Michael Heathman, Amparo de la Peña, and Jenny Chien
Eli Lilly and Company, Indianapolis, IN USA
Objectives: The most common treatment-emergent adverse events for an experimental drug (LY) were gastrointestinal, with nausea (N) and vomiting (V) among the most frequently reported. The incidence of N/V was dose dependent and greatest after the first dose of LY, at approximately the time of maximum concentration (day 2 to 3). Incidence of N/V declined rapidly after the first 2 weeks of treatment. The development team was interested in whether patients would benefit from dose titration at the Phase 3 dose levels.
The objective of this analysis was to develop an exposure-response model to characterize the relationship between LY concentration and the onset, duration, and severity of N/V, as well as the development of tolerance. The model was used to simulate N/V under various dose titration regimens.
Methods: A Markov Chain model was developed, with N/V severity incorporated as different states within the chain. The onset and duration of events were governed by the transition probabilities among these states. As N/V incidence was observed to track the LY concentration profile, longitudinal predicted concentrations from a previously developed pharmacokinetic model were used for exposure-response. These concentrations were included as modifiers on the transition probabilities to affect onset, duration, and severity. Development of tolerance was incorporated using effect compartments.
The resulting model was used to simulate N/V incidence over the course of various dose titration regimens.
Results: Increased LY concentration was found to increase the probability of N, regardless of the previous state, and to increase the probability of remaining in a moderate/severe N state. Probability of V depended on the previous N state, and increased with increasing concentration. Sustained exposure to LY was found to cause tolerance, decreasing the probability of both N and V.
There was no significant reduction in the model-estimated incidence of N/V with titration regimens that initiated with a low LY dose for 1 or more doses, before titrating to a higher dose.
Conclusions: The Markov Chain approach characterized the LY concentration-response relationship for onset, duration, and severity of N/V events, as well as the development of tolerance. The model-based evaluation of N/V supported that dose titration would not improve the overall incidence of N/V during initial treatment.
Reference: PAGE 23 (2014) Abstr 3083 [www.page-meeting.org/?abstract=3083]
Poster: Drug/Disease modeling - Safety