Clinical Risk Prediction Model for Neoadjuvant Therapy in Resectable Esophageal Adenocarcinoma.

Abstract

Clinical staging with endoscopic ultrasound (EUS) and positron emission tomography (PET) is used to identify esophageal adenocarcinoma (EAC) patients with locally advanced disease and therefore, benefit from neoadjuvant therapy. However, EUS is operator dependent and subject to interobserver variability. Therefore, we aimed to identify clinical predictors of locally advanced EAC and build a predictive model that can be used as an adjunct to current staging methods. This was a cross-sectional study of patients with EAC who underwent preoperative staging with EUS and PET scan followed by definitive therapy at our institution from January 2011 to December 2017. Demographic data, symptoms, endoscopic findings, EUS, and PET scan findings were obtained. Four hundred and twenty-six patients met the study criteria, of which 86 (20.2%) patients had limited stage EAC and 340 (79.8%) had locally advanced disease. The mean age was 65.4±10.3 years of which 356 (83.6%) were men and 393 (92.3%) were White. On multivariable analysis, age (above 75 or below 65 y), dysphagia [odds ratio (OR): 2.84], weight loss (OR: 2.06), protruding tumor (OR: 2.99), and tumor size >2 cm (OR: 3.3) were predictive of locally advanced disease, while gastrointestinal bleeding (OR: 0.36) and presence of visible Barrett's esophagus (OR: 0.4) were more likely to be associated with limited stage. A nomogram for predicting the risk of locally advanced EAC was constructed and internally validated. We constructed a nomogram to facilitate an individualized prediction of the risk of locally advanced EAC. This model can aid in decision making for neoadjuvant therapy in EAC.