ObjectiveTo construct and validate the value of constructing nomogram model based on clinical-radiological parameters in preoperative prediction of lymph node metastasis in advanced gastric cancer (AGC).
MethodsA retrospective analysis was conducted on 216 gastric cancer patients confirmed by pathology, who were randomly divided into a training group (n=158) and a validation group (n=58). The clinical data and computed tomography (CT) imaging features of patients were collected for univariate and multivariate logistic regression analysis, the validation group was used to validate, the nomogram model was constructed with R 3.5.3 software package, the prediction efficacy of the nomogram model was evaluated using receiver operating characteristic (ROC) curve, and the clinical practicality of the model was validated by calibration curve and decision curve.
ResultsAmong the 216 patients, 130 cases were positive for lymph node metastasis and 86 cases were negative for lymph node metastasis. In the training group and validation group, there were statistically significant differences in alcohol consumption history, peritumoral fat infiltration, degree of enhancement, CT-lymph node status, and platelet to lymphocyte ratio (PLR) in preoperative prediction of lymph node metastasis in gastric cancer patients (P<0.05 to P<0.01). Multivariate logistic regression analysis showed that alcohol consumption history, peritumoral fat infiltration, CT enhancement degree, CT-lymph node status, and PLR>161 were independent influencing factors for preoperative prediction of lymph node metastasis in gastric cancer patients (P<0.05). A namogram model was constructed to predict lymph node metastasis in gastric cancer patients based on alcohol consumption history, peritumoral fat infiltration, CT enhancement degree, CT-lymph node status, and PLR. The area under the ROC curve of the model was 0.789 (95%CI: 0.719-0.860) in the training group and 0.791 (95%CI: 0.678-0.905) in the validation group, respectively. The sensitivity and specificity of the model were 67.4% and 78.3% in the training group, and 62.5% and 84.6% in the validation group, respectively. The calibration curve and decision curve confirmed the clinical practicality of the model.
ConclusionsAlcohol consumption history, peritumoral fat infiltration, CT enhancement degree, CT-lymph node status, and PLR are independent influencing factors for the occurrence of lymph node metastasis in gastric cancer patients. The namogram model constructed based on them has good prediction efficacy and can assist clinical decision-making to some extent.
DownLoad: