Diagnostic accuracy of the McGill thyroid nodule score in paediatric patients

Introduction Thyroid cancer is the most common paediatric endocrine cancer; accurate diagnosis and prompt management of paediatric thyroid nodules is critical. The McGill Thyroid Nodule Score (MTNS), based upon clinical, ultrasound (US) and cytology criteria, has recently been modified and studied in a pilot paediatric group with good results. We aim to describe the diagnostic accuracy of the paediatric modified MTNS (PMTNS) in a large paediatric cohort. Methods We utilized an established retrospective cohort between 1996 and 2015 of 99 patients <= 21 years old with 131 thyroid nodules. Two experienced paediatric radiologists, blinded to pathology and radiology reports, reviewed US features. We abstracted cytology, histology and laboratory results, assigning each nodule a PMTNS. PMTNS performance was compared to FNA and histology. Results Approximately 33% of nodules were malignant. The cohort was predominantly adolescent (mean age 15.4 +/- 3.8 years). The average PMTNS for malignant and benign nodules, based on final histology, was 12.7 +/- 4.3 and 1.7 +/- 2.9, respectively. A PMTNS >= 8 resulted in a 93.2% sensitivity and 93.1% specificity for detecting malignancy, while a PMTNS >= 9 resulted in a 90.9% sensitivity and 96.6% specificity. However, Bethesda cytology category >= 4 independently had a 97.7% sensitivity and 94.0% specificity for detecting malignancy. The PMTNS had diminishing diagnostic accuracy in younger children compared with older children. Conclusion Paediatric modified McGill Thyroid Nodule Score predicts malignancy, perhaps due to the score's emphasis on cytology results; however, the score is less accurate in younger patients. While cytology results remain reliable, further work is needed to develop a non-invasive scoring system to predict malignancy in children.