The comBined Insights From the Unified RAIN and COBIS bifurcAtion regisTries (BIFURCAT) database was created by merging two large datasets (16). The COBIS III (COronary BIfurcation Stenting [COBIS] III, NCT03068494) registry is a multicenter, nationwide cohort in South Korea that includes consecutive patients who underwent PCI for bifurcation CAD with second-generation drug-eluting stents (DESs) from January 2010 to December 2014 (17). The retrospective RAIN (veRy Thin Stents for Patients With Left mAIn or bifurcatioN in Real Life [RAIN], NCT03544294) registry included consecutive patients whose coronary bifurcations and/or unprotected left main coronary artery disease (LMCA) were treated with very thin DESs (<100 mm) in 15 European centers from June 2015 to December 2017 (18). The detailed information for both registries has been previously reported.

PCI was performed as per the current practice guidelines by the Korean Society of Interventional Cardiology and the European Society of Cardiology (12). Periprocedural anticoagulation was performed using either low-molecular-weight or unfractionated heparin to maintain an activated clotting time of 250–300 s during PCI. All patients received aspirin (300 mg) and P2Y12 inhibitor loading doses (300 or 600 mg for clopidogrel, 180 mg for ticagrelor, and 60 mg for prasugrel) before PCI, unless they had previously received these antiplatelet medications. The stenting strategy (provisional or upfront two-stent technique), access sites, type of implanted DES, and use of intracoronary imaging were left to the discretion of the operator. All patients were prescribed standard maintenance doses of dual antiplatelet drugs after bifurcation PCI. The duration of dual antiplatelet drugs was left to the decisions of the caring physicians.

A total of 5,537 patients from Korea and Italy were included in the BIFURCAT registry (Supplementary Figure S1). A detailed description of the study population, including the demographic, angiographic, and procedural data, was obtained using a web-based system. Additional data, such as the duration of prescribed antiplatelet therapy, was obtained from a review of electronic health records or telephone contact if necessary. All clinical outcomes from the participating centers were reviewed and validated by an independent clinical event-adjudicating committee.

The clinical characteristics of older patients (aged ≥75 years) were compared to those of younger patients (aged <75 years). The primary outcome was a composite of target vessel myocardial infarction (MI), clinically driven target lesion revascularization (TLR), and stent thrombosis (ST) at two years post-PCI. All-cause death and any MI were secondary outcomes, as were the single components of the primary outcome. MI was defined as an elevation of the cardiac enzyme level greater than the upper limit of normal with ischemic symptoms or signs indicative of ischemia, not related to the index PCI (19). TLR was considered as a repeat PCI of the lesion within 5 mm of the inserted stent. ST was classified as definite, probable, or possible using the definitions of the Academic Research Consortium (20). Both definite and probable cases were included as STs in this analysis. Outcomes of interest were appraised according to age at the time of the index PCI (aged ≥75 vs. <75 years).

Continuous variables were reported as means ± standard deviations, whereas categorical variables were reported as counts (percentages). Group comparisons were performed using parametric (unpaired t-test) and nonparametric (Mann–Whitney U test) tests for continuous variables, whereas the chi-square test was used for categorical variables. The cumulative incidence of clinical outcome events was calculated based on Kaplan–Meier censoring estimates, and intergroup comparisons were assessed using the log-rank test. The relationships between age and clinical outcomes were explored using restricted cubic splines (21). A multivariable Cox model was used to calculate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs). Variables with significance in the univariable analysis (p < 0.2) or clinically relevant variables were included in the final model to determine independent predictors of the primary outcome. The final model included age, hypertension, diabetes mellitus, chronic kidney disease (CKD), prior history of PCI and/or coronary artery bypass graft surgery, left ventricular (LV) ejection fraction <40%, presentation of acute coronary syndrome (ACS), LMCA, true bifurcation lesion, severe calcification, and stenting technique. True bifurcation lesions were defined as Medina (1,1,1; 1,0,1; and 0,1,1) lesions (22). To account for the effect of age, the sub-distribution HR was calculated using the Fine-Gray competing risk model (23). In addition, a propensity score (PS) matching analysis was also performed, and each patient in the older group was matched with those in the younger group at a 1:1 ratio using the nearest neighbor method, with a caliper width equal to 0.2 of the standard deviation of the logit PS. Statistical analyses were performed using R Statistical Software (version 4.1.2; R Foundation for Statistical Computing, Vienna, Austria). Except for the subgroup interaction analysis (p < 0.1), p values of 0.05 were considered statistically significant.

A total of 5,537 patients were included in the current analysis, and 1,415 patients (26%) were aged ≥75 years (the older group). The median age of all study patients was 67 (IQR, 58–75) years (Supplementary Figure S2). Table 1 presents the clinical characteristics, including baseline demographics. The mean age was 80.1 ± 4.0 years and 65.2% were men in older patients. Comorbidities, including hypertension (77.4% vs. 62.2%), CKD (23.7% vs. 9.3%), and history of MI (22.5% vs. 15.2%), were more frequent in older patients. Older patients presented more frequently with LMCA (33.1% vs. 28.6%, p = 0.001) and more complex coronary lesions, including a higher proportion of true bifurcation lesions (51.8% vs. 46.4%) and severe calcification (24.5% vs. 15.0%). Overall, PCI was performed with the two-stent strategy in 17.6% of patients, and older patients showed a shorter stent length of the main branch with a smaller diameter than younger patients.

During a median follow-up of 2.1 (IQR, 0.9–2.2) years, the cumulative incidence of the primary outcome was similar between older and younger patients (p = 0.208; Figure 1). Older patients had similar rates of target vessel MI, clinically driven TLR, and ST compared to younger patients (Table 2). After adjustment, advanced age was not associated with the primary outcome (adjusted HR, 1.02; 95% CI, 0.73–1.41). In contrast, higher rates of any MI (4.9% vs. 2.7%) or combined all-cause death and any MI (14.1% vs. 5.2%) were observed in older patients (Figure 2). As a continuous variable, age had no effect on the adjusted risk of the primary outcome, and it had a curvilinear effect on the adjusted risk of combined all-cause death and any MI (Figure 3).

The independent predictors of the primary outcome were CKD, LMCA, and the two-stent strategy (Table 3). LMCA was the strongest predictor of the primary outcome (adjusted HR 2.07; 95% CI, 1.53–2.79). After adjustment with the Cox regression model, age ≥75 years was not an independent factor associated with the primary outcome. When all-cause death was modelled as a single competing outcome, the adjusted sub-distribution HR of the primary outcome was 0.98 (95% CI, 0.69–1.38; p = 0.91) for the older age group (Table 4). The multivariable-adjusted independent predictors of the primary outcome according to age subgroup are shown in Supplementary Table S1. In the older group, the presence of LMCA was a significant predictor (adjusted HR, 2.35; 95% CI, 1.31–4.21; p = 0.004), and CKD (adjusted HR, 1.85; 95% CI, 0.94–3.63; p = 0.073) showed a borderline significance. In the younger group, hypertension, diabetes, CKD, LV ejection fraction <40%, LMCA, and the two-stent strategy were important in predicting the risk of the primary outcome. In the subgroup analysis according to diabetes, advanced age (≥75 years) showed a significant interaction for the primary outcome (p for interaction = 0.036; Supplementary Figure S3).

After 1: 1 PS matching, 2,634 patients were selected (the older group, n = 1,317). Demographic and clinical parameters at baseline were well balanced, and the standardized mean differences between the groups were <10.0% for all variables (Supplementary Table S2). The incidences of the primary outcome at 2 years were comparable between groups, and the rates of combined all-cause death and any MI were higher in older patients (Supplementary Table S3).

Although our study included a large number of Korean (47.8%) and European (52.2%) patients, it has a few limitations. Using real-world registry data, this retrospective observational study design was associated with an inherent selection bias. It should also be acknowledged that PCI techniques have improved since the initiation of each registry. Moreover, information on the use of intracoronary imaging was not available, making it difficult to reflect the contemporary performance of PCI. The COBIS III registry previously reported 39.8% (1054/2648) of intravascular ultrasound guidance during PCI (13). In addition, the present dataset included only the results of PCI with second-generation DES implantation, not allowing any direct comparison with older patients who were treated with medical therapy alone for bifurcation CAD. Some important patient-centered variables, such as frailty and functional disability in older patients, were not available, providing limited information to clarify high-risk patients who could derive the greatest benefit from complex invasive procedures. Lastly, the BIFURCAT database lacks information on bleeding events, which could be an additional relevant outcome in older patients undergoing complex PCI.