The PROGRESS trial: questions about the effectiveness of angiotensin converting enzyme inhibitors

The health benefits associated with blood pressure (BP) lowering in persons with a history of cerebrovascular disease have not been studied extensively. Thus, the results of the recent Perindopril pROtection aGainst REcurrent Stroke Study (PROGRESS), a randomized double-blind placebo-controlled trial in patients who had had a previous stroke or transient ischemic attack (TIA), are welcome.1 In PROGRESS, the risk of recurrent stroke was reported to be 28% lower (95% confidence interval [CI] = 17% to 38%) in perindopril-treated than placebo-treated patients. Moreover, the risk of total major vascular events was reported to be reduced by 26% (95% CI = 16% to 34%). For those with either a recent or distant stroke or TIA, according to the study authors,1 “the general practitioner should consider starting treatment at the patient's next visit to surgery” (page 1039). Unfortunately, the nature of the trialists’ design, presentation, and interpretation of PROGRESS leave the practitioner unsure about exactly what therapy they ought to consider starting in their patients who have had a previous stroke or TIA. The results of the PROGRESS trial as presented provide little useful information for clinical practice. The design of the PROGRESS trial was complex. To be eligible, patients had to have a previous stroke or TIA within 5 years, had to have no definite indications or contraindications for an angiotensin-converting enzyme (ACE) inhibitor, and had to be compliant during the run-in phase of the study. There were no BP entry criteria; if the patients were hypertensive, they could be treated with antihypertensive drugs other than ACE inhibitors. Before randomization, physicians classified each potential subject according to whether the patient had or did not have a specific indication or contraindication to diuretics. Ran-domization was stratified on this characteristic. Among those with an indication or contraindication to diuretics, patients were randomized to 4 mg of perindopril per day or an identical-looking perindopril placebo. Among those without an indication for or a contraindication to diuretics, patients were randomized to combination therapy with 4 mg of perindopril per day plus the diuretic indapamide (2.5 mg per day) or to combination therapy with identical-looking perindopril plus indapamide placebos. Randomization was also stratified on study center, age, sex, entry systolic BP, and qualifying event. The comparison between combination therapy with perindopril-plus-indapamide and single-drug therapy with perindopril was one of the two prespecified group analyses. After an average of 3.9 years of follow-up, active treatment was associated with an overall reduction in the risks of both stroke and vascular events (Table 1). But the results differed markedly according to type of treatment, a prespecified factor on which randomization had been stratified. In subjects randomized to perindopril alone, BP lowering averaged 4.9/2.8 mm Hg, and treatment was associated with a 5% reduction in the occurrence of stroke (95% CI = −19% to 23%). In subjects randomized to perindopril plus a diuretic, the mean BP lowering was 12.3/5.0 mm Hg, and treatment was associated with a 43% reduction in the occurrence of stroke (95% CI = 30% to 54%). The effect of treatment differed significantly between randomized strata (P < .001). This unexpected observation has important implications for the analysis and interpretation of PROGRESS—implications that were apparently overlooked by the authors. Summary of primary findings from PROGRESS PROGRESS = Perindopril pROtection aGainst REcurrent Stroke Study; SBP = systolic blood pressure; DBP = diastolic blood pressure. P < .001 for difference between combination therapy and perindopril alone for both stroke risk and vascular event risk. Summary of primary findings from PROGRESS PROGRESS = Perindopril pROtection aGainst REcurrent Stroke Study; SBP = systolic blood pressure; DBP = diastolic blood pressure. P < .001 for difference between combination therapy and perindopril alone for both stroke risk and vascular event risk. In randomized trials,2 stratifying the randomization on factors such as age, sex, and study center, is designed to assure between-group comparability on factors that may affect prognosis. Thus, if one center drops out of the study, randomization stratified on study site assures that the balance in prognostic factors at the other sites would not be affected. On the other hand, stratifying the randomization on types of treatment intervention creates in the instance of PROGRESS two parallel trials. The analytic approach needs to assess not only the association between treatment and outcome in each of the two parallel trials but also whether the treatment effects seen in the two trials are similar. Insofar as they lack heterogeneity, combining the results is appropriate and reasonable. If, on the other hand, the findings in the two parallel trials differ to a substantial degree, then the findings need to be presented separately and interpreted separately. In PROGRESS, there were two distinct findings (Table 1): 1) the administration of perindopril alone was associated with virtually no reduction in the risk of stroke or vascular events; and 2) the combination of perindopril and indapamide was associated with major reductions in the risks of stroke and vascular events. The claim that a “blood pressure lowering regimen, which included perindopril for all patients and indapamide for 58%,… reduced the risk of stroke by more than a quarter” refers to an averaged effect that does not properly characterize the results seen in either group. The evaluation of the effectiveness of perindopril-based therapy, with or without diuretics, as an intervention among patients with cerebrovascular disease leaves at least one important question unaddressed—the effectiveness of indapamide alone in the prevention of recurrent cerebrovascular disease. If the trialists had wished to answer this question, they would have used a factorial design with subjects randomized to one of four arms: placebo, perindopril alone, indapamide alone, or perindopril plus indapamide. This design would have allowed the trialists to test at once the main effects of and, at the same time, the possibility of interactions between therapies on the occurrence of total stroke or vascular events. The major limitation of the PROGRESS trial was the failure to include a group randomized to indapamide alone. In the absence of a factorial design, we can speculate about the fraction of the benefit attributable to indapamide alone. If we assume that there was no interaction between therapies on an additive scale, then the risk reduction associated with indapamide use is estimated to be 38% (43% for combined therapy minus 5% for perindopril alone). This point estimate of a 38% risk reduction associated with the use of this diuretic in the secondary prevention setting is consistent with the 34% (95% CI = 22% to 45%) risk reduction seen for low-dose diuretics in the primary prevention setting.3 Moreover, the estimate of 38% for indapamide in PROGRESS is also consistent with the finding of a 29% (95% CI = 12% to 42%) reduction in the risk of recurrent stroke associated with indapamide use in the Post-stroke Antihypertensive Treatment Study.4 Thus, it is plausible that most or all of the risk reduction associated with combination therapy in PROGRESS can be attributed to the use of indapamide. The small BP differences in the two parallel trials cannot explain large treatment differences. Although the trialists invoke an hypothesis of nonlinearity to explain this discrepancy, an alternative explanation is simply that, independent of level of BP lowering, some drug treatments have different effects on major disease end points than other drug treatments. In the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT),5 for instance, the 2- to 3-mm Hg BP differences between doxazosin and chlorthalidone cannot explain the twofold difference in the risk of heart failure.5 In one large clinical trial that compared diuretics or β-blockers to placebo,6 the in-trial BP reduction accounted for only a small proportion of the overall reduction in the stroke risk. The other predefined group analysis in PROGRESS involved the prognostic factor, hypertension, defined as a BP more than 160/90 mm Hg at baseline. For the primary end point of recurrent stroke, the risk reduction associated with combination therapy was similar in hypertensive subjects (44%, 95% CI = 28% to 57%) and nonhypertensive subjects (42%, 95% CI = 19% to 58%) (Table 1). Therefore, even in patients with mild elevations of BP—the nonhypertensive group with entry BP less than 160/90 mm Hg—active combination treatment was associated with an important reduction in the risk of recurrent stroke. This finding provides support for the current U.S. recommendations that define high BP as a level more than 140/90 mm Hg7 and that have recently emphasized the importance of treating mild to moderate elevations of systolic BP.8 What treatment should doctors offer to their patients with a history of cerebrovascular disease? In PROGRESS, the diuretic indapamide was used at a moderate dose. The findings of other trials suggest that low-dose diuretics such as chlorthalidone and hydrochlorothiazide are equally effective in stroke prevention.3 The results of the PROGRESS trial provide additional support for the use of diuretics as the first-line BP-lowering treatment for the prevention of stroke. Whether the addition of ACE inhibitors improves the risk reduction remains unclear. The dose of perindopril used in PROGRESS, 4 mg/day, may have been too low. Moreover, many of the older adults eligible for this trial probably had low-renin hypertension, which is less likely to respond to ACE inhibitors. Low-dose diuretics are currently recommended as the first-line agent for uncomplicated hypertension.7 On the basis of the findings from PROGRESS, a history of cerebrovascular disease appears to be a new compelling indication for low-dose diuretic therapy.