Time-to-Event Modeling of Peripheral Neuropathy – Platform Mega-Analysis of Eight vc-MMAE Antibody-Drug Conjugates (ADCs)
Matts Kågedal (1), Divya Samisen (1), Bill Gillespie (2), Dan Lu (1), Bernard Fine (1), Y J Choi (1), Chunze Li (1), Sandhya Girih (1), Jin Yan Jin (1)
(1) Genentech, Inc., South San Francisco, CA. (2) Metrum institute, Tariffville, CT
Objectives: valine-citrulline-MMAE (vc-MMAE) ADCs are the most commonly used linker-drug combinations in the ADC platform currently under clinical investigation. Peripheral neuropathy (PN) was one of the adverse effects frequently observed in clinical studies with vc-MMAE ADCs resulting in treatment discontinuation, thereby limiting the duration of treatment with these ADCs. The objective of the this analysis was to develop a pan-ADC PN model to describe the exposure-response (E-R) and the risk factors associated with PN in support of the clinical development of the vc-MMAE ADCs.
Methods: A time-to-event (TTE) parametric survival model to the onset of PN grade ≥ 2 events across multiple vc-MMAE ADCs was developed using NONMEM® 7 to describe the relationship between the conjugate analyte exposure and the incidence of PN. Data from phase I and II studies across the tested dose ranges of 0.1-3.2 mg/kg were included in the analysis. In this model, hazard is a function of the drug concentration in a hypothetical effect compartment, time and potential PN risk factors such as demographics (age, body size metrics, gender), prior chemotherapy, prior incidence of PN, diabetes, malignancy type, ADC type and albumin. The model performance was verified by comparison of observed PN event to events generated by simulation from the model by the use of visual predictive checks.
Results: The developed TTE model successfully predicted the time course of the PN incidence for the individual ADCs tested across multiple oncology indications. The model described the exposure-response across a range of tested doses across the individual ADCs. Covariate analysis identified body weight as a statistically significant risk factor for PN. A trend of higher risk of peripheral neuropathy was observed in patients with higher body weight even after accounting for exposure. The difference between the ADCs in terms of PN risk was minimal after accounting for differences in the exposure, treatment duration and body weight. No significant influence of additional pre-disposing risk factors on the incidence of PN was identified.
Conclusions: The cross-molecule mega-analysis demonstrates the utility of using a pan-ADC model-based approach to characterize the incidence of PN for patients dosed with vc-MMAE ADCs. Such platform model could be used to simulate different dosing scenarios to inform dose strategy for vc-MMAE ADCs.