Development and Multicenter Validation of a Novel Immune-Inflammation-Based Nomogram to Predict Survival in Western Resectable Gastric and Gastroesophageal Junction Adenocarcinoma (GEA): The NOMOGAST

Abstract

Background. More than 50% of operable GEA relapse after curative-intent resection. We aimed at externally validating a nomogram to enable a more accurate estimate of individualized risk in resected GEA. Methods. Medical records of a training cohort (TC) and a validation cohort (VC) of patients undergoing radical surgery for c/uT2-T4 and/or node-positive GEA were retrieved, and potentially interesting variables were collected. Cox proportional hazards in univariate and multivariate regressions were used to assess the effects of the prognostic factors on OS. A graphical nomogram was constructed using R software’s package Regression Modeling Strategies (ver. 5.0-1). The performance of the prognostic model was evaluated and validated. Results. The TC and VC consisted of 185 and 151 patients. ECOG:PS > 0 (p p p p = 0.016) were independent prognostic values in the TC. They were used for the construction of a nomogram estimating 3- and 5-year OS. The discriminatory ability of the model was evaluated with the c-Harrell index. A 3-tier scoring system was developed through a linear predictor grouped by 25 and 75 percentiles, strengthening the model’s good discrimination (p Conclusions. We externally validated a prognostic nomogram to predict OS in a joint independent cohort of resectable GEA; the NOMOGAST could represent a valuable tool in assisting decision-making. This tool incorporates readily available and inexpensive patient and disease characteristics as well as immune-inflammatory determinants. It is accurate, generalizable, and clinically effectivex.

Keywords

• gastric cancer
• prognosis
• nomogram