| Purpose: Accurate diagnosis of metastasis is an important basis for mapping out followup treatment plans for patients undergoing gastrectomy.However,the clinical imaging techniques and pathological examinations for tumor metastasis have high false positive and/or false negative rates.Therefore,the metastasis-related molecular biomarkers for gastric cancer is difficult to reproduce in independent datasets.On the other hand,gastric cancer patients with high risk of recurrence after curative surgery or non-radical surgery are usually accompanied by 5-fluorouracil(5-FU)-based chemotherapy to prolong the survival time of patients and improve their quality of life.But only a few patients can benefit from it.At present,some studies have attempted to identify 5-FU resistance predictive signature by using patients with 5-FU-based adjuvant chemotherapy after surgery,but these studies have mixed some of patients who have low-risk of recurrence without chemotherapy,which has a certain impact on the identification of the 5-FU resistance predictive signature.Therefore,this study aims to identify postoperative recurrence risk signature and 5-FU resistance predictive signature.Methods: We first developed and validated a postsurgery recurrence risk(micrometastasis)signature consisting of 17 gene pairs based on the within-sample relative expression orderings(REOs)of genes.By combining this signature with TNM staging system,the double-identified method is used to reclassify the metastasis status of samples.Then we identified the differentially expressed genes(DEGs)between the reclassified metastasis(including micrometastases)and non-metastasis samples,computed the reproducibility of DEGs in independent datasets and performed functional enrichment analysis of reproducible DEGs in different datasets.Subsequently,the signature was applied to 5-FU-based adjuvant chemotherapy patients.Filtering some of chemotherapyirrelevant low relapse risk patients,and using the patients with high-risk of recurrence(micrometastasis)to identify a 5-FU resistance predictive signature consisting of 37 gene pairs,and verified in two independent datasets.Result: The results showed that,compared with TNM staging system alone,more and more significant DEGs related to metastasis were found after reclassification with TNM staging system and 17-GPS in the three datasets,and the concordance score(reproducibility)of the overlapped DEGs between every two datasets was higher than 97%,which was highly unlikely to happen by chance(binomial test,all p<2.20E-16).Functional enrichment analysis(hypergeometric distribution model,FDR<0.05)showed that the reproducible up-regulated genes in metastasis group were significantly enriched in the metastasis-related pathways such as PI3K-Akt signaling pathway,Hippo signaling pathway,ECM receptor interaction and cell adhesion.It was confirmed from the side that the signature can effectively recognize the reproducible metastatic molecular characteristics of gastric cancer.The 5-FU resistance predictive signature can effectively identify the gastric cancer patients who can benefit from 5-FU-based chemotherapy in three datasets.Conclusion: The recurrence risk signature for early or middle stage gastric cancer identified in this study can effectively identify patients who underwent resection and had a low risk of postoperative recurrence.Based on the recurrence risk signature,the coupled 5-FU resistance predictive signature can effectively determine whether patients with high-risk of recurrence can benefit from 5-FU-based chemotherapy.Providing a basis for different patients to take appropriate postoperative treatment,so as to avoid postoperative patients with low risk of recurrence and patients who cannot benefit from chemotherapy to receive over-treatment and side effects.And it is also possible to choose other adjuvant treatments as soon as possible for patients who cannot benefit from 5-FU-based chemotherapy. |
PDF Full Text Request