Background: Pediatric thyroid nodules are more likely to be malignant compared to those in adults and may have different concerning ultrasound (US) features. Recent adult guidelines stratify malignancy risk by US features. Our aim is to (1) describe and confirm US features that predict pediatric malignancy, and (2) apply the Adult American Thyroid Association (ATA) Risk Stratification Guidelines to a large pediatric cohort.
Methods: We identified 112 children with 145 thyroid nodules from 1996 to 2015. Two blinded pediatric radiologists independently read all US images, described multiple features, and reported their overall impression: benign, indeterminate, or malignant. Each nodule was assigned an ATA risk stratification category. Radiologists' impressions and ATA risk stratification were compared to histology and cytology results.
Results: Multiple US features including a solid composition, presence of microcalcifications, irregular margins, increased blood flow, and hypoechogenicity were associated with increased odds of malignancy. ATA risk stratification correlated with the radiologists' overall impression (p < 0.001). The sensitivity for detecting malignancy was comparable between both ATA stratification (91%) and the radiologists' overall impression (90%). The specificity of the radiologists' malignant overall impression (80%) was better compared to the ATA high risk stratification (54%).
Conclusions: At our institution, pediatric radiologists' overall impressions had similar sensitivity but better specificity for detecting malignancy than the ATA risk stratification tool by our convention. However, neither US-based methods perfectly discriminated benign from malignant nodules, supporting the continued need for fine needle aspiration for suspicious nodules. Further work is needed to develop an US-based scoring system specific to pediatric patients.
Keywords: Children; Pediatrics; Thyroid cancer; Thyroid nodules; Ultrasonography.
© 2018 S. Karger AG, Basel.