Risk prediction of occult lymph node metastasis in patients with clinical T1 through T2 N0 esophageal squamous cell carcinoma

Objectives: To investigate long-term survival outcomes and develop a risk model for occult lymph node metastasis (LNM) in patients with clinical T1 through T2 N0 esophageal squamous cell carcinoma. Methods: From 2006 to 2018, 675 patients with clinical T1 through T2 N0 esophageal cancer who underwent upfront surgery were analyzed. The survival of patients with occult LNM was compared with that of 116 patients with clinical T1 through T2N+ cancer who underwent neoadjuvant therapy plus surgery. After randomly dividing the patients with clinical T1 through T2 N0 tumors into the training and testing sets, a risk model for occult LNM was developed and validated. Results: Among patients with clinical T1 through T2 N0 esophageal cancer, occult LNM was found in 147 (21.8%) but not in 528 (78.2%). Patients with occult LNM had significantly worse prognosis than those without (P < .001), but showed similar outcomes to patients with clinical T1 through T2 N+ cancer (P = .981). According to the risk model, tumor maximum standardized uptake >3.8 (P = .002), histological differentiation grade (P = .015), tumor length >25 mm (P < .001), and advanced clinical T stage (P < .001) were independent risk factors for occult LNM in clinical T1 through T2 N0 cancer. A risk scoring system based on this model showed high accuracy (0.81) and good discriminant ability in both training sets (area under the receiver operating characteristic curve, 0.759 and testing area under the receiver operating characteristic curve, 0.743). Conclusions: Our risk scoring system for predicting occult LNM in clinical T1 through T2 N0 esophageal cancer has high accuracy and good discriminant ability.