Although resting hemodynamic load has been extensively investigated as a determinant of left ventricular (LV) hypertrophy, little is known about the relationship between provoked hemodynamic load and the risk of LV hypertrophy. We studied central pressure-flow relations among 40 hypertensive and 19 normotensive adults using carotid applanation tonometry and Doppler echocardiography at rest and during a 40% maximal voluntary forearm contraction (handgrip) maneuver. Carotid-femoral pulse wave velocity (CF-PWV) was measured at rest. Hypertensive subjects demonstrated various abnormalities in resting and induced pulsatile load. Isometric exercise significantly increased systemic vascular resistance, aortic characteristic impedance (Zc), induced earlier wave reflections, increased augmentation index, and decreased total arterial compliance (TAC; all P ≤ 0.01). In hypertensive subjects, CF-PWV was the strongest resting predictor of LV mass index (LVMI) and remained an independent predictor after adjustment for age, gender, systemic vascular resistance, reflection magnitude, aortic Zc, and TAC (β = 2.52 m/s; P < 0.0001). Age, sex, CF-PWV, and resting hemodynamic indexes explained 48% of the interindividual variability in LVMI. In stepwise regression, TAC (β = −17.85; P < 0.0001) during handgrip, Zc during handgrip (β = −150; P < 0.0001), and the change in the timing of wave reflections during handgrip (β = −0.63; P = 0.03) were independent predictors of LVMI. A model that included indexes of provoked hemodynamic load explained 68% of the interindividual variability in LVMI. Hemodynamic load provoked by isometric exercise strongly predicts LVMI in hypertension. The magnitude of this association is far greater than for resting hemodynamic load, suggesting that provoked testing captures important arterial properties that are not apparent at rest and is advantageous to assess dynamic arterial load in hypertension.
- arterial load
- arterial compliance
- wave reflections
- aortic stiffness
left ventricular (lv) hypertrophy (LVH) is one of the most powerful predictors of cardiovascular morbidity and mortality in subjects with hypertension (8, 9, 12, 15). In addition, regression of LVH in response to antihypertensive therapy has been shown to independently predict a more favorable prognosis in adults with hypertension (7, 32).
LV afterload is a key determinant of LVH. In the presence of a normal aortic valve, LV afterload is largely dependent on the properties of the arterial tree (“arterial load”) (19, 22). Understanding the components of arterial load that are associated with LVH in treated hypertension is important, since targeting these with additional therapy may result in more effective regression of LVH. The relationship between different components of hemodynamic load or arterial abnormalities and LVH has been the focus of previous studies (3, 4, 6, 24), which evaluated resting physiological indexes. Given the dynamic nature of the arterial tree, it is unlikely that measurements of arterial load performed at rest fully reflect the true hemodynamic load imposed by the vasculature on the LV on a day-to-day basis. Provocative maneuvers, such as submaximal isometric handgrip exercise, may reflect characteristics of the vasculature that are not captured at rest (10, 23). Handgrip exercise at 30–40% of maximal voluntary contraction induces a variable increase in blood pressure, with abnormal responses in hypertensive individuals (10, 30). Given that induced hemodynamic changes may be more informative than resting measurements, the study of the relationship between such responses and LVH has been pursued in multiple investigations, often with discordant results (1, 2, 11, 21, 29, 31). However, previous studies used brachial pressure responses as indexes of hemodynamic load.
It is increasingly recognized that brachial pressures are inaccurate surrogates of arterial load (19, 22). The discrepancy between brachial pressures and central hemodynamics is likely to be even more pronounced in the setting of acute interventions (such as isometric handgrip exercise) that affect heart rate and the vasculature via passive effects due to changes in blood pressure and active modulation via changes in vasomotor tone. Such modifications may affect arterial wave reflections and aortic-to-brachial pulse pressure amplification (16, 22) and may have unpredictable effects on volumetric aortic flow, which affect the interpretation of changes in blood pressure.
In this study, we aimed to characterize the changes in hemodynamic load in response to isometric exercise with detailed analyses of central arterial pressure-flow relations and test the hypothesis that hemodynamic load provoked by isometric exercise predicts LVH in treated hypertensive adults, above and beyond resting hemodynamic indexes. A secondary objective of the study was to characterize the (normal) hemodynamic response to isometric exercise in a reference sample of normotensive adults.
MATERIALS AND METHODS
We enrolled 40 hypertensive adults, aged 20–80 yr, referred for an echocardiographic examination at the Hospital of the University of Pennsylvania (Philadelphia, PA). Exclusion criteria were as follows: 1) pregnancy; 2) congestive heart failure; 3) LV ejection fraction <50%; 4) personal or family history of hypertrophic cardiomyopathy or presence of asymmetric septal hypertrophy; 5) history of coronary heart disease or wall motion abnormalities detected by echocardiography; 6) poor acoustic windows, impeding adequate assessment of LV mass; 7) inability to provide informed consent; 8) more than trace mitral regurgitation or aortic regurgitation; 9) any degree of aortic stenosis; 10) any changes in antihypertensive therapy over the last 3 mo. Antihypertensive therapy was not held for the study because we intended for our arterial load assessments to be representative of the arterial load to which the ventricle was exposed over the 3 mo before the examination, therefore not obscuring any associations between ongoing arterial load and LV remodeling. This was also the rationale behind including only subjects with stable antihypertensive therapy over that period of time.
To describe the normal changes in arterial load in response to isometric exercise, we also studied a group of normal controls (n = 19) recruited through flyer advertisements, who were not taking any vasoactive medications and did not have a history of smoking, diabetes mellitus, hypertension, dyslipidemia, cardiovascular disease, or any other major health condition. This study was approved by the University of Pennsylvania Institutional Review Board, and all subjects provided written, informed consent.
Echocardiographic examinations were performed with a Vivid-7 ultrasound platform (GE Healthcare; Bridgeport, CT). Pulsed wave Doppler measurements of flow velocities in the LV outflow tract were performed and recorded at high sweep speed (200 mm/s). LV mass was calculated with the area-length method and indexed for body height in meters to the allometric power of 2.7. Ventricular length was measured in a three-dimensional (3D) full-volume data set obtained from the apical view. This 3D method avoids cavity foreshortening, therefore improving the accuracy of the measurements.
We used a commercially available applanation tonometry system (SphygmoCor Vx System, AtCor Medical, Sydney, Australia) equipped with a high-fidelity Millar tonometer (Millar Instruments, Houston, TX). We recorded carotid artery waveforms simultaneously with LV outflow tract flow velocities at rest and during isometric exercise. Carotid pressure waveforms were calibrated according to brachial mean and diastolic pressure measured with an oscillometric device (Hewlett Packard 78352c; Hewlett Packard, Palo Alto, CA). We used mean and diastolic pressures rather than systolic and diastolic pressures for calibration of the wave form because mean and diastolic pressures exhibit little variation between the arm and the central arteries, in contrast to systolic pressure, which increases (variably) from the aorta to the brachial artery (pulse pressure amplification). Carotid-femoral pulse wave velocity (CF-PWV), an index of large artery stiffness, was measured with sequential carotid and femoral arterial tonometry, to compute the time delay between the two sites, using the QRS complex as a fiducial point. The distance between the suprasternal notch and the carotid recording point was subtracted from the distance between the sternal notch and the femoral recording point, and CF-PWV was calculated as distance/Δtime.
Assessment of LV outflow tract cross-sectional area.
Using a dedicated offline workstation (EchoPAC, GE Healthcare, Chalfont St. Giles, UK), multiplanar reconstructions were performed from full-volume parasternal 3D-echocardiographic data sets. Using a midsystolic frame, two planes orthogonal to each other running parallel to the LV outflow tract long axis were selected. Subsequently, a third plane orthogonal to both long-axis planes was selected and translated to the position immediately proximal to the aortic valve leaflets, corresponding to the minimal cross-sectional area, which was digitally traced. In contrast to the two-dimensional (2D) approach, this 3D method is not affected by LV outflow tract eccentricity and, therefore, provides greater accuracy of LV outflow tract area measurements, which are essential for the computation of volumetric flow from Doppler flow velocities.
Methods for pressure-flow analyses have been previously described in detail (28). Both pressure and flow velocity data were processed offline using custom-designed software written in Matlab (The Mathworks, Natick, MA). Flow velocities were obtained from Doppler DICOM images and appropriately corrected for interrogation angle. Instantaneous flow velocities were multiplied by LV outflow tract cross-sectional area to obtain volumetric flow. Visual time alignment of pressure and flow curves was performed to maximize the following criteria: 1) concordance of the rapid systolic upstroke of pressure and flow; 2) concordance of the pressure dichrotic notch and cessation of flow; 3) zero value of the phase angle of higher frequency harmonics (7th to 10th) of input impedance; 4) linearity of the early systolic pressure-flow relationship (22, 28).
Characteristic impedance (Zc) of the proximal aorta was calculated in the time domain as the ratio of early systolic pulsatile pressure to flow, as previously described (20, 28). Reflection magnitude was computed using wave separation analysis (22, 28). In this method, after separation of the pressure wave form into its forward and backward components, reflection magnitude is computed as the ratio of the amplitudes of backward to forward pressure wave (22, 28). The time of return of the reflected wave was measured using the pressure and flow wave forms, as previously described (27).
Augmentation index was calculated as the amplitude of the second systolic peak divided by the amplitude of the first systolic peak (P2/P1) multiplied by 100. We used this method as opposed to the more widely used method proposed by Murgo et al. (21a), in which augmentation index is calculated as (P2 − P1)/pulse pressure, because the later often results in negative values of augmentation index, which are incompatible with data transformations often needed to satisfy linear modeling assumptions and because percent changes associated with the handgrip maneuver, as presented herein, would represent, in many instances, an inadequate expression of the magnitude and direction of change in the relative amplitudes of P1 and P2 compared with baseline. Total arterial compliance (TAC) was calculated with the pulse-pressure method, as previously described (28). Pulse pressure amplification was calculated as the ratio of peripheral pulse pressure to central pulse pressure.
Each subject performed three maximal voluntary dominant forearm contractions with a Stoelting handgrip dynamometer (Stoelting, Wood Dale, IL). The force of contraction was averaged, and a submaximal target of 40% was used for a sustained handgrip effort until fatigue. In addition to resting data, pressure and flow were recorded immediately before termination of the handgrip maneuver and used for pressure-flow analyses, as described above.
We based our power calculations on previous data from Gardin et al. (8), which included an analysis of a combination of demographic, clinical, and hemodynamic factors as predictors of LV mass using multiple regression. We used this particular study because it did not include measures of LV size or systolic function in the models, which made these data suitable for our purposes, since we were interested in focusing our analyses on arterial load. Based on the R2 value obtained in this study (0.37), we powered our study to detect an increase in the amount of prediction by at least 50% (R2 increase from 0.37 to 0.55), which was considered to be a significant increase from both the clinical and epidemiological standpoints. We estimated that a sample size of 39–40 subjects was needed to achieve at least 85% power to detect such an increase, using an F-test with α = 0.05.
Continuous values are expressed as means ± SD or median (interquartile range), as appropriate. We tested changes induced by isometric exercise in hypertensive and normotensive adults with repeated-measured ANOVA, with time of measurement (baseline or during handgrip) as the within factor and hypertensive status as the grouping factor. By use of this analysis, the effects of the maneuver, the differences between hypertensive and normotensive subjects, and the interaction between hypertensive status and response to the maneuver (i.e., different responses in normotensive vs. hypertensive subjects) were evaluated simultaneously, avoiding multiple comparisons, which tend to inflate the type I error rate. Given that hypertensive subjects demonstrated higher body weight than normotensive subjects and given that indexes of arterial load bear a physiological relationship with body size, for these models we normalized cardiac output, stroke volume, and indexes of arterial load for body surface area, whereas aortic Zc was normalized for body surface area raised to the allometric power of 0.64, as previously described (5).
Among hypertensive subjects, the relationship between arterial load and LV mass index (LVMI) was analyzed using bivariate linear regression and multivariate stepwise regression with forward selection to identify selected independent predictors of LVMI. Normality of model residuals and lack of important colinearity were ascertained. Finally, to assess potential confounding by body size relationships, we used log-log models that adjusted for body surface area and confirmed these estimations with nonlinear regression. In addition to model R2, we computed adjusted R2 values, a more conservative approach to estimate the proportion of predicted variability by multivariate models, which penalizes for increasing number of predictors. Sample size calculations were performed using PASS for Windows (NCSS, Kaysville, UT). All other analyses were performed using SPSS for Windows versioin 17 (SPSS, Chicago, IL).
Important clinical characteristics of the study population are summarized in Table 1. Subjects in the hypertensive group were significantly older and had higher body mass index, LV mass, and LVMI than subjects in the normotensive group. LV mass was, on average, 33 g higher in the hypertensive group (P = 0.001). Similarly, LVMI was significantly higher in the hypertensive group (35.6 vs. 28 g/m2.7; P = 0.004). LVMI percentiles 75, 90, and 95 among hypertensives were 40, 45, and 71 g/m2.7, respectively.
Fatiguing isometric exercise.
The mean force during sustained isometric exercise was 10.6 kg among hypertensive individuals and 11.8 kg among normotensive individuals (P > 0.05). Predictors of fatiguing force (40% peak force) were age (β = −0.8 kg per 10-yr increase; P < 0.01) and male sex (β = −3.2 kg; P < 0.01). Handgrip force did not correlate with LVMI in either univariate analysis (R2 = 0.05; P > 0.05) or after adjustment for age, sex, and hypertensive status (R2 increase = 0.03; P > 0.05).
The mean time of sustained isometric exercise was 213 s among hypertensive individuals and 252 s among normotensive individuals (P = 0.02). Predictors of time of exercise included age (β = −28 s per 10-yr increase; P < 0.001) and male sex (β = 56 s; P < 0.001). The difference in exercise time between normotensive and hypertensive individuals was no longer significant after adjustment for age and sex (P = 0.33). Exercise time did not correlate with LVMI in either univariate analysis (R2 = 0.04; P > 0.05) or after adjustment for age, sex, and hypertensive status (R2 increase < 0.005; P > 0.05).
Changes in hemodynamic indexes induced by isometric handgrip exercise.
Table 2 shows mean values and 95% confidence intervals of various hemodynamic indexes at rest and during isometric exercise in hypertensive subjects. Table 3 shows the corresponding values in normotensive subjects. Figure 1 shows values of heart rate (Fig. 1A), stroke volume (Fig. 1B), cardiac output (Fig. 1C), systemic vascular resistance (SVR) (Fig. 1D), aortic Zc (Fig. 1E), TAC (Fig. 1F), and effective arterial elastance (Ea) (Fig. 1G) at rest and during handgrip in normotensive and hypertensive subjects. To allow comparisons between the groups, shown values are indexed for body surface area. Solid lines represent the hypertensive group, whereas dashed lines represent the normotensive group. A significant within-group P value indicates a statistically significant change associated with the maneuver. A significant between-group P value indicates a significant difference in mean values between the groups. A significant group ∗ time P value indicates a significant difference between the groups in the magnitude (slope) of the change associated with the maneuver.
Isometric exercise induced a significant increase in heart rate and a nonsignificant decrease in stroke volume, without a significant change in cardiac output. There was a pronounced increase in SVR in both groups (P = 0.009), but there were no significant differences between the groups in SVR or the changes in SVR associated with isometric exercise. Aortic Zc was significantly higher in hypertensive subjects (P = 0.03) and increased significantly in response to isometric exercise. TAC was markedly lower in hypertensive subjects and decreased with the handgrip maneuver in both groups (P = 0.002). Ea was not significantly different between the groups and increased with isometric exercise in both groups.
Figure 2 and Tables 2 and 3 show results from wave separation analysis, the timing of wave reflections, and carotid augmentation index at rest and during isometric exercise in normotensive and hypertensive subjects. Forward (Fig. 2A) and backward (Fig. 2B) wave amplitudes were significantly greater in the hypertensive group, but did not significantly change with isometric exercise. However, reflection magnitude (Fig. 2C) significantly increased, with a greater increase among normotensive subjects. The change in reflection magnitude among hypertensive subjects did not reach statistical significance (P = 0.47). The time to return of the reflected wave (Fig. 2D) was significantly shorter in hypertensive subjects and became shorter during isometric exercise in both groups (P < 0.0001), indicating an earlier arrival of wave reflections. There was a pronounced increase in the central augmentation index in both groups (Fig. 2E).
Figure 3 and Tables 2 and 3 show brachial and central pressures at rest and during isometric exercise in both groups. Mean arterial pressure (Fig. 3A) significantly increased with isometric exercise, and both baseline values and changes during exercise were essentially identical between the groups. Central pulse pressure (Fig. 3B) was significantly greater in hypertensive subjects and increased during isometric exercise. In contrast, brachial pulse pressure (Fig. 3C) did not significantly change with isometric exercise. Consequently, there was a significant decrease in pulse pressure amplification with isometric exercise (Fig. 3D). Central systolic (Fig. 3E) and brachial systolic (Fig. 3F) pressures, as well as diastolic blood pressure (Fig. 3G), all demonstrated a pronounced increase with the maneuver in both groups.
Figure 4 shows an example of pressure and flow wave forms at rest and during isometric exercise. In this particular case, an 11-mmHg increase in central systolic blood pressure (92–103 mmHg) and a 9-mmHg increase in diastolic blood pressure (49–58 mmHg) were seen, along with a 10-mmHg increase in mean arterial pressure (69–79 mmHg), a pronounced increase in SVR (1,240–1,428 dyn·s·cm−5), but a minimal change in central pulse pressure (43–45 mmHg). There was, however, a clear change in the pressure wave form, with more prominent late systolic peak pressure augmentation and an increase in augmentation index (from 108 to 137), along with an earlier return of the reflected wave (11-ms decrease). There was also a decrease in TAC (from 1.29 to 1.15 ml/mmHg) and an increase in aortic Zc (from 112 to 124 mmHg·ms·l−1).
Relationship between resting indexes and LVMI in hypertensive individuals.
The results of simple linear regression and correlation analyses regarding hemodynamic indexes and LVMI among hypertensive subjects are shown in Table 2. CF-PWV was the strongest correlate of LVMI (β per m/s increase = 2.36; R = 0.63; P < 0.0001). The carotid second systolic peak was also a significant predictor of LVMI (β per mmHg increase = 0.40; R = 0.56; P < 0.0001). The first carotid systolic peak and brachial systolic and diastolic blood pressure also demonstrated significant, but weaker relationships with LVMI. Backward wave amplitude (R = 0.41; P = 0.008) demonstrated a somewhat stronger relationship to LVMI than forward wave amplitude (R = 0.32; P = 0.047), although there was not a significant relationship between reflection magnitude (or carotid augmentation index) and LVMI.
After adjustment for age, sex, and key indexes of hemodynamic load at rest (SVR, aortic Zc, TAC, and reflection magnitude), CF-PWV was the only significant independent predictor of LVMI (β per m/s increase = 2.52; P < 0.0001). There was little change in the value of the estimate for the CF-PWV term compared with the unadjusted estimate, indicating little confounding in this relationship by the other terms in the model. A multivariate model, including CF-PWV and resting hemodynamic indexes, explained 48% of the interindividual variability in LVMI.
Relationship between provoked hemodynamic indexes and LVM.
The results of univariate analyses regarding the relationship of various hemodynamic indexes during isometric exercise and LVMI are shown in Table 4. In univariate analyses, various provoked hemodynamic indexes correlated with LVMI, including SVR, Ea, forward and backward wave amplitude, and both central and brachial systolic blood pressure (Table 4).
For stepwise multiple regression, we used the following candidate predictors of LVMI: sex, age, body weight, resting CF-PWV, and the following key hemodynamic parameters measured at rest and during isometric exercise (as well as terms for the change associated with isometric exercise): TAC, proximal aortic Zc, SVR, reflection magnitude, time to return of the reflected wave, and heart rate. Results of stepwise regression are shown in Table 5 (model 1). Predictors selected by stepwise regression included CF-PWV, TAC during isometric exercise, aortic Zc during isometric exercise, heart rate during isometric exercise, and the change in the time of return of the reflected wave associated with isometric exercise. This model explained 72% of the interindividual variability in LVMI. In this model, CF-PWV (resting value) remained a significant, independent predictor of LVMI (β per m/s increase = 3.61; P < 0.0001). In contrast to resting TAC, TAC during handgrip was a strong independent predictor of LVMI (β per ml/mmHg increase = −17.85; P < 0.0001). Heart rate during isometric exercise and a larger reduction (negative change) in the time to return of the reflected wave (earlier wave reflections) were also significant independent predictors of a higher LVMI. In multivariate analysis, proximal aortic Zc was a significant independent negative predictor of LVMI (β per mmHg·s·ml−1 increase = −150; P < 0.0001).
Given that heart rate during exercise was negatively related to LVMI, we considered the possibility that β-blocker use could underlie this association. We constructed a model in which β-blocker use was added to the predictors shown in Table 5. In this model, β-blocker use was not associated with LVMI (P = 0.33) and did not produce an appreciable change in the regression estimates, the partial R value associated with heart rate at isometric exercise (R = −0.45; P = 0.007), the overall model R2 (0.73), or the overall adjusted model R2 (0.68). Furthermore, we tested this model, including only subjects who were not using β-blockers, and results were essentially identical (not shown).
Finally, since LVMI is a body size-adjusted measurement of LVH and arterial load is also related to body size, we performed further analyses to adjust for body size as a potential confounder. Because relationships between body size and physiological measurements (such as LV mass and arterial load) are frequently nonlinear, we used log-log models (in which variables were transformed to their natural logarithms), which adhere to general allometric relationships. In these models (log), LV mass (rather than LVMI) was the predicted variable, and (log-transformed) predictors included the variables shown in model 1 (Table 5), in addition to body surface area. The results of this model are shown in Table 5 (model 2). In this model, the change in the timing of wave reflections with handgrip was the strongest predictor of LV mass, although resting CF-PWV, TAC during handgrip, and Zc during handgrip were also important independent predictors. Estimates obtained with nonlinear regression were very similar.
In this study, we report on comprehensive analyses of central hemodynamic changes associated with submaximal fatiguing forearm isometric exercise in normotensive and treated hypertensive subjects and demonstrate that hemodynamic load measured during this maneuver strongly predicts LVMI in treated hypertensive adults. The magnitude of this association was far greater than that observed between LVMI and indexes of resting hemodynamic load. Our findings suggest that assessment of central hemodynamics is essential to adequately measure the hemodynamic changes induced by isometric exercise, and that provoked hemodynamic testing may be advantageous to assess dynamic LV afterload in hypertension.
We found that, although there were significant relationships between various hemodynamic indexes and LVMI in univariate analyses, after adjustment for CF-PWV, resting hemodynamic indexes did not significantly predict LVMI. In contrast, indexes of provoked arterial load were selected in stepwise regression and independently predicted LVMI in multivariate models. Our findings differ from those reported by al'Absi et al. (1), in which no relationship between LV mass and brachial blood pressure reactivity was found. Our results are in line with the recent report by Lydakis et al. (16), which showed that submaximal fatiguing forearm isometric exercise is associated with a decrease in pulse pressure amplification, implying that brachial systolic blood pressure is lower relative to central pressures during this maneuver, therefore limiting the usefulness of brachial blood pressure assessments in response to isometric exercise. Indeed, we could not detect a change in brachial pulse pressure during isometric exercise in this study, whereas marked changes occurred in indexes of pulsatile hemodynamic load obtained from analyses of central pressure and flow. It should be noted that arterial pressure is determined by cardiac factors, such as heart rate, the pattern of ventricular ejection, and stroke volume on the one hand, and arterial size, wall properties, peripheral wave reflections, and SVR on the other. Ventricular afterload can only be fully characterized by analysis of pressure-flow relations in the proximal aorta (22, 28). By analyzing central pressure-flow relations, we assessed these different factors independently and their response to isometric exercise. We found that the handgrip maneuver induces an increase in central augmentation index, a decrease in the time to the arrival of the reflected wave, a decrease in TAC, and an increase in SVR, which induced an increase in mean arterial pressure. Interestingly, although an increase in reflection magnitude was seen in many hypertensive subjects, the overall change was less pronounced than the change seen in normotensive adults, indicating that the overall increase in central augmentation index seen in the hypertensive group was predominantly induced by earlier, rather than more pronounced, wave reflections, further illustrating the benefit of assessing both central pressure and flow to study central hemodynamic phenomena. The less pronounced change in wave reflection magnitude among hypertensive subjects was likely due to the effect of vasodilator drugs in muscular arteries. Importantly, although the hypertensive adults we studied had, on average, controlled hypertension (Table 1) and demonstrated values of mean arterial pressure and resistive load (SVR) that were very similar to those seen in normotensive adults (both at rest and during isometric exercise), they demonstrated clear “residual” abnormalities in indexes of resting and provoked pulsatile hemodynamics, such as increased aortic Zc, TAC, earlier wave reflections, and higher (resting) CF-PWV.
Although significant changes in both resistive and pulsatile hemodynamic load were seen with isometric exercise among hypertensive subjects, only indexes related to pulsatile load were selected as independent predictors of LVMI. In this model, TAC emerged as a strong, independent predictor of LVMI. There was also a significant independent relationship between earlier wave reflections during handgrip and LVMI. This is an interesting fact, given that the hypertensive group did not demonstrate significant exercise-induced increases in reflection magnitude, indicating the importance of wave reflection timing on hemodynamic load. Earlier wave reflections appear to constitute an important abnormality that “persisted”, despite current standard therapy in our group of treated (and, on average, controlled) hypertensive subjects, as indicated by the significantly shorter time to return of the reflected wave seen in hypertensive subjects in our study (Fig. 2D). In line with these findings, hypertensive subjects had significantly increased CF-PWV, a marker of large-artery stiffness that was strongly related to LVH in this population. Similarly, earlier wave reflections during isometric exercise were likely induced by an increase in CF-PWV during the maneuver, as a result of increased mean arterial pressure, which further increases “effective” stiffness of the arterial wall by recruiting stiffer collagen fibers as distending pressure increases (22). Our findings regarding changes in the characteristics of the pulse wave, which suggest an increase in large-artery stiffness with the handgrip maneuver, are consistent with previously reported data (17). In addition to changes in distending pressure, it is possible that short-term regulation in arterial wall stiffness occurs during this maneuver via neurohormonal mechanisms or endothelial regulation. Interestingly, the timing of wave reflections during isometric exercise was the stronger predictor of LVMI when body size was accounted for. This is explained by the fact that, for any given change in CF-PWV with isometric exercise, the change in the timing of wave reflections increases with increasing body height (due to more distant reflection sites). Since LVMI is computed by dividing LV mass by body height2.7, increasing body height decreased LVMI in this population, obscuring the body size-independent relationship between timing of wave reflections and LVH. This relationship was best characterized when body size was accounted for, illustrating the importance of accounting for body size in studies of arterial hemodynamics.
Finally, we found a negative relationship between proximal aortic Zc and LVMI. These findings should be carefully interpreted in the context of the multivariable model. It is worth noting that this model included CF-PWV at rest and time to return of the reflected wave during isometric exercise (which is a close correlate of CF-PWV during exercise). CF-PWV is a measure of large-artery stiffness and is mildly dependent on aortic diameter. In contrast, aortic Zc is moderately dependent on wall stiffness and highly dependent on aortic diameter (19). Therefore, the presence of CF-PWV in the model likely “adjusted out” the stiffness component, leaving aortic diameter as the predominant determinant of the relationship between Zc and LVMI. In hypertensive subjects, Zc increased modestly with isometric exercise, despite the pronounced increase in mean arterial pressure (which increases effective wall stiffness), suggesting that passive distension of the proximal aortic wall during the maneuver may have played a role in preventing a more marked increase in aortic Zc in the provoked state. This distension may occur due to increases in mean arterial pressure during the maneuver, although it is also possible that forceful LV ejection from hypertrophic ventricles contributed to proximal aortic wall distension during isometric exercise.
A lower heart rate during the isometric exercise was negatively associated with LVMI. Importantly, this association was independent of hemodynamic load and β-blocker use. Our findings are in agreement with previous data demonstrating an association between LVH and chronotropic incompetence (13), which was likely responsible for lower heart rates during handgrip in subjects with higher LVMI.
CF-PWV measured at rest was a strong, independent predictor in multivariable models that included resting and provoked hemodynamic load. CF-PWV has been consistently shown to be an adverse, independent prognostic indicator (14, 18, 33), and its effects on cardiovascular risk are thought to be at least partially mediated through its role in central pressure pulsatility and LV afterload. As stated above, CF-PWV is a measure of large-artery stiffness, which also affects the timing of wave reflections and TAC, both of which were associated with LVMI. Given that arterial stiffness is affected by both long-term structural changes (22) and short-term regulation by mediators such as nitric oxide (26), many factors can influence CF-PWV, including age, chronic mechanical stress, extracellular matrix remodeling, neurohormonal abnormalities, metabolic abnormalities, and obesity, as previously reviewed (25). Therefore, in addition to hemodynamic load, multiple biological pathways may mediate the association between CF-PWV and LVH in hypertension.
Our study included a clinical hypertensive population. As such, these individuals demonstrated various co-morbidities and were receiving conventional medication regimens. As a result of treatment, average systolic and diastolic blood pressures were not in the hypertensive range. Although medication use may have affected “native” abnormalities that determine LVH in untreated hypertension (such as reflection magnitude), our study addressed the clinically relevant question of residual abnormalities in conventionally treated patients, which, in turn, impact LV mass, an important determinant of outcomes in this population of patients. We also acknowledge that our population of hypertensive subjects had only mildly increased LVMI compared with our reference normotensive subjects, who underwent measurements of LV mass with identical methods; however, this is not surprising for a population of controlled hypertensive subjects. Furthermore, our population did include subjects with moderate and severe LVH. We acknowledge that systematic underestimations in LV mass measurements can occur with the echocardiography-based area-length calculations due to inaccurate assessments of true ventricular length. Nevertheless, this should not affect the assessment of the continuous relationship between LVMI and arterial load in our study population.
Our findings indicate that large-artery wall stiffness is a key determinant of residual LV mass in this population. It can be anticipated that therapies that improve large-artery wall stiffness would favorably impact CF-PWV, the timing of wave reflections, aortic Zc, and TAC.
Our study is limited by its cross-sectional nature and the relatively small sample size, which is powered to detect only quantitatively important associations. To achieve high-quality pressure-flow measurements during isometric exercise, we chose not to measure CF-PWV during handgrip. Strengths of this study include the carefully made prospective measurements and offline analyses of arterial load and LV mass aided by 3D echocardiography, which likely played a role in the strength of our predictive models. Detailed hemodynamic testing with 3D echocardiography and arterial tonometry is highly feasible to accomplish an accurate and detailed mechanistic understanding of arterial hemodynamics in clinical research settings.
I am not aware of financial conflict(s) with the subject matter or materials discussed in this manuscript with any of the authors, or any of the authors' academic institutions or employers.
- Copyright © 2010 the American Physiological Society