Surrogate endpoints as prognostic factors for long-term outcomes among patients receiving axicabtagene ciloleucel in frontline high-risk large B cell lymphoma

Abstract Background. ZUMA-12 (NCT03761056), a multicenter phase 2 study, evaluated axicabtagene ciloleucel (axi-cel), an autologous anti-CD19 chimeric antigen receptor T-cell therapy, as part of first-line treatment in patients with high-risk large B-cell lymphoma (LBCL). Axi-cel showed an 86% complete response (CR) rate and 3-year PFS rate of 75% in efficacy-evaluable patients, demonstrating high rates of durable response with long-term follow-up (Chavez et al., Blood 2025). Here, we assessed the prognostic role of surrogate endpoints on longer term outcomes from ZUMA-12 in 40 treated patients after ≥3 years of follow-up. Methods. Patients had high-risk LBCL, with MYC and BCL2 and/or BCL6 rearrangements (double- or triple-hit histology) and/or International Prognostic Index (IPI) score ≥3 plus a positive interim PET per Lugano classification (Deauville score 4/5) after 2 cycles of anti-CD20 monoclonal antibody plus anthracycline-containing regimen. Response was assessed at Month 1, every 3 months from Month 3 to Month 24, and then annually from Month 36 on. Long term outcomes included event-free survival (EFS), progression-free survival (PFS), overall survival (OS), and duration of response (DOR). CR at 6 months (6M) was defined as investigator-assessed CR at or prior to 6M post-infusion and maintained through the 6M evaluation. Kaplan-Meier (KM) and multiple Cox regression methods were used to summarize and calculate unadjusted and adjusted effects of CR at 6M on long term outcomes. Baseline characteristics for adjustments were selected based on significant or clinically meaningful differences between the CR at 6M groups and complete availability of data. These include age group (<65, ≥65 years), sex, IPI score, ECOG performance status, double/triple hit status, double expression status, C-MYC expression status, cell of origin and best response to two cycles of standard-of-care immunochemotherapy. Pearson correlations between CR at 6M and EFS, PFS, OS, and DOR at 6, 12, 18, 24, and 36 months (all binary variables) were calculated. Contemporaneous correlations between EFS, PFS, and OS at these timepoints were also assessed. We assumed all censored observations were non-events, and sensitivity analyses were repeated under the opposite assumption of events. Results. The Investigator-assessed CR rate among response-evaluable patients (n=37) increased after the primary analysis from 78% to 86% (95% CI, 71-95%). The reverse KM median follow-up for OS was 38.1 months (95% CI 36.5-42.0). There were no significant demographic or clinical differences in baseline characteristics between the CR and the no CR at 6M groups. At 36 months, KM estimates for the CR vs non-CR at 6M groups were 88.9% vs 30.0% for EFS, 88.9% vs 33.8% for PFS, 92.6% vs 50.0% for OS, and 88.9% vs 50.0% for DOR (all p<0.05). Patients achieving CR at 6M had better long-term outcomes as indicated by lower Cox regression-adjusted hazard ratios (aHR) for long term endpoints compared to those not achieving CR: EFS aHR 0.07 (95% CI, 0.02-0.32; p=0.001), PFS 0.07 (0.01-0.32; p=0.001), OS 0.11 (0.02-0.61; p=0.012), and DOR 0.17 (0.03-1.10; p=0.063). From 18 to 36 months post-infusion, Pearson correlations with CR status at 6M ranged from 0.59 to 0.51 for EFS and PFS, respectively. Correlations ranges were 0.47, 0.56, and 0.48 for OS for 18, 24, and 36 months, respectively. All correlations were significant, except for the 6- and 12-months correlations with OS, as events were rare over this timeframe. Strong contemporaneous correlations of 0.90-0.93 were observed between EFS and PFS at 6, 12, 18, 24, and 36 months, indicating substantial overlap in these outcomes. Correlations between EFS with OS and PFS with OS at 18, 24, and 36 months timepoints ranged from 0.65-0.80 and 0.70-0.85, respectively. EFS and PFS correlations with OS were lower (0.38-0.57) at 6 and 12 months. All correlations were significant (p<0.05). Results were similar in sensitivity analyses conducted with all censored observations as events. Conclusion. CR at 6 months was significantly associated with long-term EFS, PFS and OS and has potential to serve as a surrogate endpoint for long-term outcomes in future clinical trials evaluating axi-cel as part of first-line high-risk LBCL treatment. Correlations between EFS and PFS at 18 to 36 months were strong, and moderate to strong between EFS and OS and between PFS and OS, confirming the strong association between these long-term outcomes.