Lowering blood pressure in acute stroke: lessons learnt from the SCAST trial

Keywords::

• acute stroke
• blood pressure
• candesartan
• outcome
• randomized controlled trial

The management of high blood pressure (BP) during the acute phase of stroke has remained an enigma for more than 25 years, exemplified by a debate published back in 1985 [1]. While observational studies show that high BP is independently associated with a poor outcome, pathophysiology argues that lowering BP will reduce cerebral blood flow (CBF) in the face of dysfunctional autoregulation. Thus, lowering BP might improve or worsen outcome, depending on whether epidemiology or pathophysiology is more important. Such uncertainty in this common complication of stroke (80% of patients are affected) demands large definitive randomized controlled trials, and the first such study has recently been published. The 2029-patient Scandinavian Candesartan Acute Stroke Trial (SCAST) examined whether careful BP lowering with oral candesartan, an angiotensin receptor antagonist (ARA), was safe and beneficial in patients with acute ischemic stroke (85%) or primary intracerebral hemorrhage (PICH; 14%), and raised BP (systolic blood pressure [SBP]: ≥140 mmHg; mean: 171/90 mmHg) [2]. As patients had to be able to swallow, baseline stroke severity was mild (Scandinavian Stroke Scale: 41). The mean time to recruitment was 18 h up to a maximum of 30 h.

The results for the coprimary outcomes in SCAST were surprising; in comparison with placebo, treatment with candesartan was associated with a nonsignificant worse functional outcome (assessed using the modified Rankin Scale) at 6 months (adjusted common odds ratio: 1.17; 95% CI: 1.00–1.38; p = 0.048) and no reduction in vascular events (adjusted hazard ratio: 1.09; 95% CI: 0.84–1.41; p = 0.52) [2] (as there were two coprimary outcomes, the trend to a worse functional outcome is nonsignificant, since p > 0.025). So, why was SCAST carried out, and how can these results be explained?

Scandinavian Candesartan Acute Stroke Trial was the natural follow-on from an earlier trial, Acute Candesartan Cilexetil Therapy in Stroke Survivors (ACCESS) [3], which also assessed candesartan in acute ischemic stroke. Although the primary outcome in ACCESS – function measured as the Barthel Index at 3 months – was neutral (with a negative trend), a 52% reduction in a composite vascular outcome (comprising death, cerebrovascular and cardiovascular events at 12 months) was observed, a finding that captured the imagination of many physicians. The results of ACCESS [3] for vascular events (but not functional outcome) are divergent with those of SCAST however, notably, it was six-times smaller and so its results probably reflect chance – the trial was stopped early because of the significant reduction in vascular events, a secondary outcome, and it only included 339 patients out of a planned 500. Furthermore, BP did not differ between candesartan and placebo, and the difference in vascular events only occurred after 7 days when all patients were taking open-label candesartan.

One other ARA (losartan) and several angiotensin-converting enzyme inhibitors (ACE-I, captopril, lisinopril and perindopril) have been assessed in patients with acute or recent stroke [4–7], but all the trials were small and too underpowered to assess effects on function. Other antihypertensive drug classes have also been assessed, including β-receptor antagonists, calcium channel blockers and nitrates. Although most trials were small and neutral, notable negative (i.e., hazardous) effects on functional outcome were seen with intravenous nimodipine in the Intravenous Nimodipine West European Stroke Trial (INWEST) [8], and atenolol or propranolol in β-Blocker Evaluation of Survival Trial (BEST) [9].

As highlighted in the 1985 BP debate [1], a concern regarding lowering BP in acute stroke is that this might lead to a reduction in cerebral perfusion. A number of small studies have assessed the effect of various antihypertensives on CBF and these have shown that BP can be lowered independently of effects on brain perfusion, as seen individually for transdermal glyceryl trinitrate, and in aggregate in a meta-analysis involving a number of different antihypertensive drugs [10,11]. Regarding candesartan, there are no data on CBF, but neither losartan nor telmisartan, other ARAs, appear to adversely affect cerebral perfusion in recent stroke [Bath PMW, Unpublished Data] [4]. Nevertheless, candesartan was associated with increased stroke progression in SCAST [2], an effect that would be predicted if the drug did reduce CBF.

Looking beyond candesartan, it is conceivable that ARAs as a class may not be effective in preventing stroke or modulating its outcome; in addition to SCAST, ARAs were ineffective in recent large trials including Telmisartan Randomized Assessment Study in ACE-Intolerant Subjects with Cardiovascular Disease (TRANSCEND; telmisartan for patients with cardiovascular disease, or diabetes with end-organ damage, and who were intolerant of an ACE-I) and Prevention Regimen for Effectively Avoiding Second Strokes (PRoFESS; telmisartan post-ischemic stroke) [12,13]. Neutral effects of telmisartan were also seen on function and recurrence in a subset of PRoFESS patients who were recruited within 3 days of stroke onset [14]. Nevertheless, losartan reduced vascular morbidity and mortality, and stroke alone, in the Losartan Intervention For Endpoint Reduction (LIFE) trial when compared with atenolol [15], so the role of ARAs in stroke prevention remains unclear.

Other potential explanations for a trend to hazard can probably be excluded. First, an imbalance in prognostic factors at baseline was not present; such differences have ‘damaged’ the results of several previous acute stroke trials. Second, the BP reduction was very modest (just 5/2 mmHg), and not as extreme as seen with nimodipine in INWEST [8], so excessive BP reduction is unlikely to be the overall explanation for the SCAST results. Nevertheless, in additional data presented at the 2011 European Stroke Conference in May, the SCAST trialists showed that the relationship between early deterioration (progression, recurrence and symptomatic hypotension) and change in BP during treatment (with adjustment for baseline prognostic factors) was U-shaped, with a poor outcome more likely if BP rose or fell considerably [Sandset E, Pers. Comm.]. Therefore, it is conceivable that a small number of patients in SCAST had a large fall in BP and this contributed to a worse outcome. Third, there were no noteworthy differential effects in the prespecified subgroups so a positive effect in one subgroup was not reversed in another. In particular, there were no treatment interactions by stroke type (ischemic stroke versus PICH), baseline SBP (140–160, 160–180, 180–200 and >200 mmHg), or time to treatment (<6, 6–11, 12–23 and >24 h).

It is always possible that the results of SCAST are simply wrong, although this seems unlikely. First, the trial was fairly large in size with good statistical power. However, SCAST stopped short of its intended recruitment of 2500 patients because of slow recruitment and completion of the research grant. Although the failure to reach the original sample size might have damaged the statistical power, the shortfall was made up by using a more efficient statistical analysis of functional outcome based on ordinal rather than binary analysis [16]. Ironically, it is possible that this newer statistical approach would have resulted in the trial being overtly negative had the full sample size been realized. Second, the results were consistent across a wide range of outcomes, including stroke recurrence, myocardial infarction and symptomatic venous thromboembolism; in each case, the risk ratio showed a trend against candesartan. Finally, the results for functional outcome are consistent with the results of ten previous modestly sized trials of BP lowering, as shown in a meta-analysis published with the trial (odds ratio for a poor outcome: 1.04; 95% CI: 0.95–1.14) [2].

The results of SCAST, and the accompanying meta-analysis, might suggest that the question of BP lowering in acute stroke is now closed. However, the meta-analysis is neutral (even if the trend is negative) and it is conceivable that alternative methods of lowering BP might have a different effect. Two other large trials involving >2500 patients are ongoing [17]. The Efficacy of Nitric Oxide in Stroke (ENOS) trial is assessing the safety and efficacy of lowering BP with transdermal glyceryl trinitrate, and of continuing versus stopping prestroke antihypertensive medication, in 3500 patients with either acute ischemic or PICH [18]. Nitrates may be highly appropriate for stroke since they release nitric oxide, a potential neuroprotectant and key driver for collateralization via pial artery dilatation. The second trial, Intensive Blood Pressure Reduction in Acute Cerebral Haemorrhage Trial (INTERACT)-2, is comparing intensive BP lowering (achievement SBP <140 mmHg within 1 h) versus guideline BP lowering (maintaining SBP <180 mmHg) in 2800 patients with hyperacute PICH, and has a similar design to its predecessor, INTERACT [19]. Investigators may choose which intervention(s) to use to lower BP. A smaller trial, Antihypertensive Treatment in Acute Cerebral Hemorrhage (ATACH)-2, is similar in design to INTERACT-2 and is ongoing in the USA [20]. The lessons from SCAST relevant to investigators in ENOS and INTERACT-2 are to recruit patients with strokes ranging in severity (i.e., not just mild) and BP as early as possible, and to use optimal ways for statistically analyzing functional outcome (i.e., take advantage of all seven levels of the modified Rankin Scale) [16].

Hopefully, ENOS and INTERACT-2 will bring closure to this longstanding question of how to manage high BP in acute stroke. But in the meantime, if clinicians cannot enter their patients into these trials, then SCAST suggests that active lowering of BP is not warranted in the first week after acute stroke. This recommendation applies to all of the subgroups assessed in SCAST, including by stroke type, time to treatment and BP level.

Financial & competing interests disclosure

Philip Bath is Stroke Association Professor of Stroke Medicine. Philip Bath was a member of the SCAST Trial Steering Committee and received travel grants from AstraZeneca to attend committee meetings. He is Chief Investigator of ENOS, and was on the executive or steering committees of INTERACT and PRoFESS. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.