Abstract 2135: Characterization of robust and prognostic molecular subtypes in prostate cancer

Abstract

Abstract Introduction: Multiple biomarkers and molecular subtypes have been proposed in prostate cancer (PCa) with limited overlap, in part due to the heterogeneity of the disease and issues with data scaling. By applying Almac Diagnostic's unique software driven tool, claraT, cancer gene expression datasets can be characterized according to six of the well-established Hallmarks of Cancer using published signatures, proprietary assays and single gene targets. This retrospective study aimed to identify robust molecular subtypes of PCa that are prognostic and validated by applying claraT to multiple gene expression datasets. Experimental procedures: claraT was applied to three primary PCa gene expression datasets, TCGA (n = 331, Illumina HiSeq), Walker et al Discovery (n = 91, Almac Prostate DSA) and Walker et al Validation (n = 322, Almac Prostate DSA). Six sets of signature scores were generated, each representing a different hallmark and consensus clustering was performed using the signature scores. Clinical relevance was assessed using metastatic-free survival (MFS) and disease-free survival (DFS) as end-points. Using statistical tests, association of the identified subtypes with variables was investigated. Results: Of the six hallmarks investigated, robust and stable molecular subtypes were established using the Genome-Instability (GI) signature set. Four subtypes, two with good prognosis and two with poor prognosis, were characterized in the Walker Validation (MFS log-rank, p=6.798e-10), TCGA (DFS log-rank, p=0.0012) and Walker Discovery (chi-square - metastases, p=0.004). Subtypes were significantly associated with Gleason, biochemical recurrence, and Almac's Prostate Dx assay (chi-square, p<0.00001). Clustering analysis revealed the poor prognosis subtypes to have high GI signature expression e.g. chromosome instability and homologous recombination deficiency signatures. The median number of copy number alterations (CNAs) amongst the subtypes was higher in the poor prognosis subtypes (Wilcoxin, p=1.0363e-18) thus validating the genome-unstable phenotype association with the poor prognosis subtypes. Conclusion: Of the six hallmarks investigated, Genome-Instability was the most robust and prognostic. Molecular subtypes with distinct prognoses have been identified and validated. These poor prognosis subtypes are characterized by an active genome instability biology and high levels of CNAs. Therefore, these poor prognosis subtypes of PCa, with a dominant genome instability biology, may represent a larger target subgroup of PCa's which may benefit from DNA repair targeted therapies such as PARP inhibitors. Citation Format: Cathal McKinney, Nuala McCabe, Paul Harkin, Richard Kennedy, Jaine Blayney. Characterization of robust and prognostic molecular subtypes in prostate cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2135.