Lacombe Caroline, Untereiner Valérie, Gobinet Cyril, Zater Mokhtar, Sockalingum Ganesh D, Garnotel Roselyne
Université de Reims Champagne-Ardenne, Equipe MéDIAN, Biophotonique et Technologies pour la Santé, UFR de Pharmacie, 51 rue Cognacq-Jay, 51096 Reims, France.
Analyst. 2015 Apr 7;140(7):2280-6. doi: 10.1039/c4an01942c.
Classic galactosemia is an autosomal recessive metabolic disease involving the galactose pathway, caused by the deficiency of galactose-1-phosphate uridyltransferase. Galactose accumulation induces in newborns many symptoms, such as liver disease, cataracts, and sepsis leading to death if untreated. Neonatal screening is developed and applied in many countries using several methods to detect galactose or its derived product accumulation in blood or urine. High-throughput FTIR spectroscopy was investigated as a potential tool in the current screening methods. IR spectra were obtained from blood plasma of healthy, diabetic, and galactosemic patients. The major spectral differences were in the carbohydrate region, which was first analysed in an exploratory manner using principal component analysis (PCA). PCA score plots showed a clear discrimination between diabetic and galactosemic patients and this was more marked as a function of the glucose and galactose increased concentration in these patients' plasma respectively. Then, a support vector machine leave-one-out cross-validation (SVM-LOOCV) classifier was built with the PCA scores as the input and the model was tested on median, mean and all spectra from the three population groups. This classifier was able to discriminate healthy/diabetic, healthy/galactosemic, and diabetic/galactosemic patients with sensitivity and specificity rates ranging from 80% to 94%. The total accuracy rate ranged from 87% to 96%. High-throughput FTIR spectroscopy combined with the SVM-LOOCV classification procedure appears to be a promising tool in the screening of galactosemia patients, with good sensitivity and specificity. Furthermore, this approach presents the advantages of being cost-effective, fast, and straightforward in the screening of galactosemic patients.
经典型半乳糖血症是一种常染色体隐性代谢疾病，涉及半乳糖代谢途径，由1-磷酸半乳糖尿苷转移酶缺乏引起。半乳糖积累会在新生儿中引发多种症状，如肝病、白内障，若不治疗，败血症会导致死亡。许多国家已开发并应用新生儿筛查，采用多种方法检测血液或尿液中半乳糖或其衍生产物的积累。高通量傅里叶变换红外光谱法作为当前筛查方法中的一种潜在工具得到了研究。从健康、糖尿病和半乳糖血症患者的血浆中获取红外光谱。主要光谱差异出现在碳水化合物区域，首先使用主成分分析（PCA）进行探索性分析。PCA得分图显示糖尿病患者和半乳糖血症患者之间有明显区分，并且随着这些患者血浆中葡萄糖和半乳糖浓度的分别增加，这种区分更加明显。然后，以PCA得分作为输入构建支持向量机留一法交叉验证（SVM-LOOCV）分类器，并在来自三个群体的中位数、平均值和所有光谱上对该模型进行测试。该分类器能够区分健康/糖尿病、健康/半乳糖血症和糖尿病/半乳糖血症患者，灵敏度和特异度率在80%至94%之间。总准确率在87%至96%之间。高通量傅里叶变换红外光谱法与SVM-LOOCV分类程序相结合似乎是筛查半乳糖血症患者的一种有前景的工具，具有良好的灵敏度和特异度。此外，这种方法在筛查半乳糖血症患者方面具有成本效益高、快速且直接的优点。
