Assessing Risk for Poor Outcomes in Antiphospholipid Syndrome - PROJECT SUMMARY/ABSTRACT Background: This proposal is designed for Medha Barbhaiya, MD, MPH to grow towards becoming an independent investigator focused on developing and validating a novel method to identify antiphospholipid syndrome (APS) in large cohorts, evaluating its utility in identifying poor APS outcomes in multi-center EHR cohorts, and identifying discrete APS subphenotypes based on longitudinal autoantibody and biomarker data. Currently, there is no accurate way to identify APS patients in large cohorts, limiting understanding of the role of modifiable risk factors for outcomes across sociodemographic groups. Additionally, while certain aPL profiles may confer increased thrombotic risks, the extent to which novel biomarkers predict clinical outcomes is unknown. Preliminary data: For Aim 1, we have begun assessing the feasibility of developing the first algorithms for APS identification using structured EHR data. In the Hospital for Special Surgery (HSS) electronic health record (EHR), we have applied a broad screening filter to identify all potential APS patients (n=1,318 potential adult APS cases with ≥1 APS ICD-10-CM [D68.61] code since January 1, 2016). Under the guidance of her mentors, Dr. Barbhaiya will randomly select 200 of these subjects as a ‘training set’ for chart review to identify their true case status. For Aim 2, we have recently descriptively evaluated the APS ACTION registry, the largest and longest prospective registry of patients with antiphospholipid antibodies (aPL) to study associations between aPL profile and clinical events. We will now apply bioinformatics approaches to evaluate the association of novel APS biomarkers with aPL profile and clinical outcomes in this ongoing prospective registry. Methods: As part of this K23 award, we will develop algorithms using structured and unstructured data using natural language processing and machine learning approaches. We separately plan to subphenotype antiphospholipid antibody (aPL) patients in the ongoing prospective APS ACTION registry using novel biomarkers and aPL profile. We will use classical clustering methods as well as unsupervised machine learning to cross-sectionally and longitudinally evaluate for an association with APS clinical outcomes after adjusting for demographic and other clinical factors. Career Development: This proposal employs novel methods to address gaps in knowledge related to APS epidemiology. Dr. Barbhaiya is an Assistant Professor in Medicine and Population Health Sciences at Weill Cornell Medicine and an Assistant Attending at Hospital for Special Surgery with access to outstanding services and environment at these institutions. She has assembled a strong multi-disciplinary mentoring team and will be able to complete formal training in bioinformatics, biomarker assay interpretation, immunology, leadership and mentoring skills. This project has the potential to lead to future grants and studies, and will position Dr. Barbhaiya to achieve her objective to become an R01-funded independent investigator focused on improving APS outcomes.
