Model-based analysis of patient-derived organoids for evaluating anticancer drugs
Davide Ronchi1, Elena Maria Tosca1, Silvia De Siervi2, Cristian Turato2, Marco Gaetano Lolicato2, Paolo Magni1
1) Department of Electrical, Computer and Biomedical Engineering, University of Pavia, Pavia, Italy; 2) Unit of Immunology and General Pathology, Department of Molecular Medicine, University of Pavia, 27100 Pavia, Italy.
Introduction: Cancer organoids are a 3D in vitro cancer model that faithfully mimic the characteristics of human tumors from which they originated. Patient-derived organoids (PDOs) have been successfully generated for various cancer types, including intrahepatic cholangiocarcinoma (CCA) [2]. Half between 2D cell cultures and in vivo models, organoids have garnered great attention in the scientific community due to their multiple potential applications, such as drug screening, preclinical-to-clinical translation, and personalized medicine [4]. Despite the excitement, several issues still hamper the use of PDOs in translational cancer research [5]. In particular, quantitative approaches to analyze data are currently lacking.
Objectives: This work aims to exploit mathematical modeling and simulation (M&S) to boost and improve the use of cancer organoids for the preclinical assessment of anticancer drug activity. A model describing tumor growth and anticancer drug effect in CCA PDOs has been developed. It can be potentially used to predict conditions not experimentally explored, providing a tool for testing hypotheses, and derive drug-specific and experiment-independent metrics of drug efficacy, supporting the rank and screen of candidate agents.
Methods: Two PDOs studies were performed and analyzed. Fresh tumor biopsies from two CCA patients were minced and digested in small cell clusters, which were then seeded into Matrigel. After achieving stable growth, PDOs were treated with two investigating small molecules (D1 and D2), targeting the Voltage Dependence Anion Selective Channel isoform 1 (VDAC1), at different concentrations (5,10,15 uM for D1 and 1,5,10,15 uM for D2) or with the standard-of-care drug, i.e. gemcitabine (GMB) at 10 uM. Untreated (i.e. control) and treated PDOs were monitored for 72h through brightfield microscopy imaging. Collected images were processed through a semi-automated pipeline based on a literature-validated rendering software [6] to obtain longitudinal data of morphological parameters, i.e. area, surface area, volume, and sphericity, for all the individual 3D structures composing the organoid cultures. A mathematical model describing tumor growth and its inhibition after drug exposure was developed based on the time profiles of organoid volumes. Due to the high variability between 3D structures composing the organoids, the median volumes, stratified by study and treatment arm, were considered. Several commonly used tumor growth inhibition (TGI) models were evaluated. Simultaneous fitting of the control and treated arms was performed for each PDO study.
Results: Among the tested models, the Simeoni TGI model [7] with a saturation phase best described the time profiles of volume in untreated and treated CCA PDOs. More in detail, for both the investigated agents the anticancer effect, resulted not linearly proportional to concentration. An Emax and Hill (coefficient=2) models were used to grasp the non-linear effect of D1 and D2, respectively. In addition, a delayed effect on tumor growth was observed for GMB and modeled by adding an effect compartment. This delay was not observed for the two small molecules, likely due to their smaller size and faster mobility. The availability of multiple concentration levels for D1 and D2 allowed testing of the predictive performance of the developed TGI models. For both cases, the model was repeatedly identified leaving out one of the treated arms at a time that was then used as an external validation dataset. Mean prediction error (MPE) and mean absolute prediction error (MAPE) were 2.24%, -3.26%, -3.19% and 7.87%, 18.1%, 28.7%, respectively, for the 5,10, and 15 uM arms of D1. While, for the D2 at 1,5,10 and 15 uM the obtained MPE and MAPE were 0.47%, 9.14%, -6.1%, 13%, and 5.59%, 38.78%, 11.11%, 23.8%, respectively.
Conclusions: For the first time, a mathematical model describing TGI in cancer PDOs was developed. The proposed approach was able to successfully grasp the unperturbed tumor growth and the anticancer efficacy exerted by GMB and two investigated agents on CCA PDOs, showing also good predictive performances. Results strongly encouraged to explore the potential of M&S to derive quantitative measurements of anticancer drug efficacy from PDO studies, which are currently lacking, and to predict untested administration schedules. M&S can serve to increase the establishment of PDOs in nonclinical practice and improve in vitro in vivo translatability, in accordance with the FDA's 3R guidelines [8].
References:
[1] Drost J, Clevers H. Organoids in cancer research. Nat Rev Cancer. 2018 Jul;18(7):407-418. https://doi.org/10.1038/s41568-018-0007-6
[2] De Siervi S and Turato C. Liver Organoids as an In Vitro Model to Study Primary Liver Cancer. Int J Mol Sci. 2023 Feb 25;24(5):4529. doi: 10.3390/ijms24054529.
[3] Banales, J.M., Marin, J.J.G., Lamarca, A. et al. Cholangiocarcinoma 2020: the next horizon in mechanisms and management. Nat Rev Gastroenterol Hepatol. 2020; 17: 557–588. https://doi.org/10.1038/s41575-020-0310-z
[4] Zhou, Z., Cong, L., & Cong, X. Patient-Derived Organoids in Precision Medicine: Drug Screening, Organoid-on-a-Chip and Living Organoid Biobank. Frontiers in Oncology. 2021 Dec; 11. https://doi.org/10.3389/fonc.2021.762184
[5] Verduin, M., Hoeben, A., De Ruysscher, D., & Vooijs, M. Patient-Derived Cancer Organoids as Predictors of Treatment Response. Frontiers in Oncology. 2021 Mar; 11. https://doi.org/10.3389/fonc.2021.641980
[6] Piccinini F, Tesei A, Arienti C, Bevilacqua A. Cancer multicellular spheroids: volume assessment from a single 2D projection. Comput Methods Programs Biomed. 2015 Feb;118(2):95-106. doi: 10.1016/j.cmpb.2014.12.003. Epub 2014 Dec 23. PMID: 25561413.
[7] Simeoni M, Magni P, Cammia C, De Nicolao G, Croci V, Pesenti E, Germani M, Poggesi I, Rocchetti M. Predictive pharmacokinetic-pharmacodynamic modeling of tumor growth kinetics in xenograft models after administration of anticancer agents. Cancer Res. 2004 Feb 1;64(3):1094-101. doi: 10.1158/0008-5472.can-03-2524. PMID: 14871843.
[8] Han JJ. FDA Modernization Act 2.0 allows for alternatives to animal testing. Artif Organs. 2023 Mar;47(3):449-450. doi: 10.1111/aor.14503. Epub 2023 Feb 10. PMID: 36762462.