Marçal Bravo i Padrós 1, Christine Veyrat-Follet 1, Samit Ganguly 1, Min Zhu 1, Lutz  Harnisch 1
1 Regeneron Pharmaceuticals, Inc. (Tarrytown,, United States)
Introduction:
In first‑line (1L) DLBCL, rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone/prednisolone (R‑CHOP) remains a common treatment. However, up to 40% of patients experience relapsed/refractory (R/R) disease [1]. There is an unmet need to improve and deepen 1L responses, especially in patients with high‑risk features. Odronextamab, a CD20×CD3 bispecific antibody, demonstrated activity and generally manageable safety as monotherapy in heavily pretreated B-cell non‑Hodgkin lymphoma (B‑NHL), including R/R DLBCL, in the Phase 1 (ELM-1, NCT02290951) and Phase 2 (ELM-2, NCT03888105) studies [2–5]. OLYMPIA‑3 (NCT06091865) is a Phase 3, randomized study of odronextamab‑CHOP (O-CHOP) versus R‑CHOP in patients with previously untreated DLBCL and intermediate‑/high‑risk features. The study comprises Part 1A (dose escalation), Part 1B (dose optimization), and Part 2 (O‑CHOP vs R‑CHOP).
Objectives:
Evaluate treatment effect and clinical response, using joint modeling of tumor size (TS) and PFS outcomes, to support selection of the recommended Phase 3 dosing regimen for odronextamab plus CHOP in OLYMPIA-3 Part 2.
Methods:
Data were obtained from three clinical studies of odronextamab in patients with DLBCL [2–4]. Odronextamab was tested at full doses of 160 mg QW/320 mg Q2W (ELM-1/ELM-2) and 80 QW/160 Q2W (OLYMPIA-3 Part 1A); different dosing frequencies (160 QW/160 Q3W or 160 QW/320 Q2W) were assessed in OLYMPIA-3 Part 1B. The previously developed population-pharmacokinetics (PK) model of monotherapy in R/R B‑NHL (n=507) [6] was updated to further characterize odronextamab PK with CHOP. Exposure to treatments over time was introduced in the tumor growth inhibition (TGI) model using concentrations predicted by the odronextamab population-PK model and the kinetic-pharmacodynamic model for CHOP. Once the best structural TGI model was selected, the relationship between baseline covariates and model parameters was evaluated using the automatic covariate selection using stepwise covariate modeling method, with improvement on corrected Bayesian information as inclusion criteria. Only significant covariates with Wald-test p<0.05 remained in the final model.  In the second step, PFS data were modeled using a parametric proportional hazard model and baseline covariates were selected using the same approach as for the longitudinal model.  Finally, after building the longitudinal and survival models separately, several joint models were developed to find the link function that best captured the association between TS kinetics and PFS. Significant covariates found in the previous steps were verified if they remained significant in the joint model. A joint model of TS dynamics and PFS was then developed for a subset of R/R and 1L DLBCL patients with available tumor assessments (Lugano criteria) [7]. Results: A non-linear joint model of TS and PFS was developed using 272 patients, including 56 with previously untreated DLBCL from OLYMPIA-3. The joint model accounted for the dynamics of sensitive and resistant tumor cells, an exposure‑driven effect on tumor shrinkage on sensitive and resistant cells, and a signal delay to account for prolonged tumor regression observed after end-of-treatment. The effects of odronextamab on TS shrinkage rate were captured by an Emax function. CHOP had an additive effect on tumor shrinkage when combined with odronextamab. Disease progression by change from nadir TS was the best on‑treatment predictor for PFS. Albumin, lactate dehydrogenase, and bulkiness at baseline were significant covariates for TS at baseline. Model predictions suggested the odronextamab concentration that achieved 90% of tumor cell killing (EC90) was 37.1 mg/L. EC90 concentration was reached at study Week 7 (third full dose) for both the 160 QW/160 Q3W +CHOP and 160 QW/320 Q2W +CHOP treatment groups, and the predicted changes from baseline in TS at Week 9 (mid‑induction) and Week 18 (end-of-induction) were similar in both groups. Conclusions: Model predicted simulations of TS after patients received their actual odronextamab treatment at 160 QW/160 Q3W +CHOP or 160 QW/320 Q2W +CHOP indicated similar changes in TS from baseline at Week 9 (mid-induction) and Week 18 (end-of-induction) between the two treatment groups. Similarities in the combination therapy set-up are explained through the additive effect of CHOP, as the Emax relationship of odronextamab and tumor cell killing clearly benefits R/R patients that received 160 QW/320 Q2W monotherapy. Based on the totality of data, and leveraging the benefit/risk profile, odronextamab 160 mg QW/160 mg Q3W in combination with CHOP was selected as the recommended Phase 3 dose for OLYMPIA-3 Part 2. References: 1. Vodicka P, Klener P, Trneny M. Diffuse large B-cell lymphoma (DLBCL): early patient management and emerging treatment options. Onco Targets Ther 2022;15:1481–501. 2. Bannerji R, Arnason JE, Advani RH, Brown JR, Allan JN, Ansell SM, et al. Odronextamab, a human CD20×CD3 bispecific antibody in patients with CD20-positive B-cell malignancies (ELM-1): results from the relapsed or refractory non-Hodgkin lymphoma cohort in a single-arm, multicentre, Phase 1 trial. Lancet Haematol 2022;9(5):e327–e39. 3. Kim WS, Kim TM, Cho S-G, Jarque I, Iskierka-Jażdżewska E, Poon LM, et al. Odronextamab monotherapy in patients with relapsed/refractory diffuse large B cell lymphoma: primary efficacy and safety analysis in Phase 2 ELM-2 trial. Nat Cancer 2025;6(3):528–39. 4. Matasar M, Turgut B, Tessoulin B, Guidez S, Altuntas F, Iqbal N, et al. Phase 3 trial evaluating efficacy and safety of odronextamab plus CHOP vs rituximab plus CHOP in previously untreated diffuse large B-cell lymphoma (DLBCL; OLYMPIA-3). J Clin Oncol 2024;42(suppl 16):TPS7086. 5. Topp MS, Matasar M, Allan JN, Ansell SM, Barnes JA, Arnason JE, et al. Odronextamab monotherapy in R/R DLBCL after progression with CAR T-cell therapy: primary analysis of the ELM-1 study. Blood 2025;145(14):1498–509. 6. Bravo Padrós M, Conrado DJ, Srinivasan K, Harnisch LO, Davis JD, Zhu M. Pharmacokinetics of odronextamab, a bispecific T-cell-engaging antibody, in adult patients with relapsed or refractory B-cell non-Hodgkin lymphoma. CPT Pharmacometrics Syst Pharmacol. 2025,15(1):e70162. 7. Cheson BD, Fisher RI, Barrington SF, Cavalli F, Schwartz LH, Zucca E, et al. Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol 2014;32(27):3059–68.
Reference: PAGE 34 (2026) Abstr 12175 [www.page-meeting.org/?abstract=12175]
Poster: Drug/Disease Modelling - Oncology