喉癌患者血清蛋白质谱人工神经网络预测模型的初步研究

Pilot study on serum protein finger printing coupled with ANN to establish diagnostic pattern of laryngeal cancer

目的应用表面增强激光解析电离飞行时间质谱技术(surface enhanced laser desorption/ionizationtime-of-flight mass spectrometry,SELDI-TOF-MS)分析喉癌患者与对照人群的血清蛋白质谱,筛选喉癌患者血清的差异表达蛋白,并利用人工神经网络(artificial neuralnetwork,ANN)建立血清蛋白质分子诊断模型,以期构建可用于喉癌早期诊断敏感和特异的新方法。方法采集血清标本后利用SELDI-TOF-MS技术检测血清蛋白质谱数据,将获得的蛋白质谱图用Ciphergen ProteinChip 3.0软件进行数据的校正和分析,筛选喉癌患者组与对照组差异蛋白。利用筛选的差异蛋白作为标志物,结合ANN技术建立预测模型,评价该模型在喉癌诊断中的价值。结果喉癌组与对照组有79个差异蛋白质,其中差异有显著意义的蛋白质峰(分子量2000～20 000 Da,t=5.143,P<0.05)共24个。经过反复训练,筛选其中9个明显差异表达蛋白建立ANN诊断模型,灵敏度(SEN)为87.1%,特异度(SPE)84.8%,诊断指数为171.9%,其中区分喉癌与癌前病变准确率为100%。结论利用SELDI-TOF-MS技术筛选出的ANN蛋白质分子诊断模型能够较准确的区分喉癌与非喉癌人群,其在喉癌的诊断和血清肿瘤特异性蛋白质生物标志物的筛选方面具有一定临床应用价值。

OBJECTIVE To study serum protein of normal persons and patients with laryngeal cancer by using surface enhanced laser desorption/ionization time-of-flight mass spectrometry（SELDI-TOF-MS） and to screen the different expressed protein.Then,using the artificial neural network（ANN）,to construct a molecular diagnostic SELDI model and new method which has superiority in sensitivity and specificity for early detection of laryngeal cancer.METHODS The SELDI-TOF-MS and gold protein chip were performed to detect mass spectrogram for serum protein signature analysis.The mass spectrum of the protein was analyzed and corrected with Ciphergen ProteinChip 3.0 software.Then,the different expressed markers were screened from the maps by Biomarker Wizard 3.1 software and used to build an ANN model.RESULTS In total,there were 79 different expressed protein peaks were detected between the group of laryngeal cancer and healthy controls.Among them,there were 24 proteins of which the protein peaks had significant differences（t =5.143,P 0.05）.After training,nine proteins with significant differences were selected to set up an ANN model,of which the sensitivity（SEN）was 87.1%,the specificity（SPE）was 84.8%,the diagnostic index was 171.9%,and the accuracy to distinguish the laryngeal cancer and premalignant laryngeal lesions was 100%.CONCLUSION The molecular diagnostic SELDI model could distinguish the patients with laryngeal cancer accurately from normal ones.And it is useful for SELDI-TOF-MS technology to diagnose laryngeal cancer and select tumor-specific serum protein biomarker.