S1: Use of genomic dose-response for hazard characterization in the NTP

Use of genomic dose-response for hazard characterization in the National Toxicology Program

S1: Use of genomic dose-response for hazard characterization, OpenTox USA 2017

Scott Auerbach 


National Toxicology Program


Toxicoinformatics Group Leader


Abstract: The National Toxicology Program (NTP) has recently begun to incorporate in vivo dose-response transcriptomics into its chemical assessments, the most notable being the Elk River chemical spill studies. The primary purpose of these in vivo dose-response transcriptomic studies is to identify a sensitive point of departure based on molecular biological effects that can estimate the no effect levels determined in longer, more resource intensive studies. The typical protocol uses 6-8-week-old rats that are administered chemical by gavage for 5 repeated doses. The number of dose levels range from 5-7, however more recent studies have added more dose groups and reducing biological replicates, with the goal of increasing the accuracy of the estimated biological effect levels. Target organs are selected for transcriptomics based on the consideration of route, general patterns of chemical toxicity and known properties of the chemical or related chemicals. It is often the case that more traditional end points such as organ weight and clinical pathology are assessed along with the genomics to provide a degree of anchoring and context for interpretation of the genomic data. Data analysis is carried out using BMDExpress 2.0, an updated version of an analytical tool developed by Yang and colleagues (BMC Genomics, 2007). A collection of empirically derived parameter settings and thresholds are used for pre-filtering the data, performing curve fitting, and determining pathway level benchmark dose values. To date NTP has evaluated approximately 20 test articles in in vivo dose-response transcriptomics studies. In most cases the biological effect levels closely approximate the points of departure identified by longer term studies. Furthermore, in many of the cases we find that the affected biological processes that are identified in the genomics are consistent with the more traditional endpoints that are measured.  We anticipate further refinement of the protocols as more is learned about the impact of study parameters on the identified biological effect levels. Based on these early results, it is anticipated that the short-term in vivo toxicogenomics studies will become a frequently used tool in a tiered testing strategy in the NTP evaluation of chemical exposures.