Abstract B034: PHYFUM: Reconstructing the evolutionary dynamics of human tissues using fluctuating methylation clocks

Abstract What? We present a new suite of bioinformatic tools (PHYFUM, PHYFUMflow, and PHYFUMr) to track the evolution of human tissues organized in clonal stem-cell populations using methylation data from multiple samples per patient. We demonstrate the validity of our methods, characterize their accuracy under a broad simulation parameter space, and use them to analyze a total of 40 colon, 28 small intestine, and 31 endometrium samples extracted from 23 patients. We found statistically significant differences (p. adj 0. 001, Dunn’s test with Holm correction) between the shapes of the endometrial and intestinal trees (both types), highlighting the power of our method to develop evolutionary biomarkers for cancer progression. Why? The development of cancer from healthy somatic cells is fundamentally an evolutionary process. Understanding this process is crucial for enhancing its clinical management and developing more effective treatment strategies. We cannot use human cell-labeling techniques in vivo; however, tracing the fates of somatic cells would advance somatic evolutionary theory and our understanding of normal tissue function and its changes throughout development, aging, and cancer progression. Population genetic and phylogenetic models can leverage the information recorded in heritable alterations to reconstruct the history of the unobserved somatic cells. Most such methods are limited by the few parameters they can estimate and restricted to neoplastic samples due to the lack of adequate evolutionary signal in healthy somatic cells. How? Gabbutt et al. recently introduced fluctuating methylation clocks (FMCs) —tissue-specific CpG sites that randomly change their methylation state—and a model that uses them to reconstruct the evolution of a clonal stem-cell population that can be sampled together, like crypts (e. g. , intestine, Barrett’s esophagus) and glands (e. g. , endometrium). We have developed a multi-sample extension of this model that simultaneously reconstructs the evolutionary dynamics of stem cells within a niche and the niches within the tissue. This phylodynamic model generates additional estimates, including ancestral relationships between sampled clonal populations, their calendar divergence times, the number of effective evolutionary niches, and how these parameters change over time. So what? The stochastic nature of neoplastic progression makes it unlikely for any particular somatic alteration to be highly predictive of its outcome. This can explain why developing reliable traditional biomarkers has been so difficult. Alternatively, evolutionary biomarkers measure the characteristics of the process itself and thus should apply to all neoplasms. Parametric phylogenetic reconstruction models like the one introduced here will enable us to develop such universal biomarkers. Our ultimate goal is for evolutionary biomarkers to join evolutionary therapies in the revolution of cancer clinical management. Citation Format: Diego Mallo, Pablo Bousquets-Muñoz, Calum Gabbutt, Heather E. Grant, Darryl Shibata, Trevor A. Graham, Carlo C. Maley. PHYFUM: Reconstructing the evolutionary dynamics of human tissues using fluctuating methylation clocks abstract. In: Proceedings of the AACR Special Conference in Cancer Research: Cancer Evolution: The Dynamics of Progression and Persistence; 2025 Dec 4-6; Albuquerque, NM. Philadelphia (PA): AACR; Cancer Res 2025;85 (23Suppl): Abstract nr B034.