Editorial Comment—Secondary Prevention of Stroke: Beyond Meta-Analyses
Secondary prevention of stroke by blood pressure lowering drugs has been assessed in randomized controlled trials for more than 30 years,1 but the formal demonstration of a clinical benefit awaited the results from 2 major trials: PATS in 19952 and PROGRESS in 2001.3 Rashid et al have summarized the results from these and other available randomized controlled trials, to assess the effectiveness of these drugs on vascular events. In addition, they address other clinically relevant questions in exploring the heterogeneity between trials, and the relationship between blood pressure fall and risk reduction.
Overall, the use of blood pressure-lowering drugs was associated with significant reductions in stroke, myocardial infarction, and total vascular events. Beneficial trends observed for vascular or total mortality were not statistically significant. The methodology of randomized controlled trials allows affirmation that drugs per se provoked these reductions.
Is “Hypertension” Needed for Expecting a Benefit From Blood Pressure-Lowering Drugs?
The benefit observed in the 4 trials that included participants irrespective of their blood pressure level was of the same magnitude of that observed in the others. This reinforces the results obtained in subgroups from isolated trials,3,4 strongly advocates the prescription of the evaluated drugs in people without hypertension for the prevention of stroke recurrence, and puts into question the definition of hypertension.
Which Drug Can Be Used as a First-Line Therapy?
Exploring the heterogeneity of results suggested that it could be partially explained by the class of the first-line drug. In particular, the beta blocker atenolol was not associated with any benefit, and ACE inhibitors alone (ramipril and perindopril) reduced only the risk of myocardial infarction. On the contrary, diuretics (thiazide or indapamide) reduced the risk of stroke, whereas the association of indapamide with ACE inhibitor (perindopril) reduced the risk of all vascular events.
It should be emphasized that these results, from indirect comparisons, are only exploratory, with a weak level of evidence due to potential confounders: a lot of other factors may play a role in this heterogeneity, such as stroke subtype but also a variety of characteristics that will not be possibly assessed even in a meta-analysis on individual patient data. Among these factors, the choice of drug is one of the few that can be controlled: the superiority of a drug over another could be properly evaluated through direct comparisons.
What Is the Role of Blood Pressure-Lowering in This Benefit?
That the drugs from different classes did not reduce the risk similarly, whereas they reduced blood pressure in an approximately similar extent, suggests that other mechanisms than blood pressure reduction may be at play.
Of note is the fact that the majority of included trials explored the benefit associated with fixed doses of drugs, irrespective of the observed blood pressure response to drug. This demonstrates that such a simple strategy is efficient. Moreover, there is no evidence that this strategy is less efficient than others targeted toward a given level of blood pressure. We should keep in mind that, as the definition of hypertension, the definition of blood pressure target on treatment remains arbitrary, without any appropriate validation.
As others did on a larger data set,5 the authors explored the association between blood pressure fall and the risk reduction through a regression approach. They found this association significant from a statistical point of view. However, as others did, they obtained such a result by including in the regression equation a quadratic factor. This quadratic factor makes the results uneasy to interpret: a curvature in the regression function suggests that the association is not straight, and that intensifying blood pressure reduction does not lead to any further benefit, or even provoke adverse events that reduce the size of the benefit (J-curve phenomenon). Once again,6 it should be noted that association does not mean causality: it could be explained by confounders, such as other drug properties, or underlying risk of the studied population.7 That the provoked blood pressure fall explains at least partly the risk reduction observed on such drugs is more than likely. However, evidence accumulates to prove that the lower is not always the better.
The Future of Cardiovascular Prevention Drugs
The results from Rashid et al remind us of the limitations of the concepts at the basis of past and current guidelines on the management of blood pressure drugs:
The definition of hypertension is arbitrary, and people without hypertension, whatever its definition, may benefit from drugs that, among other properties, lower blood pressure.
The level of risk to be prevented is one of the major factors of medium-term (eg, 5-year) benefit: in secondary stroke prevention, the level of 5-year risk is very high and concerns a category of cardiovascular events that is best prevented by blood pressure-lowering drugs.
It is likely that for a given level of blood pressure decrease, a given choice of drug class does not lead to the same benefit.
Fixed dose drug strategy may be, on average, as efficient as traditional blood pressure targeted strategy.
Other individual characteristics, such as the nature of the qualifying stroke (ischemic or hemorrhagic), the level of risk of subsequent vascular events (stroke, myocardial infarction), the ethnic origin (Asian, European, African), etc, in addition, or more likely in interaction with the nature of drug treatment, may explain variations of the size of benefit.
The greatest progress to expect from pharmacological cardiovascular prevention may come from the optimization of the use of available drugs, rather than the discovery of the new miraculous drug. We will not always ignore that people with a given phenotype profile, and soon a given genotype, will not have their blood pressure lowered with drug A but with drug B, even if they tolerate both as well, and if A appears more efficacious, due to a greater placebo effect. But more importantly, if A lowers blood pressure better than B, C will be the most efficient to reduce overall cardiovascular risk, even if blood pressure-lowering is between that of A and B.
To access such clinically relevant knowledge, we need to explore extensively data that accumulate from epidemiology and randomized controlled trials, but also to obtain new data from compared fixed-dose strategy in a variety of populations, looking accurately at genome, environment, and their interactions.
Carter AB. Hypotensive therapy in stroke survivors. Lancet. 1970; i: 485–489.
Boissel JP, Gueyffier F, Boutitie F. Risk reduction for stroke and coronary events. Lancet. 2002; 359: 1249.
Boutitie F, Gueyffier F, Pocock SJ, Fagard R, Boissel JP, on behalf of the INDANA Project Steering Committee. Relationship between blood pressure and mortality in hypertensive patients randomised in non-active controlled clinical trials: a meta-analysis on individual patient data. Ann Intern Med. 2002; 136: 438–448.