Large B-cell lymphoma after CD19-CAR-T failure: Updated real-world outcomes, prognostic factors, and prediction models

Abstract Background While CD19-CAR-T cell therapy has revolutionized treatment for relapsed/refractory (R/R) large B-cell lymphoma (LBCL), a substantial proportion of patients experience treatment failure. Previous studies demonstrated poor outcomes with conventional salvage therapies. However, the impact of novel treatments, particularly CD19-CD3 bispecific antibodies, on outcomes in this patient population remains unclear. Additionally, prognostic factors to predict outcomes after CAR-T failure have not been well characterized. MethodsWe conducted a single-center, retrospective analysis of all R/R LBCL patients who received CD19-CAR-T therapy at Oregon Health glofitamab or epcoritamab), radiotherapy alone (13%), anti-CD19 therapies (6%; tafasitamab or loncastuximab), other targeted treatments (12%), and salvage chemotherapy ± rituximab (7%). The highest complete response (CR) rates were achieved with bispecific antibodies (38%), followed by anti-CD19 agents (33%) and polatuzumab-based regimens (27.8%). Salvage chemotherapy ± rituximab failed to induce CR in any patient. Overall response rates (ORR) were highest with polatuzumab-based regimens (55.6%), other targeted therapies (50.0%), and bispecific antibodies (46%). Radiotherapy alone achieved CR in 23% of patients with an ORR of 31%. No statistically significant differences in response rates, PFS, or OS were observed between treatment groups. Patients relapsing after 2022 showed no improvement in survival compared to those relapsing before 2022 (median OS 8.77 vs. 7.80 months, p=0.77). Univariate analysis identified prognostic factors of 1-year HR: shorter interval from CAR-T to relapse HR 0.994 (95%CI 0.991, 0.998); p=0.002), ECOG performance status ≥2 at subsequent treatment initiation [HR 2.59 (95%CI 1.36, 4.95); p=0.006, and LDH 1.5 x upper normal limit HR 3.41 (95%CI 1.81, 6.40); p=0.002. Age at subsequent treatment ≥65 years showed a trend toward worse survival (HR 1.76 95%CI 0.93, 3.32, p=0.079). Prediction models were developed to estimate 12-month overall survival at subsequent treatment using these prognostic factors. The apparent and optimism-corrected c-statistic of the best performing model was 0.78 and 0.76, respectively. ConclusionOutcomes for patients with LBCL after CD19-CAR-T failure remain poor despite the availability of novel therapies including bispecific antibodies. The identified prognostic factors and prediction model may assist clinicians in patient counseling. Further research is needed to optimize therapeutic strategies for these challenging patients.