应用SELDI-TOF-MS技术初步建立结直肠癌Ⅰ期分类树模型

Establishing a model of classification tree for stage Ⅰ colorectal cancer using surface-enhanced laser desorption ionization time-of-flight mass spectrometry

目的:分析结直肠癌Ⅰ期患者血清蛋白质组学变化,初步建立结直肠癌Ⅰ期分类树模型。方法:应用表面增强激光解析离子化飞行时间质谱仪(surface-enhanced laser desorption ionization time-of-flight mass spectrometry,SELDI-TOF-MS)和固相金属螯合(immobilized metal affinity capture,IMAC30-Cu2+)蛋白芯片检测血清样本得到质谱图,应用Biomarker Wizard软件对结直肠散发性中分化腺癌Ⅰ期31例和非癌组35例(其中健康志愿者16例,结直肠良性疾病患者19例)进行蛋白峰值鉴定和聚类。应用Biomarker Pattern软件分析得到分类树模型,在测试模式下行双盲验证。另外,应用电化学发光免疫测定法检测样本癌胚抗原(carcinoembryonic antigen,CEA)的表达。结果:软件识别了61个质峰,由质荷比为2 787、3 777、3 816、3 852、5 065、5 828、5 855和4 172等8个蛋白构成的分类树模型可以有效鉴别结直肠癌Ⅰ期与非癌组样本,在学习模式下敏感度、特异度分别是93.55%和91.43%,在测试模式下敏感度、特异度和阳性预测值分别是70.97%、82.86%和78.57%,明显优于CEA的检测,差异有统计学意义(P<0.05)。结论:SELDI-TOF-MS检测得到的血清蛋白组分类树模型可以准确鉴别结直肠癌Ⅰ期与非癌组,对结直肠癌Ⅰ期的筛查有重要的意义。

Objective：To make a proteomic analysis on the serum of stage I colorectal cancer patients and establish a classification tree model for identifying stage I colorectal cancer preliminarily.Methods：Surface enhanced laser desorption ionization time-of-flight mass spectrometry（SELDI-TOF-MS） and immobilized metal affinity capture（IMAC30 Cu2＋） protein biochips were used to detect the serum samples to obtain the mass spectra.The protein peak was identified and clustered using the Biomarker Wizard software in 16 healthy volunteers and 19 patients with colorectal benigh disease.The model of classification tree was generated using Biomarker Pattern software in the learning mode.The classification tree was then double validated in the testing mode.We additionally determined carcinoembryonic antigen（CEA） in all the serum samples using an electrochemiluminescent immunoassay method.Results：The software identified an ave-rage of 61 mass peaks/spectrum and classification tree constructed by 8 proteins with M/Z of 2 787,3 777,3 816,3 852,5 065,5 828,5 855 and 4 172 effectively identified stage Ⅰcolorectal cancer from non-malignant individuals.Its sensitivity and specificity were 93.55% and 91.43%,respectively in learning mode,and the corresponding sensitivity,specificity,and positive predicting value was 70.97%,82.86%,and 78.57%,respectively in testing mode,which were superior to CEA detection（P〈0.05）.Conclusion：Classification tree of serum proteins detects by SELDI-TOF-MS mass spectrometry correctly distinguishes stage I colo-rectal cancer from healthy and non-malignant individuals,and has significance for screening stage I colorectal cancer.