The Role of Calcium Antagonists in Stroke Prevention

Abstract

Lowering elevated blood pressure significantly reduces the risk of stroke, but whether any specific type of antihypertensive medication is better for this purpose, independent of blood pressure‐lowering effects, is controversial. Compared with placebo or no treatment, the point estimate for stroke reduction is the largest with calcium antagonists; all such studies used dihydropyridine compounds. When all clinical trials published through December 2004 are combined in meta‐analysis, stroke is not significantly reduced when an initial dihydropyridine or nondihydropyridine calcium antagonist is compared with an initial diuretic or P blocker, but when the two subclasses are combined, stroke reduction (9%) is significant. The risk of heart failure, however, is significantly increased (by 29%) with the initial calcium antagonist. These estimates may change as newer trials are reported.

Since 1958, stroke has been the third most common cause of death in the United States, and one of the leading causes of long‐term disability. On a population basis, the most important risk factor for stroke is hypertension, and many clinical trials have shown that lowering blood pressure (BP) with various antihypertensive drugs prevents stroke. Depending on which meta‐analysis one prefers, and which trials are included, active antihypertensive drug therapy is associated with a 38%–42% reduction in stroke in trials against placebo or no treatment.

In traditional meta‐analyses of clinical trials comparing the various drug classes vs. placebo or no treatment, each of the major BP‐lowering drugs is effective: strokes are reduced with diuretics by 34% (95% confidence interval [CI], 22%–46%), (3 blockers by 29% (95% CI, 17%–41%), angio‐tensin‐converting enzyme inhibitors by 31% (95% CI, 23%–39%), and calcium antagonists by 40% (95% CI, 25%–52%). ¹ No long‐term trials have compared nondihydropyridine calcium antagonists with placebo or no treatment, whereas six trials (Prospective Randomized Evaluation of Vascular Events With Norvasc Trial [PREVENT], ² Systolic Hypertension in Europe [Syst‐Eur], ³ Shanghai Trial of Nifedipine in the Elderly [STONE], ⁴ Systolic Hypertension in China [Syst‐China], ⁵ Irbesartan Diabetic Nephropathy Trial [IDNT], ⁶ and Nisoldipine in Coronary Artery Disease in Leuven [NICOLE]) ⁷ have used a drug from the dihydropyridine subclass.

Stroke prevention is highly dependent on the degree of BP reduction, which is difficult (if not impossible) to control prospectively in case‐control studies or cohort studies. In addition, results of these study designs may be confounded by “indication bias,” which occurs when individuals at higher risk are non‐randomly allocated to a specific treatment group. This may be one reason why calcium antagonists have been associated in case‐control and cohort studies with a higher rate of cardiovascular events, including death. ⁸ , ⁹ It is much easier to assess the results of comparative randomized clinical trials pitting initial antihypertensive drugs against each other, which typically report the achieved BP differences across the randomized arms. In “the largest and most important BP trial ever done in the USA,” ¹⁰ the Antihypertensive and Lipid‐Lowering Treatment to Prevent Heart Attack Trial (ALLHAT), ¹¹ the dihydropyridine calcium antagonist amlodipine was compared with the best‐studied diuretic, chlorthalidone. Despite a slightly smaller degree of BP reduction than the diuretic, amlodipine therapy was associated with an 8% reduction in stroke. The calculated probability value for this comparison (p=0.02) was slightly larger than the prespecified threshold value of p=0.0178, which had been adjusted for multiple comparisons and the unequal randomization scheme, so the comparison did not reach predefined statistical significance. In none of the eight prespecified subgroups (dichotomized by age over 65 years, gender, race/ethnicity, or diabetes) was the reduction in stroke with the calcium antagonist over the diuretic statistically significant. Perhaps more importantly in ALLHAT, amlodipine was associated with a highly‐significant (p<0.001) 38% increase in risk of heart failure, which was seen to a slightly lesser extent (19%; p<0.001) with lisinopril, compared with chlorthalidone. These differences were significant in all eight prespecified ALLHAT subgroups. These data suggested that, in North American patients, an initial calcium antagonist might be slightly better for stroke prevention than an initial diuretic, but the latter was far better at preventing heart failure. In absolute terms, for each stroke prevented in the amlodipine group in ALLHAT (compared with chlorthalidone), there were 6.3 more patients who suffered a fatal or hospitalized heart failure episode. ¹¹

Long‐term clinical trials have now been reported using both dihydropyridine and nondihydro‐pyridine compounds. In nearly all cases, the comparator was either an initial diuretic or (β blocker. In the Nordic Diltiazem (NORDIL) ¹² trial and the Swedish Trial in Older Patients With Hypertension‐2 (STOP‐2), ¹³ no data are available concerning how many events occurred among patients who received the initial diuretic, as opposed to the initial (3 blocker, so most analyses ¹ , ¹⁴ , ¹⁵ have used the category “diuretic or (β blocker” as the traditional drug arm. This strategy also allows combination of such disparate studies as Verapamil Hypertension Atherosclerosis Study (VHAS), ¹⁶ which used chlorthalidone as the first‐line comparator, and the International Verapamil/Trandolapril Study (INVEST), ¹⁷ which used atenolol initially.

Four outcome‐based trials have compared an initial nondihydropyridine calcium antagonist with an initial diuretic or (β blocker. NORDIL used diltiazem; verapamil was used in VHAS, the Controlled Onset Verapamil Investigation of Cardiovascular Endpoints (CONVINCE), ¹⁸ and INVEST. For fatal and nonfatal stroke, the meta‐analysis showed no significant inhomogeneity (p=0.80) across studies, despite major differences in the number of strokes (VHAS had <1% of the total) and offsetting trends (the point estimate for both CONVINCE and VHAS was opposite to the two other studies). When the studies are combined by the traditional Mantel‐Haenszel method, the nondihydropyridine calcium antagonist was associated with an 8% (95% CI, −6% to 21%; p=0.14) reduction in fatal or nonfatal stroke, compared with an initial diuretic or β blocker.

The analogous meta‐analysis of an initial dihydropyridine calcium antagonist compared with an initial diuretic or β blocker combines seven trials (ALLHAT, International Nifedipine Study: Intervention as a Goal in Hypertension Treatment [INSIGHT], ¹⁹ STOP‐2, European Lacidipine Study of Atherosclerosis [ELSA], ²⁰ Systolic Hypertension in the Elderly Long‐Term Lacidipine [SHELL] ²¹ trial, Multicenter Isradipine Diuretic Atherosclerosis Study [MIDAS], ²² and National Intervention Cooperative Study in Elderly Hypertensives [NICS‐EH] ²³ ). Of these, ALLHAT has the largest number of participants and reported strokes (about 53% and 59% of the totals, respectively). Not surprisingly, therefore, the overall results of the meta‐analysis are influenced by ALLHAT, the largest contributor of strokes. Of the seven trials, only the point estimate for MIDAS does not favor the calcium antagonist, and there is no significant inhomogeneity for fatal or nonfatal stroke (p=0.96). When these seven studies are combined by the traditional Mantel‐Haenszel method, the dihydropyridine calcium antagonist was associated with an 8% (95% CI, −1% to 17%; p=0.09) reduction in fatal or nonfatal stroke, compared with an initial diuretic or β blocker.

If one compares stroke prevention across all trials that used an initial dihydropyridine calcium antagonist vs. any other initial antihyper‐tensive drug, more patients are added from both ALLHAT and STOP‐2, as well as the results from IDNT, Appropriate Blood Pressure Control in Diabetes‐Hypertension substudy (ABCD‐HTN), ²⁴ and the recent Valsartan Amlodipine Long‐Term Use Evaluation (VALUE). ²⁵ Except for MIDAS and ABCD, all the point estimates favor the dihydropyridine calcium antagonist, and there is no significant inhomogeneity (p=0.21). However, the reduction in fatal or nonfatal stroke is not significant (6% in favor of the dihydropyridine calcium antagonist; 95% CI, 13% to −2%; p=0.12).

If one combines the results from all studies involving both subclasses of calcium antagonists, statistical power increases, and the influence of ALLHAT diminishes (as it now contributes only 40% of the total number of strokes). When all 11 trials are included in the meta‐analysis, there is no significant inhomogeneity (p=0.51)—the point estimate remains at 8%, but now it is significant (95% CI, 0.9%–15%; p<0.039)(see Figure).

Figure.

Results of a meta‐analysis of stroke in all clinical trials comparing an initial calcium antagonist (CCB) with an initial diuretic (D) or β blocker (β). OR=odds ratio. Trial acronyms expanded in text.

Before a calcium antagonist can be recommended over a diuretic or β blocker as providing significantly better stroke prevention, however, a broader look at other end points reveals at least one other important difference. Major cardiovascular events (defined for this purpose as first occurrence of coronary heart disease event, stroke, or cardiovascular disease‐related death, and estimated for ALLHAT, INVEST, and INSIGHT), coronary heart disease events, all‐cause mortality, and cardiovascular mortality were quite similar across the 11 trials (odds ratios varied between 0.99 and 1.03) and were not significantly different. Despite the absence of data from ELSA, however, an initial calcium antagonist was associated with a highly significant (p<0.000001) 29% (95% CI, 19%–39%) increase in the risk of heart failure, compared with an initial diuretic or β blocker. Thus, in these aggregated studies, an initial calcium antagonist prevented 8% of the strokes, but at the expense of a much greater, 29% increased risk of heart failure. Some have argued that stroke is generally irreversible and typically leads to major reductions in lifespan and quality of life, whereas heart failure can be erroneously diagnosed and merely be a consequence of mild fluid overload—and is therefore not as “serious” an end point as stroke. This was not the experience in ALLHAT. ²⁶ , ²⁷ These estimates may change as results from ongoing trials (e.g., the Anglo‐Scandinavian Cardiac Outcomes Trial [ASCOT] ²⁸ ) are reported that compare calcium antagonists with other drugs; for now, the meta‐analyses results are consistent with current US guidelines ²⁹ recommending an initial low‐dose diuretic for most patients with hypertension.