Correlation between the British Thyroid Association ultrasound grading of thyroid nodules and histopathology specimens with assessment of inter-rater
Olivia Francies, Susan Jawad, Simon Morley, Sally Daniels, Sofia Otero, Radiology, University College London Hospitals
The British Thyroid Association (BTA) guidelines 2014 introduced an ultrasound (US) classification for thyroid nodules (U1-U5) with U1 being normal thyroid, U2 benign, U3 indeterminate, U4 suspicious for malignancy, and U5 likely malignant. We retrospectively reviewed thyroidectomy specimens from a twelve-month period between 2015 and 2016. 121 specimens were eligible for inclusion. The U grade given in the original US report was correlated with the histopathology result. In addition, four observers (three Consultant radiologists and one Sonographer, all with an interest in head and neck imaging) retrospectively allocated a U grade based on the saved US pictures, blinded to the patient information, original reports and any cytology or histopathology results.
The rate of malignancy for each U category (as given in the original report) was calculated: U2 = 3%; U3 = 40%; U4 = 79%; U5 = 100%.
In addition, we calculated the specificity and sensitivity for diagnosis of malignant thyroid nodules on US for the original report and four further raters. The sensitivity was high (92-98%), however the specificity was only moderate (43-49%).
The inter-rater agreement was calculated for the original reports and four further raters, with a Fleiss Kappa score of 0.51, indicating moderate inter-rater agreement.
We conclude that the BTA guidelines U classification has a high sensitivity for detecting malignant thyroid nodules but only a moderate specificity and moderate inter-rater agreement. This is likely to be due to the inherent difficulty in the assessment of thyroid nodules on US, but may also be due to the complexity of the scoring system. The next step would be to ascertain whether the BTA guidelines provide any advantage over a simplified three-tier scoring system e.g. benign, indeterminate, likely malignant. We intend to repeat the study comparing both the BTA 2014 U grading with a simplified three-tier score.