INTERNATIONAL PROGRAMME ON CHEMICAL SAFETY WORLD HEALTH ORGANIZATION Toxicological evaluation of certain veterinary drug residues in food WHO FOOD ADDITIVES SERIES 39 Prepared by: The forty-eighth meeting of the Joint FAO/WHO Expert Committee on Food Additives (JECFA) World Health Organization, Geneva 1997 FLUMEQUINE (addendum) First draft prepared by Professor F.R. Ungemach Institute of Pharmacology, Pharmacy and Toxicology Veterinary Faculty, University of Leipzig, Leipzig, Germany 1. Explanation 2. Biological data 2.1 Toxicological studies 2.1.1 Short-term toxicity 2.1.1.1 Arthropathy in dogs 2.1.1.2 Hepatotoxicity in mice 2.1.1.3 Mechanism of tumorigenicity in mice 2.1.2 Special studies on human intestinal flora 3. Comments 4. Evaluation 5. References 1. EXPLANATION Flumequine is a fluoroquinolone antimicrobial agent. This substance was evaluated by the Committee at its forty-second meeting (Annex 1, reference 110). At that time, an ADI could not be established owing to a lack of adequate information on the toxicological and microbiological hazards of flumequine: (i) necessary to identify a NOEL for hepatotoxicity; (ii) on the mechanism of tumorigenesis; (iii) on the possible induction of arthropathy; and (iv) on the microbiological safety of residues. The additional information that was provided on these issues is summarized in this monograph addendum. 2. BIOLOGICAL DATA 2.1. Toxicological data 2.1.1 Short-term toxicity 2.1.1.1 Arthropathy in dogs Erosive arthropathy is a characteristic toxic reaction to quinolones and has been observed in growing animals and in particular young dogs after moderate doses of various fluoroquinolones (Gough et al., 1979; Brown, 1996). Transient joint damage has also been reported in children (Norrby & Lietman, 1993). An early three-month study of the toxicity of flumequine in young adult beagle dogs did not address arthropathic lesions of weight-bearing joints in particular (Nelson et al., 1972). Flumequine was administered twice daily as tablets by gavage to groups of 10 three-month-old beagle dogs of each sex at doses of 0, 15, 30, 60, or 150 mg/kg bw per day for 13 consecutive weeks. Four animals of each group were killed after three weeks. The absorption of the test compound was checked periodically by high-performance liquid chromatography analysis for flumequine and its 7-hydroxy metabolite in plasma. The animals were observed daily for clinical signs, with special attention to lameness and locomotor activity. Body weight was recorded weekly. The serum activity of alkaline phosphatase was measured on three occasions during the study. Twenty animals were killed after three weeks of treatment, and the remainder were killed at 13 weeks. All animals were submitted to necropsy and checked for gross abnormalities and alterations of the articular surfaces of weight-bearing joints of the fore- and hindlimbs; the shoulder and hip joints were submitted to histopathological examination. The study was conducted according to good laboratory practice (GLP) guidelines. No deaths and only a few signs of adverse reactions were observed, including vomiting and reduced food consumption which increased in frequency with dose. No clinical signs of arthropathy, such as staggering gait and reduced locomotor activity, were reported. Females at the higher doses had markedly reduced weight gain. The serum activity of alkaline phosphatase remained unchanged. Gross necropsy revealed erosions of the joint surfaces in two of 10 dogs at the highest dose and in one of 10 females at the lowest dose. Slight histopathological lesions of the articular cartilage were observed in one of 10 dogs given 60 mg/kg bw per day and three of 10 dogs given 150 mg/kg bw per day. These lesions were characterized by erosions, cavities, a fibrillary appearance of the cartilage, and synovial hyperplasia. The severity of the arthropathic lesions was similar at three and 13 weeks. The gross lesions of the hip joints of one animal after 13 weeks at the lowest dose were not accompanied by histopathological alterations (Woehrle, 1996). The grossly observed erosions in the hip joints of one female at the lowest dose were considered to be spontaneous, since no related histopathological alterations were found and no gross lesions were observed at the next two higher doses. The NOEL for induction by flumequine of arthropathy in young dogs was thus 30 mg/kg of bw per day. 2.1.1.2 Hepatotoxicity in mice In short-term and long-term studies of toxicity evaluated previously by the Committee (Annex 1, reference 110), oral administration of flumequine caused dose-related hepatotoxic effects in rats and mice. Hypertrophy, degenerative changes, and focal necrosis of hepatocytes were observed in rats at 800 mg/kg bw per day in a three-month study (Nelson et al., 1972) and at 400 and 800 mg/kg bw per day in a two-year study (Sibinski et al., 1977a) and in CD-1/ICR mice at 400 and 800 mg/kg bw per day in an 18-month study (Sibinski et al., 1977b, 1979). The prevalence of hepatotoxic lesions increased with the duration of treatment. After cessation of flumequine administration, the liver damage was reversed (Sibinski et al., 1979). Male mice were the most sensitive to flumequine-induced liver damage (Sibinski et al., 1977b, 1979). There was no NOEL for the hepatotoxic effects of flumequine in CD-1 mice. In a 13-week study designed to investigate the hepatotoxic lesions and the activity of hepatic drug-metabolizing enzymes, flumequine was administered orally to groups of 16 CD-1 mice of the same strain as used in the previous studies. The animals were fed diets containing flumequine at concentrations providing doses of 0, 25, 50, 100, 400, or 800 mg/kg bw per day to males and 0, 100, 400, or 800 mg/kg bw per day to females. The concentration of the test compound in the diet was checked periodically. The animals were observed daily for clinical signs, and body weight, while food consumption and food conversion efficiency were recorded weekly. Plasma enzyme activities were measured once, after 12 weeks. On the last day of treatment, the test compound was shown to be absorbed by high-performance liquid chromatography analysis of flumequine and its 7-hydroxy metabolite in plasma. At the end of treatment, all animals were necropsied, and complete macroscopic examinations were conducted. The weights of the livers were recorded, and samples of the liver and other grossly abnormal tissues were submitted to histopathology. At the end of the experiment, liver microsomes were prepared to test the activity of the xenobiotic-metabolizing enzyme system by measuring total protein and cytochrome P450 content, P450-dependent dealkylation of resorufin and coumarin derivatives, and 1-naphthol glucuronidation. Microsomes from Aroclor-treated animals were used as positive controls. The study was performed in compliance with GLP. No deaths and no clinical signs of treatment-related adverse reactions were observed. The body-weight gains of the animals at the highest dose were reduced during the first week. This effect was more pronounced in males, which also showed slightly lower food consumption and efficiency of food conversion. No differences from the controls were seen in the other groups during the study. Plasma enzyme activities indicated liver damage at high doses, with a significant increase in the activities of alanine aminotransferase and alkaline phosphatase at doses of 400 and 800 mg/kg bw per day flumequine and of lactic dehydrogenase and aspartate aminotransferase at 800 mg/kg bw per day. Liver weights were increased at the two highest doses. Histopathological examination of the livers revealed dose-dependent degenerative alterations of hepatocytes, with hypertrophy and fatty vacuolation (in males at doses greater than 25 mg/kg bw per day and in females at doses greater than 100 mg/kg bw per day), increased ploidy, intranuclear inclusions, and centrilobular necrosis (at doses greater than 100 mg/kg bw per day). The effects were more pronounced in male animals. Increased mitotic activity was observed in males at the highest dose. Flumequine at doses up to 800 mg/kg bw per day had little or no effect on P450-dependent hepatic drug-metabolizing enzymes or on glucuronyltransferase activity. It was concluded that (i) flumequine has no remarkable inducing or inhibitory effect on the hepatic P450-dependent xenobiotic-metabolizing enzyme system or on glucuronidation, and (ii) the liver is the target organ of flumequine in mice. The degeneration of hepatocytes with focal necrosis, accompanied by increased mitotic activity indicating regenerative processes, was seen only in male mice at the highest dose. The slight hypertrophic alterations of liver cells with minimal degenerative alterations in males at doses of 50 and 100 mg/kg bw per day were regarded as signs of hepatotoxic lesions rather than metabolic overload. The NOEL was thus 25 mg/kg bw per day on the basis of hepatotoxic lesions in males (Stewart, 1995). 2.1.1.3 Mechanism of tumorigenicity in mice The results of long-term studies with rodents previously evaluated by the Committee (Annex 1, reference 110) showed no carcinogenic effects in rats (Sibinski et al., 1977a), but a dose-related increase in the incidence of benign and malignant liver tumours was observed in CD-1 mice at doses greater than 100 mg/kg bw per day. The tumour incidence was parallelled by hepatotoxic changes and was significantly higher in male mice, which are known to be sensitive to liver tumour induction (Sibinski et al., 1977b, 1979; McClain, 1990). As the compound was inactive in a range of tests for genotoxicity, including assays for gene mutation in bacteria and mammalian cells in vitro and for chromosomal aberrations in mammalian cells in vivo, the mechanism of tumorigenicity was unclear. The available toxicological database on flumequine and data from the open literature were reviewed in order to discern the genotoxic or non-genotoxic ('epigenetic') mechanism of the hepatocarcinogenicity of flumequine (Marzin, 1996). By definition, a genotoxic carcinogen acts directly on DNA in the target tissues, inducing DNA damage, strand breaks, or mutations, which can be assessed in vitro in assays for genotoxicity and in short-term assays in rodents. A non-genotoxic carcinogen is devoid of such activity. The neoplastic response to a non-mutagenic carcinogen is characterized by a steep dose-response curve and a threshold dose. Non-genotoxic carcinogenicity is considered to be brought about by potent induction of cytotoxicity and cell proliferation, which may increase the frequency of tumours in the target organs by virtue of sustained mitogenic stimuli. Demonstration of a lack of genotoxic potential and of the induction of a dose-related increase in cell proliferation thus remains critical for identifying non-genotoxic carcinogens (Faccini et al., 1992; Purchase, 1994; Shaw & Jones, 1994). As negative results were obtained in various assays for genotoxicity, including reverse mutation in Salmonella typhimurium, gene mutation at the hprt locus in mouse lymphoma cells, gene mutation in Chinese hamster cells, and chromosomal aberrations in bone-marrow cells of rats in vivo, it is unlikely that flumequine or its metabolites have direct genotoxic or mutagenic activity. Flumequine, like other 4-quinolones, exerts its antibacterial effects at the level of DNA by inhibiting bacterial topoisomerase II (DNA gyrase) (Hussy et al., 1986; Sato et al., 1989). Although bacterial and eukaryotic topoisomerases II share some structural homology, they differ markedly in structure and function, which may explain the very different sensitivities to the inhibitory activity of 4-quinolones (Liu & Wang, 1991). Various 4-quinolones, including fluoroquinolones, have an affinity for the mammalian enzyme that is several orders of magnitude lower than that for the bacterial gyrase. Thus, the median inhibitory dose (ID50) for calf thymus topoisomerase II exceeds the ID50 for gyrase of Escherichia coli by at least 100 to 2460-fold (in the case of loxacin) (Hussy et al., 1986; Sato et al., 1989). This low affinity is assumed to be a common feature of all fluoroquinolones. Although the inhibition of mammalian topoisomerase II by flumequine was not investigated, it is unlikely to exist, even at therapeutic doses. This conclusion is supported by the lack of mutagenic potential of flumequine, whereas specific inhibitors of the eukaryotic topoisomerase II are genotoxic and induce frameshift-type mutations (Huff & Kreuzer, 1991). Non-genotoxic tumorigenesis in the rodent liver can occur by various mechanisms, including compound-related hormonal activity, peroxisomal proliferation, induction of hepatic drug-metabolizing enzymes, and hepatotoxicity. The toxicological database, including the results of studies of reproductive toxicity, revealed no evidence for any hormonal activity of flumequine; and in short- and long-term assays for toxicity in mice, no histopathological alterations of liver cells indicating peroxisomal proliferation were reported. Various inducers of the hepatic P450-dependent xenobiotic biotransformation system, such as phenobarbital and halogenated cyclic hydrocarbons, act as tumour promoters in rodent liver when administered at high doses over a prolonged period (Diwan et al., 1990; McClain, 1990; Grasso & Hinton, 1991). Since flumequine had only negligible effects on the hepatic P450 enzyme system in the 13-week study of toxicity in CD-1 mice at doses up to 800 mg/kg bw per day (Stewart, 1995), induction of the hepatic xenobiotic-metabolizing enzyme system can be excluded as a non-genotoxic mechanism of the hepatocarcinogenicity of flumequine. Flumequine is hepatotoxic, causing hepatocellular degeneration and focal necrosis in male and female mice, followed by a mitogenic response in male mice at the highest dose (Stewart, 1995). The dose-related severity of these hepatotoxic lesions parallelled the incidence of benign and malignant liver tumours (Sibinski et al., 1977b). Various non-genotoxic hepatotoxins have been shown to induce liver tumours in rodents (Drinkwater et al., 1990; Butterworth & Goldsworthy, 1991; Grasso & Hinton, 1991). High doses and prolonged exposure increase the frequency of mutations and the likelihood of neoplastic transformation at the cellular level by still hypothetical mechanisms, such as expression of protooncogenes and growth factors (Thompson et al., 1986; Dubois, 1990). In rodents, the mechanism is believed to be increased liver-cell proliferation due to repeated hepatocellular necrosis-regeneration cycles, leading to the development of foci of phenotypically altered hepatocytes (so-called 'preneoplastic lesions'), which finally progress to neoplasms (Pitot et al., 1990; Butterworth & Goldsworthy, 1991; Grasso & Hinton, 1991). The occurrence of foci of altered hepatocytes is an important link in the cascade of hepatotoxicity-induced liver tumorigenesis. Although the different types of preneoplastic lesions can readily be detected by conventional haematoxylin and eosin staining of liver tissue sections (Bannasch & Zerban, 1994), as was done in the studies of the toxicity of flumequine, no such lesions were reported in previously evaluated studies, including the 18-month study of carcinogenicity in CD-1 mice (Sibinski et al., 1977b; Annex 1, reference 111). In the 13-week study in CD-1 mice, clear-cell foci of altered hepatocytes, which are one type of preneoplastic lesion, were observed in one male at 400 mg/kg bw per day and in one male and one female at 800 mg/kg bw per day (Stewart, 1995). The lesions were not characterized histochemically. In a special study to assess marker enzymes of hepatic preneoplastic lesions, the effect of flumequine on the activity of gamma-glutamyltransferase and glutathione S-transferase, which in its placental form is a characteristic marker of foci of phenotypically altered hepatocytes (Bannasch & Zerban, 1994), was investigated in homogenates of livers from some of the CD-1 mice used in the 13-week study. The activity of gamma-glutamyltransferase remained unchanged. Administration of flumequine at doses of 400 and 800 mg/kg bw per day resulted in marked stimulation of glutathione S-transferase activity towards 1-chloro-2,4-dinitrobenzene in females, whereas the enzyme activity in males was only slightly affected. Doses up to 100 mg/kg bw per day had no effect (Stewart, 1996). Because of its inadequate design, this study did not allow a valid assessment of marker enzymes of preneo-plastic lesions, which are confined to a small number of phenotypically altered hepatocytes. Therefore, this study was not considered further in the evaluation. In conclusion, flumequine was considered to be a non-genotoxic hepatocarcinogen, and the induction of hepatocellular necrosis-regeneration cycles by hepatotoxicity was considered to be the relevant mechanism for induction of liver tumours. Since cytotoxic effects are a prerequiste of hepatocarcinogenicity, tumours are induced only at hepatotoxic doses (Cohen & Ellwein, 1990; Pitot et al., 1990). Therefore, the NOEL for the hepatotoxicity of flumequine, 25 mg/kg bw per day, was considered to be the threshold for both the hepatotoxicity and the associated carcinogenicity of flumequine. In evaluating the safety of flumequine, it must be kept in mind that the NOEL for hepatoxic lesions was derived from a short-term (13 weeks) study and was extrapolated to the level required for tumour formation observed at the end of a lifetime study (18 months) in mice. 2.1.2 Special studies on human intestinal flora No experimental data on the effect of flumequine on the bacteria of the human gut microflora were available at the time of the previous evaluation (Annex 1, reference 110). Additional studies have been provided to assess the effects of flumequine and its 7-hydroxy metabolite on bacterial isolates from human intestinal microflora in vitro. Studies in vivo have not been performed. In the first study, the minimum concentrations resulting in 50% inhibition (MIC50) and 90% inhibition (MIC90) and the geometric mean of the MIC50 were determined for 100 bacterial strains isolated from the faeces of healthy volunteers, comprising 10 isolates of 10 aerobic and anaerobic bacterial genera typical of the human gut microflora. The tests were performed in agar with serial dilutions under anaerobic and aerobic (Escherichia coli) conditions. Three bacterial tester strains were tested for reference. The inoculum density was 107 colony forming units (cfu) per ml. The GLP status of the study was not reported, but the protocol and conduct met accepted standards for such studies. The results of the MIC determinations are summarized in Table 1. E. coli was the most sensitive species, the mean MIC50 value for the 10 strains tested being 0.33 µg/ml. The mean MIC50 values for the most sensitive predominant species isolated from the human gastrointestinal tract, Clostridium and Fusobacterium, were 0.95 and 1.0 µg/ml, respectively (Richez, 1994a). In a second set of experiments under similar experimental conditions, the MIC values of 7-hydroxyflumequine were determined against 10 strains each of E. coli, Clostridium spp., and Fusobacterium spp. E. coli was much less sensitive to the metabolite than to the parent substance, with no inhibition at a concentration of 2 µg/ml 7-hydroxyflumequine and 100% inhibition at 4 µg/ml. The strains of Clostridium and Fusobacterium spp. were not sensitive to the highest concentration tested (16 µg/ml) (Richez, 1995). Table 1. Susceptibility of human intestinal bacteria to flumequine in vitro Bacterial species MIC (µg/ml) (10 strains each) Range MIC50 MIC90 Geometric mean E. coli (aerobic) 0.25-0.50 0.33 0.48 0.47 E. coli (anaerobic) 0.25-0.50 0.33 0.48 0.47 Streptococcus spp. 16- > 32 > 32 > 32 > 32 Proteus spp. > 32 > 32 > 32 > 32 Lactobacillus spp. > 32 > 32 > 32 > 32 Bifidobacterium spp. > 32 > 32 > 32 > 32 Bacteroides fragilis 16- > 32 > 32 > 32 > 32 Eubacterium spp. > 32 > 32 > 32 > 32 Clostridium spp. 0.50-2.0 0.95 1.70 1.32 Fusobacterium spp. 1.0-32 1.0 5.1 3.25 Peptostreptococcus spp. > 32 > 32 > 32 > 32 In a further experiment, the influence of inoculum size on the MIC was investigated for 10 strains of E. coli isolated from human faeces. Each strain was tested under anaerobic and aerobic conditions with an inoculum of either 107 or 109 cfu/ml. No effect of inoculum density was seen (Richez, 1994b). In a study reported in the open literature, the effects of gastrointestinal factors and pH on the MIC50 of flumequine were studied with relevant bacterial species. The tests were performed by the broth dilution technique in the presence of cooked meat or a combination of meat and milk, at pH values of 3-7.5. The obligate anaerobes remained insensitive to flumequine (MIC50 > 40 µg/ml), whereas the MIC50 values for the E. coli strains were increased by two- to eightfold (Nouws et al., 1994). The intestinal bioavailability of flumequine to enteric bacteria was tested by giving 830 mg 14C-flumequine orally to five healthy volunteers. The levels of radiolabel were then monitored in plasma, urine, and faeces for up to five days. A total of 84% (76-92%) radiolabel was recovered in excreta, with 9% (5.7-13%) in faeces and 75% (70-81%) in urine. It was concluded that about 10% of a dose of flumequine is available to the gut microflora (Riker Laboratories, Inc., 1994). No studies were submitted on the selection of intestinal bacterial resistance or on the inhibitory effects on microorganisms used in industrial processing of foodstuffs of animal origin. It was concluded that E. coli is the most sensitive of the relevant bacterial strains of human gut microflora tested in vitro. The absence of an effect of inoculum size indicates that flumequine has similar reactivity even at the high bacterial density in the human colon. The metabolite 7-hydroxyflumequine has markedly less antibacterial activity and can be considered to exert no relevant adverse effects on the human intestinal microflora. About 10% of ingested flumequine is available to the microflora in the human gut. The thirty-eighth Committee concluded that the most relevant parameter in vitro for assessing the risk to human intestinal flora is the geometric mean MIC against the most sensitive intestinal microorganism (Annex 1, reference 97). The MIC50 of flumequine for E. coli of 0.33 µg/ml should thus be considered the concentration that has no effect on human intestinal microflora and be used to establish the ADI. E. coli, which is very sensitive to fluoroquinolones in general, is, however, a minor component of the gastrointestinal flora (Moore & Moore, 1995). It is therefore more appropriate to consider the effects of flumequine on the most sensitive obligate anaerobes, which are the bacterial species that predominate in the human gut (Moore & Moore, 1995). The MIC50 values for Clostridium and Fusobacterium spp., 0.95-1.0 µg/ml, were considered to be the concentrations with no effect on human intestinal microflora. Flumequine is a fluoroquinolone and thus has a broad spectrum of activity against aerobic gram-negative bacteria. In humans, this class of antimicrobial agents is used clinically for selective elimination of potential aerobic and facultative anaerobic pathogens from the gastrointestinal tract while preserving the predominant anaerobic bacterial gut flora. Furthermore the administration of therapeutic oral doses of fluoroquinolones such as ciprofloxacin and norfloxacin to humans has not been shown to alter the intestinal bacterial ecology or to weaken the barrier effect. Anaerobic bacteria such as Bifidobacterium, Bacteroides, Eubacterium, Fusobacterium, and Peptostreptococcus spp., the main components of the flora in the human gastrointestinal tract, are largely unaffected by these compounds (Midtvedt, 1990; Nord, 1995). When assessing the effects of flumequine on human gastrointestinal flora, it is important to interpret the MIC50 values for selected intestinal bacterial species in the context of the overall gut ecosystem. Since the obligate anaerobic bacteria that are predominantly isolated from the gastrointestinal tract are relatively insensitive to fluoroquinolones, disturbance of the human gut ecosystem by residues of flumequine is unlikely. 3. COMMENTS The Committee considered additional information on the induction of arthropathy in young dogs, the hepatotoxic and liver enzyme-inducing effects of flumequine in mice, the possible mechanism of the hepatocarcinogenicity of flumequine, and its effect on human gut microflora. The studies were carried out according to appropriate standards for study protocol and conduct. In order to test the effects of flumequine on articular cartilage, it was administered as tablets to groups of 10 three-month-old beagle dogs at doses of 0, 15, 30, 60, or 150 mg/kg bw per day for 13 weeks; four animals from each group were killed after three weeks. The animals showed no overt clinical signs of arthropathy. Gross necropsy revealed erosions of the joint surfaces in two of 10 dogs at the highest dose and in one of 10 animals at the lowest dose. Histopathological lesions of the articular cartilage were observed in one of 10 dogs given 60 mg/kg bw per day and three of 10 dogs given 150 mg/kg bw per day. The severity of the lesions was similar at three and 13 weeks. The Committee considered that the gross lesions in the one animal at the lowest dose were not compound-related, since no histopathological alterations were found and no gross lesions were observed at the next two higher doses. Therefore, the NOEL for induction of arthropathy in young dogs was 30 mg/kg bw per day. In a 13-week study designed to investigate hepatotoxic lesions and the activities of hepatic drug-metabolizing enzymes, flumequine was administered to male CD-1 mice in the feed at doses equal to 0, 25, 50, 100, 400, or 800 mg/kg bw per day and to females at 0, 100, 400, or 800 mg/kg bw per day. The effects observed were reduced body weight, significantly increased plasma activities of alanine and aspartate aminotransferases, alkaline phosphatase and lactic dehydrogenase, and increased liver weights at 400 and 800 mg/kg bw per day. Histopathological examination of the liver revealed dose-dependent hypertrophy, degenerative alterations, and centrilobular hepatocellular necrosis. The hepatotoxic lesions were more pronounced in male than in female mice and were observed at all doses greater than 25 mg/kg bw per day. Increased mitosis was observed only in males at the highest dose. Flumequine caused little or no induction of hepatic cytochrome P450-dependent drug-metabolizing enzymes or glucuronyltransferase when given at doses up to 800 mg/kg bw per day. The NOEL was 25 mg/kg bw per day on the basis of induction of hepatotoxic lesions in male mice. At its forty-second meeting, the Committee noted that there was evidence of compound-related tumorigenic effects in the livers of CD-1 mice. The hepatotumorigenic activity of flumequine was more pronounced in male mice, which are known to be sensitive to liver tumour induction. As the compound was inactive in a range of tests for genotoxicity, including assays for gene mutation in bacteria and mammalian cells in vitro and for chromosomal aberrations in mammalian cells in vivo, the mechanism of this tumorigenesis was unclear. The present Committee noted that, although an inhibitory effect of flumequine on mammalian topoisomerase II, leading to DNA damage, was not investigated, bibliographical data on structurally related fluoroquinolones indicate that this mechanism is unlikely to operate. The Committee concluded that there is no evidence that flumequine has genotoxic potential. Non-genotoxic tumorigenesis in the liver can be due to various mechanisms, including compound-related hormonal activity, peroxisomal proliferation, induction of hepatic drug-metabolizing enzymes, and hepatotoxicity. The toxicological database, including studies of reproductive toxicity, revealed no evidence for any hormonal activity of flumequine. In short- and long-term studies of toxicity in mice, no histopathological alterations of liver cells that indicate peroxisomal proliferation were reported. Induction of the hepatic cytochrome P450 enzyme system can be excluded by the results of the 13-week study in CD-1 mice. Flumequine is hepatotoxic, causing hepatocellular degeneration and focal necrosis in male and female mice, which was followed by a mitogenic response in male mice at the highest dose in the 13-week study described above. The dose-related severity of these hepatotoxic lesions parallelled the incidence of benign and malignant liver tumours. Various non-genotoxic hepatotoxins have been shown to induce liver tumours. The mechanism is believed to be increased liver-cell proliferation due to repeated hepatocellular necrosis-regeneration cycles, leading to the development of foci of phenotypically altered hepatocytes (so-called 'preneoplastic lesions'), which finally progress to neoplasms. In the 13-week study in CD-1 mice, clear-cell foci of altered hepatocytes, which are one type of preneoplastic lesion, were observed in one male at 400 mg/kg bw per day and in one male and one female at 800 mg/kg bw per day. In a study to assess marker enzymes of hepatic preneoplastic lesions, the effects of flumequine on the activity of gamma-glutamyltransferase and glutathione S-transferase were investigated in homogenates of livers from some of the mice used in the 13-week study. Because of its inadequate design, including the lack of histochemical characterization of the foci of altered hepatocytes, this study was not considered further in the evaluation. The Committee considered that induction of hepatocellular necrosis-regeneration cycles by hepatotoxicity is the relevant mechanism for induction of liver tumours by flumequine. Therefore, the NOEL for the hepatotoxicity of flumequine, 25 mg/kg of bw per day, was considered to be the threshold for both the hepatotoxicity of flumequine and its associated carcinogenicity. The Committee noted that hepatotoxicity would have been better explored in a study of longer duration. The effect of flumequine on human intestinal microflora was assessed by determining the MIC50 values for 100 bacterial strains isolated from human faeces, comprising 10 isolates from 10 aerobic and anaerobic bacterial genera typical of the human gut microflora. These included Escherichia coli, Streptococcus spp., Proteus spp., Lactobacillus spp., Bifidobacterium spp., Bacteroides fragilis, Eubacterium spp., Clostridium spp., Fusobacterium spp., and Peptostreptococcus spp. The inoculum density was 107 colony forming units per ml. E. coli was the most sensitive bacterial species tested, with an MIC50 value of 0.33 µg/ml. The MIC50 value was not dependent on the size of the inoculum. E. coli was markedly less sensitive to 7-hydroxyflumequine, with an MIC50 value of 4 µg/ml. The mean MIC50 values for the most sensitive predominant species typically isolated from the human gastrointestinal tract, Fusobacterium and Clostridium, were 1.0 and 0.95 µg/ml, respectively. In a study of the influence of gastrointestinal factors and pH on the MIC50 values of flumequine for relevant bacterial species of the human gastrointestinal tract, the values for obligate anaerobes were unaffected, whereas the MIC50 values for E. coli strains were increased by two- to eightfold. The upper limit of the ADI based on the antimicrobial activity of flumequine on human gut flora was calculated on the basis of the formula described on p. 12 as follows: Upper limit 1 µg/ga × 220 g of ADI = 0.1b × 1c × 60 kg = 37 µg/kg bw a Mean MIC50 for the most sensitive predominant bacterial species, Fusobacterium and Clostridium b Fraction of oral dose available to act on microorganisms in the colon, based on a study in which 830 mg 14C-flumequine were given orally to five healthy volunteers. The levels of radiolabel were then monitored in plasma, urine, and faeces for up to five days. A total of 84% (78-92%) of the radiolabel was recovered in the excreta, with 9% (5.7-13%) in faeces and 75% (70-81%) in urine. The Committee concluded that approximately 10% of flumequine is available to the gut microflora. c A safety factor of 1 was used because relevant and sufficient microbiological data were provided. 4. EVALUATION The Committee noted that flumequine belongs to the group of antimicrobial fluoroquinolones that are active against aerobic gram-negative bacteria. In humans, this class of antimicrobial agents is used clinically for selective elimination of potential aerobic and facultative anaerobic pathogens from the gastrointestinal tract while preserving the predominant anaerobic bacterial gut flora. The Committee also recognized that administration of therapeutic oral doses of fluoroquinolones such as ciprofloxacin and norfloxacin to humans has no appreciable effect on the intestinal bacterial ecology or on the barrier effect. In addition, anaerobic bacteria such as Bifidobacterium, Bacteroides, Eubacterium, Fusobacterium, and Peptostreptococcus spp., the main components of the human gut flora, are largely unaffected by these compounds. E. coli, however, which is very sensitive to fluoroquinolones in general, is a minor component of the gastrointestinal flora. The Committee considered that, in assessing the effects of flumequine on the bacteria of the human gastrointestinal flora, the MIC50 values for the selected intestinal bacterial species should be interpreted in the context of the overall ecosystem of the gastrointestinal tract. Since the obligate anaerobic bacteria that predominate in the gastrointestinal tract are relatively insensitive to fluoroquinolones, disturbance of the human gut ecosystem by residues of flumequine is unlikely. Therefore the Committee decided to base the ADI on the toxicological properties of flumequine and not on its effect on the intestinal microflora. The Committee considered the NOEL of 25 mg/kg bw per day for hepatotoxicity in male CD-1 mice in the 13-week study to be the most appropriate toxicological end-point for consumer safety. An ADI of 0-30 µg/kg bw was established by applying a 1000-fold safety factor, which was chosen to account for the short duration of the study and the lack of histochemical characterization of the foci of altered hepatocytes. 5. REFERENCES Bannasch, P. & Zerban, H. (1994) Preneoplastic and neoplastic lesions of the rat liver. In: Bannasch, P & Gœssner, W., eds, Pathology of Neoplasia and Preneoplasia in Rodents, Stuttgart, Schattauer, pp. 18-30. Brown, S.A. (1996) Fluoroquinolones in animal health. J. Vet. Pharmacol. Ther., 19, 1-14. Butterworth, B.E. & Goldsworthy, T.L. (1991) The role of cell proliferation in multistage carcinogenesis. Proc. Soc. Exp. Biol. Med., 198, 683-687. Cohen, S.M. & Ellwein, L.B. (1990) Cell proliferation in carcinogenesis. Science, 249, 1007-1011. Diwan, B.A., Rice, J.M. & Ward, J.M. (1990) Strain-dependent effects of phenobarbital on liver tumour promotion in inbred mice. In: Mouse Liver Carcinogenesis: Mechanisms and Species Comparisons, New York, Alan R. Liss, Inc., pp. 69-83. Drinkwater, N.R., Hanigan, M.H. & Kemp, C.J. (1990) Genetic and epigenetic promotion of murine hepatocarcinogenesis. In: Mouse Liver Carcinogenesis: Mechanisms and Species Comparisons, New York, Alan R. Liss, Inc., pp. 163-176 Dubois, R.N. (1990) Early changes in gene expression during liver regeneration. What do they mean? Hepatology, 11, 1079-1082. Faccini, J.M., Butler, W.R., Friedman, J.C., Hess, R., Reznik, G.K., Ito, N., Hayashi, Y. & Williams, G.M. (1992) IFSTP guidlines for the design and interpretation of the chronic rodent carcinogenicity bioassay. Exp. Toxicol. Pathol., 44, 443-456. Gough, A., Barsoum, N., Mitchell, L., McGuire, E. & de la Iglesias, F. (1979) Juvenile canine drug-induced arthropathy: Clinicopathological studies on articular lesions caused by oxolinic acid and pipemidic acid. Toxicol. Appl. Pharmacol., 51, 177-187. Grasso, R. & Hinton, R.H. (1991) Evidence for and possible mechanisms of non-genotoxic carcinogenicity in rodent liver. Mutat. Res., 248, 271-290 Huff, A.C. & Kreuzer, K.N. (1991) The mechanism of antitumour drug action in a simple bacteriophage model system. In: Potmesil, M. & Kohn, K., eds, DNA Topoisomerases in Cancer [publishers not given], pp. 215-229. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Hussy, P., Maass, G., Tummler, B., Grosse, F. & Schomburg, U. (1986) Effect of 4-quinolones and novobiocin on calf thymus DNA polymerase a primase complex, topoisomerase I and II, and growth of mammalian lymphoblasts. Antimicrob. Agents Chemother., 29, 1073-1078. Liu, L.F. & Wang, J.C. (1991) Biochemistry of DNA topoisomerases and their poisons. In: Potmesil, M. & Kohn, K., eds, DNA Topoisomerases in Cancer [publishers not given], pp. 13-22. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Marzin, D. (1996) Flumequine MRLs file. Unpublished report from Laboratoire de Toxicologie, Insitut Pasteur de Lille, France. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. McClain, R.M. (1990) Mouse liver tumours and microsomal enzyme-inducing drugs: Experimental and clinical perspectives with phenobarbital. In: Mouse Liver Carcinogenesis: Mechanisms and Species Comparisons, New York, Alan R. Liss, Inc., pp. 345-365. Midtvedt, T. (1990) The influence of quinolones on the faecal flora. Scand. J. Infect. Dis., 68 (Suppl.), 14-18. Moore, W.E.C. & Moore, L.H. (1995) Intestinal floras of populations that have high risk of colon cancer. Appl. Environ. Microbiol., 61, 3202-3207. Nelson, R.A., Case, M.T., Glick, P.R. & Steffen, G.R. (1972) Ninety (90) day oral subacute toxicity of flumequine in beagle dogs. Unpublished report from Riker Laboratories Inc., St Paul, MN, USA. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Nord, C.E. (1995) Effect of quinolones on the human intestinal microflora. Drugs, 49 (Suppl.), 81-85. Norrby, S.R. & Lietman, P.S. (1993) Safety and tolerability of fluoroquinolones. Drugs, 45 (Suppl. 3), 59-64. Nouws, J.F.M., Kuiper, H., van Klingeren, B. & Kruyswijk, P.G. (1994) Establishment of a microbiologically acceptable daily intake of antimicrobial drug residue. Vet. Q., 16, 152-156. Pitot, H.C., Dragan, Y., Xu, Y., Pyron, M., Laufer, C. & Rizvi, T. (1990) Role of altered hepatocyte foci in the stages of carcinogenesis. In: Mutation and the Environment, New York, Wiley- Liss, Inc., pp. 81-95. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Purchase, I.F.H. (1994) Current knowledge of mechanisms of carcinogenicity: Genotoxins versus non-genotoxins. Hum. Exp. Toxicol., 13, 17-28. Richez, P. (1994a) Antibacterial activity of flumequine against human gut microflora. Unpublished report No. SF008 from DataVet, Vendargues, France. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Richez, P. (1994b) Antibacterial activity of flumequine against Escherichia coli strains isolated from the human gut microflora: The effect of the inoculum size. Unpublished report No. SF008-2 from DataVet, Vendargues, France. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Richez, P. (1995) Antibacterial activity of 7-OH flumequine against human gut microflora. Unpublished report No. SF008-3 from DataVet, Vendargues, France. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Riker Laboratories, Inc. (1994) Metabolism of R802 (flumequine) in humans. Unpublished report form Riker Laboratories Inc., St Paul, MN, USA. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Sato, K., Hoshing, K., Une, T. & Osada, Y. (1989) Inhibitory effects of ofloxacin on DNA gyrase of Escherichia coli and topoisomerase II of bovine calf thymus. Rev. Infect. Dis., 2 (Suppl. 5), 5915-5916. Shaw, I.C. & Jones, H.B. (1994) Mechanisms of non-genotoxic carcinogenesis. TiPS, 15, 89-93. Sibinski, L.J., Steffen, G.R. & Case, M.T. (1977a) Two-year oral toxicity-carcinogenicity study of R-802 (flumequine) in rats. Unpublished report from Riker Laboratories, Inc., St Paul, MN, USA. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Sibinski, L.J., Steffen, G.R. & Case, M.T. (1977b) Eighteen (18) month oral carcinogenicity study of R-802 (flumequine) in mice. Unpublished report from Riker Laboratories, Inc., St Paul, MN, USA. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Sibinski, L.J., Steffen, G.R. & Case, M.T. (1979) Special 18-month toxicity study of R-802 (flumequine) in male mice. Unpublished report from Riker Laboratories, Inc., St Paul, MN, USA. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Stewart, J.S. (1995) Flumequine: Toxicity study by dietary administration to CD-1 mice for 13 weeks. Unpublished report No. 94/0678 from Pharmaco LSR, Eye, Suffolk, United Kingdom. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Stewart, J.S. (1996) Flumequine: Supplementary analysis to a toxicity study by dietary administration to CD-1 mice for 13 weeks. Unpublished report No. 96/0443 from Huntingdon Life Sciences Ltd, Eye, Suffolk, United Kingdom. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France. Thompson, N.L., Mead, J.E., Braun, L., Goyette, M., Shank, P.R. & Fausto, N. (1986) Sequential protooncogene expression during rat liver regeneration. Cancer Res., 46, 3111-3117. Woehrle, F. (1996) Flumiquil comprimés 200 mg -- 13 week oral (tablet) tolerance study in the young beagle dog with interim necropsy after 3 weeks. Study No. 311/536. Unpublished report from Pharmakon Europe, L'Arbresle, France. Submitted to WHO by Sanofi Santé Nutrition Animale, Libourne, France.
See Also: Toxicological Abbreviations Flumequine (JECFA Food Additives Series 51) Flumequine (WHO Food Additives Series 53) Flumequine (WHO Food Additives Series 33) FLUMEQUINE (JECFA Evaluation)