Modeling the efficacy of different anti-angiogenic drugs on treatment of solid tumors using 3D computational modeling and machine learning

Milad Mousavi; Mahsa Dehghan Manshadi; Madjid Soltani; Farshad M Kashkooli; Arman Rahmim; Amir Mosavi; Michal Kvasnica; Peter M Atkinson; Levente Kovács; Andras Koltay; Norbert Kiss; Hojjat Adeli

doi:10.1016/j.compbiomed.2022.105511

Modeling the efficacy of different anti-angiogenic drugs on treatment of solid tumors using 3D computational modeling and machine learning

Comput Biol Med. 2022 Jul:146:105511. doi: 10.1016/j.compbiomed.2022.105511. Epub 2022 Apr 18.

Authors

Affiliations

¹ Cancer Institute of Iran, Tehran University of Medical Sciences (TUMS), Tehran, Iran.
² Cancer Institute of Iran, Tehran University of Medical Sciences (TUMS), Tehran, Iran; Department of Mechanical Engineering, K. N. Toosi University of Technology, Tehran 1999143344, Iran.
³ Department of Electrical and Computer Engineering, University of Waterloo, Ontario, Canada; Centre for Biotechnology and Bioengineering (CBB), University of Waterloo, Waterloo, Ontario, Canada; Department of Mechanical Engineering, K. N. Toosi University of Technology, Tehran 1999143344, Iran.
⁴ Department of Physics, Ryerson University, Toronto, ON, Canada; Department of Mechanical Engineering, K. N. Toosi University of Technology, Tehran 1999143344, Iran.
⁵ Department of Radiology and Physics, University of British Columbia, Vancouver, BC, Canada; Department of Integrative Oncology, BC Cancer Research Institute, Vancouver, BC, Canada.
⁶ Institute of Software Design and Development, Obuda University, 1034, Budapest, Hungary; National University of Public Service, Budapest, Hungary; Institute of Information Engineering, Automation and Mathematics, Slovak University of Technology in Bratislava, Bratislava, Slovakia. Electronic address: amir.mosavi@uni-obuda.hu.
⁷ Institute of Information Engineering, Automation and Mathematics, Slovak University of Technology in Bratislava, Bratislava, Slovakia.
⁸ Faculty of Science and Technology, Lancaster University, Lancaster, UK; Geography and Environmental Science, University of Southampton, Highfield, Southampton SO17 1BJ, UK; Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, 11A Datun Road, Beijing 100101, China; Lancaster Environment Centre, Lancaster University, Bailrigg, Lancaster LA1 4YR, UK.
⁹ Biomatics Institute, John von Neumann Faculty of Informatics, Obuda University, 1034, Budapest, Hungary; Physiological Controls Research Center, University Research and Innovation Center, Obuda University, 1034, Budapest, Hungary.
¹⁰ National University of Public Service, Budapest, Hungary.
¹¹ Department of Biomedical Informatics and Neuroscience, The Ohio State University, Columbus, OH, 43220, USA.

PMID: 35490641
DOI: 10.1016/j.compbiomed.2022.105511

Abstract

Accurate simulation of tumor growth during chemotherapy has significant potential to alleviate the risk of unknown side effects and optimize clinical trials. In this study, a 3D simulation model encompassing angiogenesis and tumor growth was developed to identify the vascular endothelial growth factor (VEGF) concentration and visualize the formation of a microvascular network. Accordingly, three anti-angiogenic drugs (Bevacizumab, Ranibizumab, and Brolucizumab) at different concentrations were evaluated in terms of their efficacy. Moreover, comprehensive mechanisms of tumor cell proliferation and endothelial cell angiogenesis are proposed to provide accurate predictions for optimizing drug treatments. The evaluation of simulation output data can extract additional features such as tumor volume, tumor cell number, and the length of new vessels using machine learning (ML) techniques. These were investigated to examine the different stages of tumor growth and the efficacy of different drugs. The results indicate that brolucizuman has the best efficacy by decreasing the length of sprouting new vessels by up to 16%. The optimal concentration was obtained at 10 mol m^-3 with an effectiveness percentage of 42% at 20 days post-treatment. Furthermore, by performing comparative analysis, the best ML method (matching the performance of the reference simulations) was identified as reinforcement learning with a 3.3% mean absolute error (MAE) and an average accuracy of 94.3%.

Keywords: Anti-angiogenic drugs; Artificial intelligence; Bevacizumab; Brolucizumab; Cancer; Ranibizumab; Solid tumor; Tumor growth.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Angiogenesis Inhibitors* / adverse effects
Computer Simulation
Humans
Machine Learning
Neoplasms* / pathology
Neovascularization, Pathologic / drug therapy
Ranibizumab / adverse effects
Vascular Endothelial Growth Factor A

Substances

Angiogenesis Inhibitors
Vascular Endothelial Growth Factor A
Ranibizumab