Bioconcentration in fish
The very low bioconcentration in fish determined in the present study corresponds to the chemical properties of DCF. Diclofenac is an ionisable chemical with a pKA of 3.99 to 4.16 14, 23. Thus, water solubility and octanol-water distribution coefficient are pH-dependent. The water solubility of the DCF salt is distinctly higher than the solubility of pure DCF. The log KOW (octanol-water partition coefficient), determined for nonionised DCF at pH 3, is 4.51, which is rather high 14. However, the octanol-water distribution coefficient (log KD) is only 1.9 at pH 7.0 23 and 1.31 at pH 7.4 14. Thus, for the estimation of fish bioconcentration by the octanol-water distribution coefficient, the KD values of DCF at the environmentally relevant pH range of 6 to 9 should be taken into account.
Schwaiger et al. 10 determined the bioconcentration of DCF in different organs of rainbow trout after exposure for 28 d. Their BCFs ranged from 69 in muscle to 2732 in liver at 1.0 µg/L, and decreased continuously with increasing test concentration up to and including 500 µg/L. Other authors have published results on the bioconcentration of DCF in the bile, liver, and blood plasma of rainbow trout (summarized in Table 3). Bioconcentration of chemicals in these fish compartments can be a useful indicator for exposure for environmental monitoring purposes. However, for potential biomagnification and secondary poisoning of a chemical within the aquatic food chain, the body burden in whole fish is normally more relevant because fish-eating birds and mammal predators usually consume whole fish. The bioconcentration of DCF in whole fish was therefore measured in the present study.
Table 3. Bioconcentration factors of diclofenac in present fish study compared to published studies
|10 d||Rainbow trout 1 year||Bile||1.7||320–950|| |
Kallio et al. 18 a
|10 d||Rainbow trout 1 year||Bile, blood plasma||1.8/43||476/797, 5.7/4.9|| |
Lahti et al. 19 b
|14 d||Rainbow trout 6 months||Blood plasma, liver||1.6–81.5||4.0, 2.5|| |
Cuklev et al. 25 b
|14 d||Rainbow trout young (1.1–1.2 g wet wt)||Whole fish||2.1/18.7||5/3||Present studyc|
|21 d||Rainbow trout adult||Bile||0.5/5/25||657/534/509|| |
Mehinto et al. 13 b
|28 d||Rainbow trout 22 months||Liver||1.06–501.2||12–2,732|| |
Schwaiger et al. 10 b
The steady-state plateau was reached within a short period of approximately 10 to 14 d, and the small amounts of DCF taken up by the fish were rapidly depurated, with a half-life DT50 of approximately one day. Similar fast depuration of DCF in fish was also observed in other studies 19, 24. The very low concentrations measured in the fish at the end of the depuration period made it difficult to assess if small amounts of radioactivity still remained in the fish after the 14-d depuration phase, as indicated by the fitted depuration curve. One reason for small amounts remaining might be enterohepatic circulation of DCF in fish, as was described by Hoeger et al. 24.
All BCFs in the present study were below 10 and showed no concentration-dependency within the tested concentration range of 2 to 19 µg/L. Also, Lahti et al. 19 and Cucklev et al. 25 found no concentration-dependency for bioconcentration of DCF in fish plasma and liver. According to the technical guidance document (no. 27) for deriving environmental quality standards (https://circabc.europa.eu/d/a/workspace/SpacesStore/0cc3581b-5f65-4b6f-91c6-433a1e947838/TGD-EQS%20CIS-WFD%2027%20EC%202011.pdf), evidence for a relevant bioaccumulation is indicated at BCF's ≥ 100. Thus, the very low BCF values of the present study indicate no potential of DCF to bioconcentrate in fish. Consequently, also, the potential of DCF for secondary poisoning by fish is low.
Many studies of the acute or (sub-)chronic toxicity of DCF to aquatic organisms have been published in the past. Toxicity data for DCF after short-term exposure (up to 4 d) have been reported for several aquatic species including algae, water plants, crustacean, bivalves, and some fish species 5–9. Overall, fish seem to be most sensitive. The acute toxicity of DCF to juvenile fish is relatively low, with 96-h LC10 of 8 mg/L for Japanese medaka 26 and 96-h LC50 of 167 mg/L for zebrafish 27. As the measured DCF concentrations in surface waters are mainly in the range of ng/L to low µg/L, the acute risk for the different organism groups and trophic levels in the aquatic environment can be neglected. Human pharmaceuticals are typically discharged continuously from patient excretion via sewage effluent to surface waters. Thus, chronic exposure is much more relevant for the environmental risk assessment (ERA) than pulsed short-term exposures of, for instance, pesticides. Consequently, chronic toxicity data, as actually required by the European Medicines Agency (EMA), are needed for the aquatic ERA of pharmaceuticals 28.
Most chronic fish toxicity studies with DCF used histopathology, sub-cellular, or other endpoints. The NOEC's cover a broad range of concentrations. Histopathological findings (Table 4) often were the most sensitive endpoints. Kidney, liver, and fish gills are possible target organs for DCF, due to their detoxification ability and/or direct contact to xenobiotics. For DCF, histopathological changes in the kidney are considered likely, given the population decline of Asian vultures after consuming carcasses of DCF-treated cattle, resulting in irreversible short-term damage to their kidneys 29–31. The liver may be exposed to increased local concentrations of DCF during detoxification processes. The gills also are a potential target organ, due to the extensive exposure to xenobiotics and their high permeability for chemicals in the water.
Table 4. NOEC's (in µg/L) for histopathological effects of diclofenac in fish organs
|Exposure period||28 d||21 d||21 d||28 d||95 d|
|Fish species||Rainbow trout||Brown trout||Rainbow trout||Zebrafish||Rainbow trout|
Other sublethal fish studies conducted with DCF resulted in distinctly lower NOEC's compared to the results of the present studies (Table 4). After the exposure of adult rainbow trout for 28 d to DCF (1.0–500 µg/L), Schwaiger et al. 10 reported statistically significant histopathological effects in the kidney and gills starting at 5.0 µg/L (NOEC 1.0 µg/L). No histological effects were obtained in that study in liver, intestine, and spleen (using light microscopy). The same fish were analyzed by Triebskorn et al. 11, 32 at the ultrastructural level using electron microscopy. In that study, subcellular cytopathological effects on liver, kidney, and gills were determined already at 1.0 µg/L. Hoeger et al. 12 exposed 18 month-old brown trout to DCF (0.5, 5.0, 50 µg/L) for 21 d and conducted histopathology in gills, trunk kidney, and liver. Mild to moderate effects were observed in all organs. A NOEC of 0.5 µg/L was set, based on the effects in the liver. However, neither Schwaiger et al. 10, Triebskorn et al. 11, 32, nor Hoeger et al. 12 found a clear dose-response for their histopathological findings. Therefore, a clear-cut NOEC is difficult to identify from these results. Mehinto et al. 13 exposed juvenile rainbow trout to DCF (0.5, 1, 5, 25 µg/L) for 21 d. Histopathological evaluations showed tubular necrosis in the kidney and alterations in the intestine, but no morphological changes in the liver. Based on the histological findings, they proposed a NOEC of 1.0 µg/L. However, Praskova et al. (E. Praskova et al., University of Veterinary and Pharmaceutical Sciences Brno, Department of Veterinary Public Health and Toxicology, Brno, Czech Republic, unpublished manuscript) did not find any histopathological effects in a 28-d toxicity study with juvenile zebrafish and exposure at 0.02 to 60 mg/L. Taken together, the data in the available literature show inconsistent results that make it difficult to determine a no-effect level for DCF.
The low threshold effect levels of 0.5 to 1.0 µg/L obtained in some of these published studies could not be confirmed in the present ELS studies with rainbow trout and zebrafish despite the very long exposure periods (compared to other studies) and the use of typically very sensitive early life stages, and despite the combination of both histopathology and population-relevant endpoints in the trout study. Histopathological alterations were found in trout gills at 1084 µg/L, but no relevant histopathological symptoms were observed in kidney and liver up to the highest test concentration. However, none of the population-relevant (apical) endpoints such as hatching, development, growth, or survival were affected up to and including the highest test concentration of 1084 µg/L. The mean body weight and length of the trout at the higher test concentrations were even slightly larger compared to the control fish. Consequently, the present study with rainbow trout demonstrates an overall NOEC (including all monitored population-relevant endpoints as well as histopathology of the potentially targeted organs gills, kidney, and liver) at 320 µg DCF/L.
While the present ELS test with rainbow trout demonstrated that DCF clearly had no inhibitory effect on fish growth up to and including the highest test concentration of 1000 µg/L, the growth effects of DCF on zebrafish in the present study were less straightforward to interpret. The moderate but more or less constant size reduction of the zebrafish at the end of the study over a wide concentration range (32–1000 µg/L) can be interpreted in two different ways. Based on the statistical results, interpretation A suggests a NOEC for the growth of the zebrafish at 10 µg/L, because the mean length and weight at 32 µg/L were statistically significantly reduced. However, the mean values at the next higher test concentrations were not always significantly different from the control. Interpretation B considers the moderately reduced zebrafish growth from 32 to 1000 µg/L as an artifact of some sort, which is not a treatment-related, repeatable, adverse effect of DCF.
In our opinion there are three reasons to favor interpretation B. The first reason is the absence of a dose-response relationship in the wide concentration range of 32 to 1000 µg/L. No plausible toxicological explanation is known to us, which could explain this plateau effect of a constant inhibition in growth over such a large concentration range. If DCF at a concentration of 32 µg/L does produce an adverse effect on fish growth, this effect would be expected to be progressively larger at 100, 320, and 1000 µg/L DCF, respectively. In contrast, fish size at these concentrations showed a typically normal pattern for fish growth, with no trend to decrease with increasing test concentrations. At the next higher test concentration of 3200 µg/L, both the mean length and weight of the zebrafish were clearly reduced (reduction in length by 32%, in wet wt by 61% compared to the control). Thus, a typical dose-response relationship was observable only between 1000 and 3200 µg/L. The second reason supporting interpretation B comes from the results of a recently submitted 28-d growth study with zebrafish according to the OECD test guideline 215 33. Praskova et al. (E. Praskova et al., University of Veterinary and Pharmaceutical Sciences Brno, Department of Veterinary Public Health and Toxicology, Brno, Czech Republic, unpublished manuscript) found no significant inhibitory effect of DCF on the growth of zebrafish up to 5 mg/L. Once confirmed by peer review, this result would strongly support our conclusion that it is unlikely that DCF inhibits the growth of zebrafish in the present ELS study within the concentration of 32 to 1000 µg/L. The third reason supporting interpretation B comes from the findings of the ELS study in rainbow trout, where clearly no inhibitory effect on growth was obtained up to the highest test concentration of 1000 µg/L. A 100-fold difference in growth sensitivity of two teleost fish species toward DCF resulting in a NOEC of 10 µg/L in zebrafish and a NOEC of ≥1000 µg/L in rainbow trout seems very unlikely. No reason is known to us to explain this apparently extreme susceptibility of zebrafish compared to rainbow trout.
In general, for a chemical like DCF, with a receptor mediated, specific mode-of-action (MOA), similar sensitivities can be expected for all teleost fish. For example, all species of fish investigated to date with the human pharmaceutical ethinylestradiol displayed relatively similar sensitivities 34, as a consequence of all fish species possessing estrogen receptors, which are the key targets for that particular pharmaceutical. In humans, DCF acts essentially as a cyclooxygenase (COX) inhibitor. If DCF also acts via this specific MOA in zebrafish and trout (which is likely, due to the largely conserved structure of COX genes in zebrafish 35, trout 36, and mammals, respectively), then we would expect both species to demonstrate relatively similar sensitivities to DCF. Based on the three reasons discussed above we consider the moderately reduced zebrafish growth in the concentration range of 32 to 1000 µg/L as an artifact of some sort, but not as a treatment-related, repeatable, adverse effect of DCF. In our view, a real adverse effect on growth may have been present first at 3200 µg/L. Our assumption is that a faster growth of the zebrafish in the control group was the reason for the difference in fish size in the concentration range of 10 to 1000 µg/L. A faster growth rate can happen by an unknown mechanism as a consequence of the test design or even by chance. For example, Owen et al. 37 reported very similar findings to those observed in the present zebrafish study. They tested the effect of clofibric acid in a fish growth study with rainbow trout according to OECD test guideline 215 33 and observed a significant reduction in fish weight and growth rate already at the lowest test concentration of 0.1 µg/L. A very similar inhibition of the growth rate by approximately 50% was obtained at all test concentrations up to 10,000 µg/L. Due to this unexpected result without any dose-response, Owen et al. 37 repeated parts of the study by testing the lower test concentrations again but this time with more replicates per treatment, to increase the statistical power. In the study repeat no adverse effects on fish growth were obtained, that is, the effects observed in the first test were not reproducible. After a detailed evaluation of the results of these two tests, Owen et al. 37 came to the conclusion that the results of the first test “could be attributed primarily to an exceptionally fast growth rate in the control fish.”
The reason for the differences in the growth of the zebrafish in the middle test concentration range of the present ELS study remains unknown. However, when the present zebrafish and rainbow trout studies, the published genetic evidence, and the emerging zebrafish study of Praskova et al. (E. Praskova et al., University of Veterinary and Pharmaceutical Sciences Brno, Department of Veterinary Public Health and Toxicology, Brno, Czech Republic, unpublished manuscript) are considered together, we strongly believe that the reduced growth in the zebrafish study in the concentration range of 32 to 1000 µg/L could be attributed primarily to an exceptionally fast growth in the control fish compared to the growth in the treated fish. This leads us to the conclusion that DCF has, with high probability, no inhibitory effect on fish growth up to at least 320 µg/L. This proposed NOEC of 320 µg/L for zebrafish is identical to the overall NOEC of the present trout study. The only difference is the NOEC trigger in zebrafish, which was a reduction in survival and possibly in growth at 1000 µg/L, while the trigger in the trout study were the histopathological findings.
A large discrepancy remains between the NOEC's that have been postulated in some of the published studies and those determined in the recent toxicity tests with rainbow trout and zebrafish. The NOEC of the present ELS test in rainbow trout is up to 640-fold higher than the lowest postulated NOEC of 0.5 µg/L 12. The reasons for this discrepancy cannot be explained with absolute certainty by the present study. Comparing the published histopathological effects of DCF, they show several inconsistencies. The interstudy impact of DCF on different fish organs and the specified symptoms are partly contradictory (Table 4). For example, Mehinto et al. 13 found no pathological effects in trout liver, but did report effects in the kidney and in the intestine. Schwaiger et al. 10 described the most prominent effects in gills, followed by kidney, but no effects in trout liver or gastro-intestinal tract. In contrast, Hoeger et al. 12 found the strongest effects in liver, followed by gills and kidney. All these studies were conducted with brown or rainbow trout. It seems unlikely that the differences obtained were caused by a different mode of action of DCF in these related trout species. The inconsistencies may, however, have been caused by other factors. One factor may be the different classification systems used to evaluate incidence and severity grade of potential symptoms. Overall assessment based on severity can be biased, for example, by combining minimal or slight findings with moderate symptoms in one coarse severity class. In addition, differences between the treatments and the control can only be assessed in a reliable manner if the baseline frequency of symptoms in the control is known 38. The relatively low number of histopathologically analyzed control fish in the earlier published studies leaves some doubts as to whether such baseline frequencies were reliably quantified and considered. High experience in histopathology is needed to avoid misdiagnosis of results because histopathological findings can be influenced by many factors, such as biological variability, diseases, parasites, or other stress in the test fish, for example from too high fish density. Additionally, histopathology leaves room for subjective interpretation. The semi-quantitative results, obtained through histopathology, require special statistical methods, and most importantly, expert judgement to avoid misdiagnosis based on over-interpreted, isolated findings. This point can be exemplified by the difficulties and recent efforts to harmonize the interpretation of fish gonad pathology findings. Fish pathology experts recently developed an OECD guidance document for the technical preparation and histopathological evaluation of fish gonads 39. An official guidance document for histopathology in fish organs other than gonads is, however, still missing. Such technical guidance and a validated rating system might help to avoid the inconsistencies obtained in the case of DCF fish pathology. Presently, it is very difficult to link histopathological results obtained in fish studies to adverse effects on a fish population level. To allow such an extrapolation from histopathological findings to population-relevant apical endpoints (i.e. development, growth, survival, or reproduction), studies are needed which include both high quality histopathology as well as population-relevant endpoints. Such studies are still scarce for DCF.
In Europe, environmental quality standards are proposed as legally binding target levels for selected surface water contaminants. The technical guidance document (no. 27) for deriving environmental quality standards (https://circabc.europa.eu/d/a/workspace/SpacesStore/0cc3581b-5f65-4b6f-91c6-433a1e947838/TGD-EQS%20CIS-WFD%2027%20EC%202011.pdf) states that test results based on endpoints of whose relationship to effects at the population level is uncertain are unsuitable and should not be used for environmental risk assessment-based decision making. According to this guidance document, histopathological data (with the exception of gonad histology), or findings on a sub-cellular level such as changes in enzyme induction or gene expression, belong to this group of endpoints with unclear population relevance. In summary, histopathological symptoms or sub-cellular endpoints should for now only be used as indicators for further evaluation, but should not be used as decision criteria in ERA processes.
So far, there is limited information available on the effects of long-term exposure of DCF to aquatic organisms. No data from an OECD test guideline 210 ELS test with standardized endpoints or an OECD test guideline 305 fish bioconcentration study were available as required for the ERA, for example, for marketing authorization of human pharmaceuticals in Europe 28. The present studies, conducted according GLP and in accordance with the validated and internationally accepted OECD test guidelines, address these gaps now. Their results should be used in future for the derivation of a robust environmental quality standard for DCF in surface waters under the Water Framework Directive in the European Union.