Identification of novel long non-coding RNA biomarkers for prognosis prediction of papillary thyroid cancer

Papillary thyroid carcinoma (PTC) is the most frequent type of malignant thyroid tumor. Several lncRNA signatures have been established for prognosis prediction in some cancers. However, the prognostic value of lncRNAs has not been investigated in PTC yet. In this study, we performed genome-wide analysis of lncRNA expression profiles in a large cohort of PTC patients from The Cancer Genome Atlas and identified 111 differentially expressed lncRNAs between tumor and non-tumor samples and between recurrent and recurrence-free samples. From the 111 differentially expressed lncRNAs, four independent lncRNA biomarkers associated with prognosis were identified and were integrated into a four-lncRNA signature which classified the patients of training dataset into the high-risk group and low-risk group with significantly different overall survival (p=0.016, log-rank test). The prognostic value of the four-lncRNA signature was validated in the independent testing dataset. Multivariate analysis indicated that the four-lncRNA signature was an independent prognostic predictor. Moreover, identified four lncRNA biomarkers demonstrates good performance in predicting disease recurrence with AUC of 0.833 using leave one out cross-validation. Our study not only highlighted the potential role for lncRNAs to improve clinical prognosis prediction in patients with PTC and but also provided alternative biomarkers and therapeutic targets for PTC patients.