INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

ONE OF THE MOST DEBILITATING effects of primary aging is the progressive decline in aerobic exercise capacity. This functional decline, which is due to primary and secondary aging, often causes sedentary, very old (>78 yr) individuals to become frail. Primary causes of this decline are the age-related decreases in cardiovascular function and reserve. Although it is well known that the decrease in peak exercise cardiac output in the elderly is in part due to a decrease in peak heart rate, less well-known is the decline in peak exercise stroke volume (SV) in the elderly (20). There are several adaptive structural and functional changes in the ventriculoarterial tree that may impact peak exercise SV, cardiac output, and therefore peak O₂ consumption (O_2 peak) in the elderly. It is appropriate to correct peak exercise SV for lean body mass (LBM) because, during exercise, a high percentage of the SV is delivered to the skeletal muscle (15). The identification of the age-related changes in the ventriculoarterial tree that are potentially predictors of peak exercise SV corrected for LBM in very old persons would 1) improve our understanding of the mechanisms by which peak exercise SV is regulated and 2) provide a basis for designing interventions for improving peak exercise SV and hence improve aerobic exercise capacity. Thus the main purpose of this study was to identify those age-related changes in the ventriculoarterial system that may influence peak exercise SV/LBM in very old individuals. We hypothesized that, among the major cardiovascular changes in the octogenarians, altered left ventricular (LV) filling, as reflected in a decreased early-to-late transmitral flow velocity ratio (E/A) in this population, and LV concentric remodeling, as shown be an increased LV posterior wall thickness-to-LV radius ratio (h/r), would be independently associated with an improved peak exercise SV and that increased arterial stiffness, as demonstrated by an increased augmentation index (AI), would be associated with decreased peak exercise SV.

A secondary aim was to determine the adaptations to aging that most influence LV remodeling in the elderly because mild LV remodeling is likely to be an important adaptive response to age-related changes in the arterial tree and may affect LV function. Indeed, in hypertension, LV remodeling is an important adaptive response, and in experimental models in which LV remodeling response to pressure overload is prohibited, LV contractile function decreased (44).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects. We studied 42 women and 10 men (n = 52), 82.4 ± 3.7 (SD) yr old, selected from a pool of 258 subjects (78-95 yr of age) who had volunteered to participate in the Washington University Claude D. Pepper Older American Independence Center research studies. The selection criteria were 1) good quality echocardiographic windows and carotid pulse tracings; 2) no evidence of carotid disease, carotid bruit, or a history of a cerebrovascular accident; 3) normal sinus rhythm without frequent premature ventricular or supraventricular depolarizations or other dysrhythmias; 4) absence of any of the following clinical conditions that are associated with LV remodeling: myocardial infarction, valvular heart disease, mitral or aortic valve replacement, coronary artery bypass graft surgery, or congestive heart failure; 5) no symptoms suggestive of active cardiac disease (e.g., dyspnea, angina); 6) nonsmoking; 7) sedentary lifestyle (as assessed by physical performance testing and a telephone questionnaire); and 8) ability to exercise sufficiently long and vigorously to make possible measurement of cardiac output during peak exercise. The physical performance testing used in this study to ensure the sedentary and mild-to-moderately frail status (with scores between 18 and 32) of the patients has been validated and used previously (7, 10). Thirty-two subjects had a history of hypertension. Because of some subjects' inability to undergo some measurements, all subjects did not undergo all tests. Of the subjects, 41, 42, 51, and 51 had a reliable E/A, SV/LBM, h/r, and LV mass/LBM, respectively. All subjects signed an institutional review board-approved written consent form before enrolling in this study.

Assessment of arterial stiffness by measuring AI. We used applanation tonometry of the common carotid artery with the use of the high-fidelity Millar (model TCB-500, Millar Instruments, Houston, TX) external transducer. The arterial waveform and its harmonics obtained from this high-fidelity probe closely mimic those recorded from the central aorta (17, 18). The Millar probe was placed perpendicularly to the skin overlying the common carotid artery to record carotid arterial pulse waveforms. The subjects rested in the recumbent position for a minimum of 15 min before undergoing the applanation tonometry. Carotid arterial pulse waves were displayed on an oscilloscope, and, after a satisfactory pressure wave contour was obtained, the carotid pulse signals were recorded and stored for subsequent analysis by using the method described by Vaitkevicius et al. (41). Ten recordings of the carotid pulse wave and ECG signals were made. Blood pressure (BP) was recorded simultaneously with the applanation tonometry with the use of an automated noninvasive BP recording device (model 1846SXv). A computer software program was used to determine the timing and the amplitude of the inflection point of the signal-averaged carotid pulse waveform by using its derivatives. The AI was calculated as (Pi/PP) × 100, where Pi is the amplitude of the inflection point and PP is amplitude of the pulse pressure, with the use of the computer-generated algorithm (17, 18, 41). The coefficient variation for the applanation tonometry was 10%.

Assessment of LV geometry and concentric remodeling. Immediately after completion of the applanation tonometry studies, the subjects underwent two-dimensional echocardiographic and Doppler studies. LV structure, geometry, and function were evaluated with the use of two-dimensional guided M-mode echocardiography (model 2000 Hewlett-Packard). Left ventricular end-diastolic (EDD) and end-systolic (ESD) diameters, h, and septal thickness as well as fractional shortening were measured and calculated according to the published guidelines (33). The value of h/r was also calculated to assess LV remodeling; h/r values 0.45 were considered indicative of LV concentric remodeling (31). LV filling was evaluated by the transmitral Doppler velocity profile and the E and A were measured, and the E/A was calculated. End-systolic wall stress (_es) was calculated by using the formula described by Grossman et al. (16): _es = Pr/2h (1 + h/2r), where P is end-systolic pressure (expressed in g/cm²), r is end-systolic radius (ESD/2), and h is end-systolic posterior wall thickness. End-systolic pressure was estimated using the equation: ESP = (2 × SBP + DBP)/3, as described by Kelly et al. (17, 18). LV mass was calculated with the use of equation by Devereux et al. (13). The LV mass was also normalized to total LBM. All of the echocardiographic and applanation tonometry data were analyzed in a blinded fashion. The reproducibility for the echocardiographic data has been resported recently (40).

Cardiac output. Two or three weeks after determination of O_2 peak, peak exercise cardiac output was determined noninvasively with the use of the acetylene-rebreathing method during a separate maximal treadmill exercise test, as described previously (26, 39). Peak exercise SV was calculated as peak exercise cardiac output/peak exercise heart rate.

Body composition. LBM was estimated with the use of the dual-energy X-ray absorptiometry (Hologic, QDR-1000/W), as described by Kohrt et al. (21).

Aerobic power. O_2 peak was determined during a treadmill exercise test. After 3-4 min of warm-up during which the subjects walked on the level with a speed ranging from 0.5 to 1.2 miles/h, depending on the subjects' clinical status, they continued to walk at the warm-up speed, and the grade was increased every 2 min by 1-2% increments.

Statistics. We used univariate linear regression analyses followed by best-subsets multiple-regression analyses to determine the variables contributing to 1) LV concentric remodeling (h/r) and LV mass and 2) peak exercise SV. The data that were not normally distributed were analyzed with the nonparametric Mann-Whitney rank-sum tests. The data are given as means ± SD. A P value of < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The physiological and clinical characteristics of the subjects are summarized in Table 1. As aforementioned, we studied sedentary elderly subjects (mean age = 82 yr) within a narrow age range. The subjects, on average, were not obese (average body mass index <30 kg/m²) and had SBP readings that bordered on hypertension, as would be expected in an unbiased sampling of the very elderly. Table 2 lists the classes of medications used by the subjects and the percentage of subjects taking a particular class of medication. Of the total cohort, 9% of subjects were taking -adrenergic-blocking agents, 25% were taking calcium channel blockers, and 19% were taking angiotensin-converting enzyme inhibitors or angiotensin-receptor blockers for hypertension.

We performed O_2 peak testing to document that our subjects were sedentary individuals. The range of O_2 peak achieved was 11.2-21.3 ml · kg¹ · min¹ or 0.60-1.57 l/min, which is a a fairly narrow range that would be expected of sedentary octogenarians. O_2 peak correlated with peak exercise cardiac output (r = 0.67, P < 0.001), peak heart rate (r = 0.46, P = 0.002), EDD (r = 0.45, P = 0.003), and the h/r (r = 0.32, P = 0.038) but not with the AI, SBP, or LV mass. Multiple-regression analysis showed that, among the above variables, only peak exercise cardiac output contributed significantly to O_2 peak (R² = 0.55, P = 0.002), accounting for greater than half of the variations in O_2 peak.

To test our hypothesis that LV filling, concentric LV remodeling, and arterial stiffness were related to peak exercise SV, we first performed univariate analyses. LV filling (as measured by E/A), LV remodeling (as reflected in the h/r and LV mass/LBM), and the BSA correlated significantly with peak exercise SV normalized for LBM (Figs. 1 and 2, Table 3). Specifically, the E/A was significantly and negatively correlated with SV/LBM (r = 0.37, P = 0.03); that is, that the higher the E/A, the lower the SV/LBM. However, the AI did not correlate significantly with peak exercise SV.

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Relationship between left ventricular (LV) wall thickness-to-LV radius ratio (h/r) and stroke volume (SV)/lean body mass (LBM). Data are from 41 subjects. r = 0.31; P = 0.046.

View larger version (9K): [in this window] [in a new window]   Fig. 2.   Relationship between early-to-late transmitral flow velocity ratio (E/A) and SV/LBM. Data are from 34 subjects. r = 0.38; P = 0.031.

In a multiple-regression analysis, LV remodeling was a less important predictor of peak exercise SV than was LV filling. Only the E/A was an independent predictor of peak exercise SV/LBM (P = 0.03). Inclusion of parameters of LV remodeling, EDD/LBM, or LV mass/LBM did not add to the ability of the model to predict peak exercise SV accurately.

To test our secondary hypothesis that AI is related to LV remodeling (as measured by the LV mass/LBM) in very old sedentary persons, we first performed univariate analyses. The AI correlated significantly with multiple measures of LV structure and concentric remodeling: LV septal wall thickness (r = 0.41, P = 0.003), LV h (r = 0.58, P < 0.001), h/r (r = 0.52, P < 0.0001; Table 3 and Fig. 3A), and LV mass expressed either in absolute terms (r = 0.32, P = 0.022) or normalized for LBM (LVM/LBM; r = 0.42, P = 0.002; Table 3 and Fig. 3B). The SBP and DBP also correlated significantly with LVM/LBM (Table 3), but multiple-regression analysis showed that of the three measures related to arterial stiffness (SBP, DBP, and AI), only AI was an independent predictor of LV mass/LBM. The addition of resting or exercise SBP or DBP did not add to the ability of the model to predict LV mass/LBM.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   A: relationship between the augmentation index (AI) and h/r. r = 0.52; P < 0.0001. Data are from 51 subjects. B: relationship between the LV mass (LVM)/LBM and AI. Data are from 51 subjects. r = 0.42; P = 0.002.

As aforementioned, h/r correlated with AI. The h/r also correlated with LBM (r = 0.37, P = 0.008), height (r = 0.30, P = 0.021), and BSA (r = 0.31, P = 0.028) but not with resting SBP or DBP. Multiple-regression analysis showed that, of the variables that were significantly related to concentric remodeling, only AI (R² = 0.36, P < 0.001) was an independent predictor of concentric remodeling, as expressed in the h/r.

Because LV _es may have an impact on the ability of the LV to generate SV, we also examined the relationships between LV _es and LV remodeling, arterial stiffness, and peak exercise SV/LBM. We found that LV _es correlated with SBP (r = 0.36, P = 0.026) but not with AI. LV _es correlated with LV remodeling as measured by LV mass (r = 0.44, P = 0.002) and inversely with h/r (r = 0.51, P < 0.001) and LV systolic shortening (r = 0.57, P < 0.001). However, LV _es did not correlate significantly with peak exercise SV/LBM. Multivariate-regression analysis showed that SBP (P = 0.005), the h/r (P < 0.001), and LV mass (P < 0.001), were all independent predictors of LV _es (R² = 0.62).

Because several of the subjects in this study (n = 28) were on vasoactive medication (including diuretics) and because these medications can have an effect on ventriculoarterial coupling, we reanalyzed predictors of SV/LBM, h/r, and LVM/BM after excluding the subjects on vasoactive medications. Although the r value describing the relationship between E/A and SV/LBM did decrease to 0.55, the P value increased to 0.05. The other significant relationships between h/r and SV/LBM, between AI and h/r, and between AI and LVM/LBM were not significant after the exclusion of the 28 subjects on medication, probably because of the smaller residual sample size.

Effect of gender. Because more women than men were studied, we compared the baseline characteristics, exercise hemodynamics, and echocardiographic data of both genders. There was no significant difference in age between the men and the women tested. However, the women had a significantly lower body weight (65 ± 11 vs. 72 ± 11 kg; P < 0.05), LBM (36 ± 7 vs. 51 ± 5 kg; P < .001), BSA (1.65 ± 0.15 vs. 1.80 ± 0.19 m²; P < 0.005), and height (158 ± 6 vs. 166 ± 8 cm; P < 0.001). There were few significant differences between the genders with regard to most of the echocardiographic and exercise hemodynamic variables. The _es was lower in the women (47.3 ± 21.5 vs. 63.2 ± 19.1 g/cm²; P = 0.003), but EDD/LBM (1.3 ± 0.2 vs. 1.1 ± 0.2 mm/kg; P < 0.001) and cardiac output/LBM were larger in the older women (226 ± 51 vs. 181 ± 52 ml · min¹ · kg¹; P < 0.05). The larger LV mass/LBM in the older women was of borderline significance (P = 0.06).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main finding of this study is that LV diastolic filling, as reflected in the E/A, was the only independent predictor of peak exercise SV/LBM. E/A correlated negatively with peak exercise SV. This finding is supported by other studies that suggest that, as part of normal aging, the peak velocity of the "E" wave should decrease and that of the "A" wave should increase (19, 24, 34-37). However, many studies on the effect of aging on E/A did not include or focus on octogenarians (5, 19, 34, 35, 37). Thus our findings extend to octogenarians the concept that as one ages, E/A should decrease progressively and, furthermore, that LV diastolic function is an important predictor of SV.

The decrease in the E-wave height relative to the A- wave height with aging represents a decrease in the velocity of the blood flow from the left atrium to the LV during early diastole and an increase in the velocity of the blood flow during atrial contraction. This is thought to be due to a general increase in LV stiffness with aging, resulting in diminished LV filling (1). Our study suggests that compromised early LV filling is compensated partially by an increased contribution during late diastole (i.e., a higher A wave and a lower E/A), which preserves SV. Indeed, our study supports the notion that the link between LV diastolic function and LV systolic function, which is due to its constant-volume pump attribute, exists in the elderly (22, 23). Our finding that LV filling is an important determinant of peak exercise SV/LBM is consistent with other studies performed in younger subjects that showed reliance of exercise SV on end-diastolic volume or blood volume in the pretrained state (8, 36, 42). Furthermore, a previous study showed that there is an age-related increase in end-diastolic volume and SV during peak exercise (28).

Our data suggest that, even though the subjects did not have the diagnosis of heart failure, a higher E/A in some subjects may have been indicative of a "pseudonormal" pattern, which is associated with higher LV end-diastolic filling pressures (25). This is concordant with the clinical outcomes of the Cardiovascular Health Study and Strong Heart Study, in which a high peak Doppler "E" velocity and a higher E/A were independent predictors of incident heart failure and mortality (2, 16). These studies support the notion that a high E/A is not normal in octogenarians (2).

The effect that hypertension and medications may have had on our findings is not clear. We analyzed the data from this study after excluding all subjects who were taking vasoactive medications; however, there were too few subjects left to make firm conclusions on how hypertension and/or medications may have influenced the results. Some medications that improve LV filling, such as verapamil, have been shown to have a salutary effect on aerobic exercise performance in the elderly and thus may be considered as potential therapy for elderly persons with decreased exercise tolerance and diastolic dysfunction (12).

Another approach for the enhancement of exercise capacity in the elderly is exercise training. Others have shown that in younger subjects exercise training is associated with decreased arterial stiffness and a higher O_2 max (41, 43). Exercise is known to increase aerobic power in older individuals (6). Studies to prove that training has salutary effects on LV filling, arterial stiffness, or LV structure in octogenarians are necessary.

Although it was somewhat surprising that AI was not a predictor of peak exercise SV/LBM, one possible explanation is that we included subjects who had a history of hypertension, some of whom were treated with vasoactive medications which have differential effects on arterial tone and LV function (8, 10, 12). The study by Tanaka et al. (38) in younger subjects not on vasoactive medications other than hormone replacement therapy (ages 21-78 yr) showed that there was a significant negative relationship between AI and resting SV (38). Prospective studies on the effect of each particular drug on AI and ventriculoarterial coupling as they relate to peak exercise SV in octogenarians would be helpful in assessing the possible confounding effect of drug treatment on our study's results.

We found that increased arterial stiffness in very elderly individuals, independent of SBP or DBP, contributes to concentric LV remodeling and increased LV mass. This observation is consistent with previous studies that reported a strong association between increased arterial stiffness and LV hypertrophy in both hypertensive and normotensive subjects (8, 27, 29, 30). In addition, our data suggest that LV concentric remodeling can favorably influence SV during peak exercise, showing that the older subjects with a greater h/r and higher LV mass/LBM are more likely to have a higher SV during peak exercise, although neither measure of concentric remodeling was an independent predictor of peak exercise SV. The apparent greater influence of arterial stiffness in inducing LV concentric remodeling than that of SBP in our subjects may seem inconsistent with an in vitro study that demonstrated that increased cardiac myocyte protein synthesis and hypertrophy are mediated directly by mechanical stretch and systolic overload (32). One possible explanation to our finding is that a single casual recording of blood pressure may not adequately reflect sustained elevation of BP required for pressure overload remodeling. Twenty-four-hour BP monitoring might have been more predictive of LV remodeling in our subjects.

Although increased arterial stiffness is a universal manifestation of primary aging (3, 4, 17, 42), and in other studies has been shown to be inversely related to O_2 peak, in our study AI was not significantly associated with O_2 peak (11, 41). The absence of a significant correlation between the AI and O_2 peak in this study is most likely explained by the fact that all of our subjects were very elderly and sedentary with low O_2 peak values that varied over only a narrow range (i.e., 11.2-21.3 ml · kg¹ · min¹).

The increase in LV mass and concentric remodeling in response to higher arterial stiffness can, by reducing LV _es, help prevent a significant fall in SV. However, the benefit of this adaptation can be offset by concurrent and severe restriction of LV filling as evidenced by the pseudonormalized E/A, particularly during exercise. Therefore, in a subject with concentric LV remodeling, the magnitude of increases of cardiac output and SV during exercise, at a given preload and contractile state, is determined by the interplay of these opposing factors. Our data suggest that, in general, mild compensatory LV remodeling can, within limits, play a beneficial role in maintaining SV during peak exercise in the elderly. This may explain why, in this study, we observed a relationship between AI and measures of LV remodeling (h/r and LV mass/LBM) but not between AI and peak exercise SV. This observation does not necessarily contradict the results from previous studies that aging results in a significantly lower cardiac output (21, 26) and SV (26) during maximal exercise. Our data only suggest that mild adaptive LV remodeling is not necessarily detrimental and may be even useful in limiting the extent of fall in SV during exercise in those older subjects who have modest LV hypertrophy and concentric remodeling without a concomitant marked reduction in LV diastolic filling.

Conclusion. Our findings suggest that the single best predictor of peak exercise SV/LBM was the E/A. Thus, in contrast to young persons without heart failure who have a high E/A, a high E/A in sedentary very old persons may not be normal and, in fact, predicts a decreased peak exercise SV/LBM. We also found that increased arterial stiffness, reflected in an increased AI, is an independent predictor of LV concentric remodeling and LV mass in the elderly. The influence of arterial stiffness on LV remodeling appears to be significantly greater than that of SBP or DBP. Mild concentric LV remodeling induced by increased arterial stiffness appears to be a useful but limited adaptation contributing to a higher SV during peak exercise in the elderly subjects, probably because of a lower LV _es.