SETAC Globe - Environmental Quality Through Science
11 August 2016
Volume 17 Issue 8

Return to the Globe

Is the Fish Embryo Acute Toxicity Test Under Threat? A Perspective from the SETAC Global Animal Alternatives Advisory Group

Adam Lillicrap, Norwegian Institute for Water Research (NIVA); Scott Belanger, Procter & Gamble; Natalie Burden, NC3Rs; Michelle Embry, ILSI-HESI; Lucy Lee, University of the Fraser Valley; and Marc Léonard, L’Oreal

Water testing After nearly eight years of formal development, a previous decade of investigational science and the most rigorous validation exercise for any new ecotoxicity test guideline to demonstrate reliability, robustness and repeatability, the Organization for Economic Co-operation and Development (OECD) fish embryo acute toxicity (FET) test guideline was officially adopted in 2013.5, 4, 2 The test had already been previously adopted in Germany for assessing the acute toxicity of wastewater effluents in place of an acute fish toxicity assay to avoid unnecessary use of fish. However, for the assessment of wastewater purposes, the duration of the test was only 48 hours. During the development of the FET test, it was recognized that some chemicals (e.g., cationic polymers) may not elicit their toxic potential until the embryo is free from the protective outer shell (chorion), and for this reason the OECD FET test was extended to 96 hours to encompass hatching. As a result, the 96 hour FET test was now able to comparably predict the acute toxicity of these chemicals. One potential drawback with the FET test, which was established before the OECD test guideline was accepted, was its insensitivity to certain neurotoxicants requiring metabolic activation to cause toxicity to occur. But this is now well recognized, and with the addition of co-factors to the test solutions, comparable FET toxicity data to that in juvenile fish has been observed. Equally, interspecies differences between fish acute toxicity data for substances that require metabolic activation to elicit a toxic response is also a recognized phenomenon with some organophosphate substances differing by a factor of >100 between fish species. On the other hand, certain other chemicals causing developmental effects during embryogenesis (when approximately 90% of the genome is active) are not as acutely toxic to juvenile fish compared with the FET test.

Reviews of large data bases strongly indicate that differences between different OECD standard fish species (e.g., zebrafish vs. rainbow trout) are on par with predictivity relationships between the FET and fish.1, 3 Hence, a balanced opinion needs to be maintained when considering the suitability of either test system for assessing the acute toxicity of chemicals. Equally, substances that have a specific mode of action affecting fish embryogenesis will currently not be sufficiently assessed for environmental risk by only testing the acute ecotoxicity in juvenile fish. However, the acute ecotoxicity test battery also includes the algal growth inhibition test and acute ecotoxicity testing for a crustacean (e.g., daphnid), either of which are known again from large reviews to be more sensitive than fish around 80% of the time. Subsequently, the environmental risk assessment or environmental hazard classification of chemicals is based on the fish acute toxicity data perhaps 20% of the time on average. That said, the task of gaining regulatory acceptance of the FET test in place of the juvenile fish acute toxicity test is still a major hurdle. One concern with the FET test is its domain of applicability (i.e., what are the classes of chemicals that the FET test is applicable for hazard assessment?). This concern has resulted in a recent review being commissioned by the European Chemicals Agency (ECHA) to assess the correlation between acute fish toxicity (AFT) data and FET data. The report on the results from the correlation analysis was published online on 25 May 2016. Coincidently, this was the same day as the SETAC Global Animal Alternatives Advisory Group (AAAG) meeting in Nantes, France. The aim of this review was to “assess the capacity of the FET test in predicting acute fish toxicity and to define the applicability domain of the FET test for regulatory purposes.” The correlations were performed on FET data from an already established database with AFT data from the OECD toolbox or from the eChemPortal. The initial FET database used for the ECHA report contained results from 2,054 studies covering 1,415 different substances. However, this number was significantly reduced, based on reliability criteria and exclusion factors, to only 156 studies covering 123 chemicals. A perusal of the three major comparisons of 2009, 2013 and the recent one in 2016 find only 17 and 32 Chemical Abstract Service (CAS) numbers in common between the 2009–2016 and 2013–2016 reviews, respectively.

Grand Place, Brussels
Two CAS numbers in Scholz et al.6 were not included in Belanger et al.1 due to the chemicals being challenging compounds to test.

Clearly, chemical coverage varies greatly amongst the comparisons when different inclusion or exclusion criteria are employed. An additional complicating factor is that for many of the substances, the FET/AFT correlations were not performed using the same species of fish (since zebrafish AFT data that had corresponding FET data was often lacking). Where zebrafish AFT data were unavailable, the correlations were based on bluegill sunfish, fathead minnow and rainbow trout AFT data with FET data. One might consider that such interspecies correlations would be akin to comparing apples and oranges and should continue to be routinely explored as in Belanger et al.1 Interestingly, the ECHA report indicated that for most of the substances where other species AFT test data were used in the comparisons, these corresponded to weaker toxicity being observed in the FET test suggesting that important differences exist with respect to choice of test species in the AFT. The report further indicated that interspecies correlations of the AFT frequently differed by a factor of >10 and often >100. But again, this is not surprising considering that even intraspecies comparisons of ecotoxicity tests using the same substance can result in differences in the result. There are multiple reasons why correlation analysis of scientific data should be considered with care. A simple example of this was eloquently demonstrated at the SETAC Europe 26th Annual Meeting in May 2016 in Nantes, France, where Michiel Jonker discussed correlations between ring trial data for passive sampling of sediments and the need for standardization. The results of the ring trial indicated that it was the actual chemical analysis that was one of the main reasons for not achieving a near perfect 1:1 correlation between different laboratories performing the same test. It is for these and other reasons that assessment factors are applied to ecotoxicity data when performing environmental risk assessments to account for uncertainty in the data. Furthermore, for any environmental risk assessment to be completely sound and robust, multiple lines of evidence are needed to comprehensively conclude on the risk that a chemical poses to the environment. Henceforth, organisms from other trophic levels than fish are also required for the environmental hazard assessment of chemicals.

Although the ECHA commissioned report does not specifically indicate any firm conclusion of the ECHA as to whether FET test data would or would not be suitable as a substitute to the AFT, the conclusions indicate that from the current correlation analysis it is not possible to define the domain of applicability of the FET. It also indicates that for some substances (22%) the AFT test was more sensitive than the FET test and that for others (7%) the FET test was more sensitive. Although these comparisons are based on a relatively low number of substances, it raises the question of whether the domain of applicability of the AFT could equally be assigned and whether AFT data provides sufficient information to be protective of the environment. Furthermore, there was no retrospective analysis of whether the outlier substances, which did not correlate between the AFT and FET, would have resulted in the environmental risk assessment being based on the AFT data or if the predicted no effect concentration would have been based on either algae or daphnid test data. This would be a relatively simple exercise to include when assessing the suitability of the FET test for regulatory purposes. The report also indicated that additional FET test data are needed to improve the FET/AFT correlations and to better define the domain of applicability of the FET. However, it is naïve to think that registrants will invest in generating data that is “nice to know” when it is not actually required for regulatory purposes. One solution would be to include a prerequisite for FET test data in the REACH information requirements and chemical safety assessment guidance document, in combination with an AFT test, until the domain of applicability can be better assigned. In this instance, to minimize the number of fish that could potentially be used, a modified threshold approach for acute fish toxicity testing4 could be applied, and the threshold concentration (i.e., the single test concentration to carry out a limit test with 7 fish and 7 control fish) be based on the most sensitive endpoint from the algae, daphnid and FET test data. In the meantime, it is of concern that the outcome of the ECHA-sponsored report may have a negative impact on the development and future regulatory acceptance of animal alternatives for environmental risk assessment of chemicals. For this reason, we propose that the FET test is considered more favorably than the ECHA-sponsored report may seem to suggest it ought to be, and we believe that there will still be a need and a driver for future animal alternatives sessions at forthcoming SETAC meetings.

Authors’ contact information:,,,, and

Return to the Globe

Contact SETAC Globe
Contact the SETAC Europe office
Contact the SETAC Europe office