Learn from historical animal data – most frequently observed targets/organs in RDT studies and how can this guide the design of NAM testing batteries?

Abstract

Toxicity testing nowadays plays a critical role in decision-making, and in vivo animal studies are still required for hazard assessment. Toxicity assessment of the 21st century, however, aims to replace in vivo studies with new approach methodologies (NAMs) based on human-centric models such as in vitro and in silico approaches. One open question is the scope of the NAM testing battery. In this study, we like to learn from the existing animal studies on main target organs for the one endpoint of systemic toxicity after repeated dose exposure. For this purpose, a large inventory of rodent repeated dose toxicity (RDT) studies from high-quality databases was analyzed concerning the most often affected targets/organs in studies with repeated oral exposure. The project dataset comprised about 7,100 repeated dose toxicity studies with oral and inhalation exposure for approximately 3,000 chemical studies. The data were extracted from several high-quality databases (DBs), namely RepDose, ToxRef (US EPA), and Hess (NEDO). Studies on pharmaceuticals were extracted from eTOX (8000 RDT), which evolved as the largest preclinical toxicity database for drugs and drug candidates, which comprises more than 1,900 different substances. The analyses distinguish main target organs at the lowest observed adverse effect level (LOEL) from those observed overall. Detecting the LOEL with high probability is performed with six main TOs, i.e., body weight, liver, clinical chemistry, clinical signs, hematology, and kidney. In the second part of the analysis, the predictivity of observed target organ/apical findings from short-term was calculated for long-term studies regarding adverse liver outcomes. The histopathological findings, organ weight changes, and necropsy of the liver in subacute treatment revealed a good predictivity for adverse liver effects in subchronic treatment. This investigation was a primary step in getting insight into in vivo RDT outcomes. The coverage and prediction model can be used as a systemic tool to prove and maintain the expert-based validation of human risk assessment.