Establishing Chronic Toxicity Effect Levels for Zebrafish (Danio rerio) Exposed to Perfluorooctane Sulfonate

Zebrafish (Danio rerio) are among the aquatic species most sensitive to perfluorooctane sulfonate (PFOS). Environmental regulatory agencies and researchers use effect benchmarks from laboratory zebrafish PFOS toxicity studies in PFOS‐spiked water to calculate PFOS aquatic life criteria. Threshold values as low as 0.7 µg/L (identified in an early, limited scope study) have been used in criteria derivation and site‐specific aquatic ecological risk assessments. The present study reviews PFOS effects benchmarks for lethality, growth, and reproduction endpoints from more than 20 zebrafish toxicity studies, including a recent multigenerational study conducted by the United States Army Corps of Engineers Engineer Research & Development Center. Our review of 12 key studies examining long‐term, chronic exposures (including multigenerational exposures of 300 days or more) indicated that 0.7 µg/L should not be used as a conservative screening threshold given that effects could not be repeated at this concentration by the recent enhanced multigenerational study. Based on this finding and multiple chronic sublethal studies on PFOS in zebrafish, chronic effects on lethality, growth, and reproduction occur at concentrations two orders of magnitude higher than 0.7 µg/L. Overall, the present review indicates a no‐effect screening level of 31 µg/L and a low‐effect screening level of 96 µg/L should be used to develop PFOS aquatic life criteria and to inform site‐specific ecological risk assessments that are charged with evaluating risks to freshwater fish. Environ Toxicol Chem 2024;43:7–18. © 2023 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.


INTRODUCTION
Among the per-and polyfluoroalkyl substances (PFAS), perfluorooctane sulfonate (PFOS) is commonly observed in environmental samples from aquatic systems due to its use in aqueous film-forming foam formulations and other products, environmental behavior, and relatively large mass of releases (Jarvis et al., 2021).Perfluorooctane sulfonate has the potential to be toxic to aquatic life, particularly fish, and has been considered as bioaccumulative in aquatic species (Conder et al., 2008).As such, establishing evidence-based criteria is critical to evaluating risks to organisms from exposures to PFOS in the environment, especially for aquatic life such as fish and invertebrates that are directly exposed to PFOS in aquatic ecosystems.
A number of regulatory agencies and scientific researchers have developed risk-based criteria for PFOS in water using results from laboratory toxicity experiments in which aquatic life were exposed to PFOS-spiked waters (Beach et al., 2006;Conder the aquatic species more sensitive to PFOS.For example, in the recent draft ambient water quality chronic PFOS criterion developed by the USEPA (2022), zebrafish were listed as the most sensitive fish species and fourth most sensitive aquatic species overall.Given that zebrafish are among the most sensitive species to PFOS exposure, they exert a large influence on aquatic life screening values calculated for environmental decision making (Conder et al., 2020;DOEE, 2016;McCarthy et al., 2017;USEPA, 2022;Yang et al., 2014).
The toxicity of PFOS to zebrafish has been relatively well studied (over 130 peer-reviewed publications since 2008), but there is substantial variation in experimental designs and study quality.The purpose of the present study was to conduct a review of the available spiked-PFOS toxicity studies with zebrafish, synthesizing and highlighting key results evaluating the chronic toxicity of PFOS to zebrafish, with the goal of recommending PFOS thresholds in freshwater that can be directly used in screening-level ecological risk assessments to evaluate the toxicity of PFOS to fish populations.The primary audience for our review is ecological risk assessors and ecotoxicologists developing aquatic life criteria and evaluating site-specific risks to aquatic life.

METHODOLOGY
Controlled laboratory studies evaluating the toxicity of PFOS to zebrafish (single chemical, PFOS-spiked water exposures) were identified by searching online literature databases (e.g., PubMed ® and Google Scholar) with the keywords "PFOS" and "zebrafish," and evaluating references cited in recent reviews of PFOS ecotoxicity and various regulatory compilations (Conder et al., 2020;Giesy et al., 2010;Gobas et al., 2020;McCarthy et al., 2017;USEPA, 2022;Zodrow et al., 2020).The present review effort focused on finding data on subacute and chronic effects on apical endpoints generally considered to result in ecologically relevant population-level effects: growth (weight and length), reproduction, and survival.In four of the studies reviewed in detail (Chen et al., 2013;Cui et al., 2017;Du et al., 2008;Wang et al., 2011), short-term evaluation of the survival of offspring produced from parents exposed to PFOS was quantified as a reproductive endpoint attributed to the parental exposure.In two studies (Gust et al., 2023;Keiter et al., 2012), the subsequent F1 generation was evaluated for growth and reproduction, so data from these studies were included in this evaluation.Although condition factor has been historically used in fish as a measure of general fish health, it was not considered to be a sensitive endpoint for PFOS exposure (see Supporting Information).
The apical endpoints of survival, growth, and reproduction that are the focus of our review are typically used for regulatory screening criteria, providing quantitative benchmarks for decision-making purposes at contaminated sites or point source evaluations (Conder et al., 2020;Environment and Climate Change Canada, 2018;Environment Protection Authority Victoria, 2017; National Institute for Public Health and the Environment [RIVM], 2010; Suter, 2018; USEPA, 2022).Studies evaluating any (or all) of these endpoints were selected if the exposure to PFOS was continuous and for a duration of at least 14 days because short-term/acute studies in fish are generally considered to be less than 14 days in duration (Organisation for Economic Co-operation and Development, 2012).Most studies identified in the present review that lasted less than 14 days generally included exposures of only 96 to 120 h, a standard acute toxicity testing duration.These shorter-term studies focused primarily on survival, and most concentrations associated with lethality were generally noted to be 1000 µg/L or more (Gust et al., 2023), approximately one to two orders of magnitude higher than the thresholds for more sensitive growth and reproduction endpoints evaluated in the chronic exposures.
Overall, 12 studies met the requirements for the focused review.Each of the 12 studies was evaluated to identify the most sensitive survival, growth, or reproductive effect endpoint, with the overall goal of identifying the associated exposure concentrations of PFOS in water that represent thresholds protective of ecologically significant effects.A detailed review of each of the 12 studies is provided in the Supporting Information section.For studies with effects that were evaluated at multiple points during the exposure, the effects and associated exposures at the termination of each PFOS exposure period were considered the most accurate thresholds for representing incipient toxicity.Zebrafish exhibit the most natural variability in growth in their larval stage as they transition into adulthood, and in many cases within the studies reviewed there were changes in growth patterns that occurred at earlier points that did not persist into adulthood.For example, in the Keiter et al. (2012) study, male fish exposed to 0.73 to 268 µg/L of PFOS weighed 20 to 33% less than controls when measured at 90 days post fertilization (dpf), yet weighed only 3 to 11% less relative to the controls when measured again at 180 dpf.Similar effects were also seen in the Cheng et al. (2016) study (female body weight decreased by 10% at 90 days, but 3% when measured at 150 days).These results suggest that PFOS may have short-term effects on growth that become largely absent once the fish are older and/or have adapted to PFOS exposure.To complicate matters, the opposite has also been seen, such as in the study by Du et al. (2008).In the present study, exposure to 250 µg/L of PFOS caused a 17% increase in body weight at day 30 of exposure, a 0% change at day 70 of exposure, and a 34% reduction in body weight after 30 days in clean water.These overall patterns likely reflect natural variability in growth rather than an underlying effect of PFOS.Thus, shorter-term studies are not representative of longer-term effects, and effects are best evaluated at the longest exposure point.In two of the studies, a multigenerational design was used, and results from each of the generations were evaluated separately for endpoints and effects thresholds.
For each study and exposure concentration, the percentage magnitude of the effects on survival, growth, and/or reproduction were normalized to the control group response using the following equation: × 100, with V1 being the biological measurement in the control, and V2 being the biological measurement in the treatment.Thus, negative values indicate adverse performance relative to the controls (e.g., a −20% change in body weight would mean the treated fish weighed 20% less than controls) and positive values indicate better performance than the controls.Measured values were typically obtained from the tables provided in each study, but in some studies, values were obtained from a visual analysis of study figures (e.g., bar charts) using ImageJ image processing software (National Institutes of Health, 2023).
In the present review, effects were considered ecologically significant if the results were statistically different from controls (p of 0.05 or less) and the adverse effect size was approximately 20% or greater (e.g., treated fish weighed at least 20% less than controls).The lowest PFOS exposure meeting those criteria was identified as the lowest observed effect concentration (LOECs) and the highest exposure less than the LOEC was identified as the no observed effect concentration (NOEC).Identification of concentration-response relationships was considered in deriving NOEC and LOEC values in each data set, following USEPA (2000) guidelines.In derivation of criteria to protect aquatic life in the United States, the USEPA has used an adverse effect of 20% or greater as the threshold for ecological significance at the population level (USEPA, 2016a(USEPA, , 2016b)).Effect sizes less than 20% can be detectable as statistically different from control effects in laboratory toxicity experiments, but this level of effect has not been considered to be severe enough to result in chronic effects at the population level (Suter et al., 2000;USEPA, 2016b).Furthermore, because natural variability in the growth of zebrafish can range up to 20% from experiment to experiment, effects below 20% are not likely to be clearly and consistently distinguishable from natural levels of biological variation for zebrafish (see below discussion Recommendations for future studies, and detail in the Bao et al. [2019] study in the Supporting Information section).Some environmental regulatory agencies (Australian Government Department of Agriculture and Water Resources, 2018; Canadian Council of Ministers of the Environment, 2007;European Chemicals Agency, 2008) and stakeholders involved in site-specific decision-making consider exposures associated with the 10% level of effect (i.e., effect concentrations causing 10% effect [EC10 values]) as the threshold for ecologically significant effects.Recently, in a draft document released for public comment, USEPA (2022) proposed the use of EC10 values for derivation of aquatic life values for PFOS, despite the use of EC20 values in derivation of previous aquatic life criteria (USEPA, 2016a(USEPA, , 2016b)).Thus, in some cases exceedance of this 10% effect size could be considered potentially unacceptable and require additional investigation and/or management.For the purposes of our discussion, we have noted effects that occurred between 10% and 20% levels of effect, despite the uncertainties noted above and the fact that most effect sizes of less than 20% in the present review were not statistically detectably different from control responses, despite strong levels of replication and precise measurements of biological responses in some notable studies.

Overview of selected studies
As shown in Table 1, the present review identified 12 controlled PFOS-spiked toxicity tests with zebrafish lasting 14 days or longer (a detailed summary of all studies can be found in Supporting Information, Table S1, and a short synopsis on each study is provided in the Supporting Information).Counting each of the individual generations evaluated for growth, survival, or reproduction as individual tests in the two generations in the Keiter et al. (2012) study (growth/reproduction was not measured in the 14 day F2 generation; the short-term survival was treated as a reproductive endpoint), or the three generations in the Gust et al. 2023 study, these 12 studies provide 15 data sets for evaluating chronic PFOS toxicity in zebrafish.Survival, growth (both length and body weight), and reproduction were measured in nine, 10, and eight of the 12 studies, respectively.Study design and procedures varied considerably between studies.For example, only three of the 12 studies included in our review verified the nominal concentration of PFOS in water (via analysis of PFOS in water samples) to which zebrafish were exposed.In roughly half the studies, exposure duration was 15 to 30 days, while the other half included exposure durations of 70 to 180 days.All studies were conducted using the "pet shop A and B" (AB) zebrafish line except for the Keiter et al. (2012) study, which used the West Aquarium (WA) line.Only two of 12 studies (Gust et al., 2023;Krupa et al., 2022) had more than three replicates per exposure concentration and/or more than three exposure concentrations.Overall, many of these studies are subject to higher uncertainty due to design constraints, especially those lacking measurements of PFOS in exposure waters and those with three or fewer exposure concentrations or replicates per exposure.Studies were not quantitatively scored or ranked, although the discussion below includes consideration of study quality and the ability of the study to provide the most robust and lowest uncertainty evaluation of toxicological thresholds.

Effects of PFOS on zebrafish growth
Between the two growth endpoints evaluated in the present review (length and weight), statistically significant, dosedependent effects on body weight occurred at lower concentrations of PFOS than effects on body length, and, at the same concentration of PFOS, effects on body weight were more pronounced than effects on body length.For example, in Krupa et al. (2022), exposure to 300 µg/L of PFOS reduced body weight by 30% and length by 13%.Exposure to PFOS at concentrations of 400 µg/L or more caused a statistically significant reduction in length (20% or higher) in only two studies (Krupa et al., 2022;Shi et al., 2009; Supporting Information, Figure S2a).Length was more sensitive than body weight in only one of the 10 studies (Shi et al., 2009) that measured both length and body weight.
Statistically significant and biologically relevant effects (20% or more effect size) observed on body weight in four of the 10 studies formed the basis of some of the lowest NOEC and LOEC values (Table 1 and Figure 1A).In these studies, LOECs ranged from 40 µg/L (26% adverse effect; Guo et al. [2019]) to 260 µg/L (30% adverse effect; Krupa et al. [2022]), as indicated by the filled in symbols shown in Figure 1A.Other studies  A −3% statistically effect on length (males and females) was found at this concentration for length, while nonsignificant effects for weight were found for weight (+6% and −5% for males and females, respectively).Weight exhibited a higher effect size at the LOEC. d These studies were all conducted following the same experimental design/conditions, by the same research group. e The authors report reduced survival and increased malformations of the F1 generation from parents exposed to 250 µg/L but do not provide the data, effects on weight not adverse. f No data or statistical analysis provided; a biologically significant level of effect at 250 μg/L was assumed based on study text. g This study had three exposure groups that were all sampled at day 120 (1-20, 21-120, or 1-120) summarized here is the full exposure (1-120). h This study reports significantly increased mortality of F1 maternally exposed fish (50 and 250 µg/L) but do not provide the data or statistical comparisons (not possible to show percentage change vs. control); results are highly uncertain. i The growth data in this study is of poor quality.k Controls had only one replicate no statistical comparisons could be evaluated.
For additional study details and NOECs and LOECs that assume a lower (10%) adverse effect size, see Supporting Information, Table S1.exposing zebrafish to similar concentrations of PFOS (as high as 50-400 µg/L) indicated statistically significant reductions on body weight of approximately 10% (Figure 1B).In addition, some studies exposing zebrafish to up to 200 µg/L did not measure any dose-dependent significant effects on body weight (Figure 1C).Overall, considering effect sizes 20% or more were observed in multiple studies at approximately 40 and 100 µg/L, exposure in this range may significantly impact growth.

Effects of PFOS on reproduction
Reproduction was of similar sensitivity to growth (body weight).Dose-dependent adverse effects on reproduction were statistically significant and biologically relevant in the four of the seven studies where reproduction was measured.Reproductive endpoints included production of eggs, 24-h post fertilization (hpf) egg viability, egg hatchability (typically measured at 72 hpf), and incidence of malformations in offspring.Survival of the offspring was also treated as a reproductive endpoint attributed to the parents in cases in which no other endpoints were evaluated on the offspring.This included the F2 generation in Keiter et al. (2012)  The effects of parental exposure to PFOS on the survival of their larval zebrafish progeny are notable because exposure to PFOS does not impair fertilization rate, hatchability, or number of eggs per reproductive event (Gust et al., 2023;Keiter et al., 2012;Wang et al., 2011).Exposure to PFOS in these studies also did not impact embryo survival within the first 96 h of development.However, impaired survival/viability of the embryos begins to occur around the time the larval fish start to absorb their yolk sac and commence feeding (~5 dpf).In the present review, this effect was attributed as a reproductive impairment on the parents.In one of the seven reproductive studies (Wang et al., 2011), survival of the F1 offspring when exposed to 50 µg/L was lower than that of the parents when they were exposed to 50 µg/L.In the case of two of the seven reproductive studies (Chen et al., 2013;Du et al., 2008), lower FIGURE 1: Change in body weight (wet wt, relative to control) of fish exposed to perfluorooctane sulfonate (PFOS).(A) Studies with statistically significant adverse (negative) effect sizes of more than 20%, (B) studies with statistically significant adverse effect sizes of more than 0 to less than 20%, and (C) studies with no significant dose response effect.Studies using multiple concentrations are connected by dashed lines.Boxes filled black indicate a statistically significant difference (p of 0.05 or less).For all studies, the more sensitive sex (largest effect size/lowest concentration with a statistically significant effect) is displayed and denoted (F = females; M = males; M/F = fish too young for sex to be determined).Only 10 of the 12 studies are displayed because two did not report growth effects.Studies with multiple generations are plotted separately with their own color and shape (Keiter et al., 2012 red triangles;Gust et al., 2023 blue circles).For the F2 generation of Gust et al. (2023), the body weight was calculated as the sum body weight per replicate tank rather than the average body weight per replicate tank that was done in all other studies/generations.survival was noted in the F1 offspring of fish parentally exposed to PFOS after hatching, even though the animals were exposed to clean water.
Overall, a statistically significant reduction in survival in offspring of parentally exposed fish occurred with exposure to PFOS at concentrations of 50 to 268 µg/L (Chen et al., 2013;Cui et al., 2017;Du et al., 2008;Keiter et al., 2012;Wang et al., 2011), as shown in Figure 2A,B.Exposure to 50 µg/L resulted in a statistically significant (38%) reduction in F1 survival in two studies (Du et al., 2008;Wang et al., 2011).In the Keiter et al. (2012) study, there were 58% and 47% reductions in survival of the F1 and F2 generations, respectively, parentally exposed to 107 µg/L of PFOS; these reductions were not statistically significant due to high variance in the treatment groups.In the Gust et al. (2023) study, there was a statistically significant (36%) decrease in survival of the larval F2 generation from parents exposed to 75.5 µg/L of PFOS, but the F1 generation showed only a 1% decrease in survival (not statistically significant) at 75.5 µg/L.Overall, despite the lack of statistically significant effects in some studies, the reproducibility of these effect sizes (40-60% adverse effects) across multiple studies at approximately 50 and 100 µg/L suggests that exposures in this range may impact the reproduction of zebrafish through decreased survival of the offspring of parents exposed to PFOS.

Effects of PFOS on survival
Chronic survival was a less-sensitive endpoint than growth or reproduction, with statistically significant and biologically relevant decreases in survival occurring at concentrations in the parental generation of approximately 100 µg/L (Gust et al., 2023) to 600 µg/L or more (Krupa et al., 2022) in the studies shown in Table 1.As discussed above, as a reproductive effect, the complete mortality in F1 larval zebrafish parentally exposed to 250 µg/L, detectable at 6 dpf in the Keiter et al. (2012) study, is notable because it occurs at concentrations lower than acute toxicity and in embryos raised in clean water.In embryos directly exposed to PFOS, the LC50 was calculated to be 5847 µg/L at 96 hpf (Hagenaars et al., 2011), while in a different study there was no significant mortality up to 120 hpf in zebrafish exposed to 100 or 500 µg/L, although significant mortality occurred at 1000 µg/L or higher (Shi et al., 2008).Acute studies in zebrafish rarely last longer than 120 hpf, and it is possible that these shorter studies are not measuring toxicity that would be observed if the duration was extended to 6 dpf or more.As highlighted in the study by Warner et al. (2022), the toxicokinetics of PFOS is complex, biphasic, and significantly influenced by hatching/shedding of the chorion.When zebrafish embryos were directly exposed to PFOS in water for 30 days, the LC50 was calculated to be 490 µg/L (although there was high variability at the 260 µg/L concentration, 10%-85% survival in replicates; Krupa et al., 2022).Based on some of the studies highlighted in the section on reproductive effects, it may be likely that PFOS is more toxic to the offspring through parental exposure than direct exposure of embryos because the LC50 values are higher in studies directly exposing developing embryos to PFOS than the 100% mortality of F1 offspring parentally exposed to approximately 250 µg/L of PFOS.Overall, there was little evidence of long-term impacts of PFOS exposure to adult zebrafish survival, with LOECs generally 100 µg/L or greater.

Recommended screening values for sublethal effects
As noted in the introduction to the present study, zebrafish are among the most sensitive fish species to PFOS, and PFOS thresholds for zebrafish are often used with those from other species in determining screening level criteria that are protective of all fish species or all aquatic life.The information compiled in the present review features information on apical thresholds that would apply to long-term chronic PFOS exposures to zebrafish, thus providing a basis for identification of thresholds protective of all or most PFOS exposures of freshwater fish in aquatic ecosystems in which PFOS is detected.The goal of this section of the study was to reduce this information to two simple values that can reflect the entire dataset: a NOEC screening level for fish exposed to PFOS in water at or below which adverse effects are not expected, and a LOEC screening level for fish exposed to PFOS at or above which adverse effects may be expected.
The 12 studies reviewed in the present study (Table 1) vary considerably in experimental design and execution, and the quality and weight of the evidence for each study should be considered in a synthesis of threshold information.The Keiter et al. ( 2012) is a key study in the present review.This study has been used in several regulatory reviews to establish aquatic life criteria for PFOS.For example, the Australian regulatory authorities have included the results of this study (i.e., the 0.7 µg/L threshold) in the derivation of water quality guidelines for PFOS (Commonwealth of Australia, 2023;DOEE, 2016).In contrast, USEPA did not include Keiter et al. (2012) in its calculation of draft ambient water quality criteria for PFOS (USEPA, 2022), citing a lack of concentration-response relationships due to limitations with study design (e.g., small number of concentrations and low replication).As seen in Table 1, the Keiter et al. (2012) study noted that the lowest exposure concentration (0.73 µg/L) was associated with statistically significant effects on body weight and length.This study was a multigenerational study and observed effects on body weight (180 days) in one of the two generations tested, finding an approximate 12% decrease in parental (P) generation females at 0.73 µg/L (Figure 1A and Table 1).Only a 5% adverse effect on length was noted in P females at this concentration, and this result was reported by Keiter et al. (2012) as statistically significantly different from controls.In contrast, no statistically significant effects on body weight were observed in males (6% increase in weight relative to controls; Figure 1B) or females (5% decrease in weight relative to controls) from the first generation (F1) fish exposed to 0.73 µg/L.Thus, the effect noted by Keiter et al. (2012) at 0.73 µg/L was not consistently observed among the sexes or among the different generations.Weight in the second generation (F2) was not measured.
These results are complicated by a limited study design of only two replicates per dose and only two PFOS doses studied among all three generations (0.73 and 107 µg/L).In addition, there was limited confirmation of water PFOS concentrations (monthly).Another challenge regarding the study was that the control fish were exposed to 0.073 µg/L (10 times lower than the lowest concentration in the PFOS-spiked exposures).As shown in Table 1, the 0.73 µg/L threshold noted by Keiter et al. (2012) is approximately two orders of magnitude lower than NOEC values for other sublethal chronic thresholds, and it is approximately 50-100 times lower than other low-end thresholds identified for growth and reproduction in our review.One key difference between the Keiter et al. (2012) and the other 11 studies was the strain of zebrafish used, although, as discussed in the Supporting Information, this does not explain the disparity in the values.
The Keiter et al. (2012) result of 0.73 µg/L was considered extraordinary given the other studies with PFOS on zebrafish.Keiter et al. (2012) is encumbered with significant uncertainty due to its limited experimental design.Because of its potential influence in regulatory aquatic life guidelines for PFOS, the goals of the Keiter et al. (2012) study were recently revisited by Gust et al. (2023) using a more rigorous experimental design.The Gust et al. study included more replication (five PFOS doses instead of two PFOS doses), a larger number of concentrations tested (five instead of two to three), and more frequent confirmation of PFOS in water (weekly instead of monthly).
The Gust et al. (2023) study was unable to reproduce any of the growth effects detected by Keiter et al. (2012) at the 0.7 µg/L (nominal) concentration or at higher concentrations (3.2 and 20 µg/L [nominal]).As noted in Table 1, consistent, statistically detectable dose-dependent effects were found at the 100 µg/L (nominal) concentration, the highest concentration tested in the study.A heat map figure comparing the average percent change (relative to control) is shown in Supporting Information, Figure S1.The limited evidence indicated for potential and inconsistent effects at the 0.73 µg/L concentration (six replicates total for the three generations) in Keiter et al. (2012) is outweighed by the increased doses and replication of the Gust et al. (2023) study at the 0.1, 0.7, 3.2, and 20 µg/L concentration (60 replicates total for the three generations).For example, 180-day exposure to 0.73 µg/L caused a 5% to 12% reduction of body weight of female fish in the P and F1 generations of the Keiter et al. (2012) study, while the 180-day exposure to 0.1 to 20 µg/L did not cause adverse dosedependent effects on body weight in the P generation females in the Gust et al. (2023) study.In contrast, there is good agreement between these two studies regarding the potential for effects at 100 µg/L.Gust et al. (2023) consistently identified statistically significant growth and survival effects in this concentration (as high as 16%-36% adverse effect), and Keiter et al. (2012) also noted statistically significant adverse effects as high as 17% to 18%.Overall, the key disparity between these two studies lies in the effects noted by Keiter et al. (2012) at the approximate 0.7 µg/L concentration (six replicates total) and the absence of effects at the 0.1, 0.7, 3.2, and 20 µg/L concentrations (60 replicates total) noted by Gust et al. (2023).
Results of other studies support our review conclusions that the effect threshold for zebrafish sublethal effects are orders of magnitude higher than the 0.7 µg/L level suggested in Keiter et al. (2012).As shown in Table 1, the next most sensitive LOEC on body weight was by Guo et al. (2019), which measured a statistically significant decrease in body weight (~26%) following a 21-day exposure to 40 µg/L of PFOS.No statistically significant adverse effect on growth was detected at 20 µg/L.However, the growth responses in this study displayed an unusual temporal response pattern because there was a rapid change in body weight from day 0 and day 7 in controls and PFOS-exposed fish, followed by no further growth for the rest of the study (see detailed discussion of this study in the Supporting Information).Shi et al. (2009) exposed larval zebrafish to 100 to 400 µg/L for 14 days and reported no growth effects at 100 or 200 µg/L.A robust 30-day exposure by Krupa et al. (2022) did not observe adverse effects on growth until concentrations reached 260 µg/L or more.Overall, the information from these additional research on zebrafish sublethal effects, when combined with those of the Gust et al. (2023) study, indicate that the effects observed at 0.7 µg/L in Keiter et al. (2012) are not considered repeatable, and 0.7 µg/L should not be considered for use in representing the threshold for zebrafish PFOS effects in aquatic life criteria calculations.
Considering the extraordinary results associated with the 0.7 µg/L concentration in Keiter et al. (2012), we excluded this study from our calculation of a NOEC screening level for PFOS in water at or below which adverse effects are not expected (31 µg/L) and of a LOEC screening level for PFOS at or above which adverse effects may be expected (96 µg/L).The NOEC and LOEC screening levels were calculated from the geometric mean of the NOEC and LOEC values (Table 1) from the following data sets: P generation of Gust et al. (2023), F1 generation of Gust et al. (2023), F2 generation of Gust et al. (2023), Wang et al. (2011), Du et al. (2008), Krupa et al. (2022), andGuo et al. (2019).There is a variety of experimental designs in these studies, and no study would be considered perfect.However, the most reliable studies in terms of study design, as reviewed in the present study, are those by Gust et al. (2023) and Krupa et al. (2022).These studies had high replication (n = 4-5), a large number of PFOS concentrations tested (five to six doses per study), and frequently measured the concentrations of PFOS in the test water.Considering these robust studies only, the geometric mean for the NOECs is 28 µg/L and for the LOEC is 120 µg/L, which is comparable to the recommended NOEC screening level (31 µg/L) and LOEC screening level (96 µg/L).
Several studies were not included in the geometric mean calculation.The unbounded NOEC of Cheng et al. (2016) and the LOEC of Chen et al. ( 2013)-250 µg/L-were not included in the geometric means because these studies only featured a single concentration (i.e., 250 µg/L).The NOEC and LOEC from Cui et al. (2017) were not included because this study did not depict the data or provide results of statistical comparisons of the PFOS-exposed zebrafish and control zebrafish.The NOEC and LOEC from Sharpe et al. (2010) were not included because this study only featured one replicate in the controls, preventing statistical analysis.
The recommended NOEC and LOEC screening values based on the geometric means of the studies reviewed, 31 and 96 µg/L, respectively, are reasonably conservative and considered protective of the majority of the responses observed in the studies focused on in our review (Figure 3).Nine of the 12 LOECs shown in Figure 3 2019) are lower than the recommended 96 µg/L LOEC screening value, but as noted in the present review, the results from these papers have higher uncertainty.The NOEC and LOEC screening levels for PFOS exposure to zebrafish are also similar to the NOEC and LOEC for another freshwater fish species, fathead minnows (Pimephales promelas), of 44 and 88 µg/L, respectively (Suski et al., 2021), based on a 18% decrease in mass in fish exposed to 88 µg/L PFOS, indicating a similar level of sensitivity among fish species.
The NOEC and LOEC screening levels recommended in the present study should not be considered default or mandatory action, permitting, or remediation goals; careful consideration of the aquatic system to which they would be applied is needed for site-specific decision-making uses of these screening levels.Under typical ecological risk assessment processes, these NOEC and LOEC values can be applied to better understand the need for additional evaluation of potential risks to fish due to exposure to PFOS in the environment.Concentrations of PFOS in water that are equal to or less the NOEC screening value of 31 µg/L should not be considered to result in population-level apical effects.Adverse effect levels less than 20% would be expected and, as noted in the present review, 20% is the lowest recommended effect size that can be clearly distinguished from natural levels of biological variation for the endpoints evaluated in our review.In contrast, concentrations of PFOS in water that are equal to or higher than the LOEC screening value of 96 µg/L should be considered evidence that adverse effects of 20% or more could be possible and that additional investigation to understand the magnitude of any potential effects may be warranted.Exceedance of the LOEC at a site does not imply adverse population level effects are guaranteed, only that additional evaluation or consideration is recommended.For example, exceedances of the LOEC should be evaluated with respect to the time over which the exceedances occur because the recommended LOEC screening level in the present review is intended to be applied to a concentration in water that reflects long-term, steady-state exposure conditions as they would occur over the weeks-long time frames of the studies in Table 1.Short-term exceedances of the LOEC (on the scale of a few hours or days) may not be sufficient to elicit the adverse effects associated with the LOEC screening level.Additional investigations may evaluate other lines of evidence at a site with water exhibiting exceedance of the LOEC screening level, such as aquatic community evaluation, fish community analysis, index of biological integrity, or toxicity testing.Concentrations of PFOS in water that are in the 31 to 96 µg/L range indicate known safe levels (NOECs) are exceeded, but LOECs associated with 20% adverse effect levels in zebrafish are not yet reached.In this range, it is uncertain if PFOS-driven adverse effects would be evident, and it is unlikely that such effects would be able to be clearly distinguished from the ranges of natural biological responses in zebrafish.

Recommendations for future studies
The present review provides recommendations for the evaluation of PFOS effects on zebrafish.Additional experiments will likely be conducted with PFOS and other PFAS with this species, and other aquatic species, in support of aquatic life criteria development and site-specific evaluation of risks to aquatic life in aquatic ecosystems in which PFOS is detected.
It is clear from our review that effect size is a critical consideration in future toxicity studies with PFOS, and understanding natural variability of endpoints is essential to evaluate so that toxic effects can be clearly separated from the range of natural variability.A few studies in our review reported multiple sets of control datasets, and this allows the comparison of natural variability in the context of endpoints.For example, the Bao et al. (2019) study used two sets of control zebrafish from the same culture, and these control fish were used in two different toxicity tests: one set was initiated in the morning and the other set was initiated in the evening.When the morning and the evening control groups were compared, an 18% difference in body weight and a 4% difference in length were evident at the termination of the 21-day experiment, despite being cultured under the same conditions and maintained under the same conditions during the toxicity test.Body weight variation was also evident in other PFOS studies.For example, the control female P generation of fish in the Keiter et al. (2012) study weighed 676 mg at 180 dpf while the F1 generation females weighed 337 mg at 180 dpf (i.e., ~50% less than the P generation controls), despite being held under the same experimental conditions and no addition of PFOS.The multigeneration study by Gust et al. (2023) also exhibited differences between the P and F1 female control fish, with average weights of 332 milligrams in the P generation and 241 mg in the F1 generation (~25% less) at 180 dpf.Reproductive data are even more variable than growth data, with a study by Paull et al. (2008) noting that variability within breeding groups of zebrafish is so high that a minimum of six replicates per treatment is necessary to detect a 40% change in egg output of female zebrafish.This suggests that all the studies in the present review are underpowered to detect even a 20% change in egg production.
Inconsistencies in effects are evident in a number of papers published by a research group at the Wenzhou Medical College (Wenzhou, Zhejiang, China).This group followed a similar experimental design and experimental conditions in evaluating PFOS in zebrafish over a number of studies (Chen et al., 2018(Chen et al., , 2013;;Cheng et al., 2016;Cui et al., 2017;Wang et al., 2011).However, even these studies, performed within one research group in the same laboratory, do not consistently replicate the adverse effects of PFOS on growth, even at concentrations above the 50 to 100 µg/L low threshold range noted in the present review.For example, in the five studies where male fish were exposed to 250 µg/L for 120 to 180 days, the magnitude of effect ranged from a 14% decrease (Wang et al., 2011) to a 16% increase in body weight (Cui et al., 2017), with the remaining two studies indicating no statistically detectable adverse effect.Overall, zebrafish exposed to 250 µg/L in these studies exhibited changes in growth within the range of −20 to +20% relative to their respective control, which suggests that there is high biological variation in growth responses to PFOS.
Overall, the 20% (or more) level of natural level of variation underscores our assertion that adverse effects below the 20% effect level are difficult to attribute to toxicants such as PFOS with high certainty (even with the support of robust replication and statistical testing).Folding in this level of biological variation in the context of laboratory-to-laboratory variation in organism responses makes for complicated data interpretation with regards to toxicological thresholds intended to protect natural populations from risks.Thus, we recommend that the 20% effect level (rather than lower effect levels such as 10%) should be the target effect level for identifying a low effect threshold that can be reliably evaluated for zebrafish and used to quantify precise thresholds for aquatic life.
For some regulatory agencies, 10% effect sizes remain the focus of criteria development; however, there is no difference in the overall conclusions in the present review if one were to focus on defining NOEC and LOEC values on the basis of a 10% effect size.Of the 12 studies reviewed (Supporting Information, Table S1), only one LOEC would change using a 10% effect size rather than 20% effect size-a 12% decrease in female weight detected in F0 fish exposed to 0.73 µg/L by Keiter et al. (2012).Using a 20% threshold, the LOEC would be 107 µg/L, but using a 10% threshold the LOEC would be 0.73 µg/L.Based on the inability of studies to replicate the growth effect detected by Keiter et al. (2012) at 0.73 µg/L and the methodological limitations of the Keiter et al. (2012) experimental design, these results were not included in our recommended geometric mean NOEC and LOEC calculations.Thus, the selection of 10% versus 20% does not affect our overall conclusions regarding the weight of evidence for NOEC and LOEC values for PFOS in zebrafish.
Another key issue for future aquatic life studies with PFOS is the need for multigenerational experimental designs.The perception that the ecotoxicity of PFOS must be based on the evaluation of multigenerational effects is not supported by these review findings.Based on the multigenerational studies by Keiter et al. (2012) and Gust et al. (2023), there is little evidence of improvement or worsening of adverse outcomes over multiple generations of exposure.The best evidence for multigenerational effect mechanism is potential reproductive toxicity of zebrafish exposure to concentrations of 100 µg/L or more.As discussed previously, Sharpe et al. (2010) noted that female zebrafish transferred 10% of their PFOS body burden to their embryos during one reproductive cycle (when exposed to 72 µg/L), and this may be responsible for latent toxicity to their offspring in the developmental point at which the embryos absorb their yolk sac.This mechanism is not specific to PFOS and is well known for other organic compounds.In the Keiter et al. (2012) study, exposures at approximately 100 µg/L did not cause mortality in the P generation but did cause mortality in the F1 and F2 generations.Keiter et al. (2012) reported that the 14 dpf survival of F1 and F2 fish was approximately 40% less than controls.However, due to high variability, this effect was not statistically significant.In the Gust et al. (2023) study, there was no statistically significant effect on mortality (4% increased survival compared with control) of 14 dpf F1 fish exposed to approximately 100 µg/L, but there was a statistically significant increase in mortality (36%) detected by 14 dpf in the F2 generation exposed to the same concentration.Overall, it is uncertain if the increase in mortality observed in the F1 and F2 generations (relative to parent generations) represents a mutigenerational effect or simply experiment-to-experiment variability.
Overall, the cases in which higher lethal sensitivity of offspring (compared with parents) was noted occurred in the approximate 50 to 100 µg/L range.This threshold range is comparable to the range associated with low level effects on body weight (40-100 µg/L) observed within single-generation tests.Although the best evidence for potential for cross-generational transfer of PFOS is evident in the Chen et al. (2013) and Du et al. (2008) studies, in which multigenerational effects were evident in offspring of fish exposed to PFOS, the thresholds at which this effect was noted were not greatly different than thresholds noted for growth effects (body wt) in single-generation exposures.Thus, the present review indicates that deriving a sensitive threshold concentration for PFOS in water does not necessarily require a multigenerational exposure design because simple growth measurements in a shorter exposure appear to be just as sensitive (both approaches suggest thresholds occur in the approximate 50-100 µg/L range).Thus, multigenerational studies do not appear to be required for identifying low-level screening thresholds that would be protective of populations of fish exposed to PFOS.
Many of the studies reviewed in the present study are limited by major drawbacks in experimental design, including a lack of establishing effect sizes a priori, consideration of concentrationresponse, defining the exposure, defining the baseline, insufficient replication, and not quantifying PFOS in the exposure groups.These issues are not unique to experiments exposing PFOS to zebrafish, but are also prevalent within published ecotoxicological research.For example, Sumpter et al. (2014) found that, due to considerable limitations in study designs, the LOECs from studies exposing fish to fluoxetine span several orders of magnitude, with LOECs ranging from pg/L to µg/L, which ultimately confounds the ability to establish the environmental concentrations of fluoxetine that pose a risk to aquatic organisms.Following recommendations presented elsewhere, such as the 12 principles of sound ecotoxicology outlined by Harris et al. (2014), would largely solve this issue.
Supporting Information-The Supporting Information are available on the Wiley Online Library at DOI:10.1002/etc.5768.
concentrations at which an effect was significantly different from the control (p ≤ 0.05) and biologically relevant (≥20% difference from control).b A negative percentage change represents adverse effects (decreased growth/survival/reproduction), positive values represent better performance compared with controls.c

j
Two replicate tanks (authors considered n = 80 because there were 40 fish in each tank).
Bold and underlined values indicate statistically significant differences from control.dpf = days post fertilization; F = females; hpf = hours post fertilization; L = growth (length); LOEC = lowest observed effect concentration; M = males; N/A = fish too young to be sexed, or exposed fish all one sex; NOEC = no observed effect concentration; N.S. = not statistically significant (data not provided, n = 1, etc.); PFOS = perfluorooctane sulfonate; R = reproduction; S = survival (often only for larval fish); W = growth (weight).

FIGURE 2 :
FIGURE 2: Average change in reproduction, as measured by survival of offspring to parents exposed to perfluorooctane sulfonate (PFOS; a negative % is a reduction in survival, relative to control), time in days that survival was measured is shown in the caption).(A) Embryos raised in clean water from fish parentally exposed to PFOS, and (B) embryos continuously exposed to PFOS (parentally and in rearing medium).Statistically significant decreases in survival are filled in black.

FIGURE 3 :
FIGURE 3: Summary of no observed effect concentrations (NOECs) and lowest observed effect concentrations (LOECs) in zebrafish exposed to perfluorooctane sulfonate (PFOS; µg/L).The effect size of LOECs is displayed above the value as percentage change relative to control.Endpoints associated with these effect sizes are noted by the following abbreviations: L = length; R = reproduction; S = survival; W = body weight (wet wt).All endpoints measured in a study dataset are presented in the x axis.For example, the Gust et al., (2023) F1 data set measured survival, body weight, length, and reproduction, and noted four concentrations with no effects (open green symbols), with the highest NOEC symbol (filled in green).The most sensitive endpoint was weight, indicated by W (filled orange symbol) with the significant effect (17% decreased body wt) indicated above the triangle.The recommended NOEC and LOEC screening levels of 31 and 96 µg/L, respectively, are indicated by the solid green and orange lines.

TABLE 1 :
Overview of no observed effect concentration (NOEC) and lowest observed effect concentration (LOEC) of perfluorooctane sulfonate (PFOS) associated with sublethal growth reproduction effects observed in toxicity tests with zebrafish, assuming an approximate 20% adverse effect size is the minimum ecologically significant effect size