Abstract 5174: Lymph node dissemination gene-miRNA signature as a prognosis marker in human thyroid cancer

Abstract

Abstract Most of the thyroid cancers (TCs) have excellent prognosis if detected early and treated appropriately. 50% of the deaths are due to aggressive variants of metastatic papillary and follicular TCs which have higher risk of recurrence, shortened disease free survival and death. The aim of this study is to identify a gene and/or miRNA signature associated with the lymph node dissemination as a predictive marker for early detection and intervention of TCs and spare patients with non-aggressive TCs from unnecessary surgery and non-surgical treatment. The bioinformatics analysis suggested a gene/miRNA signature of 25 genes (8 up regulated in N1 samples and 17 down regulated) and 2 miRNAs (1 up and 1 down regulated), based on machine learning regression algorithms classifying the samples between their N0/N1 stage with the highest predictive power estimated by the AUC (Area Under Curve) of a ROC (Receiver Operator Curve) analysis of 0.83 for the gene signature and 0.68 for the miRNA signature. A score, reflecting the expression of the gene/miRNA signatures differentiates the N0 and N1 samples with a p-value inferior to 2x10-16. Mann-Whitney tests were performed to estimate the score's correlation with the clinical parameters. The gene signature seems to be dependent of the cancer type (papillary versus follicular) and independent of the BRAF mutation status and TNM stages. A combination of a correlation analysis and predictive algorithms have been used to identify potential interactions between the 25 genes of the signature and the 9 differentially expressed miRNAs. Two clusters of genes and two clusters of miRNAs have been identified based on their potential interaction and expression. CLCNKA, MORN5 and AGXT2L1 genes are downregulated in the N1 samples and are potentially targeted by up regulated miR-526b-5p, miR-520a-3p and miR-206. At the opposite, the DSG3, KRT5, FOXR2, PRKAG3 and VGF genes are up regulated in the N1 samples, are potentially targeted by down regulated miR-873-5p, miR-184, miR-483-3p and miR-483-5p. The gene/miRNA signatures are currently validated. The predictive power of the signature scores has been estimated in other RNAseq datasets. While the q-PCR analysis of thyroid cancer specimen has been designed following a laser capture microdissection. The predictive score will be expected to estimate the probability of disease dissemination to the lymph nodes, computing the expression scores and the probability of the gene/miRNA signature to correctly classify the samples. Moreover, the genes/miRNA clusters interactions will be evaluated by dual luciferase assay and their functions in thyroid cancer dissemination will be assessed. The preliminary results indicate that the gene/miRNA signature could be a predictive marker for the classification of TCs based on their lymph node dissemination. Targeting this interactome could lead to a new therapeutic strategy for the aggressive thyroid carcinoma. Citation Format: Emmanuelle Ml Ruiz, Chih-Hong Wang, Tianhua Niu, Mohamed Hassan, Emad Kandil. Lymph node dissemination gene-miRNA signature as a prognosis marker in human thyroid cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5174.