WHO FOOD ADDITIVES SERIES: 51

First draft prepared by
Professor Fritz R. Ungemach
Institute of Pharmacology, Pharmacy and Toxicology, Veterinary Faculty, University of Leipzig, Leipzig, Germany

The quinoxaline-1,4-dioxide compound carbadox (methyl-3-[quinoxalinyl-methylene]carbazate-N¹,N⁴-dioxide) is used in pigs as a growth-promoting agent for the improvement of weight gain and feed efficiency and as an antibacterial drug for prevention and control of dysentery and bacterial enteritis in pigs.

Carbadox was evaluated by the Committee at its thirty-sixth meeting (Annex 1, reference 92). At that time, the Committee was not able to establish an ADI because carbadox and some of its metabolites (desoxycarbadox and hydrazine) were found to be genotoxic and carcinogenic in rodents. The final metabolite, quinoxaline-2-carboxylic acid (QCA), was not found to be carcinogenic or mutagenic in animals. Studies of residues showed rapid depletion of the parent substance and its genotoxic metabolites in liver and muscle within 72 h, to concentrations of < 2 µg/kg, within the limit of detection of the analytical method available at that time (MacIntosh et al., 1985). QCA was the most persistent metabolite and was the only residue detectable in edible tissues of pigs 72 h after dosing. After 28 days’ withdrawal, its concentration was < 30 µg/kg in liver and 5 µg/kg in muscle, representing the limits of quantification of the analytical method used at that time, which was based on extraction by alkaline hydrolysis. Uncertainty remained about the nature of the bound residues in liver; however, the Committee concluded that the bound residues in liver 28 days after treatment would not represent a risk for consumers. On the basis of data from studies on the toxicity of QCA, on the metabolism and depletion of carbadox and on the nature of the compounds released from the bound residues, the Committee concluded that residues resulting from use of carbadox in pigs were acceptable provided that the recommended maximum residue limits of 0.03 mg/kg in liver and 0.005 mg/kg in muscle, based on the levels of and expressed as QCA, were not exceeded. It concluded that use of carbadox according to good practice in veterinary medicine (no use in finisher pigs and a withdrawal period of at least 28 days) does not represent a dietary hazard to human health.

Carbadox was banned for use in food-producing animals in the European Union in 1999 because of its genotoxic and carcinogenic properties and for the safety of workers. Health Canada issued an order to stop the sale of carbadox in 2001. Reports of misuse and cross-contamination, combined with a better analytical capacity to detect desoxycarbadox, showed that this carcinogenic residue was present in tissues and rendered products obtained from leftover tissue of pigs that had not been adequately withdrawn. These resulted in serious concern about the safety of this drug (Vilim & Lambert, 2001).

At the request of the Canadian Government, the Codex Committee on Veterinary Drug Residues in Foods placed carbadox on the list of priorities for the sixtieth meeting of the Expert Committee and asked it to review all relevant data on the toxicology and residues of carbadox and its metabolites in porcine tissues, including analytical methods for their detection that had been generated since the previous evaluation by JECFA.

At its previous evaluation, the Committee considered studies on the reproductive toxicity of carbadox and its persistent end-metabolite QCA. In a three-generation study, with two litters per generation, Charles River C-D rats were given diets containing carbadox at concentrations providing a dose of 0, 1 or 2.5 mg/kg bw per day. No evidence of reproductive toxicity was found in the first two generations. Occasional differences from control group parameters were sporadic and considered not to be related to treatment. It was concluded that carbadox had no effect on fertility, lactation or the neonate at doses up to 2.5 mg/kg bw per day. The potential reproductive toxicity of QCA was studied in rats and rabbits. In a three-generation study of reproductive toxicity in rats given diets containing QCA at concentrations providing doses up to 100 mg/kg bw per day, no treatment-related effects were observed. In studies of developmental toxicity in New Zealand white rabbits given an oral dose of 0, 25, 50 or 100 mg/kg bw per day on days 7–18 of gestation, there was no evidence of maternal toxicity, embryotoxicity or teratogenicity. The NOEL was 100 mg/kg bw per day.

A study of the teratogenicity of carbadox in rats has been reported since the Committee’s previous evaluation. No statement was included about the compliance of the study with good laboratory practice. The protocol and conduct of the study were compliant with recognized testing guidelines for developmental toxicity, except that there were only 10 animals per group and treatment was not begun before day 8 of gestation. An aqueous solution of carbadox was administered by oral gavage once daily to groups of 10 pregnant Sprague-Dawley rats at a dose of 0, 10, 25, 50 or 100 mg/kg bw on days 8–15 of gestation. The dams were observed daily for body-weight gain. The animals were killed on day 21 of gestation, the fetuses were removed surgically, and the numbers of implants, resorptions and live pups were counted. The fetuses were weighed and examined for sex ratio and for external, skeletal and internal malformations. The viability of the pups and the number of corpora lutea were not reported.

Maternal body-weight gain was significantly reduced in a dose-related manner at all doses, but none of the treated dams died. After discontinuation of treatment, the body-weight gain recovered on days 15–21, except for animals at the highest dose. At doses of up to 50 mg/kg bw per day, the number of live pups was no different from that in the control group, with a mean of 13.5 fetuses per litter. Embryolethality was seen at the highest dose, the number of live fetuses being reduced to 2.4 per litter; the percentage of late resorptions was 82%, with complete resorptions in five dams, compared with 3.4% in the control group. No treatment-related effects on early resorptions were reported. A dose-related reduction in fetal body weight was seen, which was statistically significant at doses > 25 mg/kg bw per day. At the highest dose, carbadox induced a significantly increased incidence of malformations: in 24 fetuses examined for external malformations, short tail was recorded in 11, kinky tail in five, brachygnathia in five, ectrodactyly in three, club foot in one and generalized oedema in one. The skeletal malformations observed were mainly fused vertebrae, found in 8 of 17 fetuses. Internal examination revealed hydrocephaly in two of seven fetuses. The fetal sex ratio was affected at 100 mg/kg bw per day, but not statistically significantly so.

The author concluded that maternally toxic doses of carbadox caused embryo-toxicity and embryolethality. Carbadox was also considered to be teratogenic. The occurrence of malformations and severe maternal toxicity at the highest dose tested suggested that adverse effects on maternal animals and a direct action on the conceptus contributed to fetal malformations and embryo and fetal deaths. The NOEL was 10 mg/kg bw per day for embryotoxicity and 50 mg/kg bw per day for teratogenicity. No NOEL for maternal toxicity could be identified (Yoshimura, 2002).

The NOEL of 10 mg/kg bw per day is well above the NOEL of 2.5 mg/kg bw per day found in studies of reproductive toxicity in rats and also exceeds the lowest dose of 1 or 5 mg/kg bw per day that had adverse effects in rats in long-term studies of toxicity evaluated previously by the Committee (Annex 1, references 86 and 92). QCA, the predominant residue after carbadox treatment, showed no adverse effects in studies of reproductive toxicity at doses ­ 100 mg/kg bw per day and was considered not to be teratogenic. Therefore, the results of this new study do not affect the previous evaluation of the Committee with regard to reproductive and developmental toxicity and its overall conclusion on risk characterization of carbadox.

The main findings of studies of reproductive toxicity with carbadox are summarized in Table 1.

Table 1. Summary of pivotal studies of reproductive and developmental toxicity with carbadox and quinoxaline-2-carboxylic acid (QCA) given orally

In its previous evaluation, the Committee considered the results of a range of tests for genotoxicity in bacteria and mammalian cells with carbadox and its metabolites desoxycarbadox, methyl carbazate and its possible hydrolysis product hydrazine, and QCA. Positive results were obtained with carbadox in 14 of 15 tests. Desoxycarbadox gave negative results in many test systems, but positive findings were reported for cell transformation in mouse cells, for chromosomal aberration in rat bone marrow and for reverse mutation in Salmonella typhimurium with microsomes from rats treated with polychlorinated biphenyls. Hydrazine gave positive results in three mutagenicity test systems. Methyl carabazate gave negative results in three and equivocal results in one test system. The mutagenic potential of QCA was tested in only two assays, for reverse mutation in bacteria and for chromosomal aberrations in human lymphocytes, with negative results. The results of the mutagenicity tests indicate that carbadox, desoxycarbadox and hydrazine have genotoxic potential, while the genotoxic potency of desoxycarbadox appears to be lower than that of the parent substance carbadox.

The Committee also previously reviewed several long-term studies in which carbadox and its metabolites were administered in the diet to rodents. The main findings are summarized in Table 2.

Table 2. Tumour incidences observed in long-term studies of toxicity and carcinogenicity with carbadox and its genotoxic metabolites evaluated by the Committee at its thirty-sixth meeting (Annex 1, reference 92)

In rats, carbadox was reported to cause a dose-dependent increase in the incidence of benign and malignant hepatic tumours and of mammary tumours at doses > 1 mg/kg bw per day. Doses > 25 mg/kg bw per day were toxic, and administration could not be continued. The results of all the studies demonstrated the carcinogenic potential of carbadox, even thought there were relatively few animals per dose. Studies in monkeys and studies of ‘relay toxicity’ (in which laboratory animals are fed meat from farm animals treated with the agent) in rats and dogs were not adequate to assess carcinogenicity. In two studies with long-term administration of desoxycarbadox to rats, increased incidences of tumours were recorded at all doses. Desoxycarbadox was not only a potent hepatocarcinogen but increased tumour incidence in a dose-related manner at other sites, including the mammary gland and skin. Hydrazine was shown to be carcinogenic in mice and rats but not in hamsters. No treatment-related increase in tumour incidence was recorded in rats treated with methyl carbazate at doses ­ 10 mg/kg bw per day for 21 or 23 months or in rats or mice given a diet containing QCA at doses ­ 100 mg/kg bw per day.

The results of the studies with methyl carbazate and the persistent end-metabolite QCA thus provided no evidence of genotoxic or carcinogenic potential. The NOEL of QCA was 50 mg/kg bw per day. The Committee concluded that carbadox, desoxycarbadox and hydrazine are genotoxic carcinogens. The Committee also noted that the tumorigenic potential of desoxycarbadox was apparently greater than that of the parent compound and that it therefore probably makes a significant contribution to the tumorigenic activity of carbadox in rats. The results of studies of the metabolism of carbadox suggested, however, that desoxycarbadox is a relatively short-lived intermediate between carbadox and QCA. No NOEL could be identified for carbadox, desoxycarbadox or hydrazine in long-term studies in rodents treated in the diet. Because of the genotoxic and carcinogenic nature of carbadox, desoxycarbadox and its possible metabolite hydrazine, the Committee was not able to establish an ADI or to determine a safe concentration of total residues in the absence of any threshold.

An evaluation of carbadox residues was completed in 1998 which resulted in codification of a revised tolerance for residues of carbadox and its metabolites of carcinogenic concern in edible tissues and their risk to the consumer (Food and Drug Administration, 1998). That document referred to the review of the Committee at its thirty-sixth meeting and stated that the end-points of toxicological concern for carbadox and its metabolites were their genotoxicity and carcinogenicity. The studies of genotoxicity, the long-term studies of toxicity and carcinogenicity summarized in Table 2 and a study of toxicity in monkeys given repeated doses were reviewed. From the results of these studies, ‘no residue’ levels were determined from S₀ values for the carcinogenic compounds carbadox, desoxycarbadox and hydrazine. The S₀ values were transformed to S_m values to correct for human food intake. ‘No residue’ of a compound is considered to remain in edible tissue when the residue of carcinogenic concern in the total diet does not exceed the S₀. The S₀ is defined as the concentration of total residue of carcinogenic concern of the test compound in the total diet of test animals that corresponds to a maximum lifetime risk of cancer in the test animals of 1 in 1 million. The S₀ values were calculated for carbadox and each of its carcinogenic metabolites from a low-dose linear statistical model, as 106 ng/kg for carbadox, 61 ng/kg for desoxycarbadox and 11 ppb (µg/kg) for hydrazine. The lowest value of 61 ng/kg obtained for desoxycarbadox was designated as the S₀ of carbadox residues. The S_m is the permitted concentration of residues of carcinogenic concern in a specific edible product (Food and Drug Administration, 1998).

The S₀ values were calculated with 99% confidence limits by use of the one-hit linear extrapolation model (for review, see Calabrese, 1983) on the basis of hepatic tumours in male and female rats combined. The result obtained with the model was more conservative than that found with the Mantel Bryan extrapolation (for review, see Calabrese, 1983), anticipating an S₀ of 680 ng/kg with 99% confidence limits and a 10^–6 risk for desoxycarbadox (L. Friedlander, personal communication, 2003). As further details of the assignment of S₀ values (e.g. studies included) were not available, the results of the extrapolations could not be confirmed independently.

Because the total human diet is not derived solely from food-producing animals, a correction for food intake is made to determine the concentrations of residues of carcinogenic concern that should be permitted in edible animal tissue. Given a total daily diet of 1500 g for humans, up to 500 g are assumed to be due to the consumption of meat, with 300 g comprised of muscle, 100 g comprised of liver, 50 g comprised of kidney and 50 g comprised of fat. Thus, for each edible tissue the S_m is calculated on the basis of the lowest S₀ for desoxycarbadox, as shown in Table 3.

Table 3. Consumption figures and calculated S_m values for total carbadox-derived residues of carcinogenic concern in edible pig tissues

S₀, concentration of total residue of carcinogenic concern of the test compound in the total diet of test animals that corresponds to a maximum lifetime risk of cancer in the test animals of 1 in 1 million; S_m, permitted concentration of residues of carcinogenic concern in a specific edible product (Food and Drug Administration, 1998)

At the time of the review, no regulatory analytical method was available to monitor carbadox residues of carcinogenic concern at the level of the S₀ and S_m in muscle and liver of pigs. The results of a study of carbadox residues in porcine tissue subjected to digestion with either pepsin or pancreatin and with a more sensitive detection method indicated the presence of desoxycarbadox at concentrations that substantially exceeded the S_m value in liver throughout the experimental period of 15 days after withdrawal. When the concentration of the marker residue, QCA, had fallen below the MRL of 30 µg/kg, the concentration of residues of desoxycarbadox in liver still exceeded the established S_m by a factor of 4–5 (Heird, 2002/2003). Therefore the MRL of 30 µg/kg for liver recommended by the Committee at its thirty-sixth meeting is inadequate to monitor the absence of residues of carbadox of carcinogenic concern.

The Committee has not to date used the approach of extrapolating irreversible effects of chemicals at low doses for which thresholds have not been identified to a ‘virtually safe dose’ by using mathematical models, such as the one-hit linear model or the Mantel Bryan model, in evaluating residues of veterinary drugs in food. The models used for extrapolating to a ‘virtually safe dose’ have been criticized on several bases, such as lack of validity, the absence of consistent low-dose linearity in experimental systems, arbitrary selection of the slope in the Mantel Bryan probit analysis, the considerable diversity of predictions in the very low dose range with variations of several orders of magnitude, insufficient recognition of the effects of pharmacokinetics, and lack of consideration of individual variation in susceptibility within the human population. Although such estimates of risk have substantial inherent uncertainties, the one-hit low-dose extrapolation model is assumed to be a conservative approach, as it is least likely to result in underestimation of carcinogenic risks in humans and has been reported to show the closest fit to human response rates derived from epidemiological studies (Calabrese, 1983).

In a study of developmental toxicity, carbadox was administered orally to pregnant rats at a dose of 0, 10, 25, 50 or 100 mg/kg bw per day on days 8–15 of gestation. None of the treated dams died. Maternal body-weight gain was significantly decreased in a dose-related manner at all doses; weight gain recovered after cessation of treatment, except in animals at the highest dose. The animals were killed on day 21 of gestation, and their fetuses were removed surgically. Carbadox was embryotoxic and fetotoxic, as indicated by a dose-related reduction in fetal body weight, which was statistically significant at doses of 25 mg/kg bw per day and higher. The number of live fetuses and the resorption rate in dams at doses of 50 mg/kg bw per day or less were not different from those of the control group. Embryolethality occurred at the highest dose, the percentage of late resorptions being 82% and the number of live pups being reduced by more than 80% when compared with controls. At the highest dose, carbadox induced external, skeletal and internal malformations at rates of 47%, 45% and 28%, respectively. The abnormalities recorded were short tail, kinky tail, brachygnathia, ectrodactyly, club foot, generalized oedema, fused vertebrae and hydrocephaly. Carbadox was considered to be embryotoxic and fetotoxic as a consequence of its strong maternal toxicity and to be teratogenic in rats. The NOEL was 10 mg/kg bw per day for embryotoxicity and 50 mg/kg bw per day for teratogenicity. An NOEL for maternal toxicity could not be identified. The Committee noted that the NOELs for developmental toxicity in this study were well above the NOEL of 2.5 mg/kg bw per day found in the studies of reproductive toxicity in rats previously evaluated by the Committee.

No new experimental data on the metabolism of carbadox or on the genotoxicity or carcinogenicity of carbadox and its metabolites were provided. The Committee was aware of an evaluation based on linear extrapolation to estimate a ‘virtually safe dose’ for carbadox and its metabolites of carcinogenic concern. This estimate was made on the basis of the incidence of hepatic tumours in rats in long-term studies of carcinogenicity previously evaluated by the Committee. The evaluation resulted in codification of tolerance for residues of carbadox and the metabolites of carcinogenic concern in edible tissues and their risks to the consumer. The Committee did not consider this approach appropriate owing to the substantial inherent uncertainties involved.

The results of a recent study of residues of carbadox in pigs indicated longer persistence of substantial amounts of the carcinogenic metabolite desoxycarbadox in liver at the time when the concentration of the marker residue, QCA, fell below the MRL of 0.03 mg/kg and to the end of the experiment 15 days after withdrawal.

Carbadox was reviewed by the present Committee primarily on the basis of new information on residue levels, which indicated that the metabolite desoxycarbadox was present in edible tissues even at the end of the 15-day experimental withdrawal period. The Committee confirmed that the information previously submitted indicated that both carbadox and desoxycarbadox should be regarded as carcinogens that act by a genotoxic mechanism. Although the Committee was aware that linear extrapolation has been used to estimate a ‘virtually safe dose’ of carbadox, the Committee concluded that it was not possible to identify a dose of carbadox that poses an acceptable risk to consumers. The Committee therefore did not establish an ADI for carbadox.

Calabrese, E.J., ed. (1983) Principles of Animal Extrapolation, New York: John Wiley & Sons, pp. 555–574.

Food and Drug Administration (1998b) Mecadox® 10 type A medicated article (carbadox). Freedom of information summary, supplement to NADA No. 041-061, Washington DC.

Heird, C.E. (2002/2003) Concentration and depletion of carbadox, deoxycarbadox and quinoxaline-2-carboxylic acid (QCA) in tissue and rendered material of growing swine after consumption of carbadox at 50 g/ton of feed. Unpublished report from Southwest Bio-Labs, Inc. Submitted to WHO by Phibro Animal Health, Fairfield, New Jersey, USA.

MacIntosh, A., Lauriault, G. & Neville, G.A. (1985) Liquid chromatographic monitoring of the depletion of carbadox and its metabolite desoxycarbadox in swine tissues. J. Assoc. Off. Anal. Chem., 68, 665–671.

Vilim, A. & Lambert, G. (2001) Health risk assessment: Carbadox in swine. Human Safety Division, Bureau of Veterinary Drugs, Health Products and Food Branch, Ottawa: Health Canada.

Yoshimura, H. (2002) Teratogenic assessment of carbadox in rats. Toxicol. Lett., 129, 115–118.

    See Also:
       Toxicological Abbreviations
       Carbadox (ICSC)
       Carbadox (WHO Food Additives Series 27)
       CARBADOX (JECFA Evaluation)