Optimal criteria for predicting lymph node metastasis in esophageal squamous cell carcinoma by anatomical location using preoperative computed tomography: a retrospective cohort study

Purpose Predicting lymph node metastasis (LNM) in esophageal squamous cell carcinoma (ESCC) is critical for selecting appropriate treatments despite the low accuracy of computed tomography (CT) for detecting LNM. Variation in potential nodal sizes among locations or patients’ clinicopathological background factors may impact the diagnostic quality. This study explored the optimal criteria and diagnostic ability of CT by location. Methods We retrospectively reviewed preoperative CT scans of 229 patients undergoing curative esophagectomy. We classified nodal stations into six groups: Cervical (C), Right-upper mediastinal (U R ), Left-upper mediastinal (U L ), Middle mediastinal (M), Lower mediastinal (L), and Abdominal (A). We then measured the short-axial diameter (SAD) of the largest lymph node in each area. We used receiver operating characteristics analyses to evaluate the CT diagnostic ability and determined the cut-off values for the SAD in all groups. Results Optimal cut-offs were 6.5 mm (M), 6 mm (C, L, and A), and 5 mm (U R and U L ). Diagnostic abilities differed among locations, and U R had the highest sensitivity. A multivariate analysis showed poor differentiation to be an independent risk factor for a false-negative diagnosis ( p = 0.044). Conclusions Optimal criteria and diagnostic abilities for predicting LNM in ESCC varied among locations, and poor differentiation might contribute to failure to detect LNM.