The risk of stroke is reduced by calcium channel blockers (CCBs), including amlodipine, and the size of the blood pressure (BP) reductions is an important determinant of the size of this treatment effect.1 It remains uncertain, however, whether mechanisms independent of BP also affect the size of the treatment benefit to stroke, and there have been few formal attempts to investigate any such drug-specific effects. In their meta-regression analysis of amlodipine, Wang and associates2 plotted only 5 odds ratios (ORs) for stroke from two individual trials3,4 and three “pooled trials” with reference treatment of the same class: placebo,5–7 angiotensin-converting enzyme inhibitors (ACEIs),3,5 and angiotensin receptor blockers (ARBs).7–9The Cochrane Handbook for Systematic Review of Intervention,10 however, recommends that meta-regression should generally not be considered when there are fewer than 10 studies in a meta-analysis. We conducted re-analyses to determine the relative contribution of BP-dependent and independent mechanisms to the reduction in risk of stroke produced by amlodipine.
From the most comprehensive meta-analysis1 published in 2009, we identified 10 ORs in 7 trials of amlodipine.3–9 The BP difference was assigned a positive value when the follow-up BP level was higher in the control compared with amlodipine. Percentage reductions in risk were estimated as [(1 − OR) × 100]. The association between the difference in follow-up systolic BP levels and the log OR for stroke was investigated using unrestricted maximum likelihood meta-regression analysis with inverse variance weighting. The intercept of the regression line estimated the value of the log OR when the difference in BP reduction is zero. The slope of the regression line estimated the log OR for a unit change in follow-up systolic BP difference. Both were reported with 95% confidence intervals (CIs). Analyses were carried out using Comprehensive Meta-Analysis version 2 (Biostat, Englewood, NJ).
The magnitude of the risk reduction achieved for stroke was positively associated with the size of BP reduction. Treatment with amlodipine achieved a 6.1% (95% CI, 1.1–10.8; P=.01817) reduction in the risk of stroke for each 1-mm Hg reduction in BP (slope of the regression line, −0.06241; 95% CI, −0.11420 to −0.01063; Figure). For stroke, amlodipine conferred additional protection beyond that conferred by BP reduction alone. At zero BP reduction, the estimated relative risk reduction for stroke was 10.3% (95% CI, 2.9–17.2; P=.00754) (intercept of the regression line, −0.10867; 95% CI, −0.18838 to −0.02896; Figure).
The present analyses, based on 10 ORs from 7 trials, confirm that the size of the reduction in BP achieved with amlodipine is a major determinant of the size of the reduction in stroke. In addition, these analyses have identified a potentially important BP-independent protective effect of amlodipine on the risk of stroke. In particular, there was clear evidence of protection against stroke with amlodipine even in the absence of any reduction in BP. This BP-independent effect was equivalent to the estimated effect of an additional 1.7-mm Hg reduction in systolic BP.
A meta-regression analysis1 found the significant relationship between BP reduction and a favorable effect of CCBs (not limiting to amlodipine) for cardiovascular death, major cardiovascular events, heart failure, and stroke. Although Wang and collaborators2 showed BP-dependent and -independent effects of amlodipine on the risk of stroke in their meta-regression analysis, they plotted only 5 ORs. Therefore, we re-analyzed 10 ORs from 7 trials and confirmed BP-dependent and -independent effects with amlodipine for the risk of stroke.