AIMS

Uveal melanoma (UM) is the most common and aggressive intraocular tumor in adults, and long-term survival of UM patients remains poor. Abnormal competitive endogenous RNA (ceRNA) networks promote the initiation and progression of many tumors and may thus serve as useful prognostic indicators. Here, we do a comprehensive analysis of long non-coding RNA (lncRNA)-microRNA (miRNA)-mRNA ceRNA networks as prognostic markers for UM.

MATERIALS AND METHODS

The Cancer Genome Atlas UM dataset was used to identify survival-related mRNA and lncRNA modules through weighted gene co-expression network analysis (WGCNA). Prognostic miRNAs were identified using univariate Cox proportional hazard regression. We then used Cox and least absolute shrinkage and selection operator regression to screen for prognostic hub mRNAs and establish a hub ceRNA network. A nomogram of five hub mRNAs was constructed and Kaplan-Meier survival analysis performed.

KEY FINDINGS

Six mRNA modules were constructed, two of which involved 1490 mRNAs that significantly correlated with survival. Among the three lncRNA modules constructed, one involved 199 survival-related lncRNAs. Five hub prognostic mRNAs were identified and a hub ceRNA network constructed, consisting of six lncRNAs, four miRNAs, and five mRNAs, with high prognostic value.

SIGNIFICANCE

We describe a hub ceRNA network of survival-associated lncRNAs, miRNAs, and mRNA that may underlie a critical post-translational regulatory mechanism determining UM aggression. These hub RNAs may be valuable prognostic markers and therapeutic targets in UM.