Abstract A027: Detecting and modeling frequency-dependent evolution in serially passaged primary triple negative breast cancer patient derived xenografts

Abstract

Abstract Tumor evolution is a contributing factor to most failed cancer treatments. The intratumor heterogeneity of cellular phenotypes found in the tumor microenvironment acts as an evolutionary substate for cancer cells during tumorigenesis, subsequent growth, and after the patient is in the clinic. Selective forces imposed by certain medications can cause the emergence of a tumor with a resistant phenotype. Chemotherapies that target rapidly dividing cells will select for slower dividing ones which are less sensitive to such agents. However, many selective pressures acting on the tumor microenvironment are yet to be elucidated. For example, the effect one cancer cell has on another’s growth through competitive (i.e., the struggle for resources) or cooperative (e.g., exchange of growth factors) interactions remains poorly understood. Our central hypothesis is that cancer cells with distinct karyotypes will have a varying growth rate influenced by the karyotype landscape of the entire tumor. That is, we expect the fitness of a cancer cell to be largely determined by its karyotype and the karyotypes of its neighbors. Thus, we predict that frequency-dependent selection has a significant impact on clonal evolution in cancer. To test this, we have developed two novel methods for respectively identifying likely instances of frequency-dependent selection and modeling the long-term evolutionary trajectory in such cases. To detect frequency-dependent effects, we investigate the relationship between the frequency of one clone and the growth rate of all others (including itself). If a statistically significant correlation can be found, we proceed to modeling clonal evolution. To do this, we use a penalized least squares method and find the best-fitting parameterization of the replicator equation to the clonally resolved frequency data. This provides us with a payoff matrix describing the intensity of competition between each pair of clones. Furthermore, it allows us to model the long-term evolutionary dynamics of the tumor and to make predictions regarding co-existence of clones or clonal sweeps. Recently, investigators serially passaged primary triple negative breast cancer (TNBC) patient derived xenograft (PDX) tumors in NRG mice for over a year, inferring clonal frequencies at multiple timepoints using single cell whole genome (scWGS) and RNA sequencing (scRNA-seq). From this data, we were able to detect statistically significant frequency-dependent effects in the majority of serially passaged PDX tumors. Of these, most are predicted to result in a clonal sweep. Three of the PDX tumors were treated with cisplatin which revealed clone-specific drug response. This presents an opportunity for adaptive therapy scheduling which takes advantage of fitness tradeoffs under different contexts. A greater understanding of the effects frequency-dependent selection has on tumor evolution will contribute to the design of therapies intended to preclude or delay the emergence of resistance to treatment. Citation Format: Thomas A. Veith, Richard Beck, Noemi Andor. Detecting and modeling frequency-dependent evolution in serially passaged primary triple negative breast cancer patient derived xenografts [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr A027.