Prevention of Heart Failure in Hypertension-Disentangling the Role of Evolving Left Ventricular Hypertrophy and Blood Pressure Lowering: The ALLHAT Study

Abstract

Background Hypertension is a known risk factor for heart failure ( HF ), possibly via the mechanism of cardiac remodeling and left ventricular hypertrophy ( LVH ). We studied the extent to which blood pressure ( BP ) change and evolving LVH contribute to the effect that lisinopril, doxazosin, and amlodipine have on HF compared with chlorthalidone. Methods and Results We conducted causal mediation analysis of ALLHAT (Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial) data (1994-2002; in-trial follow-up). ALLHAT participants with available serial ECG s and BP measurements were included (n=29 892; mean age 67±4 years; 32% black; 56% men): 11 008 were randomized to chlorthalidone, 5967 to doxazosin, 6593 to amlodipine, and 6324 to lisinopril. Evolving ECG LVH and BP lowering served as mediators. Incident symptomatic HF was the primary outcome. Linear regression (for mediator) and logistic regression (for outcome) models were adjusted for mediator-outcome confounders (demographic and clinical characteristics known to be associated both with both LVH /hypertension and HF ). A large majority of participants (96%) had ECG LVH status unchanged, but 4% developed evolving ECG LVH . On average, BP decreased by 11/7 mm Hg. In adjusted Cox regression analyses, progressing ECG LVH (hazard ratio [ HR ] 1.78 [95% CI 1.43-2.22]), resolving ECG LVH ( HR 1.33 [95% CI 1.03-1.70]), and baseline ECG LVH (1.17 [95% CI 1.04-1.31]) carried risk of incident HF . After full adjustment, evolving ECG LVH mediated 4% of the effect of doxazosin on HF . Systolic BP lowering mediated 12% of the effect of doxazosin, and diastolic BP lowering mediated 10% of the effect of doxazosin, 7% of the effect of amlodipine, and borderline 9% of the effect of lisinopril on HF . Conclusions Evolving ECG LVH and BP change account for 4% to 13% of the mechanism by which antihypertensive medications prevent HF . Clinical Trial Registration URL : http://www.clinicaltrials.gov . Unique identifier: NCT 00000542.

Trial registration: ClinicalTrials.gov NCT00000542.

Keywords: ECG; antihypertensive agent; heart failure; hypertension; left ventricular hypertrophy.

Figures

Figure 1

Flow diagram of exclusion criteria applied to achieve the final study population for this secondary analysis of ALLHAT (Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial) data. BP indicates blood pressure; LVH, left ventricular hypertrophy; ITT, intention to treat.

Figure 2

Directed acyclic graph to illustrate possible structural relationships between randomized antihypertensive treatment (Rx) in intention‐to‐treat (ITT) analysis evolving ECG left ventricular hypertrophy (LVH) (A) or blood pressure (BP) lowering (B), and incident heart failure (HF). CC denotes common causes (confounding factors), measured and unmeasured. The mediated effect is represented by the pathway from antihypertensive Rx to incident HF that goes through (A) evolving ECG LVH or (B) BP lowering. The direct effect is the pathway from antihypertensive Rx straight to incident HF.

Figure 3

Scatterplots of (A) in‐trial systolic blood pressure (SBP) change (Y‐axis) against baseline SBP (X‐axis), and (B) diastolic blood pressure (DBP) change (Y‐axis) against baseline DBP (X‐axis). A line of the best linear fit is shown.

Figure 4

Unadjusted Kaplan‐Meier curves for probability of (A) incident symptomatic heart failure (HF) and (B) hospitalized or fatal HF in all treatment groups ALLHAT (Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial) participants with evolving ECG left ventricular hypertrophy (LVH) development (blue dotted line), resolution (green dashed line), or without evolving ECG changes (red solid line).

Figure 5

Unadjusted Kaplan‐Meier curves for probability of incident symptomatic heart failure (HF) in (A) doxazosin, (B) chlorthalidone, (C) amlodipine, and (D) lisinopril treatment groups. Evolving ECG left ventricular hypertrophy (LVH) groups as described in Figure 3 legend. CHF indicates congestive heart failure.

Figure 6

Adjusted (by age, sex, and race/ethnicity) risk of symptomatic congestive heart failure (HF) associated with achieved in‐trial greatest (A) systolic blood pressure (SBP) and (B) diastolic blood pressure (DBP) changes in all participants. Restricted cubic spline with 95% CI show change in hazard ratio (Y‐axis) in response to BP change (X‐axis). The 50th percentile of BP change is selected as the reference. BP indicates blood pressure.

Figure 7

Adjusted risk of symptomatic congestive heart failure (HF) associated with achieved in‐trial greatest systolic blood pressure SBP (I) and diastolic blood pressure DBP (II) changes in (A) doxazosin, (B) chlorthalidone, (C) amlodipine, and (D) lisinopril treatment groups. See Figure 5 legend for details. HR indicates hazard ratio.