Predictive value of the Delphian and level VI nodes in papillary thyroid cancer

Background: Recent published data has shown that metastatic involvement of the prelaryngeal or Delphian lymph node (DLN), the highest of the central (level VI) cervical lymph nodes, is highly predictive of advanced nodal disease in papillary thyroid cancer (PTC). The aims of this study were to determine the diagnostic accuracy of all the level VI cervical nodes in PTC and to determine which node group, if any, is the most accurate in predicting lateral node (N1b) disease. Methods: This was a retrospective cohort study. Data were obtained from the University of Sydney Endocrine Surgical Unit Database and through a review of the histopathology records. The study cohort was composed of 177 consecutive patients with a final diagnosis of PTC who underwent total thyroidectomy and lymph node dissection, spanning the period from May 2001 to December 2006. Results: Of the 177 patients with PTC, 86 had the DLN removed, 51 had a pretracheal node removed and 76 had the paratracheal group removed. DLN, paratracheal and pretracheal node disease was present in 21%, 39% and 46%, respectively. Lateral node (N1b) disease was present in 35%. Paratracheal node involvement was mildly predictive of further disease with patients 1.7 times more likely to have lateral node involvement (sensitivity = 55%, specificity = 68%). Pretracheal node involvement was moderately predictive of further disease with patients three times more likely to have lateral node involvement (sensitivity = 72%, specificity = 74%). DLN involvement was highly predictive of further node involvement with patients nine times more likely to have lateral node disease (sensitivity = 53%, specificity = 94%) and 40 times more likely to have any nodal disease (sensitivity = 41%, specificity = 100%). Conclusion: This is the first study to examine the diagnostic accuracy of all level VI lymph nodes in PTC. While, metastatic involvement of all central nodal groups is indicative of further disease, the DLN is the most accurate predictor.