Ultrasound risk stratification for malignancy using the 2015 American Thyroid Association Management Guidelines for
Jennifer E Lim-Dunham1, Iclal Erdem Toslak2, Khalid Alsabban2
1Department of Radiology, Loyola University Chicago Stritch School of Medicine, 2160 S. First Ave., Maywood, IL, 60546, USA. jlim2@lumc.edu.
Background:
The 2015 American Thyroid Association (ATA) Management Guidelines for Children with Thyroid Nodules and Differentiated Thyroid Cancer provides selection criteria for nodules prior to ultrasound-guided fine-needle aspiration biopsy.
Objective:
To evaluate the diagnostic performance of pediatric thyroid nodule risk stratification for predicting malignancy when applying the ultrasound (US) criteria recommended.
Materials And Methods:
US characteristics of 39 thyroid nodules in 33 pediatric patients who underwent US fine-needle aspiration biopsy were reviewed by two radiologists. Based on the aggregated US criteria from the ATA Guidelines, each nodule was assigned a level of malignancy risk. Kappa coefficients were estimated to assess intra- and interobserver reliability. Using each patient's largest nodule observation (n = 33), univariable exact logistic regression analyses of US parameters were then conducted to estimate the odds of a malignant pathology diagnosis. A penalized Firth correction was employed in the univariable models analyzing composition, shape and level of suspicion due to quasi-complete data separation.
Results:
Twenty-seven nodules in 21 patients (median age: 16 years; 17 female) were benign and 12 nodules in 12 patients (median age: 16.5 years; 11 female) were malignant. Intraobserver agreement was substantial to almost perfect for composition, echogenicity, shape and margins. Interobserver agreement was almost perfect for composite level of suspicion. High level of suspicion was assigned to all 12 malignant nodules versus 9/21 (43%) of the benign nodules. Level of suspicion, solid/predominantly solid composition, irregular margins and echogenic foci emerged as significant predictors of malignancy with odds ratios (OR) of 8.5 (95% confidence interval [CI]: 1.7-1,130, P = 0.001), 10.5 (95% CI: 1.1-1,417, P = 0.04), 53.2 (95% CI: 5.1-2,988, P < 0.0001) and 3.5 (95% CI: 1.1-23.2, P = 0.03), respectively.
Conclusion:
The composite, US-based risk stratification criteria from the 2015 ATA Guidelines may provide an appropriate and reproducible method for estimating risk of malignancy for pediatric thyroid nodules.
