Prepared by:
          The 50th meeting of the Joint FAO/WHO Expert
          Committee on Food Additives (JECFA)

        World Health Organization, Geneva 1998


    First draft prepared by
    Dr H. Fernandez and Dr M. Miller
    US Food and Drug Administration
    Washington DC, USA

    1.   Explanation
    2.   Biological data
         2.1  Absorption, distribution, and excretion
         2.2  Effects on enzymes and other biochemical parameters of the
              intestinal flora
         2.3  Microbiological studies
              2.3.1   In vitro
             Minimal inhibitory concentration
             Continuous flow chemostat system inoculated
                              with tetracycline
             Effect of tetracyclines on colonization
             Selection of bacterial strains resistant to
              2.3.2  Animal models
         2.4  Studies in humans
    4.   Evaluation
    5.   References


         The tetracyclines oxytetracycline, chlortetracycline, and
    tetracycline are drugs with a long history of use. These products were
    first evaluated as  a group at the twelfth meeting of the Committee
    (Annex 1, reference 17), and a temporary ADI of 0-0.15 mg/kg bw was
    established. Oxytetracycline was reevaluated at the thirty-sixth
    meeting of the Committee (Annex 1, reference 91), which concluded

    *    The absorption of oxytetracycline in humans after oral
         administration is greater than in mice and pigs (about 60%
         compared with 4-9%). After absorption, oxytetracycline is widely
         distributed in the body and eliminated mainly in the urine as
         parent drug.

    *    There was no evidence that oxytetracycline is carcinogenic in
         mice or rats. Toxicity was observed in experimental animals only
         at doses higher than those that caused microbiological effects.
         The lowest NOEL in the toxicological studies in animals was 18
         mg/kg bw per day.

    *    The NOEL for the induction of resistant coliforms in the
         intestinal microflora in a six-week study in dogs was 0.05 mg/kg
         bw per day, while in humans the NOEL was 2 mg/person per day,
         equivalent to 0.03 mg/kg bw per day. With a safety factor of 10
         to account for variation in the intestinal microflora of humans,
         an ADI of 0-0.003 mg/kg bw was established. The next highest dose
         in the study in humans was 20 mg/day.

         Chlortetracycline and tetracycline were reevaluated at the
    forty-fifth meeting of the Committee (Annex 1, reference 119), which
    concluded that:

    *    In humans receiving oral therapeutic doses, 30% of the dose of
         chlortetra-cycline was absorbed, as compared with 60-80% of an
         oral dose of tetracycline. The absorption is impaired by divalent
         cations such as calcium and magnesium, which form complexes with
         the drug. Tetracycline is excreted mainly in the urine, and
         chlortetracycline is excreted in urine and faeces. Both drugs are
         widely distributed in the body, the highest concentrations of
         residue being found in kidney and liver.

    *    Tumour incidences were not increased in mice or rats treated with
         tetracycline or chlortetracycline.

    *    Tetracycline and chlortetracycline did not have significant
         reproductive or developmental effects, and induced general
         toxicity in experimental animals only at high doses. Data on the
         reproductive toxicity of tetracycline were not available.

    *    The geometric mean MICs for chlortetracycline and oxytetracycline
         against selected microorganisms of the human intestinal
         microflora are similar.

    *    The Committee concluded that antimicrobial activity is the most
         sensitive end-point for determining the ADI for tetracycline and
         chlortetracycline. A group ADI of 0-0.003 mg/kg bw was
         established for oxytetracycline, tetracycline, and
         chlortetracycline separately or in combination because of their
         similar antimicrobial activity. This ADI is based on the NOEL of
         2 mg/person per day determined in a study in humans with
         oxytetracycline that was reviewed at the thirty-sixth meeting.
         The Committee noted that this ADI provides an adequate margin of
         safety when compared with the lowest NOEL for toxicological
         effects of 100 mg/kg bw per day for chlortetracycline in dogs.

         At the tenth session of the Codex Committee on Residues of
    Veterinary Drugs in Foods, the Expert Committee was requested to
    reconsider its previous evaluation of the tetracycline group.


    2.1  Absorption, distribution, and excretion

         The tetracyclines are incompletely absorbed from the human
    gastrointestinal tract; 30% chlortetracycline and 60-80%
    oxytetracycline and tetracycline are absorbed from an empty stomach.
    When humans were treated orally with 1 g/d of tetracycline, the
    concentration in the faeces was 10-97 g/g (Van Marwyck, 1958).

    2.2  Effects on enzymes and other biochemical parameters of the 
    intestinal flora

         Because the tetracyclines are incompletely absorbed from the
    gastrointestinal tract, high concentrations are readily achieved in
    the intestine, which perturb the intestinal microflora within 48 h of
    daily treatment. Both therapeutic and subtherapeutic doses of
    tetracyclines can perturb the intestinal microflora by inducing
    resistant strains, altering the metabolic activity and the
    colonization resistance of the flora (barrier effect) to allow
    overgrowth of pathogenic, opportunistic, or resistant microorganisms,
    and altering the ecological balance of the flora with no identified
    deleterious effect (Corpet & Brugre, 1996).

         The tetracyclines have similar microbiological activity and are
    considered to be 'broad spectrum' antibiotics. Although these drugs
    were originally very active against gram-positive bacteria, many of
    those bacteria have become resistant to tetracyclines, and many
    gram-negative bacilli have also become less susceptible. Anaerobic
    bacteria that are resistant to the tetracyclines have also emerged.
    The bacteriostatic effect of the tetracyclines is due to inhibition of
    bacterial protein synthesis. They bind to 30S subunits of bacterial
    ribosomes and prevent attachment of aminoacyl-tRNA to the ribosomal
    receptor site, thus altering protein synthesis and therefore cell
    multiplication. The inhibitory effect of tetracyclines is usually
    reversible when the drug is removed. Although tetracycline appears to
    have a greater effect on the selection of resistant bacterial strains
    than the other tetracyclines, microorganisms that are resistant to one
    of the tetracyclines are frequently resistant to the other compounds
    in this class (Brown, 1988; Sande & Mandell, 1990; Chopra et al.,
    1992; Roberts, 1996).

         Alterations in the metabolism of the intestinal microflora have
    been demonstrated in human volunteers given therapeutic doses of
    oxytetracycline. The drug caused a marked reduction in the
    esterification and transformation of faecal neutral sterols, which are
    biochemical processes carried out by the intestinal microflora
    (Korpela et al., 1984). Oxytetracycline at therapeutic doses has also
    been shown to increase faecal concentrations of conjugated oestrogens
    in men, probably due to the decreased hydrolytic effect of
    beta-glucuronidase produced by the intestinal microflora (Hamalainen
    et al., 1987). Women receiving the birth control drug
    ethyniloestradiol and treated with tetracycline had a markedly

    decreased average circulating half-life of the oestrogen and increased
    faecal excretion of ethyniloestradiol due to decreased intestinal
    reabsorption of the oestrogen. Administration of tetracycline reduced
    the beta-glucuronidase activity of the microflora and therefore
    altered the metabolism of oestrogens (Gorbach, 1993). When therapeutic
    doses of tetracycline were given to healthy volunteers who were
    subsequently treated with salicylazosulfapyridine, a drug used in the
    treatment of ulcerative colitis, the metabolism of
    salicylazosulfapyridine was altered. While untreated humans excrete
    the break-down products, the subjects given tetracycline excreted
    salicylazosulfapyridine intact in the faeces, indicating a reduction
    in bacterial azoreductase activity (Gorbach, 1993). It has also been
    demonstrated that the metabolism of digoxin, a cardioglycoside that is
    normally metabolized in the intestine, is altered by the action of
    antibiotics on the intestinal microflora. The metabolic activity of
    the microflora, mainly the anaerobic bacteria  Eubacterium lentum, is
    responsible for reduction of the lactone ring and hydrolysis of the
    glycoside moiety. Tetracycline therapy decreases urinary excretion of
    the digoxin reduction products, which can result in an increased serum
    concentration of digoxin and may cause digitalis-induced toxicity
    (Lindenbaum et al., 1981).

         Diarrhoea and superinfections are common side-effects that limit
    the therapeutic use of tetracyclines. Tetracyclines inhibit the growth
    of many aerobic and anaerobic coliform microorganisms and
    gram-positive spore-forming bacteria. As the faecal coliform count
    decreases, tetracycline-resistant organisms, particularly yeasts and
    the enterococci  Proteus and  Pseudomonas, overgrow, and
    superinfections can develop. Of the superinfections that occur as a
    result of tetracycline therapy, antibiotic-induced diarrhoea and
    pseudomembranous colitis due to a cytotoxic toxin produced by the
    overgrowth of  Clostridium difficile are the most important in
    humans. The incidence of gastrointestinal distress increases with
    dose. In general, the normal intestinal flora is restored several days
    after treatment is withdrawn (Nord et al., 1984; Sande & Mandell,
    1990; Ramos et al., 1996).

    2.3  Microbiological studies

    2.3.1  In vitro

         When  Staphylococcus aureus ATCC 9144 was exposed to 30 ppb
    oxytetracycline for 14 days, the MICs of oxytetracycline were similar
    to those of bacteria that were not exposed to the antibiotic,
    suggesting that 30 ppb is not sufficient to select for resistant
    bacteria  in vitro. When oxytetracycline was combined with low
    concentrations of other antimicrobials (neomycin, erythromycin,
    sulfamethazine, and/or dihydrostreptomycin), the MIC increased (Brady
    et al., 1993). Similar results were obtained by Brady and Katz (1992)
    when they studied the effect of low doses of antibiotics on
     Staphylococcus aureus ATCC 9144 and  Enterobacter cloacae B520
    after 14 days of incubation. These data showed that a combination of
    antimicrobials can cause a decrease in bacterial sensitivity.

         In an investigation of the ability of low concentrations of
    antimicrobials to affect the conjugal transfer of resistance among
     Escherichia coli strains, oxytetracycline added at a concentration
    of 0.01, 0.1, or 1.0 g/L to flasks containing recipient and donor
     E. coli strains did not increase the transfer of resistant plasmids.
    Oxytetracycline at 0.01 g/L, however, decreased the transfer of
    resistant plasmids (Brady & Katz, 1988). In a similar study, the
    development of resistance was studied with  E. coli CS-1 (a chicken
    faecal strain) and  Enterobacter cloacae B520 (a human intestinal
    strain). Both organisms showed decreased susceptibility (measured by
    the MIC) to chlortetracycline and oxytetracycline at concentrations of
    0.15 g/L. No change in susceptibility was seen with other
    antibiotics (streptomycin, tylosin, sulfamethazine, bacitracin,
    virginiamycin, flavomycin, and monensin) at a concentration of 1 g/L.
    The results varied markedly, however, according to the strain used and
    the method for measuring resistance (Brady et al., 1988).

         In a study with a continuous flow chemostat model system
    containing an  E. coli culture and subinhibitory concentrations of
    antibiotics, 0.25 g/ml (0.25 mg/L) of tetracycline resulted in
    selection of an R-plasmid strain that grew faster than the sensitive
    strain (Egger & Lebek, 1985). Interpretation of this study was
    difficult, because the culture did not contain species representative
    of human intestinal microflora.  Minimal inhibitory concentration

         MICs were determined with four strains of each of the 50 most
    numerous species in the human colon by the method recommended by the
    National Committee for Clinical Laboratory Standards (NCCLS) for
    anaerobic bacteria. The tests were performed in anaerobiosis, at pH
    6.0, 6.5, or 7.0, with inoculum densities of 105 and 107 cells per
    spot. The bacteria used included strains isolated from normal human
    intestinal microflora over a period of about 30 years. The results for
    the  Bacteroides species showed that their resistance has increased
    over the years. Strains with MICs for tetracycline < 0.5 g/ml were
    isolated between 1933 and 1969; strains with MICs > 1 g/ml were
    isolated after 1970. These results cannot readily be compared with
    those for more recently isolated bacterial strains because the
    susceptibility of strains isolated several years ago differs markedly
    from that of strains isolated today.

         The MICs of 10 strains of each of 10 species of human colonic
    bacteria  (E. coli, Bifidobacterium spp.,  Bacteroides fragilis, 
     Eubacterium spp.,  Clostridium spp.,  Streptococcus spp.,
     Fusobacterium spp.,  Lactobacillus spp.,  Proteus spp., and
     Peptostreptococcus spp.) and three reference strains  (E. coli 
    7624,  B. fragilis 7716 T, and  C. perfringens 6019) were
    determined. All of the bacterial species were isolated from healthy
    volunteers in 1992 and 1993. The MIC50, MIC90, and geometric MIC
    were calculated for each bacterial species and each antimicrobial
    agent, and the sensitivity of the bacterial strains to oxytetracycline
    and tetracycline was compared. The concentrations of these drugs that

    had no effect on any of the 10 strains were also presented. In
    general, the MIC50, the MIC90, the geometric mean, and the
    concentrations of oxytetracycline and tetracycline that had no
    inhibitory effect on bacterial growth were very similar; however,
    tetracycline was more active against  Bifido-bacterium spp.,
     Eubacterium spp., and  Fusobacterium spp. and less active against
     Streptococcus spp. The geometric mean MIC for all the tested strains
    (with a value of 32 g/ml given to resistant strains) was very similar
    for oxytetracycline (3.81 g/ml) and tetracycline (3.25 g/ml). The
    most sensitive relevant genera were  Fusobacterium spp. and
     Clostridium spp., with an MIC50 of 0.2 g/ml (Richez, 1994).

         MICs were also determined by the Centre National d'Etudes
    Vtrinaires et Alimentaires (1997) in France (Table 1). The tests
    were performed with bacterial strains isolated from the faeces of
    human volunteers and tested by the NCCLS method. In general, aerobic
    and anaerobic bacteria were susceptible to tetracycline, i.e. more
    than 50% of the strains were inhibited below the break-point value of
    8 g/ml. A large number of aerobic strains  (Enterococcus and
     E. coli) isolated in brainheart infusion agar and enterococcus
    media had an average MIC50 of 1 g/ml. A few strains of  E. coli 
    isolated aerobically in McConkey medium had an average MIC50 of
    8 g/ml, but their MIC90 was much higher (64 g/ml). Anaerobic
    strains had an average MIC50 of 1 g/ml in both blood-supplemented
    brainheart infusion agar and Schaedler media. Gram-positive anaerobic
    flora were more susceptible to tetracycline than gram-negative flora.
    In general, the results of this study are very similar to those
    obtained by Richez (1994).  Continuous flow chemostat system inoculated with tetracycline

         In a series of experiments carried out by Techlab, Inc. (1997),
    single-chambered chemostats with continuous flow were inoculated with
    a pool of faeces from healthy volunteers and maintained under
    conditions mimicking the human colon. The medium used in the chamber
    contained elements found in omnivorous diets (casein, peptone, and
    complex carbohydrates from plants) supplemented with bile, L-cysteine,
    salts, haemin, vitamins, and Tween 80. The lot of medium in the
    control chemostat (no drug) was changed at the same time as that in
    the test chemostats. After inoculation with faeces, the chemostat
    culture was allowed to reach steady-state (generally at about two
    weeks), which was confirmed by measuring the following
    microflora-associated characteristics: (i) general microscopic
    appearance of the culture fluids; (ii) cellular fatty acids; (iii)
    molar ratio of short-chain fatty acids; (iv) sulfate reduction; (v)
    counts of target organisms within the normal flora, including
    facultative coliforms and enterococci; (vi) bacterial enzymes
    including beta-glucosidase, beta-glucuronidase, nitroreductase, and
    azoreductase; and (vii) bacterial metabolism of bile acids. When
    steady-state was reached, each chemostat was inoculated with 0
    (control), 0.15, 1.5, or 15 mg/ml of tetracycline. These doses
    correspond to ADI values of 0, 0.025, 0.25, and 2.5 mg/kg bw, assuming
    that about 100 g of tetracycline are present in 1 g faeces or colon

        Table 1. Susceptibility to tetracylcine of identified anaerobic strains isolated from faeces of 
    four healthy donors


    Isolation medium     MIC      No. of    Bacterial species                MIC50       MIC90
                         (g/ml)  strains   (no. of strains)                 (g/ml)     (g/ml)

    Brain-heart          0.5      4         E. faecalis (3)                  1           64
    infusion agar                           E. faecium (1)
                         1        18        E. faecalis
                         2        3         E. coli
                         8        1         E. coli
                         16       2         E. coli
                         64       9         E. faecalis
                         128      2         E. durans (1)
                                            E. faecalis (1)
                         >128     2         E. faecalis

    McConkey             1        2         E. coli                          8           8
                         2        2         E. coli
                         8        6         E. coli

    Enterococcus         1        8         E. faecalis (7)                  1           64
                                            Lc lactis lactis (1)
                         64       3         E. faecalis

    Brain-heart          <0.125   2         Clostridium spp. (1)             1           32
    infusion agar                           Bacteroides spp. (1)
                         0.25     3         Propionobacterium acnes (1)
                                            Bifidobacterium breve (1)
                                            Clostridium spp.
                         0.5      7         Gram-negative bacilli (1)
                                            Mobiluncus spp. (3)
                                            Gram-positive bacilli (1)
                                            Bifidobacterium breve (2)
                         1        6         Bifidobacterium spp. (4)
                                            Bifidobacterium infantis (1)
                                            Bifidobacterium longum (1)
                         8        1         Gram-positive bacilli (1)
                         16       2         Bacteroides spp. (1)
                                            Clostridium spp. (1)
                         32       2         Bacteroides caccae (1)
                                            Bacteroides vulgatus (1)
                         64       2         Bacteroides distasonis (1)
                                            Bacteroides uniformis (1)
                         128      1         Bacteroides vulgatus (1)

    Table 1 (continued)


    Isolation medium     MIC      No. of    Bacterial species                MIC50       MIC90
                         (g/ml)  strains   (no. of strains)                 (g/ml)     (g/ml)

    Schaedler            <0.125   2         Clostridium spp. (1)             1           64
                                            Bacteroides merdae (1)
                         0.25     1         Bacteroides spp.
                         0.5      5         Gram-positive bacilli (1)
                                            Bifidobacterium infantis (1)
                                            Bifidobacterium breve (1)
                                            Bacteroides spp. (1)
                                            Actinomyces israeli (1)
                         1        10        Mobiluncus spp. (4)
                                            Bifidobacterium spp. (3)
                                            Bifidobacterium longum (1)
                                            Clostridium ramosum (1)
                                            asaccharolyticus (1)
                         16       3         Gram-positive bacilli (1)
                                            Bacilli (2)
                         32       4         Gram-positive bacilli (1)
                                            Bacteroides spp. (1)
                                            Bacteroides uniformis/ovatus (2)
                         64       5         Bacteroides ovatus (1)
                                            Bacteroides vulgatus (3)
                                            Bacteroides spp. (1)
    content (Van Marwyck, 1958) and that the bacterial content of the
    chemostats simulates that of the colon on a daily basis. Daily samples
    were taken before treatment (days 17-24) and during treatment (days

         In a separate set of experiments, chemostat faecal cultures were
    used to evaluate the emergence of resistant strains and loss of
    colonization resistance to a  Clostridium difficile strain. The
    methods were validated for short-chain fatty acids, enzymes, bile
    acids (primary and secondary analytes), protein concentration,
    cellular fatty acids, and sulfate reduction. Additional studies were
    performed to validate the procedures for maintaining the chemostat

          (i)  Microfloral ecology

         In the studies described below, the chemostats were inoculated
    with the faecal pool and allowed to reach steady-state, which required
    about two weeks. Samples were taken daily before (days 17-24) and
    during treatment (days 25-33).

          Cellular fatty acid profile: Cellular fatty acid profiles were
    used to determine whether the bacterial cultures were changing and the
    antibiotic was altering the bacteria. Each sample was centrifuged and
    washed in buffer, and the bacterial pellet was collected. The cellular
    fatty acids were extracted, derivatized, and identified and quantified
    by gas chromatography. The profiles were compared statistically with
    Microbiological Identification Inc. (Newark, DE) software developed to
    identify unknown strains of bacteria. The unit of relatedness of two
    cellular fatty acids profiles is the Euclidian distance (ED). Similar
    profiles have a low ED and different profiles have a high ED.
    Statistically significantly identical profiles (e.g. the same sample
    run twice) are related by < 2 ED; different strains of the same
    subspecies are related by an ED of 2-6; different subspecies of the
    same species are related by an ED of 6-10; and different species from
    a single genus are related by an ED of 10-25. These values were
    established by analysing thousands of previously characterized
    anaerobic bacteria from culture collections.

         In the chemostats, the flora on day 1 of inoculation resembles
    that of the faeces. After inoculation, the flora changes until it
    reaches steady-state, i.e. an unchanging bacterial community. Higher
    EDs indicate increasing differences between the successive daily
    profiles of the chemostat and the faecal flora. As the flora reaches a
    steady-state, the ED is lower. For example, the results on incubation
    day 7 differ from those on day 6 by an ED of 2, indicating that the
    flora is almost identical on those two days and that the chemostat is
    reaching steady-state. The steady-state flora and faeces had different
    cellular fatty acid profiles, however, indicating a difference in
    either the number of bacterial species or the cellular fatty acid
    composition of the flora.

         After establishment of the steady-state about two weeks after
    inoculation, tetracycline was added to the chemostats at 0, 0.15, 1.5,
    or 15 mg/L, and the range of relatedness values generated for each
    successive daily sample was compared with the total range of values.
    An effect of the drug would be reflected as an increase in the ED
    value with increasing dose of tetracycline in the chemostat; however,
    as seen in Table 2, the results showed no difference in ED at the four
    doses tested when daily samples were analysed before and during
    treatment. This suggests that tetracycline does not alter the overall
    composition of cellular fatty acids in the chemostats.

          Short-chain fatty acids: The three main short-chain fatty acids
    produced by the intestinal microflora as a product of bacterial
    metabolism are acetic, propionic, and butyric acids. These fatty acids
    are absorbed from the colon. Butyrate is the main energy source for
    colonic enterocytes. Propionate is transported to the liver and
    metabolized and reduces serum cholesterol. Acetate enters the
    peripheral circulation, where it is metabolized. Since the objective
    of this study was to model the environment of the intestinal flora,
    the net concentrations of short-chain fatty acids were used as a
    measure of bacterial metabolism.

         Because of the complexity of the intestinal microflora and lack
    of knowledge about bacterial fatty acids  in vivo, there are no fatty
    acid markers for particular groups of bacteria. This is however a
    sensitive assay commonly used for analysing the total intestinal
    microflora. Short-chain fatty acids were extracted from chemostat
    samples, derivatized, and analysed by gas chromatography. This method
    provides information only on changes in the fatty acid profiles due to

         The short-chain fatty acid content of the chemostats reached
    steady-state at the beginning of the third week. Samples were taken
    daily before and during treatment. The total short-chain fatty acid
    concentrations of the chemostats containing the drug were similar to
    those of the control chemostat. In addition, tetracycline had no
    effect on the concentrations of individual fatty acids or on the
    combined concentration of the minor short-chain fatty acid components.
    The concentrations were almost uniform between the chemostats,
    constant within chemostats at steady-state, and unaffected by the
    addition of tetracycline. The only exception was the control
    chemostat, in which the concentration of the combined minor
    short-chain fatty acids rose at the expense of the propionate. Since
    fatty acids are common indicators of changes in the metabolism of
    bacteria in the intestinal microflora, these results show that
    tetracycline does not alter the metabolic activity of the flora or
    that the flora can compensate for the effect of the drug at the
    concentrations tested.

    Table 2. Relatedness of cellular fatty acid profiles of faecal
    flora after treatment with tetracycline in a continuous flow
    chemostat system


    Dose                                    Relatenessa

    No drug             (0 ADI)             11.5
    0.15 mg/L           (1 ADI)             17.1
    1.5 mg/L            (10 ADI)            11.1
    15 mg/L             (100 ADI)           13.3

    a Euclidian distance

          Bile acids: The concentration of bile acids produced during
    lipid metabolism is related to fat intake and is a high-risk factor
    for metastatic bowel disease. High concentrations of bile acids
    produced in response to high fat intake stimulate the growth of
    certain bacterial species, such as the  Bacteroides. While these
    bacteria are dying and before they are excreted in faeces, their
    enzymes are liberated. Some of these enzymes can metabolize primary
    bile acids into secondary genotoxic metabolites. When the fat intake
    is high and fibre intake low, these genotoxic secondary metabolites
    concentrate in the intestinal lumen and remain in the colon for a
    longer time, allowing high concentrations of genotoxic metabolites to
    come into contact with the intestinal cells for prolonged periods.
    Primary (mammalian) and secondary (bacterial metabolites) bile acids
    can be measured in two ways: as the levels of individual and total
    analytes and as the percentage of the total that is secondary bile
    acids. The latter was measured in this study, since the concentration
    of primary bile acids in the medium was fixed. Two primary bile acids,
    cholic and chenodeoxycholic acids, were added to the medium.

         The results did not show a clear effect of the drug. For example,
    the percent molar ratio of chenodeoxycholic acid increased at 1.5 mg/L
    and decreased at 0, 0.15, and 15 mg/L during the steady-state period,
    for no apparent reason. The percent of secondary bile acids was
    similar and uniform at all concentrations, except in the control
    chemostat where the concentrations continued to rise steeply during
    the steady-state phase. In general, the dose of tetracycline had no
    effect on the percent of secondary bile acids, but the background
    (control) data were very variable.

          Neutral sterols: The main difference between bile acids and
    primary neutral animal sterols is that the latter lack the carboxylic
    acid group of bile acids; however, their biological properties are
    similar. Primary neutral sterols are converted to secondary sterols by
    bacterial metabolism. The commonest animal neutral sterol is
    cholesterol, and its concentrations in the culture medium can be

    adjusted. In the intestine, cholesterol is converted to coprostanol
    via an intermediate metabolite, coprostanone.  Eubacterium spp. are
    important cholesterol converters in the human intestine. The parameter
    measured in the study was the percentage of secondary neutral sterols
    present in the medium before and during treatment, indicative of
    bacterial metabolism. Tetracycline had no noticeable effect on the
    conversion of cholesterol to coprostanol.

          (ii)  Bacterial enzymes

         Bacterial enzymes were analysed in chemostat sample effluent with
    chromogenic substrates. The results were expressed as activity per
    milligram of protein.

          Sulfate reduction: The concentration of sulfate in the human
    colon ranges from 1 to 16 mmol/d, depending on the diet. The daily
    flow of inorganic sulfate in the chemostat was about 1.5 mmol, organic
    sulfate being present at some proportion. The chemostat medium is thus
    a good approximation of the human sulfate colon content. The parameter
    measured in this study was the content of sulfide, a metabolite of
    sulfate generated by bacterial metabolism. Sulfate reduction was
    uniform at steady-state and remained unchanged throughout treatment in
    all chemostats. Thus, no effects of tetracycline were seen.

          Azoreductase: Anaerobic colonic bacteria that use azo compounds
    as an energy source reduce azo dyes. No clear drug-related pattern was
    observed on this parameter; however, although the short-chain fatty
    acid concentrations indicated that the chemostats had reached
    steady-state, the azoreductase concentrations of the treated
    chemostats continued to decrease in the fourth week of the study.

          Nitroreductase: Nitroreductase, a bacterial oxygen-labile
    enzyme, was also measured anaerobically. The results were difficult to
    interpret. Two chemostats, containing 0.15 and 15 mg/L, showed steady
    declines in nitroreductase activity during the presumed steady-state
    phase; however, the control chemostat and that containing 1.5 mg/L had
    constant activity at steady-state. After the addition of 15 mg/L
    tetracycline, the decline was reversed. No conclusions were drawn from
    this study because the effect seen at the high dose was not seen at
    the lower doses and the data were highly variable for all of the other

          beta-Glucosidase: beta-Glucosidase is produced by several
    colonic bacteria, including  Bacteroides, Streptococcus, and
     Lactobacillus  spp.; it hydrolyses many sugar conjugates. When this
    enzyme is present at high concentrations in the colon, the
    glycone-linked compounds are deconjugated and enter the enterohepatic
    circulation. Since many conjugated compounds, including bile acids,
    are potentially genotoxic, high concentrations of glucosidase activity
    have been linked to an increased risk for colon cancer. High
    concentrations of glucosidase are present in people who eat high-fat,
    low-fibre diets. In this study, the four chemostats had different
    levels of beta-glucosidase activity at steady-state; the levels varied

    during treatment but showed no trend after the addition of

          beta-Glucuronidase: Most of the beta-glucuronidase activity in
    the colon is due to  Bacteroides spp., although  E. coli and other
    species also produce this enzyme. When the glucuronide conjugate is
    hydrolysed by bacterial metabolism, it enters the enterohepatic
    circulation. The activity of beta-glucuronidase is influenced by diet,
    as high-fat diets induce production of this enzyme. In this study, the
    activity of beta-glucuronidase was still falling in the control
    chemostat and that containing 1.5 mg/L and rising in those with 0.15
    and 15 mg/L even at steady-state. Addition of tetracycline did not
    change the trends.

         These studies on several bacterial enzymes failed to show a clear
    effect of tetracycline. While some parameters used to monitor
    bacterial ecosystems indicated that a steady-state was achieved, the
    enzyme concentrations were still variable. This inherent variability
    may limit the usefulness of enzymatic activity for determining the
    effect of antibiotics  in vitro.

          (iii)  Bacterial counts

         Techniques developed at the Virginia Polytechnic Institute and
    State University (VPI) Anaerobe Laboratory and those described in the
     Wadsworth Anaerobic Bacteriology Manual were used to count anaerobic
    bacteria. In addition, coliforms and enterococci were cultured
    aerobically, and the ratios of coliforms or enterococci to total
    anaerobe counts were reported.  Bacteroides fragilis, fusobacteria,
    bifidobacteria, clostridial spores, eubacteria, veillonella,
    peptostreptococci, peptococci, megasphaera,  E. coli, and enterococci
    were also counted. Although bacterial counts are traditionally used to
    assess perturbations in intestinal microflora, total bacterial counts
    lack specificity for different treatment groups, and the counts for
    matched samples vary widely. Although total anaerobes were more
    numerous in the chemostat after addition of the highest dose of
    tetracycline, it had no major effects on bacterial counts.  Effect of tetracyclines on colonization resistance

         The effects of tetracyclines on resistance were studied in
    another series of studies conducted by Techlab Inc. (1997).
    Steady-state was reached about two weeks after chemostats were
    inoculated with faeces. After the pretreatment period (days 17-24),
    tetracycline at doses of 0, 0.15, 1.5, or 15 mg/L was added on days
    25-33. On days 33, 34, and 35, each chemostat was inoculated with
    107 viable cells of  Clostridium difficile VPI 10463 in order to
    study the ability of the flora to prevent colonization with a foreign
    microorganism. Samples were drawn between days 34 and 42. Control
    samples were taken before and during tetracycline treatment to obtain
    the background count and toxin titre. Samples were cultured for
     C. difficile in selective medium [D-cycloserine, cefotoxin, and
    fructose agar (CCFA)] with and without prior washing with ethanol for

    additional selection of spores. The CCFA medium allows growth of other
     Clostridium species and of at least one  Enterococcus sp. if
    washing with ethanol is not performed. The concentrations of
     C. difficile toxin A were also measured. A positive control
    chemostat was inoculated with  C. difficile VPI 10463 but no faeces
    in order to show that  C. difficile can colonize the chemostat. The
    counts in this chemostat decreased during the first 72 h, but once
     C. difficile had adapted to the chemostat conditions (about day 6
    after inoculation), the counts rose to 105-106 microorganism per

         The chemostats were not colonized by  C. difficile, unlike the
    positive control chemostat, and toxin A production was not detected
    after the initial period of inoculation. These results could not be
    confirmed by culture of  C. difficile on CCFA because in chemostats
    inoculated with faeces the bacterial count on CCFA medium in the
    absence of  C. difficile was frequently as high or even higher than
    the count in the control chemostat with no faeces and inoculated with
     C. difficile. The species of bacteria that constituted this
    background were not identified.

         In summary,  C. difficile colonized a positive control chemostat
    (with no faeces added) lacking resistance to colonization. Addition of
    tetracycline at doses of 0.15, 1.5, or 15 mg/L to chemostats
    inoculated with faeces did not reduce the colonization resistance of
    the steady-state flora to a level that would allow  C. difficile to
    colonize.  Selection of bacterial strains resistant to tetracycline

         The numbers of tetracycline-resistant strains of enterococci,
     E. coli, and  Bacteroides fragilis were compared in control
    chemostats and in chemostats treated with 0.15, 1.5, or 15 mg/L
    tetracycline. First, the background resistance to tetracycline was
    determined in the faecal pool used to inoculate the chemostats. The
    percentages of isolated strains resistant to tetracycline are shown in
    Table 3.

    Table 3. Background resistance of faecal bacteria to tetracycline

    Tetracycline        Percent resistance
                        E. coli     Enterococci
    4                   3           ND
    8                   3           91
    16                  3           92
    32                  ND          91
    64                  ND          91

    ND, not determined

         The increase in the number of resistant strains of  E. coli was
    measured against a background of 3% at 8 g/ml.  E. coli from
    McConkey agar plates containing 20-300 colonies were plated in
    replicate onto agar containing 4, 8, or 16 g/ml tetracycline. After
    16 and 24 h of incubation, the counts on the master plates were
    compared with those on the plates with tetracycline. The highest dose
    of tetracycline increased the percent of resistant  E. coli from
    < 20 to > 50% within 24 h, rising to > 60% after a second day of
    exposure. Despite continued treatment, the percents of resistant
     E.coli then began to fall, and by day 6 after treatment it was about
    35%. In the control chemostat (no tetracycline), the percent of
    resistant strains never exceeded 5%. The chemostats with 0.15 and 1.5
    mg/L gave variable results. No effect was seen on the resistance of
    enterococci or  B. fragilis, probably because of the high level of
    resistance (60100%) of these microorganisms. Furthermore, the
    resistance of these organisms was already very high at the beginning
    of the study (91% of the enterococci in the faecal pool used to
    inoculate the chemostats were resistant to tetracycline), so that the
    addition of tetracycline did not have any measurable effect. In the
    chemostats, the lowest level of tetracycline resistance in
     B. fragilis strains was > 50%, and most of the levels were close to

    2.3.2  Animal models

         In studies of germ-free mice associated with human flora,
    treatment in drinking-water with 0.5 g/ml chlortetracycline increased
    the number of resistant  E. coli. Although the dose was similar to
    the residues allowed in food, the contribution of the
    chlortetracycline residues to resistance is probably low, because it
    has been shown that the daily increase in the number of resistant
    enterobacteria after ingestion of food can be 1 000 000 (Corpet, 1987,
    1988). In a study in which germ-free mice were associated with two
    isogenic strains of  E. coli, one of which carried an R plasmid, the
    mice were given tetracycline in drinking-water for 1015 days. The
    minimum selecting concentration of tetracycline was 6.5 g/ml, which
    resulted in a faecal concentration of 0.5 g/g tetra-cycline -
    one-half of the MIC for the plasmid-free strain. This concentration
    did not eliminate the susceptible strains from the gut; however, this
    model did not include the dominant anaerobic flora that is responsible
    for the barrier effect (Corpet  et al., 1989; Corpet, 1993). In a
    study in which germ-free mice were associated with faecal microflora
    from piglets and treated with 20 g/ml chlortetracycline in the
    drinking-water, the occurrence of lactose-fermenting bacteria
    resistant to chlortetracycline increased (Corpet, 1984). Although the
    effect of chlortetracycline was studied in piglet microflora, the
    concentration of the drug in the water was much higher than the amount
    of residues allowed in food. These studies suggest that the variations
    seen in human faecal flora after exposure to low concentrations of
    tetracyclines in food might be similar to the variation that occurs

    2.4  Studies in humans

         The emergence of resistant  E. coli strains in humans after
    therapeutic doses of tetracycline has been documented; however, the
    level of resistant bacteria decreased with time after the last
    treatment (Hirsh et al., 1973; Bartlett et al., 1975).

         In early studies, some people given oxytetracycline at 10 mg/day
    for six months showed an increased number of resistant coliforms and
    yeasts; however, the effect was transitory, and resistant coliforms
    were present even before treatment (Goldberg et al., 1961).

         A dose of 100 mg/d tetracycline given for long-term treatment of
    acne vulgaris altered the resistance patterns of the microflora,
    increasing the percentage of people with transferrable R factor
    bacteria and the number of multiresistant strains (Valtonen et al.,
    1976). In volunteers given 0, 50, or 1000 mg/d of tetracycline for
    four days, however, the shedding of  E. coli from the intestine was
    increased at the high dose but not at the low dose, indicating that 50
    mg of tetracycline do not perturb the barrier effect of the human
    flora (Hirsh et al., 1974).

         A dose-related effect of oxytetracycline was seen on intestinal
    microflora in people given 2, 20, or 2000 mg/d. The effects ranged
    from drastic disturbance of the microflora (elimination of susceptible
    anaerobes and more resistant Enterobacteriaceae with 2 g/day) to a
    tendency to a decrease in the number of susceptible Enterobacteriaceae
    and an increase in the number of resistant strains at 20 mg/d. A dose
    of 2 mg/d therefore appears to have no effect on the selection of
    resistant Enterobacteriaceae (Tancrede & Barakat, 1987).

         Healthy humans given oxytetracycline at 1 g/d orally for five
    days showed a decreased faecal concentration of saponifiable
    conjugates of bile acids, which are probably formed by metabolic
    action of the intestinal microflora. The drug also reduced bacterial
    transformation of cholesterol to coprostanol and coprostanone,
    decreased the amount of faecal sterified neutral sterols, and
    increased excretion of faecal conjugated oestrogens (Korpela et al.,
    1984, 1986; Hamalainen et al., 1987). The same dose of tetracycline
    has been shown to increase the number of tetracycline-resistant
    strains of  E. coli (Hirsh et al., 1973; Bartlett et al., 1975);
    however, the minimum effective dose of tetracyclines, 1 g/d, has been
    shown to have a minimal effect on the total population of aerobic or
    anaerobic bacteria after 8-10 days of oral administration. When given
    in divided doses, this dose of tetracycline also has infrequent, mild
    side-effects (nausea, diarrhoea, and vomiting), which disappear when
    treatment is discontinued (Kunin & Finland, 1961; Bartlett et al.,


         The studies used originally to establish an ADI for
    oxytetracycline, tetracycline, and chlortetracycline showed that these
    drugs have a low degree of toxicity, and additional toxicological
    studies have confirmed that conclusion.

         Because the tetracyclines are poorly absorbed from the
    gastrointestinal tract, high concentrations are readily achieved in
    the intestine, perturbing the intestinal microflora within 48 h of
    daily treatment. Experience with tetracycline in human medicine
    indicates that therapeutic levels of tetracyclines can perturb the
    intestinal microflora by inducing resistant strains, altering the
    metabolic activity of the microflora and the colonization resistance
    properties of the flora (barrier effect), which allows overgrowth of
    pathogenic, opportunistic, or resistant microorganisms, and altering
    the ecological balance of the flora with no identified deleterious

         Studies in humans have shown that 1-2 g/d of tetracyclines have
    adverse effects on the intestinal microflora. Resistant strains
    emerged in one study at doses of 100 mg per person per day of
    tetracycline and 10 mg per person per day of oxytetracycline, but
    resistant organisms existed before treatment and the resistance was
    transitory. Another microbiological study in humans showed that 50 mg
    tetracycline did not alter the shedding of  E. coli from the
    intestine, indicating that the barrier effect was not perturbed at
    this dose. In the study used previously by the Committee to establish
    the ADI for tetracyclines, a therapeutic dose of oxytetracycline
    (2 g/d for seven days) perturbed the balance of the intestinal flora,
    with an increased prevalence of resistant Enterobacteriaceae and yeast
    colonization. A dose of 20 mg/d had a marginal ecological impact on
    the flora by suppressing some susceptible anaerobes, but the dose of
    2 mg/d did not result in the presence of resistant Enterobacteriaceae
    in the faeces. The thirty-sixth Committee applied a 10-fold safety
    factor to the 2 mg/d dose to establish an ADI of 0-0.003 mg/kg bw.    
    Since the previous evaluation, more data from microbiological studies
    on the effects of tetracyclines on human gastrointestinal bacterial
    flora have been obtained.

         The results obtained in a continuous bacterial culture system
    (chemostat) dosed to provide the equivalent of 0.025, 0.25, or 2.5
    mg/kg bw per day of tetracycline showed that a dose of 2.5 mg/kg bw
    per day slightly increased the number of resistant  E. coli strains,
    which increased from < 20% to > 50% within 24 h of exposure and to
    > 60% after 48 h of exposure; however, the percentage fell to about
    35% by day 6 of exposure, despite continued treatment. In the control
    chemostat, which contained no drug, the percentage of resistant
    bacteria never exceeded 5%. No effects were seen in chemostats dosed
    with tetracycline at the equivalent of 0.025 or 0.25 mg/kg bw per day.
    None of the other microbiological end-points analysed in this study
    changed with any concentration of tetracycline tested. 


         The results are in agreement with findings reported in the
    literature on the effects of tetracycline on human intestinal
    microflora. Most report a very low impact of tetracycline on colonic
    flora, selection of resistant bacterial strains being the most
    sensitive end-point. On the basis of the new information on the effect
    of tetracycline on human intestinal microflora, the Committee
    concluded that no safety factor was needed, because the selection of
    resistant Enterobacteriaceae is a very sensitive end-point for
    evaluating the microbio-logical effect of tetracyclines on human
    intestinal microflora and because individuals show little variation
    with respect to this effect. On the basis of the NOEL of 2 mg/d
    (equivalent to 33 g/kg bw per day) in the study in humans, the
    Committee established an ADI of 0-30 g/kg bw for the tetracyclines,
    alone or in combination. The ADI was rounded to one significant
    figure, as is usual practice.


    Bartlett, J.G., Bustetter, L.A., Gorbach, S.L. & Onderdonk, A.B.
    (1975) Comparative effect of tetracycline and doxycycline on the
    occurrence of resistant  Escherichia coli in the fecal flora.
     Antimicrob. Agents Chemother., 7, 55-57.

    Brady, M.S. & Katz, S.E. (1988) Method to determine effect of
    antibiotics at residue levels on R-factor transfer.  J. Assoc. 
     Anal. Chem., 71, 299-301.

    Brady, M.S. & Katz, S.E. (1992)  In vitro effect of multiple
    antibiotic/antimicrobial residues on the selection of resistance in
    bacteria.  J. Assoc. Anal. Chem. Int., 75, 738-742.

    Brady, M.S., Strobel, R.J. & Katz, S.E. (1988)  In vitro analytical
    system for determining the ability of antibiotics at residue levels to
    select for resistant bacteria.  J. Assoc. Anal. Chem., 71, 295-298.

    Brady, M.S., White, N. & Katz, S.E. (1993) Resistance development
    potential of antimicrobial residue levels designated as 'safe levels'.
     J. Food Protect., 56, 229-233.

    Brown, S.A. (1988) Treatment of gram-negative infections.  Vet. 
     Clin. N. Am. Small Anim. Pract., 18, 1141-1165.

    Chopra, I., Hawkey, P.M., & Hinton, M. (1992) Tetracyclines, molecular
    and clinical aspects.  J. Antimicrob. Chemother., 29, 245-277.

    Centre National d'Etudes Vtrinaires et Alimentaire (1997) Use of
    human flora-associated rodents to study the effect of low doses of
    tetracyclines and neomycin: Pilot study. Unpublished second quarterly
    report. Submitted to WHO by Department of Health and Human Services/US
    Food and Drug Administration, Rockville, Maryland, USA.

    Corpet, D.E. (1984) The effect of bambermycin, carbadox,
    chlortetracycline and olaquindox on antibiotic resistance in
    intestinal coliforms: A new animal model.  Ann. Microbiol. (Inst. 
     Pasteur), 135A, 329-339.

    Corpet, D.E. (1987) Antibiotic residues and drug resistance in human
    intestinal flora.  Antimicrob. Agents Chemother., 31, 587-593.

    Corpet, D.E. (1988) Antibiotic resistance from food.  New Engl. J. 
     Med., 318, 1206-1207.

    Corpet, D.E. (1993) An evaluation of methods to assess the effect of
    antimicrobial residues on the human gut flora.  Vet. Microbiol., 35,

    Corpet, D.E. & Brugre, H.B. (1996) Antimicrobial drug residues in
    foods: Evaluation of the no effect level on human microflora.
     Recent Res. Devel. Antimicrob. Agents Chemother., 1, 113-123.

    Corpet, D.E., Lumeau, S. & Corpet, F. (1989) Minimum antibiotic levels
    for selecting a resistance plasmid in a gnotobiotic animal model.
     Antimicrob. Agents Chemother., 33, 535-540.

    Egger, R. & Lebek, G. (1985) Selection of R-plasmid harbouring
     E. coli by subinhibitory concentrations of antibiotics.  Zbl. 
     Bakt. Hyg. A, 260, 361-368.

    Goldberg, H.S., Goodman, R.N., Logue, J.T. & Handler, F.P. (1961)
    Long-term, low-level antibiotics and the emergence of antibiotic-
    resistant bacteria in human volunteers.  Antimicrob. Agents 
     Chemother., 80-88.

    Gorbach, S.L. (1993) Perturbation of intestinal microflora.  Vet. 
     Hum. Toxicol., 35 (Suppl. 1), 15-23.

    Hamalainen, E., Korpela, J.T. & Adlercreutz, H. (1987) Effect of
    oxytetracycline administration on intestinal metabolism of oestrogens
    and on plasma sex hormones in healthy men.  Gut, 28, 439-445.

    Hirsh, D.C., Burton, G.C. & Blenden, D.C. (1973) Effect of oral
    tetracycline on the occurrence of tetracycline-resistant strains of
     Escherichia coli in the intestinal tract of humans.  Antimicrob. 
     Agents Chemother., 4, 69-71.

    Hirsh, D.C., Burton, G.C., Blenden, D.C. & Tsutakawa, R. (1974) The
    effect of tetracycline upon establishment of  Escherichia coli of
    bovine origin in the enteric tract of man.   J. Appl. Bacteriol., 37,

    Korpela, J.T., Hamalainen, E. & Adlercreutz, H. (1984) Effect of
    oxytetracycline on bacterial intestinal metabolism of neutral sterols
    and on serum lipids.  Scand. J. Gastroenterol., 19, 401-404.

    Korpela, J.T., Fotsis, T. & Adlercreutz, H. (1986) Multicomponent
    analysis of bile acids in faeces by anion exchange and capillary
    column gasliquid chromatography: Application in oxytetracycline
    treated subjects.  J. Steroid. Biochem., 25, 277-284.

    Kunin, C.M. & Finland, M. (1961) Clinical pharmacology of the
    tetracycline antibiotics.  Clin. Pharmacol. Ther., 2, 51-61.

    Lindenbaum, J., Rund, D.G., Butler, V.P., Tse-Eng, D., & Saha, J.R.
    (1981) Inactivation of digoxin by the gut flora: Reversal by
    antibiotic therapy.  New Engl. J. Med., 305, 789-793.

    Nord, C.E., Kager, L. & Heimdahl, A. (1984) Impact of antimicrobial
    agents on the gastrointestinal microflora and the risk of infections.
     Am. J. Med., 76, 99-106.

    Ramos, K.S., Chacon, E. & Acosta, D., Jr (1996) Toxic responses of the
    heart and vascular system. In:  Casarett & Doull's Toxicology: The 
     Basic Science of Poisons, 5th Ed., New York, McGraw-Hill,  pp.

    Richez, P. (1994) Antibacterial activity of tetracycline
    oxytetracylcine against human gut flora. Study No. SVL 029.
    Unpublished report from Datavet Veterinary Drug Consultants. Submitted
    to WHO by Intervet International, Boxmeer, Netherlands

    Roberts, M.C. (1996) Tetracycline resistance determinant: Mechanisms
    of action, regulation of expression, genetic mobility, and
    distribution.  FEMS Microbiol. Rev., 19, 1-24.

    Sande, M.A. & Mandell, G.L. (1990) Tetracyclines, chloramphenicol,
    erythromycin, and miscellaneous antibacterial agents. In:  Goodman 
     and Gilman's The Pharmacological Basis of Therapeutics, 8th Ed., New
    York, Pergamon Press, pp. 1117-1125.

    Tancrede, C. & Barakat, R. (1987) Ecological impact of low doses of
    oxytetracycline on human intestinal microflora.  J. Vet. Med., 42,

    Techlab, Inc. (1997) Implementation of an in vitro system to study the
    effect of low doses of antimicrobials on the human intestinal flora.
    Unpublished 6th, 7th and 8th annual reports. Submitted to WHO by
    Department of Health and Human Services/US Food and Drug
    Adminstration, Rockville, Maryland, USA.

    Valtonen, M.V., Valtonen, V.V., Salo, O.P. & Makela, P.H. (1976) The
    effect of long term tetracycline treatment for acne vulgaris on the
    occurrence of R factors in the intestinal flora of man.  Br. J. 
     Dermatol., 95, 311-316.

    Van Marwyck, C. (1958) The excretion of tetracycline via the feces.
     Munch. Med. Wochenschr., 100, 1-5 (in German).

    See Also:
       Toxicological Abbreviations