Phillips Martin B, Sobus Jon R, George Barbara J, Isaacs Kristin, Conolly Rory, Tan Yu-Mei
U.S. Environmental Protection Agency, National Exposure Research Laboratory, 109 TW Alexander Dr, Research Triangle Park, NC 27709, USA.
U.S. Environmental Protection Agency, National Health and Environmental Effects Research Laboratory, 109 TW Alexander Dr, Research Triangle Park, NC 27709, USA.
Regul Toxicol Pharmacol. 2014 Aug;69(3):434-42. doi: 10.1016/j.yrtph.2014.05.008. Epub 2014 May 17.
Biomonitoring data are now available for hundreds of chemicals through state and national health surveys. Exposure guidance values also exist for many of these chemicals. Several methods are frequently used to evaluate biomarker data with respect to a guidance value. The "biomonitoring equivalent" (BE) approach estimates a single biomarker concentration (called the BE) that corresponds to a guidance value (e.g., Maximum Contaminant Level, Reference Dose, etc.), which can then be compared with measured biomarker data. The resulting "hazard quotient" estimates (HQ=biomarker concentration/BE) can then be used to prioritize chemicals for follow-up examinations. This approach is used exclusively for population-level assessments, and works best when the central tendency of measurement data is considered. Complementary approaches are therefore needed for assessing individual biomarker levels, particularly those that fall within the upper percentiles of measurement distributions. In this case study, probabilistic models were first used to generate distributions of BEs for perchlorate based on the point-of-departure (POD) of 7μg/kg/day. These distributions reflect possible biomarker concentrations in a hypothetical population where all individuals are exposed at the POD. A statistical analysis was then performed to evaluate urinary perchlorate measurements from adults in the 2001 to 2002 National Health and Nutrition Examination Survey (NHANES). Each NHANES adult was assumed to have experienced repeated exposure at the POD, and their biomarker concentration was interpreted probabilistically with respect to a BE distribution. The HQ based on the geometric mean (GM) urinary perchlorate concentration was estimated to be much lower than unity (HQ≈0.07). This result suggests that the average NHANES adult was exposed to perchlorate at a level well below the POD. Regarding individuals, at least a 99.8% probability was calculated for all but two NHANES adults that a higher biomarker concentration would have been observed compared to what was actually measured if the daily dietary exposure had been at the POD. This is strong evidence that individual perchlorate exposures in the 2001-2002 NHANES adult population were likely well below the POD. This case study demonstrates that the "stochastic BE approach" provides useful quantitative metrics, in addition to HQ estimates, for comparison across chemicals. This methodology should be considered when evaluating biomarker measurements against exposure guidance values, and when examining chemicals that have been identified as needing follow-up investigation based on existing HQ estimates.
通过州和国家健康调查，现在可以获得数百种化学品的生物监测数据。许多此类化学品也有暴露指导值。经常使用几种方法来根据指导值评估生物标志物数据。“生物监测当量”（BE）方法估计一个与指导值（例如，最大污染物水平、参考剂量等）相对应的单一生物标志物浓度（称为BE），然后可以将其与实测生物标志物数据进行比较。由此得出的“危害商数”估计值（HQ = 生物标志物浓度/BE）可用于对化学品进行优先排序，以便进行后续检查。这种方法专门用于人群水平的评估，在考虑测量数据的中心趋势时效果最佳。因此，需要补充方法来评估个体生物标志物水平，特别是那些落在测量分布较高百分位数范围内的水平。在本案例研究中，首先使用概率模型根据7μg/kg/天的起始点（POD）生成高氯酸盐的BE分布。这些分布反映了假设人群中所有个体都以POD暴露时可能的生物标志物浓度。然后进行统计分析，以评估2001年至2002年国家健康与营养检查调查（NHANES）中成年人的尿高氯酸盐测量值。假设每个NHANES成年人都在POD下经历了重复暴露，并根据BE分布对其生物标志物浓度进行概率解释。基于几何平均（GM）尿高氯酸盐浓度的HQ估计值远低于1（HQ≈0.07）。这一结果表明，NHANES成年人的平均高氯酸盐暴露水平远低于POD。就个体而言，对于除两名NHANES成年人之外的所有人，计算出如果每日饮食暴露处于POD水平，与实际测量值相比，观察到更高生物标志物浓度的概率至少为99.8%。这有力地证明了2001 - 2002年NHANES成年人群体中个体的高氯酸盐暴露可能远低于POD。本案例研究表明，“随机BE方法”除了HQ估计值之外，还提供了有用的定量指标，用于跨化学品进行比较。在根据暴露指导值评估生物标志物测量值以及检查已根据现有HQ估计值确定需要进行后续调查的化学品时，应考虑这种方法。
