Evaluating diagnostic efficacy of multigene testing in non-diagnostic thyroid nodules by fine-needle aspiration cytology: A prospective cohort study

ObjectiveNon-diagnostic thyroid nodules (Bethesda I) account for 5%−20% of all thyroid nodules. Accurate differentiation of benign and malignant nodules can reduce unnecessary surgeries and repeat biopsies. Herein we evaluated the diagnostic efficacy of multigene testing in non-diagnostic thyroid nodules and developed a predictive model integrating molecular and clinical data.MethodsIn this prospective cohort study, 1,175 patients with thyroid nodules were evaluated for inclusion, of which 218 patients with Bethesda I nodules met our inclusion criteria. The primary outcome was diagnostic accuracy of molecular testing, and the secondary outcome was the performance of a predictive model integrating molecular and clinical data.ResultsFinal histopathology identified 165 benign and 53 malignant nodules. Molecular testing detected 10 distinct point mutations and seven gene fusions. Among benign nodules, 147 tested negative and 18 tested positive, whereas 44 malignant nodules tested positive and nine tested negative. In nodules with ultrasound grades 4−5 and fine-needle aspiration cytology (FNAC) results categorized as non-diagnostic, molecular testing achieved sensitivity of 83.00%, specificity of 89.00%, positive predictive value (PPV) of 71.00%, negative predictive value (NPV) of 94.20%, and overall accuracy of 87.60%. The predictive model incorporated 18 clinical and 19 molecular features. Eleven non-zero predictors were selected via least absolute shrinkage and selection operator (LASSO), and the model achieved area under curve (AUC) of 0.95 in the training set and 0.96 in the testing set. Decision curve analysis indicated greater net benefit compared with conventional diagnostic approaches.ConclusionsMolecular testing significantly improved diagnostic accuracy for Bethesda I thyroid nodules. Integrating molecular and clinical data enabled the development of a robust predictive model, facilitating precise, individualized patient management and reducing the need for repeat FNAC and unnecessary surgeries.