Precision genomics for prostate cancer patient stratification

Abstract

Introduction Several unmet needs currently persist in advanced prostate cancers. Despite recent discoveries, such as identification of deficient DNA repair pathways as a key driver of prostate tumour development, relatively few precision medicine approaches are available. Robust genomics is critical to characterise and identify such subsets, but there are key limitations that must be addressed such as tumour purity and heterogeneity. Additionally, for men without a genomic stratification, treatment options may be limited to aggressive chemotherapies such as taxanes, which currently have no genomic biomarkers to improve patient stratification. Results In this study, I first show how key DNA-repair subtypes of prostate cancers (35% of cases) can bed characterise using copy-number and microsatellite analysis of targeted panel sequencing data. Both data types were well correlated with orthogonal methods. I found that tumour heterogeneity (observable at the single cell-level via pathological analysis) limited the detection of copy-number events from bulk sequencing. I went on to explore tumour purity estimation in cell-free DNA sequencing of mCRPCs treated with taxanes, and found that baseline purity measurements were strongly associated (>0.001) with survival in both univariable and multivariable analyses. Additionally, I found significant longitudinal shifts (p>0.001) in tumour purity between responders and non-responders to therapy. I sought to test specific genes and genomic loci in this cohort of taxane-treated samples, and found that copy-changes of several genes including members of the tubulin gene family had associations with changes in survival. Additionally, I identified several genomic loci with strong associations with drug response status. As these studies are limited by tumour heterogeneity, I analysed a cohort of single cells, which allowed for a high-resolution examination of tumour sub-clones. These data displayed variable genomic heterogeneity, with clinically relevant alterations identified at the sub-clonal level. Importantly, I found changes in sub-clonal proportions could be observed longitudinally at the copy number level. Conclusions In summary, I show here methods for characterising the 30-40% of mCRPC that may respond to DNA-repair targeting therapies targeted sequencing, before going on to illustrate similar approaches, and potential future biomarkers, for taxane chemotherapies. Future validation of these will be key in improving taxane response rates. Additionally, while tumour purity correction is a limiting factor for tumour NGS, it also presents as a potentially useful clinical marker of tumour burden. Key to genomic studies of mCRPC is sample selection, and I showed that single-cell analyses and micro-dissection can be employed to bypass issues of tumour heterogeneity and purity by directly revealing clinically relevant alterations present at levels undetectable by traditional bulk sequencing methods. In the future, adoption of these methods will improve patient stratification and monitoring of cancer progression.