Farkouh, Michael E.

ABSTRACT Ischemic events after contemporary percutaneous coronary intervention (PCI) are uncommon but carry high morbidity/mortality. Identifying patients at highest risk is critical to guide dual antiplatelet therapy (DAPT) intensity while minimizing bleeding. Current machine learning (ML) model‐developed risk scores do not incorporate pharmacogenetic data. To develop and externally validate ML models that integrate clinical, demographic, and CYP2C19 genetic information to predict 1‐year ischemic outcomes post‐PCI to refine clinical decision making. We analyzed 8317 patients from TAILOR‐PCI trial ( n = 4572) and the Precision PCI registry ( n = 3745). The outcome was a composite of cardiovascular death, myocardial infarction, stroke, and stent thrombosis at 1 year. Boruta feature selection identified 11 predictors. Multiple ML algorithms were trained in TAILOR‐PCI and validated in Precision PCI using cross‐validation and synthetic minority oversampling (SMOTE). Model performance was assessed by area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. The best external performance was achieved with a support vector machine (SVM, polynomial kernel) (AUC 0.667; sensitivity 0.871; specificity 0.282), while XGBoost provided a more balanced profile (AUC 0.619; sensitivity 0.442; specificity 0.688). Variable importance for the SVM polynomial model demonstrated that all 11 included Boruta feature selected predictors had relatively high importance (> 75). ML models trained on large clinical trial and real‐world registry datasets can help identify the small subset of patients at high ischemic risk after PCI. This distinction is of relevance because ischemic events are rare, and most patients may safely de‐escalate DAPT to reduce bleeding risk while maintaining ischemic protection through a multimodal approach to risk stratification. Trial Registration: TAILOR‐PCI URL: https://clinicaltrials.gov/ct2/show/NCT01742117

Abstract Aims Alcohol and its relationship with cardiovascular disease (CVD) remain contentious. Several observational studies have found cardioprotective associations, while Mendelian randomization studies have found null or harmful effects of moderate drinking. Biases, including abstainer misclassification and sick-quitter effects, may obscure true associations. It is widely understood that former drinkers are often sick quitters. However, removal of former drinkers from analyses addresses bias in the abstainer group but fails to account for induced survivorship among current drinkers. This study sought to determine whether adjusting for biases altered the association between alcohol and CVD by comparing conventional models with those using (i) occasional drinkers as the reference group instead of lifetime abstainers and (ii) ‘intention-to-treat’ reallocation of former drinkers based on past drinking patterns. Methods and results The Multi-Ethnic Study of Atherosclerosis is a prospective cohort study of 6814 participants aged 45–84 years between 2000 and 2002 without baseline CVD. A total of 6755 participants were included after excluding those with missing data. Alcohol consumption was self-reported and classified as lifetime abstainer, former, or current drinker [occasional (≤1 drink/week), light (2–7 drinks/week), moderate (8–14 drinks/week), heavy (>14 drinks/week)]. Outcomes included coronary artery calcium (CAC; 0, 1–100, 101–300, and >300) and clinical events [myocardial infarction (MI), stroke, cardiovascular mortality, composite major adverse cardiovascular events (MACEs), and heart failure]. Of 6755 participants, 20.6% were lifetime abstainers, 24.1% former drinkers, and 55.3% current drinkers. Adjusting for biases altered the observed relationship between alcohol and CVD, showing increased harm with light drinking, decreased protection with moderate drinking, and a dose–response relationship between alcohol and CAC, particularly CAC >300 in adjusted models. Protective associations between moderate drinking and MI [hazard ratio (HR): 0.64, 95% confidence interval (CI): 0.38–1.06] and cardiovascular mortality (HR: 0.81, 95% CI: 0.53–1.25) were attenuated and became non-significant. Light drinking shifted from null to increased risk for MI (HR: 1.46, 95% CI: 1.11–1.92), stroke (HR: 1.54, 95% CI: 1.17–2.03), MACE (HR: 1.33, 95% CI: 1.12–1.57), and heart failure (HR: 1.42, 95% CI: 1.10–1.84). Conclusion Associations between alcohol and CVD in observational studies are sensitive to methodologic variation. Results from models correcting for abstainer/sick-quitter biases may help explain discrepant results from previous studies.

Introduction: Identifying patients at risk for ischemic events after percutaneous coronary intervention (PCI) relies on traditional analysis of limited clinical and imaging variables. Machine learning (ML) has shown promise in effectively predicting cardiovascular risk in population studies. While existing ML models mainly predict mortality and incorporate clinical variables, there is a lack of tools that have utilized genetic data and that predict ischemic events. Aims: This study aims to develop and validate a ML model incorporating genotyping and clinical data to enhance prediction of ischemic outcomes for PCI patients utilizing large prospectively derived diverse datasets. Methods: Patients from the TAILOR-PCI trial (n=5302) were utilized for model development. 50% of the sample was utilized for Boruta feature selection and 50% for training and testing using cross validation. Features included demographics, medical history, medications, PCI characteristics, and genetic data (specifically, CYP2C19 *2, *3, *17 alleles). The primary endpoint was a composite of cardiovascular death, myocardial infarction, stroke, stent thrombosis, and severe recurrent ischemia at 12 months. Multiple ML classification algorithms, including Support Vector Machine (SVM) polynomial, Random Forest, Light gradient boost, XG Boost, among others, were benchmarked for their prediction performance on rare events. The top performing classifiers were externally validated on an independent dataset from the PRECISION PCI study (n=3,745). Results: Mean participant age of the training set was 64.2 ± 11.0 years, with 75.4% being male. During follow-up of 12 months, among 4,572 patients in the entire cohort 343 (7.5%) met the primary outcome. The SVM polynomial model demonstrated the highest area under the curve (AUC) of 0.67 for predicting the primary outcome with test dataset. The sensitivity, specificity, precision, and recall were 0.87, 0.28, 0.07, and 0.87 respectively (Figure). Peripheral arterial disease, body mass index, and age were among the top variables by feature importance. Conclusion: ML models incorporating both clinical and genetic data are feasible and highly promising in predicting major adverse cardiac events that may help guide use of anti-platelet drug therapy. The AUC values are reasonable given imbalances and misclassifications in datasets, and further model optimization with prospective utilization of the model will be paramount.

Background TAILOR‐PCI (Tailored Antiplatelet Initiation to Lessen Outcomes due to decreased Clopidogrel Response After Percutaneous Coronary Intervention) studied genotype‐guided selection of antiplatelet therapy after percutaneous coronary intervention versus conventional therapy with clopidogrel. The presence of CYP2C19 loss‐of‐function alleles in patients treated with clopidogrel may be associated with increased risk for ischemic events. We report a prespecified sex‐specific analysis of genotyping and associated cardiovascular outcomes from this study. Methods and Results Associations between sex and major adverse cardiac events (MACE: cardiovascular death, myocardial infarction, stroke, stent thrombosis, and severe recurrent ischemia) and Bleeding Academic Research Consortium (BARC) bleeding at 12 months were analyzed using Cox proportional‐hazards models. Among 5276 randomized patients, loss‐of‐function carriers were observed in ≈36% of both sexes, and >80% of carriers were heterozygotes. At 12 months, after adjustment for baseline differences, risks of MACE (HR , 1.28 [0.97 to 1.68]; P =0.088) and BARC bleeding (hazard ratio [HR], 1.36 [0.91 to 2.05]; P =0.14) were comparable among women and men. There were no significant interactions between sex and treatment strategy for MACE interaction P value ( P int =0.59) or BARC bleeding ( P int =0.47) nor for sex and genotype (MACE P int =0.15, and BARC bleeding P int =0.60). Conclusions CYP2C19 loss‐of‐function alleles were present in ≈1 in 3 women and men. Women had similar adjusted risks of MACE and bleeding as men following percutaneous coronary intervention. Genotype‐guided therapy did not significantly reduce the risk of MACE or bleeding relative to conventional therapy for both sexes. Registration URL: https://www.clinicaltrials.gov ; Unique identifier: NCT01742117.

Background: Ischemic event rates are low after percutaneous coronary intervention (PCI), conversely second generation P2Y12 inhibitors increase bleeding therefore identifying patients at risk to enable precision in prescribing dual anti-platelet therapy (DAPT) is of clinical interest. Prior machine-learning predictive analyses have been limited by use of older generation stents and lack of genetic data. We utilized clinical and genetic data of patients with new generation stents enrolled in TAILOR-PCI, a 40-center international trial to predict ischemic events after PCI. Methods: There were 4398 eligible post PCI patients with acute coronary syndromes or stable coronary artery disease who were randomly split into training (n=3299) and validation (n=1099) cohorts. The primary endpoint comprising of cardiovascular death, myocardial infarction, stroke, or stent thrombosis within 12 months of PCI was predicted with patient characteristics prior to hospital discharge. First, random forests were used for feature selection with 1,000 bootstraps comprising equal numbers of those with and without ischemic events to minimize the bias of sample imbalance. Features were selected by a consensus on variable-importance over bootstraps. Second, using these features in the training cohort, support vector machines (SVM) with a polynomial kernel was trained with repeated 5-fold cross validation to choose a model with highest area under the receiver operating characteristic curve (AUC) which was then used to predict ischemic events in the validation cohort. Results: The primary endpoint occurred in 2.4% of patients. Using 89 clinical and genetic variables in the random-forest feature selection step, variable importance-based consensus identified 50 top predictors. These 50 predictors achieved AUC=0.703 in the SVM training cohort and AUC=0.744 (specificity=0.84; sensitivity =0.43) in the SVM validation cohort. Conclusions: Machine learning methods using baseline clinical and genetic data enables prediction of infrequent ischemic events in post PCI patients receiving new generation stents. The role of such artificial intelligence algorithms in individualization of DAPT by identifying high or low risk patients needs to be explored.

Background In TAILOR‐PCI, genotype‐guided selection of P2Y 12 inhibitors after percutaneous coronary intervention did not significantly reduce the risk of ischemic events at 12 months. The Age, Body Mass Index, Chronic Kidney Disease, Diabetes, and Genotyping (ABCD‐GENE) score identifies patients with high platelet reactivity on clopidogrel at increased risk of ischemic events. The aim of this study was to investigate the value of the ABCD‐GENE score for tailoring P2Y 12 inhibitor selection after percutaneous coronary intervention. Methods and Results In a post hoc analysis of the TAILOR‐PCI, outcomes were analyzed by ABCD‐GENE score and allocation to genotype‐guided or conventional P2Y 12 inhibitor selection. Primary (death, myocardial infarction, or stroke) and secondary (cardiovascular death, myocardial infarction, stroke, stent thrombosis, or severe recurrent ischemia) outcomes were assessed. Among 3883 patients discharged on clopidogrel in the genotype‐guided and conventional therapy groups, 15.8% and 84.2% had high (≥10 points) or low (<10) ABCD‐GENE scores, respectively. At 12 months, both the primary (5.2% versus 2.6%, P <0.001) and secondary outcomes (7.7% versus 4.6%, P =0.001) were significantly increased in patients with high ABCD‐GENE score. Among 4714 patients allocated to genotype‐guided or conventional therapy, the former did not significantly reduce the 12‐month risk of the primary and secondary outcomes in both the high and low ABCD‐GENE score groups (p interaction =0.48 and 0.27, respectively). Conclusions Among patients with percutaneous coronary intervention on clopidogrel, the ABCD‐GENE score was helpful in identifying those at higher risk. The ABCD‐GENE score may potentially enhance the precision of tailored selection of P2Y 12 inhibitors, which needs to be confirmed in prospective investigations. Clinical Trial Registration URL: http://www.clinicaltrials.gov . Unique Identifier: NCT01742117.

Introduction: The benefits of optimal medical therapy (OMT) with or without percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG) surgery have been evaluated in patients with Type 2 Diabetes Mellitus (T2DM) and stable coronary artery disease (CAD) but analyses have not been stratified by detailed angiographic CAD burden or left ventricular ejection fraction (LVEF). Methods: A patient-level meta-analysis was undertaken among T2DM patients enrolled in COURAGE, BARI 2D and FREEDOM to assess the impact of randomization to OMT, PCI + OMT, or CABG + OMT on a composite end-point of Death/MI/Stroke and stratified by LVEF (≥ 50%, < 50%), 1, 2 or 3 vessel disease (VD) (% diameter stenosis ≥ 50%) and presence or absence of proximal left anterior descending (pLAD) disease. Hazard ratios (HR) for treatment groups were calculated from Cox regression models including all patients adjusting for trial and revascularization stratum in BARI 2D. Multivariable adjustment for risk factors was performed as a sensitivity analysis. Results: There were 5,034 patients included in the analysis (BARI 2D: n=2,368, COURAGE: n=766, FREEDOM: n=1,900) of whom 17% had LVEF < 50%, 29% had pLAD, and 77% had 2 or 3 VD. A total of 1,591 patients were randomized to OMT alone, 2,118 to PCI + OMT, and 1,325 to CABG + OMT. There were 1,116 events during a median 4.5 year follow-up. There was no significant effect on outcomes with OMT vs PCI + OMT irrespective of LVEF, presence/absence of pLAD and number of VD. By contrast, CABG + OMT significantly reduced the risk of Death/MI/Stroke in patients with 2 or 3 VD, irrespective of LVEF and presence/absence of pLAD (see Table). Results were similar with multivariable adjustment. Conclusions: In patients with T2DM and stable CAD, OMT alone appears to be as effective as PCI + OMT while CABG + OMT significantly reduces cardiovascular events in those with multivessel CAD, irrespective of LVEF or presence of pLAD.

Abstract Background/Introduction TAILOR-PCI is the largest cardiovascular genotype-based randomized trial (NCT#01742117) investigating whether genotype-guided selection of oral P2Y12 inhibitor therapy improves ischemic outcomes after percutaneous coronary intervention (PCI). The TAILOR-PCI Digital Sub-Study tests the feasibility of extending original follow-up of 1 year to 2 years using state-of-the-art digital solutions. Deep phenotyping acquired during a clinical trial can be leveraged by extending follow-up in an efficient and cost-effective manner using digital technology. Purpose Our objective is to describe onboarding and engagement of participants initially recruited in a large, pragmatic, international, multi-center clinical trial to a digital registry. Methods TAILOR-PCI participants, within 23 months of their index PCI, were invited by letters containing a URL to the Digital Sub-Study website (http://tailorpci.eurekaplatform.org). These invitations were followed by phone calls, if no response to the letter, to determine reason for non-participation. A NIH-funded direct-to-participant digital research platform (the Eureka Research Platform) was used to onboard, consent and enroll participants for the digital follow-up. Participants were asked to answer health-related surveys at fixed intervals using the Eureka mobile app and desktop platform. To capture hospitalizations, participants could enable geofencing to allow background location tracking, which triggered surveys if a hospitalization was detected. Result(s) Letters were mailed to 893 of 929 eligible participants across 22 sites in the United States and Canada leading to 226 homepage visits and 118 registrations. There were 107 consents (12.0% of invited; mean age: 66.4±9.0; 19 females [18%]): 47 (44%) participants consented after the letter, 36 (34%) consented after the 1st call and 24 (22%) consented after a 2nd call. Among those who consented, 100 were eligible (7 did not have a smartphone) 81 downloaded the study mobile app and 73 agreed for geofencing (Figure 1). Among the 722 invited participants who were surveyed, 354 declined participation: due to lack of time (146; 20.2%), lack of smartphone (125; 17.3%), difficulty understanding (41; 5.7%), concern about using smartphone (34; 4.7%), concern of data privacy (14; 1.9%), concerns of location tracking (6; 0.8%) and other reasons (57; 7.9%). Conclusion Extended follow-up of a clinical trial using a digital platform is feasible but uptake in this study population was limited largely due to lack of time or a smartphone among participants. Based on data from other digital studies, uptake may also have been limited since digital follow-up consent was not incorporated at the time of consent for the main trial. Figure 1. Onboarding of the digital substudy Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): National Institute of Health (NIH), National Heart, Lung, and Blood Institute (NHLBI)