应用IMAC3蛋白芯片分析食管鳞癌患者血清蛋白质谱的变化

背景与目的:血清蛋白指纹图谱技术有助于筛选与食管上皮癌变相关的分子变化。本研究通过比较食管鳞癌患者与正常对照血清蛋白表达谱的差异,筛选食管鳞癌相关血清蛋白标志物并建立诊断模型,同时探讨其在食管鳞癌血清学诊断中的意义。方法:收集68例食管鳞癌患者和44例正常对照的血清,其中建模型组90例(55例为食管鳞癌,35例为正常对照),盲法筛选组22例(13例为食管鳞癌,9例为正常对照)。采用固定金属亲和表面(immobilized metalaffinity capture,IMAC3)芯片,经表面增强激光解吸离子化飞行时间质谱(surface enhanced laser desorption/ionization-time of flight-mass spectrometry,SELDI-TOF-MS)测定得到蛋白质谱,通过Biomarker Wizard软件比较两组人群的血清蛋白质谱的差异,经生物信息学分析得到决策树模型并进行盲法验证。结果:在质荷比(m/z)1.5～20ku范围内,共检测到78个有效蛋白峰,其中25个峰差异有统计学意义(P<0.001)。对建模型组的蛋白质谱数据,通过1000次随机抽样,得到1000个决策树。结合3倍交叉证实方法,构建了"食管鳞癌-正常对照"分组诊断的20个决策树模型。用这20个决策树来对22个盲法筛选样本进行归类预测,预测正确的样本为18个,盲法验证的灵敏度为92.31%,特异度为66.67%。结论:应用SELDI-TOF-MS技术可以筛选出食管鳞癌相关的血清蛋白标志,建立的决策树模型可以对食管鳞癌做出较为准确的预测判断。

Serum proteomic spectra of esophageal squamous cell carcinoma patients analyzed with IMAC3 protein chip

BACKGROUND & OBJECTIVE: Serum protein fingerprinting technology can help to identify the molecular changes related to esophageal carcinogenesis. This study was to screen serum markers and establish the predictive models that may be of help to serologic diagnosis of esophageal squamous cell carcinoma (ESCC). METHODS: Serum samples were collected from 68 ESCC patients and 44 age-and sex-matched healthy subjects, and randomized into a training set (55 ESCC patients and 35 healthy subjects) and a blind testing set (13 ESCC patients and 9 healthy subjects). Serum samples were applied to immobilized metal affinity capture (IMAC3) proteinchip surfaces and tested by surface-enhanced laser desorption/ionization-time of flight-mass spectrometry (SELDI-TOF-MS). The data were analyzed by Biomarker Wizard software to screen serum proteomic biomarkers. Decision classification tree models were established by bioinformatics. Double-blind test was used to determine the sensitivity and specificity of the classification tree models. RESULTS: A total of 78 effective protein peaks were detected at the molecular range of 1.5 to 20 ku, among which 25 were significantly different between ESCC patients and healthy subjects (P<0.001). All the peptide pattern data were sampled randomly for 1,000 times using 3-cross validation approach, and 1,000 decision tree models were obtained. Twenty decision trees with the highest cross validation rate were chosen to construct the classification models which can differentiate ESCC patients from healthy subjects. With these decision trees, 18 samples were correctly forecasted from 22 blind testing samples, with a sensitivity of 92.31% and a specificity of 66.67%. CONCLUSION: SELDI-TOF-MS technique combined with decision tree model can help to identify serum proteomic biomarkers related to ESCC and the predictive models can discriminate ESCC patients from healthy people effectively.