The expression of seven key genes can predict distant metastasis of colorectal cancer to the liver or lung.

Abstract

It is unclear how primary colorectal cancer (CRC) cells select to metastasize to the liver or lungs, the most frequent distant metastasis of CRC. We aimed to identify the key genes and pathways that may predict the distant metastasis of CRC to these sites. Three gene expression array datasets from the Gene Expression Omnibus were analyzed. Protein-protein network analyses, best subsets regressions and backward stepwise regression analyses were used to screen the key genes and their expressions were used to construct a predictive logistic regression model. Expression data from local clinical samples were used as a validation dataset. The receiver operating characteristic (ROC) curve was used to test the performance of the predictive model. In total, 59 differentially expressed genes (DEG) related to liver-metastasis, 90 related to lung metastasis and 45 related to both liver and lung metastasis were identified. The KEGG pathways and gene oncology (GO) terms that were enriched in liver and lung metastasis were identified. A predictive logistic regression model consisted of SPARC, COL1A2, MMP9, COL11A1, COL3A1, CXCL12 and THBS2 was established. The area under the ROC curve reached 0.839 in predicting liver and lung metastasis, using our clinical samples as the validation dataset. Seven key genes capable of predicting liver and lung metastasis of colorectal cancer were identified, which provides clues for exploring the mechanism of selecting target organs during the metastatic process in CRC and inspires the researches for new potential targets for therapy to inhibit metastasis.