Construction of a nomogram model for predicting peritoneal metastasis in gastric cancer: focused on cardiophrenic
Xiaolong Gu1, Yang Li1, Gaofeng Shi1
1Department of Radiology, The Fourth Hospital of Hebei Medical University, Shijiazhuang, 050000, Hebei, PR China.
Background:
A different treatment was used when peritoneal metastases (PM) occurred in patients with gastric cancer (GC). Certain cancers' peritoneal metastasis could be predicted by the cardiophrenic angle lymph node (CALN). This study aimed to establish a predictive model for PM of gastric cancer based on the CALN.
Methods:
Our center retrospectively analyzed all GC patients between January 2017 and October 2019. Pre-surgery computed tomography (CT) scans were performed on all patients. The clinicopathological and CALN features were recorded. PM risk factors were identified via univariate and multivariate logistic regression analyses. The receiver operator characteristic (ROC) curves were generated using these CALN values. Using the calibration plot, the model fit was assessed. A decision curve analysis (DCA) was conducted to assess the clinical utility.
Results:
126 of 483 (26.1%) patients were confirmed as having peritoneal metastasis. These relevant factors were associated with PM: age, sex, T stage, N stage, enlarged retroperitoneal lymph nodes (ERLN), CALN, the long diameter of the largest CALN (LD of LCALN), the short diameter of the largest CALN (SD of LCALN), and the number of CALNs (N of CALNs). The multivariate analysis illustrated that the LD of LCALN (OR = 2.752, p < 0.001) was PM's independent risk factor in GC patients. The area under the curve (AUC) of the model was 0.907 (95% CI 0.872-0.941), demonstrating good performance in the predictive value of PM. There is excellent calibration evident from the calibration plot, which is close to the diagonal. The DCA was presented for the nomogram.
Conclusion:
CALN could predict gastric cancer peritoneal metastasis. The model in this study provided a powerful predictive tool for determining PM in GC patients and helping clinicians allocate treatment.
