(T-011) A multi-stage modeling approach towards predicting cytokine release syndrome (CRS) in patients receiving Mosunetuzumab

Rik Schoemaker, Ph.D. – Senior Consultant, Occams; Zao Li, M.Sc. – Senior Scientist, Genentech, Inc.; Feifei Li, M.Sc. – Scientist, Genentech, Inc.; Shweta Vadhavkar, M.Sc. – Senior Principal Modeling And Simulation Analyst, Genentech, Inc.; James Lu, Ph.D. – Distinguished Scientist, Genentech, Inc.; Monica Susilo, Ph.D. – Senior Principal Scientist, Genentech, Inc.; David Turner, Ph.D. – Distinguished Scientist, Genentech, Inc.; Chi-Chung Li, Ph.D. – Distinguished Scientist, Genentech, Inc.; Pascal Chanu, Pharm.D. – Distinguished Scientist, Genentech, Inc.

Objectives: T-cell directed bispecific antibodies (TDBs) are a promising new class of antibodies for cancer treatment. Mosunetuzumab (Mosun), a TDB recently approved for follicular lymphoma, targets CD3 expressing T cells and CD20 expressing malignant B cells, thereby triggering T-cell-mediated killing of tumor cells. TDBs may also trigger cytokine spikes (e.g., IL6), leading to cytokine release syndrome (CRS). To mitigate CRS for TDBs, strategies such as step-up dosing and tocilizumab (Toci) treatment have been employed. Toci binds soluble and membrane bound IL6 receptors (sIL6R and mIL6R), preventing IL-6 mediated inflammation and subsequent CRS. Using Mosun clinical data, we present a multi-stage modeling approach towards predicting IL6 and CRS response to Mosun treatment.

Methods: Stage 1: A population pharmacokinetic (PK) model was developed from 439 patients receiving intravenous Mosun, with doses from 0.05 – 60 mg. Equilibrium binding dynamics were integrated into the PK model to predict the Mosun CD20 receptor occupancy (RO%) time course [1].
Stage 2: A population PK‒pharmacodynamic (PKPD) model was developed to describe IL6 response. For patients receiving Toci, a target-mediated drug disposition (TMDD) modeling approach was added to integrate Toci PK, total IL6, and total sIL6R levels. Mosun concentration-time and RO-time profiles, from the empirical Bayes estimates in Stage 1, were assessed as PK drivers for IL6 response.
Stage 3: Exposure Response (ER) models were developed to predict the probability of Grade ≥2 CRS. PK metrics from Stage 2 (e.g., maximum Mosun concentration (Cmax), maximum Mosun CD20 RO% (ROmax), and maximum IL6 (IL6max)), were assessed as exposure predictors.
Stage 4: Population PKPD model simulations (from Stage 2; N=5000) were conducted to predict the Mosun ROmax and IL6max for the approved Mosun dose of 1/2/60/30 mg. Grade ≥2 CRS rates for the Cycle 1 dose windows (i.e., 1 mg, 2 mg, and 60 mg) were predicted using the ER models (from Stage 3; N=1000 bootstraps).

Results: The final PKPD model well-described the IL6 response, and consisted of an IL6 precursor pool, IL6 circulation pool, and placebo effect compartments. Mosun CD20 RO% stimulated the release of IL6 from the pool into circulation via an Emax model. The TMDD components of the model well-described the Toci PK, sIL6R, and IL6 data. Linear-logistic ER models, using ROmax as the exposure metric, predicted the Grade ≥2 CRS rates of 3.5%, 1.6%, and 15% for the three Cycle 1 dose windows at the 1/2/60/30 mg dose regimen (N=218).

Conclusions: The multi-stage modeling and simulation paradigm predicted the observed Mosun CD20 RO%, IL6 response, and Grade ≥2 CRS rates. This paradigm can support Mosun development with regard to CRS mitigation strategies. An external model evaluation with the clinical Mosun subcutaneous (SC) dose of 5/45/45 mg q3w is pending.

Citations: [1] Bender B., et al. Population Pharmacokinetics and CD20 Binding Dynamics for Mosunetuzumab in Relapsed/Refractory B-Cell Non-Hodgkin Lymphoma (R/R NHL). Blood (2023) 142 (Supplement 1): 2810. https://doi.org/10.1182/blood-2023-182086