Serum thyroglobulin is a poor diagnostic biomarker of malignancy in follicular and Ḧurthle-cell neoplasms of the thyroid

Abstract Background Serum thyroglobulin (Tg) is the most accurate biomarker for thyroid cancer recurrence. However, some clinicians measure preoperative Tg as a diagnostic cancer marker despite lack of supporting evidence. We examined whether Tg accurately predicts malignancy in follicular or Hurthle-cell neoplasms. Methods We reviewed 366 patients who underwent thyroidectomies for follicular/Hurthle-cell neoplasms. We compared Tg in malignant versus benign tumors by univariate and receiver-operator characteristic analyses. We also examined several Tg-derived indices that normalized Tg to known confounding factors including nodule size, thyroid function, and type of Tg assay. Results Thirty-nine patients met inclusion criteria for analysis. There were no differences between malignant (n = 16) and benign (n = 23) lesions in Tg or any of the normalized indexes. Receiver-operator characteristic analysis revealed an area under the curve of .59. Lesions with Tg levels greater than 500 μg/L had a positive predictive value of .75. Conclusions Tg has poor accuracy for predicting malignancy in follicular or Hurthle-cell thyroid neoplasms.