INDIVIDUAL RISK CALCULATOR TO PREDICT LYMPH NODE METASTASES IN PATIENTS WITH SUBMUCOSAL (T1B) ESOPHAGEAL ADENOCARCINOMA: MULTICENTER COHORT STUDY

Aims A prognostic model may help to identify patients at risk for lymph node metastases (LNM) in order to stratify between a conservative approach or additional surgery, after endoscopic resection of pT1b esophageal adenocarcinoma (EAC). The aim of this study was to develop a prediction model for the risk LNM or distant metastases in patients with pT1b EAC.

Methods This is a nationwide, retrospective, multicenter cohort study in collaboration with the Netherlands Cancer Registry. All patients who were diagnosed with pT1b EAC and treated with endoscopic resection and/or surgery between 1989 and 2017 were included. Primary endpoints: the presence of LNM in surgically resected specimens (≥12 resected lymph nodes) or the development of pathologically confirmed LNM or distant metastases during follow-up. Histopathological reassessment of all resection specimens was performed by three gastrointestinal pathologists. Cox proportional hazard analysis was performed to identify independent risk factors associated with metastases. The discriminative ability of this model was assessed using the c-statistic.

Results 283 patients were included (median age 66 years [IQR: 58-72], 87% male). Endoscopic resection was performed in 100 patients and surgery in 183 patients. Ninety-three (32.9%) patients had metastases, LNM mainly identified in surgical specimens (78/93). In multivariable analysis, the risk of developing metastases increased with worse differentiation grade (G2 vs G1: HR 3.2, 95% CI 1.2-9.0; G3 vs G1: HR 3.1; 95% CI 1.1-8.9), deep submucosal invasion (Sm3: HR 2.4; 95% CI 1.3-4.5) and lymphovascular invasion (HR 3.0; 95% CI 1.9-4.5). The c-statistic of the prediction model was 0.74 (95% CI 0.68-0.79).

Conclusions One-third of patients with pT1b EAC had metastases. Risk factors are moderate and poor differentiation grade, deep submucosal invasion and the presence of lymphovascular invasion. A personalized risk for LNM can be predicted based on the presence or absence of each of these separate risk factors with a c-statistic of 0.74.