Redefining post-CD19 CAR T-cell surveillance with ctdna: Real-world insights from post- axi-cel and liso-cel therapy

Abstract Introduction: Circulating tumor DNA (ctDNA)-based measurable residual disease (MRD) assessment is a promising noninvasive biomarker to guide treatment response and surveillance in Large B-cell lymphoma (LBCL). While PET-CT surveillance remains standard of care following anti-CD19 CAR T-cell therapy (CAR19), its limited positive predictive value, radiation exposure, and reduced accuracy due to post-treatment inflammation highlight the need for more specific and dynamic monitoring tools. ctDNA enables early relapse reduction, risk stratification, and real-time monitoring. Prior studies have shown its additive prognostic value to PET CT following third-line axi-cel (clonoSEQ, Frank et al., JCO 2021) and liso-cel (PhasED-Seq, Stepan et al., Blood 2023) administration. Here, we evaluate prospective ctDNA monitoring using clonoSEQ in patients receiving axi-cel or liso-cel in second line era, and explore ctDNA-guided surveillance to refine or reduce conventional use of PET-CT. Method: In this prospective, observational study of real-world patients receiving CD19 CAR T-cell therapy, pre-planned plasma collection at pre-lymphodepletion (PLD), days 14, 28, 90 and 180 post CAR19 resulted in 1105 samples and (VDJ) clonotype was identified from archival paraffin-embedded tissue to track MRD. PET-CT scans were performed before and at 1, 3, 6, and 12 mo post CAR19; responses assessed per Lugano criteria (Deauville 1-3 was considered PET-negative). Results: 142 patients were included (axi-cel, n=116; liso-cel, n=26). The median age was 66 years (23–83) for axi-cel and 79 years (55–88) for liso-cel; 61% vs 42% were male, and 77% vs 84% were stage III/IV. Both groups had a median of 1 prior line of therapy (range 1–4) with 61% vs 58% receiving treatment in second line, and 64% vs 80% had an R-IPI score of 3–5. Median follow-up was 17.2 months (range 0.9–36.5) for axi-cel and 11.5 months (2.2–28.1) for liso-cel. Best ORR and CR rates were 85% and 80% for axi-cel; 88% and 69% for liso-cel. Progression occurred in 45(38.7%) axi-cel patients and in 14(53.8%) liso-cel patients. For axi-cel, ctDNA levels were prognostic at all timepoints (PLD, Days 14, 28, and 90). High pre-treatment ctDNA burden (>1000 lymphoma genomes/mL plasma) was associated with inferior progression-free survival (PFS) compared to those <1000 LG/mL (median PFS: 1.87 vs 31.4 months). When excluding those who progressed at or prior to each timepoint, those with undetectable ctDNA was associated with significantly improved PFS at Day 14 (HR 4.2, 95% CI 2.2–7.9; p<0.001), Day 28 (HR 2.4, 1.1–5.8; p=0.009), and Day 90 (HR 2.5, 0.5–11.4; p=0.02). For liso-cel, undetectable ctDNA at Day 14 (HR 1.58, 95% CI 0.49–5.05; p=0.41) and Day 28 (HR 1.24, 0.24–6.39; p=0.77) was not significantly associated with PFS, with Day 90 being significant (HR 7.17, 0.46–111.2; p=0.05), consistent with prior PhasED-Seq data. These findings likely reflect differences in T-cell tumor killing kinetics between CAR19; we observed axi-cel more rapidly cleared ctDNA compared to liso-cel. This highlights the potential need for product-specific ctDNA surveillance strategies. For axi-cel patients with remission >12 months (n=54), ctDNA was undetectable in 41 of 52 patients (79%), 40 of 48 (83%), and 40 of 42 (95%) at Day 14, 28 and 90, respectively. By Day 180, ctDNA was undetectable all patients. Among the 59 relapses observed in this entire cohort (45 axi-cel, 14 liso-cel), 55 occurred within a year. Importantly, detectable ctDNA consistently emerged at or prior to clinical relapse, in 94% patients (n=52 of 55), regardless of CAR19 given. Three false-negative MRD cases occurred, one with a single 0.8 cm cervical lesion, CSF only relapse, and single 0.6 cm skin lesion in leg-type DLBLCL; the latter two cases were also a false negative by PET-CT. Overall, the negative predictive value (NPV) of undetectable ctDNA for concurrent PET-detectable disease, was 100% at Day 28 and 95.6% at Day 90. Conclusion: ctDNA-based MRD monitoring allows for early prognostication following CAR19, particularly by day 14 in axi-cel-treated patients. While its prognostic value in liso-cel emerges later at day 90, ctDNA detectability preceded or coincided with clinical relapse across products. These findings support the integration of ctDNA into post–CAR T surveillance with potential for reducing routine PET-CT imaging, particularly in undetectable MRD patients or to guide additional workup in patients with inconclusive results.