A phase II trial of ixabepilone in Asian patients with advanced gastric cancer previously treated with fluoropyrimidine-based chemotherapy

Integrative omics analysis reveals effective stratification and potential prognosis markers of pan-gastrointestinal cancers

Gastrointestinal (GI) tract cancers are the most common malignant cancers with high mortality rate. Pan-cancer multi-omics data fusion provides a powerful strategy to examine commonalities and differences among various cancer types and benefits for the identification of pan-cancer drug targets. Herein, we conducted an integrative omics analysis on The Cancer Genome Atlas pan-GI samples including six carcinomas and stratified into 9 clusters, i.e. 5 single-type-dominant clusters and 4 mixed clusters, the clustering reveals the molecular features of different subtypes, other than the organ and cell-of-origin classifications. Especially the mixed clusters revealed the homogeneity of pan-GI cancers. We demonstrated that the prognosis differences among pan-GI subtypes based on multi-omics integration are more significant than clustering by single-omics. The potential prognostic markers for pan-GI stratification were identified by proportional hazards model, such as PSCA (for colorectal and stomach cancer) and PPP1CB (for liver and pancreatic cancer), which have prominent prognostic power supported by high concordance index.

•

Pan-cancer multi-omics strategy reveals homogeneity and heterogeneity of pan-GI cancers

•

Identify 9 iclusters with significantly different survival and molecular features

•

Potential prognostic markers have prominent power supported by concordance index

Prediction of cancer drugs by chemical-chemical interactions

Cancer, which is a leading cause of death worldwide, places a big burden on health-care system. In this study, an order-prediction model was built to predict a series of cancer drug indications based on chemical-chemical interactions. According to the confidence scores of their interactions, the order from the most likely cancer to the least one was obtained for each query drug. The 1(st) order prediction accuracy of the training dataset was 55.93%, evaluated by Jackknife test, while it was 55.56% and 59.09% on a validation test dataset and an independent test dataset, respectively. The proposed method outperformed a popular method based on molecular descriptors. Moreover, it was verified that some drugs were effective to the 'wrong' predicted indications, indicating that some 'wrong' drug indications were actually correct indications. Encouraged by the promising results, the method may become a useful tool to the prediction of drugs indications.