Gastric Cancer Peritoneal Carcinomatosis Risk Score

Abstract

BackgroundGastric cancer (GC) peritoneal carcinomatosis (PC) is associated with a poor prognosis. Although grade, histology, and stage are associated with PC, the cumulative risk of PC when multiple risk factors are present is unknown. This study aimed to develop a cumulative GCPC risk score based on individual demographic/tumor characteristics.MethodsPatient-level data (2004–2014) from the California Cancer Registry were reviewed by creating a keyword search algorithm to identify patients with gastric PC. Multivariable logistic regression was used to assess demographic/tumor characteristics associated with PC in a randomly selected testing cohort. Scores were assigned to risk factors based on beta coefficients from the logistic regression result, and these scores were applied to the remainder of the subjects (validation cohort). The summed scores of each risk factor formed the total risk score. These were grouped, showing the percentages of patients with PC.ResultsThe study identified 4285 patients with gastric adenocarcinoma (2757 males, 64.3%). The median age of the patients was 67 years (interquartile range [IQR], 20 years). Most of the patients were non-Hispanic white (n = 1748, 40.8%), with proximal (n = 1675, 39.1%) and poorly differentiated (n = 2908, 67.9%) tumors. The characteristics most highly associated with PC were T4 (odds ratio [OR], 3.12; 95% confidence interval [CI], 2.19–4.44), overlapping location (OR 2.27; 95% CI 1.52–3.39), age of 20–40 years (OR 3.42; 95% CI 2.24–5.21), and Hispanic ethnicity (OR 1.86; 95% CI 1.36–2.54). The demographic/tumor characteristics used in the risk score included age, race/ethnicity, T stage, histology, tumor grade, and location. Increasing GCPC score was associated with increasing percentage of patients with PC.ConclusionBased on demographic/tumor characteristics in GC, it is possible to distinguish groups with varying odds for PC. Understanding the risk for PC based on the cumulative effect of high-risk features can help clinicians to customize surveillance strategies and can aid in early identification of PC.