Abstract 4676: Deconvolution of extrinsic and intrinsic factors influencing tumor growth in solid cancer PDX models to support design, analysis and application of PDX-based pharmacology studies

Abstract

Abstract PDX models are one of the gold standard models in preclinical drug development. Their intra- and intertumoral heterogeneity is accepted as a model immanent feature. However, the model performance directly impacts the study design as well as the interpretation of experimental data using those models. To discriminate between intrinsic and extrinsic factors driving heterogeneous tumor growth of PDX models we investigated the model performance of >500 subcutaneously growing PDX models of solid cancer by examining growth curves of control arms run in the framework of > 5000 independent experiments. The relative tumor volume (RTV) over time was correlated with the number of passages in mice, the respective mouse strain, and the applied control vehicle per model as well as across tumor types. The analysis of ten different colon cancer PDX models revealed statistically significant differences between RTV on a specific day across different passages in a specific model as well as across all ten models (Kruskall-Wallis test). However, neither the variance nor the doubling time of the tumor correlated with the number of passages in mice. Based on these results we extended the analysis across six different tumor types with 33 distinct models in six different mouse strains. Again, an influence of the passage number on the tumor volume on a specific experimental day was depicted. The range of mean RTV related to a specific passage were least pronounced on day 7 (71% - 1626%), most notable on day 14 (65% - 4543%) and decreasing again on day 21 (34% - 3633%). In this larger cohort a subset of data points was more similar in performance to each other than any other set. The clustering was based on performance on experimental day 7. This subset received either saline or was untreated. This demonstrates that non-saline ingredients have an evident effect on tumor progression. Due to limited size of the dataset the influence of the mouse strain was evident only on a descriptive level. Whereas the use of different nude mouse strains, such as CD1-, athymic, or NMRI nude, did not affect tumor volume over time, the use of more immunodeficient mice such as SCID, NSG or NOG led to a faster tumor growth in some of the models across different passages. In the next iteration we will analyze the full dataset overcoming current limitations due to small subsets in a specific condition. The actual aim of this project is to better understand and measure tumor heterogeneity in PDX models. Based on these analyses the output of PDX-based studies will be optimized, as we get a better understanding how extrinsic factors (mouse strain and vehicle) and intrinsic factors (passage number) influence tumor growth. In a longer perspective the ability to model tumor volume over time in a specific model will reduce the need of control animals in future experiments, thereby actively supporting the concept of 3R. Citation Format: Mathew Clark, Kanstantsin Lashuk, Edward McGowan, Julia Schueler. Deconvolution of extrinsic and intrinsic factors influencing tumor growth in solid cancer PDX models to support design, analysis and application of PDX-based pharmacology studies. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4676.