Using dynamic nomogram to modify TI-RADS and reduce the unnecessary FNA of thyroid nodules

Abstract

Background:

Current guidelines recommend Fine Needle Aspiration (FNA) based on nodule size and ultrasound characteristics; however, these guidelines still lead to a certain amount of unnecessary FNAs.

Objective:

To develop a dynamic nomogram prediction model based on the ACR Thyroid Imaging, Reporting, and Data System (TI-RADS) to reduce unnecessary FNAs.

Methods:

This multicenter study analyzed 3313 thyroid nodules undergoing FNA. Univariate and multivariate logistic regression models were constructed. Patients were divided into a training cohort and two validation cohorts to compare diagnostic performance and unnecessary FNAs.

Results:

This nomogram achieved performance of Area Under the Curve (AUC) 0.914 (95%CI: 0.894–0.934), 0.923 (95%CI: 0.900–0.946), 0.948 (95%CI: 0.918–0.978) in the training, internal and external validation cohort. Using this model, the unnecessary FNA rates for nodules in ACR TI-RADS category 3 (TR3) have decreased from 99.4% to 0%, in TR4 from 77.6% to 47.1%, and in TR5 from 25.4% to 18.9% in Center1, in TR3 have decreased from 91.9% to 0%, in TR4 from 60.0% to 21.1%, and in TR5 from 11.5% to 4.5% in Center 2 (p < 0.01).

Conclusions:

This dynamic nomogram achieved better prediction of malignant thyroid nodules compared with the mentioned risk stratification system, leading to a more rational FNA strategy.