| 摘要: |
| 目的：建立十二指肠乳头肿瘤术前病变性质预测模型，并筛选主要预测因子。方法：采用病例对照研究设计，以十二指肠乳头恶性肿瘤患者为病例组，十二指肠乳头良性肿瘤患者为对照组。研究对象均来自长海医院。通过查阅病历和面谈等方式调查其人口学特征、初诊时的主要临床表现、实验室检查结果和影像学检查结果。采用χ2检验、t检验或方差分析等方法进行单因素分析。选择单因素分析中P值小于或等于0.25的因素进行多因素分析，建立病变性质Logistic回归预测模型。结果：共纳入199例经病理证实的十二指肠乳头肿瘤患者。其中，病例组166例，对照组33例。血红蛋白含量(Hb)、总胆红素(Tbil)、直接胆红素(Dbil)、门冬氨酸转氨酶(AST)、碱性磷酸酶(AKP)、γ-谷氨酰转肽酶(GGT)和癌胚抗原(CEA)是病变性质的独立预测因子，其OR值(95%可信限)分别为0.981(0.959~1.003)、0.867(0.794~0.948)、1.207(1.075~1.355)、1.028(1.008~1.048)、0.996(0.992~1.000)、1.002(1.000~1.004)和0.974(0.953~0.994)。结论：包含Hb、Tbil、Dbil、AST、AKP、GGT和CEA的Logistic回归模型可以在术前较为准确地预测十二指肠乳头肿瘤的病变性质，其临床应用价值有待于进一步验证。 |
| 关键词: 十二指肠乳头肿瘤 预测模型 Logistic回归 |
| DOI：10.3724/SP.J.1008.2008.00376 |
| 投稿时间：2008-03-25修订日期：2008-03-25 |
| 基金项目: |
|
| Preoperative predictors for nature of duodenal papillary tumors |
| LI Jing,CAI Quan-cai,ZHU Wei,LI Zhao-shen* |
| (Department of Gastroenterology,Changhai Hospital,Second Military Medical University,Shanghai 200433,China) |
| Abstract: |
| Objective： To establish a preoperative forecasting model for the nature of duodenal papillary tumors and to discuss its main predictors. Methods: A case-control study was conducted; the case group included patients with malignant duodenal papillary tumors and the control group included patients with benign duodenal papillary tumors. All the patients were from Changhai Hospital. Data of patients,including the demographic characteristics,clinical symptoms during onset,laboratory findings, and radiological data,were collected by face-to-face interviews or by reviewing the medical history. Chi-square,t-test or ANOVA were employed to performed univariate analysis. All factors with P values less than or equal to 0.25 in the univariate analysis were used as independent variables for multivariate analysis,and a Logistic regression forecasting model for the nature of duodenal papillary tumors was established. Results: Totally 199 patients with pathologically-confirmed duodenal papillary tumors were included in the present study,with 166 in the case group and 33 in the control group. Multivariate analysis showed that hemoglobin(Hb),total bilirubin(Tbil),direct bilirubin(Dbil),aspartate transferase(AST),alkaline phosphatase(AKP),gamma glutamyl transpeptidase(GGT),and carcinoembryonic antigen(CEA) were independent predictors for nature of duodenal papillary tumors,with their odds ratios(95% confidence interval) being 0.981(0.959,1.003),0.867(0.794,0.948),1.207(1.075,1.355),1.028(1.008,1.048),0.996(0.992,1.000),1.002(1.000,1.004),and 0.974(0.953,0.994),respectively. Conclusion: The Logistic regression model,which takes into consideration of Hb,Tbil,Dbil,AST,AKP,GGT,and CEA,can be used to predict the nature of duodenal papillary tumors,and its clinical value need to be further studied. |
| Key words: duodenal papillary tumor predictive model Logistic regression |
