(T-137) Mechanistic physiologically-based pharmacokinetic platform model to characterize risk of cytochrome P450 based drug-drug interactions for bispecific T cell engagers in oncology patients

Vijay Upreti, NA – Executive Director, CPMS, Amgen

Objectives: Bispecific T cell engagers (Bi-TCEs) have revolutionized the landscape of cancer treatment across solid and liquid tumors. Based on mechanism of action, cytokine elevations after initial doses of Bi-TCEs in oncology patients can result in potential cytochrome (CYP) 450 based drug-drug interaction (DDI), albeit the cytokine elevation is transient in nature. Hence the risk could be theoretical but yet is unknown. Over the decades, we have gained extensive clinical development experience of Bi-TCEs with 3 generations of molecules in development and the first-in-class approval. This study aims to establish a mechanistic physiologically-based pharmacokinetic (M-PBPK) platform to holistically assess the DDI liability of cytokine elevation, including not only Interleukin (IL)-6 but also interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α), and the clinical relevance for 3 generations of Bi-TCEs across liquid and solid tumors in oncology patients.

Methods: The M-PBPK model was designed to evaluate the concurrent suppression of CYP 450 enzymes by IL-6, IFN-γ, and TNF-α. The cytokine profiles were modeled assuming 0-order input and 1st-order elimination rates to describe the observed data. Parameters for CYP enzyme suppression, such as minimum enzyme activity (Emin) and half-maximal suppression concentration (EC50), were collected from literature for CYP 3A4 and CYP 2D6. Given the limited data and high variability, a conservative approach was used to assess the worst-case scenario by selecting the lowest reported Emin and EC50 values for each cytokine-CYP interaction pathway. After validation, the model was used to predict the DDI liabilities for a wide range of Bi-TCEs.

Results: The model predictability was validated using clinical DDI data from chronic and acute inflammation-induced IL-6, IFN-γ and TNF-α elevation and in vitro cytokine cocktail study for CYP 3A4 and CYP 2D6. Bi-TCEs-induced cytokine elevation generally peaked at 6 – 48 h after dosing. The maximum suppression generally occurred at ~24 h post peak cytokine. Maximum suppression for CYP 3A4 and CYP 2D6 predicted to be less than 35 % and 15%, respectively. For the most sensitive prob substrates midazolam and dextromethorphan co-administered with Bi-TCEs at the dosing days, predicted area under the curve (AUC) ratios are 1.0 – 1.2 for midazolam and 1.0 – 1.1 for dextromethorphan, indicating no DDI liability. To capture the maximum DDI potential, midazolam or dextromethorphan were also administered at the time of maximum CYP suppression. The predicted AUC ratios were < 2 for CYP 3A4 indicating weak DDI, and < 1.25 for CYP 2D6 indicating no DDI liability.

Conclusions: Unlike disease-drug-drug interactions observed in chronic inflammation conditions such as rheumatoid arthritis, the cytokine increases evaluated for 3 generations of Bi-TCEs in oncology patients are transient and not likely to translate to clinically meaningful DDIs for CYP 3A4 and CYP 2D6 substrates.

Citations: NA