Optimal diastolic blood pressure range in intensive systolic targets: Resolving the controversy between observational and intention‐to‐treat analyses

The Systolic Blood Pressure Intervention Trial (SPRINT) randomized 9361 non‐diabetic patients with systolic blood pressure (BP) ≥130 mm Hg and a high cardiovascular risk profile to intensive therapy targeting a systolic BP <120 mm Hg versus standard therapy targeting a systolic BP <140 mm Hg.1 SPRINT was prematurely terminated after a median follow‐up of 3.26 years because of an impressive cardiovascular benefit of intensive BP lowering. Compared with the standard arm, intensive therapy provoked a 25% reduction in the occurrence of the primary composite outcome of myocardial infarction, other acute coronary syndromes, stroke, heart failure, or cardiovascular death.1 This landmark trial opened new horizons in the management of hypertension and provided scientific basis for the reappraisal of BP targets in the 2017 American Heart Association/American College of Cardiology (AHA/ACG) guidelines.2 It has to be noted, however, that the concept of intensive BP lowering is not universally and unconditionally accepted.3 For example, the 2018 European Society of Hypertension/European Society of Cardiology (ESH/ESC) guidelines are more conservative and recommend that BP should be targeted to levels <130/80 mm Hg in most patients provided that intensification of antihypertensive therapy is well tolerated.4

Reluctance in adoption of intensive BP targets in daily clinical practice is not only an issue of diverse health care policies around the world.3, 5 It could be argued that discrepancies across guideline recommendations may be explained by the presence of gaps or uncertainties in the currently available evidence. In this direction, SPRINT demonstrated an impressive cardiovascular benefit of intensive systolic BP reduction, but it is not yet fully clear whether a low baseline or a low achieved diastolic BP modifies or attenuates this benefit. From a pathophysiological standpoint, the issue of diastolic hypotension is of major clinical relevance because a low pressure during diastole may lead to myocardial hypoperfusion and promote end‐organ damage, particularly in susceptible patients with left ventricular hypertrophy or clinically overt coronary artery disease. This issue becomes even more important in light of large‐scale observational studies showing a J‐shaped or U‐shaped association of achieved diastolic BP with the risk of cardiovascular events and all‐cause mortality.6, 7

In this issue of the Journal of Clinical Hypertension, Sobieraj et al8 report a non‐prespecified, post hoc analysis of 1470 SPRINT participants with prior cardiovascular disease aiming to explore the association of achieved diastolic BP with clinical outcomes. The predictor in this analysis was defined as the mean on‐treatment (achieved) diastolic BP from the follow‐up visit at 6 months till the trial completion. The outcome of this analysis was similar to the primary composite cardiovascular outcome of main SPRINT analysis. By applying hazard ratio plots with cubic spline regression, the analysis showed a U‐shaped relationship between achieved diastolic BP and the risk for the composite cardiovascular outcome.8 Participants with achieved diastolic BP ranging from 68.6 to 78.6 mm Hg appeared to carry the lowest risk for the composite cardiovascular outcome. When the study population was stratified into subgroups according to the level of achieved diastolic BP, 33.8% of participants were within the “optimal” diastolic range, whereas 51.5% and 14.9% of participants had an achieved diastolic BP lower and higher than the optimal range, respectively. Notably, those with a lower than “optimal” achieved diastolic BP were older in age, had more common history of chronic kidney disease (CKD), and had higher pulse pressure (PP) levels at baseline.8

Strength of this post hoc analysis is the use of the appropriate methodology of restricted cubic splines in Cox regression hazard analysis that enabled the identification of a non‐linear pattern in the relationship of achieved diastolic BP with the composite cardiovascular outcome. In addition, the analysis was adjusted for several confounding factors, including age, sex, body mass index, smoking status, on‐treatment systolic BP, and presence of CKD. It has to be noted, however, that interpretation of the above results should take into consideration the observational nature of the study design and the inherent limitations of an analysis based on achieved BP that contrasts with intention‐to‐treat analyses from randomized trials directly comparing a lower versus a standard BP target.9, 10, 11 In fact, this post hoc analysis did not follow the principle of intention‐to‐treat comparison between the intensive arm and standard arm of SPRINT. Subsequently, it could be argued that despite the careful multivariate adjustment, the analysis may be still subjected to biases originating from residual confounding. In other words, the observed association of lower achieved diastolic BP with higher risk for the composite cardiovascular outcome may be not directly attributable to the level of achieved diastolic BP per se, but to other confounding factors that might have resulted in a decline in diastolic BP and were not inserted in multivariate models. Since participants in the subgroup of lower than “optimal” achieved diastolic BP were older and had higher PP levels at baseline, one possible source of residual confounding might be the underlying process of arterial stiffening.12 In the absence of aortic pulse wave velocity data at baseline for additional adjustment, the authors cannot confirm or refute our hypothesis.

In an earlier prespecified post hoc analysis of SPRINT, Beddhu et al9 explored the role of baseline diastolic BP as a risk modifier, hypothesizing that a low diastolic BP at baseline would attenuate the cardiovascular benefit of intensive systolic BP reduction. In line with the results of the present study, baseline diastolic BP exhibited a U‐shaped association with the risk of the primary composite cardiovascular outcome of SPRINT.9 When the analysis followed the intention‐to‐treat principle, however, the level of diastolic BP at baseline did not modify the cardiovascular protection offered by intensive BP lowering. In the lowest diastolic BP quintile (mean baseline diastolic BP: 61.0 ± 5.0 mm Hg), the hazard ratio for the composite cardiovascular outcome between the intensive and standard arm was 0.78 (95% confidence interval: 0.57‐1.07). Similarly, in the upper 4 quintiles (mean baseline diastolic BP: 82.0 ± 9.0 mm Hg), the hazard ratio was 0.74 (95% CI: 0.61‐0.90) with an interaction P value of 0.78.9 On this basis, the authors of this post hoc analysis concluded that a low diastolic BP should not be an impediment to the intensive treatment of hypertension.

Additional support to the notion that observational analyses based on achieved BP provide differential results from intention‐to‐treat analyses of targeted BP is provided by the African American Study of Kidney Disease and Hypertension (AASK).13 In the AASK trial, 1084 African Americans with hypertension and CKD were randomized to a mean arterial pressure (MAP) target of 102‐107 mm Hg versus a lower MAP goal of <92 mm Hg. In intention‐to‐treat analysis, compared with the usual MAP target, intensive BP lowering did not improve the primary composite renal outcome of ≥50% decline in glomerular filtration rate (GFR), requirement for dialysis or death (risk reduction: 2%; 95% confidence interval: −22% to 21%).13 By contrast, in a secondary analysis stratified according to the level of achieved MAP, each 10‐mm Hg increment in MAP was associated with 0.35 mL/min/1.73 m² (95% confidence interval: 0.08‐0.60 mL/min/1.73 m²) faster GFR decline during follow‐up as well as with 17% (95% confidence interval: 5%‐32%) higher risk for the aforementioned composite renal outcome.10

How the above methodological issues should be translated into clinical practice recommendations is the question that we should answer. From a methodological standpoint, randomized trials comparing directly a lower versus a standard BP target provide more unbiased evidence, whereas observational studies exploring risk associations of achieved BP levels are subjected to confounding from other factors that may determine the level of achieved BP. These confounding factors may be related to the level of illness, presence of comorbid conditions, and adherence to therapy.10, 11 The present study by Sobieraj et al incorporated data from a high‐risk SPRINT subpopulation with prior cardiovascular disease and showed a U‐shaped association of achieved diastolic BP with cardiovascular morbidity and mortality. Although risk associations observed in this analysis are not directly causal or may reflect the prognostic value of other underlying processes that resulted in a decline in diastolic BP (ie, the arteriosclerotic process),14 clinicians should take into consideration the level of diastolic BP when intensifying antihypertensive therapy. A lower achieved diastolic BP may reflect a poor vascular health of their patients, implying that intensified treatment in such patients should be careful and with particular attention to the tolerability and occurrence of adverse events.

CONFLICT OF INTEREST

The authors have no conflicts of interest to disclose.