Retrospective Analysis of Clinical Laboratory Parameters in Han Wistar Rat Controls

Rupert Kellner^1*Alexander Amberg²Frank Bringezu³Dragomir Ivanov Draganov⁴Alexander Gurjanov⁵Annika Kreuchwig⁵Wolfgang Muster⁴Guillemette Duchateau-Nguyen⁴Nils Oberhauser⁶Paolo Piraino⁷Markus Schaefer²Nelly Simetska¹Thomas Steger-Hartmann⁵Sylvia Emmi Escher¹

• ¹Fraunhofer-Institut fur Toxikologie und Experimentelle Medizin ITEM, Hanover, Germany
• ²Sanofi, Preclinical Safety, Frankfurt, Germany
• ³Merck Healthcare KGaA, Darmstadt, Germany
• ⁴F Hoffmann-La Roche AG, Basel, Switzerland
• ⁵Bayer Pharma AG Forschung und Entwicklung, Berlin, Germany
• ⁶Novartis AG, Basel, Switzerland
• ⁷Organon SRL, Bucharest, Romania

Introduction: Despite the availability of control animal data sets from toxicological studies, the influence of external factors, such as age of animals, test site and study conditions, on clinical laboratory parameters in rats is only sparsely characterized. Objective: In order to analyze the covariates of study design, we leveraged the largest available curated collection of control animal data from toxicological studies, sourced from five European pharmaceutical companies. We investigated the influence of external factors on commonly measured clinical chemistry, enzyme activity and hematology parameters in Han Wistar rats of both sexes. Materials and Methods: 457,605 control group clinical laboratory data points from 1,288 legacy toxicity studies on Han Wistar rats were curated and analyzed by ANOVA and partial eta squared to discern their effect sizes. Results: Our analysis revealed that bodyweight, used as a surrogate for age in rats, significantly influences some parameters, while demonstrating stability in others. Descriptive statistics and tolerance intervals are provided for 20-gram body weight class intervals. The effect size of these body weight classes, as calculated by partial eta squared, is large for parameters that change during development (e.g. phosphate or alkaline phosphatase) but was negligible for more stable parameters (e.g. calcium and alanine aminotransferase). For parameters which are less dependent on body weight class, the relative influence of other factors namely the company providing the study data, as well as study year is more prominent. These factors likely act as summary factors for various influences such as changes in analytical protocols, diet or housing conditions. Conclusion: This analysis provides a comprehensive overview of parameter variability and offers critical guidance for parameters which need to be controlled when utilizing historical control data to establish reference intervals or generate virtual control groups.