液相色谱-四极杆/飞行时间高分辨质谱组合化学计量学方法快速筛查婴幼儿配方乳粉中化学危害物

Chemometric approach based on liquid chromatographyquadrupole / time of flight high resolution mass spectrometry to screen chemical hazards in infant formula

采用液相色谱-四极杆/飞行时间(LC-Triple TOF~)高分辨质谱组合化学计量学方法建立了婴幼儿配方乳粉中化学危害物的快速筛查模型。实验采用人工添加外源化学物质(沙丁胺醇、林可霉素、磺胺嘧啶、螺旋霉素、双氯西林和醋酸甲地孕酮)模拟婴幼儿配方乳粉中未知化学危害物污染,15种不同的乳粉等量混合后用水复溶,配制代表性乳粉溶液样品。依据添加与否将样品分为参照组(不含添加的化合物)和添加组。乳粉溶液试样加入乙腈进行提取,经Captiva ND^(Lipids)固相萃取柱净化,CORTECS^(TM) UPLC C_(18+)柱液相色谱分离,以0.3%(v/v)甲酸-5%(v/v)水-乙腈溶液和0.3%(v/v)甲酸-5%(v/v)乙腈-水溶液为流动相梯度洗脱;飞行时间质谱全扫描(TOF MS)-信息依赖性采集(IDA)-子离子扫描(product ion scan)正离子模式采集样品数据。数据经MarkerViewTM软件预处理后导入SIMCA-P软件,对比分析参照组和添加组样品数据,拟合构建正交偏最小二乘法-判别分析(OPLS-DA)模型,并利用此模型进行样品组别区分和添加化合物识别。结果表明,实验构建的筛查模型拟合度和预测能力良好(R2X(cum)=0.742,R2Y(cum)=0.997,Q2Y(cum)=0.905),参照组和添加组样品被显著区分,6种添加的化合物(最低添加水平为50μg/kg,以乳粉计)全部被可靠地识别。该研究建立的方法为婴幼儿配方乳粉质量安全主动分析监控体系提供了有益技术参考。

A chemometric model for screening chemical hazards in infant formula has been built based on liquid chromatography coupled to quadrupole / time of flight（Triple TOF）mass spectrometry. Six chemicals,naming as salbutamol,lincomycin,sulfadiazine,spiramycin and megestrol acetate,were deliberately selected as the representatives of unknown chemical hazards to be spiked into infant formula. Fifteen different infant formula samples,with equal amount, were dissolved into water and mixed homogeneously to prepare test samplesrepresenting typical infant formula matrices. All test samples were divided into reference group（free from spiked chemicals）and spiked group. The test samples were extracted with acetonitrile,and then cleaned-up with Captiva NDLipidsfiltering solid phase extraction column. The liquid chromatographic separation was made on a CORTECSTMC18＋column using mobile phases of0. 3%（v / v）formic acid-5%（v / v）water in acetonitrile and 0. 3%（v / v）formic acid-5%（v / v）acetonitrile in water with gradient elution. Mass spectrometric analysis was conducted by a mixed acquisition mode of TOF full mass scan-information dependent acquisition（IDA）-product ion scan. Acquired data were exported to MarkerViewTMsoftware（version 1. 2. 1. 1,SCIEX,USA）for peak finding,alignment and filtering. Then the preprocessed data were subject to SIMCA-P software（version 14. 0,Umetrics AB,Sweden）for discriminant analysis. An orthogonal partial least squares-discriminant analysis（OPLS-DA）model was built with good fitness and good prediction ability（R2X（cum）= 0. 742,R2Y（cum）= 0. 997,Q2Y（cum）= 0. 905）. The reference and spiked samples were well grouped and the six chemicals（spiking level down to50 μg / kg,based on weight of infant formula）were reliably identified by S-Plot of OPLS-DA model. The technological system for monitoring quality and safety of infant formula could be benefited from the proposed approach.