Risk Factors Of Occult Peritoneal Metastasis Of Gastric Cancer And The Establishment And Verification Of Risk Prediction Models

Objective:1.Explore the high-risk factors of occult peritoneal metastasis of gastric cancer.2.Establish a risk prediction model for occult peritoneal metastasis of gastric cancer based on high-risk factors,use Nomogram to display it,and verify it with internal and external data.3.Provide a reference for the diagnosis and treatment decision of gastric cancer peritoneal metastasis.Methods:First,a case-control study method was used to screen out the high-risk factors for occult peritoneal metastasis of gastric cancer.Then,845 study subjects from the Affiliated Hospital of Yan’an University from 2014.1.1 to 2020.7.31 were used as the training group.According to the results of single-factor and multi-factor Logistic regression analysis,a risk prediction model for gastric cancer occult peritoneal metastasis was established,and the Nomogram chart was used to display it.The training set data was used for internal verification,and 323 subjects from the Second Hospital of Yulin City from 2017.1.1 to 2019.12.31 were used for external verification.The model was evaluated from the three aspects of discrimination,calibration and clinical practical value.Results:1.In the training group data of the Affiliated Hospital of Yan’an University,independent risk factors for occult peritoneal metastasis of gastric cancer include:imaging showing abdominal lymphadenopathy(P<0.001),presence of ascites(P<0.001),T stage(P<0.001),N stage(P<0.001),CA19-9≥23.18U/m L(P=0.021),CA72-4≥2.68U/m L(P<0.001).The independent risk factors for occult peritoneal metastasis of gastric cancer screened out from the data of Yulin Second Hospital’s validation group include: imaging revealed abdominal lymphadenopathy(P=0.006),low pathological differentiation(P<0.001),T stage(P<0.001),CA19-9≥23.18U/m L(P=0.021).2.Use six factors based on the multi-factor analysis of the training group to build the occult peritoneal metastasis risk prediction model in gastric cancer and display it(shown in Table 2-5 and Figure 2-2).In internal verification,AUC=0.895(95%CI 0.872-0.917);the calibration curve fits well;the sensitivity is 49%;the specificity is 94%;the decision curve analysis(DCA)shows the positive net income in the range of 0-100% probability.In external verification,AUC=0.899(95%CI 0.863-0.936);sensitivity is 60%;specificity is 88%.The threshold score is 120.Conclusion:1.The high-risk factors for occult peritoneal metastasis of gastric cancer in the Affiliated Hospital of Yan’an University mainly include images showing swollen abdominal lymph nodes,the presence of ascites,T stage,N stage,CA19-9≥23.18U/m L,CA72-4≥2.68U/m L.And the high risk factors for gastric cancer peritoneal metastasis in Yulin No.2 Hospital include: imaging shows swollen abdominal lymph nodes,low pathological differentiation,T stage,CA19-9≥23.18U/m L.2.The risk prediction model of occult peritoneal metastasis represented of gastric cancer composed of the above six factors is obtained and shown in the Nomogram.3.The established model is economical and convenient,and has certain clinical practical value in assisting the diagnosis of peritoneal metastasis of gastric cancer and screening more high-risk peritoneal metastasis in gastric cancer patients who fit to more active treatment measures.