Machine learning-based staging system to predict chronic limb-threatening ischemia disease progression and survival: A secondary analysis of the BEST-CLI trial data

PROJECT SUMMARY This proposed study will comprehensively stage individual disease burden and mortality risk at initial diagnosis and across the course of care for patients with chronic limb threatening ischemia (CLTI). This debilitating, progressive condition is the most severe form of peripheral artery disease, and successful treatment requires ongoing combinations of endovascular or open revascularization, medical and wound/podiatric care. Because CLTI manifests in the extremities, limb-based classification systems like the existing Wound, Ischemia, and foot Infection (WIfI) staging system are important tools; however, the goals for CLTI treatment are often broader than limb salvage and include freedom from recurrent disease and survival. Despite the availability of multiple life- and limb-preserving treatments, success in improving health outcomes of patients afflicted with CLTI has been limited. The recently completed BEST- CLI randomized controlled trial with 1830 adult participants determined that surgical vein bypass is the most effective initial revascularization approach for patients with CLTI, but almost 10% required re-intervention to maintain blood flow to the limb, 10% underwent major limb amputation, and 33% died within two years. The need for frequent re-intervention for patients in the BEST-CLI study demonstrates that long-term CLTI and survival outcomes are dependent on the sequence of treatments beyond the initial revascularization strategy. Yet, how post-primary revascularization treatment impacts outcomes among different types of patients is unknown. Our proposed study addresses these limitations. Using precision medicine analytics and machine learning, we will leverage the rich BEST-CLI trial data to identify clusters of limb characteristics, anatomic patterns of atherosclerosis, and comorbidities that associate with differing levels of CLTI-free survival. Moreover, with the BEST-CLI longitudinal patient assessments, we will quantify the impact of different post-primary revascularization treatments on CLTI-free survival. This study will provide a deeper understanding of the limb- and life- preserving effect of different combinations of medical and surgical treatments over time, especially among patients who do not respond to first-line surgical interventions. Together, our results will help clinicians set expectations and, if warranted, change practice based on likely CLTI-free survival at initial diagnosis and over the lifespan of each patient. In the future, the evidence generated from our study will be incorporated into precision medicine clinical trials such sequential multiple assignment randomized trials (SMARTs) for discovering and testing optimal, individually tailored treatment regimens for CLTI.