Prediction of the Outcome for Patients with Glioblastoma with lncRNA Expression Profiles

BACKGROUND: Progress in gene sequencing has paved the way for precise outcome prediction of the heterogeneous disease of glioblastoma. The aim was to assess the potential of utilizing the lncRNA expression profile for predicting glioblastoma patient survival. MATERIALS AND METHODS: Clinical and lncRNA expression data were downloaded from the public database of the cancer genome atlas. Differentially expressed lncRNAs between glioblastoma and normal brain tissue were screened by bioinformatics analysis. The samples were randomly separated into the training and testing sets. Univariate Cox regression, least absolute shrinkage, selection operator regression, and multivariate Cox regression were performed to develop the prediction model with the training set, which was presented as a forest plot. The performance of the model was validated by discrimination and calibration analysis in both the training and testing sets. Patient survival between model-predicted low- and high-risk subgroups was compared in both the training and testing sets. RESULTS: One thousand two hundred and fifty-five differentially expressed lncRNAs between glioblastoma and normal brain tissues were screened. After univariate Cox regression and the least absolute shrinkage and selection operator regression, a 12 lncRNA constituted prediction model was developed by multivariate Cox regression. Of the 12 lncRNAs, 4 lncRNAs were independent risk factors for patient survival. The areas under the receiver operating characteristic curves of the model for predicting 0.5-, 1-, 1.5-, and 2-year patient survival was 0.788, 0.824, 0.874, and 0.886, respectively in the training set and 0.723, 0.84, 0.816, and 0.773 in the testing set. The calibration curves of the prediction model fitted well. Significant survival disparity was observed between the model dichotomized low- and high-risk subgroups in both the training and testing set. CONCLUSIONS: LncRNA expression signature can predict glioblastoma patient survival, promising lncRNA-based survival prediction.