Abstract 6928: Expressed mitochondrial variants capture patterns of tumor evolution at single-cell level

Abstract Single-cell lineage tracing (scLT) has recently emerged as a promising tool to investigate cancer evolution. Among other scLT markers, mitochondrial variants (MT-SNVs) have been successfully used to track clones in a variety of experimental and clinical settings. However, in spite of the dynamic accrual of such genetic mutations, analysis of MT-SNVs variation has often been limited to standard clustering procedures, not exploiting the information associated with these markers at full potential. Here, we benchmarked expressed MT-SNVs-based single-cell phylogenies. To this end, we generate two single-cell multi-omics datasets: i) a benchmarking dataset, with joint profiling of MT-SNVs, gene expression and ground truth (GT) clonal identity (from lentiviral barcodes cells), and ii) a clinical cohort of wild type and mutated SRFS2 Acute Myeloid Leukemia patients, at diagnosis, with joint profiling of MT-SNVs, gene expression and targeted nuclear SNVs. We first leveraged our benchmarking dataset for validation purposes, taking advantage of its different levels of clonal complexities (in vitro clonal mixtures vs in vivo xenografts) and its longitudinal nature (i. e. , matched primary tumor-lung metastasis couples). Supervised analysis detected (GT) clonal-specific mutations rarer than previously shown, but significantly associated with GT clonal labels. The performance of tested feature selection methods varied wildly in terms of GT MT-SNVs recover, with higher False Positive than False Negative Rates. Accurate detection of GT-clones from MT-SNVs correlated with samples clonal complexity, with a substantial number of GT-clones sub-optimally detected in the highest complexity sample generated. Longitudinal analysis of matched primary tumor-metastasis couples revealed remarkable stability of MT-SNVs, corroborating evidence from other studies without longitudinal GT lentiviral clones. We then asked whether expressed MT-SNVs hold enough signal to reconstruct reliable single-cell phylogenies. Transfer Bootstrap (TS) supports of reconstructed trees correlated with clades size, with a significant number of deep clades showing high (. 70) TS values across samples and bootstrapping procedures. Moreover, GT clonal labels significantly clustered across phylogenies, suggesting high phylogenetic signal. Finally, we investigated evolutionary patterns in our clinical cohort. Here, we found independent acquisition of both nuclear- and MT-SNVs by distinct normal and malignant lineages, and characterized patterns of gene expression inheritance along recovered phylogenies. All in all, our data support the emerging role of MT-variants based lineage tracing approaches in cancer evolution studies, highlighting intrinsic limitations and potential of these endogenous markers for retrospective, phylo-phenotypic analysis. Citation Format: Andrea Cossa, Alberto Dalmasso, Andrea Tirelli, Zhan Yinxiu, Chiara Caprioli, Giulia Perticari, Pier Giuseppe Pelicci. Expressed mitochondrial variants capture patterns of tumor evolution at single-cell level abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts) ; 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84 (6Suppl): Abstract nr 6928.