conditioning is a powerful way to counteract this unfavorable oxidative condition. Navarro et al. (29) have recently described that moderate exercise, namely 52 wk of 5-min daily session throughout the week of treadmill running, decreased the aging-associated development of oxidative stress in other tissues by preventing the increase of lipid and protein oxidation, decrease in antioxidant enzyme activities, and decrease in mitochondrial oxidase and reductase enzyme activities. Finally, it might be argued that the lipid peroxidation data presently described were not fair enough because of the nonspecificity and low sensitivity of the methodology employed for measuring lipid peroxidation, i.e., the MDA reaction with thiobarbituric acid-reactive substances (16). However, the differences observed were so expressive that if any correction had to be made, this would be for an underestimation of the differences described, therefore not interfering with our interpretation of the data. The main and most interesting point we addressed in this study was the aging effects on the structure and responsiveness of the ileum, considering that intestinal problems such as constipation, inflammation, and cramps are very frequent complaints among elderly people (14). We thus raised the hypothesis that aging-associated intestine dysfunctions could be related to the increase of its oxidative stress, which could be counteracted by prolonged moderate exercise. Thus appropriate exercise could promote a favorable shift of the oxidant-antioxidant balance as a consequence of the activation of tissue antioxidant mechanisms triggered in response to ischemia-reperfusion tissue hypoxia (17, 32). In fact, ileum hypoxia induced by submaximal exercise in old rats has recently been demonstrated (27). So, we have studied the aging effects on ileum structure by means of light and electron microscopy studies, lipid peroxidation, and contractile responsiveness as a functional test. In addition, we investigated whether a regular long-term moderate exercise program might contribute to preserve intestinal tissue against aging effects. Murine ileum was sensitive to aging regarding structure and redox status, because it showed clear signs of degeneration at both cellular and ultracellular levels, and higher oxidative stress (Figs. 3–5). On the other hand, the introduction of daily moderate treadmill running sessions during a 15-mo period, corresponding to 80% of the animal life span, exerted a striking protective influence against aging deleterious effects in C57BL/6 mice (Figs. 4 and 5). Moreover, this exercise program not only diminished aging oxidative stress by reducing ileum lipid peroxidation but also remarkably improved the redox status of the senescent intestine even compared with young animals (Fig. 3). Surprisingly, all these morphological and redox status changes did not have any functional consequences on tissue responsiveness, because neither the electromechanical coupling nor the muscarinic pharmacomechanical coupling was influenced by aging (Fig. 6). This might indicate that either the signaling transduction pathways are well preserved or there are some compensatory mechanisms leading to the same contractile response. The scarce literature focusing on ileum responsiveness in aged animals or in experimentally induced ileum hypertrophy conveys the notion that the contractile response is diminished. Lofgren et al. (24), when inducing guinea pig ileum hypertrophy by partial occlusion, observed modification of the ileum contractile function, mainly by altering the expression of myosin isoforms. The only study dealing with aging on ileum was reported by Ochillo and Tsai (30), who observed a diminished responsiveness of the muscarinic receptor activation in the longitudinal muscle of guinea pig ileum, which could not be explained by a reduction in the number of receptors in aged animals. Besides the distinct animal species, the discrepancy of the present results relative to those described by Ochillo and Tsai may be attributed to the different muscarinic agonists used. The influence of aging on acetylcholine-evoked contraction has been explained by an effect on the activity of the enzyme acetylcholinesterase (21), but in the present study the influence of this enzyme was ruled out because the ileum was stimulated with choline instead of acetylcholine (Fig. 6). So, the absence of aging and/or exercise effect on ileum responsiveness, despite the loss of cholinergic nervous terminations and swallowing of nitrergic ones (12), is likely due to the large reserve capacity of this tissue as a constituent organ of the gastrointestinal system (33). In summary, we demonstrated for the first time that a lifelong physical activity pattern is a suitable strategy to counteract the observed cellular and ultracellular morphological alterations and the oxidative stress caused by aging on the isolated murine ileum. In addition, the beneficial effects of this kind of exercise are also the reversion of the well-known effects of aging, such as impairment of the physical performance, sarcopenia, heart hypertrophy, and enhanced oxidative stress of skeletal and cardiac muscles. GRANTSThis work was supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and Fundação de Amparo à Pesquisa do Estado de São Paulo. E. F. Rosa was on a fellowship from CNPq. FOOTNOTESWe are grateful to M. L. C. Almeida for interesting discussions with Eloi F. Rosa during his fellowship period and to Mauro Cardoso Pereira for technical assistance with the animals. REFERENCES
Page 7despite the impressive amount of literature on the subject, a clear definition of myocardial ischemia remains elusive (17). However, the term is generally used to describe an imbalance between myocardial oxygen supply and demand that leads to a shift from oxidative to anaerobic metabolism (11, 17). These metabolic disturbances affect the high energy stores of the myocardium, causing ionic, functional, and ultrastructural changes, which may ultimately lead to cellular injury and infarction (21, 23, 32). Thus, because it has been proven that the extent and severity of such changes are modulated by the duration of the oxygen supply-demand imbalance and by the level of myocardial collaterization present in the region at risk (22, 26, 37, 38), any system aiming to detect ischemia correctly should be sensitive to these factors. Electrical impedance, a passive electrical property, has been shown to reflect the myocardial response to ischemia, being correlated not only with the associated functional abnormalities (19, 36), but also with the underlying ionic (2, 3, 28, 39) and metabolic (10, 14) changes in the myocardium. However, despite the general agreement that myocardial electrical impedance (MEI) increases significantly with supply ischemia (induced by coronary artery occlusion), the relevancy of its early time course to the understanding of the ischemic process is not clear. Using arterially blood-perfused rabbit papillary muscles, Kleber and colleagues (25) found that MEI increases in a biphasic manner during ischemia: an early increase immediately after the ischemic insult, a plateau phase, and thereafter a rapid increase thought to mark cellular uncoupling (2, 4, 28, 39). We, like other investigators (4, 10, 14, 39), have observed (unpublished) similar patterns in dogs and pigs (7, 19) (see Fig. 1), two representative large-animal models traditionally used to study myocardial supply ischemia as can occur in humans (e.g., during beating-heart coronary artery revascularization). However, these species have well-documented physiological differences, especially regarding collaterals (26): whereas dogs possess a well-developed collateral system, pigs (like humans) generally lack it (26, 46) and therefore are more susceptible to supply ischemia (13, 16, 38). Thus we hypothesize that a formal comparison between the MEI response of dogs and pigs to coronary artery occlusion should highlight those MEI parameters that are able to differentiate between subjects with (e.g., dogs) and without (e.g., pigs) resistance to supply ischemia. To date, such comparison is lacking. Fig. 1.Sample myocardial electrical impedance (MEI) data (z[n], ○) measured on dog (A, obtained from Ref. 19) and swine (B, obtained from Ref. 7) showing the MEI biphasic rise after coronary artery occlusion (t = 0 min): early increase immediately after the ischemic insult (I), a plateau phase (II), and rapid increase thought to mark cellular uncoupling (III). Additionally, we have recently presented in vivo and in situ measurements of MEI during beating-heart coronary artery bypass surgery in humans, in which MEI was shown to respond to coronary artery occlusion-reperfusion during the grafting process (9, 20). The availability of such data, collected under similar conditions to those from pigs and dogs, provides a unique opportunity: the side-by-side validation of two large-animal (supply) ischemia models, canine and swine, against their clinical counterpart. From such a comparison, a better understanding of how well these two models approximate (both in magnitude and temporal behavior) the passive electrical changes seen in human myocardium early during coronary artery occlusion could be gained, further testing the argument that pigs constitute a better model of human disease than dogs (26). Hence, this study was designed to compare the early temporal behavior of MEI during acute occlusion of the left anterior descending coronary artery (LADa) in humans with that of dogs and pigs. METHODSMEI ischemic data were obtained from three previously presented studies (performed by this laboratory with prior institutional approval). Animal protocols were approved by the Institutional Lab Animal Care and Use Committee at this institution and adhered to the statutes of the Animal Welfare Act and the guidelines of the Public Health Service. The human protocol was approved by the Biomedical Sciences Institutional Review Board and complied with the Codes of Federal Regulations Title 45-Part 46 of National Institutes of Health and Title 21-Part 50 of the Food and Drug Administration. The studies were divided (by species) into three groups: 1) Data on dogs were taken from work performed by Howie et al. (19), who recorded MEI during acute coronary artery occlusion and reperfusion in anesthetized male dogs. For this interspecies validation study, all animals [n = 10, 22.8 kg (SD 2.0)] from the 120-min LADa occlusion group were chosen. 2) Pig data were collected by us as part of a study examining the preconditioning effects of high-dose adenosine on MEI (7). Here, seven (3 male, 4 female) juvenile pigs [n = 7, 28.1 kg (SD 1.0)] from the control (placebo) group, subjected to a 10-min acute LADa occlusion, were selected. 3) Humans were measured intraoperatively (with prior informed consent) during elective beating-heart left anterior myocardial revascularization (9, 20). At our institution, such procedure involves complete acute occlusion of the coronary artery being grafted (LADa), to achieve a clean operating field. In this study, we report on the MEI response to such procedural complete occlusion [13.2 min (SD 3.96)] of the LADa in seven (5 male, 2 female) patients [n = 7, 91.4 kg (SD 13.6)] with moderate preoperative (angiographically demonstrated) LADa stenosis (70–80%, inclusive). Despite different aims, all experiments (for the selected groups) shared the same structure: a baseline period of at least 3 min, during which no interventions were made, followed by complete LADa occlusion. In all cases, isoflurane was used for anesthesia maintenance, hemodynamic variables (including temperature) were maintained within their normal physiological range, and MEI was measured with identical equipment and technique. As previously described (8, 19), in open-chest conditions, two standard (commercial) temporary pacing wires (Medtronic Streamline 6500 or A&E Medical MYO/WIRE M-25, 8–13 mm2 of exposed surface area) were sutured completely into the midmyocardial wall, ∼1 cm apart in the LADa distribution, and distal to the occlusion site (and stenosis, in the human case). From these leads, confirmed (visually) to be in the center of the region rendered ischemic by LADa occlusion, an MEI monitor developed at this laboratory was used to measure the complex MEI spectrum (8), a combination of the true electrical impedance spectrum of the myocardium and that of the electrode-tissue interface. In short, a computer-controlled circuit stimulated the myocardium with a subthreshold zero-mean bipolar current, consisting of two alternating rectangular pulses (±5 μA, 100 μs wide) generated 200 ms apart. The complex MEI spectrum was calculated as the ratio (at each frequency) of the current and voltage spectra resulting from the ensemble averages of 10 (positive) stimulus pulses and their respective responses (8). The frequency-domain resolution was either 5.4 (pigs and humans) or 100 (dogs) Hz (7, 8, 9, 19). Regardless, all studies reported (every 3 s) on the mean MEI modulus over the studied nonuniform frequency range (0.27–5.90 kHz, i.e., the spectrum's main lobe). Because MEI has been shown to be sensitive to temperature, the electrode system's geometry, and its location in the myocardium (33, 35, 41–45), to reduce intersubject variability, we studied MEI changes expressed as percentages from baseline (normalized MEI, Z[n]). Here, baseline MEI (ζb) was defined as the average of the 15 samples immediately preceding the coronary artery occlusion marker (no, inclusive). Normalized data were analyzed by a moving-window technique. A rectangular window of length N (WN [n], N = 30 or 1.5 min) was used, and successive windows overlapped by N−1 samples; within each window, the best linear fit (L[n], in a least-squares sense) to the MEI data was calculated. The slope of the fitted line was used as an estimate of the MEI first derivative at the center of the window (Δζ[n]). The estimated MEI first derivative served to detect the ischemic plateau onset time sample (np) as defined in Ref. 7; i.e., the sampled time point at which Δζ[n] has fallen to 3% of its maximum value (Δζp) (see Fig. 2). Once np was determined, the MEI ischemic plateau value (ζp) was defined as the average of the 15 MEI measurements immediately after it (inclusive). Fig. 2.Sample data analysis procedure on swine data (obtained from Ref. 7). Top: original MEI measurements (z[n], ○) indicating moving analysis window (WN) and best (mean-square) linear fit (L) at sample ni. Bottom: MEI first derivative (Δζ[n]) estimated from the linear fit (L[n]), indicating ischemic plateau threshold (Δζp, i.e., 3% of the maximal derivative) and value at ni (Δζ[ni]). Occlusion marker (no) and ischemic plateau sample (np) are used to calculate baseline and ischemic plateau MEI values (ζb and ζp, respectively). Note: the fact that Δζ[n] appears to increase before coronary artery occlusion (no) is an artifact of the noncausal windowing technique used. MEI baseline, time to plateau onset and normalized plateau value are presented as means (SD). Mean intergroup (cross-species) differences in these parameters were evaluated by the nonparametric Kruskal-Wallis ANOVA test. If significant differences were observed, then post hoc pairwise comparisons between all groups were made by Dunn's method. On the other hand, for each individual, MEI measurement uncertainty is dominated by measurement noise (Gaussian) (44); thus mean differences between baseline and plateau values were evaluated by using a paired Student's t-test assuming unequal variances (two-tailed). In all cases, P < 0.05 was considered statistically significant. RESULTSAs it has been previously shown (7, 9, 19), MEI increased immediately and significantly (P < 0.05) from baseline after LADa occlusion, reaching an ischemic plateau value subsequently. This held true for all subjects and species studied. However, significant interspecies differences, in both the magnitude and timing of such changes, were observed (see Table 1 and Fig. 3), with the time to MEI plateau onset showing the most remarkable variation. Fig. 3.Sample normalized MEI data (Z[n]) after coronary artery occlusion. Original MEI measurements on representative pig (○, from Ref. 7), dog (▵, from Ref. 19), and human (☆, from Ref. 9) indicating occlusion marker (no, t = 0 min). Insets: estimated MEI first derivatives for each subject, indicating ischemic plateau threshold (Δζp, i.e., 3% of its maximum value). Observe that Δζ[n] in the dog does not fall below the ischemic plateau threshold, i.e., MEI does not reach ischemic plateau.
In dogs, MEI ischemic plateau was reached after 46.3 min (SD 12.9) of LADa occlusion, a significantly (P < 0.05) longer period compared with that of pigs [4.7 min (SD 1.2)] and humans [4.1 min (SD 1.9)]. Similarly, baseline MEI values on preischemic (preocclusion) myocardium differed (P < 0.05) between canines [781 Ω (SD 39)] and the other species studied [swine: 444 Ω (SD 67), humans: 489 Ω (SD 135)]. Although baseline measurements showed significant differences between dogs and pigs, no differences could be observed between these two species regarding their normalized ischemic plateau value, a parameter traditionally used to correlate MEI with ischemia (2, 3, 10, 14, 19, 25, 28, 39) and other diseased states of the myocardium (15, 29). Normalized MEI reached 19.6 (SD 2.6) and 15.3% (SD 4.7) at plateau in dogs and pigs, respectively. Interestingly, humans, who also showed baseline differences with canine data, had a lower normalized ischemic plateau value [11.0% (SD 6.0), P < 0.05]. It should be noted that, regarding all MEI parameters studied (time to plateau onset, baseline and normalized ischemic plateau value), swine and human groups were only distinguishable by the wider distribution of values observed in humans (as expected owing to the intrinsic variability of coronary artery disease). DISCUSSIONThe degree of myocardial collaterization is recognized as an important interspecies differentiating factor (13, 16) and has been shown to influence the extent and severity of myocardial ischemic injury directly (22, 38). Thus, because canines possess a well-developed collateral system, unlike humans or pigs, they have been shown to have remarkably higher resistance to supply ischemia as reflected by reduced rates of ATP depletion (38) and smaller infarcts (13). Here, the MEI ischemic time course, represented by the time from coronary artery occlusion to MEI ischemic plateau onset, was shown to reflect such interspecies differences in collaterization. Although originally attributed to a rapid collapse of the intravascular and interstitial spaces (“vascular collapse”) (25), the exact mechanism determining the early behavior of MEI under acute ischemia remains unknown. Kleber et al. (12, 47) demonstrated that initial impedance rise during zero-flow ischemia is sensitive to osmotically induced cell swelling (i.e., to changes of the extra- to intracellular volume relationship). As such, the time-dependent concentrations of intra- and extracellular ions (such as calcium, [Ca2+]i, hydrogen, [H+]i, and potassium, [K+]e; brackets denote concentration) (3, 28, 39), anaerobic by-products, and ATP (10, 14, 40) found on the ischemic myocardium have been shown to play an important role. For example, extracellular potassium ([K+]e) accumulation during ischemia is documented to follow a triphasic time course (18, 24) similar to that of MEI. Furthermore, the secondary rise (i.e., third phase) of both parameters is closely coupled in time (2, 28, 39). Owens et al. (28) demonstrated that such terminal (secondary) rise signals ischemic electrical cell-to-cell decoupling, and only occurs after a significant accumulation in [Ca2+]i and [H+]i [as initially suggested by Cascio and colleagues (2)]. However, using paired ventricular myocytes, Sugiura et al. demonstrated that (electrical) junctional conductance changes with ATP concentration, independently of intracellular free gap-closing ions (such as [Ca2+]i) (40), thus suggesting also a direct relationship between active ionic transport and the MEI. In other words, they proposed that the depletion of myocardial high-energy stores (e.g., during ischemia) leads to increases in intercellular electrical impedance (the reciprocal of electrical conductance) by means of modifying the ATP-mediated ionic conductivity through the cellular membranes. Interestingly, the activation of ATP-dependent potassium channels has been shown not only to be mediated by myocyte swelling (30), but also to affect the timing and magnitude of early [K+]e accumulation (34), a well-established cause of early electrical disturbances during ischemia (27). The role of these channels as part of the underlying mechanism behind changes in the passive electrical properties of ischemic myocardium is further strengthened by the recent observations of Bollensdorff et al. (1). They demonstrated that ATP-dependent potassium-channels mediate extracellular sodium influx to the myocytes, and, therefore, [Ca2+]i overload leading to cell-to-cell uncoupling. Hence, a slower MEI progression to plateau after coronary artery ligation (as observed on canine myocardium) is suggestive not only of delayed breakdown in cellular (ionic) homeostasis but also of slower ATP depletion rates (i.e., of a less damaging ischemic process). This is in good agreement with the classic results of Schaper and colleagues (38) and with the slower [K+]e accumulation in ischemic canine myocardium reported by David et al. (6) [evident compared with results from swine (18, 39)]. As a result, the MEI timing similarities observed between swine and human data indicate that the passive electrical properties of human myocardium early during coronary artery occlusion are closely modeled by those in pigs. Furthermore, given MEI's metabolic ties, this supports the argument that pigs are a better model than dogs for acute myocardial supply ischemia as can occur in humans (26), at least for moderately diseased patients (as those studied here) expected to have a poorly developed collateral system (5, 31). On the other hand, whereas close MEI time courses may reflect comparable myocardial metabolisms during ischemia, the observed interspecies differences in MEI baseline and normalized plateau values (the other parameters studied) could be suggestive of specialization in the myocardial tissue ultrastructure, as suggested in the literature (36, 42, 45). However, as stated above, MEI measurements have been shown to be affected by temperature (44), electrode separation (33, 41), depth (43, 45), and orientation (relative to the muscle fibers) (35, 41, 42, 45). For instance, Steendijk and colleagues (42) reported significantly different MEI values for measurements made longitudinally and transversely across canine myocardial fibers [313 (SD 49) and 487 Ω·cm (SD 49) at 5 kHz, respectively]. Hence, because the spacing and orientation of the temporary pacing wires (used to acquire MEI data) on the beating heart are difficult to control, conclusions based on the absolute difference among baseline values between the species are not possible. Furthermore, as the normalized MEI plateau value indicated coronary artery occlusion precisely but failed to differentiate between the (clearly different) ischemic processes on dogs and pigs, this study emphasizes the relevance of the early ischemic MEI time course. Here, the MEI temporal behavior after coronary artery occlusion (parameterized as the time to reach MEI ischemic plateau) reflected the higher collateral density of canine myocardium, acting, perhaps, as a direct indicator of ischemic resistance and metabolism. Therefore, MEI timing parameters could be a valuable tool during surgical myocardial revascularization procedures, in which cardioprotective techniques that seek to enhance the myocardial endurance to ischemia (e.g., cardioplegic arrest, preconditioning, etc.) are currently performed blindly, i.e., with minimal online indication of their success. In conclusion, this study not only confirms MEI as a valid online myocardial ischemia monitor, but it highlights the importance of including parameters sensitive to the MEI time course in the ischemia monitoring process. FOOTNOTESREFERENCES
Page 8toxic shock syndrome (TSS) was first described by Todd et al. (39) in 1978 as a multi-system illness characterized by the rapid onset of fever, hypotension, and multi-organ involvement, followed by desquamation upon recovery. Menstrual TSS (mTSS) has been associated with tampon use during menstruation. Despite the low incidence of mTSS, the illness remains of interest because tampons are widely used, and although rare, mTSS can be life threatening. In a descriptive research study, Czerwinski (11) reported that ∼80% of study participants (women from California under of the age of 41) used tampons during menstruation. It has also been reported that about 70% of women in the USA, Canada, and much of Western Europe use tampons during menstruation (25). TSS is thought to be caused by colonization with Staphylococcus aureus capable of producing the superantigen, toxic shock syndrome toxin-1 (TSST-1) that penetrates through the mucosal surface (12) in a person who lacks neutralizing antibodies to TSST-1. S. aureus has previously been determined to colonize the nares, axillae, vagina, vulva, anus, pharynx, or damaged skin of 30–50% of healthy adults (8, 18). This figure could be much higher based on results of recent analyses using culture-independent methods to identify S. aureus (40). Although S. aureus is capable of producing several superantigens, TSST-1 is considered to be the cause of nearly all cases of mTSS and at least 50% of nonmenstrual cases (2). The development and progression of TSS may depend on host susceptibility and factors that regulate TSST-1 production by S. aureus. A large percentage of the population develops specific antibodies to TSST-1 during the first decade of life (41). Lack of anti-TSST-1 neutralizing antibodies has been associated with cases of mTSS (5, 32). In addition, standard culture-based methods have shown that only about 10–20% of vaginal S. aureus isolates produce TSST-1 (14, 20, 21). Toxin production has been reported to be higher during menses, possibly due to the altered levels of iron, O2, CO2, pH, hormones, and osmolarity, which could affect the total number of bacteria present, the number of colonizing species (7, 36, 42), and/or gene expression (45). Tampax tampons were introduced in the United States in 1936. One of the first clinical studies indicating the safe use of tampons was published in 1942 (22). Their continued safe use has been subsequently documented (35). It has been shown, however, that there is an increased risk for mTSS attributed to tampon use (13, 27, 30, 33). Epidemiological studies have implicated five other risk factors for mTSS, including young user age, tampon absorbency, continuous usage of tampons without use of pads, tampon composition, and oxygen entrapped in the tampon, although none has been definitively confirmed (3, 17, 28). Numerous in vitro studies have demonstrated the importance of O2 tension in the regulation of TSST-1 production by S. aureus (29, 38, 44, 45). Results of early work in this area prompted the evaluation of O2 as a risk factor in later epidemiological studies (17, 28). A study by Wagner et al. (43), suggested that a large, sustained bolus of O2 is introduced into the vagina as a result of tampon insertion and that this converts the vaginal environment from an anaerobic to aerobic state. This change in vaginal environment has been postulated to be the key link between tampons and mTSS; however, the placement of the large Clark electrode sensing surface against the tampon surface may have actually measured the gas tensions at the surface of the tampon, rather than in the vaginal environment. Recent advances in sensor technology now allow the simultaneous monitoring of O2 and CO2 tensions in the vagina, as well as inside the tampon during wear. Umbilical microsensors were used to address several experimental questions. Do the concentrations of O2 and CO2 change in the vagina during and post menses? Does tampon use, as well as tampon absorbency, affect vaginal O2 and CO2 concentrations during and post menses? Can menses be a source of oxygen for the vaginal environment? Does vaginal colonization with S. aureus affect the vaginal gas profiles measured during tampon use? Results from this translational research have generated new discoveries through basic scientific inquiry by the process of applying ideas, novel techniques, and discoveries that have the potential to lead to new insights into diseases of the human female urogenital system. MATERIALS AND METHODSParticipants were recruited and gave written consent before participation. The study protocol and informed consent documents were approved by The Procter and Gamble Corporate Institutional Review Board. All participants were at least 21 years of age, in good general health (self-reported), had menstrual cycles between 21 and 35 days with menstruation lasting at least 3 days, had no present or previous gynecological complaints, typically used tampons for normal menstrual protection, and agreed to abide by the study requirements and restrictions. Participants were excluded if they reported a history of TSS or symptoms consistent with TSS, streptococcal infections within the last 3 mo, difficulty wearing tampons, or body piercing in the vulvar area, or if they were currently pregnant or currently suffering from diabetes. Inspections of the external genitalia were performed to exclude any obvious pathological disorder, such as abscesses or lesions. Vaginal swabs were obtained from the majority of participants and were analyzed for the presence of vaginal S. aureus using a standard culture-based method. Table 1 shows the demographics of study participants and the studies in which they participated. Due to participant availability, not all qualified women were able to participate in all phases of these studies.
Vaginal swabs were prepared for analysis within 24 h of collection. Swabs were streaked on mannitol salt agar (Difco, Detroit, MI) and incubated aerobically overnight at 37°C. Identities of presumptive S. aureus isolates were confirmed via Staphaurex Slide Coagulation Test (Murex Diagnostics, Dartford, UK) and gram stained (Deaconess Hospital, Cincinnati, OH). The Neotrend sensors (Diametrics Medical, St. Paul, MN) used in this study were initially developed for intra-arterial monitoring of critically ill neonates (24, 37). This is the first application of these devices to the measurement of gas tensions in the vaginal environment. Each sensor is composed of four sensing units, three fiber optic-based units for simultaneously monitoring dissolved O2, CO2, pH, and a thermocouple for measuring temperature. For perspective, these sensors (<0.5 mm in diameter and 23 mm in length) are considerably smaller in diameter than the Clark electrodes (24 and 10 mm) used in the previous study (43). On the basis of their application to intra-arterial blood gas monitoring, the response time of the sensor unit is expected to be less than 15 s at 37°C. The detection ranges for Po2 and Pco2 in blood gases are expected to be 20–500 mmHg and 10–160 mmHg, respectively, and drift in the signal is expected to be less than 0.5%/h of operation. The sensors used in this study were calibrated and deployed according to the manufacturer’s general instructions for use. In addition, assessments were made to evaluate the ability of these sensors to measure Po2 and Pco2 in menstrual fluid. Two types of tampons were used in this study, Tampax regular absorbency (Lot #27N909049, #0165243038, #0165243028, The Procter & Gamble Company, Cincinnati, OH) and Kotex Super absorbency (Lot #VP 9306, Kimberly-Clark, Neenah, WI). These two tampons were selected to represent the range of entrapped air found in commercially available tampons. Tampax regular tampons contain a blend of cotton and rayon fibers with an absorbency rating of between 6 and 9 g.1 1All absorbency ratings for tampons used were based on release criteria set by in vitro absorbent capacity measurements using the syngyna test as specified in 21 CFR 801.430(f)(2). Tampons such as these have been found to contain the least amount of entrapped air (19). Kotex Super tampons are made of rayon fibers with an absorbency rating of between 9 and 12 g. Tampons such as these have been found to contain the most entrapped air (19). These findings were independently verified in our laboratory for the actual products used in this study (data not shown).The tampons were modified to monitor the environment inside the tampon during use. The tampon-sensor assembly was made by producing a small hole (<2 mm in diameter) in the center axis of the tampon. A cutoff radial artery catheter (Arrow International, Reading, PA) was sutured to the tampon, using sterile suture thread (Ethibond Excel, Ethicon, Sommerville, NJ), to support the sensor as it was deployed into the tampon. The collar acted as the anchor point of the catheter as it was sutured gently to the base of the tampon to hold the catheter in place. Samples of menstrual fluid were collected and analyzed for the levels of O2 and CO2 using several analytical approaches. Eight participants provided a total of 17 samples of menstrual fluid for analysis. One participant provided a total of seven samples over four consecutive menstrual cycles. Samples of menses were collected using a modified INSTEAD 12 Hour Feminine Protection Cup (Ultrafem, Missoula, MT) to minimize disruption of the vaginal environment and exposure of menses to atmospheric conditions before analysis. The INSTEAD Cup is a nonabsorbent, flexible cup designed and marketed to collect menses. When inserted according to the manufacturer’s instructions, the cup fits just below the cervix. The cups were individually modified in the laboratory, as described below, to allow the collection, detection, and sampling of menses for subsequent analysis of blood gases. Water vapor transmission analysis showed that the INSTEAD Cup has a low water vapor transmission (0.32 mg·in−2·h−1at 23°C and 50% relative humidity); therefore, we expected very low permeability to CO2 and O2 during the course of menses collection (30–50 min). Three holes were made in the plastic film of an INSTEAD Cup. The first hole was 0.1 cm from the outer rim. The second hole was 1 mm from the first hole, and a third hole was ∼2 cm from the center of the film. The tip of an umbilical artery catheter (Diametric Medical, St. Paul, MN) was inserted through the first hole. Suture material (Ethibond Excel green braided polyester surgical suture thread, Ethicon, Sommerville, NJ) was used to secure the plastic of the cup to the catheter by wrapping and tying the suture thread around the extended plastic and catheter. A Neotrend sensor (Diametrics Medical) was attached and deployed through catheters in the first two holes. The plastic film was bunched around the catheters and secured by an elastic orthodontic band. This improved pooling of menses in the cup and reduced potential leakage from the cup. The blue butterfly end of Vacutainer tubing (23 gauge; 3/4 in; 12 in length; Becton-Dickinson, Franklin Lakes, NJ) was inserted through the third hole, allowing the tubing to extend on the inside surface of the cup. The tubing for each device was secured by twisting an orthodontic elastic band around the end of the tubing on the inside of the cup. The tubing was then retracted until the band rested against the inner surface of the cup. The needle from the white port of the Vacutainer tubing was removed and discarded and the end was capped. Participants inserted the modified cup for the collection of menstrual fluid. After ∼15–30 min of wear, two Neotrend sensors were inserted into the catheters and deployed into the cup. Sensor outputs were allowed to stabilize for 15–20 min. When the sensor output (Po2, Pco2) indicated a change from the initial reading, menses was withdrawn from the cup using a 1-ml non-heparinized syringe attached to the luer adaptor of the tubing. Immediately after collection, menses was analyzed using the Immediate Response Mobile Analysis (IRMA) SL Blood Analysis system (Diametrics Medical) according to the manufacturer’s instructions for blood samples. In addition, the IRMA system was evaluated for its suitability for this work by comparing blood gas values obtained from standard samples (Defibrinated Sheep Blood, Cleveland Scientific, Cleveland, OH) with results obtained using a standard blood gas analyzer (Corning model 388) at Deaconess Hospital (Cincinnati, OH). Vaginal mapping evaluations were designed to evaluate participants before and after the insertion of a tampon under menstrual and nonmenstrual conditions (Table 2). The purpose was to allow participants to act as self-controls. Initial mapping evaluations during menstruation were typically performed on the second day (i.e., 24–30 h post initiation of menstrual flow) because it typically demonstrates the highest average menstrual flow (34) and mTSS has been reported to be more prevalent during the first 2 to 3 days after onset of menstruation (13, 33). In several cases, additional evaluations were performed on the third day of menstruation (Table 2). In addition to menstrual samples, nonmenstrual evaluations were conducted during mid-cycle (days 10–20) to evaluate the vaginal environment in the absence of menses.
For all evaluations, participants were in a supine position during the entire monitoring period. Neotrend sensors were positioned at two different vaginal sites to obtain baseline O2 and CO2 levels (Fig. 1A). To deploy the vaginal sensors in the desired positions (Fig. 1A), the umbilical artery catheters used to deploy the sensors were precut to lengths that differed by 2 cm. The two catheters were placed side-by-side and sutured to ensure a spacing of 2 cm from the end of one catheter to the other catheter. The catheter assembly was premarked to indicate the depth to deploy the sensors for proper placement in the vagina. Fig. 1.Placement of vaginal sensors in vaginal canal pre (A)- and post (B)-tampon insertion. The vaginal-sensor assembly was then digitally inserted to the predetermined depth into the vaginal lumen by a medical professional. The sensors were deployed by extending the sensors from the arterial catheter and allowed to equilibrate for 10–15 min. Readings from both sensors were acquired for up to 45 min to establish baseline vaginal levels of CO2 and O2 from both vaginal sites. Data points were collected every 30 s via an RS232 port and downloaded via the hyperterminal into an Excel spreadsheet. After the baseline response was recorded, the vaginal-sensor assembly remained in the vagina. To prevent damage during insertion of the tampon-sensor assembly, the vaginal sensors were retracted into the umbilical artery catheters. The tampon-sensor assembly was then inserted either digitally or using a Kotex Super absorbency tampon applicator. Once the tampon-sensor assembly was in place, the vagina sensors were carefully redeployed from the catheters (Fig. 1B). Partial pressures of O2 and CO2 were then monitored from all three sensors. Data collection continued, as described previously, for an additional 6–8 h. Tampons were weighed before assembly of the tampon-sensor device as well as on completion of menstrual mapping evaluation. Tampon menses load was determined by weight difference. Tampon menses load values obtained were either less than or equal to 3.5 g or greater than 6.0 g. For the purposes of this study, tampons with 3.5 g or less of menses were considered “low load.” Tampons containing over 6 g of menses were considered “high load.” Tampons from study participants were removed and frozen at −70°C after mapping studies and stored for analyses of the presence and location of S. aureus on the tampons pending vaginal culture results. Subsequently, tampons from the colonized subject worn during and post menses were analyzed via PCR and fluorescent in situ hybridization (FISH). To prepare samples for PCR and FISH analyses, each of the frozen tampons was aseptically cut into 12 pieces using autoclaved razor blades and flamed forceps (Fig. 2). Four sections of each tampon (tip, upper middle, lower middle, and base) were each cut into three zones (outer, bulk, and core). Each piece was placed into a sterile Whirl-Pak bag (VWR Scientific, West Chester, PA) with 3 ml of sterile water and stomached for 3 min to extract bacterial cells. Fluid was removed, aliquoted, and frozen. The stomached fibers were air dried at 37°C until the weight remained stable for 2 consecutive days. Fig. 2.Diagram of frozen tampon dissection into 12 pieces. Tampons from the participant who was colonized with S. aureus were analyzed with culture-independent methods (see Table 1). The presence of S. aureus was confirmed using PCR (6), and the concentration of S. aureus in each subsample was determined using FISH analysis (40). FISH has been shown to be more sensitive for the detection of S. aureus than culture-based methods (9, 40). Fluid from the stomached aliquots was spotted into separate wells on 5-mm-well microscope slides (Erie Scientific, Portsmouth, NH), as were negative and positive control cultures (S. epidermidis and S. aureus, respectively). The spotted fluids were allowed to air-dry, then covered with 100% ethanol and air-dried again. Cells in each well were then fixed by covering with fresh 4% paraformaldehyde for 1 h at 4°C, then rinsed with distilled water (dH2O), being careful to prevent run-off from any well passing over other wells. After air-drying, fixed cells were permeabilized by covering the cells for 45–60 min at 37°C with a permeabilization solution: 0.1 mg/ml lipase (Sigma, l-0382) and 3 mg/ml lysozyme (Sigma, l-7651), in 25 mM Tris, pH 8.0. After the permeabilization treatment, the wells were rinsed with distilled water and slides were frozen at −20°C until required for FISH analysis. FISH for the specific detection of S. aureus was conducted according to published procedures (40) using probes Saur327 and Saur72. All FISH-positive cells were enumerated by scanning the whole wells microscopically. Any counts from the negative control wells were subtracted as background. These background-subtracted counts were normalized to dry weight of the tampon subsample. Negative controls used for each sample were S. epidermidis culture with S. aureus-specific probes, S. epidermidis with no probes, and the sample with no probes. The positive control well included S. aureus with S. aureus-specific probes. The presence of S. aureus was confirmed using PCR (data not shown). Genomic DNA was extracted and isolated from each stomached fluid aliquot using a Blood Spin kit (Mo Bio Labs, Carlsbad, CA). Extracted DNA was amplified by PCR using S. aureus specific primers (nuc gene) (6), and the product was examined on agarose gel to confirm the presence of S. aureus. Before statistical analysis, the raw experimental data were examined in detail and some data were excluded from further analysis for specific reasons. The following data were excluded from the statistical analysis: 1) data recorded before equilibration of the sensors or while the sensors were withdrawn into catheters; 2) data from tampon sensors that showed spikes due to interferences associated with direct contact between blood and the sensor tip (37); and 3) results that showed evidence of obvious sensor malfunctioning during the course of the evaluation. For each gas/experiment/vaginal site value, basic time units were 5-min intervals from the time of tampon insertion. Within each time interval, gas (CO2 or O2) levels were averaged. Primary statistical significance was declared at the two-sided 0.05 significance level. PC SAS Release 8.1 (SAS/STAT User’s Guide, Version 8, 1999; SAS Institute, Cary, NC) was used to analyze all data. A mixed model analysis of covariance (ANCOVA) was conducted to compare gas levels from the two vaginal sensors (cervix and midzone). For each menstrual/nonmenstrual condition and tampon insertion phase, the fixed terms in the ANCOVA model were tampon type, vaginal site, and tampon type by site interaction. The random term was “participant,” and covariate was “menses load level.” Because of insufficient data (i.e., no Super Kotex cervix nonmenstrual data), the comparison between cervix and midzone was not estimable for some cases using this model. Thus another ANCOVA model was also conducted for each menstrual/nonmenstrual condition and tampon insertion phase, but included only “vaginal site” as a fixed factor. Again the random term was “participant,” and covariate was “menses load level.” A mixed model ANCOVA was also used to compare pre-tampon average gas levels to post-tampon average gas levels. For each vaginal source and menstrual/nonmenstrual condition, the fixed terms in the ANCOVA model were tampon type, pre- or post-tampon insertion phase, and tampon type by insertion phase interaction. The random term was participant, and the covariate was menses load level. (For some combinations of vaginal source and menstrual/nonmenstrual condition, the ANCOVA could not be conducted because of insufficient data.) ANCOVA results indicated that tampon type data could be combined. Thus a series of paired tests was conducted to use all available data. For each experiment/vaginal site, the average pre- and post-tampon gas levels were calculated. The post-tampon minus pre-tampon deltas then were derived per experiment/vaginal site. For each load category, an average delta per subject/vaginal site also was calculated. Both the deltas per experiment/vaginal site and average deltas per subject/vaginal site were analyzed for each load category, both separately and combined, via the UNIVARIATE procedure in SAS. High and low load data from subjects who had both load categories under the same menstrual/nonmenstrual condition in separate experiments were treated as independent. If the deltas were normally distributed, then paired t-tests were conducted; otherwise, signed-rank tests were conducted. The Shapiro-Wilk normality tests were conducted at the 0.05 significance level. The above statistical tests were performed by both including and excluding nonmenstrual data from an experiment on one subject, where only two valid pre-tampon insertion measurements were made per vaginal site. Results were found to be similar with and without the inclusion of these data. To evaluate the effect of tampon menses load on O2 and CO2 values within tampons during menstruation, a mixed model ANOVA was conducted. For each gas/experiment, the following parameters were derived from the tampon menstrual measurements: delta (first − last value) for O2 and CO2 profiles, maximum in CO2-O2 differences and maximum in O2/CO2. These parameters were analyzed via the MIXED procedure in SAS. The fixed terms in the ANOVA model were tampon type, load category, and tampon type by menses load category. The random term was participant. For these analyses, ANOVA with menses load category (light, moderate, and heavy) as a class factor was used, because the focus was on distinguishing the effects of specific load levels. This objective differs from that of the ANCOVA models with menses load as a continuous covariate. ANCOVA was run to provide precise as possible estimates of the effect of tampon type (Tampax vs. Kotex), tampon phases (pre- vs. post-tampon insertion), and vaginal site (cervix vs. midzone) on gas levels. RESULTSResults showed that the IRMA system was capable of measuring mean Po2 and Pco2 in standard samples with accuracy and reproducibility similar to that obtained with the Corning BGA. Good agreement was observed between measured values for mean Po2 (Corning BCA 140.20; IRMA 138.86 mmHg) and Pco2 (Corning BGA 56.40; IRMA 57.24 mmHg), as indicated by the small differences in the measured mean values (1%). In addition, the relative standard deviation (RSD) in the mean values, a measure of the variably in results between multiple samples, was found to be very similar for both approaches (Corning BGA 5%; IRMA 6%). The IRMA system was then used to analyze samples of menstrual fluid from study participants. The mean Po2 and Pco2 values obtained for seven samples of menstrual fluid taken over four menstrual cycles from participant 9 were found to be 41.6 ± 6.9 and 54.2 ± 3.4 mmHg, respectively. Compared with standard samples, the results showed an increase in the variability of the measured Po2 (from 6 to 17%), as indicated by an increase in the RSD of the mean. This increase in variability appears to reflect the normal fluctuation of Po2 within an individual. In addition, the IRMA system was also used to analyze 17 menstrual fluid samples from eight study participants. Means were obtained for each subject before calculation of the overall mean. The overall mean Po2 and Pco2 values were found to be 41.8 ± 16.3 and 43.5 ± 9.2, respectively. Results from this larger sample set showed a further increase in the variability in the mean Po2 and Pco2 (RSD; 17–39% and 6–21%, respectively). Finally, experiments were conducted to assess the ability of the Neotrend sensors to measure Po2 and Pco2 in menstrual fluid. This was done by comparing results obtained for nine samples from six study participants using Neotrend sensors with results obtained using the IRMA analyzer. Results indicated that Neotrend sensors were capable of measuring Po2 and Pco2 in menstrual fluid with accuracy and reproducibility similar to that obtained with the IRMA analyzer. Agreement in the mean Po2 and Pco2 values between the two measuring approaches for these samples was high (IRMA 37.77 ± 11.53, 48.96 ± 8.64; Neotrend 35.44 ± 11.25, 49.68 ± 8.27 mmHg). In addition, the RSD in the mean values was found to be very similar for both O2 and CO2 using both approaches (IRMA 31%; 18% Neotrend 32%; 17%). In general, the design objectives for this study were met, with only two exceptions (Table 2, participants 1 and 9). Statistical analysis of absolute vaginal gas levels showed no significant differences between the values measured by the sensors placed in the cervix or midzone regions throughout the course of these experiments (data not shown). Differences were observed in the changes in vaginal gas levels before and after tampon insertion. Similar trends were observed regardless of the sensor location (cervix or mid-zone), the type of tampon inserted (Tampax Regular or Kotex Super), or the time during the menstrual cycle that the experiments were conducted (day 2, day 3, or mid-cycle). In all cases, results showed a decrease in the calculated mean Po2 measured in the vaginal canal after insertion of a tampon. On the other hand, the calculated mean Pco2 measured in the vaginal canal after tampon insertion either increased slightly or remained the same. The data for the two types of tampons were combined to increase the power of the statistical analysis and to compare our findings with previous work (43). This was possible because statistical analysis showed that the changes in mean O2 and CO2 levels were independent of the type of tampon inserted (see Statistical analysis). Combined tampon results obtained for mid-cycle and days 2 and 3 are shown in Figs. 3 and 4, respectively. As shown in Fig. 3, insertion of a tampon during mid-cycle decreased the mean partial pressure of O2 in the vaginal environment, as measured by sensors in both the cervix (P = 0.108) and mid-zone (P = 0.043) locations. The decrease in mean O2 level measured by the mid-zone sensor was found to be significant at the two-sided 0.05 significance level (P = 0.043). Fig. 3.Mean values of O2 and CO2 measured pre- and post-tampon insertion by sensors located in the vagina during mid-cycle. Solid bars, baseline or pre-tampon insertion; open bars, post-tampon insertion. Refer to Table 2 for n values. Standard error in the deltas (differences) were cervix O2 = 14.60; midzone O2 = 5.31; cervix CO2 = 2.10; midzone CO2 = 5.06. Note that atmospheric values for O2 would be 100–140 mmHg; for CO2 5 mmHg. Fig. 4.Mean values of O2 and CO2 measured pre- and post-tampon insertion by sensors located in the vagina during days 2 and 3. Solid bars, baseline or pre-tampon insertion; open bars, post-tampon insertion. Refer to Table 2 for n values. Standard error in the deltas (differences) were cervix O2 = 7.31; midzone O2 = 4.80; cervix CO2 = 2.48; midzone CO2 = 4.18. Note that atmospheric values for O2 would be 100–140 mmHg; for CO2 5 mmHg. Conversely, the combined data set showed a significant increase in mean Pco2 in the vaginal environment (cervix sensor P = 0.008; mid-zone P = 0.031) on insertion of a tampon. Similar trends were observed in the results for vaginal O2 and CO2 obtained during menstruation (Fig. 4). Insertion of a tampon during days 2 and 3 of menstruation decreased the mean Po2 in the vaginal environment, as measured by sensors in both the cervix (P = 0.63) and mid-zone (P = 0.031) locations. The decrease in mean O2 level measured by the mid-zone sensor was found to be significant at the two-sided 0.05 significance level (P = 0.031). The findings for CO2 showed only a slight increase after tampon insertion, as measured by the cervix sensors (P = 0.101), and no change in CO2 level as measured by the mid-zone sensor (P = 0.862). It is also important to note that the mean vaginal Po2 measured before tampon insertion (mid-cycle or days 2 and 3) were in the range of 15–35 mmHg, much lower than atmospheric levels (100–140 mmHg). Conversely, the mean values of Pco2 before tampon insertion ranged from 35 to 55 mmHg, much higher than levels expected for atmospheric conditions (5 mmHg). The initial response of the tampon sensor after insertion of the tampon-sensor assembly into the vaginal canal was similar regardless of the type of tampon (Tampax Regular or Kotex Super) or the time during the menstrual cycle that the experiments were conducted (days 2 and 3 or mid-cycle). In all cases, the tampon sensor initially indicated readings consistent with atmospheric conditions (O2 between 100 and 140 mmHg and CO2 <20 mmHg). Over the course of the evaluation, the response of the tampon sensor was observed to conform to one of two general profiles (Figs. 5–7). Fig. 5.Typical O2 (solid lines) and CO2 (dashed lines) profiles obtained from a sensor embedded within a tampon for 4 participants during nonmenstrual conditions (mid-cycle). Data shown include A: participant 9, experiment (expt) C, Tampax Regular; B: participant 2, expt A, Tampax Regular; C: participant 7, expt D, Tampax Regular; D: participant 5, expt F, Tampax Regular. The O2 and CO2 profiles obtained from tampon sensors used during nonmenstrual conditions and from tampon sensors used during menstruation that subsequently were found to have low menses loads (≤3.5 g) were found to be similar in general characteristics (Figs. 5 and 6). In most cases, the measured Po2 in tampons was observed to decline slowly from an initial high value but remained high (>100 mmHg) during the course of the evaluation and did not approach the low Po2 measured in the vagina (Figs. 3 and 4). On the other hand, Pco2 in tampons were observed to increase rapidly from initial low values and approach the partial pressure measured in the vagina (50–65 mmHg). Fig. 6.Typical O2 (solid lines) and CO2 (dashed lines) profiles obtained from a sensor embedded within a tampon for 4 participants during menstrual conditions (days 2 and 3). Subsequent to these evaluations, menses loading for these tampons was found to be low (<3.5 g). Data shown include A: participant 5, expt A (day 2), Tampax Regular, menses load = 2.172 g; B: participant 5, expt D (day 3), Kotex Super, menses load = 2.89 g; C: participant 7, expt B (day 3), Tampax Regular, menses load = 1.328 g; D: participant 2, expt B (day 2), Tampax Regular, menses load = 3.509 g. Appreciably different O2 and CO2 profiles were observed from tampons subsequently found to be highly loaded with menses (>6.0 g), as shown in Fig. 7. In these cases, the measured Po2 was found to drop to below 60 mmHg and the measured Pco2 was observed to rise to levels above those typically measured in the vagina during menstruation (50–65 mmHg). Although the profiles shown in Fig. 7 indicate some variation in these trends, profiles under these conditions show an intersection between O2 and CO2 profiles. Fig. 7.Typical O2 (solid lines) and CO2 (dashed lines) profiles obtained from a sensor embedded within a tampon for 4 participants during menstrual conditions (days 2 and 3). Subsequent to these evaluations, menses loading of tampons was found to be high (>6.0 g) and to exceed the absorbency rating of the tampon (Tampax Regular 6–9 g; Kotex Super 9–12 g). Data shown include A: participant 5, expt C (day 2), Tampax Regular, menses loading = 12.645 g; B: participant 9, expt B (day 3), Kotex Super, menses loading = 15.009 g; C: participant 7, expt A (day 2), Kotex Super, menses loading = 16.842 g; D: participant 1, expt A (day 2), Tampax Regular, menses loading = 10.752 g. The data from these complex curves were subjected to statistical analyses to determine whether the trends observed were real. Multiple statistical models were used to evaluate the data, and all models provided similar conclusions. The model reported in Table 3 compared the changes in Po2 and Pco2 values at the beginning the end of the experiment to determine whether the changes were real. The results of the statistical analysis demonstrated that the Po2 and Pco2 levels in tampons were significantly different between low load and high-load tampons. The results allow us to discuss the implications of these data.
Of particular interest are tampon O2 and CO2 profiles obtained from a single participant on 2 consecutive days during menstruation wearing different tampon products (see Fig. 8). Intersections were observed in both cases, but the times at which the intersections occurred varied from ∼1 to 3 h after the tampon-sensor device was inserted. Fig. 8.O2 (solid lines) and CO2 (dashed lines) profiles obtained from participant 9 on 2 consecutive days during menstruation wearing different tampon products. A: day 2 wearing Tampax Regular tampon found to contain 11.972 g of menses (absorbency rating 6–9 g); B: day 3 wearing Kotex Super found to contain 15.009 g of menses (absorbency rating 9–12 g). Results in Table 4 show that S. aureus was detected in both tampons worn by participant 7. For one subsample the finding was below the detection limit for FISH analysis, possibly because it was a physically small sample (<50 mg). Dry weight normalization revealed densities of between 3 × 105 and 1 × 108 S. aureus cells per gram (dry weight), distributed evenly throughout all sampling zones and regions. PCR results confirmed the presence of S. aureus in >80% of subsamples. Interestingly, the highest S. aureus densities were found in the tampon worn nonmenstrually.
DISCUSSIONThe microsensors used in this seminal work allowed simultaneous in vivo measurements of the Po2 and the Pco2 in the vagina and within tampons. Findings show the vaginal environment is anaerobic before tampon insertion, which is consistent with expectations and Wagner’s earlier work (43). In contrast to his findings, the process of tampon insertion was not observed to introduce a bolus of oxygen into the vaginal environment. In fact, surprisingly, measurements by two independent vaginal sensors showed decreases in the vaginal Po2 on tampon insertion. The observed decrease in Po2 was independent of tampon types used (Tampax Regular and Kotex Super) and the time of the cycle when the measurements were made (mid-cycle or days 2 and 3). Insertion of a tampon either increased or had little impact on the Pco2 in the vaginal environment. The mean values for Po2 and Pco2 in menstrual fluid measured in this study were similar (41.8 and 43.5 mmHg, respectively) and tended to be in the range expected for venous blood. The range of values observed across the entire sample base was greater than the range of either venous or arterial blood (22.0–68.4 mmHg for Po2 and 29.5–54.2 mmHg for Pco2). Our findings suggest that menses can provide a source of oxygen to the vaginal environment. These findings may be important for genital diseases that affect women more frequently during menses such as pelvic inflammatory disease, cervical Chlamydia trachomatis, human immunodeficiency virus, and bacterial vaginosis as well as menstrual vaginal TSS (15, 16, 23, 33). Despite differences in methods used for collection and analysis of menses, as well as differences in participant demographics, our findings are in general agreement with those reported by Wagner (42). In later in vivo work, Wagner (43) concluded that tampon insertion resulted in the introduction of a bolus of oxygen and this was a major source of oxygen to the vaginal environment. Our in vivo results do not show tampon insertion to be a source of vaginal oxygen. Insertion of a tampon was found to decrease the Po2 in the vagina. Before tampon insertion, vaginal gas levels ranged from 15 to 35 mmHg for Po2 and 35 to 55 mmHg for Pco2. These values are lower and higher, respectively, than atmospheric levels (Po2 = 100–140 and Pco2 = 5 mmHg). Compared with Wagner’s (43) results, the absolute values reported here are higher for Po2 (15–35 vs. 3 mmHg) and lower for Pco2 (35–55 vs. 64 mmHg). The higher Po2 could be due to the fact that the vast majority of the participants in this study had been pregnant, whereas none of the participants in Wagner’s study had been pregnant. Prior pregnancies could allow higher levels of O2 to enter the introitus, which could account for the apparent higher Po2 measured here. Nevertheless, these findings are generally consistent with expected anaerobic conditions in the vagina. On insertion of a tampon, our results showed a decrease in the Po2 in the vaginal environment. Insertion of a tampon either increased or had little impact on the Pco2 in the vaginal environment. These findings are not in agreement with Wagner’s results (43). Wagner reported an increase in Po2 from a mean value of 3–112 mmHg and a decrease in Pco2 from 64 to 50 mmHg. Whereas several factors such as differences in participant demographics, measurement devices, and types of tampons evaluated could contribute to the conflicting findings, our analysis suggests that very small sensors that remain exclusively inside the vaginal environment are capable of monitoring changes in vaginal environment independent of changes at or near the tampon. The tampon sensor used in Wagner’s study is described as a tampon (Tampax Regular, o.b. normal, or Playtex Regular) with a Clark electrode attached. Wagner used the best technology available at the time, a Clark electrode, with a diameter of 24 mm, which is at least twice the diameter of the tampons used in that study (∼10 mm). The electrode faces (i.e., measurement surfaces) were placed toward the tampon material and then inserted into the vagina. As a result, Wagner’s post-tampon insertion findings are more likely to correlate with findings from the tampon-sensor assembly used in this study. After insertion into the vagina, tampon sensors used in this study showed initial Po2 and Pco2 to be consistent with atmospheric conditions. As the evaluation continued, the Pco2 measured in the tampon generally increased, and the Po2 in the tampon tended to decrease. In cases where the menses load was found to be extremely high, the individual gas profiles for O2 and CO2 were found to intersect each other for both tampons evaluated in this study (Tampax Regular and Kotex Super). In other cases (nonmenstrual and low load), the individual profiles were not observed to intersect within the time frame of these experiments. The wear times at which the intersections occurred were observed to vary from ∼1 to 3 h and appeared to be related to the tampon absorbency and menses loading. As hypothesized, the tampon sensor findings reported here are in some aspects consistent with post-tampon insertion results reported by Wagner (43). Wagner reported changes in gas profiles on tampon insertion similar to those observed here from tampon sensors; however, Wagner did not report intersections between O2 and CO2 profiles. This may be attributed to the relatively short duration of the majority of Wagner’s evaluations (90 min) and the narrow demographic profile of his participants (nursing students 22–24 yr of age, nulliparous, and exclusively of Northern European descent). FISH/PCR analysis of selected tampons used in this study showed S. aureus to be present and evenly distributed throughout tampons whether worn menstrually or nonmenstrually. It is important to note that this approach identifies the presence of S. aureus, but cannot discriminate between toxigenic and nontoxigenic S. aureus strains. Vaginal colonization with S. aureus (both toxigenic and nontoxigenic) in healthy women has been reported infrequently. Vaginal carriage of these strains in 495 healthy menstruating women was found to be low (2.6% toxigenic vs. 4.0–5.2% nontoxigenic; Ref. 10). The majority of healthy women who are colonized with these strains also demonstrate measurable antitoxin titers (4). As noted above, tampon sensors showed Po2 in the tampon tended to decrease and the Pco2 tended to increase over the course of the vaginal mapping experiments. Importantly, we did not measure a corresponding increase in Po2 in the vaginal environment; therefore, this consumption of O2 and subsequent production of CO2 measured within the tampons suggests that respiration may be occurring within the tampon. It has been demonstrated that vaginal bacteria can colonize the tampon during wear. It has also been shown that other microbes can colonize tampons during menstruation, and the microbial population usually reflects that which is normally cultured in the vagina (26). Vaginal colonization with S. aureus did not affect vaginal gas profiles in the subject who was culture positive for this organism. We did not observe intersections in the O2 and CO2 profiles for tampons with low menses load or for tampons worn nonmenstrually. It appears that near saturation of tampons with menses may be associated with increased bacterial respiration resulting in the intersection of the O2 and CO2 profiles within the timeframe of these experiments. The nonmenstrual O2 and CO2 profiles showed a trend that may suggest similar profiles after excessive wear times. There are multiple hypotheses as to why the respiration would be altered with tampon loading and/or longer wear times. For example, as the bacteria increase over time, a subsequent reduction in Po2 with a commensurate increase in Pco2 would be expected. As the vital nutrients are consumed due to bacteria growth, the respiration rate would lessen, explaining the shape of these curves. Superimposed on this would be the complex effects of the aerobes. We speculate that the accelerated rate of change in the O2 and CO2 profiles associated with saturated tampons is related to the enhanced nutrient medium that menses provides. It is recognized that toxin production is enhanced when the organism becomes stressed due to nutrient depletion as well as enhanced when CO2 levels rise. Although these data potentially provide theoretical underpinning for appropriate selection of wear time criteria,2 2Regulatory communication from the Food and Drug Administration to Manufacturers of Menstrual Tampons, September 13, 1993. the association of mTSS with wear time has never been clearly elucidated in epidemiological studies (27).The specific environmental conditions that promote the in vivo production of TSST-1 by toxigenic S. aureus are not known. Numerous in vitro studies have demonstrated the importance of Po2 in the regulation of TSST-1 production (29, 38, 44). It has also been shown that an increase in the Pco2 in the presence of O2 can increase the rate of TSST-1 production under in vitro conditions (45). Other studies have shown that a neutral pH environment is necessary for TSST-1 production (31), which is the pH of menses (6.8–7.0; Ref. 1). Production of TSST-1 requires a nexus of tampon conditions, the absence of antibodies to TSST-1, and the presence of the appropriate S. aureus strain. It has been found that most women colonized with toxigenic S. aureus also have measurable antibody titers (4). There are other factors known to affect gene regulation and expression that these results do not address. Nevertheless, it is possible to propose a putative model based on these findings that may help to understand the association of mTSS with menstruation and tampon usage. Before menstruation, the vaginal environment is anaerobic, and with a pH of about 4.5, neither condition is conducive to TSST-1 production. The presence of menses increases the pH of the vaginal environment and introduces a source of O2 and CO2. If toxigenic strains of S. aureus are present, conditions could be favorable for TSST-1 production. This might explain the association of mTSS with menstruation in the absence of tampon usage. The insertion of the tampon introduces a new ecological niche within the vagina. Bacteria colonize the tampon and, as menses enters the tampon, the bacteria are provided with an O2 and nutrient-rich environment. This environment, with a near neutral pH and increasing levels of CO2 due to bacterial respiration, can lead to TSST-1 production. This putative model integrates the results of our in vivo studies with previous in vitro and epidemiological studies. DISCLOSURESThis work was sponsored by The Procter & Gamble Company (P&G). All authors of this manuscript, with the exception of Dr. Schlievert, are currently or have been employees of the Procter & Gamble Company. Dr. Schlievert consulted with the group regarding the historical knowledge of human vaginal gas tensions, design of the studies, and development of the manuscript. He did not receive compensation for the work associated with this study. FOOTNOTESWe thank the following individuals for their important technical assistance: Michaelle Jones (P&G) for the design and execution of the statistical analyses, Kathy Kerr (P&G) for the PCR work, and Bill Hood (P&G) for the vapor transmission analyses. We also thank Mark Anderson (P&G), Melanie Hansmann (P&G), and Anne Hochwalt (P&G) for helpful comments. We also thank ALG Technical Communications for assistance in the preparation of this manuscript. S. Z. Wang-Weigand is currently affiliated with Takeda Pharmaceuticals North America, Lincolnshire, IL. J. A. Flood is currently affiliated with The Procter & Gamble Company, Microscopy Section, Corporate Analytical Department, Corporate Research & Chemical Technologies Division, Cincinnati, OH 45252 REFERENCES
Page 9the link between excess weight or obesity and obstructive sleep apnea has long been appreciated. In 1956, obstructive sleep apnea was recognized as a disease of obesity and hypoventilation: the Pickwickian syndrome (3). Since then, observations of patients diagnosed with obstructive sleep apnea and findings from population studies have overwhelmingly supported a strong and likely causal role of overweight in this condition, more broadly described as sleep-disordered breathing (SDB). The commonest form of SDB is obstructive sleep apnea, in which there is repetitive collapse (apnea) or partial collapse (hypopnea) of the upper airway during sleep (6). This results in decreases and pauses in breathing during sleep and intermittent transient hypoxia. These events are often terminated in arousals from deeper sleep, and the resulting sleep fragmentation can lead to excessive daytime sleepiness. The gold standard diagnostic investigation of SDB is nocturnal polysomnography to detect apnea and hypopnea events and determine whether they are obstructive or due to abnormal control of breathing. In population screening studies, the severity range is wider than seen in clinic settings, with a higher proportion of cases at the milder end of the spectrum. Usually, no distinction is made between obstructive events and those due to abnormal control of breathing, and the condition is termed SDB. To be consistent, SDB is the term used in this review, with the assumption that it reflects, in most cases, obstructive sleep apnea. The commonly used measure for SDB is the apnea-hypopnea index (AHI; the number of apnea and hypopnea events per hour of sleep). The important association of excess body weight, a modifiable risk factor, with SDB raises many questions relevant to clinical practice and public health. The topic takes on added importance with the alarming rate of weight gain in children as well as adults in industrialized nations. Although there are multiple established factors that predispose one to SDB, ranging from genetic makeup to upper airway abnormalities and to various craniofacial phenotypes, excess weight is the strongest contributing factor (57). It is therefore expected that the prevalence of SDB will increase in parallel with obesity. Recently, interest has expanded to include intermediary mechanisms by which excess weight and SDB may interact. Particularly important are the associations of SDB, obesity, and metabolic hormones. In this review, we first describe the associations of excess weight in the occurrence and progression of SDB and estimate the overall burden of SDB that may be attributed to excess weight. Next, we discuss mechanisms, including hormonal changes, that explore the excess weight-SDB link. QUANTIFYING THE ROLE OF EXCESS WEIGHT IN SDBThe high prevalence of obesity and morbid obesity in patients diagnosed with sleep apnea is well established (43). Similarly, SDB has been reported to occur in 50–77% of obese patients in other clinical settings (51). The type of body fat distribution seen in obese and overweight SDB patients has led to hypotheses that large neck girth and a high waist-to-hip circumference ratio are stronger predictors of SDB, compared with weight-based measures such as body mass index (BMI) (9, 16). However, due to selection biases favoring the referral of the stereotypical “Pickwickian” patient (i.e., obese, sleepy, middle-aged man with a thick neck) for sleep evaluations, studies based on population samples continue to be important in characterizing the excess weight-SDB link. For the past 15 yr, population studies throughout the United States, Europe, Asia, and Australia have consistently shown a high prevalence of medically undiagnosed sleep apnea in adults (55). Notably, studies have generally shown a graded increase in prevalence as BMI, neck girth, or other measures of body habitus increases (Table 1). Although the absolute values of these prevalence estimates vary somewhat due to differences in measurement techniques, definitions, and cut points, the consistency in the dose-response nature of the associations is striking. The obesity-SDB link first seen in clinical populations is clearly present in undiagnosed SDB in the general population as well. Some studies express the excess weight-SDB association with a comparison of mean BMI by AHI categories. As shown in Table 2, BMI is higher in persons with AHI of ≥5 compared with AHI of <5. Of particular importance, findings from the population studies include a wider range of body habitus values, compared with the higher weight range and morbid obesity seen in clinic patients, and demonstrate that SDB is not just a problem of morbid obesity.
With the use of statistical modeling techniques, attempts have been made to determine which body habitus measure is the best or most important predictor of SDB. However, methodological issues, including important gender differences in ranges of anthropometric measures such as neck girth, different degrees of measurement error, and high correlations between these parameters, make results difficult to interpret. At present, a “best” or most predictive body habitus measure has not been identified. Some findings from studies with large sample sizes suggest that BMI, neck girth, and central fat patterning may independently contribute to SDB (56). Most descriptive studies, both clinical and population, on obesity and SDB have been based on cross-sectional data that only allow interperson comparisons to suggest temporal dynamics of the association. For example, cross-sectional data may indicate that a person with a higher BMI, compared with a person of the same age and sex but lower BMI, is more likely to have SDB. But data on changes within individuals over time, from prospective studies, are needed to understand how SDB varies with weight loss or gain to determine how long it takes for weight fluctuations to have an effect and to project the impact of the current trend of increased obesity on sleep apnea prevalence in the future. Some findings regarding the effect of weight change on SDB are available from clinical studies of extremely obese patients undergoing surgical procedures and from longitudinal epidemiology studies begun several years ago. Despite small sample sizes and often the lack of control groups, clinical studies of weight loss in sleep apnea patients have shown a consistent trend: an ∼3% reduction in AHI is associated with each 1% reduction in weight (55). With the current increase in bariatric surgery (even among adolescents), data on how weight loss affects SDB are becoming more abundant (22, 50). In the most recent study (12), polysomnography was performed before gastric banding and at 1–4 yr after this surgery in 25 severely obese patients. The mean weight loss was 45 kg, and the mean AHI fell from 62 to 13. A few population studies have investigated the role of weight change in SDB occurrence and progression over time. Findings consistently point to the importance of body habitus in predicting the occurrence and progression of SDB. In the Wisconsin sleep cohort, 690 men and women were studied with laboratory polysomnography at baseline and at 4-yr follow-up (29); a 10% weight gain was associated with a sixfold increase in the odds of developing moderate or worse SDB (AHI ≥ 15). Similar to the findings from clinical studies, each 1% change in weight was associated with a 3% change in AHI. The 5-yr incidence of SDB was investigated in the Cleveland Family Study (47). Of 286 men and women (mean age = 36.8 yr) who had no SDB (indicated by AHI < 5) at baseline, the incidence of new SDB (defined by developing AHI > 15 at follow-up) was 3.3% for those whose baseline BMI was <24 and 22% for those whose baseline BMI was ≥31. Most recently, longitudinal data from the Sleep Heart Health Study were used to examine 5-yr changes in weight and AHI based on in-home polysomnography of 2,968 men and women ages 40–95 yr (27). Results indicated that, although weight loss predicted a decrease in AHI, the effect was weaker than that of weight gain on an increase in AHI. For example, in men, the odds ratio for a 5-yr increase in AHI of ≥15 with a gain of at least 10 kg was 5.2, but the odds ratio for a loss in AHI of at least 15 with a loss of ≥10 kg was 2.9. Understanding special vulnerabilities to the effects of excess weight on SDB is important in targeting population subgroups that may be at high risk. Gender, age, and race or ethnicity have been characteristics of interest in looking for interaction effects. In investigating interactions, samples are stratified on the characteristic of interest, and the measures of association (e.g., odds ratios for BMI and AHI) in each strata are compared to determine whether any differences are statistically significant. Although few in number, some cross-sectional and prospective studies have addressed the possibility of differences in the nature of the obesity-SDB link by gender, age, and race. Before the publication of population-based studies of SDB that included both women and men, gender differences in SDB occurrence and risk factors were poorly understood. SDB was believed to be rare in women, particularly those who had not reached menopause. The few premenopausal women diagnosed with SDB were reported to be strikingly more obese than the men or postmenopausal women diagnosed with this condition (26, 52). These observations suggested that women were less vulnerable to the effects of weight on SDB, particularly before menopause. Population-based studies have now shown that the increased SDB prevalence associated with male sex and menopause are significant but smaller in magnitude compared with those expected from clinical observations. Although the gender differences are less striking than once thought, there is considerable interest in understanding whether body habitus differences explain gender and menopausal differences. A greater male vulnerability to SDB from obesity has been reported in some but not all studies. In a study of progression of SDB in the Cleveland Family Study, the interaction of age, sex, and BMI was significant in predicting changes in AHI (34). In women, the increase in AHI with increased weight was less than that observed in men, regardless of age and baseline BMI. Similar findings were reported by Newman et al. (27) in the recent analysis of 5-yr prospective data of the Sleep Heart Health Study. AHI was more likely to increase in men compared with women for a given weight increase. For a weight gain of ≥10 kg, the odds ratio (95% confidence interval) for a progression in AHI of ≥15 was 5.2 (2.4, 11.5) for men and 2.6 (1.0, 6.6) for women. In two cross-sectional community studies on middle-aged Chinese men and women in Hong Kong, men, compared with women, had higher AHI at any given BMI (17, 19). Although suggestive, a statistically significant interaction of gender and BMI or other obesity marker on SDB (e.g., a higher odds ratio for SDB and BMI in men vs. women) was not found in two other large population studies (4, 5, 29). The lack of agreement among the various study findings may be due in part to the precision of estimates in the BMI-gender strata when other factors, such as age and menopausal status, are accounted for. For example, Bixler and colleagues (4) noted that, in the Pennsylvania cohort sample (5), all of the premenopausal women with SDB (AHI > 15) were obese (BMI ≥ 32), compared with only 42% of postmenopausal women who were not using hormone replacement therapy. In a report from the Wisconsin cohort, at any given BMI, prevalence of SDB (indicated by AHI ≥ 5) was higher in postmenopausal compared with menopausal women, but SDB did occur in both post- and premenopausal women who were not obese (53). Clinical reports of greater obesity of premenopausal women, compared with postmenopausal women, suggest that association between excess weight and SDB is stronger for menopausal women. Thus both population and clinical studies indicate that BMI has a weaker effect on SDB in premenopausal women compared with postmenopausal women as well as with men. Investigation of an interaction of age in adults and obesity on SDB is limited by the small number of studies with a sufficient sample size at the higher range of the age spectrum, i.e., over age 75 yr. Ancoli-Israel and colleagues (2) reported on an 18-yr follow-up of community-dwelling adults age >65 yr at baseline. Interestingly, BMI at baseline and change in BMI over time, but not aging, were predictive of changes in AHI. In the Western Collaborative Group study of 281 men, ages 75–91 yr, AHI was measured with an in-home monitor and correlated with data collected during the previous 30 yr on BMI and waist girth (7). The prevalence of AHI of ≥5 was 26 and 35% in men with BMI of ≤28 and >28, respectively. The authors found that, of the body habitus measures, only midlife waist girth and subsequent increases over 30 yr were independently associated with AHI in older age. The authors commented that neck girth in elderly men may be a weaker correlate of SDB than it is in younger men. Cross-sectional studies with samples that include a wide age range have generally shown that the association of obesity and SDB appears to weaken with increasing age. In the Sleep Heart Health Study, SDB in people age 70 yr or older was only weakly related to BMI or other measures of body habitus (56). Studies of sleep apnea prevalence in Western nations, Eastern nations, and of populations in the United States with a diversity of ethnic groups show BMI or other measure of excess weight to predict SDB (17, 19, 21, 35, 48). Few studies, however, report data that can answer the question of whether there is a special vulnerability of a race or particular ethnic group to the effects of excess weight on sleep apnea. Recent studies of men and women in Hong Kong (17, 19) and Korea (21) and men in India (48) have reported prevalences of sleep apnea similar to those of Western nations and positive association of BMI and SDB, with odds ratios similar to those seen in reports of Western nations. Based on these population studies, the increased odds for AHI ≥ 5 with one standard deviation increment in BMI (∼3–4 BMI units) are 4.0 for Korean men and women (21), 5.7 for Indian men (48), and 2.4 and 3.0 for Chinese men and women in Hong Kong, respectively (17, 19). The magnitude of increased odds from these studies on Asian populations is strikingly similar to the increased odds seen in the United States. The increased odds for AHI ≥ 5 with one standard deviation increment in BMI (∼5 kg) was 4.0 in the Wisconsin sleep cohort (54). Most striking is that the similarities in the excess weight-SDB link exist despite the relatively lower BMI in Eastern compared with Western nations (as shown in Table 2). In addition to lower mean values and ranges of BMI, cut points for what is considered obese is lower (i.e., BMI > 23) in Eastern compared with Western nations (i.e., BMI > 30). Li and colleagues (25) have previously noted that far-east Asian men with sleep apnea are nonobese compared with white patients and hypothesized that the interaction of craniofacial anatomy and obesity may differ in Eastern and Western populations. In support of this interaction, craniofacial parameters were a stronger correlate of SDB in leaner men in a study of largely Caucasians in the United States (11). There have been no studies to date on SDB in African populations. In most of the studies in the United States that include African Americans, prevalence estimates are commonly adjusted for BMI, and thus subgroup vulnerability cannot be ascertained. However, in the Sleep Heart Health Study longitudinal study of weight change and SDB (27), the authors noted that no significant interaction of race and BMI with SDB was found. Similarly, although BMI was a risk factor for SDB in African Americans in the Cleveland Family Study, the effect did not differ in magnitude from the effect in Caucasians in the same study (35). What proportion of cases of SDB might be attributable to excess weight? The answer depends on the presence of a causal association between excess weight and SDB, the magnitude of the association, and the prevalence of excess weight in the population. Here, we describe a “ballpark estimate” of the proportion of SDB disease resulting from overweight and obesity. We use data from the US Centers for Disease Control and Prevention’s Behavioral Risk Factor Surveillance System (8), the US census, and our own data from the Wisconsin sleep cohort for the calculation. METHODSAlthough several research groups have presented prevalence estimates for SDB by age and sex, no published data are available at the resolution necessary (age- and sex-specific estimates within multiple categories of excess weight) to make reasonable attributable prevalence calculations. Thus we performed new analyses of in-laboratory overnight polysomography data from the population-based Wisconsin Sleep Cohort Study (54) to perform estimates of SDB prevalence. We chose to use cross-sectional attributable prevalence estimates rather than more traditional attributable risks because 1) SDB is reversible and appears to change rapidly in response to weight change (loss or gain as discussed in Longitudinal Studies) and 2) attributable prevalence per se (the amount of SDB disease in a population at a point in time attributable to excess weight) is a population health parameter of significant interest. We estimated the prevalence of SDB [“mild or worse SDB” (AHI ≥ 5) and “moderate or worse SDB” (AHI ≥ 15)] attributable to excess body weight in a multistep process. We graphically examined the relation between BMI and SDB prevalence in age- and sex-specific groups. Prevalence of SDB did not rise with increasing BMI among persons with BMI less than ∼25 kg/m2 but did with increasing BMI beyond that point (data not shown). Thus we chose BMI of <25 kg/m2 as the reference group for which to estimate prevalence ratios within age- and sex-specific categories. We examined several variables that may covary with the relation between excess weight and SDB, such as alcohol and cigarette use. These were not found to be important confounders of the association. We calculated prevalences of SDB by categories of age, sex, and BMI (<25, 25–29, 30–39, and ≥40 kg/m2). As with standard attributable risk estimation (39), the prevalence of SDB in the BMI < 25 kg/m2 category was assumed to be a baseline level of SDB prevalence that would be expected in a population for which all persons had “normal weight” (BMI < 25 kg/m2). Excess prevalence in age- and sex-specific categories of increasing BMI, beyond that seen in the baseline category, was calculated and “attributed” to excess weight. Finally, the age-, sex-, and BMI category-specific prevalences were extrapolated to the US distribution of adults aged 30–69 yr old (the age span represented in the Wisconsin sleep cohort). Sex and age distribution data were from 2003 census estimates (49). US population distribution of BMI categories within age and sex groups were calculated using 2003 data available from the US Centers for Disease Control’s Behavioral Risk Factor Surveillance System (8). RESULTSTable 3 presents the prevalence estimates for SDB by sex, age, and BMI category. In each age and sex group, SDB prevalence increases steeply with increasing BMI. Generally, SDB is more prevalent in men and older persons.
Using the prevalence of SDB in persons with a BMI of <25 kg/m2 to calculate excess SDB in the higher BMI categories and weighting over the estimated US population distribution of BMI categories within each age and sex group, Fig. 1 demonstrates age- and sex-specific estimates of US prevalences of SDB and the proportions attributed to overweight and obesity. Generally, a greater proportion of SDB is attributable to excess weight in younger persons relative to older persons. A greater proportion of more severe SDB prevalence (AHI ≥ 15), compared with SDB indicated by AHI of ≥5, which includes the milder part of the spectrum, is attributable to excess weight. Fig. 1.Estimated prevalence of mild or worse sleep-disordered breathing [SDB; apnea-hypopnea index (AHI) ≥ 5 events/h] and moderate or worse SDB (AHI ≥ 15 events/h) and prevalence of SDB attributable to excess weight (body mass index of ≥25 kg/m2) by sex and age. Among adults ages 30–69 yr, averaging over the estimated US 2003 age, sex, and BMI distributions, we estimate that ∼17% of adults have mild or worse SDB (AHI ≥ 5 ) and that 41% (7% of the total population) of those adults have SDB attributable to having a BMI of ≥25 kg/m2. Similarly, we estimate that ∼5.7% of adults have moderate or worse SDB (AHI ≥ 15) and that 58% (3.3% of the total population) of those adults have SDB attributable to excess weight. Note that these are rough estimates that depend on sampling error variance not only from the Wisconsin sleep cohort estimates but also from the census (for an intercensal year, 2003) and Centers for Disease Control and Prevention data. The prevalence of SDB in persons with BMI of <25 kg/m2, although much less than in higher BMI categories, is still of substantial public health importance, with the prevalence of mild or worse SDB ranging from 2.5% in normal weight younger women to 26% in normal weight older men. However, much mild or worse SDB seen in the population (41% by our conservative estimate) might be due to the excess weight in persons with BMI of ≥25 kg/m2 (∼2 out of 3 US adults ages 30–69 yr in 2003, and growing). Furthermore, we estimate that most (58%) of more severe SDB (AHI ≥ 15 events/h) is due to overweight and obesity among US adults. We present these attributable prevalences as rough estimates because the calculations are fraught with difficulties and assumptions, including sampling errors in several different underlying parameters, concerns about making causal interpretations, and extrapolating estimates from one study population to a more heterogeneous US population. For two important reasons, we expect that these attributable proportions are likely to be conservative underestimates. First, we use BMI as an imperfect proxy for whatever underlying body habitus parameters are most important in impacting SDB. This was necessary since population estimates of BMI distribution are available, whereas other population distribution patterns of, perhaps, more salient parameters such as fat deposition patterns in the upper airway are not available. Second, the amount of data available to us to examine SDB prevalence by categories of BMI, age, and sex allowed for only a few BMI categories, which imposed a type of “round-off” error. Both of these issues are likely to have resulted in underestimates of the proportion of SDB attributable to excess weight. Clearly, if the expanding epidemic of obesity seen in the United States continues, the prevalence of SDB will almost certainly increase, along with the proportion of SDB attributable to obesity. Excess body weight has been hypothesized to affect breathing in numerous ways, including alterations in upper airway structure (e.g., altered geometry) or function (e.g., increased collapsibility), reduced chest wall compliance, disturbance of the relationship between respiratory drive and load compensation (43), and exacerbation of obstructive sleep apnea events via obesity-related reductions in functional residual capacity and increased whole body oxygen demand. These putative mechanisms suggest that specific anatomical locations of excess fat deposition may be important (9, 14, 16, 24, 26, 37, 38). As discussed in Longitudinal Studies, longitudinal population and clinical studies have demonstrated a fairly rapid response of SDB, as indicated by AHI level, to actual change in weight. Little is known about how changes in physical activity levels, and resulting characteristic changes in body fat distribution and fitness, affect SDB apart from changes in weight. One study has demonstrated an inverse association of weekly exercise and SDB, independent of body weight (28). It is also possible that the fatigue and sleepiness that are often symptomatic of SDB disinclines afflicted persons to physical activity, further exacerbating weight gain, and thus SDB, in a feedback cycle. The potential for such behavioral feedback mechanisms in SDB is largely unexplored. Hormones, including those that influence sex characteristics and metabolism, may have a diverse role in the development or exacerbation of SDB. Clinical and epidemiological studies have demonstrated an increase in prevalence of SDB in menopausal, compared with premenopausal women (55). It is generally believed that depletion of estrogen and progesterone is responsible for an increased vulnerability to SDB, but data relevant to pathophysiological mechanisms are sparse. Progesterone stimulates breathing, but experimental studies have failed to show a protective effect regarding upper airway collapse. Hormonal changes in the menopausal transition may indirectly contribute to SDB by increasing body fat, specifically abdominal or central fat deposition, which are known risk factors for SDB. Although menopause is indeed associated with increased central body fat, it is unlikely that this mechanism can account for all of the increased SDB risk. In the Wisconsin sleep cohort, menopausal compared with premenopausal women had a higher prevalence of SDB regardless of BMI or indicators of central fat distribution (53). However, a negative interaction of menopause and BMI was demonstrated in the Pennsylvannia cohort study, in which the effect of BMI was lower in premenopausal women (i.e., SDB only occurred in obese premenopausal women) (4), indicating a greater susceptibility related to increased BMI with menopause. Data in support of a direct role of female hormones in SDB comes from the Sleep Heart Health Study, showing a decrease in SDB prevalence in postmenopausal women who do take hormone replacement therapy compared with those who do not (40). However, in most clinical trials, hormone replacement therapy has not resulted in a significant drop in AHI (32). Recently, there has been great interest in a putative interaction between SDB, insulin resistance, and metabolic hormones, thus suggesting that SDB is an important new facet of the metabolic syndrome. Particular attention has been paid to hormones released from fat cells (adipocytes), the adipocytokines. Because metabolic hormones are closely associated with body weight, investigators, using different approaches, have attempted to examine whether SDB and related hypoxemia are independently associated with the levels of these hormones. This is important because treatment of SDB may help reverse metabolic abnormalities. Other studies have suggested that changes in metabolic hormones that occur with obesity may aggravate/alleviate breathing in SDB. One adipocytokine recently studied in association with SDB is leptin (30). Leptin is a 16-kDa hormone that is released by adipocytes to signal fat stores to the hypothalamus (23). Leptin therefore tends to reduce appetite. Leptin levels are higher in obese individuals, suggesting that leptin resistance exists in obesity. A mechanism for the leptin resistance in obesity may be transfer of leptin across the blood-brain barrier. It has been reported that SDB is associated with higher leptin than would be expected based on BMI alone (30). Leptin has also been associated with obesity hypoventilation and responses to hypercapnea (31). It has been proposed that SDB patients have greater leptin resistance than commonly seen in obesity and that this further impairs their breathing. Increased leptin and insulin resistance, as seen with SDB, may in turn contribute to perpetuating obesity. SDB is associated with sleep fragmentation and therefore short sleep duration at night. It is of interest that laboratory sleep restriction in young volunteers (42) and short sleep duration in our population study (45) are paradoxically associated with low leptin levels. SDB is closely associated with visceral obesity, which is key to insulin resistance, Type 2 diabetes, and the metabolic syndrome. Studies investigating the association between SDB and insulin resistance have used multiple approaches. Population studies have been mainly cross sectional. Studies include associations between symptoms (snoring and witnessed apneas) (1, 13, 15, 20, 36) or objective measures of SDB (AHI, oxygen desaturation) (10, 18, 24, 33, 45, 46) and insulin resistance. Both population and case-control studies have been carried out. Measures of insulin resistance have mainly included self-reported diabetes, fasting glucose and insulin, glucose levels in an oral glucose tolerance test, and surrogates of insulin sensitivity such as the homeostatic model assessment. Various mechanisms proposed for the relationship between SDB and insulin resistance include alterations in adipocytokines, decreased sleep duration due to sleep fragmentation (resulting in alterations in insulin secretion and sensitivity, decreased growth hormone, and increased cortisol) (41), increased sympathetic nervous system activity, and direct effect of intermittent hypoxia on the glucose homeostatic system. Considerable disagreement exists between studies, which have reported either no association between SDB and insulin sensitivity or an association ranging from minor to highly significant. A large number of studies have not sufficiently controlled for possible confounding factors (especially body weight) or have been sufficiently powered to answer questions related to the association between SDB and insulin sensitivity. Recently, our laboratory investigated the above-discussed relationships in a large population-based sample from the Wisconsin Sleep Cohort Study (45). We examined possible associations between habitual sleep and polysomnographic sleep duration and serum leptin, insulin, glucose, QUICKI, homeostatic model assessment, and adiponectin. Adiponectin is an adipocyte-derived hormone associated with insulin sensitivity. All analyses were corrected for age, sex, and BMI, and if any associations were noted, further correction was carried out for identified potential counfounding factors, including AHI. We found a significant association between sleep duration and serum leptin. Short sleep duration was associated with low leptin levels, suggesting that it may predispose to increased appetite and potentially obesity. However, we did not find any association between sleep duration and measures of insulin sensitivity. We also examined associations between AHI and the above measures, independent of BMI. We found no significant association between AHI and any of the measures. We also tested the possibility of U-shaped relationships between the metabolic measures and AHI but found no significant associations. It is of interest that, in our study, SDB did not alter the relationship observed between short sleep duration and leptin. From our data, as expected, obesity has a strong and clear association with insulin resistance, the metabolic syndrome, and metabolic hormone levels (leptin and insulin). Any observed association between SDB and metabolic factors are therefore likely to be primarily driven by the association with BMI and, in particular, visceral obesity. FUTURE DIRECTIONSExcess weight clearly contributes to the incidence and progression of SDB. Based on current data, a considerable proportion of the future SDB burden would be eliminated by the prevention and reduction of overweight and obesity. To develop feasible strategies to address the current and growing burden, a better understanding is needed of what characteristics of body habitus are indeed significantly modifiable (i.e., beyond their genetic basis) besides weight. More information is needed on the natural history of physical fitness, as well as overweight and obesity with respect to SDB. For example, does childhood, adolescent, or lifelong obesity play a greater role in SDB pathogenesis than contemporary weight gain? Intervention studies are needed to determine the preventive role that exercise and improved physical fitness may play in coping with a possible epidemic of SDB in parallel with that of obesity. Although a causal role of excess weight in SDB is evident, the complex interactions of altered sleep, metabolic hormones, SDB, and obesity and their compound effect on morbidity are other important areas for future research. REFERENCES
Page 10obstructive sleep apnea (OSA), a disorder characterized by repetitive collapse of the upper airway during sleep, is one of the most common respiratory disorders, with an estimated prevalence of 24% and 9% among middle-aged American men and women, respectively (92). Obesity in modern Western society is also extremely common. As of 2002, nearly one-third of US adults met clinical criteria for obesity, and this prevalence appears to only be increasing (27). That obesity and OSA often colocalize is of no surprise. Other than perhaps male gender, obesity is the strongest risk factor for the development of OSA. Relative to those with a stable weight, a 10% increase in weight over 4 yr is associated with a sixfold increase in the risk of developing moderate to severe OSA (58). Weight loss trials have found significant reductions in apnea severity with moderate weight loss (69, 74). However, the mechanisms by which obesity causes OSA are not completely defined, with many potential pathways hypothesized (93). Perhaps the most obvious is that fat deposition in the neck and airway lumen may lead to increased collapsibility of the upper airway. In addition, adiposity in the chest and abdomen may result in reductions in lung volumes. Recent studies suggest reduced lung volume may independently predispose to upper airway collapse (37). Another possible mechanism relating obesity with sleep apnea is via the hormonal effects of adipose tissue. Leptin, a hormone produced by adipose tissue, has important effects on weight regulation by stimulating hypothalamic satiety centers (6, 34, 57). Human obesity is typically associated with elevated leptin levels, suggesting a state of leptin resistance (15). Besides regulating weight, leptin may have important effects on ventilatory drive. Leptin-deficient mice hypoventilate and have a blunted response to hypercapnia (53, 84). Administration of leptin corrects these abnormalities independent of changes in weight (53). Through these effects on ventilatory control, elevated leptin levels (or the underlying leptin-resistant state) may play a pathogenic role in the development of OSA. It is also possible that OSA may play a causal role in the development of obesity. OSA patients have elevated leptin levels compared with weight-matched controls, and OSA treatment lowers these levels (39, 53, 55). These findings suggest OSA may be a cause of leptin resistance and thus a propensity for further weight gain. Not only does sleep disruption, a common result of OSA, reduce serum leptin levels, it also increases levels of the appetite-stimulating hormone ghrelin (76, 82). These findings were associated with increases in hunger and appetite scores (76). Epidemiological studies consistently implicate reduced sleep as a risk factor for the development of obesity (36, 47, 70, 82, 87, 89). Although overall weight does not appear to fall with treatment of OSA, one study did find the amount of fat inside the abdominal cavity (visceral fat) diminishes (10). This fat compartment appears to have unique hormonal effects as the amount of visceral adipose tissue is more strongly associated with metabolic complications of obesity such as insulin resistance and dyslipidemia than measures of subcutaneous fat or overall obesity (28, 29). FAMILIAL AGGREGATION OF OBESITY AND SLEEP APNEAThat obesity is in large part genetically determined has been known for almost three decades (26). A large twin study estimated the heritability of weight to be 78% (79). That is to say, 78% of the variability in weight across a population is explained by shared intrafamilial factors. Subsequent studies demonstrated that adopted children have a body size more closely resembling their biological parents than their adopted parents (80). Studies in Finland and the United Kingdom have estimated the heritability of body mass index (BMI) to be 60–80% (42, 61). Linkage to obesity-related phenotypes has been investigated in some 50 genomewide scans to date with dozens of candidate loci and genes identified (3, 59). Similarly, familial factors have been known to influence OSA risk for nearly 25 years (77). Since then, several groups have quantified the familial risk of OSA by studying widely different populations. All of these studies have identified a strong heritable component to OSA (30, 32, 49, 62, 66). Family-based studies suggest that the risk of OSA is approximately twice as great among relatives of apneic persons (30, 66). In addition, a dose-response relationship exists such that the risk of OSA increases with increasing number of apneic relatives (66). Quantitative apnea-related phenotypes also demonstrate substantial heritability. A study of elderly twins found the heritability of both the respiratory disturbance index and the oxygen desaturation index to be nearly 40% (7). SHARED SUSCEPTIBILITY GENESGiven the strong genetic components to both obesity and apnea phenotypes as well as the tight association with multiple interweaving links between these two diseases, it would not be surprising for there to exist common susceptibility genes for both obesity and OSA. In fact, it has been suggested that the familial aggregation of OSA may simply be a reflection of that found in obesity. This is clearly not the case. Even after controlling for BMI, significant familial aggregation for OSA persists (32, 66). Furthermore, a study of nonobese apneic patients also demonstrated strong heritability of OSA (49). That this should be the case should be of no surprise given that other pathophysiological pathways to apnea development such as craniofacial structure and ventilatory control have also been demonstrated to have a heritable component (13, 32, 49, 63, 65). Thus the susceptibility genes for OSA are not exclusively the same genes as those modulating obesity. That does not mean, however, that no overlap exists. Clearly, given the strong effect of obesity on OSA pathogenesis, any genetic variant that predisposes to obesity will secondarily also lead to the development of OSA. Thus any obesity gene might also be considered to be an apnea gene. However, obesity is not a monolithic process. Clearly, there are some forms of obesity that have a more important role in the pathogenesis of sleep apnea. For example, fat deposition in the neck is much more important than fat deposition in the limbs (17). An android fat deposition pattern (excess subcutaneous truncal-abdominal fat) and visceral adiposity have also been found to more closely predict OSA than overall obesity (31, 72). Therefore genetic polymorphisms that influence fat deposition in these sites will be more important OSA genes. There is a wealth of evidence that these fat distribution pattern phenotypes are genetically driven. Large differences in the patterns of fat deposition exist across inbred strains of cattle, suggesting an important role for genetics (83). Among humans, even after correction for overall obesity, measures of fat deposition pattern aggregate within families. The ratio of subscapular skinfold thickness to subscapular plus suprailiac thicknesses, a measure of the android fat pattern, has been reported to have a heritability of 43% (71). Measures of central obesity such as BMI-adjusted waist circumference and ratio of trunk to extremity skinfold ratio have heritabilities of 29–48% (40). These findings strongly support the hypothesis that there are genetic polymorphisms that specifically promote fat deposition in the subcutaneous regions of the torso or around the abdominal viscera. Through this mechanism, these variants would promote the development of both obesity and OSA. Another type of potential genetic interaction between obesity and OSA is one in which a particular polymorphism leads to both obesity and OSA through independent mechanisms (Fig. 1). For example, by impacting leptin function, a genetic polymorphism may reduce satiety and also increase ventilatory instability. This is what is referred to as genetic pleiotropy. Pleiotropic genetic effects have clearly been described in other situations. A well known example is at the APOE locus, which codes for apolipoprotein E. The ε4 allele of APOE is an important risk factor for both atherosclerotic heart disease as well as Alzheimer’s dementia (24, 91). The circadian regulatory gene, CLOCK, also appears to influence multiple biological systems. CLOCK-knockout mice have profound disruptions in circadian rhythmicity, and so it is not surprising that these animals would have abnormal timings for feeding and activity (88). In addition, however, these mutant animals demonstrate an overall increase in caloric intake associated with a phenotype of obesity and metabolic syndrome (85). Similarly, given the large number of neurological, metabolic, and mechanical overlaps between obesity and OSA, it is likely that a polymorphism affecting one biochemical system may affect the risk for both disorders via multiple paths. For example, disruptions of the orexin system could potentially result in a common link between obesity and OSA. Orexinergic neurons in the lateral hypothalamus play an important role in sustaining wakefulness with projections to all of the wake-promoting areas of the brain (21). Loss of these neurons is associated with a narcolepsy phenotype (60). These neurons, as the name orexin implies, also play an important role in stimulating appetite, projecting on to the arcuate nucleus of the hypothalamus (68). Thus mutations affecting orexin, the orexin receptor, or proteins involved in the downstream signaling of orexin binding might simultaneously affect metabolic and sleep-related function. Several other potential candidate genes are listed in Table 1. Fig. 1.Both obesity and sleep apnea are heavily influenced by underlying genotype. Some susceptibility genes act directly on one phenotype and through the causal relationships between obesity and sleep apnea have indirect effects on the other. Other loci have pleiotropic effects, impacting susceptibility to both obesity and sleep apnea via independent mechanisms.
The recently reported linkage scans from the Cleveland Family Study represent the first genomewide linkage studies of OSA phenotypes (54, 55). The results provide insight into possible genetic overlaps between obesity and OSA. Linkage to the apnea-hypopnea index (AHI) as a measure of OSA and BMI as a measure of obesity was tested across the autosomal chromosomes in both a Caucasian and an African-American cohort. The heritability of AHI in both groups was ∼33%, whereas the heritability of BMI was over 50%. After controlling for BMI, significant heritability for AHI remained, supporting the notion that the genetic susceptibility to OSA is not completely defined by weight. In further multivariate modeling of a larger subset of the Cleveland Family Study, obesity measures such as BMI and serum leptin explained 50–55% of the genetic variance in AHI (56). This suggests that about half of the genetic determinants of AHI are obesity related and half are obesity independent. Linkage findings are described by the logarithmic odds (LOD) score, a measure of the odds ratio of linkage to no linkage. Although none of the linkage findings in either racial group achieved genomewide significance (LOD > 3.3) (46), there were several regions with intermediate LOD scores, indicative of possible linkage in the setting of complex, multifactorial diseases such as obesity and OSA (90). Furthermore, the change in linkage evidence for AHI after adjustment for BMI provides insight into mechanisms of action if a susceptibility locus is present. Among the Caucasians, a LOD score of 1.4 was found for AHI and 1.7 for BMI on chromosome 12p (54). After adjustment for BMI, the maximal LOD for AHI in this region dropped to only 0.4 whereas the LOD for BMI adjusted for AHI fell to 0.2. These data suggest that if a susceptibility gene for AHI exists in this region, it likely mediates its effect on apnea via obesity. On the other hand, a maximal LOD for AHI of 1.4 was found on 19q with no linkage evidence for BMI in this region (54). Adjustment for BMI had no effect on the AHI LOD, suggesting that a susceptibility gene in this region exerts its effect on AHI through obesity-independent mechanisms. A different pattern of linkage findings was found at the short arm of chromosome 2. Among the Caucasians, the maximal LOD for AHI is 1.6 and for BMI is 3.1 in this region, again suggesting that a susceptibility locus for both phenotypes might exist in this region (54). However, adjustment for BMI only dropped the LOD for AHI to 1.3, suggesting that the obesity effect at this locus does not fully explain the apnea effect. This may represent two independent loci, one influencing AHI and one BMI. Another possibility is that there is only one locus at 2p regulating both AHI and BMI but via independent mechanisms. A strong candidate gene for both phenotypes in this region is proopiomelanocortin (POMC). The POMC locus, found at 2p23.3, encodes for a number of hormones including melanocyte-stimulating hormone (MSH). POMC neurons in the arcuate nucleus of the hypothalamus are important in energy homeostasis via MSH. Leptin’s anorexic activity is mediated via depolarization of these neurons, and use of an MSH agonist decreases both body fat and leptin levels in humans (16, 25). Obesity phenotypes have been consistently linked to the POMC locus (14, 33, 38, 67), haplotypes in this gene have been associated with leptin levels (38, 50), and severe mutations in this gene produce a severe childhood obesity phenotype (44). Leptin’s effects on ventilatory drive in mouse models are also mediated via MSH, suggesting that this pathway may independently predispose to apnea as well (64). Evidence for pleiotropy also exists at chromosome 8q. Among African-Americans in the Cleveland Family Study, the peak LOD scores were 1.3 and 1.6 for AHI and BMI, respectively (55). Again, after controlling for BMI, the AHI LOD only dropped to 1.1, suggesting that the obesity and apnea promoting effects in this region were independent. A potential candidate locus in this region is the COH1 gene at 8q22–23. This gene encodes a transmembrane protein with a presumed role in vesicle-mediated sorting and intracellular protein transport on the basis of its structure (43). Severe mutations in this gene have been associated with Cohen syndrome, an autosomal recessive condition characterized by truncal obesity (9). Other findings include facial dysmorphism, mental retardation, and ocular anomalies. The craniofacial abnormalities include microcephaly, facial hypotonia, and laryngomalacia, all of which could predispose to OSA. Thus mutations in this gene that are milder but more common may play a role, via separate pathways, in contributing to nonsyndromic forms of both obesity and OSA. The serotoninergic system is another common pathway that could link obesity and OSA. Serotonin has important effects on the stimulation of satiety centers in the arcuate nucleus (5). In addition, serotonin potentiates hypoglossal neural output, which increases upper airway dilator muscle activity (45, 75). Thus reductions in serotoninergic activity, by increasing appetite, could promote obesity and, by lowering muscle tone in the upper airway, could promote OSA. Both linkage and haplotype association studies in a Finnish population suggest a serotonin-related gene at Xq24 is associated with obesity (51, 81). The SLC6A14 gene encodes for a sodium chloride-dependent transporter of neutral and cationic amino acids, which appears to play an important role in the transport of tryptophan, the precursor of serotonin, into the central nervous system (73). A French study has confirmed the association of this polymorphism with obesity, and an association with hunger and satiety scores was also found (19). Whether this polymorphism is associated with OSA (via or independent of any obesity effects) has not yet been studied, as the OSA linkage scans reported thus far have not included the sex chromosomes. GENE × ENVIRONMENT INTERACTIONAnother potential method of genetic interaction between obesity and OSA is a form of gene-by-environment effect where the adverse effects of obesity or OSA can be thought of as environmental stressors (Fig. 2). Each of the methods by which obesity predisposes to OSA can be influenced by an individual’s underlying genetic susceptibility. An increase in fat deposition around the upper airway will be more likely to produce apnea in individuals with a lower ability to respond to this stressor due to reduced upper airway dilator muscle tone. Conversely, obesogenic effects of OSA may be influenced by the underlying genetic milieu. Genetic polymorphisms may modulate the effect that exposure to sleep fragmentation from OSA has on leptin and ghrelin dynamics. Other polymorphisms might influence the effect that these hormonal perturbations have on producing further weight gain. Fig. 2.Sleep apnea susceptibility genes may interact with obesity through numerous mechanisms to influence sleep apnea predisposition. Genetic polymorphisms may modulate the degree to which obesity alters ventilatory drive, reduces lung volume, or narrows the upper airway. Other polymorphisms may affect the degree to which these stresses result in the development of sleep apnea. Similarly, obesity susceptibility genes may interact with sleep apnea in its potential effect on obesity. The peroxisome proliferator-activated receptor-γ (PPARG) gene located on chromosome 3p25 encodes a protein that is a key component of a nuclear transcription factor important in adipocyte differentiation (4). Variants in this gene have been implicated as risk factors for the development of obesity (4, 18, 20, 48, 86). Because hypoxia is known to suppress PPARG gene transcription (94, 95), the importance of a mild defect in PPARG function may become magnified in the setting of recurrent exposure to hypoxia from OSA. Thus PPARG allelic variants may play a much more important role in determining obesity phenotypes in individuals with OSA, and OSA may play a much more important role in promoting weight gain among those with a mutation in PPARG. Similarly, the effects of uncoupling protein-1 (UCP1) and UCP2, two other obesity candidate genes, may be influenced by OSA. These genes encode for uncoupling proteins, mitochondrial proton channels that divert energy from ATP synthesis to thermogenesis (2). In so doing, their activation increases energy consumption. Polymorphisms in both genes have been associated with obesity phenotypes (8, 12, 22–23, 52). Interestingly, sleep deprivation in rodent models has been shown to increase expression of both UCP1 and UCP2 (11, 41), suggesting the sleep disruption of OSA may influence expression of these genes and thus the relative importance of a variant at these loci in determining obesity risk. FUTURE DIRECTIONSThere appear to be multiple causal pathways linking obesity with sleep apnea. Although obesity is clearly a strong risk factor for OSA, further research is needed to better define any potential risk OSA carries for promoting weight gain. Both diseases also clearly have a strong genetic basis. Although work has begun to identify the specific genetic polymorphisms that confer risk to the development of these disorders, clearly much more needs to be accomplished in this arena. A recent workshop sponsored by the American Thoracic Society laid out recommendations for future research aimed at dissecting the genetic bases for sleep-disordered breathing (78). A chief objective was the development of novel phenotypes, including biomarkers, that more closely reflect only one or a few molecular pathways rather than the overarching syndrome defined with the AHI. These simpler phenotypes could be more amenable to genetic analysis because of the fewer sources of variance than a global measure of apnea. The same holds true for obesity research, where the use of more specific phenotypes of fat deposition patterns would provide not only better insight into the molecular mechanisms underlying obesity but also a better understanding of how obesity and OSA interrelate. The use of dense single nucleotide polymorphism maps will allow for better resolution of linkage findings so as to narrow down candidate loci. A better understanding of the neurobiology and molecular pathways underlying obesity and sleep apnea will also allow for a more rational selection of candidate genes to test for association with these disorders. Ultimately, the use of gene-knockout animal models will be important to establish causality of any identified associations. In performing these genetic studies, it will be important to simultaneously consider both disorders given their close relationship. Methodology for conducting bivariate linkage scans has already been developed, and there is evidence to suggest that such a strategy is more powerful than traditional univariate approaches (1). Not only will many genetic variants influence the development of OSA by causing obesity or vice versa but it is likely that there are also variants with pleiotropic effects predisposing to obesity and OSA via independent mechanisms. Identifying such genes and understanding their function will provide novel insights into the shared pathogenesis of these diseases. Finally, the possibility that genetic polymorphisms may affect the susceptibility that each disease confers toward the other should not be ignored. Incorporating such a model of gene × environment interaction into future study designs along with an understanding of the effects of obesity on OSA can allow for a better delineation of OSA susceptibility genes and vice versa. Similarly, use of such a model along with information about the genetics of obesity would allow for a more complete understanding of how OSA may impact obesity. GRANTSResearch support was provided by an American Heart Association Scientist Development Grant and by National Heart, Lung, and Blood Institute Grant HL-60292. REFERENCES
Page 11the respiratory response of newborn mammals [e.g., mice (21), rabbits (2, 5, 10, 22), rats (8, 14), and sheep (42)] to unrelenting hypoxia typically passes through four stages: hyperpnea, primary apnea, gasping, and secondary apnea. The onset of gasping following primary apnea occurs when the arterial Po2 decreases to ∼8–10 Torr; this is true during hypercapnic hypoxia produced by airway obstruction or during hypocapnic hypoxia produced by inhalation of a hypoxic gas mixture (16, 22). Peiper (25), Stevens (38), and Thach (40) have emphasized the importance of hypoxic gasping in “self-resuscitation” (1) or “autoresuscitation” (16) in human infants and that repeated exposure to hypoxia may lead to autoresuscitation failure and death. Recent reports by Poets et al. (26) and Sridhar et al. (35), where home memory monitor recordings of sudden infant deaths have been analyzed, have documented failure of hypoxic gasping to effect autoresuscitation and prevent death in a number of apneic infants. Considering the importance of gasping as the last operative mechanism used by mammals to ensure survival during exposure to severe hypoxia, it is important to be knowledgeable of factors that influence the onset, duration, and number of potential autoresuscitation producing gasps as a first step in understanding the integrated physiology of successful autoresuscitation as well as the pathophysiology of failed autoresuscitation from hypoxic-induced apnea. Our laboratory has recently reported that naive 5- to 6-day-old rat pups, studied at thermoneutrality, display a triphasic pattern of gasping on exposure to unrelenting hypoxia (8). In these animals, hypoxic-induced primary apnea was followed by a period of rapid gasping that lasted 1–2 min; this period of rapid gasping was followed by a period of slower gasping of 1–2 gasps/min that lasted 6–8 min; finally, there was a period of rapid gasping that eventually waned and gave way to secondary apnea and death. In our experience, gasping occurs within 60–90 s of the onset of hypoxia, and naive 5- to 6-day-old rat pups may gasp up to 15 min and exhibit as many as 86 potential autoresuscitation producing gasps. Our laboratory and others have shown that one or more of the previously mentioned gasping characteristics are modulated in this age range of rat pups by factors such core temperature (28), glucose (44), catecholamines (45), nitric oxide (13), and glutamate (12). Given that all of these factors may be altered in one way or the other by exposure to hypoxia, our present experiments have been carried out to determine whether prior exposure to hypoxic-induced apnea, such as may occur during prolonged obstructive apnea or positional asphyxia, influences gasping on exposure to unrelenting hypoxia. Specifically, our experiments were designed to test the hypothesis that prior exposure to 1, 2, 3, 4, 9, and 14 hypoxic-induced apnea/autoresuscitation cycles (HIA/AR) modulates the onset, duration, and number of potential autoresuscitation producing gasps on exposure to unrelenting hypoxia in an exposure-dependent fashion. Furthermore, we have done experiments to determine whether the aforementioned gasping characteristics upon exposure to unrelenting hypoxia return to normal within a 120-min normoxic recovery period after exposure to 9 HIA/AR. METHODSNinety five, 5- to 6-day-old Sprague-Dawley rat pups were studied. Each pup, born by spontaneous vaginal delivery, was housed with its mother and siblings (22 ± 1°C, 20–30% relative humidity in a 12:12-h light-dark cycle) until an experiment. Although 22°C is below the thermoneutral zone of newborn rats (23), each pup had the opportunity to huddle with its siblings and dam in the nest and thus to thermoregulate behaviorally. All experimental procedures described herein were carried out in accordance with the Guide to the Care and Use of Experimental Animals provided by the Canadian Council on Animal Care and with the approval of the Animal Care Committee of the University of Calgary. For an experiment, each 5- to 6-day-old pup was removed from its mother and siblings, weighed, and instrumented for measurement of cardiovascular and respiratory variables. Afterward, the pup was positioned prone in a metabolic chamber regulated to 37.0 ± 0.1°C into which flowed room air at a rate of 1 l/min. The time to first and last gasp as well as the total number of gasps to unrelenting hypoxia (i.e., 97% N2-3% CO2) was determined 5 min after each pup had experienced 0 (n = 13), 1 (n = 7), 2 (n = 7), 3 (n = 6), 4 (n = 9), 9 (n = 9), or 14 (n = 9) HIA/AR at 5-min intervals. For each HIA/AR, the gas that flowed into the chamber was changed from room air to 97% N2 and 3% CO2 until primary apnea occurred; the gas was then changed back to room air, and autoresuscitation was effected by gasping. When the gas mixture was changed, the flow rate was increased until the gas concentrations in the chamber had stabilized; the flow rate was then lowered to 1 l/min. Our laboratory has previously shown that naive 5- to 6-day-old pups studied at a thermoneutral temperature of 37.0°C tolerate an average of 15 episodes of HIA/AR before autoresuscitation failure (8). During an experiment, stages of the respiratory response to hypoxia were directly observed on the polygraph tracing. For an experiment, each 5- to 6-day-old pup was removed from its mother and siblings, weighed, and instrumented for measurement of cardiovascular and respiratory variables. Afterward, the pup was positioned prone in a metabolic chamber regulated to 37.0 ± 0.1°C into which flowed room air at a rate of 1 l/min. The time to first and last gasp as well as the total number of gasps to unrelenting hypoxia was then determined after each pup had experienced a normoxic recovery period of 5 min (n = 7), 15 min (n = 7), 30 min (n = 7), 60 min (n = 7), or 120 min (n = 7) after nine HIA/AR at 5-min intervals. To determine whether gasping had indeed returned to “normal,” comparisons were also made with data obtained in pups (n = 13) from experimental series I that had not experienced hypoxic-induced apnea before being exposed to unrelenting hypoxia. The metabolic chamber used in our experiments consisted of a double-walled Plexiglas cylinder (30 cm long, internal diameter 6 cm) into which flowed room air or 97% N2-3% CO2. Chamber ambient temperature was regulated to 37.0 ± 0.1°C by circulating water from a temperature controlled bath (Endocal Refrigerated Circulating Bath RTE-8DD, Neslab) through the space between the walls. Our laboratory has previously shown that 37.0 ± 0.1°C is the preferred ambient temperature of naive 5- to 6-day-old rat pups 20–30 min after they are placed in a thermocline with a linear temperature gradient of 25°C to 40°C (8). During an experiment, the electrocardiogram, respiratory movements and chamber CO2 levels were recorded on a model 7 polygraph (Grass Instrument) at a paper speed of 10 mm/s. A bipolar lead II electrocardiogram was recorded from multistranded stainless steel wire electrodes (AS 633, Cooner Wire) sewn on the right shoulder (− electrode) and the left thigh (+ electrode) as described by Osborne (24); the electrodes were connected to a model 7HIP5 high-impedance probe coupled to a model 7P5 wide-band EEG alternating-current preamplifier (Grass Instrument). Respiratory movements were recorded from a mercury-in-silicone rubber strain gauge (model HgPC, DM Davis) placed around the chest; the strain gauge was connected to a bridge amplifier (Biomedical Technical Support Center, University of Calgary, Calgary, Alberta, Canada) that was coupled to a model 7P03 adapter panel (Grass Instrument). Statistical analysis was carried out by ANOVA and Newman-Keuls multiple-comparison tests. All results are reported as means ± SD, and P < 0.06 was considered to be of statistical significance. RESULTSPrior exposure to HIA/AR at 5-min intervals did not significantly alter the time to first gasp (ANOVA P = 0.344) (Fig. 1), but it decreased the time to last gasp (ANOVA P < 0.001) (Fig. 2) after two HIA/AR and the total number of gasps (ANOVA P < 0.001) (Fig. 3) after three HIA/AR on exposure to unrelenting hypoxia. Exposure of naive pups to unrelenting hypoxia resulted in a reproducible respiratory response as previously reported (8): initially there was a period of hyperpnea and arousal that preceded primary apnea; primary apnea was followed by a period of rapid gasping (phase I of gasping) that was followed by a period of slower gasping (phase II of gasping) of 1–2 gasps/min; finally, there was a period of rapid gasping (phase III of gasping) that eventually waned and gave way to secondary apnea and death. The three phases of gasping became less identifiable as the number of prior HIA/AR were increased before exposure to unrelenting hypoxia (Fig. 4). Fig. 1.Lack of influence of prior exposure to 0, 1, 2, 3, 4, 9, or 14 hypoxic-induced apnea/autoresuscitation cycles on the time to first gasp on exposure to unrelenting hypoxia in 5- to 6-day-old rat pups. Values are means ± SD. P = 0.344 by ANOVA. Fig. 2.Influence of prior exposure to 0, 1, 2, 3, 4, 9, or 14 hypoxic-induced apnea/autoresuscitation cycles on the time to last gasp upon exposure to unrelenting hypoxia in 5 to 6 day-old rat pups. Values are means ± SD. P ≤ 0.001 by ANOVA; *P ≤ 0.06 vs. 0 prior hypoxic-induced apneas by Newman-Keuls. Fig. 3.Influence of prior exposure to 0, 1, 2, 3, 4, 9, or 14 hypoxic-induced apnea/autoresuscitation cycles on the total number of gasps on exposure to unrelenting hypoxia in 5- to 6-day-old rat pups. Values are means ± SD. P ≤ 0.001 by ANOVA. *P ≤ 0.06 vs. 0 prior hypoxic-induced apneas by Newman-Keuls. Fig. 4.Influence of prior exposure to 0 (A), 4 (B), 9 (C), or 14 (D) hypoxic-induced apnea/autoresuscitation cycles on the gasping pattern in 5- to 6-day-old rat pups on exposure to unrelenting hypoxia. Normoxic recovery time after nine HIA/AR significantly influenced the time to last gasp and the total number of gasps (Figs. 5 and 6) on exposure to unrelenting hypoxia but only at 120 min. After a normoxic recovery period of 120 min, the total number of gasps was similar but the time to last gasp was still decreased compared with that observed in naive animals exposed to unrelenting hypoxia. Fig. 5.Influence of normoxic recovery time on the time to last gasp on exposure to unrelenting hypoxia in 5- to 6-day-old rat pups after 9 hypoxic-induced apnea/autoresuscitation cycles. Values are means ± SD. NA, not applicable. P ≤ 0.001 by ANOVA. *P ≤ 0.06 vs. 0/NA from experimental series I by Newman-Keuls. †P ≤ 0.06 vs. 9/5 by Newman-Keuls. Fig. 6.Influence of normoxic recovery time on the total number of gasps on exposure to unrelenting hypoxia in 5- to 6-day-old rat pups after 9 hypoxic-induced apnea/autoresuscitation cycles. Values are means ± SD. P ≤ 0.001 by ANOVA. *P ≤ 0.06 vs. 0/NA from experimental series I by Newman-Keuls. †P ≤ 0.06 vs. 9/5 by Newman-Keuls. DISCUSSIONOur experiments provide new information about factors that influence the newborn’s respiratory response to hypoxia. Novel findings of our study were that although prior exposure to hypoxic-induced apnea did not alter the time to first gasp, it significantly decreased the time to last gasp as well as the total number of gasps on exposure to unrelenting hypoxia in an exposure-dependent fashion. When the normoxic recovery time after nine HIA/AR was varied from 5 to 120 min, the time to last gasp as well as the total number of gasps on exposure to unrelenting hypoxia recovered partially but only at 120 min (i.e., the total number of gasps was similar but the time to last gasp was still decreased compared with that observed in naive animals exposed to hypoxia). Thus prior exposure to hypoxic-induced apnea as may occur during prolonged obstructive sleep apnea or positional asphyxia decreases the number and duration of potential autoresuscitation producing gasps on exposure to unrelenting hypoxia for a period of up to and exceeding 120 min, respectively. Newborn mammals display a characteristic respiratory response consisting of hyperpnea, primary apnea, gasping, and secondary apnea on exposure to unrelenting hypoxia [e.g., mice (21), rabbits (2, 5, 10, 22), rats (8, 14), and sheep (42)]. The onset of gasping after primary apnea occurs when the arterial Po2 decreases to ∼8–10 Torr whether produced by airway obstruction resulting in hypercapnic hypoxia or inhalation of a hypoxic gas mixture resulting in hypocapnic hypoxia (16, 22). As seen in the present study, naive 5- to 6-day-old rats exhibit a triphasic pattern of gasping following primary apnea, which eventually wanes and gives way to secondary apnea and death (8, 13). As far as we are aware, the neurophysiological basis for the three phases of gasping that follow primary apnea is unknown. It may, however, result from firing of different populations of neurons in the lateral tegmental field of the medulla [the proposed neural substrate underlying gasping in the rat (9, 36, 43)], which have different thresholds and/or latencies to the hypoxic stimulus or perhaps it results from the influence of various neuromodulators on the firing pattern of a single population of neurons during hypoxia. In the present study, the three phases of gasping became less discernible as the number of prior HIA/AR were increased prior to exposure to unrelenting hypoxia. Although the mechanism of the altered gasping pattern after prior exposure to hypoxic-induced apnea is unknown, it may have resulted from substrate depletion, altered neuroendocrine function, or synthesis and release of neuromodulators that influence hypoxic gasping. As previously mentioned, a number of factors have been shown to govern the time to last gasp in rats during early postnatal development on exposure to unrelenting hypoxia including core temperature (28), glucose (37, 44), catecholamines (45), excitatory amino acids (12), and nitric oxide (13). Our laboratory has previously shown that exposure to hypoxia induces a “regulated” decrease in core temperature (3) and that core temperature influences the time to last gasp as well as the total number of gasps in 5- to 6-day-old rat pups on exposure to unrelenting hypoxia. Variations in core temperature, however, are unlikely to have altered the gasping pattern after prior HIA/AR in our present experiments as core temperature was clamped at or near 37°C by regulating environmental temperature (23), and it is an increase rather than a decrease in core temperature that elicits a decrease in the time to last gasp by (28). As oxygen levels decrease to very low levels, a transition from aerobic to anaerobic metabolism occurs throughout the body and energy for processes such as gasping is provided by glycolysis via the Embden-Meyerhof pathway, which utilizes carbohydrate (e.g., glucose and/or glycogen) as a substrate. With regard to provision of substrate, Stafford and Weatherall (37) have shown that neither liver glycogen nor blood glucose levels determine survival time (i.e., the time to last gasp) when newborn rats are exposed to nitrogen. Brain glucose levels, however, can decrease dramatically during anoxia despite normal or elevated plasma (and liver) glucose levels, indicating an imbalance between brain glucose supply and demand (19). For example, experiments carried out by Holowach-Thurston et al. (19) on intact newborn mice at 37°C have revealed that even though plasma glucose levels double during a 6-min exposure to anoxia, brain glucose decreases by ∼72%. Brain glucose is likely a relatively important substrate for glycolysis in the newborn compared with the adult because basal brain glycogen is low and remains relatively stable during the first few minutes of anoxia perhaps due to the absence of enzymes (e.g., phosphoglucomutase) required for the utilization of brain glycogen (20, 29, 30). In rats pups, supplemental glucose has been shown to increase the time to last gasp during anoxic exposure when administered after the first few days of postnatal life (17, 18, 37, 44). Thus it is possible that the altered gasping pattern observed in our present experiments after repeated HIA/AR may have resulted from inadequate energy production via glycolysis secondary to low brain glucose levels. This postulate warrants investigation as does an investigation of the rate at which brain glucose is replenished in the newborn after bouts of hypoxic-induced apnea. Hypoxia is a potent stimulus for the adrenomedullary secretion of catecholamines, which mediate important respiratory, cardiovascular, and metabolic adaptations to oxygen lack during the perinatal period (4, 27). In the rat, which is born relatively immature, functional innervation of the adrenal medulla by the splanchnic nerves is not apparent until the second week of postnatal life (31, 32, 34). Adrenal chromaffin cells, however, possess a developmentally regulated oxygen-sensing mechanism, similar to that of carotid body type I cells (41), which mediate a “nonneurogenic” release of catecholamines in response to hypoxia until splanchnic control of adrenomedullary catecholamine secretion is functional (27). The role of catecholamines in hypoxic-induced gasping was shown in experiments carried out by Yuan et al. (45), who reported that the time to last gasp on exposure to unrelenting hypoxia was decreased in 1- and 8-day-old rat pups after adrenalectomy compared with that observed in sham-operated controls. Considering this and the evidence that hypoxia causes adrenal catecholamine depletion in rat pups (27), the altered gasping pattern observed in our present experiments after HIA/AR may have resulted from adrenal catecholamine depletion and the lack of a “normal” adrenal catecholamine response. This postulate warrants investigation as does an investigation of the rate at which adrenal catecholamines are replenished in the newborn after bouts of hypoxic-induced apnea. Although Slotkin and Kirshner (33) have shown that it takes up to 96 h for adrenal vesicular catecholamines to return to control levels following insulin administration in adult rats, as far as we are aware, the rate at which adrenal vesicular catecholamine replenishment occurs in the newborn after hypoxic-induced depletion is unknown. Gozal et al. (13) and Gozal and Torres (12) have shown that nitric oxide and glutamate, signaling molecules that influence neuronal excitability, play important roles in initiating and modulating the pattern of gasping in rat pups during exposure to anoxia. In their experiments, pretreatment of 5-day-old rat pups with N-nitro-l-arginine, a nitric oxide synthase blocker, significantly increased the time to first gasp and gasping duration without altering the total number of gasps. Pre-treatment of 5-day-old rat pups with MK-801 {(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine hydrogen maleate, a noncompetitive N-methyl-D-aspartate glutamate receptor channel antagonist} also prolonged the duration of primary apnea and increased the time to last gasp. Thus their data support the postulate that these neuromodulators favor the early appearance of gasps but limit anoxic tolerance during exposure to anoxia. Given that exposure to anoxia results in massive glutamate release (39) and activation of the brain nitric oxide system (e.g., Ref. 6), it is possible that these neuromodulators played a role in modulating gasping in our experiments after repetitive HIA/AR in our experiments. This warrants further investigation. The results of our experiments extend the observations of Gozal et al. (11), who recently reported that prolonged exposure to intermittent hypoxemia in the fetus or newborn alters the gasping pattern of 5-day-old pups on exposure to unrelenting hypoxia. In their experiments, intermittent fetal hypoxemia was produced by alternating the dam’s fraction of inspired oxygen between 0.21 and 0.10 at 90-s intervals from day 5 of gestation to term, and intermittent newborn hypoxemia was produced by alternating the pup’s fraction of inspired oxygen between 0.21 and 0.10 at 90-s intervals within 12 h of parturition until day 5 of postnatal life. Neither perturbation altered the time to first gasp, but both decreased the time to last gasp as well as the total number of gasps at day 5 of postnatal life when exposed to unrelenting hypoxia. In the present experiments, decreases in the time to last gasp and the total number of gasps were produced by prior exposure to as few as two and three HIA/AR, respectively, which induces a more severe level of hypoxia. Although the mechanisms of action may be different, it is interesting that such different low-oxygen-exposure regimens in essence have the same effect on hypoxic gasping in the 5-day-old rat pup. In infants, spontaneous recovery from obstructive sleep apnea or positional asphyxia during sleep is thought to occur early as a result of arousal from sleep or later as a result of hypoxic gasping when it is known as autoresuscitation (15, 40). Peiper (25), Stevens (38), and Thach (40) have emphasized the importance of gasping in self-resuscitation or autoresuscitation during apnea in human infants and that repeated episodes of apnea might lead to autoresuscitation failure and death. Why autoresuscitation fails is unclear, but our present experiments show that the duration and number of potential autoresuscitation producing gasps on exposure to unrelenting hypoxia are decreased by prior exposure to hypoxia and that this “impairment” lasts upward of 120 min. If oxygen does not become available immediately on the initiation of gasping, as conceivably may occur in infants, hypoxic secondary to obstructive sleep apnea or positional asphyxia, a decrease in the duration and/or number of gasps could diminish the chance of a successful autoresuscitation. GRANTSThis study was supported by the Canadian Institutes of Health Research. FOOTNOTESThe authors thank Sherry Moore for expert technical assistance. REFERENCES
Page 12the prevalence of overweight and obesity is high and continues to rise, presenting ever-increasing challenges for individuals and health professionals. Overweight individuals are at increased risk for cardiovascular disease, diabetes, and other health disorders (2, 5, 14, 21, 24). The location of the excess weight is of particular importance, because the strength of the relation between central obesity and disease risk is well documented (6, 13, 17, 30, 32), with visceral fat often considered the major culprit (3, 4, 12, 20, 22, 23). In addition, several studies have also shown a significant relationship between abdominal subcutaneous fat and metabolic risk factors (1, 8, 18–20). It is important for interventions designed to reduce abdominal obesity to monitor their effects on both visceral and subcutaneous fat. Controversy exists regarding the minimal and/or optimal amount of exercise needed for health benefits. Interestingly, despite the importance of centrally located body fat, there are few if any prospective exercise training studies that compare the effects of different amounts and intensities of exercise on changes in parameters of central obesity. Studies of Targeted Risk Reduction Interventions through Defined Exercise, a randomized, controlled clinical trial, was prospectively designed to investigate, in an 8-mo training study, the separate effects of the amount of exercise and exercise intensity on cardiovascular risk factors in overweight men and women with mild to moderate dyslipidemia. This report summarizes the effects of exercise amount and intensity on visceral, subcutaneous, and total abdominal fat. METHODSA complete description of the Studies of Targeted Risk Reduction Interventions through Defined Exercise design, hypotheses, recruitment strategies, methods, and preliminary recruiting results are published elsewhere (16). Subjects (n = 330) from Durham, Greenville, and surrounding communities in North Carolina met inclusion criteria and were randomized into the study. Sixty-eight percent (225) completed the 8-mo study. Of the 225 completers, 78% (n = 175) had complete pre- and poststudy computed tomography (CT) scan data, and the data from these subjects were included in this analysis. There were no differences in any variables measured between those in the subgroup who had CT scans and those who did not have scans. Inclusion criteria were 40–65 yr of age, sedentary (exercise <2 times/wk), overweight or mildly obese (body mass index of 25–35 kg/m2) with mild to moderate lipid abnormalities [either low-density lipoprotein (LDL) cholesterol of 130–190 mg/dl or high-density lipoprotein (HDL) cholesterol of <40 mg/dl for men or <45 mg/dl for women]. Women were postmenopausal. Exclusion criteria were diabetes, hypertension, other metabolic or musculoskeletal diseases, current use of or intent to diet, use of confounding medications, overt presence of coronary heart disease, or unwillingness to be randomized to any group. The study was approved by an Institutional Review Board. After written, informed consent was given, subjects were randomly assigned to one of three 8-mo exercise groups (∼2 mo of ramp-up and 6 mo of steady-state training) or a 6-mo control group. Individuals initially assigned to the control group were promised to be randomized into an exercise group at the end of the control period. However, only their control data were used in any analyses. The research protocol was approved by the institutional review boards at Duke University and East Carolina University. The exercise groups were 1) high amount/vigorous intensity (equivalent to jogging 20 miles/wk), 2) low amount/vigorous intensity (equivalent to jogging 12 miles/wk), and 3) low amount/moderate intensity (equivalent to walking 12 miles/wk). Details are published elsewhere (15, 16). The actual exercise prescription was to expend 14 kcal·kg body wt−1·wk−1 for the two low-amount groups (26) and to expend 23 kcal·kg body wt−1·wk−1 for the high-amount group. Although the amount of exercise is expressed in terms of walking or jogging a certain distance to simplify the description of the exercise groups, the main exercise modalities were treadmills and elliptical trainers, with some use of cycle ergometers. Subjects could use any or all of these modalities. The specific exercise intensities were 65–80% of peak oxygen consumption for the two vigorous-intensity groups and 40–55% of peak oxygen consumption for the moderate-intensity group. The exercise capacity was determined via a graded maximal exercise test with continuous measurement of oxygen consumption. The actual work rate, correlating to the prescribed exercise intensity, was determined during a submaximal exercise test performed on a separate day during the first 2–3 wk of exercise training. All exercise sessions were verified by direct supervision or by use of a heart rate monitor that provides recorded data (Polar Electro, Woodbury, NY). Adherence was calculated weekly as a percent, equal to the actual number of exercise minutes completed each week at the appropriate intensity, divided by the total number of minutes prescribed. There was an initial ramp period of 2–3 mo where exercise duration and exercise intensity were gradually increased until the appropriate exercise prescription was obtained. This initial ramping period was followed by 6 additional mo of training at the appropriate exercise prescription. Nutrient intakes of each subject were determined via 3-day food records and 24-h recalls, before and after exercise training. To study the effects of exercise alone and eliminate the confounding effects of major weight loss, subjects were counseled not to diet or change their diet during this study. Height (to the nearest 0.64 cm) and weight (to the nearest 0.1 kg) were measured in light clothing without shoes on a digital electronic scale (Scale-Tronix 5005, Wheaton, IL). All CT scans were performed on a General Electric CT/I (GE Medical Systems, Milwaukee, WI). After a digital frontal scout radiograph of the abdomen was obtained, a single 10-mm-thick axial section was performed at the level of the L-4 pedicle. CT scans were analyzed using Slice-O-matic software from TomoVision to determine surface area of the visceral and subcutaneous abdominal compartments. Baseline descriptive statistics include means and standard deviations (see Table 1). Paired t-tests (2-tailed) were used to determine whether a change within any specific group was significant (see Table 2). To determine whether there were significant differences between groups, data were analyzed using one-way ANOVA (Statview Software, SAS Institute, Cary, NC) (see Fig. 1). All ANOVA tests performed were found to be significant. Therefore, a Fishers paired least-significant difference post hoc test was performed to determine which groups were significantly different from the others. P values of <0.05 were considered significant. Fig. 1.Comparison of the effects of continued physical inactivity (controls) and 3 different exercise training programs on mean changes (Chg) in visceral abdominal fat (A), subcutaneous abdominal fat (B), and total abdominal fat (C). Subjects in the control group maintained their normal diet and level of physical activity for 6 mo. In the exercise groups, the amount and intensity of exercise were gradually increased to the prescribed level over the course of 1–3 mo, after which time exercise was maintained at the prescribed level of 6 mo. Low-amount, moderate-intensity exercise represents the caloric equivalent of walking ∼12 miles/wk at 40–55% of peak oxygen consumption; low-amount, vigorous-intensity exercise represents the same amount of exercise at 65–80% of peak oxygen consumption. High-amount, vigorous-intensity exercise represents the caloric equivalent of jogging ∼20 miles/wk at 65–80% of oxygen consumption. Values shown represent means of individual change scores. Error bars represent standard errors.
RESULTSBaseline and exercise prescription data are presented in Table 1. There were no differences at baseline between groups for age, body mass index, caloric intake, or percentage of calories from macronutrients. The number of women and men was nearly equal in each group, and minorities made up 19.5% of the subject population. The total amount of exercise time for each group was approximately 3 h/wk for the low-amount/moderate-intensity and high-amount/vigorous-intensity groups and was 2 h/wk for the low-amount/vigorous-intensity group. Caloric intake, measured before and after exercise training, did not change significantly (P > 0.20) in any exercise group or in the control group (data not shown). In Table 2, the results from paired t-tests on the change scores for within-group comparisons are presented for each of the CT-derived abdominal fat compartments and for body weight. In the control group, visceral fat levels increased by 8.6%, which was statistically significant (P = 0.001). Visceral fat levels did not change significantly in either of the low-amount exercise groups. The high-amount exercise group experienced an average decrease in visceral fat of 6.9%, which was significant (P = 0.038). Only the high-amount exercise group had any change in subcutaneous abdominal fat amount, which decreased in this group by 7.0% (P < 0.001). The significant increase in total abdominal fat in the controls reflects the increase in visceral fat. Neither low-amount exercise group experienced significant change in subcutaneous or total abdominal fat. Body weight increased significantly in the control group (0.88 kg) and decreased significantly in all exercise groups (0.60 in both low-amount groups and 2.31 kg in the high-amount group) in a dose-response manner, i.e., greater weight loss with greater amounts of exercise (in kcal expended·kg−1·wk−1). For the purposes of defining dose-response effects, both dose and amount refer to the number of kilocalories expended via exercise per kilogram of body weight per week. Exercise intensity did not appear to have any effect on body weight or any of the abdominal fat compartments, as both low-amount groups experienced similar responses. In Fig. 1, comparisons between the four groups illustrate the effects of continued physical inactivity (controls) and different amounts and intensities of exercise training on visceral, subcutaneous, and total abdominal fat. In Fig. 1A, the effect of exercise amount on visceral fat reveals a dose-response relationship, where the control group gained visceral fat relative to all three exercise groups (albeit, the low-amount/moderate-intensity group was at the margin of statistical significance, compared with the controls). And the high-amount group lost more visceral fat than the low-dose groups, although these differences did not quite achieve statistical significance (P > 0.07 and <0.09). In Fig. 1B, the data show that little or no change in subcutaneous abdominal fat for the controls and both low-dose exercise groups, whereas the high-amount exercise group lost a significant amount of subcutaneous abdominal fat compared with the other three groups. Figure 3C illustrates the additive effects of both the subcutaneous and visceral changes where the high-amount exercise group lost significantly more total abdominal fat compared with the other three groups. When gender was added to the ANOVA model, no significant gender effects were observed (P > 0.20). The correlations between abdominal fat depots (subcutaneous and visceral abdominal fat) and variables of metabolic risk (lipid and carbohydrate variables) are shown in Table 3. Both pretreatment (baseline) correlations and change score correlations are shown. Baseline subcutaneous abdominal fat was not significantly related to any baseline metabolic risk variable, whereas baseline visceral fat was highly significantly related to all baseline metabolic risk variables tested (and presented) (P < 0.0001). All posttreatment correlations were essentially the same as pretreatment variables (data not shown). Correlations between change in abdominal fat depots (both subcutaneous and visceral) and changes in the metabolic risk variables also are shown in Table 3. The change in subcutaneous abdominal fat was significantly correlated only to the change in HDL size (P < 0.05). The change in visceral fat was significantly related to change in LDL particle number and change in insulin sensitivity index. Correlations between change in visceral fat and changes in the other metabolic variables (HDL size, LDL size, and triglyceride) were just on the border of statistical significance (P < 0.10 and >0.05).
DISCUSSIONThe important relationships between disease risk and central obesity in general and visceral obesity in particular are well described (2–4, 6, 12, 13, 17, 20, 22, 23, 30, 32). In this study, we present data from the first prospective, randomized, controlled study on the effects of different amounts of exercise on visceral, subcutaneous, and total abdominal fat. The data support several key findings. First, in sedentary, overweight adults assigned to the control group, a relatively short period of continued physical inactivity resulted in a sizeable and significant increase in visceral abdominal fat. This finding emphasizes the high cost of continuing to choose a sedentary lifestyle for overweight, middle-aged adults. A second key finding was that in both low-amount groups, no significant increase in visceral fat was observed, suggesting that this amount of exercise was adequate for preventing the deterioration seen in the inactive controls. This is an important observation, because this amount of exercise is similar to that recommended by the Center for Disease Control/American College Sports Medicine (27) and because the importance of prevention (14, 21) is highlighted by the high rate of weight-loss recidivism. Until we are able prevent weight regain after short-term dieting success, a greater emphasis toward prevention should be a major goal in the US (14, 21). Third, the observation that the high amount of exercise (approximately equivalent to 17 miles/wk of vigorous exercise) not only prevented increases in visceral fat but actually resulted in sizable and significant decreases in visceral fat, as well as in subcutaneous and total abdominal fat, suggests an exercise prescription for reversing metabolic disease. That this amount of exercise can reverse metabolic disease is supported by the present data and by previously published findings from this cohort that showed improvements in lipids and lipoproteins (15), insulin sensitivity (10), and body mass and fat mass loss (31). Taken together, the data suggest a clear dose-response relationship between exercise amount and changes in visceral fat. Our interpretation of the data as indicating a dose-response relationship between exercise amount (in kcal·kg body wt−1·wk−1) is based on two points. First, there are multiple possible responses that can be characterized as a dose-response relationship, as illustrated by Haskell (9). All fulfill the basic concept that, with greater amounts of exercise, greater biological benefits accrue. With some health benefits, the response relationship might be curvilinear, whereas with others the response may be linear. Second, if we fit a linear curve to the mean visceral fat response (as seen in Fig. 1A) vs. the mean exercise dose for each group (calculated as amount of exercise prescribed times adherence), we observe a relationship with a r2 of 0.96. Fitting a best-fit curvilinear “trend line” results in a r2 of 0.99. Although in our study all intergroup comparisons were not significantly different, a higher powered study might have found a difference. Although analysis of variance revealed a highly significant group effect, there were no significant gender effects (P > 0.20) or group × gender interactions (P > 0.20). The importance of visceral fat and its associations with risk factors for coronary heart disease and Type 2 diabetes have been well established. Although overall obesity is clearly related to increased risk for these diseases, the greater importance of the location of adipose tissue is illustrated by the finding that, compared with total body fat, visceral fat is a significantly higher correlate of insulin response to a glucose challenge, fasting triglycerides, both systolic and diastolic blood pressure, and for HDL-to-total cholesterol ratio. In fact, visceral fat explains approximately twice the amount of variance in these variables compared with total body fat (13, 28). Our data support these observations. In a recent study comparing lean insulin-sensitive subjects to lean insulin-resistant subjects and obese insulin-resistant subjects, the data revealed that differences in visceral fat explain much of the atherogenic lipoprotein profile that is associated with obesity and insulin resistance (23). Whether visceral obesity is a major contributor to disease risk or simply a covariate of other causative factors is controversial (7, 30). Either way, the consistent, significant associations between visceral fat and risk factors for coronary heart disease and Type 2 diabetes suggests that it is, at the very least, a good marker of increased risk for these diseases. Despite its importance, there are few randomized, controlled studies of the effects of exercise on visceral fat. In a 12-wk study in overweight men, Ross et al. (29) reported that an exercise program designed to increase energy expenditure by 700 kcal/day for 12 wk resulted in a weight loss of 7.5 kg and a decrease in visceral fat of 52 cm2 (reported as the cross-sectional area of fat on a single CT scan), corresponding to a decrease of 6.9 cm2 visceral fat per kilogram of weight loss. The men in their diet-only group (700 kcal deficit) had a similar decrease of 5.9 cm2 per kilogram of weight loss. In the present study, the men in the high-amount exercise group experienced a reduction of 5.6 cm2 per kilogram of weight loss. Irwin et al. (11) studied the effects of a 12-mo exercise program in overweight postmenopausal women and found a decrease of 8.5 cm2 of visceral fat and 1.3 kg of body weight, corresponding to a ratio of 6.5 cm2 per kilogram of weight change. In the high-amount exercise group from the present study, the women lost 6.9 cm2 of visceral fat per kilogram of body weight. In cross-sectional studies, visceral fat is often found to be significantly correlated with metabolic risk factors. Statistically significant correlation coefficients ranging from 0.30 to 0.60 are often reported for the relation between visceral fat and numerous lipid and carbohydrate risk factors in these studies (12, 22, 25). Our cross-sectional data reveal similar magnitude coefficients when baseline metabolic risk variables are correlated to baseline visceral fat (P < 0.0001; coefficients range from 0.27 to 0.44 for the variables reported in Table 3; data not shown). However, as can be seen in Table 3, when we correlate change in visceral fat levels with change in these metabolic variables, the magnitude of the coefficients is much lower, although the relations are significant (for LDL particle number and insulin sensitivity) or just on the border of statistical significance. Although Jansen et al. (12) reported baseline correlations ranging from 0.32 to 0.51 (all P < 0.05), they found no significant correlations between change in visceral fat levels and change in metabolic variables after weight loss from diet only or diet plus exercise. It is important to remember that most if not all of the data linking visceral fat with metabolic variables are associative, not causative. There is much controversy as to whether visceral fat is a major health culprit or simply a marker of obesity-related health problems (30). Ravussin and Smith make a compelling case that failure to develop adequate fat cell mass in the face of excess energy intake may be the primary culprit, which then leads to ectopic fat deposition and in this way link visceral fat and adiposity to disease (28a). Either way, it seems clear that visceral fat, whether causative or simply a more specific marker of disease risk than general obesity, is an important health parameter. Major strengths of this study include 1) the randomized, controlled design; 2) the dose-response testing; 3) the direct verification of time and intensity and, therefore, exposure for nearly all exercise training sessions; 4) the carefully defined and controlled exercise amounts and intensities; 5) a significant proportion of women and minorities in the study population; 6) an exercise stimulus that is identical for men and women (defined by kilocalories of energy expenditure per kilogram of body weight per week rather than the same number of minutes per week) that allows for better comparisons of exercise effects between genders; and finally 7) a large number of subjects in each group yielding good statistical power to detect important exercise exposure effects. One important limitation should be noted. Due to practical reasons, we could not compare the effects of a training regimen with high weekly exercise amount at the lower intensity. We believed that the large amount of time necessary for a high-amount/moderate-intensity group (up to 8 h/wk for low fitness subjects) coupled with the fact that volunteer participants had to be willing to be randomly assigned to any group, would seriously limit recruiting ability and thus generalizability of the findings. Instead, our study was designed to look at the effects of exercise amount separately (by looking at the 2 groups that exercised at the same intensity but completed different amounts of total exercise) and exercise intensity separately (by comparing the 2 low-amount exercise groups that exercised at different intensities). In conclusion, continued physical inactivity in a sedentary middle-aged overweight population led to significant gains in visceral abdominal fat over a relatively short period of time (6 mo). This finding emphasizes the high cost of continued physical inactivity for sedentary, overweight adults. Even a relatively modest exercise program, consistent with the activity recommendations from Centers for Disease Control and American College of Sports Medicine prevented significant increases in visceral abdominal fat. In view of the high rate of recidivism with weight-loss programs, the importance of prevention cannot be overemphasized. However, a modest increase in weekly caloric expenditure over Centers for Disease Control and American College of Sports Medicine recommendations resulted in significant decreases in visceral, subcutaneous, and total abdominal fat without significant changes in caloric intake. Both the detrimental effects seen in the inactive control group and the beneficial effects of the high-amount exercise were observed in men and women. FOOTNOTESREFERENCES
Page 13Letter to Editor: It seems to me that the erudite arguments of both Drs. Green and Tschakovsky (1) can be correct within the same paradigm; that is, a specific test of brachial artery flow-mediated dilatation (FMD), where the ischemic stimulus is provided by a cuff placed distally and the occlusion time is ∼5 min, is a good test to reflect arterial nitric oxide release, whereas other variations of FMD in other territories may reflect a variety of complex stimuli and responses. I thought your readers might enjoy a historical perspective from our early thinking about the development of the FMD test in the brachial artery as an illustration of how good luck plays as important a role as good hypotheses in the generation of novel diagnostic tests. We chose the brachial artery for testing FMD because it could be easily imaged by conventional high resolution ultrasound technology, whereas the coronary arteries were inaccessible to ultrasound in the early 1990s. We reasoned that the brachial artery was of similar size to the major coronary arteries and so might give useful insights. We wanted to test function of the carotid arteries, but could not think of a way of producing hyperemia there easily, although several of the research fellows did do exercise while rebreathing carbon dioxide to try to stimulate hyperemia in the carotid circulation! We did indeed choose distal cuff occlusion, as we thought that direct arterial ischemia might confound the endothelium dependence of our measurements. The 5-min time interval, however, was a compromise between an ability to produce significant hyperemia and a comfort level that children and young adults could tolerate. The passage of time and the wide acceptance of this technique by many research groups around the world speaks to the utility of FMD testing. It is certainly not ready for clinical “prime time” for the detection of an individual’s vascular risk, but we believe that it has been a very useful methodology in clinical research, allowing insights into the risk factors for and treatments of early arterial abnormalities in children and young adults at risk of atherosclerosis. REFERENCES
Page 14AbstractThis letter is in response to the Point:Counterpoint series “Flow-mediated dilation does/does not reflect nitric oxide-mediated endothelial function” that appeared in the September issue (vol. 99: 1233–1238, 2005; doi:10.1152/japplphysiol.00601.2005; http://jap.physiology.org/content/vol99/issue3/2005). Authors Tschakovsky and Pyke (2) summarize flow-mediated dilation (FMD) in their rebuttal to Dr. Green as being nitric oxide (NO) dependent when the technique is 1) brachial or radial artery specific, 2) restricted to 5 min of distal cuff occlusion, and 3) in healthy subjects. Condition number three, however, is deserving of further clarification. To date, sex differences in the contribution of NO to 5 min of distal occlusion have not been well classified. The majority of studies referenced by Tschakovsky and Pyke and/or Green involving intra-arterial infusions of NO-synthase inhibitors have been conducted in men; those studies involving women have generally involved an overall sample size or female population too small to detect sex differences between subjects. Given that Levenson et al. (3) found that women exhibit a greater brachial artery dilation per unit increase in shear rate after 5 min of distal (forearm) occlusion and that estrogen may modulate both relaxing (PGI2, NO, endothelium-derived hyperpolarizing factor) and constricting (thromboxane, endothelin) substances released from the vascular endothelium (4), it is possible that the mechanisms underlying FMD in women differ from those established in men. Furthermore, although an age-associated decline in brachial artery FMD resulting from 5 min of distal occlusion has been well-documented in healthy humans (1), it is unknown whether this decline represents diminished NO-dependent dilation, as the mechanisms underlying FMD have not been thoroughly characterized in older adults and may involve additional endothelial pathways. Thus we would remind readers that both sex and age may influence the NO dependence of brachial or radial artery FMD. REFERENCES
Page 15AbstractThis letter is in response to the Point:Counterpoint series “Flow-mediated dilation does/does not reflect nitric oxide-mediated endothelial function” that appeared in the September issue (vol. 99: 1233–1238, 2005; doi:10.1152/japplphysiol.00601.2005; http://jap.physiology.org/content/vol99/issue3/2005). To the Editor: Responses of small arteries to flow have been shown to be dependent, partially dependent, and indeed independent of nitric oxide (NO; see September Point-Counterpoint, Ref. 1). As such, although NO may clearly modulate flow-mediated dilatory responses, the response does not purely reflect NO-mediated endothelial function. Why then do we see these differences? Although variations in species and arterial size may contribute, the complexity of the endothelial factor system itself is likely to underpin these observations. It is well known that there may be cross-talk between the different endothelial factor synthesis pathways such that inhibition or ablation of one may be compensated for by increased activity of another. For example, in endothelial NO synthase knockout (KO) mice, flow-mediated dilation is maintained by EDHF, whereas in wild-type animals it is mediated by NO (2). As such, the contribution of NO to the response may vary with physiological and pathophysiological conditions and indeed with experimental modulation of different endothelial factor pathways (e.g., use of different inhibitors/inhibitor combinations or KO animals). A novel suggestion is that the involvement of NO may also vary with the duration of the flow stimulus; inhibition of NOS inhibited vasodilation to sustained flow but was without effect on the immediate response to increased flow (3). Thus NO appears to play a varied role in mediating flow-induced dilatory responses. The involvement of other factors in certain conditions, however, means that flow-mediated dilation does not necessarily reflect NO-mediated endothelial function. REFERENCES
Page 16To the Editor: In this Point:Counterpoint series, Green and Tschakovsky and Pyke address the issue of whether flow-mediated dilation (FMD) reflects nitric oxide (NO)-mediated endothelial function (1). However, this question is difficult to answer. It has been conclusively shown that FMD of the brachial artery after 5 min of wrist cuff occlusion can be entirely abolished by the NO synthase (NOS) inhibitor NG-monomethyl-l-arginine (l-NMMA), suggesting a complete NO dependency of the process (2). Nevertheless, when the cuff was positioned on the arm above the probe, FMD was found to be 12% and was only reduced to 7.5% by l-NMMA (2). In contrast, Hornig reported a similar vasodilatory response of the radial artery (∼12%) independent of whether the occlusion was performed upstream (upper arm) or downstream (wrist) of the site of measurement (3). Interestingly, the increase in radial artery diameter was even more pronounced after longer periods of vessel occlusion before assessment of FMD (∼13% after 8 min vs. ∼7% after 4 min). The amount of FMD abolished by l-NMMA was ∼66% in healthy individuals but ∼33% in patients with chronic heart failure (CHF), possibly due to endothelin-mediated vasoconstriction in CHF (1, 3). Therefore, FMD appears to partially reflect NO-mediated endothelial function. The abovementioned data are consistent with the hypothesis that key players other than NO mediating FMD, e.g., prostaglandins and myogenic factors, in different experimental settings depend on 1) the duration of occlusion, 2) the site of measurement in relation to the site of vessel occlusion, and 3) the subjects, in whom the measurements are performed. REFERENCES
Page 17To the Editor: What is new in flow-mediated dilatation (FMD)? Not to characterize precisely the flow stimulus nor to consider the change in basal tone or the flow-mediated release of vasoconstrictors (5, 6). Certainly more that increasing the duration of hyperemia using longer periods of ischemia or hand warming progressively decreases the efficacy of NG-monomethyl-l-arginine (l-NMMA) to abolish FMD (3–5)? This while growing evidence accumulates for a role of endothelium-derived hyperpolarizing factor in the regulation of conduit artery diameter (1, 3). Does it mean that FMD is not a reliable noninvasive estimate of the capacity of human endothelial cell to release nitric oxide (NO)? No, but these experiments suggest a decrease in the relative importance of NO with the duration of the flow stimulus, thus inciting us to reserve this interpretation to short durations of stimuli and to persevere in our effort to elucidate the exact mechanisms involved in sustained FMD using higher doses of l-NMMA and/or inhibitors of alternative pathways (3). In the same way, it cannot be said that coronary artery FMD is NO independent because much evidence has been provided (2). Nevertheless, NO-independent pathways have been identified at this level that could compensate for the decrease in NO availability in relation to cardiovascular risk factors or the presence of atheroma (2). Furthermore, Tschakovsky and Pyke (3) claim that the shorter hyperemia we reported (5) after local NO synthase inhibition could have fully abolished the radial artery FMD and converted it in vasoconstriction. This is unlikely because such reduced stimuli are associated with lower but significant FMD and correspond to a single displacement along the flow stimulus-FMD relationship toward low values of flow (6). Conversely, the effect we reported reflects the downward shift of the relationship toward negative values of FMD and thus demonstrates the suppression of the vasorelaxant influence of NO at this level after l-NMMA. In addition, to declare “sympathetic activation can account for blunted FMD in numerous pathologies” appears premature. Indeed, local administration of norepinephrine does not alter FMD despite reducing hyperemia (7). Moreover, duration of hyperemia is decreased in pathologies characterized by elevated sympathetic activity, thus participating in altering FMD (3). Nevertheless, the role of this factor has not been precisely evaluated during sympathetic activation, although it could explain the beneficial effect of phentolamine on the blunted FMD induced by lower body negative pressure (3). Furthermore, in heart failure, physical training restores the radial FMD by increasing NO availability, although sympathetic tone is classically increased in such conditions (4). In summary, when the flow stimulus and the non-endothelium-dependent relaxation are preserved, a depressed FMD can reasonably be interpreted as an altered endothelial reactivity and, in the defined experimental conditions stated by Green, but also including coronary arteries, this strongly suggests an altered NO reactivity. However, this conclusion must not be dogmatic and lead us to forget that only careful demonstrations performed case by case can have value of proof. This is important and attractive for future research because other endothelium-derived relaxing factors distinct from NO could emerge particularly in pathological states and are probably involved in FMD during sustained flow stimulations. REFERENCES
Page 18AbstractThis letter is in response to the Point:Counterpoint series “Flow-mediated dilation does/does not reflect nitric oxide-mediated endothelial function” that appeared in the September issue (vol. 99: 1233–1238, 2005; doi:10.1152/japplphysiol.00601.2005; http://jap.physiology.org/content/vol99/issue3/2005). To the Editor: Having an ultrasound Doppler machine in our laboratory but hoping not to fall into the “delusion of grandeur” category of vascular researchers identified by Dr. Green (1), we offer the following thoughts regarding the importance of considering shear stimulus and subsequent nitric oxide (NO) release when evaluating flow-mediated dilation (FMD). Under the agreed-upon guidelines of a 5-min cuff occlusion below the site of measurement, we recently observed a diminution in brachial artery vasodilation in old subjects, insinuating an age-related decline in vascular function, which is in agreement with the current dogma (3). However, the reactive hyperemia in response to cuff occlusion was also diminished with age, so that the degree of brachial artery vasodilation was similar between younger and older subjects when normalized for the shear stimulus (unpublished observations). Although these observations do not resolve the present debate as to whether the FMD response reflects NO-mediated endothelial function, they do highlight the stimulus-response specificity which Dr. Tschakovsky outlines here and has published previously (2), emphasizing the importance of quantifying the shear stimulus as a part of the overall evaluation of FMD. Doing so may serve to prevent erroneous conclusions regarding changes in NO bioactivity that could simply be attributable to a dissimilar shear stimulus rather than underlying vascular dysfunction. With the recognition that varied shear will modulate the degree of NO release, we continue to value brachial FMD as a useful, noninvasive technique for evaluating endothelial function. REFERENCES
Page 19AbstractThis letter is in response to the Point:Counterpoint series “Flow-mediated dilation does/does not reflect nitric oxide-mediated endothelial function” that appeared in the September issue (vol. 99: 1233–1238, 2005; doi:10.1152/japplphysiol.00601.2005; http://jap.physiology.org/content/vol99/issue3/2005). Letter to Editor: The question as to whether endothelium-derived nitric oxide is responsible for flow-mediated vasodilation is an important one (1). There are few techniques available to measure the bioactivity of nitric oxide in humans. One such technique that has gained considerable popularity among clinician scientists employs high-resolution vascular ultrasonography to measure the change in the brachial artery diameter after a flow stimulus, i.e., reactive hyperemia. Flow-mediated vasodilation of the brachial artery is abnormal in patients with risk factors for atherosclerosis, such as hypercholesterolemia, cigarette smoking, and diabetes mellitus, and has been shown to predict the risk of future cardiovascular events (3). My laboratory addressed the hypothesis that endothelium-derived nitric oxide is responsible for flow-mediated vasodilation a decade ago and reported its findings (2). Specifically, we measured the vasodilator responses of the brachial artery to flow (after 1 min of reactive hyperemia after an ischemic stimulus) and to intra-arterial infusions of acetylcholine and nitroprusside before and after administration of the nitric oxide synthase antagonist NG-monomethyl-l-arginine (l-NMMA). l-NMMA inhibited flow-mediated vasodilation and the vasodilator response to acetylcholine but did not affect the response to nitroprusside. These observations enabled us to conclude that flow-mediated vasodilation of the brachial artery is an endothelium-dependent process in humans, mediated by nitric oxide. It is important to point out that Dr. Green cites our article in his point and rebuttal, whereas Drs. Tschakovsky and Pyke failed to acknowledge this contribution in their point and rebuttal. Issues of this import are enlightened by debate but are solved by carefully performed experiments. Our data support the notion that flow-mediated vasodilation of the brachial artery is mediated by endothelium-derived nitric oxide. REFERENCES
Page 20AbstractThis letter is in response to the Point:Counterpoint series “Flow-mediated dilation does/does not reflect nitric oxide-mediated endothelial function” that appeared in the September issue (vol. 99: 1233–1238, 2005; doi:10.1152/japplphysiol.00601.2005; http://jap.physiology.org/content/vol99/issue3/2005). Letter to Editor: Flow-mediated dilation (FMD) is an increasingly popular clinical surrogate of endothelial function largely because of its noninvasive nature and deceptively simple methodology. However, as any oenophile will happily tell you, deliberate and thoughtful evaluation are required to appreciate the underlying complex character of a wine. As the spirited debate between Dr. Green and Drs. Tschakovsky and Pyke (1) demonstrates, the same too can be said for FMD. Even when FMD is assessed using the recommended standardized technical approach to minimize measurement (and perhaps mechanistic) variability, the problem of biological variability remains. Despite current widespread use of FMD in clinical trials and research studies, there are few published reports of the reliability and reproducibility of the measurements, and normative data have yet to be established (2–4). Therefore, an important but often overlooked component of rational experimental design for FMD studies is determining variance estimates [between subject, within subject, between days (or weeks, months, or years)] to calculate appropriate sample sizes. Power calculations are rarely reported in FMD studies, but are especially important for reducing the likelihood of false-negative outcomes (type II error). Clearly much work remains to be done to sort out the utility of FMD as a meaningful physiological endpoint, let alone to understand the basic underlying mechanisms. REFERENCES
Page 21To the Editor: As an investigator who has used the flow-mediated dilatation (FMD) technique, I read the debate (2) with great interest. Several years ago, we asked a similar question by comparing two typical approaches to study endothelium-dependent vasodilation (brachial artery FMD and intrabrachial infusion of acetylcholine). Surprisingly, we found that these two common experimental approaches do not correlate (1). Because the pharmacological approach was considered as a “gold standard” for the measurements of endothelial function, these results could be taken as the evidence against the FMD method. However, when the acetylcholine responses were inhibited by the nitric oxide inhibitor NG-monomethyl-l-arginine (l-NMMA), only about 30–40% of the responses were blocked (4). In contrast, FMD is completely abolished by the infusion of l-NMMA (3). Thus FMD appears to reflect nitric oxide-mediated endothelial function, at least more than the pharmacological approach, and is a promising noninvasive technique for the assessment of endothelial function. However, if this is going to be a useful clinical and research technique, several issues regarding the methodology (e.g., location and duration of occlusion, timing of peak hyperemia, widely different normative values between studies) and interpretation/confounders (e.g., differences in arterial inflow, inverse relation with baseline diameter, influences of sympathetic tone) should be properly addressed. REFERENCES
Page 22AbstractThis case describes the physiological maturation from ages 21 to 28 yr of the bicyclist who has now become the six-time consecutive Grand Champion of the Tour de France, at ages 27–32 yr. Maximal oxygen uptake (V̇o2max) in the trained state remained at ∼6 l/min, lean body weight remained at ∼70 kg, and maximal heart rate declined from 207 to 200 beats/min. Blood lactate threshold was typical of competitive cyclists in that it occurred at 76–85% V̇o2max, yet maximal blood lactate concentration was remarkably low in the trained state. It appears that an 8% improvement in muscular efficiency and thus power production when cycling at a given oxygen uptake (V̇o2) is the characteristic that improved most as this athlete matured from ages 21 to 28 yr. It is noteworthy that at age 25 yr, this champion developed advanced cancer, requiring surgeries and chemotherapy. During the months leading up to each of his Tour de France victories, he reduced body weight and body fat by 4–7 kg (i.e., ∼7%). Therefore, over the 7-yr period, an improvement in muscular efficiency and reduced body fat contributed equally to a remarkable 18% improvement in his steady-state power per kilogram body weight when cycling at a given V̇o2 (e.g., 5 l/min). It is hypothesized that the improved muscular efficiency probably reflects changes in muscle myosin type stimulated from years of training intensely for 3–6 h on most days. To the Editor: The concept that extensive endurance training improves cycling efficiency is intuitively appealing but not well supported by the literature. Recently, Coyle (1) has published efficiency data from Tour de France Champion, Lance Armstrong. In this case study Coyle concluded that “the physiological factor most relevant to performance improvement as he matured over the 7-yr period from ages 21 to 28 yr was an 8% improvement in muscular efficiency when cycling” (1). Case studies documenting adaptations in truly elite endurance athletes are important (3); however, we believe Coyle’s case study is insufficient to support his conclusions because of limitations in study design and methodology. Armstrong was tested five times over a period of 7 yr. Only the first and last test occurred during the same month (November), making it difficult to distinguish seasonal effects from maturation effects. Unfortunately, Armstrong’s fitness data within 3 mo of racing a Tour de France tour is not reported. The majority of the improvement in gross cycling efficiency (GE) occurred after January 1993 (21.6%) and before August 1997 (22.7%), 8 mo after cancer treatment. Consequently, if there were real changes in GE it becomes difficult to distinguish whether the improvements in GE are due to cancer treatment or important aspects of training (e.g., training load, altitude training, high-cadence training, time-trial training, or resistance training). Coyle does not present data documenting the accuracy and reliability of the techniques used to calculate cycling efficiency (oxygen uptake, carbon dioxide production, and power output). Friction-braked bicycle ergometers have been shown to be inaccurate when dynamically calibrated (4). Previous research has reported that Monark ergometers tend to underestimate power output by ∼2–8% (4). If Coyle’s Monark ergometer was inaccurate, then Armstrong’s actual GE before winning his first Tour de France may have been ∼19–21%, values similar to those reported for recreational cyclists (5). Also of concern is the observation that the accuracy of Monark ergometers can change with age (4). Without routine assessment of accuracy with a dynamic calibration rig, it is difficult to know whether accuracy of the Monark used in Coyle’s study changed over the 7-yr period of data collection. The terminology used by Coyle to describe the “same Monark ergometer (model 819) used for all cycle testing” is confusing. In the methods section, Coyle states that “the calibrated ergometer was set in the constant power mode” and in the discussion section that there was “a progressive loss of pedal cadence at constant power during the 30–60 s before exhaustion.” Although we are unaware of a constant power mode for Monark (model 819) ergometers, this mode of operation is commonly used with a Lode electromagnetic ergometer. A Lode ergometer has been used in Coyle’s laboratory (2). It is possible that either inappropriate terminology was used in the methods section or Armstrong was tested on two different types of ergometers. Without the appropriate data, Coyle is left to speculate that, during the Tour de France tours (1999–2004), Lance possessed a maximal oxygen uptake (V̇o2max) of ∼6.1 l/min (based on the September 1993 testing session) and a body mass of ∼72 kg (based on “his reported body weight”) and therefore a relative V̇o2max of 85 ml·kg−1·min−1. These estimations suggest that efficiency improved (21.2–23.1%; ∼9%), while V̇o2max rose (70–85 ml·kg−1min−1; ∼21% increase) and body mass fell (from 78.9 to 72.0 kg; ∼9% decrease). In contrast to Coyle’s conclusions, it appears that conventional physiological adaptations to modifications in diet (loss in body mass) and training (gains in aerobic power) may be equally, if not more, important to Armstrong’s performance than the 9% improvements in cycling efficiency. In summary, although great insight into human physiology can be gained from carefully controlled examinations of elite athletes, poor experimental design and methodology can lead to inappropriate conclusions, which in the case of a sporting hero can quickly become more hype than fact. Coyle’s data supporting the assumption that training can improve cycling efficiency in an elite cyclist are not compelling. It appears that other more conventional explanations describing why Armstrong is such a successful cyclist may be equally tenable. REFERENCES
japJ Appl PhysiolJournal of Applied PhysiologyJ Appl Physiol8750-75871522-1601American Physiological Society REPLYEdward F. CoyleDepartment of Kinesiology and Health Education The University of Texas at Austin Austin, Texas 78712 e-mail: [email protected]102005To the Editor: I appreciate this opportunity to answer the four points and address the terminology that Martin et al. find “confusing” (point 3).
Page 23To the Editor: Elite athletes are valuable study objects for exercise physiology: successful sportsmen offer unique insight into the extreme adaptation of the human organism to certain types of exercise and illustrate the amazing adaptation capacity of human physiology (9). Because of the unique characteristics of the study subjects, sample sizes in these investigations are usually low. Even case reports, such as in the article written by Dr. Coyle (1), can therefore be a valuable contribution to the scientific knowledge in this field. Nevertheless, such studies should respect the basic principles of scientific investigations. We feel that the investigation presented by Dr. Coyle has serious limitations in this context. The aim of the study was, according to the author, to report “the physiological changes that occur in an individual bicycle racer during a 7-yr period” and thereby illustrate “the extreme to which the human can adapt to endurance training.” Unfortunately, the data presented in the manuscript do not contain enough physiological information of the athlete in question (Lance Armstrong) to draw a picture sufficient to illustrate his physiological profile and the associated adaptations over 7 yr: in fact, no testing was performed in immediate connection with his Tour de France wins. It can be assumed that his physiological performance at that moment was much higher than the ones measured and described by the manuscript. The performance data reported in the manuscript are common to many elite cyclists (4, 5), none of whom matches the wins of Armstrong. Furthermore, the exercise tests outside the cancer period date from the months of January, November, and September; these are periods where professional cyclists, who target peak form for races in July, have barely the same condition as during their peak season. Therefore, all speculations in the manuscript on potential data during his Tour de France wins are not supported by any of the presented test results. To display a complete physiological profile of the athlete and to draw the present conclusions, at least some data from peak season testing should have been included. Interestingly, no data from the years of best performance of the athlete are presented: during the period from 2000–2005, Armstrong won five consecutive Tours de France; unfortunately, no exercise test seems to have been conducted during that time, which is rather surprising for an athlete of Armstrong’s caliber. To evaluate exercise performance and draw valid conclusions, it is essential to report data on the reliability and accuracy of the testing equipment, especially when only small changes are expected or the accuracy of the testing equipment is poor. In exercise physiology, especially the assessment of respiratory data is prone to errors linked to the testing procedure. This error, together with biological variation of maximal oxygen uptake, has been demonstrated to reach up to 5% (3, 8), thereby almost equaling the changes described in the manuscript. The same applies to the ergometry equipment: it has been demonstrated that many ergometers yield a high inaccuracy in their measurements, especially mechanically braked models, such as the one used for the present investigation (6, 10). In a comparable case report which uses the same type of mechanically braked ergometer (9), the authors included a 9% correction for their power measurements. Unfortunately, the author does not report any data on the accuracy and reliability (such as calibration data) of his testing equipment. Especially when evaluating the calculations and conclusions drawn from the data, this would be of great help. Furthermore, we are not aware of a reliable constant power mode in mechanically braked ergometers, such as the Monark model used for several tests in the present study. The author highlights the importance of improved muscular efficiency as being the main reason for Armstrong’s outstanding gain in performance. We feel that this assumption cannot be made on the basis of the presented information, because no records are available from periods where the athlete actually had peak form. In this context, Fig. 1 is not correct, because it implies that Armstrong’s gross and delta efficiency have been constantly rising since the age of 20 yr, despite a period of more than reduced physical condition during cancer treatment. On the basis of the presented data, the author cannot judge the efficiency of any other moment than the ones studied (November 1992, January 1993, August 1997, November 1999). Furthermore, the conclusion of the manuscript is even more surprising, because it has been shown that efficiency is not a key factor to differentiate between successful and unsuccessful cyclists (2, 7). Unpublished data from our laboratory support these assumptions: elite cyclists do not show higher efficiency than recreational cyclists. Furthermore, a high interindividual variability can be noted. In a longitudinal follow-up (intraindividually), however, efficiency remains remarkably stable, even when overall physiological exercise performance highly varies. It is therefore more likely that, in addition to very favorable genetic assets of the athlete, common physiological adaptations associated with endurance training, such as an improved aerobic and possibly anaerobic energy metabolism, increased power-to-weight ratio, or enhanced recovery functions, might have added to the truly outstanding sporting achievements of Lance Armstrong. It has to be considered that, aside from being determined by purely physiological factors, performance in sporting competitions is highly influenced by many other variables, such as tactical race understanding and motivational and psychological issues. Although speculative, the latter two might play a prominent role in Armstrong’s sporting achievements, especially when considering the athlete’s unique medical history and human experience as a cancer survivor. Armstrong might have gained the edge over his physiologically equally strong competitors by these means. We feel obliged to raise these issues to the scientific community on behalf of all scientists working with elite athletes. Even when the popularity of an athlete might strongly influence the interest of publishing data, both from the author working with the athlete and the editor’s side, the basic principles for scientific investigations should be respected. Published data (especially if published in a highly regarded scientific journal like the Journal of Applied Physiology) represent the base of knowledge and interpretation for future investigations and should therefore fulfill these scientific principles to allow upcoming studies to rely on the validity of their outcomes. REFERENCES
japJ Appl PhysiolJournal of Applied PhysiologyJ Appl Physiol8750-75871522-1601American Physiological Society REPLYEdward F. CoyleDepartment of Kinesiology and Health Education The University of Texas at Austin Austin, Texas 78712 e-mail: [email protected]102005To the Editor: I thank Dr. Schumacher et al. for the opportunity to discuss the reliability and validity of our methods. Regarding “scientific considerations,” this study focused on physiology and not the science of bicycle racing. Our main purpose was not to make measurements around the Tour de France or to compare this subject (Lance Armstrong) with other champions. The fact that our subject happened to eventually win the Tour de France was interesting but not the main “scientific consideration.” Changes in muscle efficiency with 7 yr of training was the focus. Reliability was most important, both in terms of the subject as well as the measurements of indirect calorimetry and power. This subject’s level of training and accessibility were most reliable from year to year in the early part of the competitive season when most of our measures were made. Besides, our study of Armstrong began before he ever competed in the Tour de France. The fact that we did not report data after this subject won his first Tour de France emphasizes, again, that our purpose was to observe the maturation and not report the characteristics of the existing champion. Schumacher et al. have requested data regarding the reliability of our respiratory testing equipment for measuring oxygen consumption. During submaximal exercise at 60–70% maximal oxygen consumption in a group of competitive cyclists (circa 1994), we have observed that oxygen consumption when measured on 8 separate days in a given individual displayed an average range of 0.08 l/min and a coefficient of variation of ±0.87% (2). See Martin et al. (5) for additional insight. The notion that a set 9% correction should be applied to all Monark ergometers is not supported by Maxwell et al. (6). The model 819 Monark ergometer used by Armstrong was calibrated statically and dynamically using pedal dynamometers and found valid to within ±3% (1, 4), and power can be held constant [as detailed in Martin et al. (5)]. Schumacher et al. state that “Fig. 1 is not correct” and then say that “on the basis of presented data, the author cannot judge the efficiency of any other moment than the ones studied (November 1992, January 1993, August 1997, November 1999).” The manuscript never “judged” or speculated about efficiency as it only reported actual data. Removing data from 1997 does not alter the line between 1992 and 1999. These data over years, to our knowledge, are the only published addressing long-term efficiency and training. These data seem to conflict with notions of Schumacher et al., because they state “efficiency is not a key factor to differentiate between successful and unsuccessful cyclists” on the basis of their own unpublished data as well as the work of others (7). We have presented a model of how numerous physiological factors interact to determine endurance performance and have discussed that efficiency by itself does not account for most of the interindividual variations in performance. In fact, in our 1991 manuscript (1), we also report that efficiency in a group of elite cyclists does not differ significantly from a group of good cyclists because of the high degree of individual variation in efficiency and fiber type. However, in a following study during which maximal oxygen consumption and lactate threshold were matched in a pairs of competitive cyclists, it was clear that performance power was significantly higher in subjects with greater gross efficiency and greater percentage of type I fibers (3). In fact, Armstrong makes this point in that his efficiency was only average when he was 21–22 yr despite the fact that he was already elite and world champion. However, his efficiency improved and he was able to generate 8% more power when cycling at a constant V̇o2 of 5.0 l/min. We appreciate that winning the Tour de France requires tactical race understanding and motivational and psychological issues, among other things. However, nonphysiological factors and the winning of the Tour de France, although interesting, are not the focus of this investigation.
|