Study on identification of diagnostic biomarkers in serum for papillary thyroid cancer in different iodine nutrition regions.

Abstract

Background: At present, there is a lack of efficient biomarkers for the diagnosis of thyroid cancer, and the influence of natural factors such as high iodine exposure on the expression of biomarkers remains unclear.

Methods: Serum samples from papillary thyroid cancer (PTC) and non-cancer controls matched 1:1 in different iodine nutritional regions were analyzed metabolomically using an ultra-high performance liquid chromatography-Orbitrap Exploris mass spectrometry (UHPLC-OE-MS) platform. Then the data were randomly divided into training and test sets in a 1:1 ratio according to the different iodine nutritional regions and different PTC status. In the training set, differential metabolites were selected by multivariate statistical analysis methods, and the prediction models were then built using Random forest (RF), Gradient boosting machine (GBM), and Support vector machine (SVM) models. At last, their diagnostic effects were examined in the test set.

Results: PTCs were significantly separated from non-cancer samples, and a total of 37 differentially expressed metabolites were selected. The results of pathway analysis showed that the PTC-related differential metabolites were mainly involved in the sphingolipid metabolism and glycerophosphate metabolism. The prediction models constructed by the 6 screened potential biomarkers could all better identify PTCs in the test set. The metabolomic fingerprinting between PTC and non-cancer groups in different water iodine regions did not show significant disturbance. However, high iodine exposure would effect on the expression of six metabolites, reflecting in a significantly different diagnostic efficacy in different water iodine regions.

Conclusion: Serum metabolites have potential value as biomarkers of PTC, and iodine status affects the expression and even diagnostic levels of certain serum metabolites.