| 研究実績の概要 |
Triple negative (TN) breast cancer can only be labeled by clinical markers, there is an urgent need to classify the similarities of TN cases consistently with clinical phenotypes to distinguish a novel treatment regiment for these patients. This study can be summarized into 3 points: 1) classification of breast cancer patients regarding mutational viewpoints to be proposed by clustering, 2) discrimination rules regarding the clinical phenotypes derived from classification algorithms to be shown, and 3) relevance between proposed categories and classification rules to be explored in biological or medical literature. We have unraveled the link of exonic somatic mutations and their combinations to clinical phenotypes, particularly those in TN patients into other clinical phenotypes. The pathway-wise features help associate the downstream effect of the somatic mutations in the development of the breast cancer. We identify the gene-wise mutation demonstrating the combination of TP53 & PIK3CA as significant, by computational methods alone, in 4.3% (9/208) of the general population of BC patients, exhibiting the TN patients with ER+/PR+ expressive phenotype. Thus, we believe our unique machine learning-based approach which integrates all the positional, gene and pathway information is a first stepping stone to help with therapy selection and unravel the improper downstream signaling which confers to antibiotic resistance.
|
