Dalcetrapib Reduces Risk of New-Onset Diabetes in Patients With Coronary Heart Disease.

OBJECTIVE: Incident type 2 diabetes is common among patients with recent acute coronary syndrome and is associated with an adverse prognosis. Some data suggest that cholesteryl ester transfer protein (CETP) inhibitors reduce incident type 2 diabetes. We compared the effect of treatment with the CETP inhibitor dalcetrapib or placebo on incident diabetes in patients with recent acute coronary syndrome. RESEARCH DESIGN AND METHODS: In the dal-OUTCOMES trial, 15,871 patients were randomly assigned to treatment with dalcetrapib 600 mg daily or placebo, beginning 4-12 weeks after an acute coronary syndrome. Absence of diabetes at baseline was based on medical history, no use of antihyperglycemic medication, and hemoglobin A1c and serum glucose levels below diagnostic thresholds. Among these patients, incident diabetes after randomization was defined by any diabetes-related adverse event, new use of antihyperglycemic medication, hemoglobin A1c ≥6.5%, or a combination of at least two measurements of serum glucose ≥7.0 mmol/L (fasting) or ≥11.1 mmol/L (random).
RESULTS: At baseline, 10,645 patients (67% of the trial cohort) did not have diabetes. During a median follow-up of 30 months, incident diabetes was identified in 403 of 5,326 patients (7.6%) assigned to dalcetrapib and in 516 of 5,319 (9.7%) assigned to placebo, corresponding to absolute risk reduction of 2.1%, hazard ratio of 0.77 (95% CI 0.68-0.88; P < 0.001), and a need to treat 40 patients for 3 years to prevent 1 incident case of diabetes. Considering only those with prediabetes at baseline, the number needed to treat for 3 years to prevent 1 incident case of diabetes was 25. Dalcetrapib also decreased the number of patients who progressed from normoglycemia to prediabetes and increased the number who regressed from diabetes to no diabetes.
CONCLUSIONS: In patients with a recent acute coronary syndrome, incident diabetes is common and is reduced substantially by treatment with dalcetrapib.

Introduction

Approximately 30% of patients with acute coronary syndrome (ACS) have a prior history of type 2 diabetes (1–3), a further 10% may be diagnosed with diabetes during hospitalization for ACS (4), and ∼10% may receive the diagnosis over the ensuing 5 years (5). The development of type 2 diabetes carries a heightened risk of microvascular and macrovascular complications (6) and is associated with a particularly poor prognosis after ACS (3). Accordingly, there has been intense interest in pharmacologic and nonpharmacologic strategies to reduce incident type 2 diabetes. Available approaches that have demonstrated efficacy in patients with prediabetes or obesity include thiazolidinediones, metformin, acarbose, valsartan, basal insulin, orlistat, lorcaserin, intensive lifestyle modification, and bariatric surgery (7–10). Compared with placebo or usual care, these approaches have been associated with a 15–85% reduction in the risk of incident diabetes. Despite this efficacy, treatments to forestall or prevent the onset of diabetes have not been adopted widely due to concerns of safety, tolerability, cost, and adherence.

Cholesteryl ester transfer protein (CETP) inhibitors increase the concentration of HDL cholesterol (HDL-C) and were developed as cardiovascular drugs. To date, large outcomes trials have demonstrated no cardiovascular benefit (11–13) or modest cardiovascular benefit (14) of treatment with CETP inhibitors compared with placebo. Reductions in plasma glucose and insulin were noted in patients treated with torcetrapib (15), and an 11% lower relative risk for new-onset diabetes was observed in patients treated with anacetrapib (14) or evacetrapib (16), with the effect of the former agent statistically significant. The mechanism of a salutary effect of CETP inhibition on incident diabetes is unknown, but HDL-C is purported to prevent β-cell endoplasmic reticulum stress and apoptosis and to promote insulin secretion (17). Such cellular effects are supported by human genetic data indicating decreased risk of diabetes among subjects with a genetically instrumented elevation in HDL-C (18).

Dalcetrapib is a CETP inhibitor with modest effects on HDL-C and minimal effects on LDL-cholesterol (LDL-C) concentration. The dal-OUTCOMES trial compared treatment with dalcetrapib or placebo in patients with recent ACS and found no overall cardiovascular benefit (13). However, a significant interaction of dalcetrapib treatment and allele type at the ADCY9 locus that encodes adenylyl cyclase 9 on cardiovascular outcomes (19) led to a large ongoing trial in 6,000 patients with ACS (ClinicalTrials.gov reg. no. NCT02525939) to determine cardiovascular efficacy of dalcetrapib in patients with the favorable allele type (20).

In this analysis, we compare the effects of treatment with dalcetrapib or placebo on incident diabetes among all dal-OUTCOMES participants without diabetes at baseline.

Research Design and Methods

Study Population

The design and principal results of the dal-OUTCOMES trial have been described previously (13,21). The study was performed between 2008 and 2012 at 935 sites in 27 countries. The Institutional Review Board of each site approved the study, and all subjects provided informed consent. Qualifying patients were at least 45 years of age, had recent ACS (acute myocardial infarction or unstable angina pectoris), and had completed all planned coronary revascularization procedures. Exclusion criteria included New York Heart Association Functional Classification III or IV symptoms of heart failure or class II symptoms with left ventricular ejection fraction ≤40%, uncontrolled hypertension (systolic blood pressure ≥180 mmHg and/or diastolic blood pressure ≥110 mmHg despite treatment), serum creatinine >2.2 mg/dL (194.5 μmol/L), or fasting triglycerides >400 mg/dL. Baseline laboratory testing included serum glucose and hemoglobin A1c. Patients were instructed to report for study visits after an overnight fast; actual fasting or nonfasting state was verified at each visit and recorded on a case report form. Randomization of 15,841 patients occurred 4–12 (median 6) weeks after the index ACS event when the patients were considered to be clinically stable. Serum glucose was measured 1, 3, 6, 9, and 12 months after randomization and then every 4 months and at the end of the trial. Hemoglobin A1c was measured 6 and 12 months after randomization and then annually and at the end of the trial. Concurrent medications and medical conditions were determined, weight was measured, and BMI was calculated at baseline and at follow-up visits. The primary end point of the trial was a composite of death from coronary heart disease, nonfatal myocardial infarction, ischemic stroke, unstable angina, or cardiac arrest with resuscitation.

Definition of Baseline and Incident Diabetes

The analyses in this report were performed post hoc. In the base-case model, diabetes at baseline was defined by at least one of the following criteria: a medical history of diabetes or a diabetes-related adverse event, current use of antihyperglycemic medication, hemoglobin A1c ≥6.5%, or serum glucose ≥7.0 mmol/L (fasting) or ≥11.1 mmol/L (random) at the screening or randomization visit.

Incident diabetes was defined by one or more of the following criteria fulfilled after randomization in patients without diabetes at baseline: an investigator-reported diabetes-related adverse event, new use of antihyperglycemic medication, at least one measurement of hemoglobin A1c ≥6.5%, or any combination of at least two measurements of serum glucose ≥7.0 mmol/L (fasting) or ≥11.1 mmol/L (random).

Using data from both treatment groups, we evaluated the association of baseline HDL-C and risk of incident diabetes. In an analysis restricted to the dalcetrapib group, we evaluated the association of the change in HDL-C from baseline to month 3 and the risk of incident diabetes after month 3.

Many patients with ACS have hemoglobin A1c or fasting glucose values that lie near dichotomous boundaries used to define diabetes, and intraindividual variability in these measurements may be considerable. We therefore performed a sensitivity analysis restricting the subgroup considered to be without diabetes at baseline to those with baseline hemoglobin A1c <6.3% and baseline fasting serum glucose <6.5 mmol/L (if fasting) or <11.1 mmol/L (if random), and with no medical history of diabetes and no current use of antihyperglycemic medication. The criteria for incident diabetes after randomization were the same as in the base-case model.

Prediabetes and Normoglycemia

Patients were classified as having prediabetes at baseline if hemoglobin A1c was at least 5.7% but <6.5%, or if fasting serum glucose was at least 5.6 mmol/L but <7 mmol/L, without fulfillment of any criterion for diabetes. Patients who did not meet criteria for prediabetes or diabetes were considered normoglycemic. Progression from normoglycemia to prediabetes was defined by normoglycemia at baseline and at least two postrandomization measurements of hemoglobin A1c of at least 5.7% but <6.5% or two measurements of fasting serum glucose of at least 5.6 mmol/L but <7 mmol/L, without fulfillment of any criteria for new-onset diabetes. Regression from diabetes to a nondiabetic state was defined by diabetes at baseline, the last two available measurements of hemoglobin A1c of <6.5%, and the last two available measurements of serum glucose of <7 mmol/L (if fasting) or <11.1 mmol/L (if not fasting), with no use of antihyperglycemic medications at the time of the last two measurements.

Insulin Resistance

In an exploratory analysis in patients without diabetes at baseline, fasting serum insulin was measured as part of a nested case-control biomarker survey (22). Measurements were available at baseline for 1,293 patients assigned to treatment with dalcetrapib and for 1,288 patients assigned to treatment with placebo and at month 3 for 1,071 patients assigned to dalcetrapib treatment and 1,197 patients assigned to placebo treatment, excluding patients who initiated treatment with antihyperglycemic medication prior to month 3. At each time point, HOMA of insulin resistance (HOMA-IR) was calculated as (23):The change in HOMA-IR from baseline to 3 months was evaluated according to treatment with dalcetrapib or placebo.

Statistical Analysis

Development of diabetes was measured as the time to new-onset diabetes. Differences between treatment groups were evaluated using proportional hazards regression and reported as hazard ratio (HR) with 95% CI. Progression from normoglycemia to prediabetes or regression from diabetes to prediabetes or from diabetes to a nondiabetic state was described by the proportion of subjects in each of these categories at the last two available observations; differences between treatment groups were evaluated using logistic regression and reported as the odds ratio (OR) with 95% CI. Proportional hazards regression analysis was also used to determine whether an effect of dalcetrapib on incident diabetes was related to baseline or on-treatment concentrations of HDL-C or to baseline or postrandomization BMI. Changes in fasting glucose, insulin, and HOMA-IR between baseline and 3 months of treatment with dalcetrapib or placebo or between baseline and end-of-trial measurements were assessed by unequal variance t test after log transformation of insulin and HOMA-IR. In all analyses, two-sided P values <0.05 were considered statistically significant.

Results

Baseline Characteristics

At baseline using base-case criteria, 5,141 patients (32.5%) were classified with diabetes (dalcetrapib, 2,573; placebo, 2,568) and 10,645 (67.1%) without diabetes (dalcetrapib, 5,326; placebo, 5,319). Diagnostic criteria were missing for 85 patients (0.5%). Among those without diabetes, 6,695 (62.9%) were classified as prediabetic and 3,950 (37.1%) as normoglycemic. Baseline characteristics of patients in each category are reported in Table 1. The sensitivity analysis that used more restrictive criteria to define absence of diabetes at baseline placed 9,646 patients (60.8%) in that category, with characteristics reported in Supplementary Table 1.

Table 1

Baseline characteristics of patients according to glycemic category and treatment group

Characteristic	Normoglycemia (N = 3,950)		Prediabetes (N = 6,695)		Diabetes (N = 5,141)
	Dalcetrapib n = 1,932	Placebo n = 2,018	Dalcetrapib n = 3,394	Placebo n = 3,301	Dalcetrapib n = 2,573	Placebo n = 2,568
Age, years*†‡	58.6 (8.7)	58.6 (8.8)	59.9 (9.1)	60.0 (9.0)	61.8 (9.1)	61.4 (9.2)
Male sex (%)*†‡	82.9	84.2	80.7	81.7	77.4	78.0
White race (%)*†‡	92.5	92.0	89.1	89.2	83.9	84.5
History of hypertension (%)*†‡	58.6	60.8	62.3	63.0	80.3	81.2
Current smoking (%)‡	18.7	18.6	22.8	23.4	20.4	19.3
Prior myocardial infarction (%)*†	12.0	11.6	14.5	14.0	21.2	19.4
Prior stroke (%)	2.6	2.7	2.6	2.6	4.8	5.1
Blood pressure, mmHg
Systolic*†‡	125.5 (16.6)	125.6 (16.7)	126.8 (16.9)	127.1 (16.6)	130.0 (17.1)	129.7 (17.1)
Diastolic	76.8 (9.6)	76.6 (9.8)	76.9 (9.8)	77.0 (9.6)	77.1 (9.9)	76.8 (9.6)
BMI, kg/m2*†‡	27.5 (4.5)	27.4 (4.1)	28.1 (4.6)	28.2 (4.5)	30.1 (5.3)	30.2 (5.7)
Laboratory tests
Fasting serum glucose, mmol/L*†‡	5.17 (0.53)	5.18 (0.53)	5.43 (0.60)	5.45 (0.59)	7.20 (2.40)	7.18 (2.32)
Hemoglobin A1c, %*†‡	5.34 (0.23)	5.35 (0.22)	5.85 (0.24)	5.84 (0.24)	6.82 (1.07)	6.80 (1.07)
Total cholesterol, mg/dL†‡	143.9 (30.6)	142.4 (31.5)	148.0 (33.6)	147.1 (31.9)	143.4 (34.0)	142.5 (33.5)
LDL-C, mg/dL*†‡	75.3 (24.1)	74.7 (24.8)	78.8 (27.1)	78.2 (24.9)	74.1 (26.8)	73.4 (27.6)
HDL-C, mg/dL*†‡	44.2 (12.6)	43.4 (12.2)	43.0 (11.4)	42.7 (11.2)	40.6 (11.2)	40.5 (11.2)
Triglycerides, mg/dL*†‡	123 (69)	122 (62)	132 (71)	132 (74)	145 (78)	143 (80)
eGFR, mL/min/1.7 m2*†	81.9 (16.6)	82.6 (17.4)	81.9 (17.0)	81.9 (17.2)	80.6 (21.5)	80.9 (20.4)
Medications (%)
Aspirin or other antiplatelet agent†	99.2	99.3	99.2	99.6	98.8	98.9
ACE inhibitor or ARB*†‡	73.8	72.7	78.6	79.1	84.7	84.2
β-Blocker*†	86.6	87.1	86.8	87.1	88.4	88.7
Statin*†	97.5	98.1	97.6	97.8	96.7	96.8

Data are mean (SD) unless otherwise indicated. All characteristics were balanced between the dalcetrapib and placebo groups (P > 0.05). BMI data missing for 27 patients (0.7%) with normoglycemia, 33 patients (0.5%) with prediabetes, and 44 patients (0.9%) with diabetes. Some laboratory data missing for 5 patients (0.1%) with normoglycemia, 5 patients (0.1%) with prediabetes, and 40 patients (0.8%) with diabetes. An additional 85 patients (0.5%) had insufficient glucose and hemoglobin A1c data to determine baseline glycemic status and are not included in this table or in the analyses. ARB, angiotensin II receptor blocker; eGFR, estimated glomerular filtration rate.

P < 0.05 for difference between normoglycemia and diabetes;

P < 0.05 for difference between prediabetes and diabetes;

P < 0.05 for difference between normoglycemia and prediabetes, comparing the aggregate of both treatment groups in each metabolic category.

Cardiovascular Outcomes, Safety, and Tolerability Under Treatment with Dalcetrapib or Placebo

Considering both treatment groups in aggregate, the risk of the primary end point was higher in patients with diabetes at baseline (cumulative incidence 11.6%) than among those without diabetes at baseline (6.5%). Treatment with dalcetrapib versus placebo did not affect the risk of the primary end point in those with diabetes at baseline (HR 1.06, 95% CI 0.91–1.25, P = 0.79) or in those without diabetes at baseline (HR 1.02, 95% CI 0.88–1.19, P = 0.45). As previously described (13), dalcetrapib had a generally acceptable safety and adverse effect profile. Adverse events of hypertension, diarrhea, and insomnia were reported more frequently in the dalcetrapib group than in the placebo group.

Hemoglobin A1c, Fasting Glucose, and Concurrent Medication Use With Dalcetrapib or Placebo

Supplementary Table 2 reports changes in median hemoglobin A1c and glucose levels over the course of the trial among patients without diabetes at baseline censored at first use of an antihyperglycemic medication. Hemoglobin A1c levels were slightly lower with dalcetrapib at 6, 12, 24, and 36 months, without differences in fasting glucose. At 12 and 24 months, there were no significant differences between groups in the use of concurrent medications, including statins, β-blockers, or diuretics (data not shown).

Incidence of Diabetes and Its Modification by Dalcetrapib

Median (interquartile range) follow-up was 30 (25–35) months. Using base-case criteria to classify patients without diabetes at baseline and considering both treatment groups in aggregate, there were 897 cases of incident diabetes (8.4%). Of these, 821 had baseline classification as prediabetes and 76 as normoglycemia. The criteria fulfilled to establish incident diabetes are reported in Supplementary Table 3.

Dalcetrapib treatment reduced incident diabetes (Table 2 and Fig. 1). As determined from base-case criteria, incident diabetes developed in 403 of 5,326 patients (7.6%) in the dalcetrapib group and in 516 of 5,319 patients (9.7%) in the placebo group, corresponding to an absolute risk reduction of 2.1%. Dalcetrapib prolonged time to onset of diabetes (HR 0.77, 95% CI 0.67–0.89, P < 0.001), with a need to treat 40 patients for 3 years to prevent 1 incident case.

Table 2

Transitions in glycemic status

Model	Dalcetrapib, n/N (%)	Placebo, n/N (%)	HR*	95% CI	P value
New-onset diabetes
Base case (prediabetes or normoglycemia at baseline)	403/5,326 (7.6)	516/5,319 (9.7)	0.77	0.67–0.89	<.001
Prediabetes to diabetes	364/3,394 (10.7)	473/3,301 (14.3)	0.74	0.65–0.85	<.001
Normoglycemia to diabetes	39/1,932 (2.0)	43/2,018 (2.1)	0.95	0.61–1.47	.80
Sensitivity analysis (prediabetes or normoglycemia at baseline)	254/4,602 (5.5)	324/4,602 (7.0)	0.78	0.66–0.92	.003
Normoglycemia to prediabetes	711/1,846 (38.5)	826/1,915 (43.1)	0.83*	0.73–0.94	.004
Diabetes to no diabetes	325/2,354 (13.8)	271/2,393 (11.3)	1.25*	1.06–1.49	.01

Base case uses standard criteria to define absence of diabetes at baseline and incident diabetes after randomization. Sensitivity analysis uses restrictive criteria to define absence of diabetes at baseline and standard criteria to define incident diabetes after randomization. Diabetes to no diabetes includes transitions from diabetes to prediabetes or normoglycemia. See text for full descriptions of criteria for each transition.

Outcomes that were assessed at final study observation points are described with ORs.

Figure 1

Cumulative incidence of new-onset diabetes in the dalcetrapib and placebo groups among all patients without diabetes at baseline (A) and among patients with prediabetes at baseline (B).

Of the 919 patients who developed diabetes, 837 (91%) had prediabetes and 82 (9%) had normoglycemia at baseline (Table 2). Considering only those with prediabetes at baseline, incident diabetes developed in 364 of 3,394 patients (10.7%) in the dalcetrapib group and in 473 of 3,301 patients (14.3%) in the placebo group, corresponding to a HR of 0.74 (95% CI 0.65–0.85, P < 0.001) (Table 2 and Fig. 1) and a need to treat 25 patients for 3 years to prevent 1 incident case. Restricting the analysis further to 3,371 patients with impaired fasting glucose (100–125 mg/dL) at baseline, incident diabetes developed in 251 of 1,681 patients (14.9%) in the dalcetrapib group compared with 321 of 1,690 (19.0%) in the placebo group (HR 0.80, 95% CI 0.68–0.95, P = 0.009).

In the sensitivity analysis using more stringent criteria to classify patients as not having diabetes at baseline, incident diabetes occurred in 254 of 4,602 patients (5.5%) in the dalcetrapib group and in 324 of 4,602 patients (7.0%) in the placebo group (HR 0.78, 95% CI 0.66–0.92, P = 0.003).

Among all participants without diabetes at baseline, the risk of incident diabetes was significantly related to the HDL-C concentration at baseline (HR for 1 mg/dL increment 0.98, 95% CI 0.97–0.99, P < 0.001), without significant interaction of treatment group (P = 0.16). Between baseline and month 3, the mean (SD) change in HDL-C was 14.9 (11.4) mg/dL in the dalcetrapib group and 1.7 (7.2) mg/dL in the placebo group. After adjusting for the 3-month change in HDL-C, the relationship between treatment and incident diabetes was no longer significant (dalcetrapib-to-placebo HR 0.97; P = 0.70). In an analysis limited to the dalcetrapib group and adjusted for baseline HDL-C, the risk of incident diabetes was significantly related to the change in HDL-C from baseline to 3 months of assigned treatment (HR for 1 mg/dL change 0.98, 95% CI 0.97–0.99, P < 0.001).

Progression From Normoglycemia to Prediabetes, Regression From a Diabetic to a Nondiabetic State, and Modification of These Transitions by Dalcetrapib

Dalcetrapib reduced the likelihood of progression from normoglycemia to prediabetes (Fig. 2). Among 3,761 patients who were normoglycemic at baseline and were evaluable for glycemic status at their last two study visits, prediabetes developed in 711 of 1,846 (38.5%) assigned to dalcetrapib compared with 826 of 1,915 (43.1%) assigned to placebo (OR 0.83, 95% CI 0.73–0.94, P = 0.004) (Table 2).

Figure 2

A: Proportion of patients with normoglycemia at baseline who progressed to prediabetes after randomization. B: Proportion of patients with diabetes at baseline who regressed to no diabetes after randomization. Definitions of transitions are provided in the text.

Dalcetrapib also increased the likelihood of regression from a diabetic to a nondiabetic state (Fig. 2). Among 4,747 patients with diabetes at baseline who were evaluable for subsequent glycemic status, 325 of 2,354 (13.8%) assigned to dalcetrapib had improved to a state of prediabetes or normoglycemia at their two final study visits compared with 271 of 2,393 (11.3%) assigned to placebo, corresponding to an OR of 1.25 (95% CI 1.06–1.49, P = 0.01), an absolute increase of 2.5%, and a number needed to treat of 40 (Table 2).

Effect of Dalcetrapib on BMI

BMI did not differ between the treatment groups at baseline (mean 27.9 kg/m2 in both). However, patients treated with dalcetrapib had significantly lower BMI by month 6 of treatment, and an intergroup difference in median BMI of ∼0.2 kg/m2 was sustained through month 36 (Supplementary Table 2). Among all participants without diabetes at baseline, the risk of incident diabetes was significantly related to baseline BMI (HR per 1 kg/m2 increment 1.06, 95% CI 1.05–1.08, P < 0.001 adjusted for treatment group) and to the change in BMI from baseline to month 12 (HR per 1 kg/m2 change 1.14, 95% CI 1.09–1.19, P < 0.001 adjusted for treatment and baseline BMI). However, the association between dalcetrapib treatment and the risk of new-onset diabetes did not appear to be due to changes in BMI. After adjustments for baseline and month 12 change in BMI, the HR for new-onset diabetes with dalcetrapib relative to placebo (HR 0.80, 95% CI 0.70– 0.92, P = 0.001) was equivalent to the unadjusted HR.

Effect of Dalcetrapib on Insulin Resistance

Supplementary Table 4 reports data for patients without diabetes at baseline who had concurrent measurements of fasting glucose and insulin at randomization (n = 2,581) and at month 3 (N = 2,168). At baseline, median (IQR) fasting glucose was 5.3 (4.9–5.7) mmol/L, insulin was 8.3 (5.5–12.4) μU/mL, and HOMA-IR was 1.9 (1.3–3.0), without differences between treatment groups. At month 3, median fasting glucose was unchanged, and fasting insulin increased slightly in both groups. However, dalcetrapib had no effect on the change in these measurements from randomization to month 3, nor did dalcetrapib have a discernible effect on new-onset diabetes at month 3.

Conclusions

This analysis demonstrates that dalcetrapib, a CETP inhibitor, reduces incident diabetes by ∼23% (absolute reduction 2.1%) over a median follow-up of 30 months in patients with ACS who do not have diabetes at baseline. On the basis of Kaplan-Meier incidences at 3 years, 40 patients would have to be treated with dalcetrapib to prevent 1 incident case of diabetes. If restricted to those with prediabetes at baseline, the number needed to treat was 25. The effect of dalcetrapib on incident diabetes was robust to a sensitivity analysis that used a more stringent definition to identify patients without diabetes at baseline. Dalcetrapib also had favorable effects on transitions between other glycemic states. Compared with placebo, treatment with dalcetrapib was associated with fewer patients progressing from normoglycemia to prediabetes and more patients regressing from a diabetic to a nondiabetic state, with a number needed to treat of 40 for the latter.

The present analysis does not define the mechanism by which dalcetrapib ameliorates the glycemic state of patients with prior ACS. Among potential mechanisms, HDL-C, possibly through its component apolipoprotein A-I, may improve β-cell function and enhance insulin secretion (19), a hypothesis supported by analyses showing an inverse relationship between genetically determined HDL-C levels and incident diabetes (18). In fact, we observed that the risk of incident diabetes was inversely associated both with baseline HDL-C and with the increase in HDL-C concentration with dalcetrapib treatment. However, it is uncertain whether the former relation represents a direct effect of HDL-C or an association of low HDL-C with insulin resistance. Similarly, the latter relation cannot distinguish whether increased HDL-C concentration mediates a decreased risk of diabetes or whether increased HDL-C concentration is a marker of dalcetrapib exposure, with the drug’s effects on diabetes mediated through other unmeasured mechanisms.

Dalcetrapib had a small but significant effect to reduce BMI compared with placebo, but this effect did not account for the reduction in incident diabetes. We did not observe an effect of dalcetrapib on fasting serum glucose, but we did observe slightly lower levels of hemoglobin A1c between 6 and 36 months after randomization. An effect on postprandial glucose excursions may explain this dichotomy but was not evaluated in the current study. Almost all patients in this cohort were treated with statins, and statin treatment has been associated with an increased risk of incident diabetes (24). It is not possible to determine whether dalcetrapib mitigated an increased risk of diabetes related to statin treatment or exerted an independent effect to prevent diabetes.

In an exploratory analysis, we measured fasting insulin and glucose at baseline and at month 3 of assigned treatment in a subset of patients and observed no intergroup difference in fasting insulin, glucose, or HOMA-IR between these time points. However, the difference between dalcetrapib and placebo groups in the incidence of diabetes emerged after this time, and we cannot exclude the possibility that an effect of dalcetrapib on insulin secretion or sensitivity developed after month 3. Interpretation of these findings is also subject to the limitations of selecting patients for insulin measurement in a case-control design rather than with random sampling or in the entire study cohort. Further targeted studies to assess β-cell function and insulin sensitivity may help to answer these questions.

Compared with other CETP inhibitors, dalcetrapib has a smaller effect to raise the concentration of HDL-C (11,12,14); however, the present findings suggest its effect to prevent incident diabetes is at least as large and possibly larger. In an analysis of 19,129 patients with occlusive vascular disease and no diabetes at baseline who were monitored for a minimum of 4 years, diabetes developed in 510 patients (5.3%) treated with the CETP inhibitor anacetrapib compared with 571 (6.0%) treated with placebo (HR 0.89, 95% CI 0.79–1.00, P = 0.05) (14). In an analysis of 3,856 patients with acute or chronic coronary heart disease without diabetes at baseline and with a median follow-up 30 months (the same duration as in the current analysis), diabetes developed in 176 of 1,911 patients (9.2%) treated with the CETP inhibitor evacetrapib compared with 200 of 1,945 patients (10.3%) treated with placebo (OR 0.89, P = 0.24) (16). There are several reasons why dalcetrapib might have a larger effect on incident diabetes than other CETP inhibitors. First, more potent CETP inhibitors, such as anacetrapib or evacetrapib, lower LDL-C while producing large increases in HDL-C. Lower levels of LDL-C due to genetic variants in cholesterol-regulating genes or intensive statin therapy have been associated with an increased risk of incident diabetes (25). Therefore, it is possible that a salutary effect of potent CETP inhibitors on incident diabetes through increased HDL-C was mitigated by lower levels of LDL-C. Second, it is possible that differing effects of CETP inhibitors on incident diabetes are related to different effects on HDL-C composition or function that are not reflected by HDL-C concentration (26). Finally, criteria for diagnosis of incident diabetes differ among studies of CETP inhibitors. In the analysis involving anacetrapib, incident diabetes was defined by diabetes-related adverse events or use of antihyperglycemic medication, and in the analysis involving evacetrapib by a hemoglobin A1c value >6.5%. In contrast, the present analysis used all of the above criteria as well as measurements of fasting or random glucose. Therefore, absolute incidence rates for new-onset diabetes should not be compared directly among studies, but comparison of ORs for active treatment versus placebo may be reasonable.

Considering all available laboratory data, there were relatively small changes in median fasting glucose and hemoglobin A1c over time within and between treatment groups (Supplementary Table 2). In contrast, bidirectional rates of transition among glycemic states were substantial (Table 2). One reason for this apparent disparity is that the former data include laboratory values obtained after introduction of antihyperglycemic medication, which may blunt observed differences. Another reason is that patients may cross dichotomous boundaries between glycemic categories with small changes in hemoglobin A1c or glucose. However, in a sensitivity analysis that restricted the analysis cohort to patients with baseline hemoglobin A1c and glucose well below criteria for diabetes, there was nonetheless a significant effect of dalcetrapib to attenuate the risk of incident diabetes.

Data from the placebo group are notable for the high rate of glycemic transitions over the course of the trial. The number of patients who regressed from diabetes to no diabetes was more than half of the number who developed new-onset diabetes. Approximately 40% of patients characterized as normoglycemic at baseline developed criteria for prediabetes over a median 30-month observation period. The dynamic nature of the glycemic state after ACS points to the importance of lifestyle modification during this period and to a possible role of pharmacologic intervention.

Strengths of this analysis include the large study cohort representing many nationalities, a detailed database that allowed for the use of multiple complementary criteria to define incident diabetes, and collection of data to verify fasting or nonfasting state at each study visit, allowing glucose data to be used appropriately to categorize glycemic state.

Limitations include a post hoc design and inherent inaccuracies in determining incident diabetes from criteria other than gold standard glucose tolerance testing. It is possible that some antihyperglycemic medication was prescribed for weight loss rather than diabetes, but few such cases are likely because the trial antedated the first approval of glucagon-like peptide 1 receptor agonists for weight loss.

Implications

In the present analysis, the observed effect size of dalcetrapib treatment on incident diabetes was smaller than that previously seen with thiazolidinediones and comparable to that observed with metformin (27,28). To date, pioglitazone is the sole available agent that has been demonstrated effective in preventing both incident diabetes and cardiovascular morbidity and mortality in patients without diabetes at baseline (29). Although dalcetrapib did not reduce cardiovascular events in the dal-OUTCOMES trial, this possibility is being investigated further in the precision medicine Effect of Dalcetrapib vs. Placebo on CV Risk in a Genetically Defined Population With a Recent ACS (dal-GenE) trial (20). Moreover, there could be a long-term cardiovascular benefit of preventing diabetes with dalcetrapib, only becoming apparent with a longer observation period than the median 30 months in the current analysis. Dalcetrapib does not share the risks of fluid retention and heart failure, weight gain, and bone fractures associated with thiazolidinediones, the hemodynamic and electrolyte effects with valsartan, or the gastrointestinal symptoms and a small potential for lactic acidosis with metformin. Therefore, dalcetrapib might have utility as a well-tolerated agent to prevent or delay the onset of diabetes in patients at high risk for that condition.