Anaïs Glatard[1], Alienor Berges[1], Claire Ambery[1], Jan Osborne[2], Randall Smith[3], Emilie Hénin[4], Chao Chen[1]
[1]. Clinical Pharmacology Modelling and Simulation, GSK, UK. [2]. Non-Clinical Safety Projects, Safety Assessment, GSK, UK. [3]. Computational Toxicology, Safety Assessment, GSK, US. [4]. Université Claude Bernard Lyon1, Lyon, France ; UMR CNRS 5558 – équipe Evaluation et Modélisation des Effets thérapeutiques. Hospices Civils de Lyon, Service de Pharmacologie Clinique et Essais Thérapeutiques.
Objectives: In drug development, a “no observed adverse-effect level” (NOAEL) is derived from animal studies and used to infer a maximal safe exposure in human[1,2]. The current practice is to determine NOAEL for first-time-in-human studies on the basis of discrete observations from pivotal toxicity studies, which are limited by specific experimental conditions. We hypothesise that this practice may not represent the optimal use of data, and may carry a risk of misrepresenting the true NOAEL[3-6]. We propose to explore by simulations a model-based approach in complement to the current practice. A real case was used to illustrate the simulation findings.
Methods: A linear logit model including i) background toxicity incidence, ii) study duration effect and iii) dose effect was used to simulate a dose-toxicity probability pattern of relevant toxicity experimental settings. Different scenarios in terms of sample size per dose group, level and number of doses and degree of drug toxicity were considered to investigate the NOAEL distribution. For each scenario, 1000 replicates of individual binary toxicity effects (presence/absence) were simulated. NOAEL was defined as the next dose level down from the dose group wherein the toxicity frequency was higher than in the concurrent control group. A real life example was used to illustrate the model-based approach to support NOAEL determination using histopathological data from 7 rat toxicity studies.
Results: The simulations showed that higher sample size or higher number of doses tends to lower NOAEL doses. In the case of a drug with flat dose-response, the NOAEL dose lower than the lowest dose is selected as frequently as the highest dose. The modelling of real data allowed characterisation of the dose-toxicity probability relationship for the example drug across studies. The NOAEL doses derived by current practice from each study could be then graphically linked to the true model-based toxicity probability
Conclusions: A model-based approach allows a quantitative characterisation of the toxicity frequency as a function of any dose (even those not tested) integrating treatment duration and uncertainty due to sample size. The simulations illustrated an issue associated with the current approach in defining and estimating NOAEL: NOAEL tends towards the lowest doses studied and this tendency is higher with larger sample size. These findings suggest that a probability-based approach might be a more robust alternative.
References:
[1] U.S. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER). Guidance for Industry, M3(R2) Nonclinical Safety Studies for the Conduct of Human Clinical Trials and Marketing Authorization for Pharmaceuticals, ICH. Revision 1. 2010.
[2] European Medicines Agency, Committee for Medicinal Products for Human Use. Guideline on strategies to identify and mitigate risks for first-in-human clinical trials with investigational medicinal products. London. 2007. Doc. Ref.EMEA/CHMP/SWP/28367/07
[3] Slob W, Pieters MN. A Probabilistic Approach for Deriving Acceptable Human Intake Limits and Human Health Risks from Toxicological Studies: General Framework. Risk Anal. 1998;18:787-98.
[4] Dorato MA, Engelhardt JA. The no-observed-adverse-effect-level in drug safety evaluations: Use, issues, and definition(s). Regul Toxicol Pharmacol. 2005;42:265-74.
[5] Filipsson AF, Sand S, Nilsson J, Victorin K. The Benchmark Dose Method—Review of Available Models, and Recommendations for Application in Health Risk Assessment. Crit Rev Toxicol. 2003;33(5):505–42.
[6] Ritz C, Gerhard D, Hothorn LA. A Unified Framework for Benchmark Dose Estimation Applied to Mixed Models and Model Averaging. Statistics in Biopharmaceutical Research. 2013;5:1,79-90.
Reference: PAGE 23 () Abstr 3247 [www.page-meeting.org/?abstract=3247]
Poster: Methodology - Other topics