Assessing the Diagnostic Accuracy of TI-RADS in Pediatric Thyroid Nodules: A Multi-institutional Review.

PURPOSE

Thyroid nodules are uncommon in children and adolescents but carry an increased risk of malignancy when present. The Thyroid Imaging Reporting and Data System (TI-RADS) is an adult-validated ultrasound-based risk assessment providing a prediction of malignant potential for thyroid nodules, thereby guiding recommendations for fine needle aspiration biopsy (FNAB). Minimal data exist regarding the applicability of TI-RADS to predict malignancy in pediatric thyroid nodules. This study aims to analyze the performance of TI-RADS for children and adolescents with thyroid nodules, hypothesizing that applying TI-RADS criteria would improve accuracy and reduce the number of recommended FNAB compared to American Thyroid Association (ATA) size criteria alone.

METHODS

A multi-institutional retrospective analysis was conducted including patients ≤21 years with a thyroid nodule by sonographic thyroid imaging between 2015 and 2020. TI-RADS scores were assigned at each institution by a pediatric radiologist trained in thyroid imaging and TI-RADS criteria. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy of TI-RADS scoring were compared to existing ATA size-based recommendation for performing a FNAB. Accounting for relative size differences between adults and children, a novel PED TI-RADS category was developed and tested, recommending FNAB for thyroid nodules with a TI-RADS 3 and ≥ 1.5 cm, TI-RADS 4 and ≥ 1.0 cm, and TI-RADS 5 any feasible size.

RESULTS

291 nodules from 260 patients (median age 14.9 years, 78.8% female) were assessed using TI-RADS. Applying adult TI-RADS criteria resulted in recommendation of FNAB for 35.1% of nodules, in contrast to 76.6% recommended by ATA guidelines (p < 0.0001) (Table). Utilizing the adult TI-RADS score ≥3 as an FNAB indicator resulted in 100% sensitivity and 28.5% specificity, with 0 cases of missed malignant nodules on pathology. When novel PED TI-RADS criteria were applied, 88 patients would have been spared an unnecessary FNAB with improved sensitivity and accuracy over ATA criteria.

CONCLUSIONS

The application of adult and PED TI-RADS scoring to thyroid nodules in pediatric patients enhances the accuracy of malignancy prediction compared to current American Thyroid Association size criteria alone. The utilization of PED TI-RADS scoring eliminated unnecessary biopsies in many children while not missing a single thyroid malignancy.

LEVEL OF EVIDENCE

Level III.