Abstract / Description
BACKGROUND: Cytopathology cannot be used to reliably distinguish follicular thyroid adenoma (FTA) from follicular thyroid carcinoma (FTC), the second most common form of thyroid cancer, because they exhibit nearly identical cellular morphology. Given the challenges in diagnosis and treatment, this study aims to identify the mechanisms underlying FTC. METHODS: Using parallel reaction monitoring-mass spectrometry (PRM-MS) assays, we identified and quantified 94 differentially expressed protein candidates from a retrospective cohort of 1085 FTC and FTA tissue samples from 18 clinical centers. Of these targeted proteins, those with the potential for distinguishing FTC from FTA were prioritized using machine learning. Co-immunoprecipitation (co-IP) and immunofluorescence co-localization assays, as well as gene interference, overexpression, and immunohistochemistry (IHC) experiments, were used to investigate the interactions and cellular functions of selected proteins. RESULTS: Using machine learning models and feature selection methods, 30 of the 94 candidates were prioritized as key proteins. Co-IP and immunofluorescence co-localization assays using FTC cell lines revealed interactions among insulin-like growth factor 2 receptor (IGF2R), major vault protein (MVP), histone deacetylase 1 (HDAC1), and histone H1.5 (H1-5). Gene interference and overexpression experiments in FTC-133 cells confirmed the promotional role of these proteins in cell proliferation. IHC assays of patient samples further confirmed elevated expression of these four proteins in FTC compared with that in FTA. CONCLUSIONS: Our findings underscore the utility of advanced proteomic techniques in elucidating the molecular underpinnings of FTC, highlighting the potential significance of IGF2R, MVP, HDAC1, and H1-5 in FTC progression, and providing a foundation for the exploration of targeted therapies.