Intensive Blood Pressure Lowering in Patients With and Patients Without Type 2 Diabetes: A Pooled Analysis From Two Randomized Trials

Hypertension and type 2 diabetes mellitus (T2DM) are both highly prevalent in the Western world and increase the risk of cardiovascular disease (CVD) (1–3). The benefit of a systolic blood pressure (SBP) target of <120 mmHg vs. <140 mmHg on the reduction of cardiovascular events was assessed in two large randomized clinical trials (4,5). The Action to Control Cardiovascular Risk in Diabetes Blood Pressure (ACCORD-BP) trial and the Systolic Blood Pressure Intervention Trial (SPRINT) assessed the effect in persons with T2DM at high risk (ACCORD-BP) and in persons without T2DM at high risk (SPRINT).

ACCORD-BP did not demonstrate a statistically significant benefit of intensive blood pressure lowering on the primary composite end point, while SPRINT did. The current guidelines recommend an SBP target of <140 mmHg in patients with T2DM based on ACCORD-BP, while the results of SPRINT have not been incorporated in the guidelines (6–9).

The difference in outcomes in ACCORD-BP and SPRINT have been attributed to differences in study design, sample size, interaction between T2DM and the intervention, or play of chance (5). Furthermore, the primary composite end point of SPRINT consisted of two additional components, unstable angina and acute cardiac decompensation, compared with ACCORD-BP (myocardial infarction, stroke, and cardiovascular death). This increased the event rate and possibly the treatment effect of intensive blood pressure lowering in SPRINT.

The objectives of this study were as follows: 1) to evaluate the treatment effect of intensive blood pressure lowering in patients with T2DM and patients without T2DM using the individual patient data of ACCORD-BP and SPRINT and using the same composite primary end points and follow-up durations and 2) to assess the interaction between both T2DM and baseline CVD risk and the treatment effect of intensive blood pressure lowering, as patients with T2DM have a higher 10-year risk of CVD than patients without T2DM.

The design, rationale, and outcomes of ACCORD-BP and SPRINT have previously been described including detailed inclusion and exclusion criteria (4,5,10,11) (online study protocols: https://biolincc.nhlbi.nih.gov/studies/accord/ and https://biolincc.nhlbi.nih.gov/studies/sprint_pop/). In brief, both studies were randomized, open-label multicenter trials that assigned participants at high risk of cardiovascular events with an SBP between 130 mmHg and 180 mmHg to an SBP target of either <120 mmHg (the intensive treatment group) or <140 mmHg (the standard treatment group). Action to Control Cardiovascular Risk in Diabetes (ACCORD) enrolled 10,251 participants with T2DM in a 2-by-2 factorial design in the U.S. and Canada between the years 2001 and 2005. All participants were randomly assigned to either intensive or standard glycemic control. Additionally, a subgroup of 4,733 participants were also randomly assigned to either intensive or standard blood pressure control: the ACCORD-BP trial, which was used for this analysis. SPRINT randomized patients directly to intensive or standard blood pressure lowering and recruited patients in the U.S. between 2010 and 2013. Blood pressure and pulse were measured three times after 5 min of seated rest at each clinic visit in both trials with an automated device (Model 907; Omron Healthcare, Lake Forest, IL) and averaged. ACCORD was designed to have 94% power to detect a 20% reduction in the rate of the primary outcome in the intervention arm and SPRINT to have 89% power. Both studies were sponsored by the U.S. National Heart, Lung, and Blood Institute and approved by the institutional review boards of the participating study sites. The use of both data sets was approved by the institutional review board of the Academic Medical Center, Amsterdam, the Netherlands.

Participants were eligible for ACCORD-BP if they were diagnosed with T2DM, had a glycated hemoglobin level ≥7.5%, and were ≥40 years of age with a history of CVD or ≥55 years of age with anatomical evidence of a substantial amount of atherosclerosis, albuminuria, left ventricular hypertrophy, or at least two additional risk factors for CVD (dyslipidemia, hypertension, smoking, or obesity). Participants were also required to have an SBP between 130 mmHg and 180 mmHg while taking three or fewer antihypertensive medications.

Eligible SPRINT participants were required to be ≥50 years old, without T2DM, and to have an SBP between 130 mmHg and 180 mmHg and an increased risk of cardiovascular events, defined as clinical or subclinical CVD other than stroke, chronic kidney disease with an estimated glomerular filtration rate between 20 and 60 mL/min/1.73 m² body surface area, a 10-year risk of CVD ≥15% on the basis of the Framingham risk score, or age ≥75 years.

The ACCORD-BP composite primary end point consisted of myocardial infarction, stroke, and cardiovascular death. The SPRINT composite primary end point consisted of the same end points plus acute coronary syndrome not resulting in myocardial infarction and acute decompensated heart failure. After a mean follow-up of 5 years, the hazard ratio for the occurrence of cardiovascular events in ACCORD-BP was 0.88 (95% CI 0.73–1.06), nonsignificantly in favor of intensive blood pressure control. SPRINT enrolled 9,361 participants without T2DM in the U.S. After a mean follow-up of 3.26 years, the study was terminated early owing to the significantly lower event rate in the intensive treatment arm. The hazard ratio for the occurrence of cardiovascular events was 0.75 (95% CI 0.64–0.89), significantly in favor of intensive blood pressure control. The secondary outcomes acute heart failure, cardiovascular mortality, and all-cause mortality were also significantly reduced. A committee blinded to treatment assignment adjudicated the clinical outcomes in each study.

The data of ACCORD-BP and SPRINT were merged. In the ACCORD-BP subset, the SPRINT composite end point was constructed by combining the ACCORD-BP primary end point (myocardial infarction, stroke, and cardiovascular death) with unstable angina and acute decompensated heart failure events. Additionally, the follow-up duration was truncated in ACCORD-BP to the median follow-up duration of SPRINT: 3.26 years (Supplementary Figs. 1 and 2). The 10-year risk score of CVD was calculated in the subset of patients without a history of cardiovascular events (12). The safety end point from the original studies was used and consisted of serious adverse events related to blood pressure–lowering therapy.

For the analysis, patients enrolled in ACCORD-BP were considered to have T2DM and those enrolled in SPRINT were not considered to have T2DM. Baseline characteristics were compared using Student t test, Mann-Whitney U test, or Fisher exact test when appropriate. Time-to-event analysis of the composite primary and secondary end points in the two treatment groups was performed with Cox proportional hazards regression. The status of T2DM (yes vs. no) was added to the Cox model, and subsequently an interaction term between T2DM and treatment allocation was added. The model fit with and without the interaction term was compared using Akaike information criterion (AIC) and ANOVA. The same was done for the baseline 10-year risk score of CVD grouped per 20% increments in patients without a history of CVD. Post hoc power calculations were performed using the SPRINT composite end point and a median follow-up of 3.26 years. For serious adverse events related to blood pressure–lowering therapy, the proportions were calculated using the crude numbers from the main outcomes paper, as these data were not part of the public ACCORD data set. P values <0.05 were considered statistically significant. All statistical analyses were conducted in R Studio and R version 3.3.1 (13).

The pooled cohort of ACCORD-BP and SPRINT consisted of 14,094 patients (Supplementary Fig. 3). Table 1 shows baseline characteristics of participants per study and per treatment allocation. At baseline, the two studies differed significantly except for baseline SBP, smoking, and aspirin use. The distribution of 10-year risk score of CVD was different in the two studies (Supplementary Fig. 4). Patients in the pooled cohort had a mean age of 66 ± 8.9 years, and 39.7% were female.

The randomization remained intact when the baseline characteristics were judged both by absolute numbers and statistically. The mean SBP at baseline was 139.5 ± 15.6 mmHg and the mean diastolic blood pressure was 77.4 ± 11.5 mmHg. In the pooled cohort, 33.6% of patients had T2DM and the median 10-year risk of CVD was 25% (quartiles 17, 26).

The incidence of the primary end point in the pooled cohort was 7.3% during a median follow-up of 3.26 years. The event rate was 8.0% in the intensive treatment group vs. 6.6% in the standard treatment group. The hazard ratio for the primary composite end point was 0.82 (95% CI 0.73–0.93), P = 0.002 (Fig. 1). The individual secondary end points were not significantly different between the treatment groups, except acute heart failure events (Fig. 2). Cardiovascular mortality and all-cause mortality were nonsignificantly lower. The post hoc calculated power in the pooled cohort for the primary composite end point was 88.9%.

The event rate in patients with T2DM was higher than in patients without T2DM (9.8% vs. 6.0%, P < 0.001) (Fig. 3). In patients with T2DM, the event rate was lower in the intensive group, 10.3% vs. 9.4%, but this difference was nonsignificant (P = 0.32). The event rate in patients without T2DM was also lower in the intensive group: 6.8% vs. 5.2% (P < 0.001). The interaction between the intervention and T2DM was nonsignificant when an interaction term was added to the Cox model (P = 0.13). Also, the model fit did not improve when an interaction term was forced in the Cox model (AIC 19,205 vs. 19,205, P = 0.13), indicating that there is no evidence that intensive blood pressure lowering to 120 mmHg has a different relative effect in patients with T2DM compared with patients without T2DM. The post hoc calculated power in ACCORD-BP using the SPRINT composite end point was 17.5% vs. 91.6% in SPRINT at 3.26 years median follow-up duration. With the observed treatment effect in ACCORD-BP, ∼50,000 patients would be needed to reach 90% power.

In patients without a history of cardiovascular events, those with T2DM had a significantly higher calculated baseline risk of CVD compared with those without T2DM (median 32% vs. 22%, P < 0.001). The baseline risk scores were stratified per 20% increment in 10-year risk (Supplementary Fig. 5). None of the subgroups for risk had a significant interaction with treatment allocation. Also, the model fit did not improve when an interaction term was forced in the Cox model (AIC 10,243 vs. 10,248, P = 0.84). This indicates that there is no evidence that the relative treatment effect of intensive blood pressure lowering depends on the 10-year risk of CVD. In the subgroup of patients with a history of CVD (n = 3,470), intensive treatment resulted in a nonsignificant reduction of the primary end point (13.8% vs. 12.5%, P = 0.29).

Serious adverse events related to blood pressure–lowering therapy occurred more often in the intervention group (6.5% vs. 4.6%, P < 0.001). The interaction between T2DM and intensive blood pressure lowering for safety events could not be investigated, as these data were not publicly available for all patients.

Our study suggests that there is no difference in the relative treatment effect of intensive blood pressure lowering in patients with and patients without T2DM at increased risk of CVD and with an SBP between 130 mmHg and 180 mmHg. Second, when the broader SPRINT composite primary end point that additionally included unstable angina and heart failure is applied to ACCORD-BP, the primary end point remains nonsignificantly in favor of intensive blood pressure lowering. Third, there is no evidence for the presence of an interaction between baseline risk of CVD as calculated with the Framingham risk score in primary prevention patients and the relative treatment effect of intensive blood pressure lowering on the primary end point.

In the current analysis, we sought to eliminate the effect of difference in study design by harmonizing the composite primary end point and the follow-up duration in the two studies. The main outcomes of ACCORD-BP remained unchanged when the end point and follow-up duration of SPRINT were applied. The post hoc calculated power in ACCORD was low owing to the smaller relative risk reduction and sample size.

By combining the data sets of two studies that applied the same intervention for the same indication to patients with and patients without T2DM, we were able to investigate whether an interaction exists between intensive blood pressure lowering and T2DM. In this analysis, there was no evidence for such an interaction. This implies that the primary outcome of the pooled cohort (hazard ratio 0.82 [95% CI 0.73–0.93]) should not be interpreted differently for patients with T2DM and patients without T2DM (Fig. 2). The presence of T2DM may be a marker of increased baseline risk of CVD, which in turn could affect the relative treatment effect of intensive blood pressure lowering. Therefore, the presence of an interaction between 10-year risk score in patients without a history of CVD and intensive blood pressure lowering was assessed, but no significant interaction was found.

Several analyses have addressed the question of whether the effect of intensive blood pressure lowering is different in patients with versus patients without T2DM (14–16). These analyses did not have access to individual patient data. Therefore, overall treatment effects could not be pooled and differences in baseline characteristics, outcomes, and follow-up duration could not be taken into account. A recent meta-analysis that examined differences in outcomes in patients with and without T2DM according to attained blood pressure levels concluded that, in contrast to patients without T2DM, there is little or no further benefit in lowering SBP below 130 mmHg. Our analysis is the first to pool individual patient data from the only two randomized clinical trials that assessed the effect of targeting an SBP <120 mmHg vs. <140 mmHg. After harmonization of the composite primary end point and the follow-up duration, we were able to calculate the treatment effect on the same end point over the same follow-up duration. The individual patient data provide a more precise estimate of the treatment effect. We show that, in contrast to the divergent benefit below an SBP level of 130 mmHg in a recent meta-analysis (16), the benefit of lowering blood pressure targeted at an SBP <120 mmHg is not different in patients with and without T2DM. In addition, this analysis allowed us to answer the question of whether baseline cardiovascular risk (expressed as the 10-year risk of CVD) is an effect modifier of the benefit of intensive blood pressure lowering, as patients with T2DM have a higher 10-year cardiovascular risk compared with those without. We did not find such a modifying effect caused by baseline cardiovascular risk.

Observational data have suggested that aggressive blood pressure lowering may increase cardiovascular events, indicating the potential presence of a “J-curve” phenomenon in some high-risk patients. However, in this pooled cohort, patients were experimentally exposed to different blood pressure targets and care providers were not restricted to specific blood pressure–lowering drugs. Possibly, the increase of harm in trials such as the ROADMAP study (Randomized Olmesartan and Diabetes Microalbuminuria Prevention), ALTITUDE (Aliskiren Trial in Type 2 Diabetes Using Cardiovascular and Renal Disease Endpoints [Core and Extension Phases]), and ONTARGET (Ongoing Telmisartan Alone and in Combination with Ramipril Global Endpoint Trial) may be due to double renin-angiotensin system blockade instead of intensive blood pressure lowering (17–19).

The latest guidelines from the American Heart Association, the American Diabetes Association, and the European Society of Cardiology recommend an SBP target of <140 mmHg in patients with T2DM, while <130 mmHg may be appropriate for certain individuals, such as younger patients (6–8). The results of SPRINT have not been incorporated in these guidelines, as the results were published recently. We found that based on the available data, there is no evidence that T2DM patients do not benefit from intensive blood pressure lowering. The reduction of the primary end point and the increase in serious adverse events related to intensive blood pressure lowering is similar. Therefore, tailoring the treatment for the individual patient seems appropriate based on the goals of the individual patient, life expectancy, and the likelihood of the adverse events to occur.

This study is a post hoc analysis, in which the results of two independent, randomized, open-label studies are pooled and the end points harmonized. Pooling of the studies was deemed appropriate based on both clinical and methodological grounds, with the inclusion and exclusion criteria of both studies and the study intervention taken into account. The null hypothesis was that the treatment effect is not different in patients with T2DM or with a higher 10-year risk score. This hypothesis was not rejected in this analysis, and the results should be interpreted as such. Therefore, this analysis does not exclude a difference in the treatment effect of intensive blood pressure lowering between patients with T2DM and patients without T2DM. It is important to note that in ACCORD-BP an observer was present during blood pressure measurements, whereas in SPRINT no observer was present (20). The direction of bias is that SPRINT patients may have had slightly higher blood pressures when at home compared with those who participated in ACCORD-BP, as patients in ACCORD-BP were subject to the white coat effect (21). However, the absolute difference in blood pressure in both trials was of the same magnitude.

The most important limitation of this analysis is that the serious adverse event data were not available at patient level in the ACCORD-BP public data set. As a result of this, we were not able to assess the interaction between serious adverse events related to intensive blood pressure–lowering therapy and T2DM. However, it is clear that intensive blood pressure lowering results in more serious adverse events both in patients with T2DM and in patients without T2DM. Therefore, it is important to weigh the potential benefit and harm of intensive blood pressure lowering for the individual patient (22).

In this pooled cohort, intensive blood pressure lowering to an SBP target <120 mmHg reduces the cardiovascular events in patients with an increased cardiovascular risk and an SBP between 130 mmHg and 180 mmHg. Cardiovascular mortality and all-cause mortality were not significantly different. These analyses do not provide any evidence for a differential effect of intensive blood pressure lowering in patients with T2DM or a higher 10-year risk of CVD.

Acknowledgments. The investigators acknowledge and thank the ACCORD-BP and SPRINT investigators and the National Heart, Lung, and Blood Institute for conducting the trials and making both data sets publicly available.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. T.F.B., J.T.V., D.N.K., and W.R.B. participated in acquisition, analysis, or interpretation of data. All authors participated in critical revision of the manuscript for important intellectual content. T.F.B. and J.T.V. performed statistical analysis. B.-J.H.v.d.B., R.J.P., and R.E.K. participated in study supervision. T.F.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.