Ultrasound Criteria of Malignancy for Cervical Lymph Nodes in Patients Followed Up for Differentiated Thyroid Cancer

Sophie Leboulleux, Elizabeth Girard, Mathieu Rose, Jean Paul Travagli, Nadia Sabbah, Bernard Caillou, Dana M. Hartl, Nathalie Lassau, Eric Baudin, Martin Schlumberger
2007 Journal of Clinical Endocrinology and Metabolism  
Context: Neck ultrasonography (US) has become a keystone in the follow-up of patients with differentiated thyroid cancer. Objective: The aim of this study was to determine specificity and sensitivity of ultrasound criteria of malignancy for cervical lymph nodes (LNs) in patients with differentiated thyroid cancer. Design: We prospectively studied 19 patients referred to the Institut Gustave Roussy for neck LN dissection. All patients underwent a neck US within 4 d prior to surgery. Only LNs
more » ... were unequivocally matched between US and pathology were taken into account for the analysis. Results: One hundred three LNs were detected on US, 578 LNs were surgically removed, and 56 LNs were analyzed (28 benign and 28 malignant). Sensitivity and specificity were 68 and 75% for the long axis (Ն1 cm), 61 and 96% for the short axis (Ͼ5 mm), 46 and 64% for the round shape (long to short axis ratio Ͻ 2), 100 and 29% for the loss of fatty hyperechoic hilum, 39 and 18% for hypoechogenicity, 11 and 100% for cystic appearance, 46 and 100% for hyperechoic punctuations, and 86 and 82% for peripheral vascularization. Conclusion: Cystic appearance, hyperechoic punctuations, loss of hilum, and peripheral vascularization can be considered as major ultrasound criteria of LN malignancy. LNs with cystic appearance or hyperechoic punctuations are highly suspicious of malignancy. LNs with a hyperechoic hilum should be considered as benign. Peripheral vascularization has the best sensitivity-specificity compromise. Round shape, hypoechogenicity, and the loss of hilum taken as single criteria are not specific enough to suspect malignancy.
doi:10.1210/jc.2007-0444 pmid:17609301 fatcat:2qm7btshvbgqncq3d3odt3jvim