蛋白质指纹图结合支持向量机生物信息学分析在早期胰腺癌检测中的应用  

Serum protein fingerprinting combined with artificial neural network in detection of early pancreatic cancer from general population

作　　者：刘建[1] 余捷凯[2] 邹璎[1] 毛捷鸿[1] 彭佳萍[2] 刘颖斌[3] 林汉庭[3] 郑树[2] 

机构地区：[1]浙江中医药大学附属第二医院外科,杭州310005 [2]浙江大学肿瘤研究所 [3]浙江大学附属第二医院外科

出　　处：《浙江医学》2012年第15期1251-1253,1257,共4页Zhejiang Medical Journal

基　　金：浙江省科技厅资助项目（2008C33067）

摘　　要：目的建立早期胰腺癌的血清蛋白质质谱模型。探讨蛋白质指纹图结合支持向量机生物信息学分析在早期胰腺癌诊断中的应用价值。方法选取用22例胰腺癌患者及20例健康者，采用表面增强激光解吸／电离时间飞行质谱（SELDI—TOF—MS）结合弱阳离子交换芯片（CM10）检测分析样本。采用支持向量机方法建立早期胰腺癌和健康者辨别模型。结果共筛选出11个蛋白质峰建立了早期胰腺癌和健康者辨别模型，辨别的敏感性为100％，特异性为95％。结论SELDI—TOF—MS技术结合生物信息学方法检测早期胰腺癌具有较高的敏感性和特异性。Objective TO evaluate the application of serum protein fingerprinting combined with artificial neural networking in detection of early pancreatic cancer. Methods A total of 42 serum samples were analyzed in this study, including 22 cases of early pancreatic cancer and 20 healthy subjects. The samples were first analyzed by SELDI-TOF-MS and the model for distinguishing two groups by support vector machine arithmetic （SVM） was constructed. Results Eleven protein peaks were chosen to establish distinguishing model for early diagnosis of pancreatic cancer. The specificity and sensitivity of the model in distinguishing early pancreatic cancer from healthy individuals were 95% and 100% respectively. Conclusion SELDI-TOF-MS technique combined with bioinformatics has a high sensitivity and specificity in detection of early pancreatic cancer.

关 键 词：早期诊断 胰腺癌 表面加强激光解吸/电离飞行时间质谱 支持向量机 蛋白质质谱 

分 类 号：R735.9[医药卫生—肿瘤]