AbstractAimsThe best interventional strategy for the treatment of drug-eluting stent (DES) in-stent restenosis (ISR) is still unclear and no data from randomized trials beyond 3-year follow-up are available. We aimed to define 10-year comparative efficacy and safety of plain balloon (PB), paclitaxel-coated balloon (PCB), and paclitaxel-eluting stent (PES) for percutaneous coronary intervention (PCI) of DES-ISR.Methods and resultsClinical follow-up of patients randomly assigned to PB, PCB, and PES in the ISAR-DESIRE 3 trial was extended to 10 years and events were independently adjudicated. The primary endpoint was a composite of cardiac death, target vessel myocardial infarction, target lesion thrombosis, or target lesion revascularization. The major secondary safety endpoint was a composite of cardiac death, target vessel myocardial infarction, or target lesion thrombosis. The major secondary efficacy endpoint was target lesion revascularization. Incidences by the Kaplan–Meier method were compared by the log-rank test. Risk estimation was primarily performed by Cox proportional hazards regression and supplemented by weighted Cox regression accounting for non-proportional hazards and Royston–Parmar flexible parametric regression with a time-varying coefficient. Primary results were further assessed by landmark, lesion-level, per-protocol, and competing risk analyses. A total of 402 patients (500 lesions) with DES-ISR were randomly assigned to PB angioplasty (134 patients, 160 lesions), PCB angioplasty (137 patients, 172 lesions), and PES implantation (131 patients, 168 lesions). Clinical follow-up did not significantly differ among treatments [PB, 9.62 (4.50–10.02) years; PCB, 10.01 (5.72–10.02) years; PES, 9.08 (3.14–10.02) years; P = 0.300]. At 10 years, the primary composite endpoint occurred in 90 patients (72.0%) assigned to PB, 70 patients (55.9%) assigned to PCB, and 72 patients (62.4%) assigned to PES (P < 0.001). The pairwise comparison between PCB and PES resulted in a non-significant difference [multiplicity-adjusted P = 0.610; Grambsch–Therneau P = 0.004; weighted Cox: hazard ratio (HR) 1.10, 95% confidence interval (CI) 0.80–1.51; Cox: HR 1.10, 95% CI 0.79–1.52; Royston–Parmar: HR 1.08, 95% CI 0.72–1.60]. The major secondary safety endpoint occurred in 39 patients (34.1%) assigned to PB, 39 patients (34.0%) assigned to PCB, and 42 patients (40.0%) assigned to PES (P = 0.564). Target lesion revascularization occurred in 71 patients (58.0%) assigned to PB, 55 patients (43.9%) assigned to PCB, and 42 patients (38.6%) assigned to PES (P < 0.0001). The pairwise comparison between PES and PCB resulted in a non-significant difference (multiplicity-adjusted P = 0.282; Grambsch–Therneau P = 0.002; weighted Cox: HR 0.83, 95% CI 0.56–1.22; Cox: HR 0.81, 95% CI 0.54–1.21; Royston–Parmar: HR 0.75, 95% CI 0.47–1.20). Lesion-level and per-protocol analyses were consistent. At landmark analyses, an excess of death and cardiac death associated with PES compared with PCB was observed within 5 years after PCI, though 10-year differences did not formally reach the threshold of statistical significance after adjustment for multiplicity. Competing risk regression confirmed a non-significant difference in target lesion revascularization between PCB and PES and showed an increased risk of death associated with PES compared with PCB.ConclusionTen years after PCI for DES-ISR, the primary and major secondary endpoints between PCB and PES were not significantly different. However, an excess of death and cardiac death within 5 years associated with PES and the results of the competing risk analysis are challenging to interpret and warrant further analysis. PES and PCB significantly reduced target lesion revascularization compared with PB.