Objective: To investigate the risk factors for lymph node metastasis in patients with early gastric cancer and establish a model for prediction of risk. Methods: The cohort of this retrospective observational study comprised 1096 patients who had undergone radical gastric cancer surgery combined with standard D1 lymphadenectomy and been diagnosed with early gastric cancer by postoperative pathology in Zhongshan Hospital affiliated with Fudan University from January 2016 to July 2022. The patients were allocated to groups with and without lymph node metastases. Clinicopathological characteristics were compared between the two groups and multi-factor logistic regression analysis used to identify independent risk factors for lymph node metastasis in patients with early gastric cancer. Indications for endoscopic resection in the Japanese Gastric Cancer Association (JGCA) guideline were also incorporated into construction of the model. The patient cohort was divided into training and validation sets in a 6:4 ratio. The identified independent risk factors were used to construct a predictive nomogram. Receiver operating characteristic curves were plotted separately and the difference between them in predictive efficacy was compared using the area under the curve (AUC). Results: A total of 1,096 patients with early gastric cancer were included, with 750 males and 346 females. Their average age was (61.4±10.9) years old, and the mean tumor diameter was (23.8±11.4) mm. Among them, 188 patients (17.2%) had positive lymph node metastasis, with 109 cases in N1 stage, 42 cases in N2 stage, and 37 cases in N3 stage. Additionally, 462 patients were in T1a stage, while 634 patients were in T1b stage. Univariate analysis showed that tumor diameter, location, Lauren classification, gross morphology, histological type, intravascular invasion, ulceration, differentiation type and tumor T stage were associated with lymph node metastasis after radical gastrectomy for early gastric cancer (all P<0.05). Multifactorial analysis showed that the presence of intravascular invasion (OR=14.822, 95%CI: 9.323-23.572, P<0.001), undifferentiated type (OR=3.095, 95%CI: 1.649-5.811, P<0.001), tumor T1b (OR=1.798, 95%CI: 1.053-3.079, P=0.032), and tumor diameter ≥2 cm (OR=1.229, 95%CI: 1.031-1.469, P=0.022) were independent risk factors for lymph node metastasis. The baseline data of the training set and validation set were consistent in terms of balance (all P>0.05). We used the above variables to establish a predictive nomogram for lymph node metastasis in patients with early gastric cancer. The AUC values obtained from the validation of the model in the training and validation sets were 0.880 (95%CI: 0.849-0.911) and 0.881 (95%CI: 0.841-0.921), respectively, and were significantly better than the predictive efficacy based on the JGCA guideline (AUC=0.777, 95%CI: 0.746-0.809, P<0.001). Conclusions: Patients with early gastric cancer and intravascular invasion, undifferentiated tumors, tumor T1b, and diameter ≥2 cm are at higher risk of postoperative lymph node metastasis than other patients. The predictive model developed in this study more accurately predicts lymph node metastasis in patients with early gastric cancer than previously proposed methods.
目的： 分析早期胃癌淋巴结转移的危险因素，并建立列线图预测模型。 方法： 采用回顾性观察性研究方法，收集2016年1月至2022年7月在复旦大学附属中山医院接受胃癌根治术联合标准D1淋巴结清扫，且术后病理为早期胃癌患者的临床资料。根据有无淋巴结转移将患者分为转移组与无转移组，比较两组患者临床病理特征，并将差异有统计学意义的数据纳入多因素logistic回归分析，进一步评估早期胃癌发生淋巴结转移的独立危险因素。按6∶4的比例将患者分为训练集和验证集。选取独立危险因素建立列线图预测模型，同时选取日本胃癌协会（JGCA）指南中内镜切除适应证的相关因素以相同方式建立模型，分别绘制受试者工作特征（ROC）曲线，通过曲线下面积（AUC）比较两者预测效能差异。 结果： 共纳入1 096例早期胃癌患者，男性750例，女性346例；年龄（61.4±10.9）岁；肿瘤长径（23.8±11.4）mm。其中淋巴结转移阳性患者188例（17.2%），N1、N2和N3期分别为109、42和37例，T1a和T1b期分别为462和634例。单因素分析显示，肿瘤长径、肿瘤位置、肿瘤Lauren分型、肿瘤大体形态、肿瘤组织学分型、有无脉管侵犯、有无溃疡、肿瘤分化类型和肿瘤T分期与早期胃癌根治术后淋巴结转移有关（均P<0.05）。Logistic多因素分析显示，脉管侵犯（OR=14.822，95%CI：9.323~23.572，P<0.001）、肿瘤未分化（OR=3.095，95%CI：1.649~5.811，P<0.001）、肿瘤T1b期（OR=1.798，95%CI：1.053~3.079，P=0.032）和肿瘤长径≥2 cm（OR=1.229，95%CI：1.031~1.469，P=0.022）均是影响早期胃癌淋巴结转移的独立危险因素。训练集与验证集的基线均衡性一致（均P>0.05）。基于以上变量所构建的早期胃癌淋巴结转移列线图预测模型在训练集和验证集验证所得的AUC值分别为0.880（95%CI：0.849~0.911）和0.881（95%CI：0.841~0.921），且明显优于基于JGCA指南内镜切除适应证的预测效能（AUC=0.777，95%CI：0.746~0.809，P<0.001）。 结论： 对于存在脉管侵犯，肿瘤未分化、T1b期和长径≥2 cm的早期胃癌患者，应警惕术后淋巴结转移风险；本研究构建的预测模型能够更有效地预测早期胃癌淋巴结转移。.