Abstract 3532: Preclinical-to-clinical translation of optimal schedule for MLN0128, an investigational novel small-molecule selective mTOR1/2 inhibitor, based on exposure-efficacy modeling

MLN0128 is an investigational, orally bioavailable, potent and highly selective, ATP-competitive inhibitor of TORC1/2. A phase 1 study was initiated to evaluate the safety and maximum tolerated dose (MTD) of single-agent MLN0128 at different dose levels and schedules (QW, QD, 3-on/4-off and 5-on/2-off) in patients with advanced nonhematologic tumors. As multiple MTDs were achieved, we used exposure-efficacy modeling to interpret the achieved MTDs in the context of preclinical antitumor efficacy, thus supporting the selection of dosing schedule in the clinic. To this end, we conducted xenograft efficacy studies using the 786-0 renal cell carcinoma xenograft model (PTENmut,VHLdel) in athymic nude mice. MLN0128 was orally administered at a range of different doses under several different schedules (QD, bid QW, 3-on/4-off and 5-on/2-off). Single and repeat-dose PK data was also obtained for a range of different time-points and a linear two-compartment PK model was built. This PK model was used to simulate pharmacokinetic profiles corresponding to the arms of the efficacy studies, for the study duration (504 hours, or 21 days). The average MLN0128 concentrations (Cavg, 0-504) for each arm of the efficacy study were then compared against the tumor growth rate, and found to be broadly proportional, with the exception of the most infrequent (QW) dose schedule, for which efficacy was sub-exposure proportional. A dynamic tumor growth model was found to provide an accurate description of antitumor effect for all schedules, and to predict the schedule effect from the most infrequent schedule. A mathematical analysis demonstrated that the schedule effect occurs due to the saturability of the underlying PK/kill curve, with high infrequent doses resulting in large portions of the PK profile at concentrations where the marginal benefit of additional drug levels approaches zero. A sensitivity analysis conducted on this fitted model found that it was insensitive to tumor growth rate and tumor sensitivity to drug, but not to half-life. This dynamic model will be extended in a translational context, to account for the differences in half-life between mice and humans, as well as to account for the uncertainty in parameter estimation and the human PK variability. Our work thus provides a fundamental basis for the understanding of the PK/efficacy relationship for mTOR inhibitors, which, contrary to some reports in the literature, appears to be mostly AUC-proportional. Highly infrequent dosing was found to be sub-exposure proportional, and our work provides a rational explanation for this effect. Taken together, the approach of constructing a dynamic PK/efficacy model and eventually extending this model into a human setting by adjusting the PK half-life provides a rational basis for the preclinical-to-clinical translation of schedule effects. Citation Format: Mayankbhai Patel, Jocelyn Staunton, Jerome Mettetal, Santhosh Palani, Jeffrey Ecsedy, Chuang Lu, Elly Barry, Eric Westin, Mark Manfredi, Wej Chyi Shyu, Katayoun Jessen, Esha A. Gangolli, Arijit Chakravarty. Preclinical-to-clinical translation of optimal schedule for MLN0128, an investigational novel small-molecule selective mTOR1/2 inhibitor, based on exposure-efficacy modeling. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 3532. doi:10.1158/1538-7445.AM2013-3532