Is There Accord in ACCORD? Lower Blood Pressure Targets in Type 2 Diabetes Does Not Lead to Fewer Cardiovascular Events Except for Reductions in Stroke

Approximately 8% of the US population—or 24 million people—have diabetes, with type 2 diabetes mellitus (T2DM) making up 90% to 95% of the cases. Cardiovascular (CV) disease remains the number one cause of death in T2DM, and CV risk is increased 2‐ to 3‐fold at every level of systolic blood pressure (BP). In those with versus without T2DM, optimal management of diabetes consists of controlling the major associated CV risk factors of hyperglycemia, hyperlipidemia, and hypertension, but which specific glycosylated hemoglobin, lipid, or BP goal is associated with optimal event reduction continues to be unclear. Recent guideline recommendations of the Seventh Report of the Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7) and the American Diabetes Association (ADA) recommend a target BP of <130/80 mm Hg in patients with T2DM, ¹ , ² even though there is no randomized clinical trial data to support this particular recommendation. Many would argue that the best available evidence suggests the goal BP should be <140/80 mm Hg. The Action to Control Cardiovascular Risk in Diabetes (ACCORD) BP trial, ³ part of the overall ACCORD trial program, was designed to determine whether a target systolic BP of <120 mm Hg would lead to a reduction in major CV events when compared with a target of <140 mm Hg among patients with T2DM.

ACCORD‐BP

Design and Eligibility

The ACCORD trial was a prospective, randomized, double 2×2 factorial clinical end point study that randomized 10,251 “high‐risk” participants with T2DM at 77 clinical sites in the United States and Canada to either intensive or standard glycemic control. To be included in ACCORD, patients had to have known stable T2DM for at least 3 months; a glycohemoglobin level of at least 7.5% but not more than 11% and be at least 40 years of age with established CV disease or at least 55 years of age with evidence of atherosclerosis, albuminuria, or left ventricular hypertrophy, or at least 2 additional risk factors for CV disease (dyslipidemia, hypertension, smoking, or obesity). A serum creatinine of >1.5 mg/dL or a body mass index of >45 kg/m² were exclusion criteria. In addition to being randomized to one or the other of the glycemic targets, participants were randomized to be included in either a lipid or BP trial. It was planned that approximately 4200 participants would be assigned to and randomized in the BP trial and 5800 participants would be assigned and randomized to the lipid trial.

To be eligible for the BP study, patients had to have a systolic BP between 130 mm Hg and 180 mm Hg (130–160 mm Hg and be taking no more than 3 antihypertensive medications, 161–170 mm Hg and be taking no more than 2 antihypertensive medications, or 171–180 mm Hg and be taking 0 or 1 antihypertensive medication) and have either a dipstick protein <2+, a spot urine protein/creatinine ratio <700 mg/g creatinine, or a 24‐hour urine for protein excretion of <1.0 g. ACCORD‐BP was a nonblinded study where patients were randomized to a systolic BP goal of <140 mm Hg (standard arm) or <120 mm Hg (intensive arm) and treated in an open‐label fashion using approved and commonly used antihypertensive medications, all of which had previously been demonstrated to reduce CV events in persons with diabetes. BP medications included diuretics, β‐blockers (BBs), calcium channel blockers (CCBs), angiotensin‐converting enzyme (ACE) inhibitors, or angiotensin receptor blockers (ARBs). Patients assigned to the intensive arm were recommended to begin therapy with a 2‐drug regimen that included a thiazide‐type diuretic plus an ACE inhibitor or BB. Drugs were added and/or titrated at each visit to achieve a systolic BP <120 mm Hg. At periodic milepost visits the addition of another drug was required if not at goal. Patients assigned to the standard intervention arm were only told to intensify therapy if a systolic BP ≥160 mm Hg occurred at one visit or ≥140 mm Hg occurred at two consecutive visits. In addition, investigators were to either down‐titrate treatment or remove a medication to achieve a systolic BP <130 mm Hg at one visit or <135 mm Hg at two consecutive visits.

After randomization, intensive BP participants were evaluated and had BP control assessed monthly for the first 4 months and every 2 months thereafter, whereas standard participants were seen and had BP assessed at months 1 and 4 and every 4 months thereafter. Additional visits occurred as needed in both groups to monitor and ensure that BP goals were being met. At each management visit, systolic BP, diastolic BP, and pulse rate were based on the average of 3 measurements using an automated device (Omron 907, Omron Healthcare, Bannockburn, IL) after patients rested and were seated in a chair for 5 minutes. Information on study outcomes, blood samples, and adverse events was collected every 4 months until the conclusion of the study.

The prespecified primary end point was a composite of nonfatal myocardial infarction (MI), nonfatal stroke, or CV death. The eight prespecified secondary outcomes were the combination of the primary outcome, plus revascularization or hospitalization for heart failure (termed the expanded macrovascular outcome); the combination of a fatal coronary event, nonfatal MI, or unstable angina (termed major coronary disease events); nonfatal MI; fatal or nonfatal stroke; nonfatal stroke; death from any cause; death from CV causes; and hospitalization or death due to heart failure. With a planned sample size of 4200 participants, the ACCORD‐BP trial was designed to have 94% power to detect a 20% difference in the rate of the primary outcome at 5 years, with an assumed primary event rate of 4% per year in the standard therapy group.

Baseline Characteristics and Results

Of the 10,251 overall ACCORD patients, 4733 were assigned to the ACCORD‐BP study (2362 to intensive BP control and 2371 to standard BP control). The mean age of the participants included in the BP study was 62.2 years, with 48% being female, 61% being non‐Hispanic white, 24% black, and 7% Hispanic. The mean duration of having diabetes was 10 years, and approximately one third had a history of CV disease at baseline. Mean baseline hemoglobin A_1c (HbA_1c) was 8.3%, mean baseline LDL‐C was 110 mg/dL, and mean BP on entry was 139/76 mm Hg. One year after randomization, and essentially throughout the remainder of the trial, the average systolic BP was 119.3 mm Hg for the intensive group and 133.5 mm Hg for the standard group, with the corresponding achieved diastolic BP 64.4 mm Hg in the intensive group and 70.5 mm Hg in the standard group. The mean difference in BP between groups was 14.2/6.1 mm Hg, which was associated with a mean difference of 1.1 antihypertensive medications at the end of the trial (3.4 and 2.3 antihypertensive medications used in the intensive and standard arms, respectively). At the last visit in the intensive group, 90% were taking either an ACE inhibitor or ARB, 80% were taking a diuretic, 42% a CCB, 61% a BB, 23% an α‐blocker, and 57% a statin. In the standard group, 80% were taking either an ACE inhibitor or ARB, 56% were taking a diuretic, 24% a CCB, 43% a BB, 11% an α‐blocker, and 58% a statin. There were no differences in the use of any of the glucose‐lowering medications or aspirin between groups. Of note, at the last visit, where measurements were made, achieved HbA_1c was reduced to 7.6% and LDL‐C was reduced to 98 mg/dL in both BP groups.

After a mean follow‐up of 4.7 years (with 94.8% of the potential follow‐up), there was no difference in the incidence of the primary composite outcome between the two groups (hazard ratio [HR] with intensive therapy, 0.88; 95% confidence interval [CI], 0.73–1.06; P=.20) (Table I). Of note, the primary composite outcome event rate in the standard therapy group of 2.09% per year was approximately 50% of the predicted rate. While other prespecified secondary outcomes did not reach statistical significance, a significant effect was shown for total stroke in the intensively treated patients (HR, 0.59; 95% CI, 0.39–0.89; P=.01 for fatal stroke and HR, 0.63; 95% CI, 0.41–0.96; P=.03 for nonfatal stroke). As stroke was a secondary outcome and because the number of major coronary events was far greater than the number of total strokes (253 vs 36 in the intensive therapy group and 270 vs 62 in the standard therapy group), this was felt to be of nominal clinical significance. Of note, there were no significant interactions among prespecified subgroups.

Table I.

 Primary and Secondary Outcomes of ACCORD‐BP

Outcome	Intensive Therapy (n=2362), No. of Events	Intensive Therapy (n=2362), % Per Year	Standard Therapy (n=2371), No. of Events	Standard Therapy (n=2371), % Per Year	Hazard Ratio	P Value
Primary outcome	208	1.87	237	2.09	0.88	.20
Prespecified secondary outcomes
Nonfatal MI	126	1.13	146	1.28	0.87	.25
Stroke
Any	36	0.32	62	0.53	0.59	.01
Nonfatal	34	0.30	55	0.47	0.63	.03
Death
Any cause	150	1.28	144	1.19	1.07	.55
CV cause	60	0.52	58	0.49	1.06	.74

Abbreviations: ACCORD‐BP, The Action to Control Cardiovascular Risk in Diabetes–Blood Pressure Arm; CV, cardiovascular; MI, myocardial infarction.

Serious adverse events, including hypotension, bradycardia or arrhythmia, hypokalemia (potassium <3.2 mmol/L), hyperkalemia but not serious hyper‐kalemia (potassium >5.9 mmol/L), elevations in serum creatinine, and having an estimated glomerular filtration rate of <30 mL/min, were all significantly more frequent in the intensively treated group. The frequency of orthostatic hypotension was similar between the groups.

The authors concluded that the ACCORD‐BP study found that in “high‐risk” T2DM patients treated for almost 5 years, targeting a systolic BP of <120 mm Hg as compared with <140 mm Hg, did not reduce the overall rate of CV events and was associated with more adverse events.

Interpretation and Recommendations

Current guideline recommendations calling for a systolic BP goal of <130 mm Hg in persons with diabetes are largely unsupported by clinical trial data. Previous studies, including the United Kingdom Prospective Diabetes Study (UKPDS) ⁴ , ⁵ and post hoc subgroup analysis of the Hypertensive Optimal Treatment (HOT) study ⁶ , ⁷ suggested a benefit of more intensive BP control in patients with diabetes, but, in both cases, the achieved systolic BP was >140 mm Hg (144 mm Hg for both). Additional post hoc analysis of UKPDS did find that for every additional 10‐mm Hg reduction in systolic BP, there was a corresponding reduction in both macrovascular and microvascular events, and there was no absolute threshold for benefit. In UKPDS, the lowest event rate was actually found in patients with diabetes whose systolic BP was <120 mm Hg. Since it appeared that an achieved systolic BP difference of about 10 mm Hg between groups led to a significant reduction in CV and renal events in UKPDS, it was of great interest to the hypertension community when the ACCORD‐BP trial reported that a therapeutic strategy targeting a systolic BP <120 mm Hg does not reduce CV disease events compared with a strategy targeting a systolic BP <140 mm Hg in patients with type 2 diabetes at high risk for CV disease events, although it did reduce fatal and nonfatal strokes, a secondary endpoint. The question now is how one reconciles these findings? There are a number of intriguing possibilities.

First, with the current standard of care, ACCORD may have been underpowered and of too short a duration to discern a benefit. The event rate in the BP arm of ACCORD was only about half of what was expected (approximately 2% per year). In addition, the duration of the trial (mean follow‐up 4.7 years) may not have been of sufficient duration to assess for long‐term complications. As such, the CIs were wide and do not exclude a 27% benefit for the primary end point at 5 years.

Second, the concept of “incremental benefit” comes into play. The lower‐than‐expected event rate occurred despite the fact that more than 30% of the ACCORD‐BP population had a history of previous CV events and all had either some evidence of atherosclerosis, target organ damage, or multiple CV risk factors. While according to the investigators this makes the ACCORD population at high risk, the aggressive treatment of other CV risk factors may have lowered absolute risk to a point from where it was difficult to demonstrate further incremental benefit from more aggressive treatment of a single risk factor, ie, BP. At the end of the study, only 9.5% of the intensive group and 7.5% of the standard group were smoking. Moreover, 58% of both groups were on statin therapy with a final LDL‐C of 98 mg/dL; the HbA_1c at last visit was 7.6%; and most were on aspirin therapy.

The synergistic effects of multiple risk factor intervention in patients with diabetes have previously been demonstrated. The STENO‐2 trial ⁸ was an open‐label, parallel group, 7‐year Danish trial of 160 patients (mean age 55) with T2DM who were randomized to a more or less intensive approach to multiple CV risk factors. In STENO‐2 there was generally not a substantial difference in the achieved values of most targeted risk factors between groups, and only the systolic BP goal of <130 mm Hg (achieved in only 15% of the intensive group) and the serum cholesterol goal of <175 mg/dL (achieved in 70% of the intensive group) reached individual statistical significance. Yet, despite this fact, the intensified multiple risk factor modification approach led to a significant 53% reduction in CV events. One important difference between STENO‐2 and ACCORD, however, is the aggressiveness of the targets. In general, the intensive group in STENO‐2 was treated less aggressively than the standard therapy group in ACCORD. Given the aggressive global risk reduction seen in the entire ACCORD population, there may not have been enough residual CV risk to demonstrate incremental benefit from just more aggressive BP control. This concept is supported by the overall lower‐than‐expected event rate seen in the standard therapy group.

Third, the effects of intensive BP control in the setting of good lipid and glucose control may differ for cerebrovascular and coronary events, ie, better for the brain than the heart. Despite a 14‐mm Hg difference in systolic BP, the intensively treated group in ACCORD did not have a significant reduction in the primary outcome, a composite of nonfatal MI, nonfatal stroke, or death from CV causes. Yet, there was a significant reduction for the secondary end point of fatal and nonfatal stroke in the intensively treated group, which was felt to be of “nominal” clinical significance based on the small number of total strokes compared with coronary events seen during the trial. While this difference in total strokes did not contribute much to the primary composite end point, it may actually be extremely clinically relevant (one would need to treat 89 patients more intensively for 5 years to prevent 1 fatal or nonfatal stroke) and biologically plausible.

The devastating nature of stroke needs to be taken into account in any risk‐benefit analysis. Stroke, more than any other vascular event, is directly related to the level of systolic BP, while the pathogenesis of coronary disease may be related more to plasma lipids and other metabolic factors. This effect has been suggested by multiple previous clinical trials; in fact, recent data from the Treat to New Targets (TNT) study that have been presented at the American Society of Hypertension (ASH) annual Scientific Sessions, but not yet published, demonstrated that aggressive BP control was more closely associated with decreased stroke risk, while aggressive lipid control was more closely associated with decreased coronary heart disease risk. Ignoring this factor will jeopardize the results of future trials if we do not take it into account in their design. The Kaplan‐Meier (KM) curves for stroke begin to separate at 3 years, exactly the point at which the trend for overall CV events begins to become apparent. Because there is no suggestion in these data that more aggressive BP control leads to more coronary events, we can hypothesize that with longer follow‐up, the benefit on stroke incidence would have had a more substantial effect on the overall risk of CV events. Although they would require larger numbers of participants in order to be appropriately powered, future studies examining the role of intensive BP targets probably need to evaluate stroke as a primary endpoint.

If ACCORD‐BP demonstrated that there was no incremental benefit to lowering systolic BP to < 120 mm Hg for CV disease but did show a benefit for stroke, is there a disadvantage to more intensive BP control? In general, it took one additional antihypertensive agent to achieve the lower BP goal. This would of course increase the cost of treatment. More concerning, however, is patient safety, as the rate of serious adverse events was significantly higher in the intensive‐therapy group, including the risk of hypotension, bradycardia or arrhythmia, creatinine elevation, reduction in estimated glomerular filtration rate, hyperkalemia (but not >5.9 mmol/L), and hypokalemia (<3.2 mmol/L). Despite the increased relative risk, overall serious adverse events occurred only in 3.3% of all intensive patients treated, an absolute rate that suggests statistical but less clinical significance.

Finally, we need to determine the overall implications of this study for the development of future guidelines. Current guidelines from JNC 7 and the ADA recommend a BP goal of <130/80 mm Hg for patients with T2DM. The diastolic BP goal of <80 mm Hg can be supported from the diabetic subgroup of HOT, and based on the findings of UKPDS and HOT, the systolic BP goal should be at least <140 mm Hg. There remains no randomized clinical trial data to clearly support a lower BP goal in patients with T2DM. Based on the lack of overall clinical benefit (which was likely due to the results cited above) and the increased adverse event rate (as infrequent as it was) seen with more intensive therapy in ACCORD, a systolic BP goal of <120 mm Hg cannot be justified in all patients with T2DM. Since we are not likely to see a clinical trial any time soon that is of sufficient size and duration to compare a systolic BP goal of <130 mm Hg, as current guidelines recommend, with that of <140 mm Hg or that uses < 120 mm Hg as a goal with stroke as a primary end point in those with T2DM, the best course of action may be to individualize treatment in diabetic patients based on the current study (Table II). We believe that future guidelines should suggest a systolic BP goal of <140 mm Hg (which hopefully would lead to an achieved BP of about 135 mm Hg) in the T2DM patient who is a nonsmoker, is taking appropriate statin‐based lipid‐lowering therapy, has been considered for antiplatelet therapy when appropriate (especially with underlying CV disease), and has reasonable glycemic control (HbA_1c approaching 7%), all of which were characteristics of the overall ACCORD BP population (Table II). While we await future subgroup analysis clarifying which particular groups benefit, in patients who are at truly higher risk, especially for stroke either due to previous personal or family history of stroke, inability to tolerate statin‐based lipid‐lowering therapy, poor glycemic control, or multiple poorly controlled cardiometabolic risk factors, we recommend a more aggressive systolic BP treatment goal be considered. Whether the goal in these truly high‐risk patients should be <120 mm Hg or another target remains unclear. What is clear is that if practitioners are going to target a systolic BP <120 mm Hg in patients with T2DM to reduce stroke risk while continuing to have a favorable effect on CV events, care must be taken to monitor for bradycardia, hypotension, potassium disturbances, and azotemia. In treating BP, as in all of medicine, first, do no harm, infrequent as it was in ACCORD‐BP.

Table II.

 Recommended Systolic Blood Pressure Goals Based on ACCORD BP

	Systolic BP Goal (mm Hg)	Strength of Evidence
Most patients with diabetes	< 140	Strong, based on primary endpoint
Diabetes Patients with higher stroke riska	< 120b	Less strength, based on secondary endpoint

Abbreviations: ACCORD‐BP, The Action to Control Cardiovascular Risk in Diabetes–Blood Pressure Arm; BP, blood pressure; TIA, transient ischemic attack. ^aDefined as personal or family history of TIA/stroke, smoker, unable to tolerate statin therapy or antiplatelet therapy when appropriate, or poor glycemic control; ^bWith close monitoring for bradycardia, hypotension, changes in electrolytes, and worsening renal function.

Disclosures:  Dr. Basile is an ACCORD investigator and his opinions expressed are his own; he is not speaking on behalf of the ACCORD trial.