Methods

Results

Baseline characteristics are summarized in Table 1. For all patients, EF increased by 7±7 points from baseline to the 6 month time period therapy (from 25±9 to 32±9, P<0.0001). Changes in EF were strongly correlated with changes in stroke volume (Figure 1A), but not with changes in end-diastolic volume (Figure 1B). Increased chamber contractility (R_es), decreased HR, and decreased TPR each correlated with changes in EF (Figure 1C through 1E, respectively). Because E_a is directly related to TPR and HR, it also correlated strongly with changes in EF (Figure 1F).

Responders and nonresponders did not differ in etiology of heart failure or carvedilol dose at any time during treatment. Only a higher resting HR at baseline in responders differentiated these 2 groups (Table 2). Stroke volume increased in responders but not in nonresponders (Figure 1 and Table 2). EDV trended lower in responders and contractility, indexed by R_es, increased more in responders (P<0.001). HR decreased in both groups, but the decrease was greater in responders (P=0.01) so that HR was comparable in both groups after 6 months. TPR did not change in responders but increased in nonresponders. E_a decreased significantly in responders, primarily because of the decrease in HR. SV increased in responders in part because of decreases in end-systolic volume (P=0.01). Neither LV mass nor the sphericity index changed significantly in either group. Thus, changes in HR, R_es and, to a lesser extent, TPR each correlated with increases in SV and EF.

The relative contribution of these factors to changes in EF was determined by multiple linear regression analysis which showed that: ΔEF ≈18.1ΔR_es−0.25ΔHR−0.82ΔTPR (r²=0.89). This relation provided a high degree of accuracy to account for changes in EF (Figure 2). Using data obtained from responders (who exhibited an average 0.18 mm Hg/mL increase of R_es, a 23 bpm HR reduction and a 0.14 mm Hg^.s/mL reduction in TPR), the regression model indicates that 56% of the change in EF was statistically correlated with the HR reduction, 28% was correlated with the increase in R_es and 16% was correlated with the decrease in TPR.

Changes over time (Figure 3) show that HR decreased in both groups, but by a greater amount in responders. EDV decreased over time in responders and ESV decreased in responders so that stroke volume (not shown) steadily increased. Mean arterial pressure decreased at 2 months and then returned to baseline in both groups. TPR did not change significantly, except for an increase in nonresponders at 6 months. In responders, E_a decreased progressively, primarily because of the decrease in HR and R_es increased. Ventricular-vascular coupling ratio improved in responders. LV mass did not change significantly.

Group mean estimated pressure-volume loops were constructed using the EDV, ESV, and estimate of end-systolic pressure¹⁶ (Figure 4). R_es and E_a are represented graphically by lines connecting the estimated end-systolic pressure-volume point to the origin or to the end-diastolic volume point on the volume axis, respectively. A comparison of these loops (paired baseline in blue, specified time of treatment in red) shows that for nonresponders, there was no significant effect at 2 weeks. At 2 months, there was a reduction in arterial pressure due to reductions in R_es and E_a, the latter due primarily to a decrease in HR. Both recovered by 6 months so that there was essentially no difference from the baseline condition. The recovery of E_a was due to offsetting effects of a decrease in HR and increase in TPR. In responders, R_es increased progressively while E_a decreased, mainly due to a decrease in HR; EDV trended to decline. These changes resulted in a progressive increase in stroke volume and EF with maintenance of blood pressure.

Limited data were available at 12-month follow-up (15 subjects, 5 nonresponders, and 10 responders). As at the other time points, the improvement in EF was sustained in the 6-month responders and the change in EF was correlated with changes in stroke volume (r=0.69, P=0.005) but not with changes in EDV (r=−0.31, P=0.26). The reductions in HR and E_a, and the increase in R_es were each sustained in the 6-month responders and did not change any further in nonresponders. After 12 months of therapy, LV mass declined (250±17 to 222±16 g, P=0.0531 in the entire cohort) but EDV did not change significantly (209±17 to 192±19, P=0.175).

Discussion

As in prior studies^9,21 EF increased by an average of ≈7 percentage points within 6 months of initiating carvedilol. During this time period, a decrease in HR and an increase in contractility, indexed by R_es, were the primary factors contributing to the increase in EF. The importance of HR reduction as a mechanism for increased EF is further supported by the observation that patients whose EF increased ≥5 points (responders) had higher baseline resting HRs and greater HR reductions than nonresponders, findings that are consistent with prior studies.^5,21,22 Reduction of TPR contributed quantitatively little to improved EF.

Prior studies of ventricular-vascular coupling have provided quantitative understanding of the dependence of EF on contractility, preload, and afterload¹⁷ with EF=(EDV−V_o)/(1+E_a/E_es)/EDV. This equation, combined with our data permit estimation, on a physiological basis, of the relative contributions of the different parameters to changes in EF (see Appendix 1 in the online-only Data Supplement). One limitation in directly applying this construct, however, is that we did not measure ESPVRs; rather, we estimated a single end-systolic pressure-volume point and relied on R_es (the end-systolic pressure-volume ratio) to quantify contractility. Although R_es is simple to measure and has the advantage of combining changes in E_es and/or V_o into a single measure for assessing changes in contractility, it cannot be used directly in the equations above for EF. β-blocker enhancement of LV contractility may be manifest on the pressure-volume diagram as either an increase in E_es²³ or a decrease in V_o.²⁴ These 2 scenarios are displayed in Figure 5, where we estimate baseline V_o as 50 mL.^25,26 For either a constant V_o and variable E_es (Figure 5A) or a constant E_es and variable V_o (Figure 5B), the decrease in HR contributes ≈40% and the increase in E_es contributes ≈45% to the increase in EF (see Appendix). Even for the case when V_o is assumed to be 0 (so that E_es=R_es) the relative contributions of HR and contractility to the changes in EF are similar to the estimates above (40% and 35%, respectively). Thus, the conclusions using the theoretical construct are not critically dependent on an accurate choice of V_o or whether V_o is constant between time points. These findings confirm, on a mechanistic level, the conclusions derived based on statistical considerations of our patient data (Figure 2) that HR reduction in and of itself is a major factor contributing to increased EF during carvedilol treatment.

The impact of cardiac cycle length on ventricular-vascular coupling was derived analytically over 2 decades ago, facilitating the ability to predict the stroke volume resulting from the complex mechanical interactions between the ventricle and its arterial system.¹⁷ These findings and those of others^27,28 demonstrated that HR effects ventricular-vascular coupling in a manner similar to changes in arterial resistance. This becomes evident by examining the derivation of the well established concept of indexing ventricular afterload by the effective arterial elastance (E_a). Intuitively, the influence of HR on ventricular-vascular coupling can be appreciated by realizing that at slower HRs there is more time for distal run off of arterial blood volume, thereby decreasing the pressure head against which the heart has to eject, thus an apparent reduction in afterload. Conversely, at high HRs, less time for peripheral runoff translates to higher pressure head and an apparent increase in afterload.

A meta-analysis of 35 trials, including >22 000 patients followed for an average of 9.6 months demonstrated that there was a close relation between all-cause annualized mortality and HR reduction among patients with heart failure receiving β-adrenergic antagonists and that a similarly strong correlation was observed between HR reduction and increases in EF. This suggests that HR reduction is a major contributor to the clinical benefits of β-blocker therapy in systolic heart failure. Concordant with these epidemiological findings, human data support the importance of HR reduction as a mechanism for improvements in EF with β-adrenergic antagonists. Among patients with class III heart failure and an EF <40% who were paced >70% of the time, EF decreased with increasing HR (EF of 25±6% at a HR of 90 beats per minute compared with and EF of 33±6% and 30±6% at 60 and 70 bpm, respectively).^7,17 Additionally, a study with a similar design showed that in patients on β-adrenergic therapy, pacing at a HR of 60 bpm was associated with a 2.2±5.4% increase in EF, while pacing at rate of 80 bpm was associated with a 4.5±4.4% decrease in EF.⁸ These data suggest that the benefits of β-blockade on left ventricular structure and function are attenuated or reversed by increasing HR even to a modest extent.

In addition to the direct hemodynamic effects, prior studies suggest that chronic HR reduction has beneficial effects on contractility. Nagatsu et al⁶ showed that maintenance of high HR in dogs with mitral regurgitation-induced heart failure largely negated the improvement of myocardial contractility observed during chronic β-blocker therapy.

Additional evidence for the important role of HR reduction in the improvement observed in EF in patients with left ventricular systolic dysfunction can be found with newer drugs, other than β-adrenergic antagonists, that slow HR. Ivabridine, an I_f channel sinus node inhibitory agent, has been demonstrated in animals to preserve cardiac output despite the decrease in HR because of an increase in stroke volume owing to a decrease in left ventricular end-systolic diameter without modification of left ventricular end-diastolic diameter.²⁹ Also, in humans, ivabridine, has been shown to significantly reduce left ventricular volumes in comparison with placebo, with a mean 5% increase in EF with ivabradine, compared with a mean 0.5% decrease in corresponding patients receiving placebo.³⁰ Such an effect was acutely observed in patients with severe systolic heart failure, in whom HR reduction was accompanied by increases in stroke volume.³¹

In the original VHeFT trial, an increase in EF of >5 from baseline at 6 months and at 1 year in the V-HeFT II trial were the strongest predictors of mortality benefit after adjusting for baseline EF.³² Such data have been duplicated with β-blocker therapy.³³ Because the magnitude of HR reduction is correlated with an increase in EF with β-blocker therapy and, in our data, HR reduction was 1 of the 2 primary explanations for the observed increase in EF, HR reduction is an important goal of β-blocker therapy in patients with systolic heart failure.³⁴

Our findings are in agreement with prior studies regarding both the magnitude^9,20,35 and time course³⁶ of EF change following β-blocker initiation. The relatively long time scale over which changes occur emphasizes that they are not attributable to immediate, direct pharmacological actions but rather relate to longer term biological effects on myocyte contractile function.^37,38 Our results showing the relevant contribution of change in ESV to increase in EF are further concordant with other results which demonstrated that for metoprolol,²⁴ bucindolol,²³ and carvedilol,^39,40 changes in EF were mainly due to increased stroke volume, which in turn were mainly due to decreases in ESV.

In our data the responders at baseline had significantly higher HRs than nonresponders at baseline and these differences were no longer present at 6 months. Accordingly, because HR reduction was shown to the primary mechanism by which improvements in EF were achieved with carvedilol, the utility of treating patients with lower HRs with carvedilol is raised. Despite the close and statistically significant relation we observed between EF and HRR, our data suggest that there are mechanisms other than HRR that contribute to the improvement in EF and likely the clinical benefits of therapy. Accordingly, therapy for patients with chronic systolic heart failure should not be withheld on the basis of a low resting HR.

Although invasive measurements of LV pressure and volume provide more accurate and extensive information about systolic and diastolic LV properties than the noninvasive techniques we used, these are generally not possible in larger scale, longitudinal clinical studies. Accordingly, we could not measure ESPVRs to determine E_es and V_o values. Despite its limitations, R_es as a single contractile index simplifies statistical assessment of contractility. Although baseline chamber and vascular properties did not differ significantly between responders and nonresponders (Table 2), end diastolic volume indexed to body size trended to be smaller in the responders than nonresponders (110±30 versus 126±60, P=0.10, respectively). Accordingly, it is possible that responders had heart failure for a shorter duration than nonresponders and this confounded the ability of the ventricle to reverse remodel. Unfortunately, we did not record the duration of heart failure in these subjects and thus were unable to address this limitation. Future validation of our findings using another prospectively collected data set is required to increase the certainty and generalizability of our results. Finally, in light of the small sample size, absence of data at certain time points and short follow-up as well as the inability to achieve a target dose in all patients, we cannot exclude the possibility that reverse remodeling³⁶ and/or changes in TPR are more prominent and important after longer-term therapy.

In those heart failure patients whose EF increased at least 5 points, the contribution from HR reduction was twice that of increased contractility, with the decrease in TPR contributing less. These data are consistent with results of a recent meta-analysis showing that HR reduction is the major factor associated with mortality benefit. Collectively, these findings support alternate efforts to reduce HR, particularly for patients intolerant to β-blocker therapy or for those whose HR does not decrease adequately in response to β-blockers.