P-019: Correlative Analysis to Define Patient Profiles Associated with Manufacturing and Clinical Endpoints in Relapsed/refractory Multiple Myeloma Patients Treated with Idecabtagene Vicleucel (Ide-Cel)

8021 Background: The anti-BCMA autologous CAR T cell therapy ide-cel has achieved deep and durable responses in RRMM patients (Munshi, N Engl J Med 2021). Prior studies have described the correlation between patient characteristics and clinical outcomes from CD19 CAR T cell therapy (Finney, J Clin Invest 2019; Fraietta, Nat Med 2018). We sought to define patient profiles correlated with manufacturing and clinical endpoints in RRMM patients treated with ide-cel in clinical trials using unsupervised clustering and multivariate machine learning across multiple key variable domains. Methods: Clinical and manufacturing data were harmonized across 164 RRMM patients treated with ide-cel in KarMMa (NCT03361748) and KarMMa-2 cohort 1 (NCT03601078). Based on individual multivariate importance, 10 selected peripheral blood mononuclear cell (PBMC), drug product (DP), and in-process cell growth variables were used to define patient clusters. Random forest classifiers were generated to identify patient characteristics associated with manufacturing and clinical endpoints. Wilcoxon rank sum and Kruskal-Wallis tests were used to compare 2 and 3+ categorical groups; Cox proportional hazards were used to compare groups with time-to-event data. Results: Using an unsupervised method, 4 patient clusters were identified that represented 10%, 57%, 15%, and 18%, respectively, of the total 164 patients. Cluster 4 was the most favorable and was characterized by a higher frequency of T cells in PBMCs; increased T-cell size during manufacturing; higher DP T-cell transduction, potency, and vector copy number; and ultimately a > 3-fold higher CAR T cell yield compared with the least favorable cluster 1. Cluster 2 contained most patients and was associated with intermediate manufacturing endpoints. Patients in cluster 4 had a higher complete response rate, longer progression-free survival, and were defined by lower tumor burden, higher absolute lymphocyte count (ALC), and longer washout period after alkylator treatment, among others (Table). Conclusions: The current study identified patient profiles in RRMM using accessible laboratory or medical history data that correlated with longitudinal outcomes. These findings may inform patients likely to achieve improved outcomes with CAR T cell therapy. Clinical trial information: NCT03361748. [Table: see text]