Blood pressure and outcomes in clinical trials

Hypertension treatment studies have utilized conventional sphygmomanometry to measure the effect of different drug therapies on brachial blood pressure and morbidity and mortality. When different treatment regimens are compared, differences in brachial blood pressure have been apparent in many trials, especially early in the study [1,2]. These differences have tended to diminish over time as additional blood pressure-lowering drugs have been added in an endeavour to achieve the blood pressure goal. When one treatment regimen has proven to be superior to another, debate has ensued as to whether such differences in outcome are explained by the differences in achieved blood pressure, or whether they are related to the benefits to the treatment regimen ‘beyond blood pressure’. The reality is that the data provided concerning ‘in trial’ blood pressures are invariably inadequate to inform this debate. Put simply, our understanding of the integrated blood pressure load experienced by patients over the duration of a trial cannot be accurately estimated from simple and infrequent office recordings of blood brachial pressure. Statisticians have gone further and used this inadequate data, together with a bewildering array of techniques and assumptions, to ‘correct’ for the differences in office blood pressure. Such a concept has been more useful to the pharmaceutical industry than to clinical science. Even more remarkable is the tendency to focus on differences in blood pressure at the end of the trial, when they are invariably at their smallest. It appears to have escaped the attention of those with enthusiasm for ‘end of trial’ blood pressures that these are recorded from patients who have survived the trial! We are surely most interested in the blood pressure profiles of those who suffered a cardiovascular event or death earlier in the trial, when blood pressure differences are more likely to have been greater. Thus, there is much work to be done in the design and assessment of blood pressure in future trials to ensure that data about the most efficacious property of a drug or drug regimen being tested (i.e. its ability to lower blood pressure) is more thoroughly and accurately studied. The time has surely come for an international group to set a clinical trial standard for future trials that mandates: (i) standardized methods for blood pressure measurement; (ii) a minimum frequency of office blood pressure recordings; (iii) a recommended percentage of patients who should undergo 24-h ambulatory and/or home blood pressure recordings to better inform the debate on blood pressure profiles on different treatment regimens; and (iv) clear and unequivocal methods for the reporting of blood pressure data in relation to clinical events In this issue of the journal, Hirata et al. [3] have introduced another blood pressure parameter that might ultimately influence clinical outcomes. In recent years, there has been a resurgence of interest in the possibility of using alternatives to simple brachial blood pressure measurements to non-invasively estimate circulatory pressures by applanation of tonometry and pulse-wave analysis. Hirata et al. [3] have used these techniques to demonstrate, albeit in an acute study, that two different blood pressure-lowering drugs (an angiotensin-converting enzyme inhibitor and a β-blocker), produce quite marked differences in estimated aortic blood pressures, relative to their ability to reduce brachial blood pressure [3]. The authors speculate that this differential effect of different drugs on central arterial haemodynamics could drive differences in clinical outcomes, beyond brachial blood pressure, but still be dependent on blood pressure per se. Their study should be qualified by a number of caveats: (i) it is an acute study that provides no information about what would happen in the real world of a clinical trial when the addition of other blood pressure-lowering drugs could modify the differences in central blood pressures that they have measured. This is most relevant because of the potential of different drugs regimens to have different effects on resistance vessel structure, which would influence arterial wave reflection, the timing of which is a powerful determinant of central blood pressure as measured by the techniques used in their study. (ii) Their study design resulted in significant differences in brachial blood pressure, which potentially confounds the interpretation of the central blood pressure data because of the influence of brachial blood pressure on conduit artery stiffness and wave form characteristics. (iii) Their study depends on the estimation of central arterial pressures using applanation tonometry of the radial pulse and pulse-wave analysis, a technique and approach that has been enthusiastically advocated by some but not all [4]. Nevertheless, despite the seemingly similar effects of drugs on office brachial blood pressure, the study of Hirata et al. [3] does highlight that different classes or blood pressure drugs have the capacity to profoundly influence the characteristics of the arterial wave form and the timing of its reflection, which could thereby generate differential drug effects on central aortic pressure. It has been argued, not unreasonably, that such effects on central arterial haemodynamics are likely to be more relevant to clinical outcomes than pressure in the brachial artery [5]. Our view is that this technology has too many confounders to be of clinical use when assessing individual patients. Nevertheless, the use of applanation tonometry and pulse-wave analysis could provide important mechanistic insights into the differential effects and benefit of different treatment regimens in clinical trials. In this regard, we have been conducting the Conduit Artery Functional Endpoint (CAFÉ) [6] study as a sub-study within the Anglo Scandinavian Cardiac Outcomes Trial (ASCOT) [7]. ASCOT has compared a conventional blood pressure-lowering regimen (atenolol ± bendroflumethazide K) with a more contemporary regimen (amlodipine ± perindopril) on cardiovascular outcomes in almost 20 000 patients. After more than 5 years, ASCOT was recently stopped early on the advice of the trial's data safety monitoring board because of significant cardiovascular outcome benefits in favour of the amlodipine ± perindopril regimen. The CAFÉ sub-study began recording pulse-wave analysis data on over 1500 patients within five UK ASCOT centres after completion of treatment escalation towards goal blood pressure, thereby minimizing any potential differences in brachial blood pressures. The CAFÉ study investigators have also recorded longitudinal data, with repeated pulse-wave analysis to assess the stability of the relationship between brachial and derived central arterial wave form and blood pressure parameters. Because there is no reason to assume that the patients recruited into the CAFÉ study are any different demographically from the populus of ASCOT, we will be able to relate any differences in central blood pressures and wave form characteristics between the two treatment regimens to differences in cardiovascular outcomes. Moreover, ASCOT has also evaluated statin therapy (atorvastatin) in a factorial design. Thus, the CAFÉ study will test the hypothesis that, beyond brachial blood pressure, different drug regimens can differentially influence central arterial haemodynamics and pressures. Furthermore, CAFÉ will examine the impact of statin therapy on the relationship between peripheral and central blood pressures and any potential interaction between statins and the blood pressure-lowering regimens. It is only by applying new technology prospectively, to well designed and well conducted large-scale outcome trials, that we will ultimately gain novel insights into mechanisms that might underpin differences in clinical outcomes that apparently go beyond predictions based on office blood pressures alone.