Silvia Maria Lavezzi (1), Laura Iavarone (1), Manu De Rycker (2), Xuan Zhou (3), Tara Yang (1), Jianping Zhang (1), Tim Miles (4), Chao Chen (5)
(1) Clinical Pharmacology Modelling and Simulation, Parexel International, (2) Drug Discovery Unit, Wellcome Centre for Anti-Infectives Research, University of Dundee, Dundee, UK, (3) Clinical Pharmacology Modelling and Simulation, GlaxoSmithKline, Shanghai, China , (4) Global Health R&D, GlaxoSmithKline, Tres Cantos, Spain, (5) Clinical Pharmacology Modelling and Simulation, GlaxoSmithKline, London, UK
Objectives: Parasitic diseases represent a global health problem. The development of new drugs for their treatment which facilitate compliance and have adequate efficacy and safety profiles is a crucial challenge. The probability of pharmacological success (PoPS) can be defined as the probability of achieving desired pharmacological and safety response rates in a treated population. As already shown in [1], PoPS computation can support progression decisions at early stages of development. A preclinical candidate compound was found to be highly efficacious against a parasitic disease in both in vitro and in vivo studies: in the present work, the methodology employed for human projections and evaluation of PoPS for informing the decision to progress the compound to first-in-human (FIH) trial will be outlined (modifying actual data to maintain the blinding).
Methods: Human pharmacokinetic (PK) parameters were predicted based on non-clinical data corrected for body weight; projections were done assuming oral BID dosing. Moderate inter-subject variability (ISV) was assumed for clearance and volume of distribution (CV%=30%), while low ISV was assumed on absorption rate constant (CV=10%). Body weight was simulated using a log-normal distribution with median 70 kg and CV%=14% [2]. Effect vs time (E(t)) was represented as a Hill curve driven by predicted human concentration profile, with drug potency parameters (EC50 and γ) derived from in vitro experiments in human cells, and maximum effect (Emax) corresponding to 100%. The pharmacodynamic endpoint associated with efficacy was identified in the preclinical model (with Phoenix WinNonlin 8.0 [3]) and computed for humans. The evaluated steady-state endpoints were AUC, time above EC50, time above EC90, and area under the E(t) curve (AUEC) over 24 hours (both based on total blood and free plasma concentration data). For the pharmacodynamic endpoint, the threshold corresponding to 95% of parasite reduction in vivo was derived. For PoPS computation, uncertainty was included on EC50 and γ, using a uniform prior distribution over a range derived based on in vitro experiments results. For each EC50 and γ combination (m=1000), and for each dose level (in the range of 100-2000 mg BID), individual data (n=1000) were generated (using NONMEM 7.4.3 [4] and R 3.5.1 [5], with the package rspeaksnonmem [6]). Success criterion was defined for the candidate compound used as monotherapy: safety (AUC24h≤100000 ng/mLxh) has to be maintained in 95% of subjects, while efficacy (i.e. 95% of parasite reduction, suggested from standard of care) has to be achieved in 80% of subjects within the same population. The PoPS was then obtained for each dose level as the success rate over the 1000 scenarios with different EC50 and γ combinations.
Results: In the preclinical model, the best predictors of efficacy were time above EC50 and AUEC over 24 h based on total blood data. AUEC24h was selected as the driver for efficacy. For the candidate compound used as monotherapy, the PoPS was greater than 0 for dose levels >200 mg BID, it increased steadily up to 500 mg BID, where it reached 48.4%, and then it went abruptly back to 0 for dose levels >500 mg BID. The PoPS derivation allowed to quantify the impact of uncertainty on success and gave more confidence in proceeding with the FIH trial.
Conclusions: Computation of the compound-specific PoPS, based on the AUEC (a time-integrated pharmacodynamic response that corrects the between-species pharmacokinetic differences) for animal-to-human efficacy translation, allowed to assess the pharmacological strength of the compound for treating the parasitic disease, thus supporting FIH progression decision. Repeating this exercise when new information is gathered (e.g. about the actual PK profile in human) allows to support progression decisions at future phases of development.
References:
[1] X. Zhou, C. Chen, O. Graff. Model-Based Estimation of Probability of Pharmacological Success for CNS Compounds. PAGE 28 (2019) Abstr 8917
[2] W.J., Millar. (1986). Distribution of body weight and height: comparison of estimates based on self-reported and observed measures. Journal of Epidemiology & Community Health, 40(4), 319-323.
[3] Phoenix WinNonlin™, Certara USA, Inc., 100 Overlook Center, Suite 101, Princeton, NJ 08540 USA
[4] https://www.iconplc.com/innovation/nonmem/
[5] https://cran.r-project.org/
[6] https://github.com/MikeKSmith/rspeaksnonmem
Reference: PAGE () Abstr 9327 [www.page-meeting.org/?abstract=9327]
Poster: Methodology - Other topics