Study design
This was a multicenter open-label randomized controlled trial, conducted at two institutes in Japan: the Kobe University Hospital and Hyogo Brain and Heart Center. This study was enforced between September 1, 2012 and March 2019. The study protocol was approved by the ethical review board of each participating institution. The study was registered in the UMIN clinical trial registry (UMIN000008620). Written informed consent to participate in this study was obtained from all patients.
Participants
Patients with stable CAD and IGT needed to satisfy the following criteria to be included: 1) in those undergoing percutaneous coronary intervention (PCI), "untreated IGT" which was "2-hour plasma/serum glucose level: 140mg/dL to 199 mg/dL in a 75 g OGTT", 2) low-density lipoprotein cholesterol < 120mg/dl and < 100mg/dl in patients with and without statin administration, respectively , 3) being between 20 and 80 years old, and 4) written consent provided for participation in the study. Patients meeting any of the following conditions were excluded: 1) under treatment for type 2 diabetes or having type 1 diabetes, 2) severe liver dysfunction, 3) severe renal dysfunction, 4) severe heart failure (New York Heart Association stage III or IV), 5) malignancies or other diseases with poor prognosis, 6) pregnant, lactating, and possibly pregnant women and those planning to get pregnant, 7) past medical history of hypersensitivity to investigational drugs, and 8) judged as ineligible by clinical investigators.
Treatment protocol
Patients who consented to participate and fulfilled the inclusion criteria were included in this study; then, PCI was performed for the treatment of culprit lesions causing myocardial ischemia and/or angina symptoms. Participants were randomly divided into those receiving 50 mg vildagliptin once a day for at least 6 months (vildagliptin group) or diet and exercise therapy (control group). Other oral hypoglycemic drugs as for combination treatment were prohibited during the study period. Moreover, the increase and addition of lipid improving agents such as statin preparations, fibrate preparations, and eicosapentaenoic acid were prohibited as well. Coronary angiography, OCT, and CGM were conducted before and 6 months after PCI.
Continuous glucose monitoring system
CGM was performed for more than 3 consecutive days before PCI and 6 months after PCI, and the daily glucose profile was analyzed using data obtained on days 2 and 3 to avoid any bias due to insertion or removal of the sensor. In all patients, CGM analysis software (CareLink iPro, Medtronic, Northridge, CA) calculated the median of the variables measured on days 2 and 3: 24-hour mean glucose levels, time in hyperglycemia/hypoglycemia, and the mean amplitude of glycemic excursion (MAGE) [11]. Time in hyperglycemia and hypoglycemia were defined as the time that blood glucose levels were above 140 mg/dL and under 70 mg/dL, respectively [7]. All patients received optimal meals (25–28 kcal/kg of ideal body weight; 60% carbohydrate, 15–20% protein, and 20–25% fat) during CGM.
OCT imaging
Angiographically non-significant coronary lesions (30% to 70% luminal narrowing via angiography) were enrolled to OCT assessment during index PCI procedure, and deemed to follow-up OCT assessment 6 months after PCI. Images were acquired using a commercially available, frequency-domain OCT imaging system (ILUMIEN; St. Jude Medical Inc., St Paul, MN, USA). In this system, a 2.7-Fr OCT imaging catheter is advanced distal to the lesion, and automated pullback is initiated in concordance with blood clearance by the injection of contrast media. Target plaques are all non-culprit lesions, which form mild to moderate stenosis.
OCT analysis
All OCT images were analyzed by two independent investigators, who were blinded to the angiographic and clinical findings, using Off-line Review Workstation. Using OCT examination, the lipid arc was measured at every 1-mm interval throughout the length of each lesion, and the values were averaged [12]. Lipid length was also measured on the longitudinal view [12]. Fibrous cap was defined as a signal-rich homogenous layer overlying the lipid-rich plaque [13]. The thinnest part of the fibrous cap was measured three times, and its average was defined as the minimum fibrous cap thickness [14]. Calcification was also recorded when an area consisted of a signal-poor or heterogeneous region with a sharply delineated border [15]. Calcification arc was measured at every 1-mm interval throughout the length of each lesion, and the values were averaged. Calcification length was also measured on the longitudinal view [7].
Outcomes
In this trial protocol, the primary endpoint was changes in coronary plaque characteristics, such as the minimum fibrous cap thickness and lipid arc detected by OCT between baseline and 6 months after intervention, and those in the MAGE.The following OCT parameters were determined for analysis: the minimum lumen area, lipid length, lipid mean arc, and minimum fibrous cap thickness for quantitative variables. As outcomes of glycemic metabolic variables, the MAGE; time in hyperglycemia/hypoglycemia; and mean, max, and min blood glucose levels were measured. As clinical outcome, cardiac death, myocardial infarction (MI), cerebral infarction, target lesion revascularization (TLR), and target vessel revascularization (TVR) were set.
Sample size
The needed sample size was not calculated based on statistical power. The sample size in this study was determined by referring to some similar studies, in which minimum sample size was 9 to 19 individuals per group. Referring to these numbers, the sample size needed in this study was determined as 25 per group.
Randomization
Sequence generation
Participants were randomly assigned to either the vildagliptin group or the control group with 1:1 allocation using simple randomization.
Allocation concealment mechanism
Randomization was performed by an allocation stuff member in the Kobe University Hospital who was not a clinician.
Implementation
After a participant who consented to participate and fulfilled the inclusion criteria was enrolled in the study, randomization (1:1 ratio, permuted-block design) was performed by the allocation stuff using an Excel-based allocation system with stratification. Participants and investigators were not masked to the allocation, but OCT images were analyzed in a blinded fashion.
Statistical methods
The data set consisted of all participants enrolled in this study. For the participant baseline data in analysis population, summary statistics were presented by group. The categorical counts and proportion were calculated for nominal variables; number of participants, mean, standard deviation, minimum value, median, and maximum value for continuous variables. For inter-group comparison of nominal and continuous variables, Fisher’s exact test and Wilcoxon’s rank sum test were used, respectively.
For continuous OCT parameters and glycemic metabolic variables, the adjusted mean difference of means for the change between baseline and 6 months after intervention between the vildagliptin and control groups was estimated, whereas, for binomial OCT parameters, the adjusted odds ratio was estimated, along with 95% confidence interval (CI). Since the observed values of OCT parameters were measured for the lesions, the above estimation was performed with the generalized estimate equation under the assumption of the intra-subject correlation structure of compound symmetry. The least square method was used for the glycemic metabolic variables because of an observed value per person. For both estimations, as independent variables, the baseline value of each parameter along with group variables were included in the statistical model.
For clinical outcomes, risk difference and 95% CI were estimated. A statistical test between the two groups based on z-statistics for OCT parameters, t-statistics for the glycemic metabolic variables, and Fisher’s exact test for the clinical outcome were performed and both sides p values were reported. We did not conduct any adjustment for multiplicity for statistical test, as concrete primary and secondary endpoint were not specified in the protocol. All statistical analyses were conducted using SAS software (version 9.4, SAS Institute, Cary, NC).