Circulating Tumor DNA (ctDNA) As an Early Outcome Predictor in Patients (pts) with Second-Line (2L) Large B-Cell Lymphoma (LBCL) after Lisocabtagene Maraleucel (liso-cel) Versus Standard of Care (SOC) Treatment (tx) from the Phase 3, Randomized Transform Study

Background: ctDNA clearance after frontline DLBCL tx has shown strong prognostic value for favorable outcomes, with the potential to improve MRD-driven therapeutic strategies and establish MRD as a surrogate endpoint in future studies (Roschewski M, et al. Blood 2022; Goldstein J, et al. Blood 2023). However, further study is needed to evaluate the association of this noninvasive approach with efficacy outcomes after 2L LBCL tx. We previously reported ctDNA analysis after liso-cel infusion in TRANSFORM (NCT03575351; Stepan L, et al. Blood 2023;142[suppl 1]). Here, we describe the predictive value of pre-tx and on-tx ctDNA levels for durable clinical benefit (CR and event-free survival [EFS]) in both liso-cel and SOC arms from TRANSFORM (Kamdar M, et al. J Clin Oncol 2024), evaluating ctDNA as an earlier surrogate for conventional clinical outcomes. Methods: Baseline (prerandomization) and longitudinal ctDNA levels were assessed in randomized pts treated with liso-cel or SOC in TRANSFORM using the phased variant enrichment and detection sequencing assay (PhasED-Seq; Foresight Diagnostics). Phased variants were identified from baseline plasma samples for longitudinal ctDNA monitoring at multiple on-tx time points, as previously described (Stepan L, et al. Blood 2023;142[suppl 1]). Randomization was defined as study Day 1. Association of ctDNA levels with response per independent review committee using Lugano 2014 criteria and EFS was investigated at various predefined time points, including Day 43 (after 2 cycles of salvage immunochemotherapy for SOC; Day 15 after liso-cel infusion), Day 64 (after 3 cycles of salvage immunochemotherapy for SOC; 1 month after liso-cel infusion), and Day 126 (within 2 months after ASCT; 3 months after liso-cel infusion). To assess the surrogacy value of ctDNA, ctDNA-MRD dynamics and ctDNA clearance were investigated within both arms. Results: In pooled analyses combining the liso-cel and SOC arms, higher baseline ctDNA levels (above the median) were associated with shorter EFS (P = 0.05). Pts achieving Day 126 CR had substantial reductions in ctDNA burden vs baseline, while pts with PD or stable disease had persistently high ctDNA levels vs baseline. Achieving undetectable ctDNA at different time points was associated with significantly longer EFS in pooled analyses, observed as early as study visit Day 43, with the strongest association at Day 126. HR (95% CI) for inferior EFS was 3.0 (1.6-5.7) at Day 43, 3.8 (2.1-7.0) at Day 64, and 4.2 (2.3-7.7) at Day 126 in pts with detectable vs undetectable ctDNA. In the analysis by tx arm, consistent with the pooled analyses, ctDNA clearance was associated with EFS benefit at all measured time points with either liso-cel or SOC. Significantly more pts achieved undetectable ctDNA with liso-cel vs SOC during the study (39/63 [62%] vs 25/65 [38%], respectively; P = 0.013), consistent with superior EFS (primary endpoint) with liso-cel in TRANSFORM. All pts in the SOC arm with evaluable ctDNA at Day 126 (n = 28) had received high-dose chemotherapy (HDCT)/ASCT. However, pts with undetectable ctDNA in the liso-cel vs SOC arm had longer EFS at all predefined time points and had statistically longer EFS at Day 126 (SOC vs liso-cel: HR, 3.9 [95% CI, 1.4-10.6]), indicating a deeper and more durable response with liso-cel vs SOC. In pts with CR, ctDNA showed significant predictive value beyond the response assessment at all time points in the liso-cel arm with a HR (95% CI) of 3.8 (1.3-11.0) at Day 64 and 7.0 (2.0-24.3) at Day 126 for EFS in pts with detectable ctDNA vs undetectable ctDNA. Among pts in the SOC arm who experienced subsequent PD despite achieving CR (all received HDCT/ASCT) and undetectable ctDNA at Day 126, serial ctDNA assessment showed consistent reversion to detectable ctDNA at later time points. Conclusions: ctDNA data from TRANSFORM confirm the value of ctDNA as a biomarker for disease burden monitoring and early prediction of durable clinical benefit after 2L LBCL treatment. Furthermore, longitudinal ctDNA molecular response results agree with previously reported clinical efficacy data and confirm the superiority of liso-cel with deeper responses over the historical SOC in 2L LBCL. ctDNA may have a key role as a surrogate endpoint in future LBCL studies, including after CAR T cell therapies.