Currently, the mainstays of treatment for patients with EGC are surgical treatment and endoscopic resection. Compared with surgical resection, endoscopic resection has the advantages of causing less trauma and yielding higher quality of life after surgery [24, 25] and is the preferred treatment modality for patients with EGC. However, because lymph node dissection cannot be achieved by endoscopic resection, the risk of recurrence after surgery is higher than that after surgical resection; thus, ECG patients with LNM still need to undergo surgery for radical tumour resection. It is very important to accurately predict the risk of LNM, and understanding the metastatic status of lymph nodes in EGC is conducive to selecting the appropriate surgical method and improving the overall efficacy of treatment.
To fully explore the risk factors for LNM in EGC, we examined a total of 13 variables; we determined that tumour size, tumour location, degree of tumour differentiation, and tumour pathological type were independent risk factors for LNM in patients with EGC and integrated them into the prediction model in the form of a nomogram. Subsequently, we evaluated the discriminatory ability and calibration of the model and performed internal validation.
Previous large-sample studies have reported LNM rates in EGC ranging from 16.7%~25.37% [26, 27]. In this study, the LNM rate in the entire cohort of EGC patients was 20.1% (201/1000); additionally, the LNM rate in the training group was 19.7%, and the LNM rate in the validation group was 21.2%, all of which are similar to the previously reported results. Many factors affect LNM in EGC, and the identified risk factors vary among other related studies; however, almost all studies have found that the depth of tumour invasion and tumour size are significantly correlated with LNM [28, 29]. Based on preoperative clinicopathological data, we analysed the risk factors for LNM in EGC, and the results showed that tumour size, tumour location, degree of differentiation and pathological type were independent predictive risk factors for LNM in EGC. In this study, the LNM rate among patients with tumours ≤ 2 cm was 11.4% (64/563), while that among patients with tumours > 2 cm was as high as 31.4% (137/437) (P < 0.05), which is similar to the findings of other researchers. Du MZ [30] et al. found not only that independent risk factors for LNM in EGC included tumour size ≥ 3.0 cm but also that the tumours were significantly larger in patients with than without LNM. To date, most studies have divided the tumour location into the upper, middle and lower thirds of the stomach, and the results of the multivariate analysis in this study indicated that tumour location is an independent risk factor for LNM in EGC (P < 0.05), with tumours in the stomach body, antrum, and pyloric region carrying greater risk of LNM than those in the cardia and gastric fundus, which is consistent with the results reported by Cui [22] et al.. Wang et al. also found that LNM may be more likely to occur in the lower part of the stomach, and they hypothesized that this tendency may be related to the occurrence of ulcerated undifferentiated invasive carcinoma or submucosal carcinoma, common in the antrum, and vascular invasion, as well as other forms of EGC treated by EMR, which were not included in their study. The degree of tumour differentiation is also a risk factor for LNM in patients with EGC. In a retrospective study of 503 EGC patients, Wang et al. found that more than half of patients with LNM had undifferentiated EGC and that the degree of tumour differentiation was an independent risk factor for LNM in EGC, which is consistent with our results. The results of this study showed that the lower the degree of tumour differentiation, the higher the rate of LNM, with a significant difference between the designated groups (P < 0.05). Signet ring cell carcinoma is a special histopathological type of gastric cancer, with general characteristics including poor differentiation, high aggressiveness, early metastasis, rapid progression and poor prognosis. In this study, the LNM rate in gastric adenocarcinoma was 14.4% (100/640), the LNM rate in signet ring cell carcinoma was 30.3% (96/360), and the LNM rate in early signet ring cell carcinoma was higher than that in non-ring cell carcinoma, which is consistent with the results reported by Nie [31] et al.
The risk factors associated with LNM in patients with EGC were included in the survival analysis, and the results showed that the pathological type and LNM status were correlated with the prognosis of patients with EGC, with a poorer prognosis for patients with signet ring cell carcinoma and LNM.
Based on the preoperative clinicopathological data of patients, we analysed the risk factors for LNM in EGC and then constructed a nomogram based on the results, which can visually display the risk of LNM. Then, we verified the predictive ability through ROC curve analysis. The AUC in the training set was 0.75 (95% CI: 0.701 ~ 0.789), and the AUC in the validation set was 0.763 (95% CI: 0.687 ~ 0.838). These results show that the prediction model has a good ability to distinguish whether LNM will occur in EGC. Additionally, the calibration curve of the model showed a high degree of fit between the predicted and actual probabilities, which indicates that the model is well calibrated and can serve as a basis for selecting the appropriate procedure. Although there have been numerous studies worldwide on nomograms for the prediction of LNM in EGC, the nomograms established in these studies include the depth of tumour invasion, the number of metastatic lymph nodes, the presence of vascular invasion, the presence of lymphangiosarcoma thrombosis, the presence of nerve invasion, and other clinicopathological data that can only be obtained after surgery. As such, these prediction models cannot serve as a reference for selecting an appropriate treatment method in patients with EGC before surgery. To our knowledge, this is the first nomogram to be constructed to predict LNM in EGC patients based on relevant risk factors that are available preoperatively. We recommend that all patients with gastric cancer undergo preoperative endoscopy and pathological biopsy to facilitate selection of the appropriate treatment.
Of course, there are some limitations to this study. (1) This was a single-centre retrospective study with certain selection bias. (2) The time span of patient enrolment patients is large, and with the continuous development of comprehensive modalities for the diagnosis and treatment of gastric cancer, the detection of LNM in EGC may also be affected by factors such as the extent of resection, the scope of intraoperative lymph node dissection, the pathological detection method and postoperative pathologist’s experience, which affect the results of LNM detection and lead to the occurrence of false negatives. (3) Due to the lack of preoperative clinical data such as imaging findings, specific tumour marker levels, and endoscopic ultrasound reports, these data could not be included in the present analysis. Finally, with the establishment and improvement of the standardized gastric cancer database at our centre, future research could incorporate data from multiple centres or more indicators to improve the diagnostic efficiency of the prediction model. Such additional data could include radiomic characteristics [32] and the results of cutting-edge sequencing technology to meet the urgent needs of precision medicine.