Purpose: With the increasing incidence of thyroid cancer (TC), the prognostic risk assessment of thyroid cancer has been becoming more and more important. The aim of this study was to screen TC-related biomarkers and identify key multi-long non coding RNA (lncRNA) signature for prognostic risk assessment of papillary TC.
Material and methods: The lncRNAs differentially expressed between TC tissue and adjacent normal tissue was identified by R language. Bioinformatics analysis was applied to screen the lncRNAs significantly associated with prognosis in TC patients and build the multi-lncRNA signature. The lncRNAs were annotated by co-expression and enrichment analysis to demonstrate the underlying mechanism of their effect on prognosis.
Results: 285 up-regulated and 174 down-regulated differently expressed lncRNAs were identified. Based on seven signature lncRNAs (AL591846.2, AC253536.3, AC004112.1, LINC00900, AC008555.1, TNRC6C-AS1, LINC01736) a prognostic risk assessment model was built. The model can segregate the patients into the high-risk and low-risk groups (P value <0.0001, CI: 0.02∼0.14). ROC analysis revealed that the area under the curve reached 0.86, indicating that this model had an excellent sensitivity and specificity. Also, the model could act as an independent prognostic indication (HR = 2.90, P value = 0.0094 with multivariate analysis). Annotation results further supported and enriched our understanding of the seven signature lncRNAs. Importantly, expression levels of three of the seven lncRNAs were confirmed in Gene Expression Omnibus (GEO) data.
Conclusions: This study has provided a promising method for the prognostic risk assessment in patients with TC.
Keywords: Bioinformatics; Biomarker; Long non-coding RNA (lncRNA); Prognostic risk; Thyroid cancer.
Copyright © 2021. Published by Elsevier B.V.