Study design

Randomized, comparator-controlled, open label, assessor-blinded multicentric trial (Södersjukhuset in Stockholm, Sweden and Sahlgrenska University Hospital in Gothenburg, Sweden). The European Cardiovascular Research Center (CERC) core lab analyzed the OCT-derived endpoints and was blinded to the patient’s identity, allocation arm and the baseline clinical characteristics. The CERC is a core lab totally independent from our research institution. The study was approved by the Ethical Review Board of the Stockholm County Regional Council, the Medical Products Agency and the Swedish Radiation Safety Authority and complies with the current ICH E6 (R2), good clinical practice guidelines and the Helsinki declaration.

Study population

We aimed to include female and male subjects aged 18 to 80 years old with known or newly diagnosed T2D with a HbA1c between 47–110 mmol/mol (DCCT 6.5–12.2%), who were eligible for PCI with implantation of DES. Indication for PCI included stable angina, unstable angina and non-ST elevation myocardial infarction (NSTEMI). Exclusion criteria are listed in Additional file 1: Table S1. Figure 1 shows the study inclusion flow. Screening failure summary is available in Additional file 1: Table S2.

Fig. 1

Study participants flow chart. Participants enrolled and finally analyzed are shown in the figure. CABG Coronary Artery By-pass Grafting, OCT Optical Coherence Tomography

Study procedures and exposure

PCI was performed including a post-PCI OCT recording. Thereafter, participants were randomized either to the study drug exenatide and standard treatment, or standard treatment alone and were followed for 12 weeks. During these 3 months, 2 telephone visits were held to ensure participant safety. Study drug adherence was controlled by counting pens and tablets. At 12 weeks, a new coronary angiography including OCT examination was performed to assess strut coverage and neo-intimal growth. Figure 2 provides a summary of the methods of the study.

Fig. 2

Methodology and main results. Panel A to C summarizes the methodology of the study. A: A coronary stenosis in a patient with T2D suitable for stenting is found, dilated, and stented with a Medtronic’s Resolute Onyx® DES. Then, OCT is performed to ensure optimal stent expansion and vessel wall apposition. B: Participants are randomized to either exenatide (subcutaneous injection once weekly) over standard treatment or standard treatment alone. During the 12 weeks of treatment patients are contacted twice and are interviewed to detect possible adverse events. C: A new coronary angiography including an OCT of the region of interest is performed to assess strut coverage, luminal and stent measurements as well as neo-intima measurements. D: Main results for each treatment group. DES Drug eluting stent, OCT optical coherence tomography, T2D type 2 diabetes, NIT neo-intimal thickness, NIHS neo-intima hyperplasia stenosis

Catheterization protocol, intracoronary measurements, and angioplasty

Cardiac catheterization was performed following local standard procedures; radial access was preferred if no contraindications existed, all patients were on chronic treatment with acetylsalicylic acid or were given a loading dose of 300 mg the day before the procedure and all patients were loaded with a P2Y12 inhibitor (ticagrelor or clopidogrel) before PCI. Routine angiography cines were acquired for complete anatomic evaluation and stenosis severity was visually assessed. If considered necessary, intracoronary physiologic assessment with fractional flow reserve was performed to establish functional lesion severity. At the decision of PCI, 0.75 mg/kg enoxaparine or 50–100 E/kg unfractioned heparin was administrated. Six French guide catheters were used for angioplasty and OCT recordings.

Stent implantation

Implantation technique was chosen at the operators’ discretion, but maximal stent expansion including post-dilatation with non-compliant balloons was encouraged in all cases. All patients received a Resolute Onyx® DES (Medtronic, Minneapolis, MN, US), which has a single wire platform with 81 µm rounded strut cross-sections, a platinum-iridium core, and a cobalt-chrome shell. Resolute Onyx® DES elutes zotarolimus through a durable polymer (BioLinx) [9]. This stent was chosen for its known good performance in terms of arterial healing, having low rates of malapposition and high percentage of strut coverage in previous observational studies [10].

OCT image acquisition protocol and analysis

After stent implantation, intracoronary frequency domain-OCT recordings were obtained using the commercially available ILUMIEN OPTIS™ or OPTIS integrated™ systems with the Dragonfly™ rapid exchange OCT catheter (Abbott, St Paul, MN, US). Intracoronary nitroglycerine was administered before starting the pullback. Blood displacement was achieved by manual injection of contrast and pullback speed was set to 54 mm/s. If suboptimal stent result such as malapposition were found in the immediate post-PCI OCT, further optimization was performed, and a new OCT was recorded. At follow-up, OCT recordings were obtained following the same protocol. OCT recordings were analyzed with Caas IV-LINQ software version 2.1 (Pie Medical Imaging Systems, Maastrich, The Netherlands) with a frame slice thickness of 0.1 mm for 54 mm/s pullbacks and 0.2 mm on the occasions where pullback speed was set to 75 mm/s.

Exposure

After PCI, participants were randomized to either exenatide, i.e. Bydureon® 2 mg once weekly in subcutaneous injection plus standard treatment or standard treatment alone. Standard treatment was defined as metformin (target dose 1 g bid) and Neutral Protamin Hagedorn (NPH) insulin (subcutaneous injection at bedtime and dose-adjusted for every specific case to achieve fasting glucose levels of 6 mmol/L). This was to ensure an optimal and comparable glycemic control throughout the study population. Drug-naïve patients were given NPH insulin and metformin at randomization and were up-titrated to achieve target glycemic control or the maximal tolerated dose, i.e. metformin. Participants who were already insulin-treated continued with their existing insulin given that a good glycemic control had been previously achieved.

Follow-up

During the 12 weeks of follow-up, participants were asked to keep record of capillary glucose measurements. Four and eight weeks after randomization, telephone visits were held to review the self-reported glucose measurements. At 12 weeks, repeat coronary angiography with OCT examination was performed as detailed above.

OutcomesPrimary outcome

The percentage of stent strut coverage as assessed by OCT. A strut was deemed covered if tissue was identified above the struts (Fig. 3, Panel 1).

Fig. 3

Definitions of OCT derived endpoints. In the schematic images struts are represented as boxes with metallic gradient filling. NI is represented as delimited translucid green areas. Panel 1: Stent strut coverage. A well-apposed stent covered by a thin NI layer. Struts a and b are considered covered by NI while struts c, d and e are classified as un-covered. Panel 2: Coverage thickness. Detail of 3 struts with different NI coverage thickness. NI coverage is measured as 100 µm for strut f, 60 µm for strut g and only 30 µm for strut h. Strut h would be considered covered for the primary endpoint but uncovered for the > 40 µm coverage thickness (secondary endpoint). Panel 3: Significant malapposition. An area of stent malapposition is seen from 11 to 2 o’clock. The abluminal side of strut i is separated 440 µm from the vessel wall and classified as significantly malapposed. Strut j is separated 260 µm from vessel wall which classifies for non-significantly malapposed. Panel 4: NI thickness. NI thickness measures 270 µm from the luminal side of the strut to the luminal border. Panel 5: Stent area. Seeable struts delimit stent area displayed as a light blue line. Panel 6: NI hyperplasia stenosis is calculated as NI volume (NI area multiplied by frame thickness) divided by stent volume (stent area multiplied by frame thickness) multiplied by 100. OCT Optical coherence tomography, NI Neo-intima

Secondary outcomesOCT derived endpoints

The following endpoints were analyzed: percentage of strut coverage > 40 µm, stent strut significant malapposition (distance between the luminal part of the strut to intimal border > 300 µm), malapposition volume, mean neo-intimal area, neo-intimal hyperplasia area at maximal obstruction, percentage of stenosis by neo-intimal hyperplasia, neo-intimal hyperplasia volume at maximal obstruction and maximal neo-intimal thickness at maximal obstruction. Luminal and stent areas and volumes as well as stent expansion were measured at baseline and follow-up. Definitions for some selected OCT derived endpoints are given in Fig. 3. Remaining definitions for OCT endpoints are found in Additional file 1: Table S3.

At baseline, visual assessment of plaque composition, calcium characterization and the presence of thrombi and tissue protrusion was performed for the site of maximal obstruction (culprit lesion) as well as proximal and distal vessel references (within 5 mm to stent edges). The total number of struts analyzed per lesion and the mean number of struts per cross-section were also reported.

Clinical endpoints

Need for target lesion revascularization (within 12 weeks or at follow-up angiography). Anthropometric and biochemical endpoints were pre-defined as changes between baseline and follow-up: body weight, abdominal circumference, blood pressure, heart rate, HbA1c, hemoglobin, creatinine, estimated glomerular filtration rate (eGFR), triglycerides and total cholesterol and its fractions. eGFR was calculated following the Chronic Kidney Disease Epidemiology Collaboration equation and based on the serum creatinine levels [11].

Participant’s safety

Participants were actively asked about possible adverse events in each visit of the trial and the participant’s medical journal was examined. All adverse effects, regardless of relationship to study drug or protocol were recorded in the adverse event report form.

Statistics

According to their distribution, continuous data are summarized as mean and standard deviation or as median and percentile 25th–75th. Categorical data are presented as absolute count and percentage. To test differences in changes between baseline and follow-up, new delta variables were computed. To test differences between treatment groups for those delta variables a Student T-test was used whenever data distributed normally and Fisher’s exact test, Mann–Whitney U test or Chi2 test was used whenever data was not normally distributed. A two-sided p-value of less than 0.05 was considered statistically significant. All analyses were performed using IBM SPSS Statistics software, version 28.0. Armonk, NY: IBM Corp.

Sample size calculation

Initially, we calculated to investigate 84 patients (42 in each arm) to be able to demonstrate a mean absolute difference of 5 percentage units (specifically 90% vs 95%) in strut coverage with an alpha error of 5% (two sided) and a power of 80%. The chosen effect size is 5 percentage units of stent strut coverage as this value has been observed as a clinically relevant threshold to be related with stent-failure outcomes [12]. The sample size calculation was performed estimating a standard deviation of 8% of strut coverage from previous clinical data [13]. We foresaw that some subjects would drop out, and therefore aimed to investigate 100 subjects.