Predicting peritoneal carcinomatosis of gastric cancer: A simple model to exempt low-risk patients from unnecessary staging laparoscopy

Abstract

Background: Peritoneal carcinomatosis (PC) of gastric cancer indicates a poor outcome and is mainly diagnosed by staging laparoscopy (SL). This study was designed to develop a risk stratification model based on the number of risk factors to exempt low-risk patients from unnecessary SL.

Methods: This was a retrospective cohort study based on a single institution between January 2015 and December 2019. SL is indicated for patients of advanced locoregional stage, and clinicopathologic characteristics of 535 consecutive patients were included. PC-associated variables were identified by logistic regression analysis. A risk stratification model based on the number of risk factors was constructed, and we defined its predictive value with a receiver operating characteristic (ROC) curve and negative predictive value.

Results: In total, 15.9% of included patients were found to have PC during SL. Borrmann type IV, elevated CA125, and tumour diameter ≥5 cm were independent risk factors of PC. These three factors combined with cT4 were selected as predictive factors, and the number of predictive variables was significantly related to the possibility of PC (2.0%, 12.8%, 20.0%, 54.2%, and 100%, respectively). When the cutoff value is more than one predictive factor, the negative predictive value is 98.0%, with an area under the curve of 0.780. This model could exempt 29.8% of unnecessary SL compared to the indication of the current NCCN guideline.

Conclusions: We constructed a simple model to predict the probability of PC using the number of predictive factors. It is recommended that patients without any of these factors should be exempt from SL.

Keywords: gastric cancer; peritoneal carcinomatosis; prediction model; risk factor; staging laparoscopy.