标题：Prognostic value of a hypoxia-related microRNA signature in patients with colorectal cancer
作者：Yang, Yongmei; Qu, Ailin; Wu, Qi; Zhang, Xin; Wang, Lili; Li, Chen; Dong, Zhaogang; Du, Lutao; Wang, Chuanxin
作者机构：[Yang, Yongmei; Qu, Ailin; Zhang, Xin; Wang, Lili; Dong, Zhaogang] Shandong Univ, Qilu Hosp, Dept Clin Lab, Jinan 250012, Shandong, Peoples R China.; 更多
通讯作者地址：[Wang, CX]Shandong Univ, Dept Clin Lab, Hosp 2, Jinan 250033, Shandong, Peoples R China.
关键词：colorectal cancer; hypoxia microRNA signature; prognosis; prediction;; nomogram
摘要：Hypoxia has been particularly associated with poor prognosis in cancer patients. Recent studies have suggested that hypoxia-related miRNAs play a critical role in various cancers, including colorectal cancer (CRC). In the present study, we found 52 differentially expressed miRNAs in HT-29 cells under hypoxic conditions versus normoxic conditions by analyzing the profiles of miRNAs. Using Cox model, we developed a hypoxia-related miRNA signature consisting of four miRNAs, which could successfully discriminate high-risk patients in the Cancer Genome Atlas (TCGA) training cohort (n=381). The prognostic value of this signature was further confirmed in the TCGA testing cohort (n=190) and an independent validation cohort composed of formalin-fixed paraffin-embedded clinical CRC samples (n=220), respectively. Multivariable Cox regression and stratified survival analysis revealed this signature was an independent prognostic factor for CRC patients. Time-dependent receiver operating characteristic (ROC) analysis showed that the area under the curve (AUC) of this signature was significantly larger than that of any other clinical risk factors or single miRNA alone. A nomogram was constructed for clinical use, which incorporated both the miRNA signature and clinical risk factors and performed well in the calibration plots. Collectively, this novel hypoxia-related miRNA signature was an independent prognostic factor, and it possessed a stronger predictive power in identifying high-risk CRC patients than currently used clinicopathological features.