(M-085) An innovative framework to find potential immunotherapy targets using QSP modeling and RNA-seq data

Mahya Aghaee – University of Florida; Victoria Cicchirillo – University of Florida; Kyle Adams – University of Florida; Alberto Riva – University of Florida; William Hager – University of Florida; Ashley Brown – University of Florida; Elias Sayour – University of Florida; Domenico Santoro – University of Florida; Rowan Milner – University of Florida; Bently Doonan – University of Florida; Helen Moore – University of Florida

Enter your core abstract text here using the suggested layout. Abstracts are text only: No figures/tables. Do not paste author or affiliation information in the body of your abstract. Max character count not including spaces for abstract body is 2800.: Title: Optimization of therapeutic regimens for a quantitative systems pharmacology (QSP) model based on RNA sequencing data
Authors: Mahya, Torri, Domenico, Alberto, Rebecca, Elias, Rowan, Bently, Helen

Objectives: Despite advances in the treatment of melanoma in the last 15 years with new therapies, there are still many melanoma patients who do not respond to therapy. We have developed novel methods to help with this; namely, identifying the most promising pathways to target, and optimizing regimens for therapies that target these pathways. In this work, we show our latest results applied to canine melanoma, a valuable comparative oncology model.

Methods: We previously developed a mechanistic mathematical model for melanoma tumor-immune dynamics and treatment response. In this work, we collected new samples of primary tumor samples and healthy tissue from canine melanoma patients and healthy canine controls. We also had samples from diseased lymph nodes and healthy lymph nodes. We identified key immune cells to include in the model by applying immune cell deconvolution to differential gene expression analysis of the bulk RNA-sequencing (RNA-seq) data. We compared this to single cell RNA-seq data. Comparison of differentially-expressed genes with the KEGG canine melanoma whole genome background helped determine pathways to include. We used this to update a quantitative systems pharmacology (QSP) model of canine melanoma tumor-immune dynamics. We analyzed the QSP model with global sensitivity analysis (Morris method, eFAST, and Sobol indices) to rank most-influential model parameters. We performed identifiability analysis on the most-influential parameters. We applied optimal control to determine the best regimens.

Results: We identified M1 and M2 macrophages, CD8+ T cells, and regulatory T cells as the most influential immune cell types in the disease, and incorporated them into our QSP model. The tumor proliferation and removal pathways were found to be highly influential, as expected. We also found the M2 macrophages signaling M2 macrophages decrease CD8+ T cell efficacy for killing melanoma cells to be an influential pathway. We validated our findings with single cell RNA-seq analysis. Optimal control determined levels of suppression or stimulation needed in the tumor-immune dynamics QSP model to achieve optimal outcomes for canine melanoma patients.

Conclusions: We previously developed a model that integrated information from the literature, canine melanoma patient and healthy canine control samples, and expert input. We used quantitative methods to determine key outcome drivers and predict interventions that optimize patient outcomes. In this new work, we analyzed and compared gene expression in additional samples, updated our previous model, compared three additional global sensitivity analysis methods, performed identifiability analysis, and re-ran optimal control to determine best regimens for this updated model. We plan to validate this method by testing predicted optimal regimens in animal studies. We will then update our QSP model with human melanoma samples and predict novel pathways to target and optimal therapeutic interventions for these targets in human melanoma.

Citations: Eftimie R, Hamam H. Modelling and investigation of the CD 4 + T cells – Macrophages paradox in melanoma immunotherapies. J Theor Biol. 2017;420:82-104. doi:10.1016/j.jtbi.2017.02.022

Erdag G, Schaefer JT, Smolkin ME, et al. Immunotype and Immunohistologic Characteristics of Tumor-Infiltrating Immune Cells Are Associated with Clinical Outcome in Metastatic Melanoma. Cancer Res. 2012;72(5):1070-1080. doi:10.1158/0008-5472.CAN-11-3218

Gallaher J, Larripa K, Renardy M, et al. Methods for determining key components in a mathematical model for tumor–immune dynamics in multiple myeloma. J Theor Biol. 2018;458:31-46. doi:10.1016/j.jtbi.2018.08.037

Moore H. How to mathematically optimize drug regimens using optimal control. J Pharmacokinet Pharmacodyn. 2018;45(1):127-137. doi:10.1007/s10928-018-9568-y

Keywords: RNA-seq Analysis, QSP modeling, Optimization