Abstract 1699: SignaTree: A tool to identify evolutionary trajectories of the activity of mutational processes in TRACERx

Abstract

Abstract Introduction: The analysis of somatic mutations has identified different mutational signatures that represent different exogenous or endogenous mutagenic processes active during tumor development. Those biological processes represent a main driver of diversity, but little is known about the dynamics of their activity over time. Based on the multi-region sequencing data of the TRACERx study we developed SignaTree, a tool to deconvolve mutational signatures for each subclone within a phylogenetic tree of a tumor to quantify the dynamics of mutational processes over time. Methods: The subclonal architecture of 396 TRACERx tumors with phylogenetic tree output were used to develop SignaTree and investigate mutational signature dynamics over time. Only mutational signatures that were identified in a de novo signature analysis of the TRACERx 421 patients were included for deconstruction. A randomized sampling technique informed by both ancestral and descendent mutations was used to mitigate the uncertainty of signature deconvolution within small subclones. A bootstrapping method was applied to reconstruct the distribution of the mutational signature activity within the ancestral clone to statistically test the change in signature activity for significance between ancestral and descendent clones. Simulations of mutational signatures within different subclones along a phylogenetic tree were used for validation purposes. Results: SignaTree was identified to better predict mutational signature activity in 83% of simulations in comparison to applying a deconstruction tool (deconstructSigs) without a randomized sampling step for small subclones. 77% of the TRACERx tumors showed at least one significant change in signature activity between ancestral and descendent clones, with the majority of tumors exhibiting multiple shifts in mutational processes during their evolutionary history. The most common signature dynamic observed involved a decrease in smoking signature (SBS4) and an increase in APOBEC (SBS2 + SBS13) related activity. 8% of tumors presented evidence for dynamic fluctuations in APOBEC activity over time (episodic APOBEC) which has previously only been described in vitro. Most of them exhibited episodic APOBEC activity only in a small subset of tumor regions highlighting local heterogeneity in addition to timing diversity of the endogenous APOBEC process. Conclusion: SignaTree allows the deconstruction of mutational signatures within subclones and improves the ability to detect signatures in clones with a low number of mutations. Using this approach, it is possible to identify different evolutionary patterns of mutational signature activity during tumor development including episodic APOBEC activity. Citation Format: Michelle Dietzen, Oriol Pich, TRACERx Consortium, Simone Zaccaria, Charles Swanton, Nicholas McGranahan. SignaTree: A tool to identify evolutionary trajectories of the activity of mutational processes in TRACERx [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1699.