Plasma Cell-Free DNA to Differentiate Malignant from Benign Thyroid Nodules.

Molecular testing is increasingly used to identify malignancy in thyroid nodules (especially indeterminate category). Measurement of cell-free DNA (cfDNA) levels from plasma has been useful in diagnosis of cancers of other organs/tissues; herein we analyze cfDNA levels in patients with thyroid nodules to explore the possibility of establishing a cutoff for identification of malignancy. Patients underwent ultrasonography (USG) and USG-guided fine needle aspiration as well as surgery, where indicated. Cell-free DNA was extracted from plasma and quantified. In initial analysis (determination of cutoff), cfDNA levels were compared between Bethesda 2 and Bethesda 5 &6 to establish a cutoff value that could differentiate malignant from benign nodules. In the subsequent analysis, the aforementioned cutoff was applied (validation of cutoff) to those with indeterminate nodules to check ability to predict malignancy. Fine needle aspiration (n = 119) yielded patients with Bethesda 2 (n = 69) Bethesda 5 & 6 (n = 13) who underwent histopathological confirmation. Cell-free DNA levels in these 2 groups were 22.85 ± 1.27 and 96.20 ± 8.31 (ng/mL) respectively. A cfDNA cutoff of 67.9 ng/mL, with area under the curve of 0.992 (95% CI, 0.97-1.0) with 100% sensitivity and 93% specificity was established to identify malignant lesions. Indeterminate group (Bethesda 3 & 4) underwent surgery (malignant n = 24), (benign n = 13), and using the previously identified cutoff for cfDNA, we were able to identify malignant lesions with a sensitivity of 100% and specificity of 92.3%. There was a very strong agreement between cfDNA-based classification with histopathology-based classification of benign and malignant nodules (Cohen's kappa 0.94; P < 0.001). Plasma cfDNA estimation could help differentiate malignant from benign thyroid nodules.

© The Author(s) 2021. Published by Oxford University Press on behalf of the Endocrine Society. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.