Transcriptomic points-of-departure from short-term exposure studies are protective of chronic effects for fish exposed to estrogenic chemicals
Transcriptomic points-of-departure from short-term exposure studies are protective of chronic effects for fish exposed to estrogenic chemicals
复制标题
DOI：
10.1016/j.taap.2019.114634
复制
发表时间：
2019-09-01
影响因子：
3.8
通讯作者：
O'Brien, Jason M.
中科院分区：
文献类型：
--
作者：
Page-Lariviere, Florence;Crump, Doug;O'Brien, Jason M.
资源限制往往要求风险评估人员利用评估因素从急性试验中推断慢性毒性。短期暴露后的转录组学剂量反应分析可能为估计慢性毒性提供更可靠和基于生物学的替代方法。在这里，我们表明，短期暴露于内分泌干扰物（EDCs）后，鱼类的转录剂量反应分析提供了慢性毒性的估计，可用作风险评估的保护出发点（POD）。基准剂量（BMD）的方法被用于公开可用的数据集（n = 5），以确定转录组POD在鱼暴露于三个内分泌干扰物（双酚A，乙炔基雌酚，和己烯雌酚）。为了测试与数据处理相关的潜在偏倚，我们的分析比较了不同归一化、过滤和BMD分组方法对转录组POD的影响。然后将所得POD与根据经验得出的每种物质的慢性LOEC进行比较。归一化和过滤方法对最终POD的影响有限。然而，我们发现，来自本体论或路径为基础的基因分组方法的POD是高度可变的，而POD分组方法，专注于最敏感的基因是更稳定的，并提供POD估计是最相似的慢性LOEC。总体而言，无论数据分析方法如何，72%的转录组POD均在慢性LOEC的1个数量级内。当应用我们推荐的分析方法时，一致性提高到100%。这些结果表明，毒理基因组学剂量反应分析有可能成为一个保护性的决策支持工具的化合物与慢性毒性，如内分泌干扰物。
Resource limitations often require risk assessors to extrapolate chronic toxicity from acute tests using assessment factors. Transcriptomic dose-response analysis following short-term exposures may provide a more reliable and biologically-based alternative for estimating chronic toxicity. Here, we demonstrate that transcriptomic dose-response analysis in fish following short-term exposure to endocrine disrupting chemicals (EDCs) provides estimates of chronic toxicity that may be used as protective points-of-departure (POD) for risk assessment. The benchmark dose (BMD) method was used on publicly available datasets (n = 5) to determine transcriptomic PODs in fish exposed to three EDCs (bisphenol A, ethinylestradiol, and diethylstilbestrol). To test for potential bias related to data processing, our analysis compared the effect of different normalization, filtering, and BMD-grouping methods on the transcriptomic PODs. The resulting PODs were then compared to the empirically derived chronic LOEC of each substance. Normalization and filtering methods had limited impact on the final PODs. However, we found that PODs derived from ontology-or pathway-based gene grouping methods were highly variable, whereas PODs from grouping methods that focused on the most responsive genes were more stable and provided POD estimates that were most similar to the chronic LOEC. Overall, 72% of transcriptomic PODs were within 1 order of magnitude of the chronic LOEC, regardless of data analysis method. When our recommended analysis approach was applied, the concordance improved to 100%. These results suggest that toxicogenomic dose-response analysis has the potential to be a protective decision-support tool for compounds with chronic toxicity, such as EDCs.