Impact of Arterial Load and Loading Sequence on Left Ventricular Tissue Velocities in Humans

Study population and design

Fifty subjects were recruited from the community to reflect a balanced mixture of ages and hypertension history, requiring 5 to 15 subjects per decade and encouraging hypertensives to enroll via advertisement. Subjects with heart failure, atrial fibrillation, carotid or valvular disease, inadequate ultrasound images, mitral annular calcification, and Raynaud’s phenomenon were excluded. Informed consent was obtained from all subjects, and the protocol was approved by the Joint Committee on Clinical Investigation of the Johns Hopkins Medical Institutions.

Subjects rested in a quiet room in the supine position for >10 min to reach stable baseline. Carotid-femoral pulse wave velocity (PWV) and carotid augmentation index (cAI) were determined using a multiarray applanation tonometry system (VP 2000, Omron Healthcare, Banockburn, Illinois); signals were digitized at 1.2 kHz and saved for off-line analysis. Comprehensive echo Doppler study with color-coded TDE was performed (Vivid7, GE Healthcare, Chalfont St. Giles, United Kingdom), followed by repeat carotid tonometry using a hand-held tonometer (Millar Instruments, Houston, Texas) with simultaneous flow velocity assessment (Parks Instruments, Aloha, Oregon) and radial tonometry (Omron Healthcare). Heart rate was continuously recorded, with flow and pressure tracings gated to the electrocardiogram.

Cardiac function analysis

All measurements represent the mean of ≥3 beats. Color-coded TDE data were obtained using the Octave strain rate preset (Vivid7) with high frame rate (>150 Hz) maintained by limiting the image size to include LV septal and lateral walls in the apical 4-chamber view. Peak mitral annular systolic (S′) and diastolic early (E′) and late (A′) tissue velocities were measured by placing a 4 × 4 mm region of interest in the midmyocardial area of the basal LV septum, 10 mm apical to the medial mitral valve annulus. Left ventricular ejection fraction (EF), dimensions, volumes, and wall stress were obtained using standard methods (25). Volumetric, flow, and resistive data were indexed to body surface area. Early (E) and late (A) transmitral filling velocities and stroke volume (SV) were determined by pulse-wave Doppler. Cardiac output was the product of SV and heart rate. Cardiac contractility was assessed by EF as well as by load-independent indexes: peak power index (peak ejection rate × systolic blood pressure [SBP]/end-diastolic volume), estimated end-systolic elastance (Ees) (0.9 × SBP/end-systolic volume [ESV]), and stroke work index (mean blood pressure × SV/EDV) (26,27).

Vascular function analysis

Oscillometric brachial BP was measured (Dinamap, Tampa, Florida), with mean and diastolic used to calibrate central pressures from carotid tonometry which were used for analysis. Total afterload was defined by the effective arterial elastance (Ea = 0.9 × SBP/SV) (28) (Fig. 1).Systemic vascular resistance index (SVRI), the mean (nonpulsatile) component of afterload, was determined by dividing mean arterial pressure and cardiac index, converted to standard units. Vascular stiffness was quantified by PWV, determined from the time delay between the foot of pressure upstroke at carotid and femoral arteries (VP2000, Omron Healthcare) divided by the surface distance between sampling. Late-systolic load was quantified by cAI and radial augmentation index (rAI), determined from zero-crossing of the fourth pressure derivative (21) from digitized carotid/radial waveforms processed by custom software, averaged from 15 to 30 consecutive steady-state beats (29). Zero-crossings occurring after peak systolic pressure were assigned negative values (21). Carotid pressure tracings were analyzed by integrating the area under the central pulse-pressure curve (pressure–time integral), with the last one-third (PTI3) being used as a separate measure of late-systolic load. Simultaneous carotid pressure and linear flow velocity data were transformed to Fourier series to determine input impedance. Carotid characteristic impedance (Zc), determined from the mean impedance modulus from 2 to 12 Hz, was used to assess proximal load. Carotid Zc served as a surrogate for aortic Zc, with velocity rather than volumetric flow; thus units are cm⁻³. Total arterial compliance (Ca) was estimated from the ratio of SV to pulse pressure.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Components of Arterial Load and Loading Sequence

(A)Net/total afterload is effective arterial elastance (Ea), the ratio of left ventricular (LV) end-systolic pressure (ESP) to stroke volume (SV). The red dotted linesshows an increase in Ea. (B)Characteristic impedance (Zc) is the mean of impedance moduli at harmonics 2 to 12 Hz. The Zc is elevated with aging and hypertension. (C)Late-systolic load is determined by the ratio of the augmented pressure (AP) to pulse pressure (PP): the augmentation index (AI). (D)The last one-third of the area under the pulse pressure-time integral curve (PTI3) is another measure of late-systolic load.

Statistical analysis

Data are expressed as mean ± SD. Differences within and between groups were compared by paired and unpaired ttests, respectively. Linear univariate and multivariate regression analysis was performed to test relations between hemodynamic load parameters and LV tissue velocities, assuming that relations were linear and that dependent variables were normally distributed for each independent variable. Multivariate modeling was determined using the a priori hypothesis that E′, E/E′, and S′ were related to given components of arterial afterload based upon a 3-element Windkessel model (30) (Zc, SVRI, Ca), cAI, and preload (EDVI). Regression β-coefficients (parameter estimates) are expressed in units of SD from the mean for each parameter (z-standardized) to allow for comparison of the magnitude of each relationship.

Load associations with early relaxation

Figure 2displays the relationships between components of arterial afterload and E′, which correlated inversely with net afterload (Ea: r = −0.42; β = 1.04; p = 0.003) as well as proximal (Zc: r = −0.47; β = 1.2; p = 0.002) and mean/nonpulsatile loads (SVRI: r = −0.44; β = 1.1; p = 0.004) and PWV (r = −0.46; β = 1.04; p = 0.004). Stronger associations were observed with indexes of arterial stiffness, wave reflection, and late-systolic load, assessed by total arterial compliance (r = 0.55; β = 1.4), cAI (r = −0.64; β = 1.6), rAI (r = −0.62; β = 1.6), and PTI3 (r = −0.62; β = 1.6; all p < 0.0001) (Fig. 3).E′ did not correlate with preload volume (EDVI: p = 0.1) or mean LV wall stress (p = 0.6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Relationships between E′ and Total, Nonpulsatile, and Early/Proximal Arterial Load

The tissue Doppler echocardiography early-diastolic velocity (E′) varies inversely with mean (systemic vascular resistance index [SVRI]), net (arterial elastance [Ea]), and early (characteristic impedance [Zc]) left ventricular afterload, as well as aortic stiffness (carotid-femoral pulse wave velocity [PWV]). Dotted linesshow 95% confidence bands for the regression slopes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Relationships Between E′, Compliance, and Late-Systolic Load

The association between tissue Doppler echocardiography early-diastolic velocity (E′) and afterload is strongest for arterial compliance and late-systolic load (carotid augmentation index [cAI], radial augmentation index [rAI], and last tertile of central pulse pressure-time integral [PTI]), suggesting that these factors may particularly affect early relaxation. Dotted linesshow 95% confidence bands for the regression slopes. Ca = total arterial compliance.

Multivariate analysis was then performed incorporating components from the 3-element Windkessel model, with cAI as a measure of wave reflections (adjusted R²= 0.50; p < 0.0001). Of the individual parameters, only cAI (p = 0.002) and Zc (p = 0.04) remained significant predictors of E′. Importantly, these relationships remained significant even after adjusting for age (both p < 0.05). The combination of cAI, Zc, and age explained 70% of the variability in E′ across the study population (adjusted R²= 0.69; p < 0.0001). The presence or absence of hypertension did not influence the relationship between load and E′.

E did not vary with afterload (not shown), such that the E/E′ ratio varied directly with afterload (Table 2).Similarly to E′, the relationship was most marked for pulsatile load.

Arterial afterload and systolic tissue velocity

S′ also varied inversely with arterial afterload and directly with compliance (Table 3).Mean, net, and late-systolic components of LV afterload showed similar magnitudes of association, but in multivariate analysis only arterial compliance remained significant. S′ varied weakly with preload volume (EDVI: r = 0.35; p = 0.01) but not with mean wall stress (p = 0.45). S′ did not directly correlate with any other parameter of contractility, including peak power index, stroke work index, Ees, and EF (all p > 0.3), although each of the latter indexes correlated well with one another (r = 0.5 to 0.9; all p < 0.001).

Arterial load and atrial function

The TDE late-diastolic velocity (A′), a measure of atrial contractile function (31), did not significantly correlate with load, nor did it vary with age or hypertension (data not shown). However, A correlated directly with afterload (Table 4).This association was strongest with indexes of late-systolic load and wave reflections, and in multivariate analysis only cAI remained significant (p = 0.003), consistent with the notion that increased late-systolic load may induce greater impairment of diastolic relaxation, reducing early filling and increasing the role of atrial filling. The E′ was inversely associated with A-wave velocity (r = −0.61; p < 0.0001).

Effects of hypertension

Table 5contrasts hypertensive and nonhypertensive subjects. As expected, hypertensive subjects had lower arterial compliance and elevated afterload, particularly late-systolic load, and this was associated with lower E′. Intriguingly, systolic tissue velocities were also lower in the hypertensive group, whereas nonload-dependent parameters of contractility, such as peak power index, Ees, and stroke work index, were higher. The higher contractility indicated by the latter measures was matched to the higher net afterload in the hypertension subjects (i.e., the ventricular-arterial coupling ratio, Ea/Ees, was similar), such that the EF was nearly identical in each group. When S′ was indexed to Ea (product of S′ and Ea), the apparent decrease in contractility suggested by lower S′ in the hypertensive group was no longer present (8.4 ± 2.1 cm·mm Hg/s·ml vs. 8.8 ± 2.4 cm·mm Hg/s·ml for hypertensive vs. normotensives; p = 0.6).

Characterization of arterial afterload

Afterload, wave reflections, and vascular stiffness all increase with aging (21,23,32–34). Associated changes in the material properties of the cardiovascular system confer increased risk of cardiovascular disease (35). In patients with hypertension, diastolic dysfunction, renal disease, and HFpEF, this increase is exaggerated (23,36). Vascular stiffening increases the resistance to pulsatile blood flow in the proximal aorta. Characteristic impedance increases with age (33) and hypertension (22) and is considered a primary determinant of proximal “early” LV afterload. Arterial compliance decreases with aging, whereas mean systemic vascular resistance tends to remain stable (32). During systole, as the incident pressure wave encounters zones of impedance mismatch (e.g., arterial bifurcations), part of the wave is reflected backward. This reflected wave sums with the incident (forward) wave, augmenting net pressure. With vascular stiffening, the pulse wave travels faster (increased PWV), such that the reflected wave returns to the heart in late systole, increasing “late” afterload (21). Thick-thin myofilament interactions may be altered by the increase in late-systolic load, such that crossbridge dissociation during isovolumic relaxation becomes impaired (6).

Arterial load modulation of relaxation

Relaxation can be quantified by the time constant of LV pressure decay, tau (3), yet this requires invasive measurement. Recently, relaxation has been assessed noninvasively using diastolic E′ (14–16). E′ correlates inversely with tau and is relatively insensitive to LV filling pressures, particularly when relaxation is prolonged (16,17). Animal studies have shown acute afterload modulation of relaxation (5–9), although the magnitude of this effect remains disputed (3). Increases in aortic pressure slow relaxation, and increases in late-systolic load appear more deleterious than those imposed earlier (5,7,8).

Human data are less abundant and primarily drawn from small acute intervention studies. Pharmacologically induced changes in aortic pressure did not affect relaxation kinetics in one study (13), but more recently Iketani et al. (11) used angiotensin and nitroglycerin to vary afterload and found that tau was inversely correlated with AI but not absolute aortic pressure. Yano et al. (12) generated an augmented reflected wave in young and older patients undergoing catheterization. In younger subjects, the reflected wave returned during diastole, augmenting coronary flow, associated with improved relaxation. In older subjects, the reflected wave returned earlier (in late systole) and was associated with prolonged relaxation. Kawaguchi et al. (10) showed that HFpEF patients display elevated vascular stiffness and that acute increases in load with stress may further impair relaxation. Indeed, conduit vessel stiffening in HFpEF is associated with impaired exercise performance (24), and treatment with agents that reduce stiffness in patients with diastolic dysfunction improves exercise capacity (36). Although diastolic dysfunction is often considered a key factor in the pathogenesis of HFpEF (2), most patients with diastolic dysfunction do not have clinical HF (1); further study is needed to clarify the role for therapies that target diastolic dysfunction. Nonetheless, the current data extend upon these earlier acute studies by showing an inverse association between E′ and afterload at steady-state, consistent with the hypothesis that part of the therapeutic improvement may be related to enhanced diastolic function.

The association of E′ with vascular load was strongest for measures of stiffening and late-systolic loading, even after adjusting for age. This is important, because E′ is known to decrease with age (14,37). Indeed, this may represent a mechanism whereby aging impairs early diastolic function, although the fact that age also remained an independent predictor in multivariate analysis indicates that age-dependent myocardial changes are also important. Earlier studies of the hemodynamic determinants of E′ have mostly been acute studies, focusing on preload rather than afterload. Acute preload modulation had no effect on E′ in an early human study (14), and, in another, E′ was found to correlate significantly with tau, with little preload dependence (31). The effects of afterload, however, were not reported. The direct relationship between A-wave amplitude and afterload is consistent with reduced early filling, lower E′, and thus greater dependence upon volume transfer in late diastole. Because A′ was not associated with arterial load, the greater dependence upon late-diastolic filling is unlikely due to enhanced atrial function.

Systolic velocity and contractility

S′ varied inversely with arterial loading, whereas other indexes of contractility did not. Reduced S′ velocities have been reported in populations thought to have normal systolic function, such as HFpEF (19), suggesting that EF may be insensitive to subtle contractile abnormalities. However, HFpEF patients typically have elevated afterload, and studies employing invasive measures of contractility have not demonstrated impaired contractility (27). Mitral annular S′ measures longitudinal, not radial, shortening, and it may be that structure-function changes associated with hypertension and aging lead to a compensatory increase in radial thickening (38). S′ was lower in hypertensive subjects, in association with an increase in afterload, yet EF was essentially the same in both groups. When S′ was indexed to load, the apparent discrepancies were no longer present. The present results question the validity of S′ as a robust indicator of contractility and suggest that indexing for arterial afterload may improve its utility in this regard.

The E/E′ ratio varies directly with LV filling pressures, because E shows much greater preload dependence than E′ (16,18,39). In the present study, E/E′ varied directly with afterload. Without invasive data, we cannot determine whether there was a true afterload-associated increase of cardiac filling pressures. The E/E′ ratio also varies directly with age, whereas E does not (37). It is conceivable that there are cutoffs for “normal” E/E′ ratios that are age specific, as has been observed for other noninvasive estimates of LV filling pressure (40).

Arterial afterload versus wall stress

Concentric hypertrophy develops with pressure overload, normalizing wall stress, but this does not mean that the original insult (increased arterial loading) is negated. Regardless of chamber geometry, vascular input impedance still potently dictates the magnitude and time course of forces generated during ejection, and this in turn influences myocardial tissue motion. The distinction between vascular afterload (aortic input impedance) and ventricular wall stress (often also considered to represent “afterload”) was demonstrated by the hypertensive group, who displayed higher arterial afterload, lower TDE velocities, yet similar wall stress. The TDE velocities did not correlate with wall stress in the present study, consistent with earlier studies showing longitudinal shortening to be unrelated to wall stress but declining with hypertrophy (38). The TDE velocities inversely correlated with both arterial load and wall thickness. Because the former is in the numerator and latter in the denominator of equations estimating wall stress, their effects cancel, leading to no net association when both remodeling and pressure overload coexist.

Study limitations

Although the measurements used are well validated with invasive data, invasive pressures were not measured. The TDE velocities observed are lower than in some studies, and therefore the E/E′ ratios are higher. This is likely because color-coded imaging was used, which determines the mean velocity of the spectral envelope rather than the peak, as in pulse-wave TDE, resulting in ∼30% lower values. The TDE measurements were obtained from the LV septum rather than lateral wall, which also contributed to lower observed E′. Impedance data was measured at the carotid rather than the aorta, but earlier studies support the use of central carotid pressures to assess central afterload (29).