Abstract

Abstract Recent breakthroughs in phylogenetic analysis of bulk tissue have allowed researchers to reconstruct the evolutionary histories of tumours. The hope is that knowledge of the order and timing of genetic mutations will allow us to characterise the properties of early-stage cancer cells, to better identify and target them. This is complicated by the phenomenon of tumour plasticity, the ability of cells to acquire the hallmarks of cancer via mechanisms other than heritable genetic mutation. In a tumour with only genetic heritability, changes in cell phenotype would occur only as a result of genetic mutations. By contrast, non-genetic heritability could lead to a disconnect between genotype and phenotype. Whilst both mechanisms are likely to impact on cancer development, we ask whether one has a significantly stronger influence than the other by simulating tumour growth under each regime and comparing the distribution of mutations obtained in either case to patient data. Here, we address this question in a real patient cohort by adapting a mathematical model of lung tumour evolution, which we use to test the validity of two scenarios. In a ‘low-plasticity’ scenario, changes to reproductive fitness occur when cells acquire mutations in the exome. Mutations may be ‘drivers’ (beneficial) or ‘passengers’ (deleterious or neutral). In a ‘high-plasticity’ scenario, mutations have no effect on cell fitness. Cells experience ‘driver-like’ or ‘passenger-like’ cell fitness changes on division, without leaving a genetic mark. The model describes three-dimensional growth of a tumour from a single cell and incorporates biologically-informed death patterns and local competition for space and resources. When the simulation has reached a realistic size, cells on the surface are sampled and sequenced to predict the relatedness of mutations present at detectable frequencies in each of several regions. The outputs are designed to allow comparison with those of the TRACERx cohort of 421 non-small- cell lung cancer patients (NSCLC), comprising multi-region whole-exome (WXS) and bulk RNA sequencing. In this ongoing work, we show results from a large cohort of simulations under both scenarios and predict corresponding patterns of genetic similarity. This allows us to use approximate Bayesian computation (ABC) to predict the mechanisms at play in the TRACERx cohort of 421 non-small-cell lung cancer patients (NSCLC). We present a novel ‘meta-inference’ approach, where evolutionary parameters are fit to each patient’s data using well-chosen summary statistics. Given the size of the TRACERx cohort, this enables the evaluation of each scenario by examining the plausibility and similarity of output parameters obtained across patients. We hope that this work will shed light on the role of heritability in lung cancer development and guide future research into therapeutic approaches. Citation Format: Helena M. K. Coggan, Carlos Martínez-Ruiz, James R. M. Black, Kristiana Grigoriadis, Nicholas McGranahan, Jasmin Fisher. An agent-based modelling framework to study cell plasticity in non-small cell lung cancer [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 PR013.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call