Prognostic value of circulating tumor DNA (ctDNA) detection by PhasED-Seq after axicabtagene ciloleucel (axi-cel) therapy in relapsed/refractory large B-cell lymphoma (LBCL).

e15043 Background: In ZUMA-7, a phase 3 randomized trial to evaluate axi-cel treatment in LBCL (NCT03391466), most patients responded well to axi-cel, but many eventually experienced disease progression (Locke et al, NEJM 2022). Monitoring of ctDNA levels in blood using the ultrasensitive phased variant enrichment and detection sequencing (PhasED-Seq) technology, has previously been used to assess measurable residual disease (MRD) in LBCL and has demonstrated prognostic value in the context of chemoimmunotherapy (Sworder et al, ASH 2023). Here, we explored the value of ctDNA detection by PhasED-Seq to predict outcomes of patients receiving axi-cel in r/r LBCL. Methods: Patients treated with either axi-cel or standard of care (SOC) with evaluable pre-infusion tumor and post-infusion peripheral tracking plasma ctDNA (n = 44) were analyzed by the Foresight CLARITY assay, powered by PhasED-Seq. Tumor informed variants were identified in biopsy tissue obtained prior to treatment and longitudinally monitored in plasma at approximately 50, 100, and 150 days post randomization. On average, 0.8 mL (range 0.2-2.1mL) of plasma per timepoint was available for phased variant (PV) tracking. Positive predictive value (PPV; MRD+ pts who relapsed or were nonresponders/total MRD+ pts ×100) and negative predictive value (NPV; MRD- pts in ongoing responders/total MRD- pts ×100) were assessed at days 50 and 150. Results: In a pooled analysis of axi-cel and SOC treatments, patients who are MRD- at day 50 and day 150 demonstrate a non-statistically significant increased duration of PFS in comparison to MRD+ patients. When considering best overall response, achieving MRD negativity at any time was significantly correlated with improved PFS (p < 0.05). Patients who were MRD- by their last evaluable timepoint achieved significantly improved PFS (p < 0.05). When considering a landmark, the NPV at day 50 was 70% and improved by day 150 to 77%. At day 50, closest to initial tumor clearance, PPV was 50% and increased modestly, reaching 57% by day 150. Overall, we observed a CR concordance rate with MRD of 73% at day 50 and 78% at day 150. Conclusions: Although a limited number of subjects and plasma volumes were available, we found that the prognostic value of MRD was limited at day 50 and improved at day 150. In contrast, both overall MRD negativity and MRD negativity at the last evaluable assessment were more strongly prognostic for PFS in r/r LBCL patients receiving axi-cel. Both the positive and negative predictive values of MRD using PhasED-seq showed a high level of accuracy by day 150. This analysis highlights the prognostic value of MRD detection by ultrasensitive measures in early lines of cellular therapy. Further exploration is warranted of this assay’s utility in monitoring response and use as a surrogate endpoint in LBCL.