血清蛋白质谱结合人工神经网络在头颈部鳞癌诊断中的研究

Serum Diagnosis of Head and Neck Squamous Cell Carcinoma Using Surface-enhanced Desorption Ionization Mass Spectrometry and Artificial Neural Network Analyses

背景与目的:头颈部鳞癌早诊率的高低是影响治疗效果的关键因素之一。本研究应用血清蛋白质谱技术结合人工神经网络建立头颈部鳞癌(head and neck squamous cell carcinoma,HNSCC)患者的诊断模型,并评价其诊断价值。方法:对74例头颈部鳞癌患者和146例健康人的血清样本随机分为训练集(148例)和测试集(72例)。首先应用表面加强激光解吸电离-飞行时间质谱(surface-enhanced desorption ioniza tiontime-of-flight mass spectrometry,SELDI-TOF-MS)技术及WCX2(weak cation-exchange)芯片检测训练集样本,结合反向传播人工神经网络(artificial neural network,ANN)的方法建立诊断模型,进一步检测测试集样本并评价该模型的诊断价值。结果:在训练集样本中筛选出4个m/z(质荷比)峰(4469u,5924u,8926u,16697u)区分头颈部鳞癌和对照人群的敏感度和特异度均达100.0%(148/148);用该模型对测试集样本进行双盲法预测的敏感度为85.7%(18/21),特异度为96.1%(49/51)。结论:该模型诊断头颈部鳞癌具有较高的灵敏度和特异度,值得进一步研究。

BACKGROUND ＆ OBJECTIVE： The early diagnosis of head and neck squamous cell carcinoma （HNSCC） is the key factor that affecting the treatment result. We performed surface-enhanced desorption ionization time-of-flight mass spectrometry （SELDI-TOF-MS） using a multi-layer artificial neural network （ANN） to develop and evaluate a proteomic diagnosis approach for HNSCC. METHODS： Serum samples from 74 HNSCC patients and 146 healthy individuals were randomized into training set （148 samples） and test set （72 samples）. At first, we detected the training set of samples using SELDI mass spectrometry and WCX2 （weak cation-exchange） chips. Using a multi-layer ANN with a back propagation algorithm, we identified a proteomic pattern that could discriminate cancer from control samples in the training set. The discovered pattern was then used to determine the accuracy of the classification system in the test set. RESULTS： Four top-scored peaks, at m/z （mass/charge） ratio of 4 469 u, 5 924 u, 8 926 u, and 16 697 u, were finally selected as the potential biomarkers for detection of HNSCC with both sensitivity and specificity of 100.0% in the training set. The classifier predicted the HNSCC with sensitivity of 85.7% （18/21） and specificity of 96.1%（49/51） in the test set. CONCLUSION, SELDI profiling is a useful tool to accurately identify patients with HNSCC.