Identification of a microRNA signature associated with risk of distant metastasis in nasopharyngeal carcinoma

Abstract

Purpose: Despite significant improvement in locoregional control in the contemporary era of nasopharyngeal carcinoma (NPC) treatment, patients still suffer from a significant risk of distant metastasis (DM). Identifying those patients at risk of DM would aid in personalized treatment in the future. MicroRNAs (miRNAs) play many important roles in human cancers; hence, we proceeded to address the primary hypothesis that there is a miRNA expression signature capable of predicting DM for NPC patients.

Methods and results: The expression of 734 miRNAs was measured in 125 (Training) and 121 (Validation) clinically annotated NPC diagnostic biopsy samples. A 4-miRNA expression signature associated with risk of developing DM was identified by fitting a penalized Cox Proportion Hazard regression model to the Training data set (HR 8.25; p < 0.001), and subsequently validated in an independent Validation set (HR 3.2; p = 0.01). Pathway enrichment analysis indicated that the targets of miRNAs associated with DM appear to be converging on cell-cycle pathways.

Conclusions: This 4-miRNA signature adds to the prognostic value of the current "gold standard" of TNM staging. In-depth interrogation of these 4-miRNAs will provide important biological insights that could facilitate the discovery and development of novel molecularly targeted therapies to improve outcome for future NPC patients.

Figure 1

(A&B) Kaplan-Meier curves showing NPC patients dichotomized based on risk score in (A) the training cohort; and (B) the validation cohort. “High Risk” is defined as a RS ≥ the median in the training cohort, and “Low Risk” is defined as a RS < the median in the training cohort. (C) ROC AUCs across various time points demonstrating the ability of prognostic models generated using all possible combinations of 1, 2, 3, or 4 miRNAs from the 4-miRNA signature to predict distant relapse in NPC patients. RS, Risk Score; HR, Hazard Ratio; CI, Confidence Interval; ROC, receiver operating characteristic; AUC, Area Under the Curve.

Figure 2

(A) Kaplan-Meier curve showing patients assigned to risk groups based on the combined N-stage and 4-miRNA signature Risk Score. (B) ROC AUCs over time demonstrating the ability of various clinical factors and the 4-miRNA signature RS to predict distant relapse in NPC patients.

Figure 3

Kaplan-Meier curves showing distant relapse in NPC patients dichotomized based on miRNA risk score in advanced stage patients (Stage III/IV) treated with (A) RT alone or (B) combined CRT. ROC, receiver operating characteristic; AUC, Area Under the Curve; RT, radiotherapy; CRT, chemoradiotherapy; HR, Hazard Ratio; CI, Confidence Interval. RS, Risk Score; “High Risk” is defined as a RS ≥ the median in the training set, and “Low Risk” is defined as a RS < the median in the training set.

Figure 4. Venn diagram showing commonly and uniquely enriched pathways across three sets of miRNA-targets

Figure 5. Cell-cycle related genes targeted by miRNAs associated with risk of distant metastasis

(A) Chart showing Cell-cycle related genes targeted by at least 2 of the miRNAs queried. Black boxes indicate a miRNA-target relationship. (B) Pathway diagram modified from the “Cell-Cycle” KEGG pathway using Cytoscape (v3.1.1) showing validated targets of miRNAs from three prognostic groups. Note: miRNAs with no validated “cell-cylce” targets (miR-154–5p from the 4-miRNA signature and miR-30e-5p from The 5-miRNA signature) were omitted from the figure. See legend for further details.