New Thyroid Imaging Reporting and Data System (TIRADS) Based on Ultrasonography Features for Follicular Thyroid Neoplasms: A Multicenter Study.

Abstract

Objective: Preoperative diagnosis of follicular thyroid neoplasm (FTN) (including follicular thyroid adenoma [FTA] and follicular thyroid carcinoma [FTC]) is difficult and the current ultrasound risk stratification systems (RSSs) for thyroid nodules are not suitable for FTN. Therefore, this study aimed to establish a new RSS for FTN as a useful preoperative evaluation to reduce missed diagnoses and unnecessary biopsies.

Methods: We included 535 FTNs from four hospitals in this retrospective study. All nodules were divided randomly into a test group (n = 370; 111 FTCs and 259 FTAs) and a validation group (n = 165; 51 FTCs and 114 FTAs). The ultrasound features of each nodule were analyzed. Thyroid Imaging Reporting and Data System (TIRADS) of FTN (FTN-TIRADS) was established based on the results of univariate analysis and logistic regression of ultrasound features. Each nodule was evaluated and classified by these four RSSs (RSS from European Thyroid Association [EU-RSS], RSS from the American Thyroid Association [ATA-RSS], TIRADS from the American College of Radiology [ACR-TIRADS], TIRADS from China [C-TIRADS]) and FTN-TIRADS, respectively. Receiver operating characteristic curve analysis was performed and the area under the receiver operator characteristic curve (AUC) was used to evaluate the diagnostic value of the RSSs mentioned above. The diagnostic value of FTN-TIRADS in the validation group was verified and compared with the test group and other four RSSs. The unnecessary rates of fine needle aspiration (FNA) were calculated and compared between FTN-RIRADS and the other four RSSs.

Results: The test group and validation group included 370 patients (86 male patients) and 165 patients (48 male patients). The following ultrasound features were independent risk factors and included in FTN-TIRADS: nodule composition, echogenicity, calcifications, halo sign, indistinct boundary with the thyroid capsule. Nodules were classified as TR2, TR3, TR4a, TR4b, TR4c, and TR5 with none to five suspicious features described, respectively. The AUC of FTN-TIRADS was 0.855, statistically higher than EU-RSS, ATA-RSS, ACR-TIRADS and C-TIRADS (0.759, 0.759, 0.753, and 0.677, respectively) (all p < 0.05). The FTN-TIRADS of the validation group had similar diagnostic performance to that of the test group (AUC 0.879 vs. AUC 0.855; p = 0.569). The unnecessary FNA rate of the FTN-TIRADS was 41.4%, which was significantly lower than that of EU-RSS (67.3%), ATA-RSS (69.0%), ACR-TIRADS (60.1%), and C-TIRADS (63.2%).

Conclusions: FTN-TIRADS achieved better differential diagnosis of FTN than current RSSs and significantly reduced the rate of unnecessary FNA. It can serve as a reliable preoperative noninvasive assessment tool for thyroid follicular tumors, reducing unnecessary FNA and unnecessary surgeries.