Outcome of patients with large B-cell lymphoma relapsing after second-line CAR-t: Insights from the descar-t registry

Abstract

Abstract Background: Relapse after CAR T-cell therapy in large B-cell lymphoma (LBCL) is associated with poor prognosis, lacking a clear salvage standard and supported by limited evidence. Patients receiving second-line (2L) CAR T-cells typically have high-risk features, such as chemo-refractory disease or early relapse after frontline therapy. Those who fail 2L CAR T-cells represent a particularly challenging subgroup. We evaluated outcomes and the impact of salvage therapies in LBCL patients relapsing after 2L CAR T-cell treatment (trt). Methods: We performed a retrospective analysis using the French DESCAR-T registry. Salvage trts were categorized as: BsAbs monotherapy (BI), BsAbs-based combinations (BIC: BsAbs + chemo or IMiDs), chemotherapy (CH: DHA(OX/C/P) or ICE), and other therapies (targeted agents, IMiDs, radiotherapy). The primary endpoint was overall survival after post–CAR T-cell relapse (OS2). Secondary endpoints included overall response rate to salvage therapy (ORR2), progression-free survival from relapse (PFS2) and PFS2/OS2 by salvage trt. Propensity score weighting was used to adjust for baseline differences. Results: Among the 893 pts with R/R LBCL treated with anti-CD19 CAR-T cells in 2L (axi-cel:85.2%; liso-cel:4.8%), 297 (33%) experienced disease progression. Among pts who experienced relapse after 2L CAR T-cell therapy, the best response prior to progression was complete or partial remission (CR/PR) in 70.4% (n=171), whereas 29.6% (n=72) had stable or progressive disease (SD/PD). At relapse, these pts had a median age of 62 years (range, 19–86), and 24% were ≥70, 47% were ECOG≥2, LDN>ULN: 48% and CRP≥30 mg/L: 7.8%. Median time from CAR T-cell to relapse was 2.7 months (IQR 1.4–3.3); 33% relapsed within 2 months, 55% between 2–6 months, and 12% after 6 months. Of the 297 pts, 231 (78%) began salvage trt within a median of 15 days post-relapse (IQR: 7–34). Baseline characteristics were similar between treated and non-treated pts, except for older age in the non-treated group (67 vs 61 years; p=0.033). Salvage regimens included BI (n=121, 52.4%), BIC (n=29, 12.6%), CH (n=29, 12.6%), and other (n=52, 22.5%). Among 156 evaluable pts, ORR2 was 39.1% including CR in 27.6% and PR in 11.5%. CR rates were higher in pts with ECOG 0–1 (25.0% vs 10.7%) and pts with late relapsed: 15.6% when <2 months, 16.5% between 2–6 months, and 30.7% when >6 months. The ORR after trt was 39.6% in the BI group, 29.5% in the BIC group, 50% in the CH group, and 33.3% in the other trt group, with corresponding CR rates of 29.6%, 10.9%, 41.7% and 18.1%. The median PFS2 was 3.4 months (IQR 1.8–11.5). Median PFS2 by trt was 3.7 months for BI, 2.9 months for BIC, 2.4 months for CH, and 1.7 months for other therapies. PFS2 was numerically longer with BI: HR 0.58 (BI vs other; 95%CI, 0.37–0.89) with a global p-value <0.001). After a median follow-up of 8.4 months (IQR 3.4–11.3), median OS2 was 6.5 months (IQR 3.2–22.4). In univariate analysis, the following variables, assessed at the time of relapse, were associated with longer OS2: lower ECOG (0–1 vs ≥2; HR 0.64, 95%CI 0.44–0.94, p=0.021), CRP<30 mg/L (HR 0.39, 95%CI 0.18–0.87, p=0.022), higher hemoglobin (per 1 g/dL; HR 0.91, 95%CI 0.83–1.00, p=0.047), and directly better prior response to CAR T-cells (CR/PR vs SD/PD; HR 0.52, 95%CI 0.35–0.76, p<0.001). No significant association was found for refractory pts to frontline therapy, histology subtype, CAR T product, response to bridging therapy, early progression after CAR T-cells, LDH/aaIPI/Ann Arbor stage/ bulk disease or neutrophil counts at progression. In multivariate analysis, CRP<30 mg/L (HR 0.41, 95%CI 0.18–0.91, p=0.029) and ECOG<2 (HR 0.59, 95% CI 0.36–0.99, p=0.046) remained independent OS2 predictors. By subgroups, median OS2 from trt start was 7.1, 4.9, 6.3 and 5.9 months for pts treated with BI, BIC, CH, and other therapies, respectively. In the weighted cohort (n=231), covariate balance was achieved across nine covariates (pairwise ASMD<0.3). OS2 was numerically longer with BI: HR 1.30 (CH vs BI; 95%CI, 0.76–2.24), 0.55 (BI vs BIC; 0.31–0.95), and 0.64 (BI vs other; 0.40–1.03), with a global p-value of 0.0094. Conclusion: This real-world analysis confirms the poor outcomes of R/R LBCL pts after 2L CAR T-cell therapy. BsAbs were the most frequently used and currently appear the most effective salvage option. These results highlight the urgent need for prospective trials and novel combination strategies in this high-risk population.