Food and Chemical Toxicology 48 (2010) S75–S80
Contents lists available at ScienceDirect
Food and Chemical Toxicology
journal homepage: www.elsevier.com/locate/foodchemtox
Application of the margin of exposure (MOE) approach to substances in food that
are genotoxic and carcinogenic. Example: Leucomalachite green
Andrew Renwick a,*, Jean-Charles Leblanc b, R. Woodrow Setzer c
a
University of Southampton, UK
French Food Safety Agency, France
c
US Environmental Protection Agency, USA
b
a r t i c l e
i n f o
Article history:
Received 8 June 2009
Accepted 21 September 2009
Keywords:
Leucomalachite green
Margin of exposure
MOE
a b s t r a c t
Leucomalachite green (LMG) is mutagenic and produces DNA-adducts in vivo, and is carcinogenic in
rodent bioassays. Dose-response modelling of the data for hepatocellular adenomas and carcinomas in
female mice gave a BMDL10 of 20 mg/kg-bw/day. Limited data are available on the concentrations present in fish for human consumption. Human exposure estimates assumed that all consumed fish is contaminated with LMG. The calculated MoEs were 4,000,000 and 400,000 respectively for average and
high exposure estimates.
Ó 2009 ILSI Europe. Published by Elsevier Ltd. All rights reserved.
Introduction
Malachite green chloride is a triphenylmethane dye used illegally
in the fish culture as an antifungal agent and in dye industries. Leucomalachite green (LMG) is formed in vivo by the reduction of malachite
green (MG). Although MG is not approved for use in aquaculture, its
low cost and high efficacy make illicit use likely, and residues of LMG
have been reported in fish for human consumption.
1. Toxicological data
1.1. Genotoxicity
The following summary is based on a joint review by the United
Kingdom Committee on Carcinogenicity (UK CoM) and Committee
on Mutagenicity (UK CoM) in 2004 (CoC/CoM, 2004). It provides an
update to the data available to the Committee on Mutagenicity in
1999 (CoM, 1999).
Negative results were obtained on the in vitro mutagenicity of
LMG from limited Salmonella and CHO/HGPRT assays, and also from
a Comet assay in CHO cells. 32P-post-labelling studies using a 28day dietary exposure have indicated that LMG induces a low level
of DNA-adduct(s) in the liver of F344 rats; only a marginal response
was however seen in B6C3F1 mice. DNA-adducts have been reported in the livers of mice treated with LMG, with a dose-related
increase in binding. LMG did not induce micronuclei in peripheral
* Corresponding author. Address: ILSI Europe, Avenue E. Mounier 83, B6 1200
Brussels, Belgium. Tel.: +32 2 771 00 14; fax: +32 2 762 00 44.
E-mail address: [email protected]
lymphocytes or HPRT mutations in splenic lymphocytes in the
28-day study. Gene mutation studies have indicated that LMG can
induce mutations in vivo in liver-DNA. Studies in Big Blue F344 rats
gave equivocal results. LMG produced an increase in lacI mutations
only at the highest dose tested (543 ppm in diet) and at a single
time point (16 weeks). No increase was seen after 32 weeks. Studies
in female Big Blue B6C3F1 mice indicated that LMG produced an increase in cII mutant frequency in liver-DNA of mice treated with
408 ppm LMG in the diet. Furthermore, it was shown that the spectrum of mutation seen in the DNA was distinct from that of the control mice. These data provide evidence of in vivo mutagenicity at the
target site in the carcinogenicity bioassay.
The 2004 statement (CoC/CoM, 2004) concluded ‘‘In view of the
demonstration of the induction of mutations in liver-DNA of female B6C3F1 mice, LMG should be regarded as an in vivo mutagen”.
1.2. Carcinogenicity
The CoC paper included a detailed description of a National Toxicology Program (NTP) cancer bioassay which has now been published on the NTP website (NTP, 2004, 2005), and a paper has
been published in Food and Chemical Toxicology (Culp et al., 2006).
1.2.1. Rats
Groups of 48 male and female F344 rats were fed diets containing 0, 91, 272 or 543 ppm LMG for 104 weeks. The dietary levels
were equivalent to average daily doses of approximately 0, 5, 15
and 30 mg LMG/kg-bw/day in the males and 0, 6, 17 and 35 mg
LMG/kg-bw/day in the females. Survival was comparable in all
groups apart from the males given 272 ppm where increased sur-
0278-6915/$ - see front matter Ó 2009 ILSI Europe. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.fct.2009.09.025
S76
A. Renwick et al. / Food and Chemical Toxicology 48 (2010) S75–S80
vival was noted. A dose-related decrease in body weight gain was
seen for both sexes amounting to a 25%-reduction near termination
in the females and 10–15% in the males. Liver weights were significantly increased at the two higher dose levels in the males, as were
relative liver weights in the females. Relative thyroid weights were
increased significantly in both sexes at the highest dose level.
At necropsy, a significant increase in mammary gland adenoma
and carcinoma (combined) was seen with the effects at the top
dose level exceeding the historic control range in experiments conducted at the test laboratory. In addition, in the male rats a positive
trend was noted in the incidence of interstitial cell adenoma of the
testes, with the increase being statistically significant at the top
dose level (37/48, 42/47, 43/48 and 45/47, respectively).
A minimal increase in hepatocellular adenomas was also seen in
the female rats given 91 and 543 ppm in the diet which exceeded
the historical control range (1/48, 3/48, 0/48 and 3/48 at 0, 91, 272
and 543 ppm in the diet, respectively). No increase in such tumours was seen in the males (2/48, 2/47, 3/48 and 2/47 at 0, 91,
272 and 543 ppm in the diet, respectively). Non-neoplastic liver lesions (eosinophilic foci, cystic degeneration and cytoplasmic vacuolisation) were increased in both males and females. Minimal
increases, not statistically significant, were seen in thyroid follicular cell adenomas and carcinomas (combined) in both the females
(0/46, 1/46, 2/47 and 1/48, respectively) and males (0/47, 2/47, 1/
48 and 3/46, respectively). Decreases in the incidence of mononuclear cell leukemia and pituitary gland adenoma were seen in both
sexes. There was a decrease with treatment in the total numbers of
animals showing a malignant neoplasm.
1.2.2. Mice
Groups of female mice were fed diets containing 0, 91, 204, or
408 ppm LMG (equivalent to average daily doses of 0, 13, 31 and
63 mg LMG/kg-bw/day) for 104 weeks. Survival and body weight
gain were comparable in all groups. Relative kidney weight was
significantly reduced in all treated groups. The only statistically
significant (P = 0.022) increase in tumour incidence noted at necropsy related to hepatocellular adenoma and carcinoma combined.
The incidence of hepatocellular adenoma alone was increased but
this was not statistically significant.
The conclusions of the NTP (as agreed by the peer-review panel,
NTP, 2005) were
i. there was equivocal evidence of carcinogenic activity in
male F344/N rats based on an increase in interstitial cell adenoma of the testes and the occurrence of thyroid gland follicular cell adenoma or carcinoma (combined) in exposed rats,
ii. there was equivocal evidence of carcinogenic activity in
female F344/N rats based on a marginally increased incidence of hepatocellular adenoma and the occurrence of thyroid gland follicular cell adenoma or carcinoma (combined)
in exposed rats, and
iii. there was some evidence of carcinogenic activity in female
B6C3F1 mice based on an increase in hepatocellular adenoma or carcinoma (combined).
Based on the peer-review it would seem probable that the data
for mice on hepatocellular adenoma and carcinoma combined
would be the endpoint to use for BMD/BMDL estimations and calculation of the MOE.
1.3. Mode of action
The mode of action was considered in the unpublished 2004
CoC report, which concluded ‘‘In view of the demonstration of
the induction of mutations in liver-DNA of female B6C3F1 mice,
LMG should be regarded as an in vivo mutagen”.
1.4. Epidemiological data
None identified.
1.5. Dose–response relationships
The raw incidence data for both rats and mice are given in the
following table.
Tumour type
Incidence at different dietary concentrations
0 ppm
(0 mg/
kg/day)
Tumours in male rats
Testes,
37/48
interstitial
cell adenoma
Mammary
0/48
adenoma or
carcinoma
0 ppm
(0 mg/
kg/day)
Tumours in female mice
Hepatocellular 3/47
adenoma or
carcinoma
91 ppm
272 ppm
(5 mg/kg/ (15 mg/
day)
kg/day)
543 ppm
(30 mg/
kg/day)
42/47*
43/48*
45/47*
2/48
3/48*
4/48
91 ppm
(13 mg/
kg/day)
204 ppm
(31 mg/
kg/day)
408 ppm
(63 mg/
kg/day)
6/48
6/47
11/47*
*
Increase reported as statistically significant (P < 0.05) Culp et al.
(2006).
The main dose–response identified by the NTP was an increase in
hepatocellular adenoma or carcinoma (combined) in female
B6C3F1 mice.
1.6. Data quality, uncertainties and limitations
The data are from a recent NTP 2 year bioassay and therefore
‘‘gold standard”.
2. Human dietary exposure analysis
2.1. Sub-populations of interest
None.
2.2. Concentration in food
Malachite green is used to treat fungal and protozoan infections
in ornamental fish. Although it is not registered for use in aquaculture for food in many countries, including Australia, Canada, the
European Union, Hong Kong, Indonesia, the United States and Vietnam, residues are detected in farmed fishes.
In 2005, the Food Standards Australia New Zealand (FSANZ)
conducted a Survey of Chemical Residues in Domestic and Imported
Aquaculture Fish. Sixty domestic and imported aquacultured fish
sampled to account for country of origin and species to best represent the aquacultured fish available for sale in Australia were collected and analysed. Leucomalachite green was quantified in 10
(17%) samples, with levels between 4 and 110 lg/kg, the limit of
detection (LOD) being 2 lg/kg. The lower bound mean (assuming
0 as the concentration for non-detects) was 5 lg/kg and the upper
bound mean (assuming the LOD, 2 lg/kg, as the concentration for
non-detects) was 7 lg/kg (FSANZ, 2005).
S77
A. Renwick et al. / Food and Chemical Toxicology 48 (2010) S75–S80
Malachite green and its metabolite leucomalachite green are
monitored in a limited number of countries. The United Kingdom
has published detailed results since 2001 (Veterinary Residues
Committee, 2001, 2002, 2003, 2004, 2005). Their monitoring programmes include:
– statutory surveillance on domestic products, following a sampling design based on national production volume, in accordance with the European Council decision 96/23/EC,
– non-statutory surveillance on imported products, targeted on
areas where residues of concern are more likely to occur,
– retail surveys aimed to assess the incidence of residues in products available from retail outlets.
Recent publications, primarily concerned with analytical methods for the determination of malachite and leucomalachite green,
have confirmed the prevalence and levels of these illegal colours
in farmed fish (Halme et al., 2007; Tittlemier et al., 2007). A study
of the effects of cooking on these colours has shown that boiling in
water has little effect on the level of leucomalachite green in fish
muscle. However, microwave cooking of fish resulted in some
decomposition of leucomalachite (Mitrowska et al., 2007).
Country
Mean consumers
(ng/kg-bw/day)
95th percentile consumers
(ng/kg-bw/day)
Australia
EU
8–21
2–7
24–65
5–16
2.3.2. International estimates
According to the GEMS/Food Cluster Diets (WHO, 2007), the
consumption level for marine and freshwater fishes, fresh and processed, ranges between 12 g/day/capita in Cluster I (South and East
Africa) and 53 g/day/capita in Cluster L (Far-Eastern Asia). There is
no distinction among fish species consumed, nor are wild or
farmed fishes differentiated in these diets. Combining the
weighted mean LMG concentration reported in the table above
(3.6 lg/kg) with the fish consumptions from the GEMS/Food Cluster Diets, assuming a body weight of 60 kg for consumers in all
clusters, results in intakes ranging between 0.7 ng/kg-bw/day
(Cluster I, South and East Africa) and 3.2 ng/kg-bw/day (Cluster L,
Far eastern).
Survery
Number
of samples
Number of samples
with residues (%)
Mean concentration
(lg/kg)
Mean concentration contaminated
samples only (lg/kg)
Max concentration
(lg/kg)
GB monitoring
programmes
2001–2005
Australian survery
1953
87 (4.5%)
3.7
36.1
406
60
10 (17%)
7.3
33.0
110
Total
2013
97 (5%)
3.6
36.1
406
2.3. Dietary exposures
2.3.1. National estimates
The FSANZ provides a national assessment of the public health
risk associated with low residues of malachite green chloride and
leucomalachite green in aquacultured fish (FSANZ, 2005) based on:
– the results of the Survey of Chemical Residues in Domestic and
Imported Aquaculture Fish,
– consumption data from the 1995 Australian National Nutrition
Survey (NNS) (N = 13858 aged 2 years and above; 989 aged 2–
6 years), assuming no distinction among fish species, and
including fish consumed on its own or as an ingredient in a
mixed food dish.
The mean dietary exposure to leucomalachite green was estimated to range between 8 and 21 ng/kg-bw/day for the whole Australian population, including young children aged 2–6 years. The
95th percentile estimate ranged between 24 and 65 ng/kg-bw/day.
Additionally, dietary exposures were estimated using food consumption data for the general European Union (EU) populationbased on data from three members states reported in an European
Food Safety Authority opinion (EFSA, 2005). The food consumptions
were combined with the mean concentration reported in the GB
monitoring program (3.7 lg/kg) to derive dietary exposure estimates for EU adult consumers of fish products in the three reported
countries. The estimates include children and consumers of fish
products at a high percentile of the total population as reported in
the Calipso survey (Bemrah et al., 2009). The estimates of dietary
exposure in the adult EU population, assuming a 60 kg average body
weight for adults and 30 kg for children, ranged between 2 and 7 ng/
kg-bw at the mean, and up to 16 lg/kg-bw at the 95th percentile.
2.4. Dietary exposure assessment (values to be used in the MOE
calculation)
Dietary exposure to LMG will be intermittent because its presence in fish muscle is due to the illegal use of malachite green in
farmed fish. Available data suggest that only about 10–15% of fish
contain residual levels of LMG. Also, the data available on fish consumption generally do not distinguish between farmed and wildcaught fish. Therefore, any estimate of LMG consumption will be
conservatively high. For contaminants with this exposure pattern,
the use of a population-based mean exposure estimate should be
used for the dietary exposure to the substance. The national population means ranged from 2 to 21 ng/kg-bw/day (down to 0.7 ng/
kg-bw/day for the international estimate using the Cluster Diet
with the lowest fish intake).
The use of 5 ng/kg-bw/day, approximately the average of the
mean national estimates reported for adults, is appropriate for calculating MOEs for leucomalachite green for average intakes, with a
high value of 50 ng/kg-bw/day used to reflect the 95th percentile
reported for Australia.
2.5. Data quality, uncertainties and limitations
Initial review of the data indicates that occurrence data for LMG
in fish are very limited. Estimates of international dietary exposure
are based on levels found in two countries, which is not likely to be
representative of LMG concentrations that occur worldwide.
All of the dietary exposure estimates are considered to be conservatively high because it has been assumed that all fish (fresh
and marine) consumed are contaminated at the mean leucomalachite green concentration reported.
S78
A. Renwick et al. / Food and Chemical Toxicology 48 (2010) S75–S80
3. Modelling
Loglogistic, logprobit, gamma, Weibull, multistage, probit and
logistic models were fitted to the data, and the output of the average model used for determining the critical tumour type for risk
assessment and estimation of the MOE values, using the method
described in Wheeler and Bailer (2007).
3.1. BMD and BMDL
The following tumour types gave a positive trend analysis in the
BMD calculations undertaken for this investigation. The data are
arranged in order of increasing value of the BMDL 10 values calculated by the different models.
The average model is discussed in another paper in this issue
(Benford et al., this issue). It is clear from the above table of individual models that the BMDL values derived from the logprobit
model have a major effect in the average model at low BMRs of
5% and especially at 1%. The reason for this is that the parameter
beta in the logprobit model is not restricted (Wheeler and Bailer,
2007) so that the fitted curve, while flat with zero slope at very
low doses, can rise very steeply over a short dose range. In this
model, a very low (virtually 0) BMD is consistent with the data.
If beta were restricted to be >1, then the BMDL would be substantially greater than 0. Thus it is clear that decisions made about
data modelling could have a major effect on the estimated BMDLs,
especially for sub-optimal dose–response data (see Section 3.3 of
Benford et al., this issue for discussion). The data for the BMDL10
BMD(L) values for tumours in animals treated with leucomalachite green.
MR – testes interstitial cell adenomas
FM – hepatocellular adenomas or carcinoma (combined)
FR – mammary adenomas
BMD10
BMDL10
BMD05
BMDL05
BMD01
BMDL01
0.91
39.57
39.07
0.0003
20.44
23.46
0.33
20.43
17.06
<0.0001
6.81
8.81
0.03
4.17
1.32
<0.0001
0.0047
<0.001
The data are in mg/kg-bw/day.
FM – female mice.
MR – male rats.
FR – female rats.
For interstitial cell adenomas in the testes of rats all models fitted
the data with P values >0.41. The response at the lowest dose is
similar to the responses at higher doses so that fitting of the entire
dataset would be consistent with a very rapid rise in response over
a very low dose range followed by a shallow increase in the range
of the administered doses. Thus, the BMDL is estimated to be a value that rounds close to 0, which means the data are inadequate to
put a lower bound on the estimate of the BMD. If interstitial cell
adenomas in the testes of rats were selected as the most sensitive
endpoint of relevance to humans the available data would be inadequate for estimation of either a reliable BMDL or a usable MOE.
Selection of this endpoint would lead to the need for a new cancer
bioassay with more appropriate dose selection.
The data for mammary adenomas plus carcinomas in female
rats (all P values >0.40) and for hepatocellular adenomas or carcinomas (combined) in female mice (all P values >0.48) were significantly fitted by all models.
Because of the doubtful relevance of tumours of the interstitial
cells of the testes in F344 rats to humans (see Cook et al., 1999 and
Benford et al., this issue) and the high and variable incidences in
historical control groups (69–90%), the very low BMDL10 for this
endpoint was not used to calculate the MOE. Because the BMDL10
for mammary tumours is higher than that for hepatocellular tumours, the BMDL values for hepatocellular adenomas and carcinomas (combined) in female mice were used as the basis for the MOE
calculation.
derived by the average model were used to calculate the MOE
(see below).
3.2. T25 calculation
The T25 was calculated from the lowest dose giving a significant
increase compared to controls in the study of Culp et al. (2006). The
extra risk was calculated by dividing the additional risk (% tumourbearing animals in the dosed group minus % tumour-bearing animals in the control group) by the non-affected fraction in the control
population. The extra risk at the lowest dose giving a statistically significant increase = [(B/100 A/100)/(1 A/100)] 100 where
A = proportion of animals with tumours in the control group and
B = proportion of animals with tumours in an exposed group.
Although increased incidences of hepatocellular adenomas and
carcinomas (combined) were found in female mice at each dose level, the lowest dose showing a significant increase (P = 0.004) was
at the top dose. The incidence in the control group was 6% and in
the 63 mg/kg-bw/day group was 23%. Therefore, the extra risk is
[(23/100) (6/100)/(1–6/100)] 100 = 18.1% and the T25 = 25/
18 63 mg/kg-bw/day = 87 mg/kg-bw/day.
3.3. MOE calculation
The most sensitive site for tumour development is the interstitial
cells of the testes in F344 rats; but there is a very high spontaneous
Modelling of data for hepatocellular adenomas and carcinomas (combined) in female mice (see Fig. 1).
Gamma
Weibull
Loglogistic
Probit
Logistic
Multistage
Logprobit
Avg. model
P value
AIC
BMD10
BMDL10
BMD05
BMDL05
BMD01
BMDL01
0.80
0.80
0.80
0.78
0.77
0.51
0.48
0.70
150
150
150
150
150
152
152
35.92
35.92
34.80
42.33
43.49
39.10
32.39
39.57
20.16
20.16
18.46
29.54
31.16
20.20
17.49
17.49
16.48
24.03
25.30
20.02
14.64
20.43
9.81
9.81
8.74
16.86
18.14
9.83
3.43
3.43
3.16
5.52
5.96
3.99
3.30
4.17
1.92
1.92
1.68
3.86
4.23
1.93
20.44
6.81
0.0047
S79
A. Renwick et al. / Food and Chemical Toxicology 48 (2010) S75–S80
0.3
Fraction Affect
Fraction Affect
0.3
0.2
0.1
0.0
0.2
0.1
0.0
0
10
20
30
40
50
60
0
10
20
dose
logistic
0.2
0.1
0.0
50
60
40
50
60
40
50
60
40
50
60
0.2
0.1
0.0
0
10
20
30
40
50
60
0
10
20
dose
loglogistic
30
dose
multistage
0.3
Fraction Affect
0.3
0.2
0.1
0.0
0.2
0.1
0.0
0
10
20
30
40
50
60
0
10
20
30
dose
weibull
dose
probit
0.3
0.3
Fraction Affect
Fraction Affect
40
0.3
Fraction Affect
Fraction Affect
0.3
Fraction Affect
30
dose
gamma
0.2
0.1
0.0
0.2
0.1
0.0
0
10
20
30
40
50
60
0
10
dose
logprobit
20
30
dose
Average Model
Fig. 1. Model fitting to data for hepatocellular adenomas and carcinomas in female mice.
rate of these tumours in F344 rats and tumours at this site usually
arise from non-genotoxic mechanisms, although the mode of action
of LMG is unclear. Because of limitations in the available data the
BMDL values could not be derived for this site. The MOE has therefore been calculated for hepatocellular adenomas in female mice.
The MOE values have been calculated based on intakes of 5 ng/
kg-bw/day, approximately the average of the mean national estimates reported for adults, with a high value of 50 ng/kg-bw/day
used to reflect the 95th percentile reported for Australia, and the
BMDL values for hepatocellular adenomas and carcinomas in female mice (BMDL10, BMDL05 and BMDL01 – calculated by the
average model) treated with LMG (Section 3.1).
The MOE values for the BMDL10 are similar to those reported by
FSANZ at the National level taking the non-neoplastic lesions in
the liver as the most sensitive endpoint (FSANZ, 2005). The MOE
values for BMDL01 are heavily influenced by the output of the logprobit model and decisions about model restriction (see above).
3.4. Modelling limitations
The design of the study was inadequate to estimate either the
BMD or the BMDL for interstitial cell adenoma in rat testes. The response at the lowest dose is similar enough to the responses at
higher doses that the best-fitting model has a very rapid rise in re-
MOEs calculated for different values of the BMDL (mg/kg-bw/day) with an intake from 5 to 50 ng/kg-bw/day (see Fig. 1).
(5 ng/kg-bw/day)
MOE for BMDL10
MOE for BMDL05
MOE for BMDL01
MOE
MOE
20.44/0.000005 = 4,000,000
20.44/0.00005 = 400,000
6.81/0.000005 = 1,000,000
6.81/0.00005 = 100,000
0.0047/0.000005 = 1000
0.0047/0.00005 = 100
S80
A. Renwick et al. / Food and Chemical Toxicology 48 (2010) S75–S80
sponse over a very low dose range to a plateau in the range of the
administered doses. Thus, the data are inadequate for placing a
lower bound on the BMD, and the BMDL is estimated to be a value
that rounds to 0. The dose–response data for hepatocellular adenomas and carcinomas (combined) in female mice provided good
BMD and BMDL values for BMRs of 10% and 5%, but the BMDL01
estimate was three orders of magnitude below the BMD01. Apart
from the logprobit model, the different models gave BMDL10 values that varied within a factor of about 3-fold.
4. Learning points
i. The critical issue is the relevance of the animal tumours to
human health; calculation of an MOE does not resolve this
issue.
ii. The data on the testicular interstitial cell tumours in F344
rats, were not used because they are of doubtful relevance
to humans. If they had been used, the very low BMDL10,
which was three orders of magnitude lower than the
BMD10, would have resulted in an MOE of less than 100.
iii. The BMDL10 values for hepatocellular adenomas or carcinoma (combined) in female mice and mammary adenomas
in female rats differ from the corresponding BMDs by a factor of 2-fold or less, but the BMDL01 values can be orders of
magnitude below the corresponding BMD01. An exception is
the logprobit model where the choice of restriction would
significantly affect the derivation of BMDLs using the average model.
iv. Only limited data are available on the concentrations of LMG
in fish.
v. The MOE values are heavily dependent on the assumptions
necessary to get any estimate of intake.
Conflict of Interest
The authors declare that there are no conflicts of interest.
References
Benford, D., Bolger, P.M, Carthew, P., Coulet, M., DiNovi, M., Leblanc, J.-C., Renwick,
A.G., Woodrow Setzer, R.W., Schlatter, J., Smith, B., Slob, W., Williams, G.,
Wildemann, T., this issue. Application of the margin of exposure (MOE)
approach to substances in food that are genotoxic and carcinogenic. Food
Chem. Toxicol.
Bemrah, N., Sirot, V., Leblanc, J.C., Volatier, J.L., 2009. Fish and seafood consumption
and omega 3 dietary intake in French coastal populations: CALIPSO survey.
Public Health Nutr. 12, 599–608.
CoM, 1999. Committee on mutagenicity of chemicals in food, consumer products
and the environment. Malachite Green and Leucomalachite Green, COM
Statement COM/99/S1 (for COT). Available at: <http://www.advisorybodies.
doh.gov.uk/com/malachit.htm>.
CoC/CoM, 2004. Joint CoM and CoC Statement on Mutagenicity and Carcinogenicity
of Malachite Green (MG) and Leucomalachite Green (LMG) COM/04/S4 and
COC/04/S7 – December 2004. Available at: <http://www.iacom.org.uk/
statements/COM04S4.htm>.
Cook, J.C., Klinefelter, G.R., Hardisty, J.F., Sharpe, R.M., Foster, P.M., 1999. Rodent
leydig cell tumorigenesis: a review of the physiology, pathology, mechanisms,
and relevance to humans. Crit. Rev. Toxicol. 29, 169–261.
Culp, S.J., Mellick, P.W., Trotter, R.W., Greenlees, K.J., Kodell, R.L., Beland, F.A., 2006.
Carcinogenicity of malachite green chloride and leucomalachite green in
B6C3F1 mice and F344 rats. Food Chem. Toxicol. 44, 1204–1212.
EFSA, 2005. European Food Safety Authority. Opinion of the scientific panel on
contaminants in the food chain on a request from the commission related to the
presence of non-dioxin-like PCB in feed and food. EFSA J. 284, 1–137.
FSANZ, 2005. Food Standards Australia and New Zealand. Report on a survey of
chemical residues in domestic and imported aquacultured fish. A National
Survey Conducted under the Coordinated Food Survey Plan with Participation
by Food Regulatory Agencies in all Australian Stated and Territories. Available
at: <www.foodstandards.gov.au/_srcfiles/Chemical%20Residues%20in%20Fish%
20Survey.pdf>.
Halme, K., Lindfors, E., Peltonen, K., 2007. A confirmatory analysis of malachite
green residues in rainbow trout with liquid chromatography–electrospray
tandem mass spectrometry. J. Chromatogr. B 845, 74–79.
Mitrowska, K., Posyniak, A., Zmudzki, J., 2007. The effects of cooking on residues of
malachite green and leucomalachite green in carp muscles. Anal. Chim. Acta
586, 420–425.
NTP, 2004. National Toxicology Program. NTP technical report on the toxicity
studies on malachite green chloride and leucomalachite green (CAS Nos. 56964-2 and 129-73-7) administered in feed to F344/N rats and B6C3F1 mice. NIH
Publication No. 04-4416. Available at: <http://ntp-server.niehs.nih.gov/ntp/
htdocs/ST_rpts/tox071.pdf>.
NTP, 2005. National Toxicology Program. NTP Technical report on the toxicology
and carcinogenesis studies of malachite green chloride and leucomalachite
green (CAS Nos. 569-64-2 and 129-73-7) in F344/N rats and B6C3F1 mice (feed
studies). NTP TR 527. NIH Publication No. 05-4463. Available at: <http://
ntp.niehs.nih.gov/files/527_FINAL.pdf or at http://ntp.niehs.nih.gov/?objectid=
070B74C5-C780-E272-6721383F7A1113BC>.
Tittlemier, S.A., Van de Riet, J., Burns, G., Potter, R., Murphy, C., Rourke, W., Pearce,
H., Dufresne, G., 2007. Analysis of veterinary drug residues in fish and shrimp
composites collected during the Canadian Total Diet Study, 1993–2004. Food
Addit. Contam. 24, 14–20.
Veterinary Residues Committee. Annual Reports on Surveillance for Veterinary
Residues in Food in the UK for 2001, 2002, 2003, 2004 and 2005. Available at:
<http://www.vet-residues-committee.gov.uk/>.
Wheeler, M.W., Bailer, A.J., 2007. Properties of model-averaged BMDLs: a study of
model averaging in dichotomous response risk estimation. Risk Anal. 27, 659–
670.
WHO, 2007. World Health Organization. GEMS/food consumption cluster diets.
Regional per capita consumption of raw and semi-processed agricultural
commodities. Prepared for the Global Environment Monitoring System/Food
Contamination Monitoring and Assessment Programme (GEMS/Food). WHO
Library, p. 40.