Glycosylation Signatures of Thyroglobulin in Papillary Thyroid Carcinoma

Thyroglobulin (Tg) is a clinically established biomarker for thyroid cancer; however, its diagnostic specificity remains limited due to confounding benign conditions. This study investigated changes in the site-specific N-glycosylation profiles of thyroglobulin protein derived from papillary thyroid cancer and adjacent healthy tissues. Using a bottom-up mass spectrometry-based glycoproteomic approach, we analyzed site-specific glycan compositions in retrospective patient tissue samples. Nine distinct N-glycosylation sites were identified and quantified in this study. Statistical analysis revealed significant differences in the site-specific glycosylation profiles. The glycan compositions HexNAc(3)Hex(5) and HexNAc(3)Hex(6) at the Asn1365 site exhibited a discriminatory potential. Dimensionality reduction and hierarchical clustering based on the top four glycopeptide variants showed observable grouping of tumor versus normal samples. Furthermore, supervised machine learning models were used to evaluate these features, with a neural network achieving the highest diagnostic performance (area under the curve [AUC] = 0.83). Feature importance analysis indicated that Asn1365-HexNAc(3)Hex(6) was the strongest contributor to the predictive power of the model. The observed alterations in specific glycan compositions, combined with machine learning evaluations, may offer improved diagnostic specificity and provide promising avenues for personalized thyroid cancer management.