Використання цитохімічних маркерів як предикторів агресивності вузлів щитоподібної залози «сірої зони»

Bethesda IV–V cytological categories are characterized by an indeterminate risk of malignancy and often require additional risk stratification. Cytochemical assessment of dipeptidyl peptidase IV activity (DPP IV/CD26; designated DAP IV in the study protocol) may provide an accessible adjunctive test; however, its diagnostic thresholds for papillary thyroid carcinoma (PTC) and lymph-node metastasis remain insufficiently defined. Objective — to evaluate the diagnostic value of cytochemical DAP IV activity and determine threshold values at which the probability of detecting PTC and lymph-node metastases is clinically meaningful in patients with Bethesda IV—V cytology. Materials and methods. This single-center retrospective diagnostic accuracy study included 657 surgically treated patients from the Ukrainian Center of Endocrine Surgery with Bethesda IV–V cytology and available DAP IV results and postoperative histopathological findings. Diagnostic performance was evaluated using ROC analysis; the AUC and 95% confidence interval were determined by the DeLong method. Results. In the Bethesda IV sample, PTC was confirmed in 139 of 352 patients (39.5%), and lymph-node metastases were present in 11 (3.1%). Grouped ROC analysis demonstrated poor discriminatory performance of DAP IV: AUC 0.568 (95% CI 0.510 — 0.625) for PTC and 0.473 (95% CI 0.342 — 0.604) for metastases, with sensitivity, specificity, PPV, NPV, and accuracy of 63.0%, 61.8%, 85.1%, 32.6%, and 62.7%, respectively. In the Bethesda V sample, PTC was identified in 213 of 305 patients (69.8%), and metastases in 16 (5.2%). In the analytical Bethesda V subsample, the AUC was 0.664 for PTC and 0.646 for metastases. At a DAP IV threshold of ³ 6, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy for detecting PTC were 54.7%, 77.6%, 85.2%, 42.1%, and 61.5%, respectively; at a threshold of ³ 8 for metastases, the corresponding values were 66.7%, 56.3%, 9.1%, 96.3%, and 56.9%. Conclusions. In patients with Bethesda IV nodules, ROC analysis for predicting PTC and its metastases yielded AUC values close to 0.5, with modest diagnostic accuracy (62.7%) and a low NPV of 32.6%, indicating poor model performance. For the Bethesda V cohort, model performance may be considered moderate (AUC range, 0.6 — 0.7). DAP IV may be used as a predictor of metastasis with a high NPV (94.6%) and a low PPV (25.6%). In the Bethesda V cohort, a threshold of ³ 6 was characterized by high PPV but low NPV for PTC, whereas a threshold of ³ 8 had high NPV but very low PPV for metastases. In the selected Bethesda IV and Bethesda V cohort, the DAP IV value cannot serve as a sufficiently reliable predictor of either cancer or metastases. Clinical use of DAP IV as a predictor of metastasis requires refinement of an integrated risk-stratification model for gray-zone thyroid nodules.