ABSTRACT
INTRODUCTION: The prognosis for large B-cell lymphoma (LBCL) patients who did not respond or relapsed after chimeric antigen receptor (CAR)-T therapy remains dismal, with no established consensus on the most effective salvage regimen. METHODS: We conducted a random-effects meta-analysis of complete response (CR) and overall response rates (ORR) to first-line treatments for CAR-T-relapsed/refractory LBCL. We followed the predefined protocol available at PROSPERO (CRD42023473854). RESULTS: We identified 41 studies evaluating the following interventions: non-CD19 CAR-T, CD19 CAR-T, bispecific antibodies (BiTEs), lenalidomide- and polatuzumab-based regimens, radiotherapy, immune checkpoint inhibitors (ICI), Bruton's Tyrosine Kinase inhibitors (BTKi). Non-CD19 CAR-T cells yielded the best CR (56%, CI: 40-71%), significantly higher than other interventions except CD19 CAR-T (CR = 30%, CI: 7-58%). BiTEs, radiotherapy, lenalidomide- and polatuzumab-based regimens (CR: 28%, 26%, 19%, 24% respectively) did not differ significantly from each other. ICI and BTKi showed the lowest CR rates (12%, CI: 5-20% and 8%, CI: 0-23%, respectively), and were also significantly inferior to BiTEs. ORR was the highest for non-CD19 CAR-T (ORR = 80%, CI: 66-92%), whereas all other regimens yielded values below 50%. CONCLUSIONS: Non-CD19 CAR-T cells were associated with higher response rates and should be considered if patients are eligible. Given the heterogeneity of the estimates, the results should be interpreted cautiously. REGISTRATION: PROSPERO CRD42023473854.