You are here:
Toxicity Screening of a Subset of the ToxCast Chemical Library Using a Zebrafish Developmental Assay: Comparison of Methods
Citation:
Padilla, S., K. Houck, S. Mosher, R. Judson, AND M. Martin. Toxicity Screening of a Subset of the ToxCast Chemical Library Using a Zebrafish Developmental Assay: Comparison of Methods. Fish and Amphibian Embryos as Alternative Models in Toxicology and Teratology, Paris, FRANCE, December 01 - 02, 2016.
Impact/Purpose:
When screening for developmental toxicity, it is important to define the critical experimental variables. This work, comparing the results of two different studies using different methods for chemical screening of approximately 1000 chemicals using the zebrafish developmental toxicity assay, is a first step in identifying those critical experimental variables.
Description:
While many laboratories engage in toxicity screening using a developmental zebrafish larval assay, studies are lacking comparing the many variables of that assay (e.g., one dose or multiple doses, dechorionation or not, dose spacing, line of fish etc.) with the outcome of those assays. There have been two, independent assessments of the toxicity of the approximately 1000 chemicals in the ToxCast Phase I and II libraries to the developing larval zebrafish. Assay A used dechorionated embryos, a highest nominal concentration of 64 uM with log spacing of the doses, and a single dose of the chemical on day 0 with no renewal of the chemical or solution throughout; Assay B used chorionated embryos, a highest nominal concentration of 80 uM with semilog spacing of the doses, and daily renewal of the dosing solution. In order to compare the outcomes for the two assays, the results from the assays were analyzed in approximately the same manner. Both Assays A and B systematically catalogued malformation and death. For purposes of comparison, scores were scaled from 0 to 100, with 0 representing no malformations, values between 0 and 100 representing the number of observed malformations and their severity, and 100 representing death. Concentration-response modeling was then performed using the tcpl R package and associated database system. Activity calls were made based on having a median score at any concentration above 20 and a curve fit model yielding the modelled top of the curve being above 20, and either a Hill or gain/loss model being selected over a constant model. Assays A and B were then compared based solely on the resulting hit calls: 186 hits for Assay A, 460 hits for Assay B, with 176 chemicals as hits in both assays. Because, in general, the chemicals that were positive in Assay A were a subset of the chemicals positive in Assay B, dechorionation may not be a major variable increasing the overall likelihood of a hit call, and/or repetitive dosing may increase the likelihood of a positive result. This abstract may not necessarily reflect official Agency policy.