Tumor Determinants of Response to CAR-T Therapy in Relapsed or Refractory Mantle Cell Lymphoma

Background: While complete response rates to CD19-directed chimeric antigen receptor T-cell (CAR-T) therapy in relapsed or refractory (R/R) mantle cell lymphoma (MCL) exceed 70%, a substantial proportion of patients’ (pts) disease will ultimately relapse. Biomarkers are urgently needed to inform resistance mechanisms and identify pts at risk for poor outcomes in this setting. It remains unclear whether traditional histological, cytogenetic, and genomic features that predict outcomes with frontline chemoimmunotherapy and other agents in MCL can serve as informative biomarkers in the CAR-T setting. Therefore, we conducted a multicenter retrospective analysis to evaluate molecular and patient-level factors influencing response and survival in pts with R/R MCL receiving CD19 CAR-T.

Methods: In this international multicenter retrospective study, we included pts with R/R MCL treated with autologous CD19-directed CAR-T therapies. Comprehensive patient-, disease-, and treatment-related data were collected. Pre-CAR-T tumor genomic data gathered from targeted next-generation sequencing panels were available for a subset of pts. We evaluated the association between tumor features (histology, cytogenetics, and genomics) and CAR-T response using univariable logistic regression models. We used Cox regression models to analyze the relationship between predictive features and overall survival (OS) and progression-free survival (PFS) and Fisher's exact test between TP53 mutations and response.

Results: We included 88 pts from 5 international centers with a median age at infusion of 66 years (range 36-86). Brexu-cel was the predominant product (N=57) followed by liso-cel (N=21) and a point-of-care CD19 CAR-T product (N=10) with a CD28 costimulatory domain (Kedmi et al., JTCT 2022). Most pts (58%) received ≤3 lines of therapy prior to CAR-T.

High-risk disease features were frequent in the population; 46% had blastoid/pleomorphic morphology, 68% had K67 ≥50%, and 54% had high risk MIPIb (54%) at apheresis. Deletion 17p was observed in 19/60 (32%) pts and P53 was detected by immunohistochemistry in 32/56 (57%) pts. Among pts with available pre-CAR-T tumor NGS testing, the most frequent mutations were TP53 (28/52, 54%) and ATM (13/50, 26%). CDKN2A alterations were detected in 5/52 pts.

Our cohort had high overall response (86%) and CR rates (82%), comparable to results in the pivotal CAR-T MCL clinical trials. Grade ≥3 cytokine release syndrome and neurotoxicity occurred in 10% and 19% of patients, respectively.

With a median follow-up of 18 months (IQR 7.8-28.1), 1-year OS and PFS were 64% (95% CI: 54-77%) and 59% (95% CI: 48-72%), respectively. Blastoid/pleomorphic morphology, Ki67 > median of 60%, P53 expression (any percentage), SOX11 expression, deletion(17p), MIBIb group at apheresis, and LDH before lymphodepletion were not associated with OS or PFS from time of infusion.

In the subset of pts with NGS testing, TP53 and ATM alterations lacked significant association with these outcomes. For example, 2-year PFS differed by TP53 mutation status (52% vs. 39%), but this difference was not statistically significant. Of the features evaluated, only receipt of bridging therapy (66% of pts) was associated with inferior PFS (HR 2.39, 95% CI: 1.12-5.03) and OS (HR 2.16, 95% CI: 1.01-4.63). Additionally, TP53-mutated MCL was associated with a lower CR rate that did not meet statistical significance (96% vs. 74%, P = 0.053).

Discussion: In this comprehensive, multicenter/multinational dataset derived predominantly from pts treated in non-trial settings, molecular features traditionally associated with inferior outcomes in MCL did not significantly influence response or survival in pts with R/R MCL treated with CD19-CAR-T. Our findings, which warrant confirmation in larger genomically characterized cohorts, highlight the utility of CAR-T even in pts with high-risk MCL features. These data also underscore the pressing need for identifying novel biomarkers to inform on resistance mechanisms and predict treatment success. Data from additional patients will be incorporated if this work is selected for presentation at the meeting.