[Establishment and validation of a predictive nomogram model for advanced gastric cancer with perineural invasion]
S H Liu,X Y Hou,X X Zhang,G W Liu,F J Xin,J G Wang,D L Zhang,D S Wang,Y Lu
DOI: https://doi.org/10.3760/cma.j.cn.441530-20200103-00004
2020-11-25
Abstract:Objective: Peripheral nerve invasion (PNI) is associated with local recurrence and poor prognosis in patients with advanced gastric cancer. A risk-assessment model based on preoperative indicators for predicting PNI of gastric cancer may help to formulate a more reasonable and accurate individualized diagnosis and treatment plan. Methods: Inclusion criteria: (1) electronic gastroscopy and enhanced CT examination of the upper abdomen were performed before surgery; (2) radical gastric cancer surgery (D2 lymph node dissection, R0 resection) was performed; (3) no distant metastasis was confirmed before and during operation; (4) postoperative pathology showed an advanced gastric cancer (T2-4aN0-3M0), and the clinical data was complete. Those who had other malignant tumors at the same time or in the past, and received neoadjuvant radiochemotherapy or immunotherapy before surgery were excluded. In this retrospective case-control study, 550 patients with advanced gastric cancer who underwent curative gastrectomy between September 2017 and June 2019 were selected from the Affiliated Hospital of Qingdao University for modeling and internal verification, including 262 (47.6%) PNI positive and 288 (52.4%) PNI negative patients. According to the same standard, clinical data of 50 patients with advanced gastric cancer who underwent radical surgery from July to November 2019 in Qingdao Municipal Hospital were selected for external verification of the model. There were no statistically significant differences between the clinical data of internal verification and external verification (all P>0.05). Univariate analysis and multivariate logistic regression analysis were used to determine the independent risk factors for PNI in advanced gastric cancer, and the clinical indicators with statistically significant difference were used to establish a preoperative nomogram model through R software. The Bootstrap method was applied as internal verification to show the robustness of the model. The discrimination of the nomogram was determined by calculating the average consistency index (C-index). The calibration curve was used to evaluate the consistency of the predicted results with the actual results. The Hosmer-Lemeshow test was used to examine the goodness of fit of the discriminant model. During external verification, the corresponding C-index index was also calculated. The area under ROC curve (AUC) was used to evaluate the predictive ability of the nomogram in the internal verification and external verification groups. Results: A total of 550 patients were identified in this study, 262 (47.6%) of which had PNI. Multivariate logistic regression analysis revealed that carcinoembryonic antigen level ≥ 5 μg/L (OR=5.870, 95% CI: 3.281-10.502, P<0.001), tumor length ≥5 cm (OR=5.539,95% CI: 3.165-9.694, P<0.001), mixed Lauren classification (OR=2.611, 95%CI: 1.272-5.360, P=0.009), cT3 stage (OR=13.053, 95% CI: 5.612-30.361, P<0.001) and the presence of lymph node metastasis (OR=4.826, 95% CI: 2.729-8.533, P<0.001) were significant independent risk factors of PNI in advanced gastric cancer (all P<0.05). Based on these results, diffused Lauren classification and cT4 stage were included to establish a predictive nomogram model. CEA ≥ 5 μg/L was for 68 points, tumor length ≥ 5 cm was for 67 points, mixed Lauren classification was for 21 points, diffused Lauren classification was for 38 points, cT3 stage was for 75 points, cT4 stage was for 100 points, and lymph node metastasis was for 62 points. Adding the scores of all risk factors was total score, and the probability corresponding to the total score was the probability that the model predicted PNI in advanced gastric cancer before surgery. The internal verification result revealed that the AUC of nomogram was 0.935, which was superior than that of any single variable, such as CEA, Lauren classification, cT stage, tumor length and lymph node metastasis (AUC: 0.731, 0.595, 0.838, 0.757 and 0.802, respectively). The external verification result revealed the AUC of nomogram was 0.828. The C-ndex was 0.931 after internal verification. External verification showed a C-index of 0.828 from the model. The calibration curve showed that the predictive results were good in accordance with the actual results (P=0.415). Conclusion: A nomogram model constructed by CEA, tumor length, Lauren classification (mixed, diffuse), cT stage, and lymph node metastasis can predict the PNI of advanced gastric cancer before surgery.