Fenollar Florence, Goncalves Anthony, Esterni Benjamin, Azza Said, Habib Gilbert, Borg Jean-Paul, Raoult Didier
Unite des Rickettsies, Centre National de la Recherche Scientifique 6020, Institut Federatif de Rechercher 48, Faculte de Medecine, Marseille, France.
J Infect Dis. 2006 Nov 15;194(10):1356-66. doi: 10.1086/508429. Epub 2006 Oct 11.
Bacterial endocarditis is a serious disease. Surface-enhanced laser desorption/ionization time-of-flight (SELDI-TOF) mass spectrometry (MS) based on serum protein profiling is a powerful approach that can generate biomarkers with diagnostic value.
To identify a protein signature associated with bacterial endocarditis, we retrospectively performed SELDI-TOF MS profiling of serum samples from 88 patients hospitalized because of clinical suspicion of endocarditis. The diagnosis was confirmed by conventional criteria for 34 patients (endocarditis positive) and was excluded for 54 patients (endocarditis negative). Serum samples were incubated with cation-exchange ProteinChip arrays. The protein profiles generated were subjected to biostatistical processing.
Fifty-nine samples (23 endocarditis positive and 36 endocarditis negative) were randomly selected for a learning set, with the 29 remaining samples (11 endocarditis positive and 18 endocarditis negative) serving as an independent testing (validation) set. Sixty-six protein peaks were differentially expressed between the endocarditis-positive and the endocarditis-negative patients. By combining partial least squares and logistic regression methods, we built a serum protein model that perfectly discriminated between endocarditis-positive and endocarditis-negative patients. Importantly, when this model was tested on the independent testing set, a correct prediction rate of nearly 90% was demonstrated. Overall, sensitivity, specificity, positive predictive value, and negative predictive value were 94%, 98%, 96%, and 96%, respectively.
SELDI-TOF MS profiling revealed a serum signature with high diagnostic potential for endocarditis.
细菌性心内膜炎是一种严重疾病。基于血清蛋白谱的表面增强激光解吸/电离飞行时间（SELDI-TOF）质谱分析法是一种强大的方法，可生成具有诊断价值的生物标志物。
为了识别与细菌性心内膜炎相关的蛋白质特征，我们回顾性地对88例因临床怀疑心内膜炎而住院的患者的血清样本进行了SELDI-TOF质谱分析。根据传统标准，34例患者诊断得到证实（心内膜炎阳性），54例患者被排除（心内膜炎阴性）。血清样本与阳离子交换蛋白芯片阵列孵育。对生成的蛋白质谱进行生物统计学处理。
随机选择59个样本（23个心内膜炎阳性和36个心内膜炎阴性）作为训练集，其余29个样本（11个心内膜炎阳性和18个心内膜炎阴性）作为独立测试（验证）集。心内膜炎阳性和心内膜炎阴性患者之间有66个蛋白峰差异表达。通过结合偏最小二乘法和逻辑回归方法，我们建立了一个血清蛋白模型，该模型能完美地区分心内膜炎阳性和心内膜炎阴性患者。重要的是，当在独立测试集上测试该模型时，显示出近90%的正确预测率。总体而言，敏感性、特异性、阳性预测值和阴性预测值分别为94%、98%、96%和96%。
SELDI-TOF质谱分析揭示了一种对心内膜炎具有高诊断潜力的血清特征。
