2024 AIChE Annual Meeting

(374d) Optimizing Personalized Treatment Policy for Cancer Chemotherapy

Authors

Phan, T. V. - Presenter, Johns Hopkins University
Li, S., Princeton University
Howe, B., Princeton University
Amend, S. R., Johns Hopkins School of Medicine
Pienta, K. J., University of Michigan, Ann Arbor
Brown, J. S., Moffitt Cancer Center
Gatenby, R. A., Moffitt Cancer Center
Austin, R. H., Princeton University
Kevrekidis, I. G., Princeton University
Cancer is a disease of uncontrolled proliferation by transformed cells subject to evolution by natural selection [1]. It is among the most devastating classes of human pathologies, resulting in millions of deaths annually worldwide. Improving cancer treatment can alleviate this suffering, enhance quality of life, and extend survival for those affected by the disease. One of the most popular treatments of cancer is via chemotherapy, which employs potent drugs to inhibit tumor growth. In adaptive chemotherapy, the administration of drugs is personalized, depending on how each patient responds to the drugs [2].

We utilize an evolutionary multi-population model of prostate cancer to describe the dynamics of the cancer population during chemotherapy, considering the standard evolutionary scenario where drug-resistant cells are present from the outset [3]. We formalize the quest of finding a treatment policy to maximize patient life expectancy into a well-defined mathematical optimization problem. Here we apply an active search method, i.e. Bayesian optimization [4] for a Gaussian process surrogate model [5], to demonstrate how both single-drug and multi-drug cancer therapy can be optimized. This method can be generalized to more complex treatment policies beyond current practice. Our findings show that a tight-control adaptive therapy at a high tumor burden can enhance life expectancy while also significantly improving drug efficiency compared to current clinical practices. As part of this ongoing work, currently have patients undergoing single-drug treatment based on the optimized treatment policies we have identified.

We explore possible computational savings by linking BO with deterministic global Gaussian Process optimization of the acquisition function at each step.

[1] Joel S. Brown, Sarah R. Amend, Robert H. Austin, Robert A. Gatenby, Emma U. Hammarlund, and Kenneth J. Pienta. Updating the definition of cancer. Molecular Cancer Research 21, no. 11 (2023): 1142-1147.

[2] Robert A. Gatenby, Ariosto S. Silva, Robert J. Gillies, and B. Roy Frieden. Adaptive therapy. Cancer research 69, no. 11 (2009): 4894-4903.

[3] Zhang, Jingsong, Jessica J. Cunningham, Joel S. Brown, and Robert A. Gatenby. "Integrating evolutionary dynamics into treatment of metastatic castrate-resistant prostate cancer." Nature communications 8, no. 1 (2017): 1816.

[4] Bobak Shahriari, Kevin Swersky, Ziyu Wang, Ryan P. Adams, and Nando de Freitas. Taking thehuman out of the loop: A review of bayesian optimization. Proceedings of the IEEE, 104(1):148–175, 2016.

[5] Artur M. Schweidtmann, Dominik Bongartz, Daniel Grothe, Tim Kerkenhoff, Xiaopeng Lin, Jaromil Najman, and Alexander Mitsos. Deterministic global optimization with gau