Assessment of Thyroid Nodules using TIRADS Classification: Interobserver Agreement and Correlation with Histopathology

Abstract

Classification of thyroid nodules (TNs) based on TIRADS category is important for predicting malignancy, avoiding unnecessary biopsy and aiding in the management of patients. A prospective nonintervention study was carried out in the period from February 2020 to February 2021 in King Fahad Specialist Hospital in Dammam (Saudi Arabia). The study included 222 patients with suspected TNs (TIRADS 3-5). The thyroid ultrasound scanning was performed with a high frequency linear array probe (9 MHz and 15 MHz). The thyroid gland and adjacent neck tissues were scanned. The diagnostic performance of the TIRADS classifications was evaluated against final histology. Three highly experienced sonographers independently rated the US features of each nodule. Each sonographer assigned points to each TN for the five separate categories (composition, echogenicity, shape, margin, and echogenic foci), according to the TIRADS protocols produced by ACR. Cohen’s kappa scale was used to measure the interobserver agreement in categorizing TNs, and then the performance of ACR TIRADS categories for predicting malignancy was assessed using fine needle aspiration (FNA) as reference standard. For estimation of the diagnostic performance of ACR TIRADS to predict malignancy, the category for each TN was assigned by the principal investigator (Observer 1). This study included only TIRADS 3-5 (mildly to highly suspicious nodules: TR3-TR5), according to ACR TIRADS classification. The agreement for all sonographic features of TN among three observers (principal Observer 1 and two sub-investigators [Observer 2 and Observer 3]) was perfect, being 0.91-1.00 (Cohen’s kappa). Out of 222 patients with TNs, the percentage of malignancy in 68 TIRADS-5 nodules was 91.2%, while in 154 TIRADS 3-4 nodules the percentage of malignancy was 29.9%. The study concluded that interobserver agreement in TIRADS classification and characterization of suspicious thyroid nodule was perfect.