胃癌腹膜转移风险评估模型的建立及其预测价值  

Establishment of risk evaluation model of peritoneal metastasis in gastric cancer and its predictive value

作　　者：赵骏杰[1] 周荣健[1] 张启[1] 束平[1] 李豪杰[1] 汪学非[1] 沈振斌[1] 刘凤林[1] 陈伟东[1] 秦净[1] 孙益红[1] 

机构地区：[1]复旦大学附属中山医院普通外科,上海200032

出　　处：《中华胃肠外科杂志》2017年第1期47-52,共6页Chinese Journal of Gastrointestinal Surgery

基　　金：国家自然科学基金（31470794,31300671,31670806）

摘　　要：目的 建立胃癌腹膜转移评估模型并评价其临床意义。方法 回顾性分析2015年4月至2015年12月复旦大学附属中山医院普通外科收治胃癌患者的临床病理资料，排除18例出现非腹膜转移的远处部位转移患者后，最终710例患者被纳入研究。采用单因素和多因素的方法分析影响胃癌腹膜转移的相关因素，将多因素分析中有意义的影响因素作为指标，通过R软件的＂rms＂包构建列线图（nomogram），构建胃癌腹膜转移的危险评估模型，并得到各项指标的评分。将每个患者的各项指标对应的分数相加，得到模型的总分；总分越高，对应的胃癌腹膜转移风险越高。采用受试者工作特征曲线（ROC）计算风险评估所建模型预测胃癌腹膜转移的准确性，用Delong. Delong. Clarke-Pearson检验比较AUC的差异，当ROC曲线下面积（AUC）最大时，对应的风险阈值使模型具有最佳的敏感性和特异性。结果 710例患者中，腹膜转移者47例（6.6%），其中男性30例（30/506，5.9%），女性17例（17/204, 8.3%）；≥60岁者31例（31/429, 7.2%）；肿瘤≥3 cm者38例（38/461，8.2％）。Lauren分型为肠型者2例（2/245，0.8%），混合型者8例（8/208，3.8%），弥漫型者11例（11/142，7.7%），其余不详；糖类抗原（CA） 19-9≥37 kU/L者13例（13/61 ，21.3%）；CA125≥35 kU/L者11例（11/36, 30.6%）；CA72-4≥10 kU/L者11例（11/39, 28.2%）；中性粒细胞／淋巴细胞计数比（NLR）≥2.37者26例（26/231, 11.3%）。多因素分析显示，Lauren分型（HR= 8.95 ，95%CI：1.32~ 60.59，P= 0.025）、CA125（HR= 17.45 ，95%CI：5.54~ 54.89 ，P= 0.001）、CA72-4（HR= 20.06，95%CI：5.05~ 79.68 ，P= 0.001）、NLR （HR= 4.16，95%CI：1.17~ 14.75，P= 0.032）是胃癌腹膜转移的独立危险因素。根据列线图分析建立的腹膜转移风险评估模型总分为241分，包括弥漫型或混合型（54分）、CA125≥35 kU/L （66分）、CA72Objective To establish an evaluation model of peritoneal metastasis in gastric cancer, and to assess its clinical significance.Methods Clinical and pathologic data of the consecutive cases of gastric cancer admitted between April 2015 and December 2015 in Department of General Surgery, Zhongshan Hospital of Fudan University were analyzed retrospectively. A total of 710 patients were enrolled in the study after 18 patients with other distant metastasis were excluded. The correlations between peritoneal metastasis and different factors were studied through univariate （Pearson＇s test or Fisher＇s exact test） and multivariate analyses （Binary Logistic regression）. Independent predictable factors for peritoneal metastasis were combined to establish a risk evaluation model （nomogram）. The nomogram was created with R software using the ＇rms＇ package. In the nomogram, each factor had different scores, and every patient could have a total score by adding all the scores of each factor. A higher total score represented higher risk of peritoneal metastasis. Receiver operating characteristic （ROC） curve analysis was used to compare the sensitivity and specificity of the established nomogram. Delong. Delong. Clarke-Pearson test was used to compare the difference of the area under the curve （AUC）. The cut-off value was determined by the AUC, when the ROC curve had the biggest AUC, the model had the best sensitivity and specificity.Results Among 710 patients, 47 patients had peritoneal metastasis （6.6%） , including 30 male （30/506, 5.9%） and 17 female （17/204, 8.3%） ; 31 were ≥ 60 years old （31/429, 7.2% ） ; 38 had tumor ≥ 3 cm （38/461, 8.2% ）. Lauren classification indicated that 2 patients were intestinal type （2/245, 0.8% ） , 8 patients were mixed type （8/208, 3.8%） , 11 patients were diffuse type （11/142, 7.7% ） , and others had no associated data. CA19-9 of 13 patients was ≥ 37 kU/L （13/61, 21.3%） ; CA125 of 11 patients was ≥ 35 kU/L （11/36, 30.6%） ;

关 键 词：胃肿瘤 腹膜转移 风险评估模型 

分 类 号：R735.2[医药卫生—肿瘤]