IPCS INCHEM Home

    CARAMEL COLOURS

    EXPLANATION

         In the period since 1972, when the fifteenth report of the
    Committee was published (Annex 1, reference 26), caramel colours have
    been classified into 4 classes which differ in their method of
    manufacture, composition, functional properties, and application.

    1. Caramel Colour I (synonyms: plain caramel, caustic caramel, and
    spirit caramel); this class is prepared by the controlled heat
    treatment of carbohydrates with alkali or acid.

    2. Caramel Colour II (synonyms: caustic sulfite process caramel); this
    class is prepared by the controlled heat treatment of carbohydrates
    with sulfite-containing compounds.

    3. Caramel Colour III (synonyms: ammonia caramel, ammonia process
    caramel, closed-pan ammonia process caramel, open-pan ammonia process
    caramel, bakers' caramel, confectioners' caramel, and beer caramel);
    this class is prepared by the controlled heat treatment of
    carbohydrates with ammonium compounds.

    4. Caramel Colour IV (synonyms: ammonia sulfite process caramel,
    sulfite ammonia caramel, sulfite ammonia process caramel, acid-proof
    caramel, beverage caramel, and soft-drink caramel); this class is
    prepared by the controlled heat treatment of carbohydrates with
    ammonium-containing and sulfite-containing compounds.

         Caramel colours were reviewed at the eighth, thirteenth,
    fifteenth, sixteenth, eighteenth, twenty-first, and twenty-fourth
    meetings of the Committee (Annex 1, references 8, 19, 26, 30, 35, 44,
    & 53). The thirteenth meeting concluded that a toxicological
    distinction between caramels produced commercially and caramel formed
    in cooked foods or when sucrose is heated is unwarranted except with
    caramel prepared by processes using ammonia or ammonium salts. This
    conclusion was endorsed by the fifteenth meeting and an ADI "not
    limited" was allocated to caramel colours prepared by processes other
    than those involving ammonia or ammonium salts; a temporary ADI of
    0-100 mg/kg b.w. was allocated to caramel colours produced by the
    ammonia process. The sixteenth meeting reviewed further information on
    the composition of caramel colours produced by the ammonia process,
    including the 4-methylimidazole content. Revised specifications were
    prepared and the previously-established temporary ADI was maintained. 
    The specifications for this class were further revised by the
    eighteenth meeting and the temporary ADI was extended pending the
    results of long-term and reproduction studies on caramel colours
    prepared by the ammonia or ammonia sulfite process.

         The twenty-first meeting of the Committee noted that the
    specifications for caramel colour (ammonia process) were ambiguous,
    since they appeared to cover caramel colours manufactured by the
    ammonia sulfite process as well. Separate specifications were prepared
    for caramel colour (ammonia process) and caramel colour (ammonia
    sulfite process), which have since been designated caramel colours III
    and IV, respectively. The two classes have been shown to differ in
    toxicity. The principal toxic effect of caramel colour III is
    depression of circulating lymphocytes and leucocytes, and a no-effect
    level could not be determined; accordingly, the temporary ADI was
    revoked. The temporary ADI of 0-100 mg/kg b.w. was retained for
    caramel colour IV pending the submission of reports of adequate
    carcinogenicity/teratogenicity studies.

         The twenty-fourth meeting extended the temporary ADI for caramel
    colour IV and confirmed that caramel colour II had no ADI since it was
    not included in the ADI for caramel colour I nor the temporary ADI for
    caramel colour IV.

         Since the previous evaluations, additional data have become
    available and are summarized and discussed in the following
    monographs. Previously-published monographs have been expanded and are
    reproduced in their entirety under the class of caramel colour to
    which they relate.


    CARAMEL COLOUR I

    EXPLANATION

         At the thirteenth and fifteenth meetings (Annex 1, references 19
    & 26), the Committee concluded that caramel colour I is a natural
    constituent of the diet and is acceptable as an additive. An ADI "not
    limited" was allocated at the fifteenth meeting.

    BIOLOGICAL DATA

    Biochemical aspects

         No information available.

    Toxicological studies

    Special study on mutagenicity

         Two samples of caramel colour I with different colour intensities
    were subjected to the Ames test using Salmonella typhimurium strains
    TA98, TA100, TA1535, TA1537, and TA1538. Caramel colour I was neither
    mutagenic nor cytotoxic, either with or without activation by rat
    liver S-9 fraction, at concentrations up to 20 µl per plate (Richold &
    Jones, 1980a,b).

    Acute toxicity

         No information available.

    Short-term studies

    Rats

         Caramel colour I was administered to groups of 20 weanling female
    Wistar rats at dietary levels of 0, 15, or 30% for 8 weeks followed by
    a 4-week recovery period. Diarrhoea was observed in the treated
    animals and food efficiency was decreased, but the growth rate was
    normal. Haematological indices, in particular leucocyte counts, were
    normal throughout the study. The relative caecal weights were
    increased after eight weeks, but returned to normal by the end of the
    4-week recovery period. Discolouration of the mesenteric lymph nodes
    was observed in animals of both treatment groups after eight weeks,
    but the discolouration diminished during the recovery period. No other
    gross or microscopic pathological changes were reported (Sinkeldam &
    van der Heyden, 1976).

    Long-term studies

         No information available.

    Observations in man

         No information available.

    Comments

         Caramel colour I is free of the heterocyclic compounds associated
    with convulsant activity or depressed lymphocyte counts which occur in
    caramels prepared using ammonia or ammonium salts, and displays a low
    order of short-term toxicity. None of the data suggest a need to
    revise earlier JECFA recommendations.

    EVALUATION

    Estimate of acceptable daily intake for man

         ADI "not specified".

    REFERENCES

    Richold, M. & Jones, E. (1980a). Ames metabolic activation test to
         assess the potential mutagenic effect of ETA-38-2H. Unpublished
         report No. FDC 8/80359 from Huntingdon Research Centre,
         Huntingdon, England. Submitted to WHO by International Technical
         Caramel Association.

    Richold, M. & Jones, E. (1980b).  Ames metabolic activation test to
         assess the potential mutagenic effect of ETA-38-1W. Unpublished
         report No. FDC 8/80361 from Huntingdon Research Centre,
         Huntingdon, England. Submitted to WHO by International Technical
         Caramel Association.

    Sinkeldam, E.J. & van der Heyden, C.A. (1976).  Short-term feeding
         test with three types of caramels in albino rats. Unpublished
         report No. R4789 from CIVO/TNO, Zeist, The Netherlands. Submitted
         to WHO by International Technical Caramel Association.
    

    CARAMEL COLOUR II

    EXPLANATION

         The twenty-fourth meeting of the Committee (Annex 1, reference
    53) drew attention to the fact that caramel colour II has no ADI since
    it is not included in the ADI of caramel colour I. This material has
    not previously been evaluated by the Committee.

    BIOLOGICAL DATA

    Biochemical aspects

         No information available.

    Toxicological studies

    Special studies on mutagenicity

         Caramel colour II at concentrations of 2.5-20 µl/plate was
    neither mutagenic nor cytotoxic in the Ames test using Salmonella
    typhimurium strains TA98, TA100, TA1535, TA1537, and TA1538, with or
    without metabolic activation by rat liver S-9 fraction (Richold &
    Jones, 1980).

         In a similar test, caramel colour II was neither mutagenic nor
    cytoxic at concentrations of 50-5000 µg/plate (Richold et al.,
    1984).

         Caramel colour II was tested for potential mutagenic activity
    based on induction of DNA repair (unscheduled DNA synthesis) in
    cultured human epithelial (HeLa 53) cells. Caramel colour II
    was incorporated in the culture medium at concentrations of
    25-51,200 µg/ml and the test was performed on 2 occasions both in the
    presence and absence of rat liver S-9 mix. In both tests, in the
    absence of the S-9 mix a small but statistically significant increase
    in the number of silver grains over nuclei was observed at a
    concentration of 25,600 µg/ml; no significant increases were seen at
    higher concentrations in this test nor at any concentration in the
    repeat test (Allen & Proudlock, 1984).

         Caramel colour II was not clastogenic to cultured Chinese hamster
    ovarian cells at concentrations of 500, 2500, or 5000 µg/ml either in
    the presence or absence of rat liver S-9 mix (Allen et al., 1984).

    Acute Toxicity

         No information available.

    Short-term studies

    Rats

         Five groups of 20 male and 20 female weanling Fischer F-344 rats
    were given caramel colour II in drinking water at dose levels of 0, 4,
    8, 12, or 16 g/kg b.w./day for 90 days. Body weights and food and
    water intakes were recorded weekly. Haematological examinations, blood
    bichemical studies, and urinalysis were performed during the seventh
    week and at termination on all animals. Animals were fasted overnight
    prior to bleeding from the orbital sinus under anaesthesia; the
    animals were also fasted during the 24-hour urine collection period.
    At necropsy, organs were weighed and all animals were examined
    macroscopically. Histopathological examinations were conducted on all
    animals in the control and high-dose groups and on animals from the
    low- and mid-dose groups that died or were sacrificed in extremis.

         All animals showed a weight loss and reduced food intake between
    weeks 7 and 8, presumably due to the fasting prior to bleeding and
    during urine collection. In addition, caramel colour II caused a dose-
    related reduction in body-weight gain and in food and fluid intake in
    both sexes. Males treated with 8 g caramel colour II per kg b.w., and
    females treated with 4 and 8 g/kg b.w., had significantly lower food
    consumption than controls for 4 of 13 weeks, 3 of 13 weeks, and 6 of
    13 weeks, respectively; mean food consumption, of both males and
    females treated with 12 and 16 g/kg b.w., was significantly lower than
    food consumption of their respective controls. All treated groups had
    lower mean fluid intake than their respective controls, usually
    significantly lower. Males receiving 12 and 16 g/kg b.w. had
    significantly lower mean body weights than controls from weeks 7-13
    and weeks 4-13, respectively. Females treated with 8 g/kg b.w. had
    significantly lower mean body weights than controls from week 11-13
    while, beginning at week 7, females receiving 12 and 16 g/kg b.w. had
    lower mean body weights than controls.

         Occasional statistically-significant changes were seen in some
    haematological and clinical chemical parameters, but the mean values
    were within the normal range for Fischer 344 rats and were not
    considered to be of toxicological significance. There was a dose-
    related reduction in urinary volume and pH, and an increase in urine
    specific gravity.

         At necropsy, treatment-related increases were observed in kidney
    weights and in full and empty caecum weights, but no significant
    histopathological changes were observed in these or other tissues.
    Dose-related staining of the gastrointestinal tract and mesenteric
    lymph nodes was noted, and deposits of yellow pigment were seen
    histopathologically in the caecal submucosa and mesenteric lymph nodes
    of the top-dose groups (the only treated animals examined
    comprehensively). Reactive hyperplasia was not associated with the
    pigment deposition.

         The investigators concluded that the reduced body weights, food
    and fluid intakes, and increased kidney weights were due to water
    imbalance reflecting poor palatability of the drinking solution rather
    than toxic effects of caramel colour II per se (MacKenzie, 1985).

    Comments

         The above data are insufficient to evaluate caramel colour II for
    an ADI, but the substance has a low sub-chronic toxicity and no
    frankly pathological effects were seen in the 90-day study.

    EVALUATION

    Estimate of acceptable daily intake for man

    No ADI allocated.

    REFERENCES

    Allen, J.A., Brooker, P.C., Birt, D.M., & McCaffrey, K.J. (1984).
         Analysis of metaphase chromosomes obtained from CHO cells
         cultured in vitro and treated with caramel colour (II).
         Unpublished report No. ITC 3A/84965 from Huntingdon Research
         Centre, Huntingdon, England. Submitted to WHO by International
         Technical Caramel Association.

    Allen, J.A., & Proudlock, R.J. (1984). Autoradiographic assessment of
         DNA repair in mammalian cells after exposure to caramel colour
         II. Unpublished report No. ITC 2A/84750/2 from Huntingdon
         Research Centre, Huntingdon, England. Submitted to WHO by
         International Technical Caramel Association.

    MacKenzie, K.M. (1985). 90-day toxicity study of caramel color
         (caustic sulfite process) in rats. Vols. I & II. Unpublished
         report No. 6154-105 from Hazleton Laboratories America Inc.,
         Madison, WI, USA. Submitted to WHO by International Technical
         Caramel Association.

    Richold, M. & Jones, E. (1980). Ames metabolic activation test to
         assess the potential mutagenic effect of ETA-38-3N. Unpublished
         report No. FDC 8/80356 from Huntingdon Research Centre,
         Huntingdon, England. Submitted to WHO by International Technical
         Caramel Association.

    Richold, M., Jones, E., & Fenner, L.A. (1984).  Ames metabolic
         activation test to assess the potential mutagenic effect of
         caramel colour II. Unpublished report No. ITC 1A/84708 from
         Huntingdon Research Centre, Huntingdon, England. Submitted to WHO
         by International Technical Caramel Association.
    

    CARAMEL COLOUR III

    EXPLANATION

         In earlier evaluations, the presence of 4-methylimidazole in
    caramel colour III was noted, particularly since this compound was
    likely to be the causal agent of convulsions in cattle and sheep fed
    ammonia-treated molasses. However, at its twenty-first meeting, the
    Committee (Annex 1, reference 44) no longer considered this a cause
    for concern since the introduction of chemical specifications limits
    the concentration of 4-methylimidazole in caramel colour III. The
    twenty-first meeting identified the principal toxic effect of
    ammoniated caramels as the depression of circulating lymphocytes and
    total leucocytes for which a no-effect level had not been determined;
    consequently, the temporary ADI for caramel colour III was revoked.

    BIOLOGICAL DATA

    Biochemical aspects

    Absorption, distribution, and excretion

         In groups of 2-4 rats, the absorption of the colour-giving
    components of caramel was determined by faecal extraction. Recoveries
    varied widely for the 10 or 20% caramel solutions examined despite
    pre-treatment for 100 days before testing. About one-third of the
    colour-giving components appeared to be absorbed, but no conclusions
    could be drawn regarding the absorption of colourless components
    (Haldi & Wynn, 1951).

    Toxicological studies

    Special study on 4-methylimidazole (4-MeI)

         4-MEI has been shown to be the most likely toxic component in
    ammoniated molasses, its being a convulsant to rabbits, mice, and
    chicks at oral doses of 360 mg/kg b.w. (Nishie et al., 1969).

    Mice

         Male albino mice (20-25 g) were used to determine the median
    convulsive dose (CD50) and the median lethal dose (LD50) of a few
    imidazoles. The results are given in the following table:

    Convulsant and lethal effects of imidazoles
 
                                                                          

                             CD50±SE                   LD50±SE
                             in mg/kg                in mg/kg b.w.
                         b.w i.p     oral          b.w. i.p    oral
                                                                          

    4-methylimidazole    155 ± 5     360 ± 18      165 ± 3     370 ± 15
    1-methylimidazole    380 ± 8.2   1,400 ± 79    380 ± 8.2   1,400 ± 79
    2-methylimidazole    500 ± 12    1,300 ± 70    480 ± 18    1,400 ± 114
    imidazole            560 ± 34    1,800 ± 45    610 ± 7.4   1,880 ± 45
                                                                          
 
         All the imidazoles tested produced varying degrees of tremor,
    running, restlessness, dialorrhea, Straub tail, opisthotonus and tonic
    extensor seizure that ended in death (Nishie et al., 1970).

    Chickens

         The CD50 and the LD50 of 4-MEI by i.p. injection in 1-day-old
    chicks were 174±10 mg/kg b.w. and 210±15 mg/kg b.w., respectively.
    Orally, the CD50 was 580±30 mg/kg b.w. and the LD50 was 599±50 mg/kg
    b.w. Doses of 100 mg/kg b.w. i.p. caused tremors, peeping, and
    spreading of the wings. Doses over 150 mg/kg b.w. i.p. caused
    opisthotonus, prostration with clonic leg movements, and terminal
    tonic extensor seizure (Nishie et al., 1970).

    Special studies on reduction of total lymphocyte counts

         Following the request in the twenty-first report of the Committee
    for information on the factor(s) responsible for the haematological
    effects of caramel colour III, a series of studies have been performed
    to investigate the mechanism(s) underlying these effects and the
    agent(s) responsible.

    Rats

         The effects of vitamin E, folic acid, pyridoxine and choline on
    the capacity of caramel colour III to reduce total lymphocytes in the
    blood of rats fed caramel colour III were examined. Four groups of 10
    male weanling rats were fed Spratt's diet containing 8% caramel colour
    III and supplemented with vitamin E (100 mg/kg), folic acid
    (10 mg/kg), pyridoxine hydrochloride (10 mg/kg), or choline chloride
    (1000 mg/kg). Groups on Spratt's diet alone or on Spratt's diet with
    8% caramel colour III without any supplement were used as controls.
    After 12 days, there was a marked reduction in total white blood cells
    and lymphocytes in rats fed the diet containing 8% caramel colour III
    and neutrophil counts were increased. Dietary supplementation with
    vitamin E, folic acid, or choline did not noticeably affect lymphocyte

    counts. However, rats receiving the diet supplemented with pyridoxine
    had white blood cells and lymphocytes in numbers similar to those fed
    the basal diet without caramel colour III. The basal diet was found to
    contain 2.3 mg/kg pyridoxine.

         In a further study using CIVO stock diet (pyridoxine
    concentration 3 mg/kg), rats receiving 8% caramel colour III in the
    diet showed a reduction in total white blood cell and lymphocyte
    numbers; these changes were ameliorated by the addition of 10 mg/kg
    pyridoxine in the diet. In this study, neutrophils were not affected
    by the caramel colour III treatment, but the plasma pyridoxal
    phosphate levels were reduced by treatment (Sinkeldam et al., 1984).

         In order to quantify the relationship between dietary pyridoxine
    and reduction of lymphocyte counts by caramel colour III, and to study
    the effects of age, two 14-day studies were performed, one in
    weanling, and the second in mature, Wistar rats; in other respects the
    protocols were identical. Groups of 10 male animals were fed diets
    containing approximately 2.5, 6, 12, or 24 ppm of pyridoxine. At each
    dietary level groups were given caramel colour III1 in the drinking
    water at levels of 0, 1, 4, or 8%. A clear inverse dose relationship
    between the severity of lymphocyte depression and the pyridoxine level
    of the diet was observed. In weanling rats fed a diet containing
    2.5 ppm pyridoxine, statistically significant lymphocyte reduction
    occurred at all caramel colour III levels on days 6 and 13. Groups fed
    diets containing 6, 12, or 24 ppm pyridoxine did not show
    statistically significant reductions in lymphocyte counts on day 6.
    However, on day 13, groups receiving 4 or 8% caramel colour III and 6
    ppm pyridoxine in the diet had a statistically significant reduction
    in lymphocytes. On day 13 at a dietary level of 12 ppm pyridoxine only
    the 8% caramel colour III group had a statistically significant
    reduction in lymphocytes. There were no significant reductions in
    lymphocyte counts in animals fed 24 ppm pyridoxine at any level of
    caramel colour III intake. In mature rats fed 4 and 8% caramel colour
    III a statistically significant reduction in lymphocytes occurred at
    dietary pyridoxine levels of 12 ppm and lower on day 6. On day 13
    lymphocytes were significantly decreased at caramel colour III doses
    1, 4, and 8% in animals fed a diet containing 2.5 ppm pyridoxine.
    There were no significant decreases in lymphocytes at these dose
    levels of caramel colour III in animals fed 6, 12, or 24 ppm
    pyridoxine (Sinkeldam, 1981; 1982a).

              

    1  This sample of caramel colour III contained 107 mg/kg
       2-acetyl-4(5)-tetrahydroxybutylimidazole (THI) on a colour
       equivalent basis, 204 mg THI/kg on an 'as is' basis, and 295 mg
       THI/kg on a solids basis.

    Special studies on 2-acetyl-4(5)-tetrahydroxybutylimidazole (THI)

         An isolation procedure was developed and a fraction of caramel
    colour III that contained the lymphocyte-depressing activity was
    isolated. The single component of this fraction that was responsible
    for the activity was identified as THI (Kroplien et al., 1984).

    Rats

         THI was administered to groups of 10 male weanling Wistar rats at
    levels of 0, 2, 5, or 20 ppm in drinking water for 7 days; a fifth
    group received a 1% solution of caramel colour III in drinking water
    as a positive control. THI produced a marked depression of lymphocyte
    counts at all dose levels and a dose-dependent increase in
    neutrophils. The lymphocyte-depressing potency of 2 ppm THI was
    comparable to that of 1% caramel colour III in the drinking water,
    indicating a level of approximately 200 mg/kg THI in the caramel
    sample. This compares with the result of chemical analysis of a sample
    of this batch of caramel colour III, which indicated a value of
    204 mg/kg THI on an 'as is' basis (Sinkeldam, 1982b; Kroplien, 1984).

    Special studies on carcinogenicity (see also long-term studies)

    Mice

         Three groups of 50 male and 50 female B6C3F1 mice were given
    drinking water containing 0, 1.25, or 5% caramel colour III for 96
    weeks followed by 8 weeks of drinking water without caramel colour;
    the animals were fed CFR diet ad libitum. The caramel colour III
    used in this study contained less than 25 mg/kg THI.

         The animals were observed daily for abnormalities and mice
    showing signs of ill-health were isolated to be returned to the group
    if their condition improved but otherwise killed and autopsied.
    Individual body weights were recorded weekly for the first 14 weeks,
    then every-other-week. Food and water consumption were recorded over a
    2-day period before each weighing. During week 104, fresh urine
    samples were obtained from all survivors and analysed for pH, protein,
    glucose, bilirubin, ketones, occult blood, and urobilinogen. At
    termination the animals were killed by exsanguination under ether
    anaesthesia and haematological examinations were performed, which
    consisted of measurements of haemoglobin concentrations, haematocrit
    values, erythrocyte counts, leucocyte counts, platelet counts, and
    differential leucocyte counts. At autopsy, gross findings were
    recorded and the following organs weighed: brain, heart, liver,
    spleen, adrenals, and testes or ovaries. Samples of these organs and
    of the salivary gland, trachea, lungs, thymus, lymph nodes, stomach,
    small intestine, pancreas, urinary bladder, pituitary, thyroid,
    prostate, seminal vesicle, uterus, mammary gland, skeletal muscle,

    eye, Harderian glands, spinal cord, sciatic nerve, and any other
    tissues of abnormal appearance were examined histologically
    (haematoxylin- and eosin-stained). Histopathological examinations were
    also performed on mice that died and on those that were killed in
    moribund condition. During the in-life phase no consistent differences
    were noted between the test and control groups with respect to growth
    or water intake.

         The cumulative mortality of males given 5% caramel colour III was
    higher than that of controls from week 100 to the end of the
    experiment, but there were no clear pathological differences in any
    organs and no treatment-related abnormalities in urinalyses.

         The only statistically significant differences in haematological
    parameters between control and treated groups were elevations of the
    total leucocyte counts in males of both treatment groups, but the
    observed values were within the range encountered for the strain of
    B6C3F1 mice used in the study. No treatment-related gross pathology
    was noted during or at the end of the experiment. Malignant lymphoma/
    leukaemia, hepatocellular carcinoma, and sub-cutaneous fibrosarcoma
    and/or malignant fibrous histocytoma were frequent, but no significant
    differences were found in their incidences between treated and control
    groups. Adenomas and adenocarcinomas of the lungs, hyperplastic
    nodules of the liver, and fibroma of the sub-cutis were frequent in
    males, but their incidences were similar in treated and untreated
    mice.

         The authors concluded that, under the conditions used, caramel
    colour III was not carcinogenic for B6C3F1 mice (Hagiwara et al.,
    1983).

    Rats

         Three groups of 50 male and 50 female F344 rats were given
    caramel colour III in drinking water at levels of 0, 1, or 4% for 104
    weeks followed by drinking water without caramel for 9 weeks. The
    animals were fed ad libitum basal diet that contained 11-12 mg/kg
    pyridoxine, and the caramel colour III used contained less than
    25 mg/kg THI. During the experimental period, all animals were
    observed daily, and clinical signs and mortality were recorded. Body
    weights were recorded weekly during the first 13 weeks of the study
    and then every 4 weeks. Moribund or dead animals and animals
    sacrificed at termination were autopsied and examined for the
    development of tumours in the following organs and tissues: brain,
    pituitary, thyroid (including parathyroid), thymus, lungs, trachea,
    heart, salivary glands, liver, spleen, kidneys, adrenals, tongue,
    oesophagus, stomach, duodenum, jejunum, ileum, caecum, colon, rectum,
    urinary bladder, lymph nodes, pancreas, gonads, accessory genital
    organs, mammary gland, skin, musculature, peripheral nerve, spinal
    cord, sternum, femur, eyes, ear duct, and nasal cavity.  These tissues
    were also examined histologically (haematoxylin- and eosin-stained).

         No treatment-related differences in growth or survival rates were
    noted. No dose-related effects were found either in the incidence or
    induction time of tumours in the various organs and tissues except in
    the pituitary gland of males of the top-dose group in which the
    incidence of tumours was significantly higher than that in controls.
    However, pituitary tumours are among the most common spontaneous
    tumours in F344 rats that occur with variable incidence; most of the
    tumours were microscopic and there were no significant differences in
    their induction times compared with controls. The authors concluded
    that the higher incidence of pituitary tumours was not related to
    caramel administration, but could be explained by the variability of
    spontaneous tumour incidence (Maekawa et al., 1983).

    Special studies on mutagenicity

         Caramel colour III (a blend of 3 commercial samples) was
    evaluated by the Ames Salmonella/microsome plate test and the
    Saccharomyces/microsome plate test. The Salmonella test organisms
    used were TA98, TA100, TA1535, TA1537, and TA1538; the Saccharomyces
    was strain D4, and the tests were conducted in the presence and
    absence of liver S-9 mix from Araclor-induced rats. The dose range
    employed was 1-50 mg/plate for TA100 and 1-20 mg/plate for all other
    test strains. Caramel colour III was non-mutagenic to the test
    organisms under these conditions (Jagannath & Brusick, 1978a).

         In a similar study using the same Salmonella tester strains,
    caramel colour III was non-mutagenic in a range of concentrations up
    to 20 µl/plate with or without metabolic activation (Richold & Jones,
    1980).

         Caramel colour III was non-mutagenic in the Ames test against
    Salmonella strains TA98 and KTA100 at concentrations of 
    0-1 mg/plate, with or without metabolic activation. Acidic and basic
    organic extracts of the caramel were also non-mutagenic (Ashoor &
    Monte, 1983).

         Thirteen commercial caramel colours (not identified) were
    examined for mutagenicity in the Ames test, using Salmonella strains
    TA98 and TA100 with and without metabolic activation, and for
    DNA-damaging effects in E. coli (Wild/pol A-, Wild/rec A-); none
    of the samples tested was active under the test conditions (Kawana
    et al., 1980).

         Five samples of commercial caramel colour (not identified) were
    tested for mutagenicity in the Ames test using Salmonella strains
    TA98 and TA100. All samples gave equivocal results with TA100 and 2 of
    the 5 samples were equivocal with TA98 (Kawachi et al., 1980).

         Five samples of caramel colour were tested in the Ames
    Salmonella plate assay using strains TA98, TA100, and TA1537 without
    metabolic activation but with a 20-minute preincubation step. The same
    samples were used in chromosome aberration tests performed on a
    cultured Chinese hamster lung fibroblast cell line, both in the
    presence and absence of S-9 fraction. All samples of caramel colour
    were designated as positive both in the Ames assay and in the
    chromosome aberration test (aberrations in 20% of metaphase cells)
    (Ishidate & Yoshikawa, 1980).

         The same series of caramel colour samples was tested by Jagannath
    and Brusick, and all were found to be non-mutagenic in the Ames test
    using Salmonella strains TA98, TA100, TA1535, TA1537, and TA1538,
    and Saccharomyces strain D4, in the presence or absence of S-9 mix.
    One of the samples in this study and in the Ishidate & Yoshikawa
    (1980) study was caramel colour III (Jagannath & Brusick, 1978b).

         The mutagenicity of a series of samples taken at various stages
    in the manufacture of caramel colour III were assayed in the Ames test
    using Salmonella strains TA98, TA100, and TA1535, with and without
    metabolic activation. No mutagenicity was seen with any sample against
    all three tester strains in the presence of S-9 fraction but and
    increased number of revertants was found with strain TA100 in the
    absence of S-9 fraction. Mutagenic activity was associated with
    samples taken late in the process (Jensen et al., 1983).

         A sample of caramel colour III was non-mutagenic in the Ames test
    against Salmonella strains TA98, TA100, TA1535, TA1537, and TA1538
    both in the presence and absence of rat hepatic S-9 fraction at
    caramel colour concentrations up to 5 mg/plate (Richold et al.,
    1984).

         The same sample as that used by Richold et al. (1984b) was
    tested for potential mutagenic activity based on induction of
    unscheduled DNA synthesis in cultured human epithelial (HeLa 53) cells
    both in the presence and the absence of rat liver S-9 fraction. The
    test was performed on two occasions at caramel colour III
    concentrations of 25-51,200 µg/ml in the culture medium. In both
    tests, caramel colour III caused a significant increase in the number
    of silver grains found over cell nuclei at concentrations of 6,400 and
    12,800 µg/ml in the absence of S-9 mix; significant increases were not
    observed at higher or lower concentrations in the absence of S-9 mix
    nor at any of the concentrations tested in the presence of S-9 mix.
    The authors concluded that, although reproducible effects on
    unscheduled DNA synthesis were demonstrated in the test, this effect
    may be moderated by metabolism (Allen & Proudlock, 1984).

         Caramel colour III was not clastogenic to cultured Chinese
    hamster ovarian cells at concentrations of 500, 2500, or 5000 µg/ml
    either in the presence or absence of rat liver S-9 mix (Allen
    et al., 1984).

         The clastogenicity of caramel colour III was evaluated in an
    in vitro cytogenetic assay using cultured Chinese hamster ovarian
    cells. A significant increase in chromosome aberrations was observed
    at concentrations of 3 mg/ml or higher in the absence of metabolic
    activation; similar findings were reported for sodium ascorbate at
    2 mg/ml (Galloway & Brusick, 1981a).

         These tests were repeated in the presence of rat liver S-9
    fraction, and neither caramel colour III at a concentration up to
    5 mg/ml nor sodium ascorbate at a concentration of 5 mg/ml was active
    (Galloway & Brusick, 1981b).

         In an in vivo mouse micronucleus test, caramel colour III was
    administered by gavage to 5 males and 5 females at doses of 0, 1.05,
    or 3.5 g/kg b.w. (2 doses 24 hours apart). Caramel colour III did not
    increase the incidence of micronuclei in polychromatic erythrocytes
    obtained from marrow and was considered by the authors not to exhibit
    clastogenic activity under the conditions of the test (Cimino &
    Brusick, 1981).

    Special studies on teratogenicity

         Teratogenicity studies were carried out on caramel colour III in
    mice, rats, and rabbits. The doses employed were 0, 16, 74.3, 345, and
    1600 mg/kg b.w. in all three species.

    Mice

         Caramel colour III was administered by gavage to groups of 22 or
    23 pregnant CD1 mice at the above doses beginning on day 6 and
    continuing through day 15 of gestation. On day 17, all dams were
    subjected to Caesarian section under anaesthesia and the numbers of
    implantation sites, resorption sites, and live and dead foetuses were
    recorded. The urogenital tract of each dam was examined for anatomical
    abnormalities and all foetuses were examined grossly for external
    abnormalities. One-third of the foetuses from each litter were
    examined for visceral abnormalities using the Wilson technique, and
    the remaining two-thirds were examined for skeletal defects after
    clearing and staining with Alizarin Red. Caramel colour III had no
    clearly discernible effects on nidation nor on maternal or foetal
    survival. The number of abnormalities of either soft or skeletal
    tissues of the test groups did not differ from those occurring
    spontaneously in controls (Morgareidge, 1974).

    Rats

         Caramel colour III was administered by gavage to groups of 21-24
    pregnant Wistar rats at the above doses beginning on day 6 and
    continuing daily through day 15 of gestation. On day 20, all dams were
    subjected to Caesarian section under anaesthesia and the numbers of

    implantation sites, resorption sites, and live and dead foetuses were
    recorded. The body weights of live pups were also recorded. All
    foetuses were examined grossly for external abnormalities. One-third
    of the foetuses from each litter were examined for visceral
    abnormalities using the Wilson technique, and the remaining two-thirds
    were examined for skeletal defects after clearing and staining with
    Alizarin Red. Caramel colour III had no clearly discernible effect on
    nidation nor on maternal or foetal survival. The number of
    abnormalities seen in either soft or skeletal tissues of the test
    groups did not differ from those occurring spontaneously in controls
    (Morgareidge, 1974).

    Rabbits

         Caramel colour III was administered by gavage to groups of 11 or
    12 pregnant Dutch-belted female rabbits at the above doses beginning
    on day 6 and continuing daily through day 18 of pregnancy. On day 29,
    all does were subjected to Caesarian section under anaesthesia and the
    numbers of corpora lutea, implantation sites, resorption sites, and
    live and dead foetuses were recorded. All foetuses were examined
    grossly for the presence of external congenital abnormalities. The
    live foetuses from each litter were then placed in an incubator for 24
    hours for evaluation of neonatal survival. Surviving pups were
    sacrificed and all pups examined for visceral abnormalities by
    dissection. All foetuses were then cleared and stained with Alizarin
    Red and examined for skeletal defects. Caramel colour III treatment
    had no effect on nidation nor on maternal or foetal survival. The
    number of abnormalities in either soft or skeletal tissues of the test
    groups did not differ from those occurring spontaneously in controls
    (Morgareidge, 1974).

    Acute toxicity
                                                                        

                      LD50
    Species   Route   (mg/kg b.w.)                    Reference
                                                                        

    Rat       oral    > 2.3 ml approx. or eq. 1,900   Foote et al., 1958
    Rat       oral    > 25 ml approx. or eq. 17,500   Chacharonis, 1960
    Rat       oral    > 30 ml approx. or eq. 20,400   Chacharonis, 1963
                                                                        

         No treatment-related effects were detected during the observation
    of animals for 14 days after administration of single doses of 12
    different caramels manufactured with ammonia or sulphate ammonia
    catalysts (Foote et al., 1958; Chacharonis, 1960; 1963). Single
    doses of caramel colour III of up to 10 g/kg b.w. in mice and 15 g/kg
    b.w. in rabbits did not cause convulsions or others signs of distress
    (Sharratt, 1971).

    Short-term studies

    Mice

         In a 4- to 6-week study, albino Swiss mice (10 males and 10
    females/group) were fed caramel colour III containing 830 ppm
    4-methylimidazole at concentrations of 0, 1, 2, 4, 8, or 16% in the
    diet. No influence of caramel colour III on appearance, behaviour, or
    food intake was observed. Growth was decreased, especially in the
    third and fourth weeks, in the groups fed 16% caramel colour III. The
    faeces of the animals fed the higher dose levels were soft, tarry in
    appearance, poorly-formed, and sticky or pasty in consistency.

         In the males fed 16% caramel colour III, an increase of
    neutrophilic leucocytes and a decrease in lymphocytes were observed.
    The mean relative weights of the caeca (full and empty) were increased
    at the 4, 8, and 16% dietary levels. No other remarkable findings were
    observed on gross examination; histopathological examinations were not
    conducted. No information is available on the THI content of the
    sample of caramel colour III used in this study, nor on the dietary
    levels of pyridoxine (Procter, 1976).

         In a pilot study to select the dose levels for a carcinogenicity
    study, B6C3F1 mice were fed caramel colour III for 13 weeks. No
    details of this study were reported (Hagiwara et al., 1983).

    Rats

         Four groups of 10 male and 10 female rats received either 0 or
    10% of 2 different samples of caramel colour III in their diet for 90
    days. Weight gains showed slight reductions compared with controls,
    but food consumption was normal in all groups. No abnormalities were
    noted regarding haematology, urinalysis, gross pathology, or
    histopathology. No information is available on the THI content of the
    samples of caramel colour III used in this study (Charcharonis, 1963).

         Four groups of rats received 0, 4, 8, or 16% caramel colour III
    in their diet for 3 months. No convulsions or other behavioural
    abnormality or signs of neurological damage were seen. No macroscopic
    pathological abnormalities were found in the central nervous system
    (Sharratt, 1971).

         Six groups of 15 rats (CFE strain) of each sex were fed diets
    containing 4, 8, or 16% caramel colour III produced by either an
    "open" or a "closed" pan process for 13 weeks; a group of 25 rats of
    each sex served as controls. Body-weight gain was decreased at all
    dietary levels of both caramel colours. Haemoglobin concentrations
    were reduced at the highest dietary levels at week 6, while at the
    lower levels this effect was less clear. After 13 weeks, the males
    at all dose levels had significantly decreased haemoglobin

    concentrations, but this effect occurred in females only at the 8 and
    16% levels. In some groups there was a less consistent decrease in the
    total number of red blood cells at 6 and 13 weeks. The total number of
    leucocytes was significantly decreased and a lymphocytopenia was
    present at all dose levels at 6 and 10 weeks; at 13 weeks these
    changes were observed only in males. At necropsy, decreased weights of
    the thymus and spleen were observed at the 8 and 16% dose levels. The
    caecal weights were increased at the 8 and 16% dose levels compared
    with controls. Increased relative liver and kidney weights suggested
    an effect on these organs. Changes in the weights of other organs were
    considered to be related to the differences of body weight between the
    groups. The volumes of urine excreted during a 6-hour period without
    water or in the 2-hour period after a water load were lower than the
    control values. The latter differences were accompanied by higher
    values for the specific gravity of the urine. The histopathological
    study did not reveal treatment-related changes. No details of the THI
    content of the caramel colours used in this study were available
    (Gaunt et al., 1975).

         Seven groups of 20 female Wistar rats were fed diets containing
    0, 15, or 30% of 3 types of caramel colour, one of which was a sample
    of caramel colour III, for 8 weeks followed by a 4-week recovery
    period. The diets containing caramel colour III caused dose-related
    decreases in body weights and food efficiency, and also caused
    diarrhoea at the 30% dietary level. Leucocyte counts at 4 weeks were
    significantly increased at the highest dose level whereas at 8, 10,
    and 12 weeks no differences were observed between test groups and
    controls. The relative weights of the caeca of animals receiving
    caramel colour III, both filled and empty, were increased at 4 and 8
    weeks but reverted to normal after the 4-week recovery period.

         Gross examination at autopsy after 8 weeks revealed a slight,
    dose-related brown discolouration of the mesenteric lymph nodes in a
    few animals of each test group. After recovery periods of 2 or
    4 weeks the discolouration was less intensive, but still visible.
    Microscopically, the lymph nodes of the test rats showed accumulation
    of pigment-laden macrophages, which was not noticeably diminished
    after withdrawal of the caramel for 2 or 4 weeks. No details of the
    THI content of the caramel colour III used in this study were
    available (Sinkeldam & van der Heyden, 1976a).

         In a 10-week feeding study, groups of 15 male and 15 female
    Sprague-Dawley rats were fed diets containing caramel colour III at
    concentrations of 0, 1.25, 2.5, 5.0, 10.0, or 15.0%. In the animals
    receiving caramel colour III at levels of 5% or higher, the faeces
    became soft within two weeks and the water content of the faeces was
    higher than the faeces of controls. Body-weight gains were generally
    reduced in animals fed caramel colour III, particularly during the
    last 2-4 weeks of the study.

         In rats fed caramel colour III, there were no changes in
    haemoglobin levels or erythrocyte counts; a significant reduction in
    lymphocyte counts and a coincident increase in the number of segmented
    neutrophils was observed at all dose levels. No macroscopic evidence
    of abnormal pigmentation of the mesenteric lymph nodes was found. An
    increase in empty caecal weight was consistently evident in animals of
    both sexes given caramel colour III at the 5, 10, and 15% levels.

         Histopathological examination did not reveal changes in the
    structure of the ileal or caecal mucosa nor in the reticuloendothelial
    components of the central or peripheral systems. No abnormal
    pigmentation of the lymph nodes was found. No details of the THI
    content of the caramel colour used in this study were available
    (Procter et al., 1976).

         Groups of 10 male and 10 female Wistar rats were fed diets
    containing caramel colour III at concentrations of 0, 1.25, 2.5, 5.0,
    10.0, or 15.0% for 10 weeks. Caramel colour III caused loose stools,
    particularly at the 5% dietary level and body weights were slightly
    decreased in both sexes at this level. Leucocyte (lymphocyte) counts
    were decreased in males and in females fed 15% caramel colour III; at
    lower dose levels this effect occurred only in females. The relative
    weight of the caecum (both full and empty) was increased by feeding
    caramel colour III. Minimal amounts of pigment were observed in
    mesenteric lymph nodes of several rats fed 1.25% and higher levels of
    caramel colour III. From this experiment it appeared that caramel
    colour III was more active than two other types of caramel tested
    concurrently in regard to growth depression, enlargement of the
    caecum, and decrease in leucocyte counts. No details of the THI
    content were available (Sinkeldam & van der Heyden, 1976b).

         In a 10-week study, weanling Wistar rats were fed caramel colour
    III in the diet at levels of 0, 0.5, 1.0, 2.0, 4.0, or 16%. There were
    15 males and 15 females in each group except for the control group,
    which had 60 animals of each sex, and the highest dose group, which
    had 10 animals of each sex. Food intake and growth rates were
    recorded, and haematological examinations and histopathological
    studies were carried out. In particular, the lymph nodes, thymus,
    spleen, and caecum were examined for distribution of pigment.

         Another 2 groups of 10 rats were given basal diet or diet
    containing 16% caramel colour III for 10 weeks followed by a 28-day
    period during which the basal diet was fed (recovery experiment). Five
    rats of each sex were killed after 7 days and the remainder after 28
    days of the recovery period.

         Caramel colour III depressed body-weight gain at dietary levels
    greater than 1%. Total leucocyte counts were decreased in males in the
    groups receiving 2, 4, and 16% caramel colour III and in females
    receiving 4 and 16% caramel colour III. However, the lymphocyte/

    neutrophil ratio was significantly decreased in both sexes at all dose
    levels. Relative liver weights were increased at dietary levels of 2%
    and higher levels of caramel colour III, and reduced spleen weights
    were observed at the highest (16%) dose level. Increased relative
    kidney weights were observed in both sexes at 2% and higher dietary
    levels of caramel colour III. Increased caecal weights were observed
    at the 16% dietary level; microscopically, pigment was observed in the
    mesenteric lymph nodes of male and female rats at this dose level.

         During the recovery phase, white cell counts, cell ratios, and
    total numbers of lymphocytes rapidly returned to normal; recovery was
    complete in both sexes by 7 days. The caecal weights had returned to
    normal by 7 days and the relative liver and spleen weights returned to
    normal during the recovery phase, while kidney weights partially
    recovered. No details of the THI content of the caramel nor of the
    pyridoxine content of diet were available (BIBRA, 1977).

         Test groups of 15 male and 15 female Sprague-Dawley rats were fed
    diets containing 10 or 15% caramel colour III for 4 weeks; a control
    group of 20 rats of each sex received basal diet. During the course of
    the study, the rats fed diets containing caramel colour III had soft
    dark-coloured faeces, particularly at the highest dietary level. There
    were no consistent differences in body-weight gain or food consumption
    and no mortality occurred. Haematological studies were conducted prior
    to feeding caramel colour III and after 2 and 4 weeks. In males, total
    white cell and lymphocyte counts were unaffected by treatment at
    either sampling interval or dose level. In females, total white cell
    and lymphocyte numbers were significantly depressed after 4 weeks but
    not at 2 weeks at both dose levels. However, significant differences
    in differential white cell counts were observed in both sexes. In
    males, the differential lymphocytes (%) were significantly depressed
    at both dose levels and treatment intervals, and there was a
    concomitant increase in segmented neutrophils. In females, the
    differential counts of lymphocytes and segmented neutrophils were
    similarly affected after 4 weeks.

         At necropsy, increased caecal weights were observed in both sexes
    and a statistically significant increase in relative weights of the
    thymus in male rats fed both doses of caramel colour III was noted.
    Histopathological studies were not conducted. Details of the THI
    content of the caramel sample used were not available (Procter, 1977).

         In a 13-week toxicity study, caramel colour III was administered
    in drinking water at concentrations of 0, 4, 6, 8, or 10% to groups of
    10 male and 10 female weanling Wistar rats. The caramel colour III
    sample used was analysed and found to contain 78 mg/kg THI on an 'as
    is' basis, 105 mg/kg THI on a solids basis. The diet fed to the rats
    contained approximately 13 mg/kg pyridoxine. The general condition and
    behaviour of the animals was checked at frequent intervals, individual
    body weights were measured weekly, food consumption was measured (on a

    cage basis) over weekly periods during the whole experimental period,
    and fluid consumption was measured daily (on a cage basis). Samples of
    blood for haematology were collected from the tip of the tail of all
    rats initially and at days 29/30, 57/58, and 84/85. Urinalysis was
    performed on individual urine samples from all animals during the last
    16 hours of a 24-hour period of deprivation of food and water at day
    87. Tail tip blood collected at day 87 was examined for glucose and
    urea nitrogen. Blood was obtained from the aorta under ether
    anaesthesia at termination (day 91 for males, 92 for females), and the
    following assays performed on the plasma; alkaline phosphatase, GPT,
    GOT, lactate dehydrogenase, total protein, and albumin. Pyridoxal
    phosphate was measured in plasma from EDTA-treated blood samples. At
    autopsy the following organs were weighed: thyroid, adrenals,
    testes/ovaries, kidneys, thymus, brain, spleen, heart, liver, and
    caecum (full and empty). Histological examinations were carried out on
    all animals of the control and top-dose groups and, in addition to the
    weighed organs, the following organs/tissues were examined: aorta,
    axillary and mesenteric lymph nodes, cervix, colon, oesophagus,
    stomach, duodenum, ilium, jejunum, lungs, epididymides, pituitary,
    prostate, skeletal muscle, skin, sternum, pancreas, trachea, urinary
    bladder, uterus, and all gross lesions.

         There was a dose-related decrease in fluid consumption, decreased
    urinary output of more concentrated urine, decreased food consumption,
    and decreased body-weight gain in both sexes. These changes were
    related to the palatability of the drinking fluid. No outstanding
    differences were observed in red blood cell analyses between test
    groups and controls. Total lymphocyte counts were relatively low in
    all test groups of both sexes. The differences from control values
    were statistically significant in all male treatment groups after
    29/30 days, in the female 8% group after 29/30 days, and in the male 4
    and 8% groups after 57/58 days. However, these differences did not
    show a clear dose-dependent relationship and at 13 weeks there were no
    significant differences among any of the treatment groups compared to
    controls.

         At necropsy, the relative weights of the kidneys and caecum were
    increased in several groups receiving caramel colour III. The increase
    in kidney weight was dose-dependent, but no treatment-related
    histopathological changes were seen in any of the organs examined and
    the enlargement of the kidneys was attributed to the decreased fluid
    consumption by the treated rats. No effects on the relative weights of
    spleen or thymus were observed and these tissues were histologically
    normal (Sinkeldam et al., 1980b).

         A 13-week toxicity study was conducted with caramel colour III
    given in the drinking water at concentrations of 0, 5, 10, 15, or 20%
    to groups of 10 male and 10 female weanling Wistar rats. The caramel
    colour III sample used in this study contained 0-3 mg/kg THI on an 'as
    is' basis. The diet fed to the rats contained 13.5 mg/kg pyridoxine.

    The protocol used was similar to that described in the previous study
    except that blood samples were collected at days 30/31, 57/58, and
    80/81, and urinalysis was performed at day 85.

         Water intake (fluid intake corrected for caramel colour solids
    content) showed a dose-related decrease in both sexes. Food intake was
    also generally lower in all treated groups. Body-weight gains of males
    were decreased in a dose-related manner, while body-weight gains of
    females were comparable to controls. Urine volumes were decreased in
    rats treated with caramel colour III and the urines were more
    concentrated in treated groups than in controls. These changes were
    attributed to the decreased water intake. The urine was darker-
    coloured at the 15 and 20% dose levels; however, urine composition was
    essentially normal. Lymphocyte counts were in general relatively low
    in all test groups in both sexes but the differences only attained
    statistical significance in males of the top-dose group after 30 days
    and males of all treatment groups after 57 days. No significant
    differences in lymphocyte counts were observed in males of any group
    after 80 days, nor in females at any dose level at any time interval.
    Mean neutrophil counts were significantly increased in females
    receiving 20% caramel colour III for 81 days.

         Although the relative weights of the caecum, liver, brain,
    kidneys, and testes were increased in several test groups, there were
    no pathological changes in any of these organs. Treatment-related
    microscopic changes consisting of increased numbers of macrophages
    containing a yellow-brown PAS-positive pigment were found in the
    mesenteric lymph nodes. These were the only treatment-related changes
    noted (Sinkeldam et al., 1980a).

         Groups of 10 male and 10 female weanling F344 rats were given
    caramel colour III in the drinking water at concentrations of 0, 0.5,
    1.0, 2.0, 4.0, or 8.0% for 4 weeks. The sample of caramel colour III
    used was found to contain 70 mg/kg THI on an 'as is' basis and the NIH
    07 Open Formula Mouse and Rat diet used in this study contained
    17 mg/kg pyridoxine.

         No differences in body-weight gain were noted for any of the test
    groups, although food consumption of the males was significantly lower
    than controls throughout the study. Transient decreases in lymphocyte
    counts were noted among the treatment groups at the midpoint of the
    study, but no significant differences between control and treated
    groups were noted at 1 month. Other haematological and clinical
    chemistry parameters were normal for all groups. At necropsy no
    differences in organ weights were noted between treated and control
    groups and there were no gross or microscopic pathological changes
    related to treatment (Heidt and Rao, 1981).

         Groups of 10 male and 10 female F344 rats were given caramel
    colour III in the drinking water at concentrations of 0, 1.25, 2.5,
    5.0, 10.0, or 20.0% for 13 weeks in order to select doses for long-
    term carcinogenicity/chronic toxicity studies. Rats were given
    20 ml/day of these solutions and basal diet (CRF-1, Charles River
    Japan, Inc.) was available ad libitum. The caramel colour III used
    contained 78 mg/kg THI on an 'as is' basis and the diet contained
    11-12 mg/kg pyridoxine. During the experimental period, all animals
    were observed daily and clinical signs were recorded. Body weights
    were measured every other week and haematological examinations were
    performed every 4 weeks in the control and 20% groups. At the end of
    the study, all survivors were sacrificed for gross and microscopic
    examination.

         Weight gains were less in all experimental groups than in the
    control group from the first experimental week, and in week 13 the
    weight gains of the 1.25, 2.5, 5.0, 10.0, and 20.0% groups were 89,
    94, 84, 76, and 76%, respectively, of that of controls for males and
    96, 98, 80, 84, and 92%, respectively, for females. Except in the male
    2.5% group and the female 1.25, 2.5, and 20.0% groups, the differences
    in weight gain from the controls were significant. No haematological
    changes were observed either during or at the end of the experimental
    period. At necropsy, no pronounced macroscopic changes were observed
    in any animals, although a few rats in the experimental groups were
    very emaciated. No histological changes related to caramel colour III
    administration were found in any experimental groups.

         From these results, the authors concluded that 1 and 4% caramel
    colour III in drinking water were the appropriate dose levels for a
    carcinogenicity study (Maekawa et al., 1983).

         In a 90-day toxicity study, 2 samples of caramel colour III were
    used, one containing approximately 15 mg/kg THI on a solids basis
    (batch A) and the other containing 295 mg/kg THI on a solids basis
    (batch B). Groups of 20 male and 20 female weanling F344 rats were
    given caramel colour III in drinking water at dose levels of 0, 10,
    15, or 20 g/kg b.w. of batch A and 20 g/kg b.w. of batch B. The NIH 07
    Open Formula Mouse and Rat diet used in this study contained more than
    10 mg/kg pyridoxine. During the study, body weights and food intake
    were recorded weekly. Fluid consumption was recorded 3 times per week
    and concentrations of the test material were adjusted on the basis of
    body weight and fluid intake to give the required dose. Haematological
    analyses were performed at 2 and 6 weeks and at termination, and
    clinical chemistry analyses were carried out at 6 weeks and
    termination. At necropsy, all animals were examined macroscopically
    and selected organ weights were determined. Histopathological
    examinations were conducted on the 20 g/kg test groups and on the
    control animals.

         Although the findings were not always consistent, the animals
    treated at the higher dose levels (15 and 20 g/kg) of batch A and
    those teated with batch B generally had decreased body weights. All
    treated groups had significantly decreased food and fluid intake. If
    fluid intake values are corrected for solid contents of the caramel
    colours, the values for water intake were markedly below those of the
    controls. Because of the rather constant fluid intake of the treated
    groups, it was necessary to increase progressively the caramel colour
    concentration to maintain a constant intake of the caramel colour on a
    body-weight basis. It is likely that the effects on body-weight gain
    and food consumption were due to the reduced water intake of the rats,
    reflecting the poor palatability of the drinking solution rather than
    toxic effects of the caramel colour per se.

         The haematology studies revealed decreased lymphocyte counts in
    the male and female rats fed batch B at 2 weeks and in the male rats
    fed batch B at 6 weeks. All lymphocyte values in these groups were
    normal at the termination of the study. No decreases in lymphocyte
    counts occurred in male or female groups fed batch A at any of the
    dose levels. There were no consistent changes in clinical chemistry
    values, with the exception of slightly increased values in blood urea
    nitrogen in the rats treated with batch B. Urinalysis revealed
    decreased urinary volume and increased specific gravity in rats of
    both sexes treated with either batch of caramel colour III at 6 weeks.
    At termination, these differences were only significant in male rats
    in the top-dose groups (both batches). Treatment-related increases
    were observed in the absolute and relative weights of the caecum (full
    and empty) in animals of both sexes at all dose levels. Dose-related
    increases occurred in absolute and relative kidney weights and were
    considered to reflect compensatory hypertrophy as a consequence of
    reduced water intake; there were no histopathological changes in the
    kidneys of any of the test groups. A dose-related decrease in the
    absolute weight of the thymus was observed, which reached statistical
    significance in the top-dose group males with both samples of caramel
    colour III and top-dose group females with batch B; this decrease was
    not evident when the thymus weight was expressed relative to body
    weight. Other differences in absolute and relative organ weights
    appeared to be a consequence of the dose-related reduction in body
    weight.

         The only treatment-related microscopic changes noted were minimal
    to moderate accumulation of pigment in the tissues of the intestinal
    tract and mesenteric lymph nodes without pathologic alteration.

         Although statistically-significant changes were identified in
    some clinical and anatomical parameters, the authors did not consider
    them to be toxicologically important. On this basis, the no-adverse-
    effect level for ammonia caramel III was considered to be 20 g/kg b.w.
    (MacKenzie, 1985).

         A paired-feeding study was conducted to determine whether poor
    palatability is the mechanism underlying the decreased body-weight
    gains frequently noted in toxicity studies of caramel colour III. A
    group of 10 male Wistar rats were fed 12% caramel colour III in the
    drinking water, and similar groups were permitted a limited intake of
    food or water equivalent to that consumed by the caramel colour III
    group. Body-weight gain and food and fluid intake were decreased in
    the group fed caramel colour III. A similar decrease in body-weight
    gain was noted in the groups of rats restricted to the equivalent
    intake of either food or water. The author concluded that the growth
    depression observed in rats when caramel colour III is fed in the
    drinking water is the result of decreased fluid and food intake. Poor
    palatability of drinking fluid containing caramel colour III was the
    probable cause of these changes (Sinkeldam, 1979).

    Long-term studies

    Rats

         Four groups of 48 male and 48 female Wistar rats were given diets
    containing 0, 1, 3, or 6% caramel colour III (ammonia catalysed "half
    open-half closed pan" caramel; no indication about the presence of
    4-methylimidazole was given) for 2 years. Food and water intake,
    growth, mortality, and organ weights were measured; haematological
    examinations, urinalyses, kidney function tests, and histopathological
    examinations also were carried out.

         A decrease in growth, which was significant in the males, was
    observed at all dose levels. This effect was accompanied by a
    reduction in the cumulative food intake. A significant reduction in
    white cell number (in the 6% group) was associated in the early part
    of the study with a lymphocytopenia, which was present until week 80
    in the male rats fed a diet containing 3 or 6% caramel. In the female
    animals the lymphocytopenia was present until week 52 in these 2
    groups (the 1% group was not tested).

         Spleen weights were reduced in a dose-related manner. The
    relative weights of the (full) caeca were clearly increased at all
    dose levels. No changes were found in the pancreata of the control
    animals, while in the test groups (not dose-related) a total of 10
    hyperplastic changes were found. However, the number of tumours of the
    pancreas showed no relation to the administration of caramel. There
    was no evidence of a carcinogenic effect.

         The authors concluded that a no-observed-effect level could not
    be established for the caramel used in this particular study (Evans
    et al., 1976).

    Observations in man

         In a number of animal studies with caramel colour III, a decrease
    in the total number of leucocytes associated with a decrease in the
    number of lymphocytes was noted. A pilot study in humans was carried
    out in which 1.5 g of caramel colour III (prepared by a closed-pan
    process) was ingested daily by 9 volunteers for 21 days. Total
    circulating leucocytes, lymphocytes, and erythrocytes, together with
    haemoglobin concentrations, were measured prior to and during the
    treatment. No changes were found that could be attributed to treatment
    with caramel colour III. In this experiment 6 of the subjects showed
    no differences from normal in stool frequency or condition. Three
    volunteers occasionally had soft stools; no control groups were used
    (BIBRA, 1976).

    Comments

         The temporary ADI for caramel colour III was revoked at the
    twenty-first meeting of the Committee due to its effects on
    circulating total leucocytes and lymphocytes. Since that
    time, a component of caramel colour III, 2-acetyl-4(5)-
    tetrahydroxybutylimidazole (THI), has been identified and shown to
    cause a depression of lymphocyte counts; the lymphocyte depression
    caused by caramel colour III has been shown to be due largely, if not
    solely, to this minor component. Comparison of the lymphocyte-
    depressing activity of pure THI with a batch of caramel colour III
    containing a known level of THI indicated that other components of
    this batch had an insignificant activity. The lymphocyte depression
    was largely ameliorated by dietary pyridoxine. In studies on samples
    of caramel containing 10 mg/kg THI, no effects on lymphocyte counts
    were observed with adequate dietary levels of pyridoxine.

         Long-term studies in rats and mice indicate that caramel colour
    III is not carcinogenic at dose levels of up to 4% in drinking water,
    which was also the no-effect level in these long-term studies.

    EVALUATION

    Level causing no toxicological effect

         As it was not possible to include caramel colour III at higher
    levels than 4% in drinking water in long-term studies, and as the
    effect of most concern, i.e. lymphopenia, could best be evaluated from
    short-term studies, the Committee based its evaluation on the 
    no-effect level of 20 g/kg b.w./day in a 90-day study in rats using
    caramel colour III which contained approximately 15 ppm THI on a
    solids basis (10 ppm on an 'as is' basis).

    Estimate of acceptable daily intake for man

    0-200 mg/kg b.w. (0-150 mg/kg b.w. on a solids basis).

    Further work or information

    Desired

         Analytical data to confirm that the sample on which the
    evaluation is based is representative of current commercial samples.

         Studies to confirm that THI is the sole component of caramel
    colour III that has lymphocyte-depressing activity and to establish a
    no-effect level for THI.

    REFERENCES

    Allen, J.A., Brooker, P.C., Birt, D.M., & McCaffrey, K.J. (1984).
         Analysis of metaphase chromosomes obtained from CHO cells
         cultured in vitro and treated with caramel colour III.
         Unpublished report No. ITC 3B/84966 from Huntingdon Research
         Centre, Huntingdon, England.  Submitted to WHO by International
         Technical Caramel Association.

    Allen, J.A. & Proudlock, R.J. (1984). Autoradiographic assessment of
         DNA repair in mammalian cells after exposure to caramel colour
         III. Unpublished report from Huntingdon Research Centre,
         Huntingdon, England.  Submitted to WHO by International Technical
         Caramel Association.

    Ashoor, S.H. & Monte, W.C. (1983).  Mutagenicity of commercial
         caramels. Cancer Letters 18, 187-190.

    BIBRA (1976). A study of the haematological effects of caramel in
         human volunteers. Unpublished report No. 1/172/76 from the
         British Industrial Biological Research Association, Carshalton,
         Surrey, England.

    Chacharonis, P. (1960). Acute and chronic toxicity studies on caramel
         colours A and B. Unpublished report No. S.A. 54219 from
         Scientific Associates Inc., St. Louis, MO, USA.

    Chacharonis, P. (1963). Acute oral toxicity study in rats on caramel
         colorings 25A-1, 30B-0, and 30F-1. Unpublished report No. S.A.
         79105 from Scientific Associates Inc., St. Louis, MO, USA.

    Cimino, M.C. & Brusick, D.J. (1981). Mutagenicity evaluation of ETA
         48-IH caramel color in the mouse micronucleus test. Unpublished
         report No. 22129 from Litton Bionetics Inc., Kensington, MD, USA.
         Submitted to WHO by International Technical Caramel Association.

    Evans, J.G., Butterworth, K.R., Gaunt, I.F., & Grasso, P. (1976).
         Long-term toxicity study in the rat of a caramel produced by the
         "half-open-half closed pan" ammonia process. Unpublished report
         No. 6/1976 from the British Industrial Biological Research
         Association, Carshalton, Surrey, England.

    Foote, W.L., Robinson, R.F., & Davidson, R.S. (1958). Toxicity of
         caramel color products. Unpublished report of Battelle Memorial
         Institute, Columbus, OH, USA.

    Galloway, S.M. & Brusick, D.J. (1981a). Mutagenicity evaluation of
         ETA-48-IH in an in vitro cytogenetic assay measuring chromosome
         aberration frequencies in Chinese hamster ovary (CHO) cells.
         Unpublished report No. 20990 from Litton Bionetics Inc.,
         Kensington, MD, USA. Submitted to WHO by International Technical
         Caramel Association.

    Galloway, S.M. & Brusick, D.J. (1981b). Mutagenicity evaluation of
         ETA-48-IH in an in vitro cytogenetic assay measuring chromosome
         aberration frequencies in Chinese hamster ovary (CHO) cells. Part
         II. Unpublished report No. 20990 from Litton Bionetics Inc.,
         Kensington, MD, USA. Submitted to WHO by International Technical
         Caramel Association.

    Gaunt, I.F., Lloyd, A.G., Grasso, P., Gangolli, S.P., & Butterworth,
         K.R. (1975). Toxicological investigations of caramels. I. A
         short-term study in the rat with two caramels produced by
         variations of the ammonia process. Unpublished report No. 14 from
         the British Industrial Biological Research Association,
         Carshalton, Surrey, England.

    Hagiwara, A., Shibata, M., Kurata, Y., Seki, K., Furushima, S., & Ito,
         N. (1983). Long-term toxicity and carcinogenicity test of
         ammonia-process caramel colouring given to B6C3F1 mice in the
         drinking water. Fd. Chem. Toxicol. 21, 701-706.

    Haldi, J., & Wynn, W. (1951). A study to determine whether or not
         caramel has any harmful physiological effect. I. Unpublished
         report from Emory University, Atlanta, GA, USA.

    Heidt, M. & Rao, G.N. (1981). Subchronic toxicity study of ammonia
         caramel color type AC2 in rats. Unpublished report No. 80036 from
         Raltech Scientific Services, Inc., Madison, WI, USA. Submitted to
         WHO by International Technical Caramel Association.

    Ishidate, M. Jr. & Yoshikawa, K. (1980). Chromosome aberration tests
         with Chinese hamster cells in vitro with and without metabolic
         activation - a comparative study on mutagens and carcinogens.
         Arch. Toxicol. Suppl. 4, 41-44.

    Jagannath, D.R. & Brusick, D. (1978a). Mutagenicity evaluation of ETA
         4-10, ETA 4-11, ETA 4-15 in the Ames Salmonella/microsome plate
         test. Unpublished report No. 20838 from Litton Bionetics Inc.,
         Kensington, MD, USA. Submitted to WHO by International Technical
         Caramel Association.

    Jagannath, D.R. & Brusick, D. (1978b). Mutagenicity evaluation of ETA
         4-8, ETA 4-9, ETA 4-12, ETA 4-13, ETA 4-14 in the Ames
         Salmonella/microsome plate test. Unpublished report No. 20838
         from Litton Bionetics Inc., Kensington, MD, USA. Submitted to WHO
         by International Technical Caramel Association.

    Jensen, N.J., Willumsen, D., & Knudsen, I. (1983). Mutagenic activity
         at different stages of an industrial ammonia caramel process
         detected in Salmonella typhimurium TA 100 following pre-
         incubation. Fd. Chem. Toxicol. 21, 527-530.

    Kawachi, T., Yahagi, T., Kada, T., Tazima, Y., Ishidate, M., Sasaki,
         M., & Sugiyama, T. (1980). Cooperative programme on short-term
         assays for carcinogenicity in Japan. In Molecular and Cellular
         Aspects of Carcinogen Screening Tests, IARC Scientific
         Publication No. 27, 323-330. International Agency for Research
         on Cancer, Lyon, France.

    Kawana, K., Akema, R., Nakaoka, T., Ikeda, H., Takimoto, T., &
         Kawauchi, S. (1980). Studies on mutagenicities of natural food
         additives. II. Mutagenicities and antibacterial activities of
         caramels. Eisei Kagaku 26, 259-263.

    Kroplien, U. (1984). Letter dated 17 December 1984 to International
         Technical Caramel Association, submitted to WHO.

    Kroplien, U., Rosdorfer, J., van der Greef, J., Long, R.C. Jr., &
         Goldstein, J.H. (1985). 2-Acetyl-4(5)-(tetrahydroxybutyl)-
         imidazole: Detection in commercial caramel colour III and
         preparation by a model browning reaction. J. Org. Chem. 50,
         1131-1133.

    MacKenzie, K.M. (1985). 90-day toxicity study of caramel color
         (ammonia process) in rats. Vols. I & II. Unpublished report No.
         6154-107 from Hazleton Laboratories America, Inc., Madison, WI,
         USA. Submitted to WHO by International Technical Caramel
         Association.

    Maekawa, A., Ogiu, T., Matsuoka, C., Onodera, H., Furuta, K.,
         Tanigawa, H., Hayashi, Y., & Odashima, S. (1983). Carcinogenicity
         study of ammonia-process caramel in F344 rats. Fd. Chem.
         Toxicol. 21, 237-244.

    Morgareidge, K. (1974). Teratologic evaluation of FDA 71-82 (caramel,
         bakers and confectioners) in mice, rats and rabbits. Unpublished
         report No. PB 234-870 from Food and Drug Research Laboratories
         Inc., Waverly, NY; USA.

    Nishie, K., Waiss, A.C., & Keyl, A.C. (1969). Toxicity of
         methylimidazoles. Toxicol. Appl. Pharmacol. 14, 301-307.

    Nishie, K., Waiss, A.C., & Keyl, A.C. (1970). Pharmacology of alkyl
         and hydroxyalklpyrazines. Toxicol. Appl. Pharmacol. 17, 1.

    Procter, B.G. (1976). A preliminary evaluation of the potential
         toxicological effects of ammonia caramel in mice. Unpublished
         report No. 5705 from Bio-Research Laboratories Ltd., Pointe
         Claire, Quebec, Canada.

    Procter, B.G. (1977). A preliminary evaluation of the potential
         toxicological effects of ammonia caramel (U.S. origin) in rats.
         Unpublished report No. 5617 from Bio-Research Laboratories Ltd.,
         Pointe Claire, Quebec, Canada. Submitted to WHO by International
         Technical Caramel Association.

    Procter, B., Berry, G., & Chappel, C.I. (1976). A toxicological
         evaluation of various caramels fed to albino rats. Unpublished
         report No. 4244 from Bio-Research Laboratories Ltd., Pointe
         Claire, Quebec, Canada.

    Richold, M. & Jones, E. (1980). Ames metabolic activation test to
         assess the potential mutagenic effect of ETA-48-1H. Unpublished
         report No. FDC 8/80358 from Huntingdon Research Centre,
         Huntingdon, England.  Submitted to WHO by International Technical
         Caramel Association.

    Richold, M., Jones, E., & Fenner, L.A. (1984).  Ames metabolic
         activation test to assess the potential mutagenic effect of
         caramel colour III. Unpublished report No. ITC 1B/84709 from
         Huntingdon Research Centre, Huntingdon, England. Submitted to WHO
         by International Technical Caramel Association.

    Sharratt, M. (1971). Unpublished report.

    Sinkeldam, E.J. (1979). Paired feeding test in rats with caramel
         (AC3) in drinking water. Unpublished report No. R 6157 from
         Centraal Instituut voor Voedingsonderzoek (CIVO/TNO), Zeist, The
         Netherlands. Submitted to WHO by International Technical Caramel
         Association.

    Sinkeldam, E.J. (1981). Quantitative relationship between dietary
         pyridoxine content and lymphocyte counts in rats fed ammonia
         caramel. Unpublished report No. V81.390/211704 from Centraal
         Instituut voor Voedingsonderzoek (CIVO/TNO), Zeist, The
         Netherlands. Submitted to WHO by International Technical Caramel
         Association.

    Sinkeldam, E.J. (1982a).  Quantitive relationship between dietary
         pyridoxine content and lymphocyte counts in mature rats fed
         ammonia caramel. Unpublished report No. V82.102/220102 from
         Centraal Instituut voor Voedingsonderzoek (CIVO/TNO), Zeist, The
         Netherlands. Submitted to WHO by International Technical Caramel
         Association.

    Sinkeldam, E.J. (1982b). Short-term (7 day) bioassay in rats with
         three dose levels of 2-acetyl-4(5)-tetrahydroxybutylimidazole
         (13-A-82). Unpublished report No. V82.291/221153 from Centraal
         Instituut voor Voedingsonderzoek (CIVO/TNO), Zeist, The
         Netherlands. Submitted to WHO by International Technical Caramel
         Association.

    Sinkeldam, E.J., Bruyntjes, J.P., & Kuper, C.F. (1980a). Sub-chronic
         (13-week) oral toxicity study with a modified ammonia caramel
         (ETA-26-1) in rats. Unpublished report No. R 6483 from Centraal
         Instituut voor Voedingsonderzoek (CIVO/TNO), Zeist, The
         Netherlands. Submitted to WHO by International Technical Caramel
         Association.

    Sinkeldam, E.J., Roverts, W.G., & Kuper, C.F. (1980b). Sub-chronic
         (13-week) oral toxicity study with ammonia caramel (AC2) in
         rats. Unpublished report No. R 6121 from Centraal Instituut voor
         Voedingsonderzoek (CIVO/TNO), Zeist, The Netherlands. Submitted
         to WHO by International Technical Caramel Association.

    Sinkeldam, E.J., de Groot, A.P., van den Berg, H., & Chappel, C.I.
         (1984). The effect of vitamin B6 on the number of lymphocytes in
         blood of rats fed caramel colour (ammonia process). Unpublished
         report submitted to WHO by International Technical Caramel
         Association.

    Sinkeldam, E.J. & van der Heyden, C.A. (1976a). Short-term feeding
         test with three types of caramels in albino rats. Unpublished
         report No. R 4789 from Centraal Instituut voor Voedingsonderzoek
         (CIVO/TNO), Zeist, The Netherlands.

    Sinkeldam, E.J. & van der Heyden, C.A. (1976b). Short-term (10 week)
         feeding study in rats with three different ammonia caramels.
         Unpublished report No. R 5120 from Centraal Instituut voor
         Voedingsonderzoek (CIVO/TNO), Zeist, The Netherlands.


    CARAMEL COLOUR IV

    EXPLANATION

         The report of the twenty-fourth meeting of the Committee (Annex
    1, reference 53) drew attention to the need for adequate
    specifications for caramel colour IV and for a long-term study of
    carcinogenicity. The temporary ADI of 0-100 mg/kg b.w. was extended
    pending the results of long-term toxicity studies.

    BIOLOGICAL DATA

    Biochemical Aspects

    Absorption, distribution, and excretion

         Pigmentation of tissues of the lymphoreticular system,
    particularly of the mesenteric lymph nodes, has been a frequent
    observation in rats fed high levels of caramel colour. A study was
    undertaken to determine the distribution of the colour component of
    caramel colour in rats and to determine whether accumulation of
    caramel colour was the cause of pigmentation in the mesenteric lymph
    nodes.

         Experimental batches of unlabelled and 14C-labelled caramel
    colour IV were prepared and a concentrate of the higher molecular-
    weight colour components was made by ultrafiltration. The absorption,
    tissue distribution, and excretion of this fraction was studied
    in male F344 rats after a single oral gavage of 2.5 g/kg b.w.
    14C-Labelled material was administered to naive animals and to
    animals which had received 2.5 g/kg b.w. unlabelled material in
    drinking water daily for 13 days prior to administration of
    14C-labelled material.

         Differences between the results in naive and pretreated animals
    were small. Most of the colour components were not absorbed, but
    instead were excreted in the faeces, mainly within 48 hours. By 96
    hours after dosing, 99.7-102.4% of the administered dose was excreted
    in the faeces, with only 1-2% in urine and insignificant amounts
    (< 0.1%) as 14CO2 in expired air.

         Groups of 4 rats were killed at intervals of 4, 8, 12, 24, and 96
    hours after receiving the single oral dose of 14C-labelled material,
    and radioactivity was measured in the following tissues/organs; blood,
    brain, heart, lungs, liver, kidneys, spleen, thymus, mesenteric and
    cervical lymph nodes, gastrointestinal tract (contents and tissues),
    and carcass. Most of the radioactivity was located in the
    gastrointestinal tract, and only low levels of radioactivity were
    found in blood and tissues. The specific activity in the thymus,
    mesenteric lymph nodes, spleen, kidneys, and liver exceeded that in

    blood, but with the exception of the mesenteric lymph nodes, the
    radioactivity was cleared rapidly over the 96-hour study period. With
    the exception of the gastrointestinal tract, the highest tissue levels
    attained, in the liver and kidneys, never exceeded 0.02% and 0.01% of
    the administered dose.

         The results demonstrate that, after administration of large doses
    of the coloured components of caramel colour IV, only a small fraction
    was absorbed, distributed in lymphoreticular tissue, and eventually
    excreted in the urine; retention by the mesenteric lymph nodes
    appeared to account for the pigmentation observed in this tissue
    (Selim, 1982).

    Toxicological studies

    Special studies on carcinogenicity (See also long-term studies)

    Mice

         A carcinogenicity study of caramel colour IV was performed on
    B6C3F1 mice using 5 groups of 50 male and 50 female mice in each
    group. Two groups of each sex served as controls and the 3 treatment
    groups received caramel colour IV in the drinking water at dose levels
    of 2.5, 5.0, or 10.0 g/kg b.w./day for 104 weeks. All animals were
    observed twice daily and palpated for tissue masses weekly from week
    27. Body weights and food intake were recorded weekly, and fluid
    intake was measured on the first, third, fifth, and seventh days of
    the first 12 weeks and over a 48-hour period each week thereafter.

         A complete necropsy was performed on all animals dying on test or
    sacrificed in a moribund condition, and on all survivors at
    termination. Histology (haematoxylin and eosin staining) was performed
    on all gross lesions and on the following organs/tissues of all
    animals: adrenals, brain (three sections), bone, bone marrow,
    epididymis, oesophagus, eyes, gall bladder, heart, intestine
    (duodenum, ileum, caecum and colon), kidneys, liver, lymph nodes
    (cervical, mesenteric, and thoracic), mammary glands, ovaries,
    pancreas, parathyroid, pituitary, prostate, salivary gland, sciatic
    nerve, seminal vesicle, skin, spinal cord, spleen, stomach, testes,
    thymus, thyroids, trachea, urinary bladder, and uterus.

         Four animals were found in the wrong cages during week 15 and had
    probably been misplaced 9 weeks earlier. These animals were removed
    from the study during week 31 and were not included in subsequent
    examinations.

         Although during the course of this study there were sporadic
    significant differences in mean body weights of treated male and
    female groups compared to controls, caramel-colour feeding did not
    have consistent effects on body weights or body-weight gains in either
    sex. Males of the 10 g/kg b.w.-dose group had lower mean food

    consumption than the control groups for 67 of the 104 weeks and males
    receiving 2.5 g/kg b.w. had lower mean food consumption for 21 of the
    104 weeks. There were no consistent differences in food intake of
    males at the 5 g/kg b.w. dose level nor of any of the females.
    Decreased fluid intake was noted, particularly in males at the higher
    dose levels; mean fluid intakes of treated females were usually equal
    to or only slightly lower than controls. There were no treatment-
    related differences in survival rates, which at termination were
    65-75% for the male groups and 67-79% for the female groups. At
    necropsy, there were dose- and/or treatment-related effects on the
    gastrointestinal tract and mesenteric lymph nodes, including dark
    gastrointestinal contents, staining of the mucosa, and diffusely red
    and congested mesenteric lymph nodes. These changes were not
    considered to be toxicologically important and there were no other
    changes of toxicological significance. There was no evidence of
    treatment-related neoplastic lesions in any organs (MacKenzie, 1985b).

    Rats

         A long-term toxicity and carcinogenicity study of caramel colour
    IV was conducted on F344 rats. In the carcinogenicity portion of the
    study, 5 groups of 50 male and 50 female weanling rats were selected
    randomly; 2 groups of each sex served as controls and the 3 treatment
    groups received caramel colour IV in drinking water at dose levels of
    2.5, 5.0, or 10 g/kg b.w./day for 24 months. In the chronic toxicity
    portion of the study, groups of 25 male and 25 female rats received
    caramel colour IV in drinking water at doses of 0, 2.5, 5.0, 7.5, or
    10 g/kg b.w./day for 1 year. The following parameters were monitored:
    mortality, clinical observations including ophthalmic changes, body
    weight, and food and fluid consumption. In the chronic toxicity
    portion clinical chemistry and haematological studies were done at
    intervals of 6 and 12 months and necropsies were performed at 12
    months.

         In the carcinogenicity portion of the study, a complete
    necropsy was performed on all animals dying on test or sacrificed in
    moribund condition, and on all survivors at termination. Histology
    (haematoxylin and eosin staining) was performed on all gross lesions
    and on the following organs/tissues of all animals: adrenals, brain
    (three sections), bone, bone marrow, epididymis, oesophagus, eyes,
    heart, intestine (duodenum, ileum, caecum, and colon), kidneys, liver,
    lung, lymph nodes (cervical, mesenteric, and thoracic), mammary gland,
    ovaries, pancreas, parathyroid, pituitary, prostate, salivary gland,
    sciatic nerve, seminal vesicle, skin, spinal cord, spleen, stomach,
    testes, thymus, thyroid, trachea, urinary bladder, and uterus.

         The feeding of caramel colour IV did not affect survival in
    either the chronic toxicity or carcinogenicity sections of this study
    and, other than dark-stained and soft faeces, there were no treatment-
    related antemortem observations. During the course of both studies
    body weights were reduced for both males and females at the 5 and
    10 g/kg b.w. dose levels. These effects on body weights were
    correlated with reduced water and food consumption at these dose
    levels and reflect the reduced palatability of the drinking fluid.

         Clinical chemistry and haematological studies at 6 and 12 months
    in the chronic toxicity study did not reveal changes of toxicological
    concern. At 6 months, serum concentrations of blood urea nitrogen and
    creatinine were reduced in male groups treated with 5.0, 7.5, and
    10 g/kg b.w. and in female groups treated with 7.5 and 10 g/kg b.w.
    caramel colour. Similar changes were noted at 12 months. Creatinine
    levels were within the normal range for the F344 rat, whereas blood
    urea nitrogen levels were slightly outside this range. Decreased
    levels of serum total protein, albumin, and globulin were also noted
    at the 6-month sampling, particularly in male rats. These changes,
    which were less marked at 12 months, were not accompanied by any
    pathology in the liver or kidneys and were not considered to be of
    toxicological importance.

         The urinalysis studies revealed generally reduced urine volume
    and increased specific gravity in both sexes.

         At necropsy the changes noted were characteristic of the feeding
    of high levels of caramel colour, which consisted primarily of brown
    staining of the gastrointestinal tract and mesenteric lymph nodes and
    caecal enlargement. Increased kidney weights were noted in both sexes
    in animals fed caramel colour IV; no histologic alterations were
    present that could be associated with the increased renal weight,
    which the authors considered to be related to the water imbalance in
    these animals.

         Microscopic examination of the tissues of these animals did not
    reveal specific toxicological changes. Pigmentation in the
    gastrointestinal tract and mesenteric lymph nodes was observed. There
    was no evidence of reactive hyperplasia to the pigment in the
    mesenteric lymph nodes of the gastrointestinal tract.

         In the chronic portion of the toxicity study, although
    statistically significant changes were noted in some parameters, they
    were not considered to be toxicologically important. The highest dose
    fed, 10 g/kg b.w., was considered to be the no-adverse-effect level.

         In the carcinogenicity portion of the study, the observations
    were generally similar to those in the chronic portion. Survival at 24
    months ranged from 64-68% for males and 82-92% for females. Random
    variations in both benign and malignant neoplasms typical of the F344

    strain and this age of animal were observed; however, there were no
    treatment-related differences. The authors concluded that the feeding
    of caramel colour IV at doses up to 10 g/kg b.w. for 24 months did not
    induce neoplastic changes or non-neoplastic changes of toxicological
    importance (MacKenzie, 1985a).

    Special studies on mutagenicity

         Caramel colour IV was evaluated for genetic activity in a series
    of in vitro microbial assays with and without metabolic activation.
    Salmonella typhimurium (strains TA1535, TA1537, and TA1538) and
    Saccharomyces cerevisiae were used. Caramel colour IV was not
    genetically active under the test conditions employed in this study
    (Brusick, 1974).

         Caramel colour IV was evaluated for mutagenicity using the Ames
    Salmonella/microsome plate test and the Saccharomyces/microsome
    plate test. The samples tested were blends of 3 samples of caramel
    colour IV (low colour intensity) and 3 samples of caramel colour IV
    (high colour intensity). The Salmonella test organisms used were
    TA98, TA100, TA1535, TA1537, and TA1538. Tests were conducted directly
    or in the presence of liver microsomal enzyme preparations from
    Araclor-induced rats. Tests were conducted over a range of
    concentrations from 1-50 mg/plate. No signs of genetic activity were
    observed with any of the samples of caramel colour IV tested using
    either the Salmonella or Saccharomyces test organisms (Jagannath &
    Brusick, 1978a).

         The clastogenicity of caramel colour IV (low colour intensity)
    was evaluated in an in vitro cytogenetic assay using cultured
    Chinese hamster ovary cells without metabolic activation. No increase
    in chromosome aberrations was observed at concentrations of test
    material between 5 µg/ml and 5 mg/ml. In the same test, sodium
    ascorbate produced a positive response at 2 mg/ml (Galloway & Brusick,
    1981a).

         The clastogenicity of caramel colour IV (high colour intensity)
    was evaluated in an in vitro cytogenetic assay using Chinese hamster
    ovary cells without metabolic activation. No increase in chromosome
    aberrations was observed at concentrations of test material between
    5 µg/ml and 5 mg/ml. In the same test, sodium ascorbate produced a
    positive response at 2 mg/ml (Galloway & Brusick, 1981b).

         Two samples of caramel colour IV of medium and high colour
    intensity were assayed for mutagenic activity in the Ames test using
    Salmonella strains TA98 and TA100 in the absence or presence of rat
    hepatic S-9 fraction. Neither sample showed mutagenic activity under
    the conditions of the test (Ashoor & Monte, 1983).

         Thirteen commercial caramel colours (not identified) were
    examined for mutagenicity in the Ames test using Salmonella strains
    TA98 and TA100, with and without metabolic activation, and for
    DNA-damage effects on E. coli (Wild/pol A-; Wild/rec A-). None of
    the caramel-colour samples tested showed mutagenic activity in these
    tests (Kawana et al., 1980).

         Five samples of commercial caramel colour were tested for
    mutagenic activity against Salmonella strains TA98 and TA100 in the
    Ames test. All samples were reported to show equivocal results with
    strain TA100, and 2 were equivocal with TA98 (Kawachi et al., 1980).

         Five samples of caramel colour were tested in a chromosome
    aberration test in a cultured cell line of Chinese hamster lung
    fibroblasts, in the presence or absence of hepatic S-9 fraction. Ames
    tests were also conducted using Salmonella strains TA98, TA100, and
    TA1537, with or without metabolic activation and with a 20-minute
    pre-incubation step. All samples of caramel colour were designated as
    positive in the Ames assay and in the chromosome aberration test
    (aberrations were detected in 20% of metaphase cells) (Ishidate &
    Yoshikawa, 1980).

         The same series of 5 caramel-colour samples as used in the above
    studies were tested for mutagenicity in the Ames test using
    Salmonella strains TA98, TA100, TA1535, TA1537, and TA1538 and
    Saccharomyces strain D4. Assays were performed both in the presence
    and absence of hepatic microsomal S-9 fraction from Araclor-induced
    rats in the conventional plate assay and after a pre-incubation step
    in strains TA100 and TA1535. The results of the mutagenicity assays
    were negative under all the test conditions and over a concentration
    range of 1-50 mg/plate for strain TA100 and 1-20 mg/plate for the
    other strains (Jagannath & Brusick, 1978b).

    Special studies on reproduction

         Fifteen male and female Wistar rats were given 0 or 10% caramel
    colour IV solution as their sole fluid source until day 100 and were
    then mated. Animals of the F1-generation (25 males and 25 females)
    were weaned and again given 0 or 10% caramel solution until day 100.
    There were no adverse effects with regard to the number of litters
    born and the number of pups/litter. No influence on haematology,
    growth, food consumption, gross pathology, or histopathology of the
    F1-generation at 100 days of age was observed (Haldi & Wynn, 1951).

         Six different samples of caramel colour IV (3 single-strength
    (SS) and 3 double-strength (DS) products), each containing a different
    level of 4-methylimidazole (between 200 and 850 ppm) were tested in a
    reproduction study in rats. Two control diets were used, an unmodified
    stock diet and the stock diet supplemented with starch and cellulose.
    The various test diets are shown in the following diagramme.

        Composition of the test diet (in g)
                                                                                       

    Group       1     2      3     4      5     6     7     8     9     10    11    12
                                                                                       

    Caramel
    in %        5     10     15    10     10    2     4     6     4     4     0     0
                                                                                       

    stock diet  100   100    100   100    100   100   100   100   100   100   100   100

    wheat
    starch      3.8   1.9    -     1.9    1.9   4.9   4.1   3.4   4.1   4.1   5.8   -

    cellulose   4.2   2.2    -     2.2    2.2   5.4   4.6   3.8   4.6   4.6   6.2   -

    SSa caramel colour
    IV + 202b   5.6   11.5   17.6

    SS caramel colour
    IV + 400                       11.5

    SS caramel colour
    IV + 600                              11.5

    DSc caramel colour
    IV + 350                                    2.3   4.5   6.8

    DS caramel colour
    IV + 639                                                      4.5

    DS caramel colour
    IV + 852                                                            4.5
                                                                                       

    a  SS=Single stength
    b  Quantity of 4-methylimidazole in ppm
    c  DS=Double strength
                                                                                       
    
         Twelve groups of 10 male and 20 female weanling Wistar rats were
    allocated to the above dietary groups and, at week 12, the rats were
    mated in sub-groups of 5 males and 10 females. After a 3-week mating
    period, the females were caged individually.

         At weaning age of the F1 generation, 40 males and 40 females
    were selected from as many different litters as possible within each
    caramel-colour group and continued on the same diet as the parent
    generation. Ten males and 10 females of each group were sacrificed
    after one year (see Sinkeldam et al., 1975, under short-term
    toxicity studies); the remaining 30 males and 30 females of each group
    were fed the test diets for 2 years (see Sinkeldam et al., 1976, under
    long-term toxicity studies). After weaning the F1 litters, the dams
    were killed and the implantation sites were counted.

         No consistent, dose-related effects on growth of the F0 animals
    were noted. No adverse effects were seen on female fertility, litter
    size, average weight and growth of the pups, or number of implantation
    sites or sex ratio of the young. In one group (10% SS caramel colour
    IV + 600 ppm 4-methylimidazole) there was a slight increase in
    mortality at birth. No teratogenic effects were found (Til & Spanjers,
    1973).

         A range-finding reproduction study was conducted on F344 rats.
    Five groups of 12 male and 12 female mature rats were given caramel
    colour IV at concentrations of 0, 10, 15, 20, or 25% in drinking water
    for 21 days prior to mating and throughout mating, gestation, and
    lactation. Animals of the F0 generation were killed when animals from
    the F1 generation were weaned. At weaning, 2 pups/sex/litter were
    randomly selected from litters that had a minimum of 2 males and 2
    females at day 21 and treatment of these animals was continued for 13
    weeks post-weaning. Haematology and clinical chemistry were performed
    on days 45/46 and at termination. Complete gross post-mortem
    examinations of both F0 and F1 animals were performed at sacrifice.
    Selected organs (caecum, spleen, thymus, liver, kidneys, heart,
    adrenals, and gonads from animals of the F0 generation were weighed
    at autopsy.

         Although the dose levels of caramel colour IV in this study were
    very high (ranging from 8-28 g/kg b.w.), no specific toxic effects
    were observed. All F0-generation animals survived the duration of the
    study, but 3 F1-generation animals were killed accidentally while
    sampling for clinical chemistry studies at day 45. In the 20- and
    25%-dose groups of both generations there was a higher incidence of
    soft stools than in the controls, and all animals of the F0
    generation showed slight to statistically significant, dose-related
    decreases in body weight. Dose-related depression of body-weight gain
    was also noted in animals of the F1 generation.

         Mating, pregnancy, and fertility rates were comparable for all
    groups, but the number of implantation sites and of pups alive at days
    0, 4, and 21 of lactation in the 20%-dose groups was significantly
    lower than control values. Litter size was decreased at the 15-, 20-
    and 25%-dose levels. Pups in the 25%-dose group showed a markedly
    higher incidence of alopecia and arched spine than controls, and a

    generalized poor condition during the last 7 days of suckling. No
    significant haematological changes were observed at 45 or 90 days
    post-weaning, except that prothrombin time of the F1 females in the
    15 and 25% groups were significantly greater than controls at day 45.
    Blood urea nitrogen values were lower than controls at 45 and 90 days
    but other clinical chemistry values were normal.

         At necropsy, dose-related increases in absolute and relative
    weights of the liver, kidneys, and caecum (full and empty) were
    observed from animals in the 15%- and higher-dose groups. The only
    gross treatment-related morphological changes reported were
    brown/black or green colouration of the contents and mucous membrane
    of the lower gut and mesenteric lymph nodes (Tierney, 1980).

    Special studies on teratogenicity

         Teratogenicity tests were performed with caramel colour IV on
    mice, rats, and rabbits. The doses employed were 0, 16, 74.3, 345, and
    1600 mg/kg b.w. in all 3 species.

    Mice

         Caramel colour IV was administered by gavage to groups of 19-22
    pregnant albino CD1 mice at the doses above, beginning on day 6 and
    continuing through day 15 of gestation. Body weights were recorded and
    all animals were observed daily for changes in appearance and
    behaviour. On day 17 all dams were subjected to Caesarean section
    under surgical anaesthesia and the numbers of implantation sites,
    resorption sites, live and dead foetuses, and body weights of live
    pups were recorded. All foetuses were examined grossly for external
    congenital abnormalities, one-third of the foetuses from each litter
    underwent visceral examination (Wilson technique), and the remaining
    two-thirds were cleared, stained with Alizarin Red, and examined for
    skeletal defects.

         The number of abnormalities seen in either soft or skeletal
    tissues did not differ from the number occurring spontaneously in
    sham-treated controls (Morgareidge, 1974).

    Rats

         Caramel colour IV was administered by gavage to groups of 21-24
    pregnant Wistar rats at the dose levels above, beginning on day 6 and
    continuing daily through day 15 of gestation. Body weights were
    recorded and all animals were observed daily for changes in appearance
    and behaviour. On day 20 dams were subjected to Caesarean section
    under surgical anaesthesia and the number of implantation sites,
    resorption sites, live and dead foetuses, and body weights of live
    pups were recorded. All foetuses were examined for gross, visceral,
    and skeletal abnormalities as in the mouse experiment.

         No clearly discernible effects on nidation or on maternal or
    foetal survival were observed. The number of abnormalities seen in the
    test groups did not differ from the number occurring spontaneously in
    sham-treated controls (Morgareidge, 1974).

    Rabbits

         Caramel colour IV was administered by gavage to groups of 11-12
    pregnant Dutch-belted female rabbits at the doses above beginning on
    day 6 and continuing daily through day 18 of pregnancy. Body weights
    were recorded and all animals were observed daily for changes in
    appearance and behaviour. The does were subjected to Caesarian section
    under surgical anaesthesia on day 29 and the numbers of corpora lutea,
    implantation sites, resorption sites, and live and dead foetuses were
    recorded.

         Body weights of the live pups were also recorded. All foetuses
    were examined grossly for the presence of external congenital
    abnormalities. The live foetuses from each litter were then placed in
    an incubator for 24 hours for evaluation of neonatal survival.
    Surviving pups were sacrificed and all pups examined for visceral
    abnormalities (by dissection). All foetuses were then cleared with
    potassium hydroxide, stained with Alizarin Red, and examined for
    skeletal defects.

         No clearly discernible effects on nidation or on maternal or
    foetal survival were observed. The number of abnormalities seen in
    either soft or skeletal tissues of the test groups did not differ from
    the number occurring spontaneously in sham-treated controls
    (Morgareidge, 1974).

    Special study on haematology

    Rats

         A study was conducted in rats to determine whether haematological
    changes were associated with the feeding of high dietary
    concentrations of caramel colour IV. Sixty-six male rats, mean initial
    body weights approximately 173 g, were fed caramel colour IV in the
    diet for 28 days. Sixteen rats were assigned to a control group and 10
    rats/group were assigned to levels of 16 or 22% caramel colour IV
    (high colour intensity) and 34 or 47% caramel colour IV (low colour
    intensity). Caramel colour III was fed at a 4% level to a positive
    control group. The parameters observed were body weight, food
    consumption, haematology (total and differential leucocyte counts),
    and terminal necropsy with organ weights (caecum - full and empty).
    Rats were bled from the retro-orbital sinus on days 16, 9, and 2
    before treatment and again on days 7, 14, and 28 during treatment.

         Feeding of these high dietary levels of caramel colour IV led to
    reduced rates of body-weight gain. The animals fed caramel colour III
    gained weight at a rate similar to the controls. Animals receiving
    caramel colour III had progressively lower relative lymphocyte counts
    commencing 1 week after treatment and increasing in severity with
    duration of treatment. This pattern of decreased relative lymphocyte
    counts was not seen with either sample of caramel colour IV and no
    significant decreases in lymphocyte counts were observed at any dose
    level of this caramel colour. Caecal weights were increased in all
    treatment groups (BIBRA, 1978).

    Acute toxicity

         No information available.

    Short-term studies

    Mice

         Groups of 10 B6C3F1 mice of each sex were given concentrations
    of 0, 10, 15, 20, or 30% caramel colour IV in drinking water for 4
    weeks. The intake of caramel colour IV expressed in g/kg b.w./day was
    more than twice the percentage concentration in the drinking water. At
    28 days the body weights of male animals at the 30%-dose level were
    significantly decreased compared to the controls, and transient
    depressions of body weights were noted at the highest-dose level in
    females. No significant depressions in body-weight gain were noted at
    the lower-dose levels in either male or female mice. Fluid consumption
    was depressed throughout the study among all treatment groups.
    However, these changes were not consistent with time or dosage. There
    were no statistically significant differences in food consumption
    between treated animals and controls. At necropsy, the only treatment-
    related effect reported was enlargement of the caecum (Tierney, 1979).

    Rats

         Groups of 5 rats received either 10 or 20% caramel colour IV
    solution (equivalent to about 10 or 20 g/kg b.w./day) as their sole
    source of fluid for 127 days. Only dark faeces and very mild diarrhoea
    were noted. No adverse effects were noted regarding general health,
    body weight, food and fluid consumption, haematology, gross pathology,
    or histopathology (Haldi & Wynn, 1951).

         Six groups of 5 male and 5 female weanling rats received 0 or 10%
    caramel colour IV solution as their sole fluid source for 100, 200, or
    300 days. No adverse effects were noted regarding growth, food and
    fluid intake, haemotology, gross pathology, or histopathology (Haldi &
    Wynn, 1951).

         Groups of 16 male and 16 female rats received either 0 or 10%
    caramel colour IV solution for 100 days and groups of 5 rats received
    20% caramel solution for 100 days. At the lower test level, there were
    no observable abnormalities as regards growth, food consumption,
    haematology, gross pathology, or histopathology. Only growth and
    haematology were examined at the higher test level, and the results
    for both parameters were found to be normal (Haldi & Wynn, 1951).

         Groups of 5 male and 5 female rats were given 1 ml/kg b.w. of
    concentrated caramel colour for 21 days. Some diarrhoea was induced in
    all animals, but no other abnormalities were noted. Gross pathology
    and histopathology revealed no significant changes due to
    administration of the test compound (Foote et al., 1958).

         Three groups of 20 male and female rats received either 0 or
    11-14 g/kg b.w. of caramel colour IV solutions for 100 days. Growth
    and food intake did not differ significantly between test and control
    animals. Gross pathology and histopathology showed no abnormal
    findings related to administration of the test compound (Haldi & Wynn,
    1958).

         Four groups of 10 male and 10 female Sprague-Dawley rats received
    0, 0.1, 1.0 or 10% caramel colour IV in their diet for 12 weeks. No
    adverse effects were noted on growth, food consumption, urinalysis,
    haematology, gross pathology, or histopathology that were related to
    administration of the caramel colour (Prier, 1960).

         Groups of 10 male and 10 female rats received 0, 5, or 10 g/kg
    caramel colour IV in their diet for 3 months. Weight gain was normal
    in all groups. Food consumption, haematology, and urinalysis were
    comparable in all groups. Gross pathology and histopathology showed no
    test-related adverse findings (Chacharonis, 1960).

         Four groups of 10 male and 10 female Sprague-Dawley rats received
    0, 5, 10, or 20% caramel colour IV (low colour intensity) or caramel
    colour IV (high colour intensity) in their diet for 90 days. In
    addition, a paired-feeding study involving 5 male rats in 2 groups was
    run for 23 days (one sample was at the 20% level), and there were no
    differences in the rate of growth. The only effects attributable to
    treatment were a mild depression in growth of male rats at the 10 and
    20% levels due to unpalatability of the test diet. No other adverse
    findings were noted in growth, behaviour, mortality, haematology,
    urinalysis, gross pathology, organ weights, or histopathology (Kay and
    Calandra, 1962a; 1962b).

         Four groups of 10 male and 10 female Sprague-Dawley rats received
    either 0 or 10% of 3 different caramel colours (one of which was a
    sample of caramel colour IV) in their diet for 90 days. Weight gains
    showed a slight reduction compared with controls but food consumption
    was normal for all groups. No abnormalities were noted with regard to
    haematology, urinalysis, gross pathology, or histopathology
    (Chacharonis, 1963).

         Four groups of 15 male and 15 female rats received 0, 5, 10, or
    20% caramel colour IV in their diet for 90 days. No adverse effects
    were noted on appearance, behaviour, survival, body weights, food
    intake, haematology, blood chemistry, urinalysis, organ weights, gross
    pathology, or histopathology (Oser, 1963).

         Four groups of 10 male and 10 female rats received 0, 0.015, 0.3,
    or 3.0% caramel colour IV in their diet for 90 days. No differences
    between test and control animals were noted regarding body weight,
    food consumption, haematology, urinalysis, gross pathology, or
    histopathology (Nees, 1964).

         Four groups of rats received 0, 4, 8, or 16% caramel colour IV
    in their diet for 3 months. No convulsions or other behavioural
    abnormalities or signs of neurological damage were seen. No
    macroscopic or microscopic pathological abnormalities were found in
    the central nervous system (Sharratt, 1971).

         Three groups of 20 female Wistar rats received stock diet to
    which 0, 10, or 20% caramel colour IV containing 202 ppm 4-
    methylimidazole was added. During the second week of the experiment
    these levels of caramel colour IV were increased to 15 and 25% and in
    week 7, the levels were increased to 25 and 30% caramel colour IV. The
    diets were administered for 16 weeks followed by a 4-week recovery
    period.

         Food consumption and growth of the test animals were comparable
    with the controls. Leucocyte counts, collected after 4, 8, 12, and 16
    weeks, did not show statistically significant differences among the
    groups. The relative weights of the caecum, both filled and empty,
    were distinctly increased after 4 weeks of feeding caramel colour IV.
    After the recovery period of 4 weeks the increases had disappeared.
    The relative weight of the thymus was not affected.

         Gross examination at autopsy after 16 weeks of feeding caramel
    clour IV revealed a dose-related, brown-greenish discolouration of the
    mesenteric lymph nodes in all test animals of the highest-dose level.
    After the recovery period of 4 weeks the colour change was less, but
    still visible. Microscopically, the lymph nodes of the test rats
    showed pigment accumulation which was not noticeably diminished after

    withdrawal of the caramel for 4 weeks. These results failed to confirm
    the decreased leucocyte counts that were observed in females fed 5 to
    15% caramel colour IV in a 1-year feeding study (Sinkeldam & van der
    Heyden, 1975).

         A 10-week feeding study with 18 groups of 10 male and 10 female
    Wistar rats was carried out with 3 caramel colours, including caramel
    colour IV (low colour intensity) and caramel colour IV (high colour
    intensity). The dietary concentrations were 0, 1.25, 2.5, 5, 10, and
    15% caramel colour IV (low colour intensity) and 0, 0.5, 1, 2, 4, and
    6% caramel colour IV (high colour intensity).

         Both caramel colours caused loose stools at the highest-dose
    levels, although body-weight gains were not affected. Leucocyte
    (lymphocyte) counts were not affected in the animals fed samples of
    caramel colour IV at any of the dose levels. Only slight indications
    of caecal enlargement were observed. Minimal amounts of pigment were
    observed in the mesenteric lymph nodes of rats fed 2.5% and higher of
    the low colour intensity sample, and 1% and higher of the high colour
    intensity sample (Sinkeldam and van der Heyden, 1976).

         In a 10-week feeding study, 17 groups of 15 male and 15 female
    Sprague-Dawley rats were given various caramel colours that included
    caramel colour IV (low colour intensity) and caramel colour IV (high
    colour intensity). The dose levels fed were 1.25, 2.5, 5, 10, and 15%
    for caramel colour IV (low colour intensity) and 0.5, 1, 2, 4, and 6%
    caramel colour IV (high colour intensity). Two control groups were
    used.

         In the animals fed caramel colour IV (low colour intensity) at
    the 15% level, the faeces became soft within 2 weeks. The water
    content of the faeces from these animals was higher than that of the
    controls, as was the water content of faeces from rats fed 6% caramel
    colour IV (high colour intensity). Body-weight gains were slightly
    decreased in male, but not female, rats fed both samples of caramel
    colour.

         There were no significant changes in total white cell or
    lymphocyte counts in animals fed either sample of caramel colour IV.
    No macroscopic or microscopic evidence of abnormal pigmentation of the
    mesenteric lymph nodes was found in any group of either sample of
    caramel colour IV. Caecal weights were generally increased in all test
    groups fed both samples of caramel colour IV (Procter et al., 1976).

         Groups of weanling Wistar rats were given caramel colour IV (low
    colour intensity) or caramel colour IV (high colour intensity) at
    concentrations of 0, 0.5, 1.0, 2.0, 4.0, or 16.0% in the diet for 10
    weeks. Each group contained 15 male or 15 female rats except the
    control group (60 animals of each sex) and the 16%-dose group

    (10 animals of each sex). Food intake and growth were recorded and
    haematological studies were carried out. Lymph nodes, thymus, spleen,
    and caecum were examined histologically for distribution of pigment.

         An additional 3 groups of 10 rats of each sex were given basal
    diet or diet containing 16% each of the caramel colours for 10 weeks.
    At the end of this period all the rats received basal diets.
    Haematological studies were performed on these animals at 3, 7, 14,
    and 28 days. Five rats of each sex were killed after 7 days and the
    remainder after 28 days of feeding basal diet (recovery experiment).

         Decreased body-weight gains were noted in animals of both sexes
    fed 16% caramel colour IV (high colour intensity). No decreases were
    observed in the groups fed caramel colour IV (low colour intensity).
    Food intake was not consistently altered in any of the groups fed
    either sample of caramel colour IV. Occasionally, values for total
    leucocyte counts were significantly higher or lower than controls in
    the groups fed both samples of caramel colour IV; however, these
    changes were not consistent in direction and they were not dose-
    related. There were no consistent or dose-related differences in
    lymphocyte counts between the groups fed caramel colour IV and the
    controls. Liver weights were significantly increased in the group fed
    16% caramel colour IV (high colour intensity). Increased relative
    kidney weights were observed in the groups fed 2, 4, and 16% caramel
    colour IV (high colour intensity) and 16% caramel colour IV (low
    colour intensity), although no histological changes were observed.
    Increased caecal weights were seen only at the 16% feeding level for
    both caramel colours. At necropsy pigmentation of the lymph nodes
    was seen at the 16% feeding level of both caramel colours.
    Microscopically, pigmentation was observed in the mesenteric lymph
    nodes in the males and females in the groups fed 16% caramel colour IV
    (high colour intensity). Relative weights of the liver and kidneys and
    caecal weights returned to normal during the recovery period (BIBRA,
    1977).

         Five groups of 30 male and 30 female weanling F344 rats were
    given caramel colour IV in drinking water at concentrations which
    provided intakes of 0, 15, 20, 25, or 30 g/kg b.w./day for 13 weeks.
    After 43 days, 10 animals of each sex from each group were randomly
    selected for collection of data on urinalysis, haematology, and
    clinical chemistry, followed by sacrifice and necropsy; similar
    examinations were performed on survivors at termination. At interim
    and terminal sacrifice, a detailed necropsy was performed on all
    animals and a complete histopathological examination was performed on
    controls and animals in the top-dose group.

         Throughout the study, rats given caramel colour IV produced dark-
    coloured, soft or sticky, liquid and/or odorous, faeces which stained
    and caused alopecia of the perianal area, most noticeably at the
    higher-dose levels. Dose-related decreases in food intake, water

    consumption (after correction for caramel content), and body-weight
    gains were observed and were attributed to the poor palatability of
    the drinking solutions.

         Caramel colour IV at the dose levels tested did not significantly
    affect haematological values at interim or terminal examinations. All
    treatment groups of both sexes had significantly-reduced blood urea
    nitrogen and alkaline phosphatase levels at both 45 and 90 days. Total
    serum protein values of both sexes in the treatment groups were lower
    than controls at 90, but not at 45, days. These effects may be due to
    reduced food intake and growth retardation.

         Treated rats had reduced urine volume and increased urine
    specific gravity, protein, ketones, and acidity, which were associated
    with decreased water consumption. Increased kidney weights were
    observed at necropsy. Treatment-related decreases in thymus and spleen
    weights and increased caecal size, with staining of the mucosa of the
    caecum and colon, were noted. Accumulation of yellowish-tan pigment
    occurred in macrophages of the mesenteric lymph nodes. No treatment-
    related histopathological changes were observed in any organs at the
    highest-dose level (Heidt & Rao, 1980).

         A 1-year rat toxicity study was conducted as a continuation of
    the reproduction study described earlier (Til & Spanjers, 1973). It
    involved the interim sacrifice of one-fourth of the animals utilized
    in the 2-year toxicity study described below.

         Wistar rats (10 males or 10 females in each group) were selected
    from the first litter of parents that were given diets containing 6
    samples of caramel colour IV, 3 of which were low colour intensity and
    3 of which were high colour intensity. The dose levels employed were
    5, 10, and 15% caramel colour IV (low colour intensity) and 2, 4, and
    6% caramel colour IV (high colour intensity). Two additional samples
    of caramel colour IV (low colour intensity) were tested at 10% and 2
    additional samples of caramel colour IV (high colour intensity) were
    tested at 4% in the diet. The animals were sacrificed after a feeding
    period of 1 year.

         No adverse effects on behaviour, growth, food intake, mortality,
    liver and kidney function tests, urine composition, or organ weights
    were observed. Haematological indices showed a slight dose-related
    decrease in total leucocyte counts in females fed one sample of
    caramel colour IV (low colour intensity) which contained 202 mg/kg
    methylimidazole. In males, no effect was noted and in a subsequent
    experiment in which the same sample of caramel colour IV was fed to
    female rats at levels as high as 15 and 30%, no indications of
    decreased leucocyte counts were observed after 4, 8, 12, or 16 weeks.

         The only finding attributed to the feeding of caramel colour IV
    consisted of increased accumulation of a yellow-brown pigment and
    pigment-laden macrophages in the mesenteric lymph nodes of males and
    females in all caramel colour IV groups. Inflammatory or degenerative
    changes of the lymphoid tissue were not found (Sinkeldam et al.,
    1975).

    Dogs

         Four groups of 3 male and 3 female adult beagle dogs received 0,
    6, 12.5, or 25% caramel colour IV in their diet 5 days per week for 90
    days. No significant adverse effects on growth, behaviour, food
    consumption, mortality, liver function, kidney function, haematology,
    urinalysis, gross pathology, or histopathology were noted (Kay and
    Calandra, 1962c).

    Long-term toxicity studies

    Rats

         Six samples of caramel colour were tested in a long-term toxicity
    study, 3 of which were caramel colour IV (low colour intensity) and 3
    of which were caramel colour IV (high colour intensity). Each group
    consisted of 40 male or 40 female weanling Wistar rats, except the
    control group which had double this number of animals. Animals were
    selected from the first litter of parents fed diets containing the
    various caramel colours from weaning age (see reproduction studies,
    Til & Spanjers, 1973). The dose levels tested were 5, 10, or 15%
    caramel colour IV (low colour intensity) and 2, 4, or 6% caramel
    colour IV (high colour intensity). Two additional samples of caramel
    colour IV (low colour intensity) were tested at 10% and 2 additional
    samples of caramel colour IV (high colour intensity) were tested at 4%
    in the stock diet.

         Observations were made of general appearance, behaviour,
    mortality, growth, food intake, haematological factors, and clinical
    chemistry of blood and urine. After about 14 months mortality
    attributed to intercurrent disease was observed in both control and
    treated groups. Approximately three-quarters of the animals died or
    were killed before the experiment was terminated at week 104. Organs
    of animals that died or were killed were weighed and extensive
    histopathological examinations were carried out. However, one-third of
    the animals could not be examined histopathologically due to
    autolysis. At week 104 organs were weighed and extensive
    histopathological examinations were carried out on all surviving rats
    of all groups.

         No clinical changes were observed in this study except for
    slightly decreased haemoglobin and haematocrit values at weeks 78 and
    98 in males fed caramel colour IV (high colour intensity). Leucocyte

    counts were decreased in females fed 10 and 15% of one sample of
    caramel colour IV (low colour intensity) at weeks 13 and 52, but these
    changes were not consistent and decreases in lymphocyte counts were
    not reported. At autopsy, an increased incidence of greenish
    discoloured mesenteric lymph nodes was observed in most groups fed
    high levels of caramel colour. Microscopically, an increased pigment-
    phagocytosis in the mesenteric lymph nodes in all test groups (except
    the group fed 2% caramel colour IV (high colour intensity)) was
    observed. No evidence of any other adverse structural or cellular
    alteration was found. Gross and microscopic examination of the other
    organs did not reveal any pathological changes attributable to the
    ingestion of caramel colour IV. An increase in the incidence of
    neoplastic lesions in the different groups was not found (Sinkeldam
    et al., 1976).

         A long-term toxicity and carcinogenicity study of caramel colour
    IV was conducted using F344 rats (MacKenzie, 1985a). Details are given
    above (see special studies on carcinogenicity).

    Observations in man

         Tolerance studies of caramel colour IV (low colour intensity) and
    caramel colour IV (high colour intensity) were conducted in human
    volunteers. The subjects, 10 men and 10 women, ingested caramel colour
    once daily in simulated soft drinks over 3 test periods of 21 days
    each separated by 7-day rest intervals. The test doses were 6 g/day
    during the first test period, 12 g/day during the second period, and
    18 g/day during the third test period.

         Haematological, clinical chemical, and routine urinary parameters
    were studied at the beginning of each ingestion period, after 10 days
    of ingestion, and at the end of each ingestion period. Most individual
    values for haemoglobin, haemotocrit, RBC, corrected sedimentation
    rate, WBC, and differentials (neutrophils, basophils, eosinophils,
    monocytes, and lymphocytes) were found to be within normal limits.
    There were a few instances of values outside the normal range,
    indicating mild neutropenia and mild lymphocytosis, but these were not
    consistent and were unrelated to the ingestion of caramel colour IV.
    On the other hand, caramel ingestion was associated with an increased
    frequency of bowel movements and softening or increased liquidity of
    faeces (Marier et al., 1977a; 1977b).

    Comments

         The pigmentation of mesenteric lymph nodes and caecal enlargement
    were considered to be non-specific effects that are of no
    toxicological significance. The carcinogenicity studies required by
    the twenty-fourth meeting of the Committee (Annex 1, reference 53)

    have been conducted in rats and mice, and no treatment-related
    neoplastic changes were observed. The material used in these studies
    conformed to recent specifications. The committee based its evaluation
    on the no-effect level in these long-term/carcinogenicity studies, to
    which (in view of the ancillary human data in which no adverse effects
    other than laxation were observed) a safety factor of 50 was applied.

    EVALUATION

    Level causing no toxicological effect

    Mouse:    10 g/kg b.w./day in drinking water

    Rat:      10 g/kg b.w./day in drinking water

    Man:      No adverse effects other than laxation at levels up to
              18 g/day.

    Estimate of acceptable daily intake for man

    0-200 mg/kg b.w. (0-150 mg/kg b.w. on a solids basis).

    REFERENCES

    Ashoor, S.H. & Monte, W.C. (1983).  Mutagenicity of commercial
         caramels. Cancer Letters 18, 187-190.

    BIBRA (1977). The short-term (10 week) feeding study with three
         caramels in rats. Unpublished report No. 177/2/77 from the
         British Industrial Biological Research Association, Carshalton,
         Surrey, England.

    BIBRA (1978). The evaluation of the effect of sulfite ammonia caramels
         on the lymphocytes of the rat. Unpublished report No. 238/2/78
         from the British Industrial Biological Research Association,
         Carshalton, Surrey, England.

    Brusick, D. (1974). Mutagenic evaluation of compound FDA 71-83,
         caramel. Unpublished report No. 2468 from Litton Bionetics Inc.,
         Kensington, MD, USA.

    Chacharonis, P. (1960). Acute and chronic toxicity studies on caramel
         colours A and B. Unpublished report No. S.A. 54219 from
         Scientific Associates Inc., St. Louis, MO, USA.

    Chacharonis, P. (1963). Acute oral toxicity study in rats on caramel
         colorings 25A-1, 30B-0, and 30F-1. Unpublished report No. S.A.
         79105 from Scientific Associates Inc., St. Louis, MO, USA.

    Foote, W.L., Robinson, R.F., & Davidson, R.S. (1958). Toxicity of
         caramel color products. Unpublished report of Battelle Memorial
         Institute, Columbus, OH, USA.

    Galloway, S.M. & Brusick, D.J. (1981a). Mutagenicity evaluation of
         ETA-36-1 in an in vitro cytogenetic assay measuring chromosome
         aberration frequencies in Chinese hamster ovary (CHO) cells.
         Unpublished report No. 20990 from Litton Bionetics Inc.,
         Kensington, MD, USA. Submitted to WHO by International Technical
         Caramel Association.

    Galloway, S.M. & Brusick, D.J. (1981b). Mutagenicity evaluation of
         ETA-63-1S in an in vitro cytogenetic assay measuring chromosome
         aberration frequencies in Chinese hamster ovary (CHO) cells. Part
         II. Unpublished report No. 20990 from Litton Bionetics Inc.,
         Kensington, MD, USA. Submitted to WHO by International Technical
         Caramel Association.

    Haldi, J. & Wynn, W. (1951). A study to determine whether or not
         caramel has any harmful physiological effect. I. Unpublished
         report from Emory University, Atlanta, GA, USA.

    Haldi, J. & Wynn, W. (1958). A study to determine whether or not
         caramel has any harmful physiological effect. II. Unpublished
         report from Emory University, Atlanta, GA, USA.

    Heidt, M. & Rao, G.N. (1980). 90-day subacute toxicity study of
         caramel color (sulfite ammonia process) type SAC2 in rats.
         Unpublished report No. 79028 from Raltech Scientific Services,
         Inc., Madison, WI, USA. Submitted to WHO by International
         Technical Caramel Association.

    Ishidate, M. Jr. & Yoshikawa, K. (1980). Chromosome aberration tests
         with Chinese hamster cells in vitro with and without metabolic
         activation - a comparative study on mutagens and carcinogens.
         Arch. Toxicol. Suppl. 4, 41-44.

    Jagannath, D.R. & Brusick, D. (1978a). Mutagenicity evaluation of ETA
         4-10, ETA 4-11, ETA 4-15 in the Ames Salmonella/microsome plate
         test. Unpublished report No. 20838 from Litton Bionetics Inc.,
         Kensington, MD, USA. Submitted to WHO by International Technical
         Caramel Association.

    Jagannath, D.R. & Brusick, D. (1978b). Mutagenicity evaluation of
         caramel colors ETA 4-8, ETA 4-9, ETA 4-12, ETA 4-13, ETA 4-14 in
         the Ames Salmonella/microsome plate test. Unpublished report
         No. 20838 from Litton Bionetics Inc., Kensington, MD, USA.
         Submitted to WHO by International Technical Caramel Association.

    Kawachi, T., Yahagi, T., Kada, T., Tazima, Y., Ishidate, M., Sasaki,
         M., & Sugiyama, T. (1980). Cooperative programme on short-term
         assays for carcinogenicity in Japan. In Molecular and Cellular
         Aspects of Carcinogen Screening Tests, IARC Scientific
         Publication No. 27, 323-330. International Agency for Research
         on Cancer, Lyon, France.

    Kawana, K., Akema, R., Nakaoka, T., Ikeda, H., Takimoto, T., &
         Kawauchi, S. (1980). Studies on mutagenicities of natural food
         additives. II. Mutagenicities and antibacterial activities of
         caramels. Eisei Kagaku 26, 259-263.

    Kay, J.H. & Calandra, J.C. (1962a). Subacute oral toxicity of caramel
         colorings (coded Sample A) - albino rats. Unpublished report from
         Industrial Bio-test Laboratories, Inc., Northbrook, IL, USA.

    Kay, J.H. & Calandra, J.C. (1962b). Ninety-day subacute oral toxicity
         of caramel coloring (coded Sample B) - albino rats. Unpublished
         report from Industrial Bio-test Laboratories, Inc., Northbrook,
         IL, USA.

    Kay, J.H. & Calandra, J.C. (1962c). Subacute oral toxicity of caramel
         coloring (coded Sample A) - dogs. Unpublished report from
         Industrial Bio-test Laboratories, Inc., Northbrook, IL, USA.

    MacKenzie, K.M. (1985a). Carcinogenicity and chronic toxicity study of
         caramel color (sulfite ammonia process) type SAC2 in rats
         (Chronic toxicity portion, Vols. I-V; Carcinogenicity portion,
         Vols. I-VIII). Unpublished study No. 81184 from Hazleton
         Laboratories America, Inc., Madison, WI, USA. Submitted to WHO by
         International Technical Caramel Association.

    MacKenzie, K.M. (1985b). Carcinogenicity study of caramel color
         (sulfite ammonia process) type SAC2 in mice. Vols. I-VI.
         Unpublished study No. 81185 from Hazleton Laboratories America,
         Inc., Madison, WI, USA. Submitted to WHO by International
         Technical Caramel Association.

    Marier, G., Berry, G., & Orr, J.M. (1977a) Tolerance study of single
         strength ammonia sulfite caramel in human volunteers. Unpublished
         project No. 5612, report No. 2, from Bio-Research Laboratories
         Ltd., Pointe Claire, Quebec, Canada.

    Marier, G., Berry, G., & Orr, J.M. (1977b) Tolerance study of double
         strength ammonia sulfite caramel in human volunteers. Unpublished
         project No. 5711, report No. 2, from Bio-Research Laboratories
         Ltd., Pointe Claire, Quebec, Canada.

    Morgareidge, K. (1974). Teratologic evaluation of FDA 71-83 (caramel,
         beverage) in mice, rats and rabbits. Unpublished report No. PB
         234-867 from Food and Drug Research Laboratories Inc., Waverly,
         NY, USA.

    Nees, P.O. (1964). Toxicological feeding study of caramel E.
         Unpublished report from Wisconsin Alumni Research Foundation,
         Madison, WI, USA.

    Oser, B.L. (1963).  Toxicological feeding study of "acid-proof"
         caramel. Unpublished report No. 83911 from Food and Drug Research
         Laboratories, Inc., Waverly, NY, USA.

    Prier, R.F. (1960). The toxicity of double strength acid proof caramel
         in rats - 12 week feeding test. Unpublished report No. 9070599
         from Wisconsin Alumni Research Foundation, Madison, WI, USA.

    Procter, B.G., Berry, G., & Chappel, C.I. (1976). A toxicological
         evaluation of various caramels fed to albino rats. Unpublished
         report No. 4244 from Bio-Research Laboratories Ltd., Pointe
         Claire, Quebec, Canada.

    Selim, S. (1982). Single and repeat oral dose pharmacokinetic and
         distribution studies of caramel color concentrate in the rat.
         Unpublished study No. IT-59r from Primate Research Institute,
         Holloman Air Force Base, NM, USA. Submitted to WHO by
         International Technical Caramel Association.

    Sharratt, M. (1971). Unpublished report.

    Sinkeldam, E.J. & van der Heyden, C.A. (1975). Short-term feeding
         study with caramel SS 202 in albino rats. Unpublished report No.
         R 4777 from Centraal Instituut voor Voedingsonderzoek (CIVO/TNO),
         Zeist, The Netherlands.

    Sinkeldam, E.J. & van der Heyden, C.A. (1976). Short-term (10 week)
         feeding study in rats with three different ammonia caramels.
         Unpublished report No. R 5120 from Centraal Instituut voor
         Voedingsonderzoek (CIVO/TNO), Zeist, The Netherlands.

    Sinkeldam, E.J., van der Heyden, C.A., & Beems, R.B. (1976). Chronic
         (two year) feeding study in rats with six different ammonia
         caramels. Unpublished report No. R 4961 from Centraal Instituut
         voor Voedingsonderzoek (CIVO/TNO), Zeist, The Netherlands.

    Sinkeldam, E.J., Willems, M.I., & van der Heyden, C.A. (1975). One
         year feeding study in rats with six different ammonia caramels.
         Unpublished report No. R 4767 from Centraal Instituut voor
         Voedingsonderzoek (CIVO/TNO), Zeist, The Netherlands.

    Tierney, W.J. (1979). A four-week dose range-finding study of caramel
         colour no. 3, sample 3-1, in mice. Unpublished project
         No. 79-2380 from Bio/dynamics Inc., East Millstone, NJ, USA.
         Submitted to WHO by International Technical Caramel Association.

    Tierney, W.J. (1980). An in-utero range-finding study of caramel
         colour no. 3, sample 3-1, in rats. Unpublished project
         No. 79-2405 from Bio/ynamics Inc., East Millstone, NJ, USA.
         Submitted to WHO by International Technical Caramel Association.

    Til, H.P. & Spanjers, M. (1973). Reproduction study in rats with six
         different ammonia caramels. Unpublished report No. R 4068 from
         Centraal Instituut voor Voedingsonderzoek (CIVO/TNO), Zeist, The
         Netherlands.
    


    See Also:
       Toxicological Abbreviations