Impact of Ezetimibe on New‐Onset Diabetes: A Substudy of IMPROVE‐IT

Statins have been associated with an increased risk for new‐onset diabetes (NOD). ¹ However, it is unclear if the addition of the nonstatin therapy ezetimibe to a statin would affect the incidence of NOD. We therefore evaluated the risk of NOD with ezetimibe added to simvastatin versus placebo added to simvastatin in IMPROVE‐IT (Improved Reduction of Outcomes: Vytorin Efficacy International Trial). ²

These analyses were carried out in patients without diabetes at baseline (no prior diagnosis of diabetes, no antihyperglycemic medication, and no elevated blood glucose [≥126 mg/dL fasting or ≥200 nonfasting] at randomization). Because glycosylated hemoglobin was not measured systematically during the trial, NOD was defined for the primary analysis as either the initiation of an antihyperglycemic medication or 2 consecutive blood glucose measurements of ≥126 mg/dL and was analyzed as time‐to‐event outcome. Sensitivity analyses using other potential indicators to define NOD were also performed (Table). To account for interval censored data due to medications and blood glucose recorded only at the study visits, parametric Weibull regression univariable and multivariable models were used to determine predictors of NOD.

Table 1.

Sensitivity Analyses for the Development of New‐Onset Diabetes in IMPROVE‐IT

Definition	Simvastatin	Simvastatin/ezetimibe	HR (95% CI)	P value
Primary definition of new‐onset diabetes	694/4740	720/4760	1.03 (0.93–1.15)	0.549
Sensitivity analyses
S1: Initiation of diabetic medication post randomization	547/6084	578/6042	1.06 (0.95–1.21)	0.300
S2: Two consecutive post. baseline glucose ≥126 mg/dL	335/4650	348/4675	1.03 (0.9–1.22)	0.655
S3: Diabetes‐related AE or SAE defined from clinical review of Medical Dictionary for Regulatory Activities codes	501/6141	510/6113	1.03 (0.91–1.17)	0.692
S4: Diabetes‐related AE/SAE or diabetic medication post randomization	671/6084	690/6042	1.02 (0.93–1.13)	0.718
S5: Diabetes‐related AE/SAE or diabetic medication post randomization or 2 consecutive postbaseline fasting glucose ≥126 mg/dL	786/4746	804/4763	1.03 (0.93–1.16)	0.530

Sensitivity analyses with respective hazard ratios (HRs), 95% CIs, and P values for the simvastatin vs simvastatin/ezetimibe arms of IMPROVE‐IT. AE indicates adverse event; IMPROVE‐IT, Improved Reduction of Outcomes: Vytorin Efficacy International Trial; and SAE, serious adverse event.

Baseline characteristics included age, sex, self‐identified race, current smoking, hypertension, waist circumference, metabolic syndrome, body mass index, systolic and diastolic blood pressure, total cholesterol, low‐ and high‐density lipoprotein cholesterol, triglycerides, glucose at randomization, use of prior statin, beta blocker, and angiotensin‐converting enzyme inhibitor/angiotensin receptor blocker. Metabolic syndrome at randomization was defined as ≥3 of the criteria defined by the American Heart Association. ³ Reported P values are 2‐sided, with <0.05 considered statistically significant. Analyses were completed using SAS version 9.4.

The data that support the findings of this study are available from the corresponding author upon reasonable request. Institutional review board approval or informed consent was not needed because this was a prespecified analysis of a clinical trial already conducted and published.

Patients were evaluated at 30 days after randomization and then every 4 months, with a minimum 2.5‐year and median 6‐year follow‐up. Of the 18 144 patients enrolled in IMPROVE‐IT, 8644 were excluded due to a history of diabetes (N=5284), or missing data (N=3360), leaving 9500 patients for the study population. After randomization, 1414 patients (14.9%) met criteria for NOD for an annualized incidence of approximately 2.7% per year. For those who developed NOD there were no differences in baseline characteristics between those who received simvastatin/ezetimibe (n=720) and those who received simvastatin/placebo (n=694). Compared with patients who did not develop NOD (n=8086), those with NOD were more likely at baseline to have a higher body mass index (hazard ratio [HR], 1.27 [95% CI, 1.20–1.35]), higher systolic blood pressure (HR, 1.07 [95% CI, 1.00–1.14]), higher triglycerides (HR, 1.33 [95% CI, 1.17–1.53]), and ≥3 components of the metabolic syndrome (HR, 1.36 [95% CI, 1.17–1.61]). There were no major differences in baseline medical therapies or low‐density lipoprotein cholesterol between any of the groups. Compared with the placebo/simvastatin arm, there was no difference in risk of NOD with the addition of ezetimibe to simvastatin (HR, 1.03 [95% CI, 0.93–1.15]). This relationship remained qualitatively consistent across sensitivity analyses using various NOD definitions including the following: initiation of a diabetes medication (HR, 1.06 [95% CI, 0.95–1.21]), 2 glucose measurements ≥126 mg/dL (HR, 1.03 [95% CI, 0.90–1.22]), diabetes‐related adverse event (HR, 1.03 [95% CI, 0.91–1.17]), diabetes‐related adverse event or initiation of diabetes medication (HR, 1.02 [95% CI, 0.93–1.13]), and diabetes‐related adverse event or initiation of diabetes medication or 2 glucose readings ≥126 mg/dL (HR, 1.01 [95% CI, 0.93–1.16]); Table). The addition of ezetimibe did not affect risk of NOD in either statin‐naïve (not on previous statin before trial qualifying event) or statin‐experienced (on previous statin before trial qualifying event) patients (HR, 1.01 [95% CI, 0.89–1.15] versus HR, 1.08 [95% CI, 0.87–1.36], respectively; P _interaction = 0.50). Furthermore, in an additional sensitivity analysis to assess the risk of NOD based on presence of metabolic syndrome at baseline, the addition of ezetimibe did not affect risk in those with metabolic syndrome or in those without metabolic syndrome (HR, 1.06 [95% CI, 0.94–1.21] versus HR, 0.98 [95% CI, 0.81–1.18], respectively, P _interaction = 0.46).

In this analysis of patients enrolled in IMPROVE‐IT, we found no increase in the risk of NOD with addition of ezetimibe to statin therapy. Our data are consistent and supported by the findings from a recent trial that found a lower risk of NOD with lower dose rosuvastatin and ezetimibe versus higher dose rosuvastatin. ⁴ Similar lack of NOD has been seen with PCSK9 inhibitors. ⁵ Limitations of the present study include not having glycosylated hemoglobin measurements or glucose tolerance tests and use of only simvastatin as the background statin therapy. However, sensitivity analyses were performed for validation, and the results are further strengthened by being obtained in the context of a large randomized clinical trial with 6 years follow‐up. Therefore, given outcomes from IMPROVE‐IT, and absence of any signal that ezetimibe increases risk of NOD, our results support the use and safety of this medication as an adjunct to statins for further low‐density lipoprotein cholesterol lowering, which, combined with recent trials and the PCSK9 inhibitor data, could be important for national guidelines.

Sources of Funding

This study was funded by Merck and Co. The sponsor had no role in the development of the analytic plan, data analysis, or data interpretation. No additional external sources of funding was obtained.

Disclosures

N.P.S. reports the following research grants: Amgen, Janssen, National Institutes of Health; consultant/advisor: Esperion, Amgen, Norvartis. D.K.M.: honoraria for trial leadership from Boehringer Ingelheim, Sanofi, Merck & Co, Pfizer, AstraZeneca, Novo Nordisk, Esperion, Lilly USA, Lexicon, CSL Behring, and honoraria for consultancy from Lilly USA, Boehringer Ingelheim, Merck & Co, Novo Nordisk, Applied Therapeutics, Sanofi, CSL Behring, Bayer, Altimmune, Intercept, and GSK. C.P.C. reports research grants from Amgen, Better Therapeutics, Boehringer‐Ingelheim (BI), Bristol‐Myers Squibb (BMS), Daiichi Sankyo, Merck, Novo Nordisk, Pfizer; consulting fees from Aegerion/Amryt, Alnylam, Amarin, Amgen, Applied Therapeutics, Ascendia, BI, BMS, Eli Lilly, Janssen, Lexicon, Merck, Pfizer, Rhoshan, Sanofi; serves on data and safety monitoring board for Applied Therapeutics and NovoNordisk. R.P.G. reports research grants: Amgen, Anthos Therapeutics: Ionis; consulting fees: Amarin, Amgen, Artivion, Bayer, Boston Scientific, Daiichi Sankyo, Gilead, Inari, Inventiva, Labcorp, Medpace, Paratek, Pfizer, PhaseBio Pharmaceuticals, St. Lukes Hospital System, Samsung, Sanofi Aventis. Y.L.: institutional grant to DCRI from Merck during the conduct of the study and institutional grants to DCRI from Amylin Pharmaceuticals Inc (a wholly owned subsidiary of AstraZeneca). A.M.T. is employed by Merck. E.B.: research support: AstraZeneca, Daiichi Sankyo, Merck, and Novartis; consultancies with Amgen, Cardurion, MyoKardia, Novo Nordisk, and Verve. The remaining authors have no disclosures to report.