We used data from patients with stable coronary artery disease (CAD) to assess the risk of new-onset diabetes mellitus (NOD) with β blockers and to determine whether angiotensin-converting enzyme (ACE) inhibition would modify this risk. The Prevention of Events with Angiotensin Converting Enzyme Inhibition (PEACE) trial randomized 8,290 patients with stable CAD to trandolapril or placebo. Presence of NOD was assessed at each study visit over a median follow-up time of 4.8 years. We examined the risk of NOD associated with β-blocker use with Cox regression models adjusting for 25 baseline covariates and tested whether this risk was modified by randomization to the ACE inhibitor. Of 6,910 patients without diabetes mellitus at enrollment (1,179 women and 5,731 men, mean age 64 ± 8 years), 4,147 (60%) were taking β blockers and 733 (8.8%) developed NOD. We observed a significant interaction between β-blocker use and randomization to ACE inhibitor with respect to NOD (p = 0.028). Participants taking β blockers assigned to the placebo group (n = 2,090) were at increased risk for NOD adjusting for baseline covariates (hazard ratio 1.63, 95% confidence interval 1.29 to 2.05, p <0.001), and this risk was attenuated in those assigned to trandolapril (n = 2,057, hazard ratio 1.11, 95% confidence interval 0.87 to 1.42, p = 0.39). β blocker use was associated with increased risk for NOD in patients with stable CAD, and this risk was decreased in patients treated concurrently with an ACE inhibitor. In conclusion, these data suggest that ACE inhibition may attenuate the risk for glucose abnormalities observed in patients taking β blockers.
Several classes of common cardiovascular medications have been shown in clinical trials to have disparate effects on blood glucose and risk for development of new-onset diabetes mellitus (NOD). β blockers have been associated with an increased risk for development of NOD. β blockers may negatively affect glucose homeostasis through increased hepatic glucose production and blockade of insulin release and may worsen insulin resistance through decreased peripheral glucose use. The effect of angiotensin-converting enzyme (ACE) inhibitors on diabetes risk has been more varied. Post hoc analyses in large trials originally suggested that ACE inhibition might delay or prevent the onset of diabetes mellitus, whereas the diabetes reduction assessment with ramipril and rosiglitazone medication (DREAM) trial did not show a benefit on frank development of diabetes but did show some improvement in glycemic control with ACE inhibitors. Mechanistically, ACE inhibitors may improve insulin sensitivity secondary to kinin accumulation and increased peripheral blood flow. The Prevention of Events with Angiotensin Converting Enzyme Inhibition (PEACE) trial was designed to test the hypothesis that an ACE inhibitor would decrease cardiovascular events in patients with stable coronary artery disease (CAD). Trandolapril therapy did not decrease the primary end point of death from cardiovascular causes, myocardial infarction, or coronary revascularization but was associated with a 17% decrease in NOD. We used data from PEACE to assess the influence of β blockers on risk for NOD and whether this risk was modified by randomization to ACE inhibition.
Methods
The PEACE trial included patients ≥50 years old with stable CAD, defined as history of myocardial infarction, coronary revascularization, or stenosis >50% on angiogram, and with normal or mildly decreased left ventricular function (left ventricular ejection fraction >40%). Patients were excluded from PEACE if at the time of screening CAD was not stable (i.e., hospitalized for unstable angina in preceding 2 months, had coronary revascularization within previous 3 months), had a planned elective coronary revascularization, a serum creatinine value >2.0 mg/dl (>177 μmol/L), or a serum potassium level >5.5 mEq/L. Patients were randomly assigned to receive the ACE inhibitor trandolapril (titrated to a target dose of 4 mg/day) or to placebo and followed for a median of 4.8 years, as previously described . The PEACE protocol was approved by each participating site’s institutional review board. All patients provided written informed consent in accordance with established guidelines for protection of human subjects.
Of the 8,290 patients randomized, we included in this analysis 6,910 patients who did not have diabetes at baseline assessed by patient report. The primary outcome variable for this analysis was NOD; diabetic status (i.e., presence of a new diagnosis of diabetes mellitus) was assessed by study personnel from patient history at each study visit every 6 months and marked on the case-report forms. No other information about the diagnosis of diabetes mellitus was available including laboratory measurements. Medications were also recorded at baseline and at each visit. Specific β blockers were not recorded.
Baseline demographics of participants taking versus not taking β blockers were compared to identify potential differences. Between-group assessments were performed using t tests for normally distributed continuous variables or Wilcoxon rank-sum tests for non-normally distributed continuous variables and chi-square or Fisher’s exact tests, as appropriate, for categorical variables. Risk for NOD associated with β-blocker use at baseline was examined with Cox proportional hazards models adjusting for baseline covariates and randomized treatment interactions. β blocker use was also explored as a time-dependent covariate. Model covariates chosen a priori included age, gender, body mass index, tobacco use, systolic and diastolic blood pressures, glomerular filtration rate, left ventricular ejection fraction, baseline cholesterol and potassium concentrations, history of CAD on angiogram, myocardial infarction, angina, percutaneous transluminal coronary arterioplasty, coronary artery bypass graft, stroke, transient ischemic attack, intermittent claudication, Canadian Cardiovascular Society functional class, and use of lipid-lowering agents, digoxin, aspirin, or antiplatelets. To test the robustness of multivariable models, we performed a propensity-adjusted analysis in which we generated a propensity score for baseline β-blocker use in a logistic regression of baseline covariates and then adjusted for this propensity score in Cox regressions. Because determination of diabetes mellitus status was assessed every 6 months, we also used discrete-time proportional hazards models taking into account the discrete nature of the NOD information captured in the study. The investigators had full access to the data and take full responsibility for the integrity of the data. All investigators read and agreed to the report as written.
Results
Of 6,910 patients without diabetes included in these analyses, 4,147 (60%) were taking β blockers ( Figure 1 ) at baseline. Of 4,147 taking β blockers, 2,090 were assigned to the treatment group and 2,057 were assigned to the placebo group. Baseline characteristics of all analyzed subjects, broken down by β-blocker use, are listed in Table 1 . Participants taking β blockers were more likely to be younger, have a higher body mass index, a history of CAD, documented myocardial infarction, and were more likely to have undergone coronary interventions compared to patients not taking β blockers.
| Variable | All Subjects (n = 6,910) | β Blocker | p Value | |
|---|---|---|---|---|
| Yes (n = 4,147) | No (n = 2,763) | |||
| Age (years), mean ± SD | 64 ± 8 | 64 ± 8 | 65 ± 8 | <0.001 |
| Women | 17% | 18% | 16% | 0.05 |
| Body mass index (kg/m 2 ) | 28.1 | 28.4 | 27.6 | <0.001 |
| Current tobacco use | 14% | 14% | 15% | 0.49 |
| Ejection fraction (%), mean ± SD | 58 ± 9 | 58 ± 9 | 58 ± 9 | 0.31 |
| Hypertension | 44% | 49% | 36% | <0.001 |
| Documented myocardial infarction | 56% | 58% | 52% | <0.001 |
| Coronary disease on angiogram | 60% | 65% | 54% | <0.001 |
| Angina pectoris | 69% | 71% | 65% | <0.001 |
| Coronary artery bypass grafting | 37% | 35% | 41% | <0.001 |
| Stroke | 4% | 4% | 4% | 0.44 |
| Medicines | ||||
| Calcium channel blocker therapy | 35% | 28% | 44% | <0.001 |
| β Blocker | 60% | 100% | 0% | NA |
| Diuretic | 12% | 12% | 12% | 0.97 |
| Aspirin/antiplatelet | 91% | 92% | 90% | 0.008 |
| Lipid-lowering agent | 71% | 74% | 66% | <0.001 |
| Anticoagulant | 5% | 4% | 5% | 0.05 |
| Digoxin | 3% | 3% | 4% | 0.97 |
| Antiarrhythmic | 2% | 1% | 3% | <0.001 |
There were 733 cases of NOD reported over the trial follow-up time of 4.8 years (event rate 2.0%/year). Randomization to trandolapril was associated with a 17% decreased risk for development of NOD (hazard ratio [HR] 0.83, 95% confidence interval [CI] 0.71 to 0.95, p = 0.009), as previously reported ( Figure 2 ). In univariate analyses, β-blocker use was associated with a 44% increased overall risk for development of NOD (HR 1.44, 95% CI 1.23 to 1.68, p <0.001) and remained associated with an increased risk for NOD after adjustment for baseline covariates and randomized treatment (HR 1.36, 95% CI 1.15 to 1.61). We observed a significant interaction between treatment assignment to trandolapril and use of β blockers on NOD in univariate (p for interaction = 0.021) and multivariable-adjusted (p for interaction = 0.028) models. Participants taking β blockers assigned to the placebo group (n = 2,090) had an adjusted increased risk for NOD (HR 1.63, 95% CI 1.29 to 2.05, p <0.001; Figure 3 ), whereas this risk was attenuated in those assigned to trandolapril (n = 2,057, HR 1.11, 95% CI 0.87 to 1.42, p = 0.39). Adjusted analyses in which β-blocker use throughout the trial was included as a time-dependent covariate yielded qualitatively similar results (placebo group HR 1.40, 95% CI 1.12 to 1.75; trandolapril group HR 1.02, 95% CI 0.81 to 1.29). Propensity-adjusted analyses yielded similar results (trandolapril group HR 1.59, 95% CI 1.26 to 2.00; placebo group HR 1.06, 95% CI 0.83 to 1.34). Additional predictors of NOD in adjusted models are listed in Table 2 . Results from discrete-time proportional hazards models (not shown) were similar to those from the standard Cox regression models reported earlier.
| Variable | Univariate HR (95% CI) | Multivariable HR (95% CI) | Chi-square |
|---|---|---|---|
| Body mass index (per kg/m 2 ) | 1.1 (1.09–1.11) | 1.1 (1.08–1.09) | 13.3 |
| β Blocker | 1.44 (1.24–1.68) | 1.56 (1.24–1.95) | 3.83 |
| Seated systolic blood pressure (per mm Hg) | 1.01 (1.0–1.01) | 1.0 (1.0–1.01) | 2.96 |
| Use of lipid-lowering agents | 0.8 (0.67–0.94) | 0.81 (0.69–0.97) | 2.26 |
| Use of potassium-sparing diuretics | 1.8 (1.29–2.54) | 1.51 (1.06–2.18) | 2.25 |
| Use of other diuretics | 1.52 (1.22–1.9) | 1.24 (0.98–1.57) | 1.8 |
| Seated diastolic blood pressure (per mm Hg) | 1.0 (0.96–1.01) | 0.99 (0.98–1.0) | 1.74 |
| History of coronary disease on angiogram | 1.32 (1.13–1.54) | 1.16 (0.98–1.37) | 1.73 |
| Use of aspirin or antiplatelet | 0.94 (0.73–1.2) | 0.81 (0.6–1.07) | 1.46 |
| Canadian Cardiovascular Society functional class | 1.18 (1.07–1.3) | 1.08 (0.97–1.2) | 1.43 |
| History of percutaneous transluminal coronary arterioplasty | 0.98 (0.84–1.13) | 0.89 (0.76–1.05) | 1.36 |
| Use of angiotensin-converting enzyme inhibitors | 0.83 (0.72–0.96) | 1.03 (0.79–1.34) | 1.34 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
