Identification of nine microRNAs as potential biomarkers for lung adenocarcinoma

Abstract

Lung cancer is a leading global cause of cancer-related death, and lung adenocarcinoma (LUAD) accounts for ~ 50% of lung cancer. Here, we screened for novel and specific biomarkers of LUAD by searching for differentially expressed mRNAs (DEmRNAs) and microRNAs (DEmiRNAs) in LUAD patient expression data within The Cancer Genome Atlas (TCGA). The identified optimal diagnostic miRNA biomarkers were used to establish classification models (including support vector machine, decision tree, and random forest) to distinguish between LUAD and adjacent tissues. We then predicted the targets of identified optimal diagnostic miRNA biomarkers, functionally annotated these target genes, and performed receiver operating characteristic curve analysis of the respective DEmiRNA biomarkers, their target DEmRNAs, and combinations of DEmiRNA biomarkers. We validated the expression of selected DEmiRNA biomarkers by quantitative real-time PCR (qRT-PCR). In all, we identified a total of 13 DEmiRNAs, 2301 DEmRNAs and 232 DEmiRNA-target DEmRNA pairs between LUAD and adjacent tissues and selected nine DEmiRNAs (hsa-mir-486-1, hsa-mir-486-2, hsa-mir-153, hsa-mir-210, hsa-mir-9-1, hsa-mir-9-2, hsa-mir-9-3, hsa-mir-577, and hsa-mir-4732) as optimal LUAD-specific biomarkers with great diagnostic value. The predicted targets of these nine DEmiRNAs were significantly enriched in transcriptional misregulation in cancer and central carbon metabolism. Our qRT-PCR results were generally consistent with our integrated analysis. In summary, our study identified nine DEmiRNAs that may serve as potential diagnostic biomarkers of LUAD. Functional annotation of their target DEmRNAs may provide information on their roles in LUAD.

Keywords: DEmiRNA; LUAD; TCGA; biomarker; lung adenocarcinoma; microRNA.

Figure 1

Hierarchical clustering analysis of DEmiRNAs and the top 100 DEmRNAs between LUAD and adjacent tissues. (A) DEmiRNAs; (B) the top 100 DEmRNAs. Rows and columns represent DEmiRNAs/DEmRNAs and tissue samples. The color scale represents the expression levels.

Figure 2

Identification of miRNA biomarkers for LUAD. (A) The variance rate of prediction accuracy for each DEmiRNA. (B) The variance rate of classification performance when increasing the number of the predictive miRNAs. (C) ROC results of combinations of nine miRNA biomarkers based on random forest classification model. (D) Hierarchical clustering analysis of the nine miRNA biomarkers between LUAD and adjacent tissues. Rows and columns represent DEmiRNAs and tissue samples. The color scale represents the expression levels.

Figure 3

LUAD‐specific DEmiRNA–DEmRNA interaction network. The ellipses and rhombuses represent the DEmRNAs and DEmiRNAs, respectively. Red and blue represent up‐ and down‐regulation, respectively.

Figure 4

The expression levels of DEmiRNAs and DEmRNAs between LUAD and adjacent tissues. The x‐axes represent normal (LUAD) and control (adjacent tissues) groups. The y‐axes represent gene expression levels. (A) hsa‐mir‐210, (B) hsa‐miR‐9‐1, (C) hsa‐miR‐9‐2, (D) hsa‐miR‐9‐3, (E) hsa‐mir‐153‐2, (F) hsa‐mir‐486‐1, (G) hsa‐mir‐486‐2, (H) hsa‐mir‐577, (I) hsa‐mir‐4732, (J) FLI1, (K) NTRK3, (L) SLC7A5, (M) TGFBR2, (N) ERG and (O) SIX1.

Figure 5

The ROC curves of selected DEmiRNAs and DEmRNAs between LUAD and adjacent tissues. The x‐axes show 1 − specificity (the proportion of false positives) and y‐axes show sensitivity (the proportion of true positives). (A) hsa‐mir‐210, (B) hsa‐miR‐9‐1, (C) hsa‐miR‐9‐2, (D) hsa‐miR‐9‐3, (E) hsa‐mir‐153‐2, (F) hsa‐mir‐486‐1, (G) hsa‐mir‐486‐2, (H) hsa‐mir‐577, (I) hsa‐mir‐4732, (J) FLI1, (K) NTRK3, (L) SLC7A5, (M) TGFBR2, (N) ERG and (O) SIX1.

Figure 6

qRT‐PCR results for DEmiRNAs between LUAD and adjacent tissues. Statistical significance was assessed by one‐way ANOVA. The error bars indicate SD. n = 3. The x‐axis and y‐axis represent DEmiRNAs and log fold change, respectively. **P < 0.01.