Malachite green in food

1.3.1.1 International and European agencies

The EFSA Scientific Panel on Food Additives, Flavourings, Processing Aids and Materials in Contact with Food (AFC Panel) reviewed the toxicological information of a large number of dyes illegally present in food in the EU (EFSA, 2005b). MG and LMG were included in the group of ‘dyes that have been used illegally in countries outside the EU from which spices originate and dyes that have been used in the past as food colours in other countries but withdrawn from food use following discovery of toxicity’. The AFC Panel concluded that both compounds should be viewed as genotoxic and/or carcinogenic.

At its 70th meeting, the Joint FAO/WHO Expert Committee on Food Additives (JECFA) evaluated MG and LMG (FAO/WHO, 2009). In short-term (28 days) toxicity studies, MG caused haematological changes and liver toxicity in male and female rats with a no-observed-adverse-effect-level (NOAEL) of 10 mg/kg body weight (bw) per day and LMG induced liver effects in male rats at all doses tested (NTP, 2005). The increased incidence of fetal anomalies in all dose groups observed in a teratogenicity study in rabbits (Meyer and Jorgenson, 1983), although without a consistent dose–response relationship, raised concern regarding the potential developmental toxicity of MG. JECFA concluded, however, that this study was inadequately conducted and reported, and that additional studies are needed to evaluate the reproductive and developmental toxicity of MG.

JECFA concluded that MG has no genotoxic potential in conventional in vitro and in vivo assays. LMG was negative in in vitro assays, but induced cII mutations in the liver of female Big Blue^® B6C3F1 transgenic mice, whereas MG did not. MG was not carcinogenic in a long-term study in female mice, but in female F344 rats a trend in the incidence of thyroid gland follicular cell adenomas or carcinomas was found. In a long-term study in female mice, LMG caused a dose-related increase in the combined incidence of hepatocellular adenomas and carcinomas. Because of the induction of cII mutations by LMG in liver cells of female transgenic mice, JECFA concluded that a genotoxic mechanism cannot be ruled out.

JECFA considered it inappropriate to establish an acceptable daily intake (ADI) for MG and used a margin of exposure (MOE) approach. Due to deficiencies in the database, it was considered inappropriate to derive an MOE for non-carcinogenic endpoints. It concluded that the induction of hepatocellular adenomas or carcinomas in female mice treated with LMG was the pivotal effect for the risk assessment, and derived a lower 95% confidence limit or a benchmark response of 10% extra risk (BMDL₁₀) of 20 mg LMG/kg bw per day as the reference point for the MOE calculation. Because there was no information on the conversion rate of MG to LMG in food, JECFA considered it prudent to evaluate the sum of the residues of MG and LMG in food, expressed as LMG. Using average and high (97.5th percentile (P97.5)) exposures (see Section 3.2.1), MOEs of about 1.3 × 10⁵ and 3 × 10⁴, respectively, were calculated, which were considered to be of low concern for human health.

The European Chemicals Agency (ECHA) evaluated the same data on MG and LMG as used by JECFA (ECHA, 2010a,b). ECHA concluded that MG may cause developmental toxicity in rabbits, but noted that the poor quality of the study cast some doubts on the reliability of the findings. Based on the induction of mutations in the liver of transgenic mice by LMG and the induction of DNA adducts in the liver of rats and mice by MG and LMG, ECHA considered it prudent to presume that MG and LMG are potential in vivo somatic cell mutagens and, based on classification, labelling and packaging (CLP) criteria, a classification of ‘Muta.2, Suspected to cause genetic defects’ was proposed for both compounds.

In female rats, there were no significant increases in tumour incidence caused by MG, and no tumour findings in female B6C3F1 mice. ECHA considered the evidence as not sufficient to warrant a classification of MG for carcinogenicity. ECHA concluded that there was limited evidence for carcinogenicity of LMG based on tumour induction in the liver of mice and equivocal evidence of induction of tumours in the liver of female rats. Although it was recognised that the evidence for carcinogenicity is only weak, the possible involvement of a genotoxic action raised concerns for the carcinogenicity of LMG. Therefore, based on CLP criteria, a classification of ‘Carc.2, Suspected of causing cancer’ was proposed.

1.3.1.2 National agencies

The UK Committee on Mutagenicity of Chemicals in Food, Consumer Products and the Environment (UK COM, 2004) and the UK Committee on the Carcinogenicity of Chemicals in Food, Consumer Products and the Environment (UK COC, 2004) evaluated MG and LMG. Based on the available conventional mutagenicity data, there was some evidence for a clastogenic potential of MG, but it was not possible to make an adequate assessment for LMG. However, due to the formation of DNA adducts in the liver from both rats and mice by MG, and the induction of mutations in liver DNA of female transgenic B6C3F1 mice by LMG, it was concluded that both compounds should be regarded as in vivo mutagens (UK COM, 2004). For MG, there was no convincing evidence for a carcinogenic effect in female F344 rats and female B6C3F1 mice, but there was equivocal evidence of carcinogenic activity of LMG in female F344 rats based on an increased incidence of hepatocellular adenomas or carcinomas (combined). Taking into account the views of the UK COM (2004), it was therefore considered prudent to regard LMG as a genotoxic carcinogen (UK COC, 2004).

The National Food Institute (NFI) of Denmark and the Technical University of Denmark (DTU) provided a risk assessment of MG in 2007 (NFI and DTU, 2007). It was reported that in the Danish surveillance programme up to 28 μg/kg of LMG was found in fish produced in Denmark. With an intake of 100 g fish/day, this led to a conservatively estimated exposure of 2.8 μg LMG/day (0.048 μg/kg bw per day for a 60-kg adult). Compared with a BMDL₁₀ of 20 mg/kg bw per day for hepatocellular adenomas and carcinomas in female mice, this led to an MOE of more than 4 × 10⁵, which would be a low concern for public health, and considered to be a low priority for risk management actions.

A health risk assessment of MG and LMG was reported by the Food Safety Commission (FSC) of Japan (FSC, 2005). Concerning genotoxicity it was concluded that the available in vitro and in vivo results ‘failed to provide a univocal explanation for various in vivo mutations including DNA adduct formation and cII mutation. Nevertheless, the results obtained so far could not deny the genotoxic potentials of MG and LMG. Further studies are required to reach a reliable conclusion’. According to the FSC, the results of the 2-year rodent studies suggested that MG acts as a weak carcinogen in the liver and mammary glands of female rats, and that LMG acts as a liver carcinogen in female mice and as a weak carcinogen in liver and thyroid of rats.

Food Standards Australia New Zealand (FSANZ) investigated the risk of exposure to MG in fish (FSANZ, 2007). Levels of MG in climbing perch of 7.8 μg/kg were found. Based on a mean fish consumption of 100 g/day, this resulted in a dietary exposure of about 0.011 μg/kg bw per day for a 70-kg adult. Based on the data on carcinogenicity and genotoxicity, it was suggested that ‘MG is a very low risk’, and the lowest-observed-effect level (LOEL) of 5 mg/kg bw for non-neoplastic lesions in the rat liver was considered as the most sensitive endpoint. The estimated dietary exposure was about 450,000 times below the LOEL, and it was therefore concluded that the health risk from MG residues in fish is extremely small.

In 2014, the Committee for Pharmacologically Active Substances and Veterinary Medicinal Products of the German Federal Institute for Risk Assessment (BfR) published a toxicological evaluation on MG (BfR, 2014). The Committee evaluated occurrence data on MG from official controls, analysis and toxicity of MG, the applicability of the threshold of toxicological concern (TTC) concept, an assessment based on the MOE and the appraisal by ECHA on MG. Based on a BMDL₁₀ value of 20 mg LMG/kg bw for hepatocellular adenomas and carcinomas in female mice and an estimated human daily exposure of 5–50 ng LMG/kg bw, the Committee calculated MOEs of 4 × 10⁶–4 × 10⁵ and concluded that a daily exposure of 50 ng LMG/kg bw is toxicologically tolerable. In addition, the Committee evaluated the applicability of the TTC concept and concluded that MG and LMG fulfil the TTC criteria, which lead to a TTC value of 0.15 μg per day for MG/LMG. In summary, the Committee stated that MG should not be administered in food-producing animals and recommended to apply the TTC value of 0.15 μg per day for the health assessment of MG/LMG residues in food as this represents the more conservative approach compared to other toxicological assessments, in particular, when taking sensitive population groups into account.

The CONTAM Panel noted the risk assessment by l'Agence française de sécurité sanitaire des aliments (AFSSA; currently ANSES (Agence nationale de sécurité sanitaire de l'alimentation, de l'environnement et du travail)) from 2002 (AFSSA, 2002). However, since the bulk of information regarding carcinogenicity of MG/LMG became available after 2002, no further information regarding this risk assessment is provided in this Scientific Opinion.

1.3.1.3 Other assessments

Rauscher-Gabernig et al. (2007) assessed the risk to the Austrian population from exposure to MG and LMG. Fish samples, mainly fresh fish produced in Austria, collected in the period 2003–2005, contained residues almost entirely of MG. The estimated intake for high fish consumers (95th percentile, P95), expressed as LMG, was 0.83 μg/kg bw per day for children (3–6 years of age), 0.42 μg/kg bw per day for adult women and 0.51 μg/kg bw per day for adult men. Applying a BMDL₁₀ of 21 mg/kg bw per day for liver tumours in female mice and thyroid tumours in male rats, these intakes resulted in MOEs of 2.5 × 10⁴, 5 × 10⁴ and 4.1 × 10⁴ for these three age groups, respectively. Therefore, the authors classified the health risk to the Austrian population from exposure to MG and LMG as low.

Based on the maximum concentration of LMG found in wild eel caught in Germany (see Section 3.1.1) and chronic fish consumption data, intakes of 0.27 and 3.8 ng/kg bw per day for children and adults, respectively, were reported by Schuetze et al. (2008). Using a LOEL of 13 mg LMG/kg bw per day for the increased incidence of neoplasms (NTP, 2005), the authors reported MOEs of 4.9 × 10⁷ and 3.4 × 10⁶ for children and adults, respectively.

The application of the MOE approach to substances in food that are genotoxic and carcinogenic was studied by a group of international scientists (Benford et al., 2010). LMG was one of the 12 compounds addressed. Available literature data suggested that only about 10–15% of fish contained residues of LMG, and data generally did not distinguish between farmed and wild-caught fish. Average and high (P95) dietary intakes were estimated to be 5 and 50 ng/kg bw per day, respectively. The authors considered these intake estimates to be highly conservative because it had been assumed that all fish consumed was contaminated at the mean LMG concentration reported in surveillance studies. Based on a BMDL₁₀ of 20.4 mg/kg bw per day for liver tumours (adenomas and carcinomas combined), the MOE for average and high consumers were 4 × 10⁶ and 4 × 10⁵, respectively. Therefore, it was concluded that LMG is likely to be a relatively low priority for risk management actions, unless specific incidents result in higher levels of exposure. Identical MOEs for LMG were reported by Renwick et al. (2010).

The risk of human exposure to LMG due to fish consumption in Taiwan was assessed by Chu et al. (2013). They conducted a probabilistic risk assessment for three population groups, adolescents (13–18 years), adults (19–64 years) and elderly (> 64 years), with elderly consumers having the highest intake of LMG. Based on a BMDL₁₀ of 18.5 mg/kg bw per day for hepatocellular adenomas and carcinomas in female mice and probabilistic estimates of the lifetime average daily dose for mean and high (P95) adult consumers, MOEs for these consumers were 4.8 × 10⁶ and 4.1 × 10⁵, respectively. Using a cancer slope factor of 0.035 mg/kg bw per day, cancer risk estimates for this age group were 4.7 × 10⁻⁷ at a mean and 1.6 × 10⁻⁶ at a 95th percentile lifetime average daily dose. The authors concluded that daily intake of LMG from fish consumption in Taiwan was of low priority for risk assessment, except for those with 95th percentile high intake. The CONTAM Panel identified some inconsistencies in the way the results were reported, and noted that had the reported point estimates for exposure of average (13.5 ng/kg bw per day) and high adult consumers (45 ng/kg bw per day) been used, MOEs of 1.4 × 10⁶ and 4.1 × 10⁵, respectively, would have resulted. However, these results do not affect the conclusions of Chu et al. (2013).

1.3.3.1 Malachite green

MG1 ([4-[[4-(dimethylamino)phenyl]-phenylmethylidene]cyclohexa-2,5-dien-1-ylidene]-dimethylazanium;chloride; Chemical Abstract Service (CAS) No 569-64-2) is a cationic dye and consists of green crystals with metallic lustre. It has the molecular formula C₂₃H₂₅N₂Cl and a molecular weight of 364.91 g/mol. The oxalate salt (CAS No 2437-29-8, molecular weight 463.50 g/mol) is also marketed. The green colour of the compound is independent of whether it occurs with the chloride or oxalate anion.

The melting point is 158–160°C. When heated to decomposition, it emits toxic fumes of nitrogen oxide and hydrogen chloride. MG occurs almost entirely in the ionised form at pH values of 5–9. It has a good solubility in water and alcohols. MG is easily converted into LMG (see Section 3.3.1).

1.3.3.2 Leucomalachite green

LMG1 (4-[[4-(dimethylamino)phenyl]-phenylmethyl]-N,N-dimethylaniline; CAS No 129-73-7) consists of an off-white to light brown powder. From alcohol and benzene, it forms needles or leaflets. It has the molecular formula C₂₃H₂₆N₂ and a molecular weight of 330.46 g/mol (Figure 1).

The melting point is 102°C. In contrast to MG, the estimated solubility of LMG in water, being 6.4 × 10⁻² mg/L at 25°C, is very low (EPI (estimation program interface) Suite, estimated). LMG is very soluble in organic solvents, such as ethanol, ethyl ether, ethylene glycol monomethyl ether and benzene.

1.3.4.1 Sampling and storage

Most of the sampling of food, and of related materials, for MG/LMG testing in foods of animal origin is undertaken in the context of the national residue monitoring plans as specified in Council Directive 96/23/EC,2 with residue testing undertaken in accordance with Commission Decision 2002/657/EC3. For details of the protocols and procedures specified for such sampling and testing, see Section 2.1.1 of this opinion.

Commission Decision 2002/657/EC states that samples shall be obtained, handled and processed in such a way that there is a maximum chance of detecting the substance. Sample handling procedures shall prevent the possibility of accidental contamination or loss of analytes. To achieve this goal, samples are stored in suitable, secure, clearly identified containers and in conditions, such as frozen storage (fish and crustaceans, water) or at refrigerated/ambient temperatures (fish feed), prior to analysis.

1.3.4.2 Specific issues relating to (leuco)malachite green analysis

Initially, testing for residues of MG in animal tissues was by methods directed at the parent compound, using high-performance liquid chromatography (HPLC) with ultraviolet/visible (UV/Vis) detection. However, studies on MG showed that residues of the parent compound are less persistent than those of the reduced metabolite LMG. Therefore, subsequent HPLC methods measured both substances using a combination of UV/Vis detection (618 nm) for MG and fluorescence detection (λ_ex 265 nm/λ_em 360 nm) for LMG, or measurement of both substances as MG following oxidation of LMG to MG. MG and LMG are the targets for residue analysis in food, while testing for the parent compound alone is limited to samples of fish feed and water.

Because MG and LMG are sensitive to air and light, particular steps need to be taken to protect both analytical standards and sample residues from deterioration. Addition of ascorbic acid to standard solutions of LMG, to prevent photo-oxidative demethylation, and of acetic acid to standard solutions of MG, to prevent transformation of MG to its carbinol form, has been applied (Hall et al., 2008; Ahn et al., 2010) and standard solutions are normally stored in amber bottles (Mitrowska et al., 2008a; Hurtaud-Pessel et al., 2011). In the case of sample treatment, protection from light is recommended and steps, such as addition of hydroxylamine hydrochloride as a reducing agent/antioxidant (Hurtaud-Pessel et al., 2011) or use of acidified acetonitrile to reduce conversion of MG to its carbinol form (Hall et al., 2008) have been applied.

1.3.4.3 Extraction and sample clean-up

Extraction of MG/LMG from fish and crustacean samples is most often carried out using mixtures of low pH buffer (McIlvane pH 3 or ammonium acetate buffers) and organic solvent (acetonitrile), but also using direct extraction with organic solvents such as acetonitrile. Typically, the solvent extract is subjected to liquid/liquid partitioning with dichloromethane or defatting by washing with hexane, and further clean-up is achieved by the addition of adsorbents, such as alumina, basic alumina or primary/secondary amines, to the extract or by using solid-phase extraction (SPE). For SPE, a range of sorbents have been applied including reversed phase (such as C18 and polymeric sorbents), normal phase (such as alumina sorbents) and cation exchange (such as propylsulfonic acid sorbents). Depending on whether the method is a screening or confirmatory method, less or more sample extract purification steps may be required.

Other approaches have been applied to extraction/clean-up of MG/LMG from fish and crustacean samples, such as the Quick Easy Cheap Effective Rugged Safe (QuEChERS) technique, molecularly imprinted polymers (MIPs) and immunoaffinity chromatography (IAC). The QuEChERS, or dispersive SPE, technique involves use of a combination of salts and SPE sorbents to achieve a combined extraction and clean-up of MG/LMG from the sample matrix. IAC involves use of MG specific antibodies immobilised on a support material, following oxidation of residues of LMG to MG with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ), to isolate MG/LMG from sample extracts (Xie et al., 2013). MIPs involve use of an imprinted polymer with selectivity for both MG and LMG (Long et al., 2009) or MG specific imprinted polymers with pre-MIP oxidation of LMG to MG using DDQ (Martínez Bueno et al., 2010; Guo et al., 2011); the MIP is packed into a column to trap the analyte(s) from the sample extract and allow for washing steps prior to elution of the analyte(s) from the column. In a variation on the MIPs approach, Huang et al. (2015) describe use of a magnetic MIP to selectively enrich MG/LMG from fish sample extracts.

In the case of water samples, some screening methods based on determination by enzyme-linked immunosorbent assay (ELISA) (Yang et al., 2007; Xing et al., 2009) or conductive carbon black paste electrode (Qu et al., 2012) use simple dilution of the filtered sample. Other methods use solvent extraction (Maleki et al., 2012), microwave-assisted ionic liquid extraction (Gao et al., 2013), hollow fibre liquid-phase microextraction (Zou et al., 2014), cloud point extraction (Pourreza and Elhami, 2007; An et al., 2010), direct SPE (Safarik and Safarikova, 2002; Mitrowska et al., 2008a) or MIP-based SPE to isolate MG alone (Lian and Wang, 2012) or MG and LMG (Li et al., 2008) from water samples.

Extraction of MG from fish feed samples was undertaken by solvent extraction with acetonitrile and clean-up of the extract by MIP-based SPE (Li et al., 2008).

1.3.4.4 Screening methods

Screening methods should measure MG/LMG with sufficient sensitivity to satisfy regulatory requirements, currently at the minimum required performance limit (MRPL) of 2.0 μg/kg for the sum of MG and LMG in meat of aquaculture products (Annex II of Commission Decision 2002/657/EC). Screening methods for MG/LMG include immunoassays (ELISAs, RNA aptamer assay, biosensors), electrochemiluminescence and HPLC.

A number of ELISAs have been developed for determination of LMG (Singh et al., 2011) and for both MG and LMG (Yang et al., 2007; Xing et al., 2009; Xu et al., 2013) with limits of detection (LOD) between 0.05 and 0.5 μg/kg in fish. Zhang et al. (2015) describe a chemiluminescent enzyme immunoassay for MG with a limit of quantification (LOQ) of 0.1 μg/kg in fish and shrimp. Commercial ELISA kits are available for determination of both MG and LMG with reported LOD values of 0.1–0.25 μg/kg in fish and crustaceans.

As an alternative to antibodies used as the binding agents in conventional ELISAs, use of an RNA-aptamer has been described by Stead et al. (2010); this method was capable of determining both MG and LMG (following oxidation to MG) in salmon tissue at levels considerably lower than the MRPL of 2 μg/kg.

Biosensor-based assays have been described, such as an electrochemical method for determination of MG using a multiwall carbon nanotube-modified glassy carbon electrode (Yi et al., 2008) and an electrochemical enzyme inhibition sensor on a screen-printed carbon working electrode for determination of MG and LMG in fish, with LODs of 0.25 μg/kg (Faridah et al., 2013; Hidayah et al., 2013).

Methods based on determination by electrochemiluminescence have been described: one using molecularly imprinted SPE for isolation of MG with an LOQ of 0.02 μg/kg (Guo et al., 2011) and another using magnetic MIPs to isolate MG and LMG with an LOQ of 0.036 μg/kg (Huang et al., 2015).

HPLC methods for determination of MG and LMG in fish and crustacean samples may involve direct measurement of both substances using UV/Vis (618–620 nm) and fluorescence (λ_ex 265 nm/λ_em 360 nm) detectors, respectively; Mitrowska et al. (2005) describe a method for carp with decision limits (CCα4) of 0.15 and 0.13 μg/kg for MG and LMG, respectively, and detection capability (CCβ5) values of 0.37 and 0.32 μg/kg for MG and LMG, respectively, while Chen and Miao (2010) describe a method for catfish with LOD values of 0.38 and 0.10 μg/kg for MG and LMG, respectively. Where oxidation of LMG to MG is performed, such oxidation may be undertaken pre-HPLC with DDQ or post-column with lead oxide (PbO₂) or with iodine. A number of methods using pre-HPLC derivatisation with DDQ have been described for salmon, shrimp and catfish with LOD values of 0.15–1.0 μg/kg (Andersen et al., 2005, 2006, 2009), for fish with an LOD value of 0.15 μg/kg (Xie et al., 2013) and for trout with CCα/CCβ values of 0.16/0.39 μg/kg (Fallah and Barani, 2014). Methods involving post-column derivatisation with PbO₂ have been described for catfish (Roybal et al., 1995) and for catfish and trout (Rushing and Thompson, 1997) with LOQ values of 0.5–1.5 μg/kg, for fish and prawns with an LOD value of 1.0 μg/kg (Bergwerff and Scherpenisse, 2003; Stoev and Stoyanov, 2007), and for fish, shrimp and shellfish with CCα/CCβ values of 0.14/0.24 μg/kg (Long et al., 2009). Long et al. (2008) describe an alternative post-column oxidation procedure for LMG using an iodine solution, giving CCα/CCβ values of 0.17/0.23 μg/kg.

Most HPLC methods for water samples test only for MG with UV/Vis detection (600–618 nm) and reported LOD values between 0.01 and 0.1 μg/L (Li et al., 2008; Lian and Wang, 2012; Maleki et al., 2012; Gao et al., 2013; Zou et al., 2014). A method for determination of MG (CCα/CCβ values of 0.03/0.08 μg/L) and LMG (CCα/CCβ values of 0.03/0.07 μg/L) in water samples using UV/Vis and fluorescence detectors was described by Mitrowska et al. (2008a).

A number of methods have been described for the determination of MG, used as a dye, in meat (Sun et al., 2013), various coloured foods (Dixit et al., 2011) and medicinal herbs (Li et al., 2015). These methods involved microwave-assisted extraction with SPE clean-up and HPLC with UV/Vis detection, solvent extraction with SPE clean-up and HPLC with UV/Vis detection, and use of a silver nanoparticle wiper with surface-enhanced Raman scattering detection, respectively.

1.3.4.5 Confirmatory methods

Liquid chromatography–mass spectrometry (LC–MS) is the method of choice for confirmatory analysis of MG/LMG in fish and crustaceans. Of the various mass spectrometric methods available, liquid chromatography linked with a triple quadrupole mass detector is most commonly used, although ion trap (Turnipseed et al., 2005; Wu et al., 2007; Andersen et al., 2009; Martínez Bueno et al., 2010) and time-of-flight (Villar-Pulido et al., 2011) detectors have also been used. Some of the published methods for MG/LMG are multiresidue methods, also including other triphenylmethane dyes, such as (leuco)crystal violet and/or (leuco)brilliant green (Dowling et al., 2007; Wu et al., 2007; Tarbin et al., 2008; Andersen et al., 2009; Chen and Miao, 2010; Hurtaud-Pessel et al., 2011, 2013; Tao et al., 2011). A method using time-of-flight-mass spectrometry (TOF-MS) has been developed for multiclass determination of residues of selected pharmacologically active substances, including MG/LMG, in shrimps (Villar-Pulido et al., 2011).

Typical MS conditions used for the confirmatory analysis of MG/LMG are positive electrospray ionisation (ESI) with two precursor to product ion transitions being monitored for each analyte, m/z 329 > 313, 208 for MG and m/z 331 > 239, 316 for LMG. Sample treatment, prior to liquid chromatography–tandem mass spectrometry (LC–MS/MS) determination of the analytes, typically involves buffer/organic solvent extraction coupled with clean-up by SPE, but extraction by automated solvent extraction (Tao et al., 2011) or by QuEChERS (Villar-Pulido et al., 2011; Hashimoto et al., 2012; Lopez-Gutierrez et al., 2013) and clean-up using MIPs (Martínez Bueno et al., 2010) have also been reported.

LC–ESI–MS/MS methods for determination of MG/LMG have been applied to a variety of fish, including catfish, trout, salmon and tilapia, and to shrimps. The range of values for CCα and CCβ by these methods, for MG and for LMG, are < 0.1–0.9 and 0.1–1.2 μg/kg, respectively (van de Riet et al., 2005; Dowling et al., 2007; Halme et al., 2007; Arroyo et al., 2009; Chen and Miao, 2010; Hurtaud-Pessel et al., 2011, 2013; Tao et al., 2011; Ascari et al., 2012; Hashimoto et al., 2012; Xu et al., 2012; Lopez-Gutierrez et al., 2013; Andersen et al., 2015); all with method sensitivity sufficient to satisfy the MRPL value of 2 μg/kg. A number of papers have been published on the determination of total MG in fish, following oxidation of any LMG present in the sample to MG, post-column using PbO₂ with reported CCα/CCβ values of 0.11/0.15 μg/kg (Scherpenisse and Bergwerff, 2005) and pre-LC using DDQ with reported CCα/CCβ values of 1.2/2.0 μg/kg (Tarbin et al., 2008).

LC–MS methods using an ion-trap detector have been developed for analysis of total MG in salmon and catfish, following oxidation of any LMG present in the sample to MG using DDQ, with reported LOD/LOQ values of < 0.15/0.15 μg/kg (Turnipseed et al., 2005), 0.24/0.75 μg/kg (Andersen et al., 2009) and 0.003/0.005 μg/kg (Martínez Bueno et al., 2010). Isotope dilution liquid chromatography/mass spectrometry (ID-LC/MS), using deuterated and/or ¹³C₆-MG and -LMG, has been developed for fish, shrimp and shellfish (Doerge et al., 1998; Wu et al., 2007; Ahn et al., 2010). LOD values of 0.02 μg/kg for MG and of 0.5 μg/kg for LMG are reported (Doerge et al., 1998; Ahn et al., 2010) and CCα/CCβ values of 0.08/0.13 μg/kg for MG and of 0.05/0.09 μg/kg for LMG are reported (Wu et al., 2007). An accurate mass full-scan TOF-MS method has been applied to the analysis of shrimp, with reported LOD values of 0.06 and 0.60 μg/kg for MG and LMG, respectively (Villar-Pulido et al., 2011).

An LC–ESI–MS/MS method has been developed for MG/LMG in water samples, following centrifugation and preconcentration on diol SPE columns, with reported CCα/CCβ values of 0.04/0.06 μg/L for MG and of 0.03/0.05 μg/L for LMG (Mitrowska et al., 2008a).

1.3.4.6 Analytical quality assurance: performance criteria, reference materials and proficiency testing

The performance criteria for methods used to test for MG/LMG are those laid down in Commission Decision 2002/657/EC for screening and confirmatory methods. Methods must have satisfactory performance for the characteristics of specificity, trueness, ruggedness, and stability of the analyte in standard solutions and in test matrices. The methods must be validated for recovery, repeatability, within-laboratory reproducibility, calibration curves, CCα and CCβ according to procedures specified in the Decision, or equivalent procedures.

MG oxalate (94.3 ± 1.4% purity) and LMG (98.8 ± 0.8% purity) reference standards have been prepared and are commercially available (Le Goff and Wood, 2008). Isotopically labelled materials, such as d₅-MG, d₅- or d₆-LMG and ¹³C₆-LMG, are available commercially for use as internal standards. No certified reference materials for MG/LMG are commercially available to date. A sample of salmon to be used in an international intercomparison study was produced and the sum of MG and LMG in this sample was determined to be 9.32 ± 0.98 μg/kg (at the 95% confidence interval) using ID-LC/MS (Hall et al., 2008).

Several proficiency tests and interlaboratory studies have been reported for MG/LMG in fish. The EURL, Anses-Fougères (France), provides proficiency testing for National Reference Laboratories in charge of control for dye residues in aquaculture products with three rounds having been organised covering trout (2005, 2009) and prawn (2013) samples (Verdon et al., 2015). In January 2008, the Proficiency Test Advisory Board, Government Laboratory, Hong Kong reported on a proficiency test (APLAC T058) aimed at evaluating the testing capability of 48 participating laboratories in the quantitative analysis of MG and LMG in swamp eels; 48% and 62% of participants had satisfactory results (z-score ≤ 2) for the determination of MG (assigned value 28.2 μg/kg) and LMG (assigned value 2.21 μg/kg), respectively (Wong, 2008). In the UK, the Food Analysis Performance Assessment Scheme (FAPAS) provides samples of fish muscle containing MG and LMG for testing.6

1.3.4.7 Concluding comments

While a combination of buffer/solvent extraction, liquid/liquid partitioning and SPE is the most commonly used sample preparation for the analysis of MG/LMG using both screening and confirmatory methods, other approaches, such as the QuEChERS technique, MIPs and IAC have been applied. Screening methods for MG/LMG include HPLC with UV/Vis and fluorescence detectors, ELISA and biosensor methods and electrochemiluminescence, all providing sufficient analytical sensitivity to meet the MRPL of 2 μg/kg. Confirmatory methods are based on various LC–MS methods and these typically provide CCα and CCβ values of < 1.0 μg/kg for a range of fish and crustaceans, again adequately meeting the MRPL of 2 μg/kg.

2.1.1.1 National residue monitoring plans

Council Directive 96/23/EC on measures to monitor certain substances and residues thereof in live animals and animal products requires that Member States draft a national residue monitoring plan for the groups of substances detailed in Annex I of this Directive. These plans must comply with the sampling rules in Annex IV of the Directive.

Within Group B, Veterinary drugs and contaminants, of Annex I, Subgroup B3e Dyes, covering MG/LMG, samples are to be taken only from aquaculture animals, their feeding stuffs, water and primary animal products. The minimum number of samples to be collected each year is specified as a proportion of the aquaculture production volume of the previous year. The compounds sought and the samples selected for analysis should be selected according to the likely use of these substances. Sampling under the national residue monitoring plan should be targeted; samples should be taken preferably at the farm, on fish ready to be placed on the market, and either at the processing plant or at wholesale level. Samples taken at farm level should be taken from a minimum of 10% of registered sites of production.

Member States submit data on the occurrence of non-compliant results determined in the residue monitoring, including for MG/LMG to the EC database on residues of veterinary medicines. Data on the occurrence of MG/LMG in food have been extracted from the EC database on residues of veterinary medicines. This database contains the annual sampling plan and the results from 2004 onwards provided by all Member States. The results are reported as aggregate data. However, there is no indication of the sample matrix tested and no concentration for the chemical residue or contaminant detected in the sample is provided. In addition, the number of samples analysed for the individual substances are reported by the Member States only if there is at least one non-compliant sample for the substance in question. Where all samples are compliant, the number of samples analysed is not reported. Furthermore, where controls are carried out at farm and processing plant, the total number of samples recorded may refer to samples taken at either farm or processing plant, depending on where the non-compliant samples were found, and this may be on a substance group basis rather than on the individual substance basis. Where non-compliant samples were found at both farm and processing plant, the number of samples represents the sum of samples taken at both sampling points.

For the years 2002 and 2003, data on MG/LMG reported by Member States have been extracted from the Commission Staff Working Papers on the implementation of national residue monitoring plans in the Member States (EC, 2004, 2005). The data presented in these papers are not at the same level of detail as found on the database for the years 2004–2014. For example, for 2003 the number of samples analysed represents the total number of targeted samples tested for all categories of substances, rather than the number of samples specifically tested for Subgroup B3e Dyes.

Data on the occurrence of MG/LMG in food have been considered for the period 2002–2014.13

In addition to the data submitted to the EC database on residues of veterinary medicines from Member States, data from Norway on the occurrence of MG/LMG for the years 2006–2014 were also considered. The latter were extracted from the annual reports ‘Monitoring program for residues of therapeutic agents, illegal substances, pollutants and other undesirables in farmed fish (in accordance with Council directive 96/23/EC)’ (NIFES, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015).

2.1.1.2 Rapid Alert System for Food and Feed

The CONTAM Panel considered the Rapid Alert System for Food and Feed (RASFF)14 database as another source of information on the occurrence of MG/LMG in food.

RASFF notifications mostly concern controls at the outer European Economic Area (EEA) borders, at points of entry or border inspection posts, when a consignment is not accepted for import into the EU. The second largest category of notifications concerns official controls on the internal market. A small number of notifications may be triggered by official controls in non-member countries.

After an inspection is conducted within a country and unfavourable results of the analysis are obtained, the risk needs to be evaluated, as does the probability that the product may be present on the market of other Member States. In the latter case, notifications are provided when non-compliant samples for a contaminant are found, providing also quantified values. However, information on the total number of samples analysed, the number of compliant samples, the concentrations and the type of analysis undertaken is rarely provided.

Searches in the RASFF database were performed for the hazard category ‘residues of veterinary medicinal products’ – hazard ‘malachite green’ or ‘leucomalachite green’ – that had been notified between 1 January 2002 and 31 December 2014.

2.2.2.1 Strategy for literature search

A comprehensive search for literature was conducted for peer-reviewed original research pertaining to adverse health effects on (experimental) animals and humans. The search strategy was designed to identify scientific literature dealing with toxicity, mode of action, toxicokinetics and human data on MG and LMG. An overview of the search terms is given in Appendix A, Section A.1.

It should be noted that a narrative approach was used for those sections dealing with methods of analysis, chemistry, occurrence and exposure as the identified papers are used only to give background information to the reader. For methods of analysis and chemistry, recent reviews, in combination with a limited literature search, were used to identify the scientific literature (see Appendix A, Section A.1.). Occurrence data were only considered when the samples had been collected since 2002, because the Member States became aware of the possible misuse of MG in aquaculture in 2002 and so control of MG was intensified by increasing the number of samples taken (EC, 2004).

The literature search was not restricted to publications in English. A first literature search was performed in April 2015 for all topics (October 2015 for papers on the occurrence in foods due to illegal use of MG as a food colouring agent). The literature search has been updated in December 2015 to identify papers dealing with toxicity, mode of action, toxicokinetics and human data on MG and LMG. Web of Science16 and PubMed17 were identified as databases appropriate for retrieving literature for the present evaluation. The references resulting from the literature search were imported and saved using a software package (EndNote18), which allows effective management of references and citations. Additionally, reviews, relevant scientific evaluations and toxicity studies by national or international bodies were considered for the current risk assessment, i.e. AFSSA (2002), NTP (2004, 2005), UK COM (2004), UK COC (2004), FSC (2005), (NFI and DTU, 2007), FSANZ (2007), FAO/WHO (2009), ECHA (2010a,b), and the BfR (2014). In addition, when relevant papers were identified during the risk assessment process (e.g. from other studies or reviews), they were also considered.

The references obtained were screened using title and abstract to identify relevant literature. The exclusion criteria used are shown in Appendix A, Section A.2.

2.2.2.2 Appraisal of studies

The information retrieved has been reviewed by the CONTAM Standing Working Group (SWG) on non-allowed pharmacologically active substances in food and feed and their RPAs, and has been used for the present assessment based on expert judgement. Any limitations in the information used are documented in this scientific opinion.

Selection of the scientific papers for inclusion or exclusion was based on consideration of the extent to which the study was relevant to the assessment and on general study quality considerations (e.g. sufficient details on the methodology, performance and outcome of the study, on dosing and route of administration and on statistical description of the results (EFSA, 2009b)), irrespective of whether they yielded positive, negative or null results.

Studies solely focusing on the efficacy of MG as a pharmacologically active substance were excluded from the assessment.

3.1.1.1 Fish and crustaceans

Bergwerff and Scherpenisse (2003) analysed trout (n = 18), eel (n = 10) and salmon products (n = 20) collected in the Netherlands for the presence of MG/LMG residues using HPLC-UV (LOD: 1 μg/kg) or LC–MS/MS (LOD: 0.2 μg/kg). The year of sampling was not indicated. LMG was detected in 13 trout samples (range: 1.3–14.9 μg/kg), five eel samples (range: 1.5–9.7 μg/kg), three samples of fresh salmon (range: 0.2–2.9 μg/kg), two smoked salmon samples (0.2 μg/kg in both samples) and in none of the canned salmon samples.

Schuetze et al. (2008) analysed wild eel (n = 45; year of sampling not indicated) caught from lakes, a river and a canal in Berlin, Germany. MG residues (reported as the sum of MG and LMG) were detected in 25 samples at concentrations ranging from 0.053 to 0.765 μg/kg. The authors stated that the occurrence could be linked directly to the presence of treated sewage discharge in the surface waters from which the fish was caught. Samples were analysed using LC–MS/MS (LOD/LOQ: 0.02/0.04 μg/kg for MG and 0.01/0.02 μg/kg for LMG).

Samples of carp (n = 42) and rainbow trout (n = 30), collected between 2009 and 2011 from fish farms in Croatia, were analysed for MG residues using an ELISA specific for MG (LOD/CCβ: 0.31/0.68 μg/kg). MG was detected in 13 samples but at concentrations below the MRPL of 2 μg/kg (Bilandžić et al., 2012).

Conti et al. (2015) analysed samples of sea bass (n = 15) and gilthead sea bream (n = 15) collected from an aquaculture plant in Italy in 2013. An ELISA method with an LOD/LOQ of 0.2/0.5 μg/kg was used. MG was detected at concentrations up to 1.21 μg/kg (mean concentration: 0.48 μg/kg).

Samples of wild European eel were collected between 2000 and 2009 in Belgium (n = 91 sites; 1 eel/site). Samples were analysed using ultra-performance liquid chromatography (UPLC)–MS/MS (LOD/LOQ: 0.25/0.05 μg/kg). LMG and MG were detected in 74.8% and 25.3% of the samples, respectively. Concentrations up to 9.61 and 0.96 μg/kg (mean concentrations: 0.56 and 0.07 μg/kg) were reported for LMG and MG, respectively (Belpaire et al., 2015).

In Australia, a national survey was conducted in April–June 2005 for which 60 samples of local and imported farmed finfish were analysed. LMG was detected in 10 samples and in two of these MG was also detected. The highest concentration detected was 880 μg/kg for LMG in fish imported from Vietnam. Samples were analysed using LC–MS/MS (LOQ: 2 μg/kg) (FSANZ, 2005).

In the framework of a Canadian total diet study, 12 composite samples of marine, freshwater and canned fish and shrimp were collected between 2002 and 2004. MG and LMG were analysed using LC–MS/MS (LOD: 0.15 μg/kg for both MG and LMG). LMG was detected in two composite samples of freshwater fish and one composite shrimp sample (0.95, 0.73 and 1.2 μg/kg) (Tittlemier et al., 2007).

Lee et al. (2010) reported on the analysis of 253 processed fish products collected from local markets in Korea (year of sampling not indicated). Samples were analysed using HPLC-diode array detection (DAD) (LOQ: 1.0 μg/kg) and positive results were confirmed by LC–MS/MS. MG/LMG residues were detected in one shrimp sample at a concentration of 2.6 μg/kg.

Samples of catfish nuggets (n = 24) were analysed in the USA for the presence of MG and LMG (year of sampling not reported). ELISA was used as the screening method (LOD: 1 μg/kg for the sum of MG, LMG, crystal violet and leucocrystal violet, using antibodies specific for MG and crystal violet following oxidation of the leuco metabolites). MG/LMG was not detected in any of the samples (Ozbay et al., 2013). The same method was also used to analyse tilapia fillets imported into the USA (n = 36; year of sampling not reported), but none of the samples tested positive for MG or LMG (Babu and Ozbay, 2013).

Samples (n = 144) of rainbow trout, collected in 2011 from fish farms in Iran, were analysed for the presence of MG/LMG using HPLC-DAD (CCα/CCβ: 0.16/0.39 μg/kg for MG). MG residues (sum of MG and LMG) were detected in 49% of the samples at concentrations ranging from 0.3 to 146.1 μg/kg and in 33% of the samples at the level of the MRPL or higher (Fallah and Barani, 2014).

3.1.1.2 Use of malachite green as a colouring agent

In 2005, the AFC Panel reviewed the toxicology of some dyes and MG was considered in the group of ‘dyes that have been used illegally in countries outside the EU from which spices originate and dyes that have been used in the past as food colours in other countries but withdrawn from food use following discovery of toxicity’ (EFSA, 2005b). Also in the scientific literature, MG has been reported to occur as a colouring agent in food.

Ashok et al. (2014) used micellar LC-DAD (LOD/LOQ: 100/250 μg/kg) to analyse samples of green peas (n = 8), ice candy (n = 5) and chili sauce (n = 5) that had been collected in India (year of sampling not reported). MG was detected in six samples of green peas (concentration range: 2,350–4,490 μg/kg), three samples of ice candy (concentration range: 2,410–4,290 μg/kg) and one sample of chili sauce (concentration: 250 μg/kg). Similar results were reported for samples collected in India before 2002 (Tripathi et al., 2007).

Dixit et al. (2011) collected samples of candy floss, sweetened puffed rice, cream biscuits, fruit cakes, coloured fried peas, sugar-coated fennel, cereal/pulse-based sweets, sugar toys and starch-based savoury products including fryums (n = 5 for each product) from a local market in India (year of sampling not reported). Analysis was done using HPLC-UV/Vis (LOD/LOQ: 0.195–0.695/0.62–2.21 mg/L or mg/kg). MG was detected together with auramine in one sample of fryums (concentration of the sum: 65,560 μg/kg). The same authors reported the analysis of 2,409 samples of milk-based sweets, non-milk/cereal-based sweets and savoury items collected in India (year of sampling not reported; Dixit et al., 2013). MG was detected in 11 samples (0.46%) at concentrations ranging from 57,400 to 231,000 μg/kg (median: 128,500 μg/kg).

The occurrence of MG in foods available on the EU market, due to use of MG as a food colouring agent, has not been reported. The CONTAM Panel noted the illegal use of MG as a colouring agent in samples from different foodstuffs collected in India. However, the data pertaining to foods in India are not considered relevant for the current risk assessment.

3.1.2.1 National residue monitoring plans

In the period 2002–2014, at least 21,00019 targeted samples of aquaculture products (ranging from at least 860–2,600 per year) were analysed for Subgroup B3e Dyes in all Member States and Norway.20 For MG/LMG, 548 targeted samples were reported to be non-compliant (levels not reported, see Section 2.1.1). The non-compliant samples were distributed across the years as shown in Figure 2. The highest number of non-compliant samples was reported in 2006. An explanation of the limitations of the data is given in Section 2.1.1.

3.1.2.2 Rapid Alert System for Food and Feed

In the RASFF database, there were 135 notification events21 reported for MG/LMG in food products for the period 2002–2014, as seen in Figure 3. The notifications covered the following product categories: fish and fish products, crustaceans and products thereof, farmed fish and products thereof (other than crustaceans and molluscs)22 and wild-caught fish and products thereof (other than crustaceans and molluscs).23 The highest number of notifications (50) was reported in 2005.

3.2.3.1 High and frequent fish consumers

There is a concern that high and frequent consumers of fish might have elevated levels of MG/LMG dietary exposure. To test such a hypothesis, the 95th percentile hypothetical chronic dietary exposure from the daily consumption of fish among ‘consumers only’ was retrieved from the Comprehensive Database for surveys in which the number of selected participants exceeded 60.

The minimum, median and maximum of the 95th percentile MG/LMG dietary exposures for fish consumers are shown in Table 2. The dietary exposure estimations in high and frequent consumers varied from a minimum of 1.3 ng/kg bw per day in the adults group to a maximum of 11.8 ng/kg bw per day in the toddlers group.

High and frequent fish consumers were also present among pregnant women. The dietary exposure to MG/LMG for high and frequent fish consumers in this population group was calculated based on food consumption data originating from one survey. The 95th percentile dietary exposure estimate (4.1 ng/kg bw per day for pregnant women) was similar to that for the adults group.

The 95th percentile of the dietary exposure to MG/LMG is higher in the toddlers group for the high and frequent fish consumers, although it is not very different from that in the other children group. However, this was the reverse of that for the dietary exposure for all population, consumers and non-consumers. This is explained by the different percentage of ‘fish consumers-only’ between these two age groups. The toddlers ‘fish consumers-only’ account for 41% of the toddlers’ population, whereas the other children ‘fish consumers-only’ account for 32% of the other children population.

3.3.1.1 Laboratory animals

3.3.1.2 Humans

No data on the toxicokinetics of MG in humans could be found in the open literature. The only reference to human data is the in vitro capacity of human faecal microflora to almost completely reduce MG to LMG (Henderson et al., 1997).

The three main human GST classes (α, μ, π)25 were expressed in Escherichia coli and tested for their sensitivity to inhibition by MG and other basic triphenylmethane dyes. All three GST classes were inhibited by MG in a non-competitive manner displaying K_i values in the order of 0.3 μM (π), 10 μM (α) or 70 μM (μ). Despite GSH–MG adduct formation being demonstrated, the nature of the inhibitor (adduct or free dye) was not established (Glanville and Clark, 1997).

3.3.1.3 Fish and crustaceans

3.3.1.4 Accumulation in edible tissue

Maximum levels of MG in edible tissue of fish, i.e. in muscle, occur immediately after exposure whereas peak levels of LMG tend to occur later (hours to days) depending on the fish species, the MG exposure concentration and duration, and environmental conditions such as temperature and pH. Limited studies on crustaceans (shrimp) indicate that levels of both MG and LMG fell below the LOD (0.5 μg/kg) after some days following treatment.

3.3.1.5 Conclusions

There is limited knowledge on the kinetics of MG and its metabolites in mammalian and piscine species, the published studies being mostly focused on occurrence of the parent compound and its derivatives in tissues. The main metabolite is LMG, the reduced form of MG, which is slowly excreted and may be stored in fish edible tissues for a long time after exposure. An oxidative pathway, likely mediated by CYPs and other (per)oxidases, has been demonstrated entailing the sequential N-demethylation of both compounds. This results in the formation of N-demethylated metabolites (structurally similar to carcinogenic amines), which have been found to occur in fish fillets. Apart from the likely spontaneous generation of a MG–GSH adduct, which is reported to undergo biliary excretion in rats, no information could be found on the involvement of other conjugation reactions in MG/LMG biotransformation.

3.3.2.1 Acute toxicity

For female Sprague–Dawley rats an oral median lethal dose (LD₅₀) for MG-oxalate (purity not given) of 520 mg/kg bw (95% confidence limits 433–624 mg/kg bw) was reported by Meyer and Jorgenson (1983). Observed effects included depression, prostration, emaciation, coma and death. The oral LD₅₀ for MG-oxalate (purity ≥ 90%) in Wistar rats was 275 mg/kg bw. No sex differences were observed. Major acute effects were hyperaemia and atonia of the intestinal walls, often in association with dilatation of the gastrointestinal tract (Clemmensen et al., 1984).

3.3.2.2 Repeated dose toxicity

Groups of male and female B6C3F1 mice (8/sex per dose group) received a diet containing 0, 25, 100, 300, 600 or 1,200 mg MG/kg diet (corresponding to 0, 4, 18, 50, 100 or 220 mg/kg bw per day for males and 0, 5, 20, 65, 120 or 250 mg/kg bw per day for females) for 28 days. Purity of MG was ≥ 94%. Female mice at the highest dose showed a decrease in final body weight (to about 92% of controls). Feed consumption was not affected. Significant but small decreases (< 7%) in erythrocyte count, haemoglobin and haematocrit were found in females at 120 and 250 mg/kg bw per day and in males in the highest dose group. Reticulocytes were significantly increased in females at doses of 65 mg/kg bw per day and higher, but in males at the highest dose only. No significant histopathological changes were observed (Culp et al., 1999; NTP, 2004).

MG-oxalate (purity ≥ 90%) was administered to groups of male and female Wistar rats (8/sex per dose group) in the diet at concentrations of 0, 10, 100 or 1,000 mg MG-oxalate/kg diet for 28 days. Age of the animals was not reported, but body weights at the start of the experiment ranged from 170 to 250 g. These concentrations correspond to doses of 0, 1.2, 12 and 120 mg MG-oxalate/kg bw per day (equivalent to 0, 0.9, 9.4 and 94.5 mg MG/kg bw per day) when applying a default factor of 0.12 for a subacute study (EFSA Scientific Committee, 2012). No clinical signs were seen at 0.9 or 9.4 mg MG/kg bw per day, but hyperactivity and a significant reduction in body weight gain and in food consumption (none of these parameters quantified) were observed at 94.5 mg MG/kg bw per day in both sexes. High-dose females showed small haematological changes: an increase in lymphocytes, a decrease in neutrophils and a decrease in haematocrit. In high-dose males, a significant increase in plasma urea was noted (Clemmensen et al., 1984).

Male and female Fischer 344 rats (8/sex per dose group) were administered a diet containing 0, 25, 100, 300, 600 or 1,200 mg MG/kg diet (purity ≥ 94%) (corresponding to 0, 3, 12, 40, 70 or 175 mg/kg bw per day for males and 0, 3, 12, 40, 75 or 190 mg/kg bw per day for females) for 28 days. A decrease in mean body weight (to about 80% of the control for females and to 87% of the control for males) was found at the highest dose. Final body weight was significantly reduced for females at 75 and 190 mg/kg bw per day for females and at 175 mg/kg bw per day for males. Feed consumption was generally similar to that of the controls. Absolute and relative liver weights were significantly increased in females from 40 mg/kg bw per day onwards and in males in the two highest dose groups. Histopathological examination revealed an increased incidence of hepatocyte cytoplasmic vacuolisation in male rats in the two highest dose groups, 1/8 and 4/8, respectively, compared to 0/8 in the control and other groups, and of 7/8 in the highest dose females, compared to 0/8 in the control and other groups. No further details on this effect were provided. In both sexes, there was a significant increasing trend in the activity of gamma-glutamyl transferase (GGT), being 4.2-fold greater in high-dose females than in the controls. Red blood-cell parameters (erythrocyte count, haemoglobin and haematocrit) showed a small (< 5%), but significant, decrease in females of the highest dose group. Male rats showed a small (< 3%), but significant, decrease in mean erythrocyte haemoglobin at 40 mg/kg bw and higher, however without a consistent dose–response relationship. No effect on reticulocytes was observed in male and female rats (Culp et al., 1999; NTP, 2004).

Male and female Fischer 344 rats (8/sex per dose group) were fed a diet containing 0 or 1,200 mg MG/kg diet (equivalent to approximately 0 or 200 mg/kg bw per day) for 4 or 21 days (purity ≥ 94%). Triiodothyronine (T3) levels were significantly higher in treated females at 21 days than in controls, and T4 levels were significantly lower in females at 4 and 21 days. There were no significant changes in T3 and T4 levels for males and no significant changes in TSH for either sex (Culp et al., 1999; NTP, 2004).

Groups of female B6C3F1 mice (8/dose group) were fed a diet containing LMG (purity ≥ 99%) at concentrations of 0, 290, 580 or 1,160 mg/kg diet for 28 days. These concentrations correspond to doses of about 0, 60, 110 or 220 mg/kg bw per day. Final body weight and absolute and relative kidney weight were significantly decreased in the two highest dose groups. Feed consumption was lower in all dose groups, on average about 80–90% of the controls, but without a dose–response relationship. All mice at the high dose had scattered dead or degenerated cells in the transitional epithelium of the urinary bladder. Many of the cells lacked nuclei, and when visible, the nuclei were condensed or fragmented, suggesting apoptosis (Culp et al., 1999; NTP, 2004).

LMG (purity ≥ 99%) was administered to male Fischer 344 rats (8/dose group) in the diet at concentrations of 0, 290, 580 or 1,160 mg/kg diet (corresponding to 0, 30, 60 or 115 mg/kg bw per day) for 28 days. Final body weight was significantly decreased in the high-dose group. Feed consumption was on average about 10% lower in the two highest dose groups, compared to the controls. Small (< 6%) but significant decreases of haemoglobin, haematocrit and the erythrocyte count were observed in the high-dose group. A significant increase in the GGT activity and phosphorous level was observed in the high-dose group. Absolute liver weight increased with dose, but the increase was significant at the high-dose only. A significant dose-related increase in relative liver weight was found in all groups. A dose-related increase in hepatocyte cytoplasmic vacuolisation was observed (0/8, 2/8, 5/8, 7/8 for control, low- , mid-, and high-dose groups, respectively), being significant in the mid- and high-dose group. This effect was not further characterised. Upon histological examination, also apoptotic follicular epithelial cells in the thyroid gland were observed at the two highest doses (2/8 in the 60 and 2/8 in the 115 mg/kg bw per day groups), and there was evidence of follicular epithelium regeneration (Culp et al., 1999; NTP, 2004).

Male Fischer 344 rats (8/dose group) were fed a diet containing 0 or 1,160 mg LMG/kg diet (equivalent to approximately 0 or 200 mg/kg bw per day) for 4 or 21 days (purity of LMG ≥ 99%). There was a significant decrease in T4 and a significant increase in TSH levels at days 4 and 21 compared to the controls. T3 levels were not affected (Culp et al., 1999; NTP, 2004).

3.3.2.3 Immunotoxicity

No immunotoxicity studies with oral administration were identified.

3.3.2.4 Developmental and reproductive toxicity

MG oxalate (purity not given) was administered by gavage to groups of 20 pregnant New Zealand white rabbits at doses of 0, 5, 10 or 20 mg MG oxalate/kg bw per day from day 6 through 18 of gestation (Meyer and Jorgenson, 1983). Intubation errors led to loss of four animals in both the low- and high-dose groups. Body weight loss occurred in dams of the two highest dose groups. The body weights of the progeny of the treated rabbits were less than those of the controls (37.1, 34.2, 35.8 and 35.2 g for control, low-, mid- and high dose, respectively), but differences were only significant at the low- and high dose. Fetal toxicity was shown at all three dose levels. There were significant increases in preimplantation losses, but actual data on this parameter were not reported by the authors. Based on the reported data for total number of corpora lutea and total number of implantation sites, the CONTAM Panel calculated that the preimplantation loss was 8%, 19%, 20% and 17% for control, low-, mid- and high dose, respectively. At all three dose levels, the average number of resorptions per dose was significantly increased (0.8, 2.1, 2.3 and 3.8 for control, low-, mid- and high dose, respectively) and the average number of live fetuses per dose was significantly decreased (7.3, 4.9, 5.2, 5.2 for control, low-, mid- and high dose, respectively). According to the authors, early resorptions accounted for most of the dead implants, but detailed information was not provided. An increased incidence in skeletal malformations (incomplete ossification of caudal vertebrae and malformed skulls) was observed in all treated groups (20%, 36%, 38% and 38% for control, low-, mid- and high dose, respectively). In treated animals, enlargement of the liver and heart and haemorrhaging of the bladder was also observed, but without a dose–response relationship. The CONTAM Panel considered this study to be of limited value, because of the losses sustained during gavage of the animals, the inadequate reporting and the lack of a consistent dose–response relationship for the observed fetal toxicity and skeletal and visceral malformations.

LMG (Purity 99.0%) was administered orally (by gavage) to groups of 24 female Sprague–Dawley rats of 11–12 weeks of age, mated with males of the same strain (Wan et al., 2011). Administration of LMG was once daily from day 6–15 of gestation, at doses of 0, 10, 80 or 160 mg/kg bw per day. At doses of 80 and 160 mg/kg bw per day, LMG induced maternal toxicity, which consisted of a severe reduction in maternal weight and feed consumption compared to the controls. The number of corpora lutea per dam, number of implants per dam, and preimplantation losses per dam were not affected by LMG, but post-implantation losses per dam and the number of dams with resorptions were significantly increased in the highest dose group. Fetal body weight and placental weight were significantly decreased at 160 mg/kg bw per day only. In the treated groups, the incidences of fetuses (and litters) with external abnormalities were not significantly different from those in the control group. LMG induced a dose-related increase in the incidence of fetuses (reported as 8.8%, 13.8%, 36.6% and 72.3% for the control, low-, mid- and high dose, respectively) and litters (reported as 34.8%, 58.3%, 73.7% and 100% for the control, low-, mid- and high dose, respectively) with skeletal abnormalities. The predominant skeletal abnormalities were bipartite ossification of the thoracic centrum (reported as 11%, 19%, 41% and 45% of the fetuses for the control, low-, mid- and high dose, respectively). These effects were significant at doses of 80 mg/kg bw per day and higher. The CONTAM Panel noted that the number of fetuses with skeletal abnormalities for each litter was not reported. In the high-dose group (160 mg/kg bw per day), LMG also induced a significant increase in the number of fetuses and litters (%) with visceral abnormalities, particularly hydronephrosis, compared with the controls. Based on the observed effects, a NOAEL of 10 mg/kg bw per day for maternal and fetal developmental toxicity in rats could be identified.

The developmental toxicity of MG has also been studied in a model with zebrafish (Danio rerio) embryos. Exposure of zebrafish embryos to MG concentrations of 0.125, 0.15 or 0.175 mg/L for 14 h significantly altered cardiovascular development causing growth retardation through blocking of the activation of the vascular endothelial growth factor receptor 2. In contrast, exposure up to 0.875 mg LMG/L for 14 h did not affect zebrafish embryo growth (Jang et al., 2009).

Minta and Wilk-Zasadna (2007) studied potential embryotoxic activity of MG and LMG in an in vitro model using micromass cultures of rat embryonic mesenchyme limb and midbrain cells. Embryonic mesenchyme limb and midbrain cells cultured at high density were exposed to 0.002, 0.005, 0.013, 0.032, 0.080, 0.200, and 0.500 μg/mL MG and 0.03, 0.12, 0.49, 1.95, 7.81, 31.25, and 125 μg/mL LMG. After 5 days of incubation, the viability and differentiation were determined. From the dose–response curves, the concentrations (μg/mL) that produced 50% inhibition of viability/proliferation (IC₅₀-P) and 50% reduction of differentiation (IC₅₀-D) were calculated. Embryotoxicity was classified according to the prediction model described by Genschow et al. (2002). MG and LMG induced dose-dependent inhibition of cell survival and differentiation in both cell cultures. MG was approximately 200–300 times more potent than LMG. The reduction of cell survival and differentiation were observed at similar concentrations. IC₅₀-P values ranged from 0.02–0.08 and 5–17 μg/mL for MG and LMG, respectively. According to the prediction model, MG was classified as strongly embryotoxic in mesenchyme limb cells (IC₅₀-D = 0.02 μg/mL) and midbrain cells (IC₅₀-D = 0.05 μg/mL), whereas LMG was classified as weakly embryotoxic in mesenchyme limb cells (IC₅₀-D = 15.6 μg/mL) and strongly embryotoxic in midbrain cells (IC₅₀-D = 3.7 μg/mL).

3.3.2.5 Neurotoxicity

No neurotoxicity studies with oral administration were identified.

3.3.2.6 Genotoxicity

The in vitro and in vivo genotoxicity studies for MG and LMG are summarised in Tables 4 and 5, respectively.

Bacterial reverse mutation assays (Ames test) of MG were performed with Salmonella Typhimurium strains TA97, TA98, TA100, TA102, TA104, TA1535 and TA1537. Because MG is highly toxic to bacteria, it was tested only at low doses, with a few exceptions. In one study, MG was mutagenic only in TA98 in the presence of S9 (Clemmensen et al., 1984). In later studies, the compound was tested at doses up to 10 μg/plate only, due to its antibacterial activity, and did not show any mutagenic activity (Fessard et al., 1999; NTP, 2004). LMG showed no mutagenic activity at doses up to 2,000 μg/plate (Fessard et al., 1999).

Assessment of the in vitro mutagenicity of MG or LMG in the Chinese hamster ovary (CHO) hypoxanthine–guanine phosphoribosyltransferase (HPRT) locus assay gave negative results, with and without S9 activation (Fessard et al., 1999). Due to severe cytotoxicity, evaluation of the mutagenicity of MG was restricted to very low concentrations (up to 0.05 μg/mL and 1 μg/mL without and with metabolic activation, respectively). LMG was less cytotoxic than MG and mutagenicity of LMG was evaluated at concentrations of 5–100 μg/mL.

In CHO cells, the capacity of MG and LMG to induce DNA damage was determined by the Comet assay (Fessard et al., 1999). When tested without metabolic activation, MG induced an increase in DNA strand breaks at concentrations ≥ 3 μg/mL, at which loss of viability ranged from 20% to 30%. The significance of the positive result observed at the 10 μg/mL dose is questionable due to a 70% loss of viability at this dose. After metabolic activation, DNA damage was observed at concentrations (15–20 μg/mL) where cell viability was reduced by less than 20%. In the same study, LMG tested at concentrations of 5–500 μg/mL did not induce DNA damage or loss of viability (Fessard et al., 1999).

MG-induced DNA strand breaks were reported also in Syrian hamster embryo (SHE) cells after exposure to 0.0125–0.1 μg/mL (Bose et al., 2005). In the same study, the authors observed MG-mediated cell cycle arrest and induction of apoptosis. Panandiker et al. (1994) showed an association of MG-induced DNA strand breaks with the formation of free radicals in SHE cells. Several studies showed that MG induces malignant transformation in vitro in SHE cells (Panandiker et al., 1993; Mahudawala et al., 1999). Au and Hsu (1979) reported no evidence of MG-induced chromosomal aberrations in vitro in cultured CHO cells.

In vivo genotoxicity studies for MG and LMG include chromosomal aberration and micronucleus assay, forward mutation assays, lacI and cII transgene mutation assays with rodents and ³²P-postlabelling studies of liver DNA adduct formation. No increase in micronucleated erythrocytes was observed in bone marrow of mice administered a single dose of 37.5 mg/kg bw MG oxalate by gavage (Clemmensen et al., 1984).

In rat bone marrow cells, following three intraperitoneal injections of MG chloride at doses ranging from 1,094 to 8,750 mg/kg bw, a very small but significant increase in micronuclei frequency was observed only at a dose of 4,375 mg/kg bw (NTP, 2004). As the frequency of micronucleated polychromated erythrocytes (PCEs) was not significant at a dose of 8,750 mg/kg and no bone marrow toxicity was detected at this dose, the test in rats was judged to be negative overall. No increase in micronuclei was observed in peripheral blood normochromatic erythrocytes (NCEs) or PCEs of male and female mice after 28 days of exposure to MG chloride in feed (up to 1,200 mg/kg; approximately 250 mg/kg bw per day) (NTP, 2004).

MG also did not induce increase in micronucleated NCEs or PCEs in female Big Blue^® B6C3F1 mice that received 450 mg/kg MG in the diet for 4 or 16 weeks (Mittelstaedt et al., 2004). On the contrary, Donya et al. (2012) reported a dose- and time-dependent increase in chromosomal aberration in bone marrow and spermatocytes in Swiss albino male mice treated daily by gavage with up to 543 mg MG/kg bw for 7, 14, 21 and 28 days. In the same study, the authors reported also dose- and time-dependent increase in sister chromatid exchanges in bone marrow and DNA fragmentation in liver cells. Induction of chromosomal aberrations, micronuclei formation and DNA damage in bone marrow have also been observed in Swiss albino female mice treated by gavage with 4 mg MG/kg bw for 30 days (Das et al., 2013).

A peripheral blood micronucleus test with LMG was performed in female B6C3F1/Nctr BR mice after 28 days of exposure to 290, 580, or 1,160 mg LMG/kg in feed. Significant increases in the frequency of micronucleated NCEs were observed in the 290 and 580 mg/kg groups. The increase of micronucleated PCEs was not significant (NTP, 2004). In female Big Blue^® B6C3F1 mice that received 204 or 408 mg LMG/kg in the diet for 4 or 16 weeks, LMG did not induce an increase in micronucleated NCEs or PCEs (Mittelstaedt et al., 2004).

It has been shown that exposure to MG (100 or 600 mg/kg feed) for 28 days induces the formation of DNA adducts in liver of male F344 rats and female B6C3F1 mice (Culp et al., 1999). Under the same exposure conditions, LMG induced the formation of DNA adducts in rat, but not in mouse liver (Culp et al., 1999). A dose-related increase in DNA adduct levels in liver has been reported also in female Big Blue rats^® fed a diet containing 9–543 mg LMG/kg for 4 weeks (Culp et al., 2002; Manjanatha et al., 2004). Because the results are based on ³²P-postlabelling analysis, without further characterisation of the nature of adducts, the CONTAM Panel noted that these adducts are not necessarily MG/LMG metabolites bound to DNA.

When livers of female Big Blue^® rats treated with 273 and 543 mg LMG/kg were analysed for lacI transgene mutations at 4, 16, and 32 weeks post-exposure, a significant increase (2.9-fold) in lacI mutant frequency was observed only in the livers of rats fed 543 mg LMG/kg for 16 weeks (Culp et al., 2002). However, molecular analysis of 80 of these liver lacI mutants revealed that 21% (17/80) were clonal in origin and that the majority (55/63) of the independent mutations were base pair substitutions involving guanine (G)-cytosine (C) to adenine (A)-thymine (T) transitions, similar to those found for control rats, suggesting that LMG is not a mutagen in the livers of female rats (Manjanatha et al., 2004). In the same female Big Blue^® rats orally exposed to LMG, the analysis of the induction of HPRT mutations in lymphocytes also showed negative results (Manjanatha et al., 2004).

In another study, transgenic female Big Blue^® B6C3F1 mice were given feed with MG (450 mg/kg) or LMG (204 and 408 mg/kg) for 4 and 16 weeks (Mittelstaedt et al., 2004). An increase in the liver cII transgene mutant frequency was observed after a 16-week treatment with 408 mg LMG/kg, indicating that LMG is an in vivo mutagen in transgenic female mouse liver. Molecular analyses of mutations from mice fed LMG had increased frequencies of G → T and A → T transversions. Statistical evaluation indicated that the spectrum of mutations for LMG-treated mice is significantly different from the controls. MG did not increase the mutation frequency of the cII transgene in the liver and neither MG nor LMG increased the HPRT mutation frequency in lymphocytes at either time point (Mittelstaedt et al., 2004).

3.3.2.7 Chronic toxicity and carcinogenicity

Groups of 48 female B6C3F1 mice were fed diets containing 0, 100, 225 or 450 mg MG/kg for 104 weeks (NTP, 2005; Culp et al., 2006). These dietary concentrations correspond to doses of approximately 0, 15, 33 or 67 mg/kg bw per day. The purity of MG was about 87%, with identified impurities of LMG (7.5%), N-desmethyl MG (3.8%) and N-desmethyl LMG (0.5%). There were no effects on body weight or on the survival rate of female mice, but absolute kidney weights were decreased in the mid- and high-dose groups. No treatment-related neoplasms were observed.

Groups of 48 female F344 rats were fed diets containing 0, 100, 300 or 600 mg MG/kg for 104 weeks (NTP, 2005; Culp et al., 2006), equivalent to average doses of 0, 7, 21 or 43 mg/kg bw per day. The purity of MG was about 87% (for impurities, see above). Body weights were decreased, particularly in the mid- and high-dose groups, being approximately 90% (mid dose) and 88% (high dose) of the control group weight at the end of the experimental period. In these groups, a small decrease in feed consumption was observed during the first year of the study compared to that of the control group. The survival rate was not affected. A small increase in the incidence of thyroid gland follicular cell adenoma or carcinoma was observed (0/46, 0/48, 3/47 and 2/46 for control, low-, mid- and high-dose groups, respectively). According to the NTP technical report (NTP, 2005), the incidence of thyroid gland follicular carcinoma was 0/46, 0/48, 2/47 and 1/46 for control, low-, mid- and high dose, respectively. The incidence of mammary gland carcinoma was increased in the high-dose group (2/48, 2/48, 1/48 and 5/48 for control, low-, mid- and high-dose groups, respectively). A dose-related increase in the incidence of hepatocellular adenomas was observed in female rats fed MG (1/48, 1/48, 3/48 and 4/48 for control, low-, mid- and high-dose groups, respectively), being significant only in the high-dose group.

Groups of 48 female B6C3F1 mice were fed diets containing 0, 91, 204 or 408 mg LMG/kg (purity about 99%) for 104 weeks (NTP, 2005; Culp et al., 2006). These dietary concentrations are equivalent to average doses of approximately 0, 13, 31 or 63 mg/kg bw per day. Body weight and survival rate were not affected by treatment with LMG, but a decrease in relative kidney weights was observed in all dose groups, however without a consistent dose–response relationship. The only neoplastic finding was a dose-related increase in the incidence of hepatocellular adenomas or carcinomas (3/47, 6/48, 6/47 and 11/47 for control, low-, mid- and high-dose groups, respectively), reported as significant in the high-dose group. The data reported by NTP (2005) indicated that hepatocellular adenomas formed the largest part of these tumours, and that only a low number of malignant liver tumours were found – the incidence of hepatocellular carcinomas alone was 0/47, 0/48, 1/47 and 2/47 for control-, low-, mid- and high-dose groups, respectively.

Male and female F344 rats (48 per sex) were fed diets containing 0, 91, 272 or 543 mg LMG/kg (purity about 99%) for 104 weeks (NTP, 2005; Culp et al., 2006). These dietary concentrations are equivalent to average doses of 0, 5, 15 or 30 mg/kg bw per day for males and 6, 17, or 35 mg/kg bw for females. Reduced body weights were found in both sexes, particularly in the mid- and high-dose groups, with body weights at termination of the experiment being approximately 92% (mid dose) and 87% (high dose) of the control group for males, and approximately 89% (mid dose) and 77% (high dose) of the control group for females. In mid- and high-dose males and females, a small decrease in feed consumption was observed throughout the study, compared to that by the control group. The survival rate was not affected. The absolute (and relative) liver weights of males were significantly increased in the mid- and high-dose groups. For females, only the relative liver weight was significantly increased in the mid- and high-dose groups. The incidence of hepatic cytoplasmic vacuolisation was increased in female rats (5/48, 5/48, 19/48, 22/48 for control, low-, mid- and high-dose groups, respectively). No further details on this effect were provided. In male and female rats, follicular cysts in the thyroid were found in the high-dose group, accompanied by an increase in relative thyroid weight. Both sexes also showed a low, not dose-related, increase in incidence of thyroid gland follicular cell adenomas or carcinomas (females 0/46, 1/46, 2/47, 1/48 and males 0/47, 2/47, 1/48, 3/46 for control, low-, mid- and high-dose groups, respectively). Female rats had an increased incidence of mammary gland adenomas or carcinomas: 0/48, 2/48, 3/48, 4/48 for control, low-, mid- and high-dose groups, respectively. The NTP (2005) report showed that the incidence of mammary gland adenomas was 0/48, 1/48, 1/48 and 2/48 and of mammary gland carcinomas 0/48/, 1/48, 2/48 and 2/48 for the control, low-, mid- and high-dose groups, respectively. Male rats had a dose-related increasing trend in interstitial cell adenomas of the testis (37/48, 42/47, 43/48, 45/47 for control, low-, mid- and high-dose groups, respectively).

3.3.5.1 Neoplastic effects

For both MG and LMG, long-term carcinogenicity studies are available (see Section 3.3.2.7). The CONTAM Panel concluded that MG may be considered as carcinogenic in rats and LMG in both mice and rats. No information was identified on the carcinogenicity of MG or LMG in humans. For the selection of the reference point for neoplastic effects, the CONTAM Panel focused on those organs in which malignant tumours had been observed.

MG induced thyroid gland follicular cell adenomas or carcinomas and mammary gland carcinomas in female F344 rats, but in female B6C3F1 mice no treatment-related neoplasms were observed (NTP, 2005; Culp et al., 2006). Based on positive results in the in vivo micronucleus tests in mice and the capacity to form DNA adducts in vivo, the CONTAM Panel considered MG to be genotoxic in vivo. Therefore, MG may be regarded as a substance which is genotoxic and carcinogenic.

LMG induced hepatocellular adenomas or carcinomas in female B6C3F1 mice and, in female F344 rats, induced thyroid gland follicular cell adenomas or carcinomas and mammary gland adenomas or carcinomas. Based on positive results in the in vivo micronucleus tests in mice, positive results in cII transgene mutations in mouse liver, and the capacity to form DNA adducts in vivo, the CONTAM Panel considered LMG to be genotoxic in vivo. Therefore, LMG may be regarded as a substance which is genotoxic and carcinogenic.

Based on these conclusions regarding the genotoxicity and carcinogenicity of MG and LMG, the CONTAM Panel concluded that the derivation of a health-based guidance value (HBGV) for these compounds is not appropriate, and decided to apply an MOE approach for the risk characterisation.

The CONTAM Panel selected the BMDL₁₀ value of 13 mg/kg bw per day (rounded figure) derived for LMG. As both the RPA and the data on occurrence of MG and LMG in fish and crustaceans are expressed for MG/LMG, and consequently, the exposure is estimated for MG/LMG, the CONTAM Panel used the selected BMDL₁₀ value as a reference point for the neoplastic effects of MG/LMG.

The CONTAM Panel noted that JECFA also selected the incidence of hepatocellular adenomas and carcinomas (combined) in female mice induced by LMG as the pivotal effect for the risk assessment. JECFA calculated BMDL₁₀ values ranging from 18.5 to 31.2 mg/kg bw per day and selected a BMDL₁₀ of 20 mg LMG/kg bw per day as the reference point for the risk characterisation of neoplastic effects. The CONTAM Panel also noted that JECFA used 15 mg/kg bw per day as the dose for the low-dose group, while in the current risk assessment a dose of 13 mg/kg bw per day, which is the dose reported in the NTP study, is used (NTP, 2005).

3.3.5.2 Non-neoplastic effects

For the evaluation of non-neoplastic effects, 28-day, long-term (104 weeks) and reproductive and developmental toxicity studies are available for both MG and LMG (see Section 3.3.2).

Based on the effect of MG on liver weight and of LMG on body weight, the CONTAM Panel selected the BMDL₀₅ value of 6 mg/kg bw per day (rounded figure), as a reference point for the non-neoplastic effects of MG/LMG. The CONTAM Panel noted that this value is higher than the acute dose that caused adverse effects in a case of intoxication of an infant. However, the composition of the product was not reported and this is further addressed in Section 3.5.

A.1.1.1 Web of Science

Used search string: TOPIC: (“malachite green” OR “leucomalachite green”) AND TOPIC: (chemistry OR analysis OR determination OR detection OR ELISA OR immune* OR GC OR GC-MS OR HPLC OR LC-MS OR ICP-MS) NOT TOPIC: (“waste water*” OR waste OR wastewater* OR adsorp* OR degradat* OR decoloriz* OR Ichthyophthirius OR mycobacterium OR tile OR ceramic OR pottery OR Saprolegnia OR salmonella OR coli); Refined by: RESEARCH AREAS: (CHEMISTRY OR FOOD SCIENCE TECHNOLOGY OR SPECTROSCOPY OR INSTRUMENTS INSTRUMENTATION OR AGRICULTURE OR VETERINARY SCIENCES OR FISHERIES); Timespan=2012-2015; Search language=Auto

Result in Web of Science: 267

A.1.1.2 PubMed

Used search string: (((((“malachite green” OR “leucomalachite green”)) AND (chemistry OR analysis OR determination OR detection OR elisa OR immune* OR GC OR gc-ms OR hplc OR lc-ms OR icp-ms)) AND (“2012/01/01”[PDAT]: “3000”[PDAT]))) NOT (“waste water*” OR waste OR wastewater* OR adsorp* OR degradati* OR decoloriz* OR ichthyophthirius OR mycobacterium OR tile OR ceramic OR potter OR saprolegnia OR salmonella OR coli)

Results in PubMed: 251

A.1.2.1 Web of Science

Used search string: TOPIC: (“malachite green” OR “leucomalachite green”) AND TOPIC: (occurrence OR exposure OR assessment OR levels OR concentrate* OR beverage OR food OR vegetable* OR fruit* OR grain OR cereal OR poultry OR meat OR eggs OR milk OR seafood OR fish OR shrimp OR prawns OR crustacean) NOT TOPIC: (“waste water” OR waste OR wastewater OR adsorp* OR degradati* OR decoloriz* OR Ichthyophthirius OR mycobacterium OR tile OR ceramic OR pottery OR Saprolegnia OR salmonella OR coli); Refined by: RESEARCH AREAS: (CHEMISTRY OR FOOD SCIENCE TECHNOLOGY OR VETERINARY SCIENCES OR FISHERIES OR SPECTROSCOPY OR INSTRUMENTS INSTRUMENTATION OR AGRICULTURE); Timespan=2002-2015;Search language=Auto

Result in Web of Science: 475

A.1.2.2 PubMed

Used search string: ((((“malachite green” OR “leucomalachite green”)) AND (occurrence OR exposure OR assessment OR levels OR concentrate* OR beverage OR food OR vegetable* OR fruit* OR grain OR cereal OR poultry OR meat OR eggs OR milk OR seafood OR fish OR shrimp OR prawns OR crustacean)) AND (“2002/01/01”[Date - Publication]: “3000”[Date - Publication])) NOT (“waste water” OR waste OR wastewater OR adsorp* OR degradati* OR decoloriz* OR ichthyophthirius OR mycobacterium OR tile OR ceramic OR potter OR saprolegnia OR salmonella OR coli OR soil)

Results in PubMed: 166

A.1.3.1 Web of Science

Used search string: TOPIC: (“malachite green”) AND TOPIC: ((adulteration or “food colour*” or “food color*”)); Timespan=2002-2015; Search language=Auto

Result in Web of Science: 13

A.1.3.2 PubMed

Used search string: ((“malachite green”) AND (adulteration or “food colour*” or “food color*”)) AND (“2002/01/01”[Date - Publication]: “3000”[Date - Publication])

Results in PubMed: 2

A.1.4.1 Web of Science

Used search string: TOPIC: (“malachite green” OR “leucomalachite green”) AND TOPIC: (cooking OR roasting OR frying OR boiling OR baking OR “thermal processing”) NOT TOPIC: (“waste water” OR waste OR wastewater OR decolor* OR biodegrad* OR photocatal* OR photodegrad* OR adsorp* OR Ichthyophthirius OR mycobacterium OR tile OR ceramic OR pottery OR Saprolegnia OR salmonella OR coli); Refined by: RESEARCH AREAS: (CHEMISTRY OR VETERINARY SCIENCES OR AGRICULTURE OR FOOD SCIENCE TECHNOLOGY OR FISHERIES); Timespan=All years; Search language=Auto

Result in Web of Science: 29

A.1.4.2 PubMed

Used search string: (((“malachite green” OR “leucomalachite green”)) AND (processing OR cooking OR roasting OR frying OR boiling OR baking OR “thermal processing” OR sterilization))

Results in PubMed: 16

A.1.5.1 Web of Science

Used search string: TOPIC: (“malachite green” OR “leucomalachite green”) AND TOPIC: (toxicokinetic* OR metabolism* OR distribution OR excretion OR “mode of action” OR absorption OR biomarker OR biotransformation) NOT TOPIC: (“waste water*” OR fluorogen OR waste OR “surface plasmon resonance” OR SPR OR “light scattering” OR film* OR wastewater* OR soil* OR humic OR humus OR nano* OR decolor* OR biodegrad* OR photocatal* OR photodegrad* OR adsorp* OR Ichthyophthirius OR mycobacterium OR tile OR ceramic OR pottery OR Saprolegnia OR salmonella OR coli); Refined by: RESEARCH AREAS: (CHEMISTRY OR BIOCHEMISTRY MOLECULAR BIOLOGY OR PHARMACOLOGY PHARMACY OR TOXICOLOGY OR CELL BIOLOGY OR ENDOCRINOLOGY METABOLISM OR VETERINARY SCIENCES); Timespan=All years; Search language=Auto

Result in Web of Science: 557

A.1.5.2 PubMed

Used search string: (((“malachite green” OR “leucomalachite green”)) AND (toxicokinetic* OR metabolism* OR distribution OR excretion OR “mode of action” OR absorption OR biomarker OR biotransformation)) NOT (“waste water*” OR fluorogenic OR waste OR “surface plasmon resonance” OR SPR OR “light scattering” OR film* OR wastewater* OR soil* OR humic OR humus OR nano* OR decolor* OR biodegrad* OR photocatal* OR photodegrad* OR adsorp* OR ichthyophthirius OR mycobacterium OR tile OR ceramic OR potter OR saprolegnia OR salmonella OR coli)

Result in PubMed: 301

A.1.6.1 Web of Science

Used search string: TOPIC: (“malachite green” OR “leucomalachite green”) AND TOPIC: (toxicity OR toxi* OR mutagen* OR teratogen* OR carcinogen* OR carcino* OR genotox* OR reprotox* OR nephrotox* OR neurotox* OR hepatotox* OR immunotox* OR haemotox* OR hematotox* OR cytotox* OR “develop* toxicity” OR endocri*) NOT TOPIC: (“waste water*” OR fluorogen OR waste OR “surface plasmon resonance” OR SPR OR “light scattering” OR film* OR wastewater* OR soil* OR humic OR humus OR nano* OR decolor* OR biodegrad* OR photocatal* OR photodegrad* OR adsorp* OR Ichthyophthirius OR mycobacterium OR tile OR ceramic OR pottery OR Saprolegnia); Refined by: RESEARCH AREAS: (TOXICOLOGY OR CARDIOVASCULAR SYSTEM CARDIOLOGY OR ANATOMY MORPHOLOGY OR PHARMACOLOGY PHARMACY OR OPHTHALMOLOGY OR BIOCHEMISTRY MOLECULAR BIOLOGY OR LIFE SCIENCES BIOMEDICINE OTHER TOPICS OR DERMATOLOGY OR HEMATOLOGY OR GERIATRICS GERONTOLOGY OR ONCOLOGY OR UROLOGY NEPHROLOGY OR CELL BIOLOGY OR PATHOLOGY OR RHEUMATOLOGY OR ENDOCRINOLOGY METABOLISM OR RESPIRATORY SYSTEM OR ORTHOPEDICS OR OBSTETRICS GYNECOLOGY OR NEUROSCIENCES NEUROLOGY OR DEVELOPMENTAL BIOLOGY OR GASTROENTEROLOGY HEPATOLOGY OR DENTISTRY ORAL SURGERY MEDICINE OR IMMUNOLOGY OR REPRODUCTIVE BIOLOGY OR ALLERGY OR PEDIATRICS OR PUBLIC ENVIRONMENTAL OCCUPATIONAL HEALTH); Timespan=All years; Search language=Auto

Result in Web of Science: 555

A.1.6.2 PubMed

Used search string: (((((“malachite green” OR “leucomalachite green”)) AND (toxicity OR toxi* OR mutagen* OR teratogen* OR carcinogen* OR carcino* OR genotox* OR reprotox* OR nephrotox* OR neurotox* OR hepatotox* OR immunotox* OR haemotox* OR hematotox* OR cytotox* OR “develop* toxicity” OR endocri*))) NOT (“waste water*” OR fluorogen* OR waste OR film* OR wastewater* OR soil* OR humic OR humus OR nano* OR decolor* OR biodegrad* OR photocatal* OR photodegrad* OR adsorp* OR Ichthyophthirius OR mycobacterium OR tile OR ceramic OR pottery OR Saprolegnia))

Result in PubMed: 132

A.1.7.1 Web of Science

Used search string: TOPIC: (“malachite green” OR “leucomalachite green”) AND TOPIC: (epidemiology OR biomarker OR cohort OR “case control” OR “case stud*” OR “adverse effect*”) NOT TOPIC: (“waste water*” OR fluorogen OR waste OR “surface plasmon resonance” OR SPR OR “light scattering” OR film* OR wastewater* OR soil* OR humic OR humus OR nano* OR decolor* OR biodegrad* OR photocatal* OR photodegrad* OR adsorp* OR Ichthyophthirius OR mycobacterium OR tile OR ceramic OR pottery OR Saprolegnia); Refined by: RESEARCH AREAS: (TOXICOLOGY OR HEMATOLOGY OR PHARMACOLOGY PHARMACY OR DEVELOPMENTAL BIOLOGY OR ANATOMY MORPHOLOGY OR UROLOGY NEPHROLOGY OR SURGERY OR PEDIATRICS OR RHEUMATOLOGY OR BIOCHEMISTRY MOLECULAR BIOLOGY OR IMMUNOLOGY OR RESPIRATORY SYSTEM OR PUBLIC ENVIRONMENTAL OCCUPATIONAL HEALTH OR DERMATOLOGY OR RESEARCH EXPERIMENTAL MEDICINE OR CARDIOVASCULAR SYSTEM CARDIOLOGY OR REPRODUCTIVE BIOLOGY OR OPHTHALMOLOGY OR ONCOLOGY OR PATHOLOGY OR GERIATRICS GERONTOLOGY OR GASTROENTEROLOGY HEPATOLOGY OR CELL BIOLOGY OR ANESTHESIOLOGY OR DENTISTRY ORAL SURGERY MEDICINE OR ENDOCRINOLOGY METABOLISM); Timespan=All years; Search language=Auto

Result in Web of Science: 63

A.1.7.2 PubMed

Used search string: (((“malachite green” OR “leucomalachite green”)) AND (epidemiology OR biomarker OR cohort OR “case control” OR “case stud*” OR “adverse effect*”)) AND (“waste water*” OR fluorogen* OR waste OR film* OR wastewater* OR soil* OR humic OR humus OR nano* OR decolor* OR biodegrad* OR photocatal* OR photodegrad* OR adsorp* OR Ichthyophthirius OR mycobacterium OR tile OR ceramic OR pottery OR Saprolegnia)

Result in PubMed: 11