(M-015) Hybrid MBMA-QSP Model Predicts T Cell Engager Plus Cereblon E3 Ligase Modulatory Drugs Combination Outcomes As A Function Of Prior Therapies In Relapsed/refractory Multiple Myeloma

Clifton Anderson – Simulations Plus, Research Triangle Park, NC 27709, USA; Kevin McCormick – Simulations Plus, Research Triangle Park, NC 27709, USA; Celeste Vallejo – Simulations Plus, Research Triangle Park, NC 27709, USA; William Duncan – Simulations Plus, Research Triangle Park, NC 27709, USA; Zeel Shah – Bristol Myers Squibb, Lawrenceville, NJ 08540, USA; Chuanpu Hu – Bristol Myers Squibb, Lawrenceville, NJ 08540, USA; Jian Zhou – Bristol Myers Squibb, Lawrenceville, NJ 08540, USA; Brian Schmidt – Madrigal Pharmaceuticals, West Conshohocken, PA 19428, USA; Bristol Myers Squibb, Lawrenceville, NJ 08540, USA; Alexander Ratushny – Bristol Myers Squibb, Lawrenceville, NJ 08540, USA; Anna Kondic – Bristol Myers Squibb, Lawrenceville, NJ 08540, USA

Objectives: This work hybridized quantitative systems pharmacology (QSP) with model-based meta-analysis (MBMA) to predict the outcomes of a novel drug combination (T cell engager (TCE) plus cereblon E3 ligase modulatory drugs (CELMoD®)) in relapsed/refractory multiple myeloma (RRMM) as a function of the number of prior therapies patients have received.

Methods: We developed a QSP model that describes RRMM response to therapy and subsequent disease progression. This model incorporates the mechanisms of action for most RRMM-tested therapies. Calibration data includes best and overall response (BOR/ORR) and progression-free survival (PFS). Mechanisms of neutropenia for both TCEs and CELMoDs are also implemented and were calibrated to trial data.

Clinical trials track the number of prior treatments (n_prior) for each patient; however, this attribute for patients in simulated populations (SimPops®) must be inferred. Within the model, a probabilistic classifier was used to assign each patient to an n_prior category based on the patient’s average response across all therapies administered in parallel.

The parameters of this classifier were calibrated using clinical trial data stratified by the median number of prior lines in the trial (1, 2-3, or ≥4 lines) with a subset of the clinical data held out for validation. Validation trials included daratumumab triplet therapies, TCE combination therapies, and CELMoD triplet therapies. All other mono and doublet therapies were used for fitting. For simplicity, all TCEs shared a common PK and PD parameterization, and as a result, TCE trial results were pooled into a single TCE arm.

For the following predictions, the model was used to simulate initial dosing with a TCE followed by combination with a CELMoD to support the NCT06163898 trial.

Results: As a result of calibration, virtual patients responding to a greater number of simulated therapies are associated with a lower number of prior therapies. The hybrid MBMA-QSP model accurately represents how ORR decreases with n_prior, with 76.4% of fitting and 74% of validation trial data for efficacy falling within the model’s 90% confidence intervals (n=165 trial arms across 37 drug combinations). TCE combinations with immunomodulatory drugs (IMiDs) show enhanced neutropenia, in line with data.

TCE and CELMoD combination predictions show enhanced neutropenia (52% Gr3+ TCE, 77% CELMoD, 93% TCE + CELMoD) and ORR (60% TCE, 37% CELMoD, 66% TCE + CELMoD) in patients with 4+ prior lines. Notably, TCE + CELMoD ORR was higher in patients assigned 1 (84%) and 2-3 (92%) prior lines.

Conclusions: This study highlights the utility of a QSP-MBMA model to predict the efficacy and safety of novel drug combinations, as well as investigate patient characteristics (n_prior) that were identified by MBMA as covariates of trial outcomes. This study hybridizes the QSP approach of virtual patients, which are inherently mechanistic, with an MBMA-identified patient characteristic that is known to affect trial outcomes but cannot be defined as a simple mechanism. The model was used to predict grade 3+ neutropenia and ORR efficacy profiles for novel TCE/CELMoD combinations across populations on different lines of therapy.

Citations: [1] Shah et al. 2024. 15th Annual Conference on Pharmacometrics (ACoP). Poster number M-099. Advancing Drug Development in Relapsed and Refractory Multiple Myeloma (RRMM): Assessing the Safety and Efficacy Landscape Utilizing Model-Based Meta-Analysis. Presented: Monday, November 11, 2024

Keywords: QSP, MBMA, Myeloma