BackgroundDifferentiated thyroid cancer (DTC) requires long-term follow-up due to the risk of delayed recurrence. Follow-up surveillance involves serial neck ultrasound (US) and thyroglobulin (Tg); however, the optimal frequency and diagnostic performance of neck US outside of specialized thyroid cancer centres in higher risk patients is not well defined. We sought to evaluate the diagnostic performance of US and serial Tg in advanced stage DTC.MethodsWe retrospectively reviewed our thyroid cancer database for patients with stage III and IV DTC from 2006 to 2018, total thyroidectomy, and at least 2 years follow-up to assess recurrence rates. Those with hemi-thyroidectomy or anti-Tg antibodies were excluded. Diagnostic performance of US and Tg were assessed using a composite reference standard of follow-up imaging and pathology. All relevant US were reviewed by a blinded expert radiologist for uniformity.ResultsOf 136 included patients (91 females, mean age 58.9), 26 (19%) had recurrence of DTC over median follow-up of 6.6 years (IQR 5.3–9.3). The sensitivity and specificity of US in diagnosing cervical recurrence were 73.3% (95% CI 0.51–0.96) and 68.3% (95% CI 0.60–0.77) based on historical reports, respectively, and 80% (95% CI 0.60-1.00) and 87.8% (95% CI 0.82–0.93) based on blinded expert review, respectively. Tg had a sensitivity of 95.5% (95% CI 0.89-1.0) and specificity of 96.2% (95% CI 0.92–0.99) in detecting cervical recurrence or distant metastases. False positive US findings on historical US and subsequent review occurred in 38 (28%) and 15 (11%) patients, respectively, while 5 (3.6%) had false positive Tg results.ConclusionSerial Tg has better sensitivity and specificity than US for detecting recurrence of advanced stage DTC. Furthermore, re-interpretation of abnormal findings using structured US reporting with a subspecialized reader may improve diagnostic performance of US and improve its utility in clinical care.