Backgroud: Tumour recurrence and metastasis lead to poor prognosis in colon cancer (COAD). Therefore We aimed to identify a lncRNA signature through an integrative analysis of copy number variation, mutation and transcriptome data to predict prognosis and explore its internal mechanism.
Methods: The lncRNA expression profile were collected from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO). TCGA data was randomly divided 3:1 into training and testing cohort. In the training, we performed integrated analyses of three candidate lncRNA sets that correlated with prognosis, copy number variations and mutations to establish a signature through Cox regression analysis. The robustness was determined in the testing and GEO.
Results: An 11-lncRNA signature that was significantly associated with prognosis was constructed in the training (P<0.0001, HR=2.014) , And this signature was validated in the testing (P=0.0019, HR=3.374) and GSE17536 (P=0.0076, HR=1.864). The signature is significantly related to MSI status and clinical prognostic factors. The prognostic-related risk scores were significantly excellent than the other five models have been reported. Furthermore, GSEA suggested that the signature was involved in COAD development and metastasis-related pathways.
Conclusions: We identified an signature has strong robustness and can stably predict the prognosis of COAD in different platforms and may be implicated in COAD pathogenesis and metastasis and applied clinically as a prognostic marker.