Abstract Introduction: Breast cancer is the most common female malignancy in Taiwan and treatment outcomes have improved enormously with early detection and advancements in adjuvant therapy. Unfortunately, a subset of early stage breast cancer still suffers from recurrence, metastasis, or even cancer-specific death despite treatment improvement, and conventional pathological factors fail to provide sufficient explanatory power accounting for the observed prognostic discrepancy. The aim of the project is to take advantage of digital RNA counting for gene expression and targeted sequencing of actionable mutations to develop a predictive model for operable breast cancers initially treated with curative intention, and the risk of recurrence, distant metastasis, as well as breast cancer-specific mortality will be speculated. Methods: Two approaches are applied for candidate genes yield; we have identified concurrent genes from the coherent patterns between genomic and transcriptional profiles, and the synthesized signatures were prognostic. In addition, concordant genes from leading edge analysis of published gene expression signatures were retrieved, and the consensus of leading edge subsets could stratify breast cancer patients into groups with distinct survival patterns. With bioinformatics tools, the harvested genes will be used to build a risk stratification model, and the residual risk after surgery and adjuvant therapy will be evaluated from archived formalin-fixed paraffin embedded (FFPE) cancer tissues, augmenting the clinical significance of the proposed signature. And finally, actionable mutations will be identified for relapsing patients through targeted sequencing. Results: There were 1584 extended concurrent genes with the threshold MAD (medium absolute deviation) of 0.9, constituting 15.89% of common genes across copy number variation and gene expression arrays. We also constructed breast cancer risk prediction using concordance of genes from leading edge analysis of various prognostic signatures; enrolled signatures were split into either the up-regulated or down-regulated gene sets within each study, and leading edge analysis of filtered up-/down-regulated gene sets from all training data sets were performed, followed by the PLS (partial least square) risk predictive model composed of consensus of leading edge subsets among all assayed subjects. The 145-gene BCeC Sig platform designated for the current study has been established, and was prognostic in 96 Taiwanese breast cancers (log-rank test: P<0.01). Among 37 breast cancers categorized into the high-risk group, there were 383 SNPs, 23 deletes, and 11 insertions. Variations per breast cancer were between 91 and 162 (average: 120). When collapsed into genes, the most altered genes were: BRCA1/2 (30 variants), FANCA(27), NOTCH(21), ALK, AR (20), ATM, BSG(14), FGFR, MET(12), JAK2, PDFGR, RET(11), CSF1R, DAB2I, and TP53(10). Discussion and conclusions: Archived pathological specimens provide an invaluable source to validate the residual risk prediction model for breast cancer patients managed with contemporary multi-modality treatment including surgery, chemo-, endocrine, and targeted therapy. Degraded nucleic acid from FFPE samples could be tackled by the latest NanoString digital RNA counting. This breast cancer residual risk model, composed of concurrent genes and consensual leading edge subset of published gene expression signatures, is believed to provide clinical applicability and substantial benefits for Taiwanese breast cancer patients in terms of personalized medicine Citation Format: Chi-Cheng Huang, Ching-Shui Huang. Residual risk stratification of Taiwanese breast cancer following curative therapies with digital RNA counting and actionable mutations sequencing [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P4-07-12.
Read full abstract