Identifying Reproducible Molecular Biomarkers For Gastric Cancer Metastasis With The Aid Of Recurrence Information

Metastasis is the primary cause of death for most gastric cancer patients. To precisely diagnose metastasis state is important for tailoring treatment strategies for individual patients. However, there is no robust or efficient biomarker for predicting the gastric metastasis so far. The routinely employed radiological and pathologic tests for tumor metastasis are lack of accuracy, with considerable high false positive and false negative rates. The misjudged samples retard the identification of reproducible metastasis-related molecular biomarkers for gastric cancer. In this research, using three microarray datasets, we firstly showed that differentially expressed genes(DEGs)between metastatic tissue samples and non-metastatic tissue samples could hardly be reproducibly detected with a proper statistical control when the metastatic and non-metastatic samples were defined by TNM stage alone. We put forward a hypothesis:the patients who do not diagnose as metastasis cases while suffer from recurrence after curative resection are the ones who have developed micrometastasis. Based on this hypothesis, we reclassified all the “non-metastatic” samples of patients, defined according to TNM stage, as metastatic samples whenever the patients experienced tumor recurrence during follow-up after tumor resection. By taking into account the recurrence information of samples accompanied with the TNM staging, we were able to found clear and reproducible DEGs between the reclassified metastatic samples and non-metastatic tissue samples with a proper statistical control. Functional enrichment analysis was done for the concordant DEGs, which enriched in a wide range of gastric cancer metastasis-related pathways, such as ECM-receptor interaction, focal adhesion,cell cycle and so on. Finally, after reclassifying gastric cancer samples with multi-omic data deposited in the TCGA according to the recurrence information, we were able found significant DNA methylation alterations distinguishing metastatic tissues and non-metastatic tissues of gastric cancer. Especially, the results showed that genes withhypermethylated CpG sites within their promoters exhibited significantly concordant under-expression in the metastatic samples compared with the non-metastatic samples,the genes with hypomethylated CpG sites within their promoters exhibited weakly but significantly concordant over-expression in the metastatic samples, compared with the non-metastatic samples. These results not only confirmed the DEGs we obtained were reliable, but also show the essential role of methylation in the gastric tumour cell migration. Some of the genes with concordant DNA methylation alternations and transcriptional changes played important roles in the tumour metastasis, such as IFNG in the regulation of autophagy pathway, which was both hypermethylated and down-regulated in metastasis tissues, might reduce cell epithelial apoptosis and decrease cell proliferation via autophagy; the COL4A3 annotated in the ECM-receptor interaction pathway was both hypomethylated and up-regulated in metastasis tissues,which might play a role in tumour metastasis. Our analyses suggested that the follow-up recurrence information for patients should be employed in the research of tumour metastasis in order to decrease the confounding effects of false non-metastatic samples with undetected micrometastases.