ObjectiveThis article aims to summarize the historical evolution of thyroid cancer molecular classification and explore the establishment of a precise classification system based on molecular characteristics and its impact on clinical applications. MethodsA literature review was conducted to analyze and organize the recent influences of molecular classification of thyroid cancer on clinical diagnosis and treatment. ResultsIn recent years, the classification of thyroid cancer has introduced molecular features such as BRAF and RAS mutations, highlighting the close association between these molecular characteristics and prognosis. For example, the BRAF V600E mutation is associated with high aggressiveness in papillary thyroid cancer, while RAS mutations suggest malignant potential in follicular tumors. With the advancement of multi-omics research, classification strategies based on multi-omics have shown significant value in the diagnosis, monitoring, treatment, and prognostic assessment of thyroid cancer. Although multi-omics integration has significantly improved the accuracy of prognostic assessments in thyroid cancer, there are still limitations, including imprecise detection of tumor heterogeneity and insufficient sensitivity and specificity of molecular biomarker detection.ConclusionsThe classification of thyroid cancer is developing towards the integration of molecular features to achieve more precise diagnosis and treatment. To accomplish this goal, it is necessary to overcome the challenges of tumor heterogeneity and the limitations of detection technologies in the future, and to promote the practical application of molecular classification in clinical settings.
ObjectiveTo evaluate the predictive value of intraoperative frozen section analysis of the Delphian lymph node (DLN) and pretracheal lymph node (PLN) for central lymph node metastasis (LNM) and recurrence risk stratification in patients with differentiated thyroid carcinoma (DTC). MethodsThis retrospective study included 133 DTC patients who underwent initial surgery with intraoperative frozen section evaluation of the DLN and PLN at the Department of Thyroid and Breast Surgery, Union Hospital, Wuhan, between January 2023 and December 2024. Receiver operating characteristic (ROC) curves were used to assess the predictive value of DLN/PLN metastasis count and ratio for central LNM and recurrence risk stratification. The concordance between intraoperative frozen pathology and final postoperative pathology was also evaluated. ResultsMultivariate analysis identified age (<20 or >50 years) as protective factor (OR=0.332, P=0.012) and capsular invasion as risk factors for DLN/PLN metastasis (OR=2.823, P=0.017). DLN/PLN metastasis number and ratio showed strong predictive performance for central LNM >5 nodes, with AUC of 0.913 [95%CI (0.841, 0.986), P<0.001] and 0.910 [95%CI (0.837, 0.983), P<0.001], and optimal cut-off values of 1.5 nodes and 45.00%, respectively. For predicting intermediate-to-high recurrence risk, AUCs were 0.818 [95%CI (0.740, 0.895), P<0.001] and 0.800 [95%CI (0.720, 0.880), P<0.001], with cut-offs of 0.5 nodes and 26.79%, respectively. Intraoperative frozen pathology demonstrated a sensitivity of 88.00% (66/75), specificity of 100.00% (58/58), positive predictive value of 100.00% (66/66), and negative predictive value of 86.57% (58/67). Concordance with postoperative pathology was high, with a Kappa value of 0.849 [95%CI (0.761, 0.937), P<0.001] and an intraclass correlation coefficient of 0.917 [95%CI (0.885, 0.940), P<0.001]. ConclusionsIntraoperative frozen section analysis of the DLN and PLN demonstrates reliable predictive value for central LNM and recurrence risk stratification in DTC. This method may help identify patients who could benefit from an extended surgical approach and is recommended as a valuable adjunct to intraoperative decision-making.