A randomized, placebo-controlled study of the cardiovascular safety of the once-weekly DPP-4 inhibitor omarigliptin in patients with type 2 diabetes mellitus

Eligible patients were at least 40 years old with a history of T2DM and established CV disease. Patients were required to be on one of the following diabetes treatment regimens that was stable for at least 12 weeks (except for pioglitazone which was required to be stable for at least 16 weeks) and be within the glycated hemoglobin (HbA1c) range associated with the treatment regimen (Fig. 1). Patients on diet and exercise alone (not on an AHA for ≥12 weeks) or on monotherapy or dual combination therapy with metformin (MF), pioglitazone (PIO), an alpha-glucosidase inhibitor (AGI) or an SGLT2 inhibitor (SGLT2i) were required to have an HbA1c of ≥6.5 and ≤10.0% (≥48 and ≤86 mmol/mol). Patients on a sulfonylurea or meglitinide either as monotherapy or as part of dual combination therapy with MF, PIO, AGI or an SGLT2i were required to have an HbA1c ≥7.0 and ≤10.0% (≥53 mmol/mol and ≤86 mmol). Patients could also be enrolled if they were on a stable insulin regimen (±metformin) with an HbA1c ≥7.0 and ≤10.0% (≥53 mmol/mol and ≤86 mmol). Insulin regimens that were allowed included basal insulin (e.g., insulin glargine, insulin detemir, NPH insulin, degludec), prandial insulin (e.g., regular, aspart, lispro, glulisine), basal/prandial insulin regimen consisting of multiple dose insulin injections of basal and prandial insulin or the use of pre-mixed insulin (e.g., Novolog 70/30^®, Novolin 70/30^®, Humalog 75/25^®, or Humulin 70/30^®). A stable insulin regimen was defined as no change in the insulin regimen [i.e. type(s) of insulin] and ≤10% change in the total daily dose of insulin. Prior to randomization, patients were required to have a site fasting-fingerstick glucose (FFSG) >126 mg/dL (7.0 mmol/L) and <260 mg/dL (14.4 mmol/L).

The criteria for having established CV disease included the presence of one of the following: (1) coronary artery disease defined as having a history of MI, surgical or percutaneous (balloon and/or stent) coronary revascularization procedure, or coronary angiography showing at least 1 stenosis ≥50% in a major epicardial artery or branch vessel; (2) ischemic cerebrovascular disease defined as having a history of ischemic stroke (strokes not known to be hemorrhagic will be allowed as part of this criterion); (3) carotid arterial disease defined as a ≥50% stenosis documented by carotid ultrasound, magnetic resonance imaging, or angiography, with or without symptoms of neurologic deficit; or (4) atherosclerotic peripheral arterial disease, as documented by objective evidence such as amputation due to vascular disease, current symptoms of intermittent claudication confirmed by an ankle-brachial pressure index of <0.9 or a toe-brachial pressure index <0.7 or history of surgical or percutaneous revascularization procedure.

Patients were excluded if they had acute coronary syndrome (e.g., MI or unstable angina), a coronary artery intervention (e.g., CABG or PTCA), a stroke or transient ischemic neurological disorder or worsening signs or symptoms of coronary heart disease or congestive heart failure within the past 3 months; type 1 diabetes, a history of ketoacidosis, or C-peptide <0.7 ng/mL; estimated glomerular filtration rate <40 mL/min/1.73 m²; active liver disease or an aspartate aminotransferase (AST) or alanine aminotransferase (ALT) level >2× the upper limit of normal (ULN); thyroid stimulating hormone outside the central laboratory normal range; triglycerides >600 mg/dL (6.78 mmol/L); hemoglobin <12 g/dL (120 g/L) (males) and <11 g/dL (110 g/L) (females); history of malignancy or hematological disorders; or previously treated with omarigliptin.

This multicenter, double-blind, randomized, placebo-controlled, parallel group, Phase 3 study (Fig. 1) was conducted in 559 centers: 10 in Argentina, 7 in Australia, 5 in Austria, 10 in Belgium, 16 in Brazil, 6 in Bulgaria, 7 in Canada, 4 in Chile, 6 in Colombia, 6 in Croatia, 11 in the Czech Republic, 7 in Denmark, 2 in Finland, 5 in France, 15 in Georgia, 24 in Germany, 8 in Hong Kong, 8 in Hungary, 11 in Israel, 14 in Italy, 4 in Lebanon, 5 in Lithuania, 9 in Malaysia, 4 in Mexico, 5 in the Netherlands, 7 in New Zealand, 12 in the Philippines, 14 in Poland, 11 in the Republic of Korea, 16 in Romania, 12 in the Russian Federation, 5 in Serbia, 11 in Slovakia, 20 in South Africa, 15 in Spain, 9 in Sweden, 6 in the province of Taiwan, 16 in the Ukraine, 16 in the United Kingdom and 180 in the United States. Patients were randomly assigned in a 1:1 ratio to omarigliptin 25 mg q.w. or placebo using an interactive voice-response system.

After Week 18, patients were to have their background AHA regimen adjusted as deemed necessary by the study investigator to achieve an appropriate individualized glycemic goal in line with local, national or international guidelines. Whenever possible, the patients were followed until study closeout, regardless of whether they continued to receive blinded study medication. The study was designed and funded by the Sponsor and conducted in conjunction with PAREXEL International Corp., Waltham, MA.

The study (Omarigliptin Protocol 018; ClinicalTrials.gov: NCT01703208) was conducted in accordance with the principles of Good Clinical Practice and was approved by the appropriate institutional review boards and regulatory agencies. All patients enrolled in the study provided written informed consent.

Safety assessment included collection of adverse events, physical examination, vital signs, standard laboratory blood chemistry (e.g., liver and renal safety tests), lipid panel, hematology, urinalysis and electrocardiogram. In addition, amylase and lipase were measured per regulatory agency request. A standard questionnaire was provided to patients to collect hypoglycemia information.

An external Data Monitoring Committee reviewed unblinded interim data from this study. External Clinical Adjudication Committees (CACs) evaluated potential CV events, cases of pancreatitis and prespecified hypersensitivity reactions in a blinded manner. The CACs were not charged with assessing causality to study medication. The adjudication of hypersensitivity was at the request of regulatory authorities and was not due to any signal observed with omarigliptin. Hypersensitivity events prespecified for adjudication were anaphylactic reaction, angioedema, asthma–bronchospasm, erythema multiforme, Stevens–Johnson syndrome, toxic epidermal necrolysis, and drug rash with eosinophilia and systemic symptoms.

CV endpoints analyzed included (1) time to first event of MACE (confirmed CV-related death, nonfatal MI, nonfatal stroke); (2) time to confirmed CV-related death; (3) time to first event of confirmed MI (fatal and nonfatal); (4) time to first event of stroke (fatal and nonfatal); (5) time to all-cause mortality; (6) time to first event of confirmed hHF; and (7) time to the composite of first confirmed event hHF or CV death. Change from baseline over time in HbA1c and non-CV safety, including hypoglycemia, was analyzed. Criteria for the adjudication of CV endpoints followed guidelines of the Clinical Data Interchange Standards Consortium for standardized definition of CV [9] and stroke endpoints in clinical trials and the Third Universal Definition of Myocardial Infarction [10], as excerpted in Additional file 1.

The number of patients randomized was based on the primary hypothesis of the study, which was non-inferiority of omarigliptin compared with placebo for the primary composite MACE endpoint. Non-inferiority was to be declared if the upper bound of the two-sided 95% confidence interval (CI) of the HR for MACE was <1.30 in the intention-to-treat (ITT) population and was to include all confirmed events that occurred up to 365 days after last dose of blinded study medication. A total of 632 patients having a primary endpoint event would provide 90% power for the test of non-inferiority, assuming a true HR of 1.00. It was estimated that the study would take approximately 8 years from initiation to completion, assuming accrual of primary endpoints at 3.0% per year and rates of enrollment and patient dropout.

Due to the early termination of the study and the attendant decrease in power to adequately test the primary hypothesis of non-inferiority, no formal statistical testing of the non-inferiority hypothesis was conducted. The population analyzed was the ITT population that included all randomized patients who took at least 1 dose of blinded study medication and all CV endpoint data, regardless of the time since a patient may have discontinued blinded study medication.

The analyses presented herein encompass the time period from first-patient-first-visit (October 16, 2012) to the final data cut-off date (May 13, 2016). The data cut-off date was approximately 30 days after investigator notification of study discontinuation, which followed the Sponsor’s announcement of its intent not to seek marketing authorization for omarigliptin in the US and Europe. At the time of the data cut-off date, 90.4% of randomized patients had been discontinued from the study. CV events that occurred after the data cut-off date in the remaining 9.6% of patients (n = 403; 209 in the omarigliptin group and 194 in the placebo group) who were ongoing in the study at the time of the data cut-off date were not adjudicated and are not included in the analyses presented herein. Nearly all of the ongoing patients were discontinued from blinded study medication shortly afterwards; however, in Brazil, based on local regulatory procedures, 75 patients remained on blinded study medication, with the last patient visit occurring on March 23, 2017.

The ITT population was used for the assessment of CV endpoints. The Cox proportional-hazards model was used to calculate the HR and two-sided 95% CI. The Kaplan–Meier method was used to estimate the time-to-first event in the two treatment groups. The Full Analysis Set, consisting of all randomized patients who received at least 1 dose of study medication and had a baseline or a post-randomization measurement, was used for analyzing the HbA1c endpoint. No subgroup analyses were performed. Safety and tolerability were assessed in the All-Patients-as-Treated (APaT) population, which included all randomized patients who received at least 1 dose of study medication, by clinical review of all relevant parameters, including adverse events, laboratory tests and vital signs.

The primary analysis of safety was assessed in the time frame consisting of the Treatment Period + 21 days. Adverse events of symptomatic hypoglycemia (episodes with clinical symptoms attributed to hypoglycemia, without regard to glucose level) were prespecified as events of interest and p values and 95% CI for between-treatment group comparisons were calculated. Confirmed events of pancreatitis and prespecified hypersensitivity adverse events were summarized regardless of time after discontinuation from study medication. Patients who discontinued blinded study medication and who did not withdraw consent were followed by telephone contact until study end to ascertain any serious adverse events that occurred after the discontinuation of blinded study medication. A second approach to safety analyses, that applied only to serious adverse events, included all safety-related data after the first dose of blinded study medication and was referred to as Treatment Period + all follow-up days.

Patient disposition is summarized in Table 1. Eight patients randomized to the omarigliptin group and two randomized to the placebo group did not take any study medication and were not included in any safety or efficacy analyses. Demographic and anthropometric characteristics of randomized patients were similar between the two treatment groups (Table 2). Approximately 70% of patients were male and 30% female; the mean age was 63.6 years. Racial percentages were 81.3% White, 10.8% Asian, 3.8% Black, and 12.7% were of Hispanic or Latino ethnicity. The mean BMI was 31.3 kg/m². Twenty-one percent of patients were from North America, 55.6% were from Europe, 6.8% were from Latin America, 9.1% were from Asia and 7.6% were from other countries.

Baseline disease state characteristics are presented in Table 2. The mean baseline HbA1c was 8.0% and mean duration of T2DM was 12.1 years. The majority of patients (88.3%) had an estimated glomerular filtration rate (eGFR) ≥60 mL/min/1.73 m² and 11.3% had an eGFR ≥30 and <60 mL/min/1.73 m². The majority had a history of hypertension (95.4%) and 15.3% had a history of heart failure (16.2% in the omarigliptin group and 14.3% in the placebo group). The most commonly used prior oral AHA medication was metformin (77.4%), followed by sulfonylurea (39.1%) and insulin (34.9%).

The median duration of exposure to study medication was similar in the two treatment groups: 90.0 weeks (range 1.0–172.0) in the omarigliptin group and 89.0 weeks (range 1.0–168.0) in the placebo group. The median duration of treatment plus post-treatment follow-up was similar in the two treatment groups: 96.1 weeks (range 1.1–178.6) in the omarigliptin group and 95.6 weeks (range 1.3–176.0) in the placebo group. Mean compliance with study medication (omarigliptin or matching placebo) was similar for the two treatment groups (97.8 and 97.9% for omarigliptin and placebo, respectively).

Table 3 summarizes the results of the CV endpoints analyzed. There was no significant between-group difference in the composite MACE endpoint, which occurred in 114/2092 patients in the omarigliptin group (5.45%) and 114/2100 patients in the placebo group (5.43%).

Figure 2 presents the Kaplan–Meier plot of time to first MACE event. There was no notable treatment difference in event rates over time through Week 156.

The rate of confirmed MACE per 100 patient-years was 2.96 for omarigliptin and 2.97 for placebo (HR, 1.00; 95% CI, 0.77–1.29). The rate of confirmed CV-related death per 100 patient-years was 0.94 for omarigliptin and 0.89 for placebo, with an HR (95% CI) of 1.06 (0.66, 1.68). The rate of confirmed fatal or non-fatal MI per 100 patient-years was 1.34 for omarigliptin and 1.55 for placebo, with an HR (95% CI) of 0.87 (0.60, 1.26). The rate of confirmed fatal or non-fatal stroke per 100 patient-years was 0.82 for omarigliptin and 0.88 for placebo, with an HR (95% CI) of 0.94 (0.58, 1.52). The rate of all-cause mortality per 100 patient-years was 1.63 for omarigliptin and 1.28 for placebo, with an HR (95% CI) of 1.28 (0.88, 1.85). There was no specific pattern of non-CV deaths. Figure 3 presents a forest plot of MACE and other CV endpoints (CV-related death, fatal and nonfatal MI, fatal and nonfatal stroke and all-cause mortality).

There was no significant difference in the rate of hHF or the rate of the composite of hHF and CV death. The rate of first hHF per 100 patient-years was 0.51 for omarigliptin and 0.85 for placebo, with an HR (95% CI) of 0.60 (0.35, 1.05). The rate of the composite endpoint of first hHF or CV death per 100 patient-years was 1.41 for omarigliptin and 1.65 for placebo, with an HR (95% CI) of 0.86 (0.60, 1.23).

The changes from baseline in HbA1c over time (up to Week 142) are shown in Fig. 4. At Week 18, prior to the adjustment of background medication to meet individualized patient goals (see above), from a mean baseline HbA1c of 8.0%, the LS mean change from baseline (95% CI) in HbA1c was −0.58% (−0.62, −0.55) in the omarigliptin group and −0.16% (−0.19, −0.12) in the placebo group, with a between treatment difference of −0.43% (−0.48, −0.37). The profile of the change in HbA1c over time indicates that the between-group difference persisted throughout the treatment period. At Week 142, the change from baseline (95% CI) was −0.36% (−0.47, −0.25) in the omarigliptin group and −0.06% (−0.17, 0.05) in the placebo group, with a between treatment difference of −0.30% (−0.46, −0.14). Over time, an increase in HbA1c was observed in both treatment groups.

In the following, the term “higher” indicates that the 95% CI for the between-group difference in the incidences of adverse events excluded 0. Adverse event summary measures were similar in the omarigliptin and placebo group, with the exception of drug-related adverse events, which was higher in the omarigliptin group compared with the placebo group (Table 4). The between-group difference in the overall incidence of drug-related adverse events was primarily due to differences in the adverse event of hypoglycemia.

The number of patients with adverse events by MedDRA system organ class (SOC) was similar between the omarigliptin and placebo groups, with the exception of skin and subcutaneous tissue disorders SOC, where the incidence was higher in the omarigliptin group compared with the placebo group (7.9% [166/2092 patients] versus 6.0% [126/2100 patients]) due to a variety of specific adverse events. The majority of cutaneous adverse events in the omarigliptin group were non-serious; however 2 (1 eczema and 1 dermatosis) were serious, but were not considered by the investigator to be related to study medication and did not lead to the discontinuation of study medication.

Among the specific adverse events, the incidence of the adverse event of AST increased was higher in the omarigliptin group compared with the placebo group (Table 5). The between-group difference in adverse event of ALT increased was less notable and there were no notable between-group differences in the incidences of exceeding the predefined limits of change (PDLC) for ALT or AST of ≥3× ULN or ≥5× ULN. Including supplemental laboratory data (measurements entered in the database from sources other than the central laboratory; primarily drawn during hospitalization), there were 16 patients in the omarigliptin group and 17 patients in the placebo group with ALT ≥5× ULN; all of the patients in the omarigliptin group had a plausible alternative explanation (e.g., acute cholecystitis, alcohol abuse, metastatic cancer) and/or resolved on study medication. There were no cases that met the combined criteria for a Hy’s law case (i.e., ALT or AST ≥3× ULN with elevations in bilirubin ≥2× ULN and alkaline phosphatase <2× ULN without alternative explanation) that would suggest the potential for drug-induced liver injury [11]. There were no between-group differences in mean change from baseline over time in ALT or AST.

Table 6 summarizes the adverse events of hypoglycemia. The incidence of symptomatic hypoglycemia was higher in the omarigliptin group (22.6% [472/2092 patients]) compared with the placebo group (19.7% [414/2100 patients]); p = 0.024. The incidence of severe hypoglycemia requiring medical assistance was 4.2% (88/2092 patients) in the omarigliptin group compared to 3.3% in (70/2100 patients) in the placebo group.

Table 7 summarizes the events of adjudication-confirmed acute and chronic pancreatitis and adjudication-confirmed prespecified hypersensitivity reactions. Confirmed acute pancreatitis events were uncommon overall and none of the cases in either treatment group were fatal. One patient in the placebo group had confirmed chronic pancreatitis (no cases occurred in the omarigliptin group).

Four patients in the omarigliptin group had a confirmed case of angioedema; all were nonserious and one case occurred in the post-treatment period (i.e., after the 21-day follow-up after discontinuation of study medication). One patient in the placebo group had two confirmed cases of angioedema; both of these cases were serious adverse events. Two patients in the omarigliptin group with a history of asthma had events of confirmed asthma–bronchospasm; one was reported as non-serious and the other as a serious adverse event. Two patients in the placebo group had confirmed cases of asthma–bronchospasm; both were reported as non-serious. One patient in the placebo group had a confirmed case of anaphylactic reaction.

Investigator reported pancreatic cancer occurred in five patients in the omarigliptin group (3 cases of pancreatic carcinoma and 2 cases of pancreatic cancer metastatic) and one patient in the placebo group (1 case of pancreatic carcinoma).

There were no clinically meaningful changes from baseline or in analysis of PDLC in laboratory safety measures. Small changes from baseline in mean serum amylase and lipase values were observed in the omarigliptin group by Week 6 and continued through Week 156 (Table 8). All mean serum amylase values over time were within the normal laboratory range (35–121 U/L) throughout the treatment period in both treatment groups. The mean serum lipase value was within the normal laboratory range (13–60 U/L) in both treatment groups throughout the treatment period except for the omarigliptin group at Week 104, when the mean value (72.3 IU/L ± 437.2) was slightly greater than the upper limit of the normal range.

There were no clinically meaningful between-group changes from baseline in eGFR at any timepoint. At Week 142, the LS mean change from baseline (SE) in eGFR calculated by the MDRD equation (mL/min/1.73 m²) was −0.97 (1.05) in the omarigliptin group and 1.45 (1.06) in the placebo group; LS mean difference (95% CI) was −2.43 (−5.36, 0.51).

There were no clinically meaningful changes from baseline over time in heart rate, blood pressure, or ECG intervals (including QTc) in either treatment group. At Week 142, the LS mean changes (SE) from baseline in body weight were −0.33 kg (0.21) in the omarigliptin group and −0.25 kg (0.22) in the placebo group.