Utility of adult-based ultrasound malignancy risk stratifications in pediatric thyroid nodules.

Abstract

Individual ultrasound (US) features have limited ability to distinguish benign from malignant thyroid nodules. Adult-based systems have been developed to integrate the sonographic features in an effort to improve diagnostic accuracy. None, however, has been validated in children, in whom the likelihood of malignancy is 2-5 times higher than adults. To assess the performance of two adult-based sonographic (US) stratification methods for assessment of thyroid nodules in children. This retrospective study comprised 124 children who underwent thyroid US. Three radiologists reviewed the US data using the American Thyroid Association (ATA) and the Thyroid Image Reporting and Data System (TI-RADS). Radiologists' accuracy and agreement was assessed. The reference standard was histopathology/cytology or 2-year follow-up of clinical outcome for nonoperative cases. We assessed 71 benign and 52 malignant nodules and excluded 1 nodule. Using the ATA pattern descriptions, 80% of malignant nodules were classified as "high" 36/52 (69%) or "intermediate" 6/52 (11%) likelihood of malignancy. A total of 20/71 (28%) benign nodules were also classified within these two categories. Using the TI-RADS, malignant nodules were classified as 2, 3, 4a, 4b, 4c and 5, with rate of malignancy of 0%, 0%, 7/52 (13.5%), 7/52 (13.5%), 32/52 (61.5%) and 6/52 (11.5%), respectively. Benign nodules were also classified in the 4a (26/71; 36.6%), 4b (17/71; 24%), 4c (14/71; 19.7%) and 5 (1/71; 1.4%) categories. The positive and negative predictive values were 68.0% and 87.5% for ATA, and 71.7% and 80.0% for TI-RADS. We validated the use of ATA and TI-RADS methods in children and showed that they have test characteristics similar to those in adults, although neither is independently sufficient to discriminate nodules' likelihood of malignancy.