胃癌高危因素评分模型构建及对筛查时机、方案选择的价值研究

背景与目的：伺机性筛查也称为个体筛查，是一种基于临床表征的筛查方法，花费少，患者依从性高，是目前提高我国早期胃癌检出率的可行途径。基于患者基线资料及血液学检查等常用指标，构建一套关于胃癌高危因素评分模型，探讨其对胃癌高危患者筛查时机、方案选择的价值，以期为临床高效筛查提供更多依据。方法：收集2014年6月—2017年12月甘肃省人民医院普外科收治的387例胃黏膜相关疾病患者为研究对象。收集幽门螺杆菌（Helicobacter pylori，HP）感染情况、血清胃蛋白酶原（pepsinogen，PG）Ⅰ及PGⅠ/Ⅱ等指标，采用病例-对照的研究方法，构建胃癌高危评分模型。结果：受试者工作特征（receiver operating characteristic，ROC）曲线显示，当PGⅠ为43.7 μg/L时，曲线下面积最大为0.736，其灵敏度为0.529，特异度为0.779。当PGⅠ/Ⅱ为2.2 μg/L时，曲线下面积最大为0.780，其灵敏度为0.578，特异度为0.849。将二者并联时，对胃癌诊断的灵敏度为71.8%、特异度为75.5%，可确定PGⅠ≤43.7 μg/L且PGⅠ/Ⅱ≤2.1 μg/L是最佳筛查临界值。单因素分析结果显示，两组患者的性别构成、年龄、饮用水类型、家族史、食用腌制品、HP感染、PGⅠ及PGⅠ/Ⅱ等差异有统计学意义（P＜0.05）。进一步行多因素Logistic分析发现，患者性别、饮用水类型、HP感染、家族史、PGⅠ、PGⅠ/Ⅱ及年龄是影响患者胃癌发生的独立危险因素（P＜0.05）。在Logistic分析基础上，对各危险因素进行赋值，建立评分模型：Y=A×年龄+30×性别+30×饮用水+30×HP（+）+50×家族史+B×PG水平（35~45岁：A=20；46~55岁：A=40；56~65岁：A=70；≥66岁：A=80。当PGⅠ≤43.7 μg/L且PGⅠ/Ⅱ＞2.1 μg/L：B=10；PGⅠ＞43.7 μg/L且PGⅠ/Ⅱ≤2.1 μg/L：B=30；PGⅠ≤43.7 μg/L且PGⅠ/Ⅱ≤2.1 μg/L：B=80）。根据构建模型对两组患者评分进行验证，结果发现，病例组评分［（209.78±46.98）分］显著高于对照组［（122.37±56.37）分］，差异有统计学意义（χ ² =13.962，P＜0.001）。ROC曲线显示，当临界值为156分时，曲线下面积最大为0.876，灵敏度为0.880，特异度为0.716，Youden指数=0.595。拟合优度经Hosmer-Lemeshow检验后发现，模型HL指标为13.492，P=0.095，表明模型拟合度较好。结论：根据建立的胃癌评分模型，对评分≥156且因消化道相关不适而就诊的患者，应视为高危人群，建议至少每年进行1次胃镜随访。

Construction of the risk factor model of gastric cancer and its value in screening opportunity and options

Background and purpose: The opportunistic screening, also known as individual screening, is a clinic-based screening method, which is a feasible way to improve the detection rate of early gastric cancer in China. Based on the data of patients and hematological indexes, we constructed a scoring model of high risk factors for gastric cancer, and explored its value inscreening gastric cancer patients. Methods: Three hundred and eighty-seven patients were enrolled in this study. We investigated the Helicobacter pylori (HP) infection, serum pepsinogen (PG)Ⅰ, PGⅠ/Ⅱ level and other indexes. The high risk assessment model of gastric cancer was constructed. Results: The receiver operating characteristic (ROC) curve showed that the maximum area under the curve was 0.736 when PGⅠ was 43.7 μg/L. When PGⅠ/Ⅱ was 2.2 μg/L, the maximum area under the curve was 0.780. PG=43.7 μg/L and PGⅠ/Ⅱ=2.1 μg/L were the best screening critical values. Gender, drinking water type, HP infection, family history, PGⅠ, PGⅠ/Ⅱ level and age were independent risk factors for gastric cancer in the patients (P<0.05). On the basis of logistic regression analysis, the risk factors were assigned and the scoring model was established: Y=A×age +30×gender +30×drinking water +30×HP (+) +50×family history +B×PG level (35-45 years: A=20; 46-55 years: A=40; 56-65 years: A=70; ≥66 years: A=80. PGⅠ≤43.7 μg/L and PGⅠ/Ⅱ>2.1 μg/L: B=10; PGⅠ>43.7 μg/L and PGⅠ/Ⅱ≤2.1 μg/L: B=30; PGⅠ≤43.7 μg/L and PGⅠ/Ⅱ≤2.1 μg/L: B=80). According to the construction model, the scores of the two groups were verified. The results showed that the score of case group (209.78±46.98) was significantly higher than that of the control group (122.37±56.37) (χ ² =13.962, P<0.001). The ROC curve showed that the maximum area under the curve was 0.876 when the critical value was 156, and the Youden index was 0.595. After Hosmer-Lemeshow test, it was found that the fitting degree of the model was better. Conclusion: According to the scoring model of gastric cancer, the patients whose score ≥156 with gastrointestinal discomfort, should be regarded as high-risk population, and it is recommended to have a follow-up gastroscopy at least once a year.