食品科学 ›› 2018, Vol. 39 ›› Issue (5): 226-233.doi: 10.7506/spkx1002-6630-201805034

• 营养卫生 • 上一篇    下一篇

基于Monte Carlo模拟法对液态乳中黄曲霉毒素M1的风险评估

耿梦梦1,2,徐明芳1,2,*,王 阳1,黎 明1,陈耕南1   

  1. 1.暨南大学生命科学技术学院,广东 广州 510632;2.暨南大学应急管理研究中心,广东 广州 510632
  • 出版日期:2018-03-15 发布日期:2018-03-14
  • 基金资助:
    暨南大学应急管理研究中心重大项目(JD2015008);广东省科技计划项目(2009B011300003)

Risk Assessment of Aflatoxin M1 in Milk by Monte Carlo Simulation

GENG Mengmeng1,2, XU Mingfang1,2,*, WANG Yang1, LI Ming1, CHEN Gengnan1   

  1. 1. College of Life Science and Technology, Jinan University, Guangzhou 510632, China; 2. Research Center of Emergency Management, Jinan University, Guangzhou 510632, China
  • Online:2018-03-15 Published:2018-03-14

摘要: 通过研究南方某市液态乳中黄曲霉毒素M1(aflatoxin M1,AFM1)污染水平及量化该市居民AFM1急性膳 食摄入风险,为其液态乳安全生产、消费及质量安全监管提供依据。基于不同品种奶源采集的379 份液态乳样品, 运用风险评估软件@Risk,构建基于Monte Carlo模拟法的非参数概率评估模型,对AFM1膳食摄入风险进行概率评 估。结果表明:采用高效液相色谱法检测的自2015年12月至2016年5月期间采集的379 份液态乳样品中,AFM1检 出率为19.79%(共75 份),污染平均值为0.19 μg/kg,最高污染水平为0.62 μg/kg。5 种不同品种的乳样中AFM1污 染平均值依次为样品D(0.22 μg/kg)>样品A(0.21 μg/kg)>样品C(0.20 μg/kg)>样品E(0.18 μg/kg)>样品B (0.13 μg/kg)。运用@Risk软件对不同液态乳中AFM1污染水平检测值进行Pareto、Expon、Laplace等分布拟合, 运用赤池信息量准则、贝叶斯信息准则、Chi-Squared、Anderson-Darling和Kolmogorov-Smirnov 5 种统计方法进 行拟合度检验。结果显示,不同液态乳样品中AFM1污染分布均符合Pareto分布,分别记为Risk Pareto(2.661 5, 0.029)、Risk Pareto(3.765 2,0.029)、Risk Pareto(3.370 8,0.029)、Risk Pareto(2.854 6,0.029)和Risk Pareto(2.423 5,0.029)。基于Monte Carlo模拟和Bootstrap抽样技术,根据液态乳AFM1污染分布特征,结合相关 的暴露参数,用@Risk软件进行模拟,当迭代次数为10 000 次时,通过饮用不同品种液态乳对AFM1日暴露量的平 均值依次为0.000 23、0.000 20、0.000 21、0.000 22、0.000 25 μg/(kg·d),第99百分位数的AFM1暴露量分别为 0.000 82、0.000 49、0.000 57、0.000 73、0.000 97 μg/(kg·d)。根据暴露评估模型,分别用联合暴露边界比和危 害指数表征不同人群通过饮用液态乳暴露AFM1的致癌和非致癌风险。结果表明不同人群通过饮用不同品种液态乳 暴露AFM1的致癌风险值均显著大于100,而当每日耐受摄入量为0.005 μg/(kg·d)时,非致癌风险危害指数均大 于1,说明针对同一奶源液态乳,虽然不同人群之间致癌和非致癌风险评估结果均存在显著差异,但不同人群通过 饮用液态乳摄入AFM1对健康的风险很小。本研究可为液态乳质量安全监管和今后系统开展液态乳质量安全风险评 估提供有益借鉴和参考。

关键词: 黄曲霉毒素M1, 液态乳, 健康风险评估, 概率评估模型

Abstract: In order to detect aflatoxin M1 (AFM1) pollution in milk from a certain city in south China and quantify the health risk of AFM1 via dietary intake of milk for residents in the city, 379 fresh milk samples from different cow breeds (A, B, C, D and E) were collected during December 2015 to May 2016. AFM1 concentration in milk samples was determined by high performance liquid chromatography and a nonparametric probabilistic evaluation model was fitted to evaluate the health risk associated with dietary intake (exposure) of AFM1 using the @Risk software package by Monte Carlo simulation. The results showed that 75 of these 379 samples (19.79%) were found positive for AFM1 with mean and maximum concentration of 0.19 and 0.62 μg/kg, respectively. The mean concentrations of AFM1 in different milk sources were in the following descending order: D (0.22 μg/kg) > A (0.21 μg/kg) > C (0.20 μg/kg) > E (0.18 μg/kg) > B (0.13 μg/kg). The AFM1 concentration values for all investigated samples were fitted with the Pareto, Expon and Laplace distribution models, and the goodness of fit of the developed equations were checked by five statistical methods of Akaike’s Information Criterion (AIC), Bayesian Information Criterion (BIC), Chi-Squared test, Anderson-Darling test and Kolmogorov-Smirnov test. The results indicated that the AFM1 pollution in all milk samples conformed to the Pareto distribution with Risk Pareto (2.661 5, 0.029), (3.765 2, 0.029), (3.370 8, 0.029), (2.854 6, 0.029) and (2.423 5, 0.029) for A, B, C, D and E, respectively. After 10 000 iterations, the average daily exposure to AFM1 from the milk sources A, B, C, D and E were determined respectively as 0.000 23, 0.000 20, 0.000 21, 0.000 22 and 0.000 25 μg/(kg·d) by means of Monte Carlo simulation and the Bootstrap method according to the distribution characteristics of AFM1 pollution and the exposure assessment parameters, and the daily exposure to AFM1 from these milk sources at the 99th percentile were 0.000 82, 0.000 49, 0.000 57, 0.000 73, 0.000 97 μg/(kg·d), respectively. The carcinogenic and non-carcinogenic risk associated with dietary exposure to AFM1 through the consumption of milk for different populations was characterized by using the dietary exposure evaluation model based on combined margin of exposure (MOE) and hazard index (HI) values, respectively. It turned out that MOE values for all populations were significantly higher than 100 and that the HI values were higher than 1 at a tolerable daily intake (TDI) of 0.005 μg/(kg·d) suggesting that although the results of carcinogenic and non-carcinogenic risk evaluation were significantly different for different populations, the health risk associated with dietary intake of AFM1 from milk was very low. In conclusion, this study may provide useful information and guidance for milk quality and safety regulation and systematic risk assessment in the future.

Key words: aflatoxin M1, milk, health risk assessment, probabilistic evaluation model

中图分类号: