Comprehensive Summary
This study evaluates the use of an AI-based model to predict early relapse in patients undergoing CAR T-cell therapy for large B-cell lymphoma (DLBCL). The electronic health records and mortality data from 416 patients treated with axi-cel were analyzed across six University of California Health centers between 2017 and 2024. Wang et al. developed a decision tree machine learning (ML) model incorporating patient age, LDH, CRP, ferritin, hematocrit, platelet count, and PT levels to classify the risk of relapse in patients within six months of treatment. The model achieved a high predictive performance (AUROC) level of 0.82 and effectively categorized patients into high and low-risk cohorts. The median progression-free survival was 10.1 months, and the median overall survival was 54.4 months, with side effects including CRS and neurotoxicity observed in 18.8% and 32.5% of patients. These results show the potential of computational tools to enhance early detection and improve health outcomes for patients undergoing axi-cel therapy.
Outcomes and Implications
Over 57% of patients experience relapse after undergoing axi-cel treatment, emphasizing the need for interventions that can identify high-risk patients post-treatment. The ML model developed by Wang et al. demonstrates a promising tool for overall risk assessment and patient health monitoring both during and after axi-cel therapy. This study highlights how AI-based approaches can be integrated into treatment plans to enhance long-term outcomes for patients with DLBCL and enable more timely, patient-focused interventions.