Circulating Tumor DNA Analysis Associates with Progression-Free Survival (PFS) with Odronextamab Monotherapy in Relapsed/Refractory (R/R) Follicular Lymphoma (FL) and Diffuse Large B-Cell Lymphoma (DLBCL): Identification of Minimal Residual Disease Status and High-Risk Subgroups from the Phase 2 ELM-2 Study

Abstract

Background Molecular characterization of B-cell non-Hodgkin lymphoma (B-NHL), through circulating tumor DNA (ctDNA) assessment of minimal residual disease (MRD) has been proposed as a tool to predict clinical outcome. Odronextamab, a CD20×CD3 bispecific antibody, demonstrated deep and durable responses and a generally manageable safety profile in patients (pts) with R/R FL or DLBCL in the Phase 2 ELM-2 study (NCT03888105; Kim TM et al. and Kim WS et al. ASH. 2022). In the overall populations, 12-month PFS rates were 64% and 29%, respectively. Using ctDNA from ELM-2, we show that these assessments associate with clinical outcomes following odronextamab treatment. Methods In ELM-2, pts received IV odronextamab in 21-day cycles, with step-up dosing in Cycle (C) 1, 80 mg (FL)/160 mg (DLBCL) QW in C2-4, then 160 mg (FL)/320 mg (DLBCL) Q2W until disease progression or unacceptable toxicity. Baseline (BL) ctDNA and tumor biopsies were used for molecular profiling. BL and on-treatment ctDNA were used for MRD determination in the biomarker population (BP; pts required ≥1 available plasma biomarker sample to be included in the BP). The first post-BL ctDNA sample was collected on C5 Day (D) 1, at the time of positron emission tomography-computed tomography (PET-CT). A modified AVENIO ctDNA analysis workflow (Roche; research only) was used for next-generation sequencing, based on the cancer personalized profiling by deep sequencing technique (Kurtz et al. J Clin Oncol. 2018). Whole blood cell pellets were used to filter out germline allele variants. MRD negativity was reported when the P value for variant allele frequency was >0.005. Results The BP comprised 53 FL pts and 63 DLBCL pts; at BL, all FL pts and all but one DLBCL pt were MRD(+). Pts remaining on study until C5D1 had similar PFS regardless of whether they were in the BP or overall population (FL, n=111; DLBCL, n=83). Pts who were MRD(-) at C5D1 had significantly longer PFS vs. those who remained MRD(+) (FL: HR 0.27 [95% CI 0.09-0.84], P=0.024 [Fig. 1a]; DLBCL: HR 0.29 [95% CI 0.12-0.71], P=0.007). In pts with FL achieving complete response (CR) by PET-CT at C5D1, a trend for prolonged PFS was observed in those who were MRD(-) (n=26) at this timepoint vs. those who were MRD(+) (n=17; HR 0.29 [95% CI 0.07-1.1], P=0.072). Pts with DLBCL who were MRD(-) at C5D1 (n=20) had similar PFS benefit regardless of PET-CT CR status (HR 0.56 [95% CI 0.10-3.11], P=0.511). Four DLBCL pts who were MRD(-) and did not achieve CR by PET-CT at C5D1 (partial response, n=3; stable disease, n=1) went on to achieve PET-CT CR at a later timepoint. Mutational analyses of BL ctDNA identified TP53 as the most frequent mutation (FL, n=34/53 [64%] mutation-evaluable pts; DLBCL, n=44/58 [76%]). In FL pts, TP53 mutations were associated with MRD(+) status at C5D1 (Fisher's exact test, P=0.002) and predicted significantly shorter PFS (HR 3.57 [95% CI 1.01-12.69]; P=0.049); this association was not observed in DLBCL. LymphGen classification (Wright et al. Cancer Cell. 2020) of BL ctDNA in pts with DLBCL identified mostly MCD and EZB subtypes (MCD, n=15; EZB, n=14; ST2, n=1; BN2, n=1; other, n=26). EZB-subtype pts treated with odronextamab had significantly longer PFS compared with MCD-subtype pts (HR 0.23 [95% CI 0.07-0.83]; P=0.025 [Fig. 1b]). Using BL ctDNA, cell of origin was determined in 43/58 pts with DLBCL; odronextamab treatment led to similar PFS in higher-risk non-germinal-center B-cell-like (non-GCB, n=18 [42%]) pts vs. GCB pts (n=25 [58%]; HR 1.62 [95% CI 0.72-3.62]; P=0.244). Further molecular assessment focused on gene fusions in DLBCL pts in the BP (39 with local laboratory gene fusion and ctDNA analyses, 24 with only ctDNA analyses). Odronextamab led to similar PFS in pts with double-hit (n=15) or triple-hit (n=5) fusions compared to those without (n=43; P=0.343 and P=0.140, respectively). Conclusions This study is among the first to analyze ctDNA in pts with R/R FL and DLBCL in a pivotal trial. This non-invasive method allows molecular characterization of pts with no available tissue, enabling identification of high-risk subgroups. CtDNA MRD status at C5D1 of odronextamab treatment was highly associated with PFS in pts with FL and DLBCL and in the future could form the basis of response-directed treatment paradigms. Molecular characterization in tumor biopsies, including CD20, will be presented. Ongoing monitoring of ctDNA may serve as an important early progression marker in R/R FL and DLBCL.