Eline van Maanen (1), Oliver Ackaert (1), Nahor Haddish-Berhane (2), Nelleke Snelder (1), Joost de Jongh (1), Hugh A. Barton (2), Alison Betts (2)
(1) LAP&P Consultants BV, Leiden, The Netherlands; (2) Translational Research Group, Department of Pharmacokinetics, Dynamics & Metabolism, Pfizer, Groton (CT), USA
Objectives: The toxicity of a series of new antibody-drug conjugates (ADCs) have been assessed pre-clinically. The outcome of these toxicity studies were summarized per tissue in different categories. The objectives of this analysis were to (i) determine an objective no-observed-adverse-effect-level (NOAEL) by establishing an exposure-response relationship for toxicity using this categorical data, (ii) rank the different ADC’s according to their toxicity for pre-clinical screening.
Methods: Single end-point studies were conducted in rats to investigate the toxicity of a series of ADC’s. The toxicity was investigated in different tissues (heart, lung, liver, kidney, bone marrow, eye) using histopathology, following administration of vehicle and 3 different dose levels (±5 animals per dose group). The results were summarized per tissue in 5 toxicity scores, ranging from 0 (not toxic) to 5 (very toxic). The proportional odds model was compared to the differential odds model to analyze this categorical data [1]. Furthermore, it was investigated how the model outcome can be used to compare the toxicity of the different compounds.
Results: The proportional odds model was selected to analyze the toxicity scores, since no statistically significant improvement in model fit was observed when using a differential odds model. The toxicity in the different tissues of the different compounds could not be compared using the slope of the exposure-response relationship for toxicity, since the slope was highly dependent on the baseline probabilities of the different scores. The different compounds were compared using EC50’s, which were directly derived from the slope of the drug effect and cumulative probabilities between the cutpoints. The EC50x was defined as the concentration where the probability of score x is reduced by half. Not only the obtained values, but also the distance between them for a compound was found to be informative on its toxicity. The EC50x values of the different compounds were normalized by their tumor static concentrations to derive toxicity scores. Ranking of the compounds according to the obtained toxicity scores gave good correspondence with an external validation method.
Conclusions: A new model-based method was developed to easily assess and compare categorical toxicity scores. This method can be used for rapid screening of ADC’s during drug development.
References:
[1] Kjellsson, MC, Zingmark P-H, Jonsson EN, Karlsson MO. Comparison of proportional and differential odds models for mixed-effects analysis of categorical data. J Pharmacokin Pharmacodyn (2008) 35: 483-501.
Reference: PAGE 23 (2014) Abstr 3049 [www.page-meeting.org/?abstract=3049]
Poster: Methodology - New Modelling Approaches