Revealing the clinical potential of cancer whole-genome data: A retrospective analysis of a breast cancer cohort in England linked with mortality statistics.

Abstract

Abstract: Background: Breast Cancer (BC) is the most frequently diagnosed cancer in women. Survival is considered generally favourable, yet some patients remain at risk of early death. Here, our primary objective is to assess whether comprehensive whole genome sequencing (WGS) linked to mortality data can add prognostic value to existing clinical measures and/or highlight potential responders to targeted-therapeutics. Methods: We present an integrative analysis of 2,445 WGS BCs (any grade/molecular- subtype) sourced from an all-comers cohort of patients recruited via England’s National Health Service. We linked 90% of cases with clinical and/or mortality data. We performed analysis on this high-depth (mean tumour ~90X, matched normal ~40X) WGS cohort, seeking somatic/constitutional drivers, mutational signatures, and compound algorithmic scores reporting homology-directed repair deficiency (HRD) and mismatch repair deficiency amongst others. Associations with cancer-specific mortality were evaluated for ER-positive HER2- negative (ER+HER2-) BCs with ~5 years follow-up. Data from 1,803 additional BCs were used to validate various findings. Findings: Genomic characteristics with immediate personalised medicine potential were observed in 27% of this population-representative cohort, including features reporting HRD (12% total BC, 6% ER+HER2- BC), highly-individualised driver events, mutations underpinning endocrine-therapy resistance, and mutational signatures indicating therapeutic vulnerabilities. 15% had translational research potential (e.g., base excision repair- strong prognosticators in ER+HER2- BC. Burden of structural variation (HR 3·9, CI(95):2·4- 6·2, p<0·0001), high levels of APOBEC signatures (HR 2·5, CI(95):1·6-4·1, p<0·0001), and TP53 drivers (HR 3·9, CI(95):2·4-6·2, p<0·0001) were independently prognostic of customary clinical measures (age/stage/grade). We thus developed an ER+HER2 prognosticator capable of detecting patients who require increased intervention and candidates for de-escalation, validating the framework using an independent Sweden Cancerome Analysis-Breast (SCAN-B) cohort. Interpretation BC genomes are rich with predictive/prognostic value today. We suggest a blueprint for effective clinical WGS application. CRUK (C60100/A23916;C60100/A25274;CGCATF-2021/100013). compromised and non-homologous end-joining dependency). WGS additionally uncovered Funding: NIHR-NIHR301627, BRC-1215-20014, BCRF-23-197, Gray Foundation, and