Prospective validation of end of treatment ctDNA-MRD by PhasED-Seq in DLBCL patients from a national trial.

7000 Background: The prognostic utility of circulating tumor DNA measurable residual disease (ctDNA-MRD) detection at end of treatment (EOT) using phased variant (PV) enrichment and detection sequencing (PhasED-Seq) has been demonstrated in patients with diffuse large B-cell lymphoma (DLBCL) receiving first-line (1L) therapy. Prior studies are limited by treatment, patient, and sample heterogeneity. Here, we independently validate the prognostic value of PhasED-Seq in a national, multi-center study of uniformly treated 1L DLBCL patients. Methods: ctDNA-MRD was assessed using Foresight CLARITY in LBCL patients enrolled on HOVON-902 from >50 centers in the Netherlands and Belgium. Patients were treated with curative-intent 1L therapy (R-CHOP or DA-EPOCH-R). We evaluated the prognostic significance of MRD status [positive (+), negative (-)] on progression-free survival (PFS) and overall survival (OS). PVs were identified from pretreatment biopsies or plasma with matched normal DNA. EOT plasma samples were used for ctDNA-MRD detection. Results: A total of 150 of 156 (96%) eligible patients had successful PV identification. Of included patients, 90%, 9%, and 1% had DLBCL, HGBL, and PBMCL, respectively. IPI distribution was 22% low, 29% low-intermediate, 27% high-intermediate, and 22% high risk; median age was 67.5. The 24-month PFS and OS in this cohort were 74% and 86%, respectively, with 31 months of median follow-up. At the EOT, 76% of patients were MRD- and 24% were MRD+. MRD+ status significantly predicted inferior PFS (2 yr PFS 88 vs 28%; HR 9.7, 95% CI 4.2-22.3, p<0.0001) and OS (2 yr OS 97 vs 50%; HR 10.6, 95% CI 4.1-27.7, p<0.0001). Moreover, in patients without complete response, MRD+ was significantly prognostic for PFS, suggesting an ability to adjudicate imaging results (HR for PFS 7.6, 95% CI 3.6-16.3, p < 0.0001). Among patients who were MRD- and achieved CMR at EOT, 2-year PFS and OS were 91% and 99%, respectively. All patients who failed to achieve CMR and remained MRD+ experienced relapse. ctDNA-MRD was prognostic for outcomes in all subgroups considered, including source of baseline sample (tumor versus plasma), best clinical response, IPI, sex, lactate dehydrogenase, stage, or extranodal disease. In multivariate analysis including ctDNA-MRD, IPI, and best overall response, ctDNA-MRD was significantly and independently prognostic for both PFS [HR for ctDNA: 7.1, 95% CI 3.5-14.3, p<0.0001] and OS [HR for ctDNA: 5.1, 95% CI 2.2-11.9, p=0.00018]. Conclusions: We validated the prognostic value of PhasED-Seq-based ctDNA-MRD in a real-world multicenter 1L DLBCL cohort. This highlights the utility of ctDNA-MRD to confirm residual disease in patients without complete response by imaging, as well as the potential to identify patients who may benefit from consolidation therapy. These results support the integration of MRD as a standard component of response evaluation in 1L DLBCL treatment.