Predictors and Long-term Incidence of In-stent Restenosis After Chronic Total Occlusion Percutaneous Coronary Intervention

This study was approved by the ethics committee of Helsinki University Central Hospital (approval number HUS/12527/2022) and complied with the Declaration of Helsinki.

Patient selection Study participants were selected from a registry that included stable coronary artery disease patients with a chronic total occlusion (CTO) treated with percutaneous coronary intervention (CTO PCI) in Helsinki University Central Hospital 2014-2021. Optimal medical therapy was prescribed to all patients at the time of CTO PCI. All patients, who received surveillance imaging, gave written, informed consent prior to the beginning of the study.

Definition Segments extending 5 mm proximal and 5 mm distal to the stent were included as part of the stent for the evaluation of in-stent restenosis (ISR). CTO stents consisted of one or more coronary artery stents, potentially involving different stent types. Stent overlap was taken into consideration in cases of multiple stents. In coronary computed tomography angiography (CCTA) radiological ISR was defined as luminal narrowing of ≥ 50 % within the CTO stent. In native coronary arteries, stenosis was classified as minimal (<25 %), mild (25-49 %), moderate (50-69 %), severe (70-99 %), and total occlusion (100 %).

In CCTA, a non-contrast calcium score was obtained initially. This was followed by third-generation dual-source CCTA (Somatom Force, Siemens) using contrast medium (Omnipaque 350, GE Healthcare). Sublingual nitroglycerin was administered prior to scanning. The imaging protocol was optimised according to the patient's heart rate and in cases where the heart rate exceeded > 70/min, intravenous metoprolol was administered.

Patients' symptomatic status at the time of surveillance imaging was assessed using the Canadian Cardiovascular Society (CCS) classification via a telephone interview. For patients without study-driven surveillance imaging, the CCS classification at clinical imaging follow-up was determined based on electronic medical records.

CCTA results and clinical data were reviewed by the interventional cardiologists of the study, and invasive coronary angiography (ICA) was offered to patients with:

1. ISR or inconclusive finding in the long CTO stent.
2. Anginal symptoms with ≥ 50 % stenosis or inconclusive finding in any native coronary artery or non-CTO stent.

In ICA, ISR was defined as ≥ 50 % luminal narrowing in the stent. Physiological assessment (fractional flow reserve) and intravascular imaging (intravascular ultrasound, optical coherence tomography) were performed during ICA at the discretion of the interventional cardiologist. Any incidental findings were evaluated separately.

Patient-, lesion-, and procedure-related factors examined in the study:

age (years), body mass index (kg/m2), sex, smoking history, medical conditions at CTO PCI (diabetes, insulin required diabetes, heart failure, peripheral artery disease, dyslipidemia, hypertension, pulmonary disease, chronic kidney disease, atrial fibrillation, cardiac pacemaker, prior myocardial infarction, prior stroke, prior PCI, prior coronary artery bypass surgery (CABG)), double-CTO (two treated CTO:s in one procedure), primary procedure in-stent CTO, a non-CTO stent implanted in the CTO PCI procedure, clinical presentation (stable angina pectoris or acute coronary syndrome), stented vessel (right coronary artery, right coronary artery vein graft, left anterior descending coronary artery, left circumflex coronary artery), Japanese-CTO score (objective index of CTO procedure difficulty), access site (femoral or radial approach), procedural characteristis (antegrade or retrograde procedure, utilisation of a re-entry device), stent parameters (number of consecutive stents, stent length (mm), stent diameter (mm)), symptomatic status at stent imaging (assessed with CCS classification grade 0-IV)

To provide clinical context for the study's main research questions, other prespecified endpoints were: complications (access site complication, coronary perforation, pericardial centisis, left ventricle assist device, acute kidney injury, procedural death), all-cause death, time to all-cause death (years), cardiovascular death, myocardial infarction, stroke, target lesion revascularisation, stent thrombosis, any coronary artery revascularisation and combined clinical outcome (including all-cause death, MI, stroke and any revascularisation)

Statistical analysis Descriptive statistics were presented as median and interquartile range (IQR) or range (minimum-maximum) for continuous variables, and as frequency (%) for categorical variables. The temporal distribution of all-cause death and ISR was analysed using Kaplan-Meier curves, with two-sided 95% pointwise confidence intervals. For ISR analysis, patients were censored at the time of death, end of follow-up, or TLR in the absence of prior ISR. Fatal clinical outcomes were presented as Kaplan-Meier estimates. Non-fatal clinical outcomes were presented as Kaplan-Meier estimates, with censoring at death or end of follow-up; for stent thrombosis, patients were censored at death, end of follow-up, or TLR. Predictors of ISR were assessed using logistic regression models that included baseline patient and procedural variables. A p-value <0.05 was considered statistically significant. All variables were first analysed in univariate models and subsequently entered into multivariate analysis if statistical significance was reached. There were no missing data for the variables used in any analyses and no imputation was performed. Statistical analyses were performed with JMP version 18.2.0 (JMP Statistical Discovery LLC).

Descriptive statistics were presented as median and interquartile range (IQR) or range (minimum-maximum) for continuous variables, and as frequency (%) for categorical variables. The temporal distribution of all-cause death and ISR was analysed with Kaplan-Meier curves. Predictors of ISR were assessed using logistic regression models, that included baseline patient and procedural variables. A p-value <0.05 was considered statistically significant. All variables were first analysed in univariate models and subsequently entered into multivariate analysis if statistical significance was reached. Statistical analyses were performed with JMP version 18.2.0 (JMP Statistical Discovery LLC, Cary, NC, USA).