Construction and validation of a novel prognostic signature of microRNAs in lung adenocarcinoma

Wanzhen Li; Shiqing Liu; Shihong Su; Yang Chen; Gengyun Sun

doi:10.7717/peerj.10470

Construction and validation of a novel prognostic signature of microRNAs in lung adenocarcinoma

PeerJ. 2021 Jan 8:9:e10470. doi: 10.7717/peerj.10470. eCollection 2021.

Authors

Wanzhen Li^#¹, Shiqing Liu^#^{2

3

4}, Shihong Su¹, Yang Chen¹, Gengyun Sun¹

Affiliations

¹ Department of Respiratory and Critical Care Medicine, The First Affiliated Hospital of Anhui Medical University, Hefei, China.
² Department of Respiratory Medicine, National Key Clinical Specialty, Branch of National Clinical Research Center for Respiratory Disease, Xiangya Hospital, Central South University, Changsha, China.
³ Key cite of National Clinical Research Center for Respiratory Disease, Xiangya Hospital, Central South University, Changsha, China.
⁴ Xiangya Lung Cancer Center, Xiangya Hospital, Central South University, Changsha, China.

^# Contributed equally.

Abstract

MicroRNA (miRNA, miR) has been reported to be highly implicated in a wide range of biological processes in lung cancer (LC), and identification of differentially expressed miRNAs between normal and LC samples has been widely used in the discovery of prognostic factors for overall survival (OS) and response to therapy. The present study was designed to develop and evaluate a miRNA-based signature with prognostic value for the OS of lung adenocarcinoma (LUAD), a common histologic subtype of LC. In brief, the miRNA expression profiles and clinicopathological factors of 499 LUAD patients were collected from The Cancer Genome Atlas (TCGA) database. Kaplan-Meier (K-M) survival analysis showed significant correlations between differentially expressed miRNAs and LUAD survival outcomes. Afterward, 1,000 resample LUAD training matrices based on the training set was applied to identify the potential prognostic miRNAs. The least absolute shrinkage and selection operator (LASSO) cox regression analysis was used to constructed a six-miRNA based prognostic signature for LUAD patients. Samples with different risk scores displayed distinct OS in K-M analysis, indicating considerable predictive accuracy of this signature in both training and validation sets. Furthermore, time-dependent receiver operating characteristic (ROC) analysis demonstrated the nomogram achieved higher predictive accuracy than any other clinical variables after incorporating the clinical information (age, sex, stage, and recurrence). In the stratification analysis, the prognostic value of this classifier in LUAD patients was validated to be independent of other clinicopathological variables, such as age, gender, tumor recurrence, and early stage. Gene set annotation analyses were also conducted through the Hallmark gene set and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, indicating target genes of the six miRNAs were positively related to various molecular pathways of cancer, such as hallmark UV response, Wnt signaling pathway and mTOR signaling pathway. In addition, fresh cancer tissue samples and matched adjacent tissue samples from 12 LUAD patients were collected to verify the expression of miR-582's target genes in the model, further revealing the potential relationship between SOX9, RASA1, CEP55, MAP4K4 and LUAD tumorigenesis, and validating the predictive value of the model. Taken together, the present study identified a robust signature for the OS prediction of LUAD patients, which could potentially aid in the individualized selection of therapeutic approaches for LUAD patients.

Keywords: TCGA; LASSO bagging cox regression test; Lung adenocarcinoma; Prognosis.

Grants and funding

This work was supported by the National Key Clinical Specialist Construction Programs of China (N3101005005025), the National Multidisciplinary Cooperative Diagnosis and Treatment Capacity Building Project for Major Diseases (Lung Cancer), and the Xiangya clinical big data project of Central South University (Clinical big data project of lung cancer). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.