Predicting target lesion failure following percutaneous coronary intervention through machine learning risk assessment models.

Aims: Central to the practice of precision medicine in percutaneous coronary intervention (PCI) is a risk-stratification tool to predict outcomes following the procedure. This study is intended to assess machine learning (ML)-based risk models to predict clinically relevant outcomes in PCI and to support individualized clinical decision-making in this setting. Methods and results: Five different ML models [gradient boosting classifier (GBC), linear discrimination analysis, Naïve Bayes, logistic regression, and K-nearest neighbours algorithm) for the prediction of 1-year target lesion failure (TLF) were trained on an extensive data set of 35 389 patients undergoing PCI and enrolled in the global, all-comers e-ULTIMASTER registry. The data set was split into a training (80%) and a test set (20%). Twenty-three patient and procedural characteristics were used as predictive variables. The models were compared for discrimination according to the area under the receiver operating characteristic curve (AUC) and for calibration. The GBC model showed the best discriminative ability with an AUC of 0.72 (95% confidence interval 0.69-0.75) for 1-year TLF on the test set. The discriminative ability of the GBC model for the components of TLF was highest for cardiac death with an AUC of 0.82, followed by target vessel myocardial infarction with an AUC of 0.75 and clinically driven target lesion revascularization with an AUC of 0.68. The calibration was fair until the highest risk deciles showed an underestimation of the risk. Conclusion: Machine learning-derived predictive models provide a reasonably accurate prediction of 1-year TLF in patients undergoing PCI. A prospective evaluation of the predictive score is warranted. Registration: Clinicaltrial.gov identifier is NCT02188355.

Clinical Outcomes of Percutaneous Coronary Intervention for Bifurcation Lesions According to Medina Classification.

Background Coronary bifurcation lesions (CBLs) are frequently encountered in clinical practice and are associated with worse outcomes after percutaneous coronary intervention. However, there are limited data around the prognostic impact of different CBL distributions. Methods and Results All CBL percutaneous coronary intervention procedures from the prospective e-Ultimaster (Prospective, Single-Arm, Multi Centre Observations Ultimaster Des Registry) multicenter international registry were analyzed according to CBL distribution as defined by the Medina classification. Cox proportional hazards models were used to compare the hazard ratio (HR) of the primary outcome, 1-year target lesion failure (composite of cardiac death, target vessel-related myocardial infarction, and clinically driven target lesion revascularization), and its individual components between Medina subtypes using Medina 1.0.0 as the reference category. A total of 4003 CBL procedures were included. The most prevalent Medina subtypes were 1.1.1 (35.5%) and 1.1.0 (26.8%), whereas the least prevalent was 0.0.1 (3.5%). Overall, there were no significant differences in patient and procedural characteristics among Medina subtypes. Only Medina 1.1.1 and 0.0.1 subtypes were associated with increased target lesion failure (HR, 2.6 [95% CI, 1.3-5.5] and HR, 4.0 [95% CI, 1.6-9.0], respectively) at 1 year, compared with Medina 1.0.0, prompted by clinically driven target lesion revascularization (HR, 3.1 [95% CI, 1.1-8.6] and HR, 4.6 [95% CI, 1.3-16.0], respectively) as well as cardiac death in Medina 0.0.1 (HR, 4.7 [95% CI, 1.0-21.6]). No differences in secondary outcomes were observed between Medina subtypes. Conclusions In a large multicenter registry analysis of coronary bifurcation percutaneous coronary intervention procedures, we demonstrate prognostic differences in 1-year outcomes between different CBL distributions, with Medina 1.1.1 and 0.0.1 subtypes associated with an increased risk of target lesion failure.

Outcomes and regional differences in practice in a worldwide coronary stent registry.

OBJECTIVE: The primary objective was to assess the performance of a new generation thin-strut sirolimus-eluting coronary stent with abluminal biodegradable polymer in an all comer population. The secondary objective was to detail differences in contemporary percutaneous coronary intervention (PCI) practice worldwide. METHODS: e-Ultimaster was an all-comer, prospective, global registry (NCT02188355) with independent event adjudication enrolling patients undergoing PCI with the study stent. The primary outcome measure was target lesion failure (TLF) at 1 year, defined as the composite of cardiac death, target vessel myocardial infarction and clinically driven target lesion revascularisation. Data were stratified according to 4 geographical regions. RESULTS: A total of 37 198 patients were enrolled (Europe 69.2%, Asia 17.8%, Africa/Middle East 6.6% and South America/Mexico 6.5%) and 1-year follow-up was available for 35 389 patients (95.1%). One-year TLF occurred in 3.2% of the patients, ranging from 2% (Africa/Middle East) to 4.1% (South America/Mexico). In patients with acute coronary syndrome, potent P2Y12 inhibitors were prescribed in 48% of patients at discharge, while at 1 year 72% were on any dual antiplatelet therapy. Lipid-lowering treatment was administered in 80.9% and 75.5% of patients at discharge and 1 year, respectively. Regional differences in the profile of the treated patients as well as in PCI practice were reported. CONCLUSIONS: In this investigation with worldwide representation, contemporary PCI using a new generation thin-strut sirolimus-eluting coronary stent with abluminal biodegradable polymer was associated with low 1-year TLF across clinical presentations and continents. Suboptimal adherence to current recommendations around antiplatelet and lipid lowering treatments was detected.