228 DOI 10.1002/mnfr.200900281 Mol. Nutr. Food Res.

2010, 54, 228239

REVIEW

Toxicology and risk assessment of coumarin: Focus on

human data

Klaus Abraham, Friederike Wohrlin, Oliver Lindtner, Gerhard Heinemeyer

and Alfonso Lampen

Federal Institute for Risk Assessment (BfR), Germany

Coumarin is a secondary phytochemical with hepatotoxic and carcinogenic properties. For the Received: June 15, 2009
carcinogenic effect, a genotoxic mechanism was considered possible, but was discounted by Revised: August 20, 2009
the European Food Safety Authority in 2004 based on new evidence. This allowed the deri- Accepted: August 24, 2009
vation of a tolerable daily intake (TDI) for the first time, and a value of 0.1 mg/kg body weight
was arrived at based on animal hepatotoxicity data. However, clinical data on hepatotoxicity
from patients treated with coumarin as medicinal drug is also available. This data revealed a
subgroup of the human population being more susceptible for the hepatotoxic effect than the
animal species investigated. The cause of the high susceptibility is currently unknown;
possible mechanisms are discussed. Using the human data, a TDI of 0.1 mg/kg body weight
was derived, confirming that of the European Food Safety Authority. Nutritional exposure
may be considerably, and is mainly due to use of cassia cinnamon, which is a popular spice
especially, used for cookies and sweet dishes. To estimate exposure to coumarin during the
Christmas season in Germany, a telephone survey was performed with more than 1000
randomly selected persons. Heavy consumers of cassia cinnamon may reach a daily coumarin
intake corresponding to the TDI.

Keywords:
Coumarin / Exposure / Humans / Risk assessment / Toxicology

1 Introduction of Europe as active principles. In foods they may have a

strong flavor but they are also toxicologically relevant.
Coumarin (1,2-benzopyrone, CAS No. 91-64-5) consists of Coumarin has a long and interesting history of use and
an aromatic ring fused to a condensed lactone ring (Fig. 1). regulation. In 1822, the substance was isolated and purified
It is a naturally occurring constituent of many plants with a for the first time from tonka beans (seed of Dipteryx odorata,
pleasant spicy odor of fresh hay, woodruff or vanilla. Along also called Coumarouna odorata). After chemical synthesis in
with safrole and estragole, it belongs to the group of 1868, coumarin was marketed and used as a food flavoring
ingredients in spices and herbs that are listed by the Council for a long time [1]. In the middle of the last century,
coumarin was discovered to cause hepatic damage in
laboratory animals [2], and the addition of synthetic
Correspondence: Dr. Klaus Abraham, Federal Institute for Risk
coumarin to foods was banned, first in the USA in 1954.
Assessment (BfR), Department of Food Safety, Thielallee
88-92, 14195 Berlin, Germany
Furthermore, the formation of tumors was observed in long-
E-mail: klaus.abraham@bfr.bund.de term animal experiments [3], and for a long time, a geno-
Fax: 149-30-8412-3763 toxic mechanism of action could not be ruled out. According
to the as low as reasonably achievable principle, in 1988 the
Abbreviations: BfArM, Federal Institute for Drugs and Medical European Union set a strong coumarin limit of 2 mg/kg for
Devices; BfR, Bundesinstitut fur . Risikobewertung; EFSA,
food in general resulting from the use of natural spices and
European Food Safety Authority; o-HPA, o-hydroxyphenylace-
taldehyde; o-HPAA, o-hydroxyphenylacetic acid; NOAEL, no herbs, with exceptions for special foods (chewing gum,
observed adverse effect level; TAMDI, theoretical added maxi- caramel confectionery, and alcoholic beverages). As for other
mum daily intake; TDI, tolerable daily intake active principles regulated in the council directive 88/388/

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
Mol. Nutr. Food Res. 2010, 54, 228239 229

In this review, we take a closer look at this clinical data,

O
which in our view has not been sufficiently evaluated before,
and use this information to derive a TDI. We also survey the
possible causes of coumarin hepatotoxicity in a subgroup of
O O O O
the human population. Coumarin exposure is discussed in
Coumarin Coumarin 3,4-Epoxide
the light of data on coumarin content in cinnamon-
containing foods from Germany, and on new consumption
CH2CHO data from a telephone survey during the Christmas season.
This information allows us to make a risk assessment for
coumarin.
HO O O OH
7-Hydroxycoumarin o-Hydroxyphenylacetaldehyde (o-HPA)

2 Hazard of coumarin
CH2COOH
2.1 Data from laboratory animals

OH Extensive data is available on the toxicity of coumarin in

o-Hydroxyphenylacetic acid (o-HPAA)
laboratory animals, and these have been evaluated in various
Figure 1. Major metabolic pathways (phase I) of coumarin. overview articles and expert opinions of scientific bodies. Of
the effects observed in vivo, the carcinogenic and hepatotoxic
properties are of major importance. Hepatotoxicity was
EEC [4], the value of 2 mg/kg represented the limit of observed not only in rodents, but also in many other
detection at that time. During the following years, however, mammal species. Tumor formation was observed in long-
compliance with this limit was evidently not monitored term experiments with rodents: adenomas and carcinomas
closely by food regulatory authorities in Europe. In 2005, the of the liver and bile ducts and adenomas of the kidney in
Chemisches und Veterin.aruntersuchungsamt in Munster, . rats, as well as adenomas and carcinomas of the lung and
Germany, discovered a coumarin content of 22 mg/kg by liver adenomas in mice. Carcinomas were found only at
chance in a sample of cinnamon star cookies, (Zimtsterne, doses higher than 100 mg/kg body weight per day. Since the
a typical German Christmas cookie) www.bfr.bund.de/cm/ findings on the carcinogenic properties of coumarin in the
245/high_daily_intakes_of_cinnamon_health_risk_cannot_ 1960s, there had been discussions about their importance
be_ruled_out.pdf [5]. This prompted a great increase in for humans and the underlying mechanism of action [3, 4].
coumarin measurements in food from the German market After evaluating of new data on DNA adduct formation in
[6], and a political debate on product recalls and the level of 2004 [911], EFSA concluded that in vivo, coumarin does not
compliance with regulatory limits by the food industry in bind in a covalent manner to the DNA of target organs and
Europe [7]. As a result of these investigations and discus- therefore that its carcinogenic effect does not have a geno-
sions, new coumarin limits for cinnamon-containing foods toxic mechanism. Instead, coumarin induces tumors by a
were laid down in the European Regulation EC 1334/2008, mechanism, which is preceded by toxicity in the same target
which replaced the earlier Directive 88/388/EEC. organ, and this allows a threshold-based approach and the
These changes in the regulation of coumarin also reflect establishment of no observed adverse effect level (NOAEL)
changes in the scientific understanding of coumarin toxi- [4]. After evaluation of the available oral animal studies for
cology. On the European level, comprehensive opinions of sub-acute and chronic toxicity, hepatotoxicity in Beagle dogs
the former Scientific Committee on Food of the European [12] was identified as the most sensitive effect. In this
Commission are available from 1994 and 1999. The study, hepatotoxic effects were evident in the animals given
Committee could not rule out a genotoxic mode of action for 25 mg/kg body weight daily, but not in the animals given
tumor formation and recommended a maximum level of 10 mg/kg body weight daily (autopsies between day 297 and
coumarin of 0.5 mg/kg in foods. These opinions were day 350). These results were used to establish a NOAEL of
revised by the Panel of European Food Safety Authority 10 mg/kg body weight daily; by using a safety factor of 100
(EFSA) (the successor of the Scientific Committee on Food) (10 for interspecies variation and 10 for interhuman varia-
in 2004 [4] on the basis of new evidence for a non-genotoxic bility), a TDI of 0.1 mg/kg body weight daily was derived [4].
mode of action. This made it possible to derive a tolerable As to the mechanism of hepatotoxicity, many in vivo and
daily intake (TDI) for the first time. in vitro studies have been performed in laboratory animals to
In the past, reports and opinions dealing with the ques- elucidate the metabolism of coumarin [3, 4]. Briefly, the two
tion of risk assessment for coumarin focused primarily on most important pathways of coumarin metabolism are
animal data and its extrapolation to humans [3, 4, 8]. 7-hydroxylation leading to detoxification, which is predo-
However, considerable human data, indicating hepatotoxi- minant in primates [13], and metabolism of the lactone ring
city, is available from use of coumarin as a medicinal drug. to form a coumarin 3,4-epoxide intermediate (Fig. 1). This

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
230 K. Abraham et al. Mol. Nutr. Food Res. 2010, 54, 228239

can be conjugated with glutathione or may spontaneously with chronic lymphedema of the ipsilateral arm after treat-
degrade with the loss of carbon dioxide to form o-hydro- ment for breast cancer [20]. They received 200 mg coumarin
xyphenylacetaldehyde (o-HPA). The latter compound was or placebo twice daily for 6 months and then the other
found to be a hepatotoxic metabolite and is detoxified by medication for the following 6 months (cross-over design).
oxidation to o-hydroxyphenylacetic acid (o-HPAA). Much Blood samples were taken for the first time after 3 to 4
less o-HPAA is formed in rats than mice, explaining the months and after 1 month during the first and second
higher susceptibility of rats to coumarin-induced hepato- treatment period, respectively. The incidence of hepatotoxic
toxicity. Therefore, differences in detoxification of o-HPA are effects was substantially higher with coumarin than with
assumed to be a determining factor for species differences placebo. In none of the women did serum aminotransferase
in sensitivity to coumarin hepatotoxicity [14]. levels reach 2.5 times the upper limit of normal during the
placebo period, whereas in nine women (6%) the levels
became high during treatment with coumarin (p 5 0.006).
2.2 Human experience from coumarin use as The most prominent instance of hepatotoxicity occurred in a
medicinal drug woman who developed jaundice with a serum bilirubin
concentration of 19.3 mg per deciliter while she was receiv-
Coumarin was approved from the 1970s onwards in various ing coumarin. In these nine women, the enzyme levels
countries as a medicinal product to treat edemas caused by returned to normal after coumarin treatment was stopped.
venous (chronic venous insufficiency) and lymphatic The authors were unable to identify any predisposing
(lymphatic edemas) drainage disorders, possibly through factors, excluding, among others, therapy with tamoxifen
stimulation of proteolysis by tissue macrophages. In addi- and high body weight.
tion, direct anti-tumor activity was reported, and the In a clinical study from Germany [2123], 231 patients
substance was used to treat renal cell carcinoma and other with chronic venous insufficiency were randomly assigned
tumors, with doses of up to 7000 mg daily [15]. Some of the verum (90 mg coumarin and 540 mg troxerutin per day,
patients developed severe hepatotoxicity (toxic hepatitis, liver n 5 114) or placebo (n 5 117) for 16 weeks. Blood liver
failure in a few cases) from a few weeks to 6 months after enzymes were monitored at baseline and at five time
commencement of treatment [1618]. These observations points during the treatment. Four weeks after the beginning
led to these products being withdrawn from the market in of the therapy, nine out of the 114 patients treated
several countries in the 1990s (Australia, Belgium, France with verum (7.9%) manifested elevated transaminase levels
and Canada) [18]. above 2.5-fold the upper limit of the normal range. They
Varying frequencies of the hepatotoxic response were were analyzed with respect to causality: three were assessed
reported, with dependence on the method used to detect as unrelated, two as unlikely, three as possible and
hepatotoxicity (clinical observation only or blood sampling one as probable. Elevated liver enzymes were also
to detect elevated liver enzyme levels), as might be expected. observed in the placebo group (number not reported).
In one report, 3 of 48 patients with metastatic prostatic The authors tried to identify risk factors (Section 2.4);
carcinoma (6.3%) treated with 3000 mg coumarin daily risk factor adjusted logistic regression was performed and
responded with elevated liver enzymes [19]. Cox et al. the basic risk of elevated liver enzyme levels was estimated
observed a study group of 2173 patients with cancer or to be 4.9 and 2.2% in the verum and placebo group,
chronic infections; they were treated with 252000 mg respectively. The results were not easy to interpret, because
coumarin daily (the majority received 100 mg daily for some of the patients had a history of hepatitis and elevated
1 month, followed by 50 mg daily for 2 years; blood samples liver enzymes before start of the treatment. In addition, it
were taken every 3 months). Seventeen patients developed has to be taken into account that coumarin was adminis-
elevated liver enzymes levels of sufficient magnitude (i.e. tered as co-treatment with troxerutin, which may have a
alanine aminotransferase levels between 115 and 960 U/L, hepatoprotective effect (study findings from isolated
at least double the normal maximum level of 3550 U/L). perfused rat liver [24]). This would be a reason to
Four of them were diagnosed as probably having other expect a stronger hepatotoxic effect when coumarin is
causes than coumarin treatment. Of the remaining 13 administered alone.
patients (0.60%), elevated liver enzymes levels returned to The published data is consistent with the existence of a
normal while still on coumarin in five, and only in the subgroup of the human population that reacts sensitively to
remaining eight (0.37%) was the hepatotoxicity attributed to coumarin with hepatotoxic effects. If recurrent blood
coumarin treatment. Several of the patients were re-exposed sampling is used to monitor liver enzyme levels, this group
to coumarin showing the same hepatotoxic response, but may amount to a single-digit percentage of the population,
with a faster onset (Section 2.4) [16]. much higher than in drug responses often classified as
A more reliable estimate of the percentage of patients idiosyncratic. No clear-cut dose-dependent increase in
sensitive to the hepatotoxic effect of coumarin can be severity of the effect was observed within the subgroup,
expected from systematic placebo-controlled studies with although liver failure was observed only in patients treated
recurrent blood sampling. Loprinzi et al. studied 140 women with high doses (more than 100 mg daily). On the other

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
Mol. Nutr. Food Res. 2010, 54, 228239 231

hand, people not sensitive to the effect evidently tolerate Case reports evaluated by Bergmann also allow an esti-
daily coumarin doses in the gram range. Drug-dependent mation of the time period critical for the onset of hepatitis in
moderate elevation of liver enzyme levels is often tolerated sensitive subjects [25]. The shortest periods documented
in clinical practice; if positive effects for the patient are were 5 days (dose: 90 mg daily), 16 days and 18 days (dose:
expected, treatment is often continued with monitoring of 30 mg daily each). Therefore, it is not acceptable to signifi-
the laboratory parameters. However, such risk-benefit cantly exceed the TDI over several weeks. The European
considerations are not permissible for foods. These have to Commission asked EFSA for an opinion on the relatively
be safe in any case, and if dietary exposure to coumarin high cinnamon exposure during the Christmas season
caused elevated blood liver enzyme levels, even if they were (Section 3.3), and EFSA concluded in 2008 that exposure to
reversible, would not be acceptable. coumarin resulting in an intake three times higher than the
TDI for 12 weeks is not a cause for concern [27].

2.3 Hazard assessment using human data

2.4 What is the cause of coumarin hepatotoxicity in
As outlined above, the frequency as well as the severity of the human subgroup?
coumarin hepatotoxicity in the human subpopulation is
relevant, and the effect should be considered in hazard As outlined above, the varying susceptibility to hepatotoxicity
assessment. Due to underreporting, the cases known to the of coumarin observed in animal species was attributed to the
authorities cannot be used to estimate the frequency of these varying ability to detoxify the substance via the 7-hydroxyla-
reactions, but they can be used to identify the lowest daily tion pathway. In the human population, 7-hydroxylation of
dose of coumarin able to cause hepatotoxicity. This dose can coumarin is catalyzed by a high-affinity CYP2A6 liver enzyme
be used as a staring point for deriving a TDI. which is the only enzyme metabolizing this reaction in the
In 1999, an expert opinion on the assessment of human liver [28, 29]. Due to this predominant pathway,
coumarin in medicinal products with regard to a hepatotoxic coumarin usually has a large first pass effect in the human
effect in humans was commissioned by the Federal Insti- liver after oral administration (9498%) [30], leading to rapid
tute for Drugs and Medical Devices (BfArM) in Germany urinary excretion of 7-hydroxycoumarin and its glucuronide.
[25]. The 82 case reports (international notifications) of Using 5 mg coumarin as test substance to investigate the
possible coumarin-associated liver damage available to the CYP2A6 phenotype, a mean total 7-hydroxycoumarin
Institute at that time were evaluated from a pharmacological formation of 64% (range 20100%) of the dose was observed
perspective. These reports included cases of liver failure in a European population of 110 volunteers; more than 95%
(survival of the patients) and seven fatalities. A dose classi- of the metabolite formed was excreted within 4 h [31]. In
fication was possible for 51 cases from France, Ireland and subjects with high excretion rates of 7-hydroxycoumarin
Germany. The most frequent daily dose was 90 mg measured after oral coumarin administration, 0.13% of the
coumarin prescribed for the main indication lymphatic dose or less was found as 3-hydroxycoumarin [32, 33]. In the
disorders and varicose veins. Five cases (10%) occurred at same investigation, mean urinary excretion of o-HPAA (via
the lowest doses (25 and 30 mg daily); of the three cases 3,4-epoxidation) was found to be 3.5% (range 1.87.0%) of a
from Germany documented in more detail, two had devel- dose of 1000 mg coumarin [34], comparable to the number of
oped hepatitis. According to the expert report, for part of the 4% (range 16%) of the dose of 200 mg coumarin reported
population liver damage cannot be ruled out at a daily dose earlier for eight volunteers [35].
of 25 mg coumarin. The human data on rapid detoxification of coumarin is
In order to extrapolate from this effect level to a human consistent with data of patients (Section 2.2) treated for
NOAEL, a factor of 5 is considered justified in the case of a weeks with high doses up to 7000 mg coumarin daily,
severe effect at the lowest observed adverse effect level. This corresponding to maximal doses of more than 100 mg/kg
results in a level of 5 mg coumarin per day, which is body weight daily (assumed weight 60 kg) tolerated without
expected to cause no adverse effects even in sensitive signs of hepatotoxicity [15]. Such a dose was lethal to two
subjects. When choosing this factor, it was borne in mind Beagle dogs within 9 and 16 days, respectively [12]. The
that knowledge on the mechanism of action in sensitive apparently low susceptibility to the hepatotoxic effect of
individuals is not available. As this group of persons must coumarin in the majority of the human population, as
already be viewed as the most sensitive subgroup in the compared with many animal species, was attributed to
population, no additional intraspecies factor was applied. detoxification via the 7-hydroxylation pathway predominant
Using the established safe daily dose of 5 mg coumarin for in humans. The ability to catalyze the 7-hydroxylation of
an adult weighing 60 kg, a (rounded) TDI of 0.1 mg/kg body coumarin was tested using liver microsomes from nine
weight was derived by the Federal Institute for Risk mammalian species. 7-Hydroxycoumarin was the major
Assessment (BfR) [5, 26]. This value based on human data metabolite (greater than 70%) in humans and monkeys, but
agrees with, and lends support to the EFSA value based on only a minor metabolite in rat (less than 1%), mouse (3%)
animal data [4]. and dogs (18%) [13]. According to these and comparable

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
232 K. Abraham et al. Mol. Nutr. Food Res. 2010, 54, 228239

results of other authors [3], it was discussed whether the blood sampling revealed a proportion of sensitive
interspecies factor used in risk assessment for extrapolation subjects in the single-digit range [20, 22]. In addition,
from animals to humans should be reduced [4, 8, 27]. CYP2A6 genotyping of 216 patients of the German
However, as outlined above, a relevant subgroup of the study [22] revealed 7.4% subjects with defective geno-
human population is much more susceptible, as is most type (CYP2A62 or CYP2A63 allele) all found to be
impressively demonstrated by the study of Loprinzi et al. [20] heterozygous for the variant alleles; of the nine patients
and with some limitations by the German study [22]. From with elevated liver enzyme levels in the verum group,
the daily doses applied in these studies (about 6.7 and only one carrier of a variant allele was identified who
1.5 mg/kg body weight, respectively, assuming a body exhibited an isolated g-GT elevation without concomi-
weight of 60 kg) and from even lower doses (down to about tant increase in transaminases. Additional genotyping
0.5 mg/kg body weight) in single cases reported to the of affected patients for the deletion of the CYP2A6 gene
authorities [25] it can be concluded that individuals of the (CYP2A64 allele) revealed no further polymorphism.
human subgroup are more susceptible than various animal Therefore, no evidence was obtained that the poly-
species investigated (NOAEL of 10 mg coumarin per kilo- morphism in CYP2A6 is a determinant of the coumarin-
gram body weight daily in Beagle dogs identified as most associated elevation of blood liver enzymes [21].
sensitive species [4]). (ii) Rietjens et al. used a physiologically based toxicokinetic
The cause of this higher susceptibility is unknown. model to predict liver levels of the toxic o-HPA metabolite
Unfortunately, no phenotype data of coumarin metabolism in rats and in human subjects with normal or deficient
is available from any patient affected by a hepatotoxic CYP2A6catalyzed coumarin 7-hydroxylation phenotype;
response following the treatment with coumarin. A genetic data from the literature as well as data obtained from in
polymorphism of CYP2A6 as underlying mechanism has vitro investigations of microsomes was used to determine
been discussed for a long time, possibly leading to an kinetic parameters for coumarin metabolism of animals
increased formation of 3,4-coumarin epoxide and o-HPA. and humans [40]. Modeling allowed the prediction of
This has been observed in a homozygous individual with an maximum tissue concentration of o-HPA in the liver of
inactivating CYP2A62 allele: Following the oral adminis- wild-type human subjects and of subjects deficient in
tration of 2 mg coumarin, about 50% of the dose was 7-hydroxylation which were three and one order of
excreted in the 8-h urine as o-HPAA, whereas 7-hydro- magnitude, respectively, lower than the values predicted
xycoumarin could not be detected; the urinary metabolite for rat liver. The authors concluded that even when
excretion of the heterozygous parents was not found to 7-hydroxylation activity is missing, the formation of the
differ from that of controls [36]. However, this is the only in hepatotoxic o-HPA metabolite will be significantly lower
vivo observation with documentation of alternative metabo- in the liver of humans than those expected in the liver of
lism via the 3,4-epoxidation pathway in a CYP2A6deficient rats when exposed to a similar (low) dose per body
subject. A very low or missing urinary excretion of weight. As the rat is a relatively coumarin-sensitive
7-hydroxycoumarin after oral administration of coumarin species [41] comparable to dogs (NOAEL about 10 mg/kg
was found in a significant proportion of Asians who lack the body weight) [12], it can be concluded that o-HPA is
CYP2A6 protein completely due to the relatively high inci- unlikely to be the toxic agent causing hepatotoxicity in the
dence of CYP2A6 gene deletion alleles in Japanese and human subgroup which is more susceptible than rats
Chinese populations. The frequency of poor metabolizers in and dogs, as outlined above.
Asian populations (up to 20%) is much higher than in
Caucasian populations [3739]. Unfortunately, the possible From these observations it can be reasoned that CYP2A6
alternative routes of metabolism were not investigated in polymorphism with deficient 7-hydroxylation of coumarin is
any of the studies of individuals identified as deficient in the not the cause of the high susceptibility to hepatotoxicity in the
7-hydroxycoumarin pathway. human subgroup. Regarding other possible causes, previous
Therefore, it is currently unknown what CYP2A6 poly- hepatitis and elevated baseline g-GT were identified as risk
morphism with deficient 7-hydroxylation of coumarin factors in the German coumarin study [22]. At least two
means with respect to the 3,4-coumarin epoxide pathway disease states seem to reduce coumarin 7-hydroxylation. In
and the possible generation of toxic metabolites. In addition patients with alcohol-induced liver disease, mean urinary
to this lack of evidence, further observations do not support 7-hydroxycoumarin excretion over 2 h was decreased in severe
the assumption of a CYP2A6 polymorphism as cause of the (18.0% of dose) and moderate (34.2% of dose), but not in
higher susceptibility in the human subgroup: mild (49.7% of dose) disease relative to controls (56.2% of
dose) [42]. In previously healthy adult patients with acute
(i) The vast majority of patients with hepatotoxic response jaundice from hepatitis virus A infection, mean total urinary
following the treatment with coumarin is of Caucasian 7-hydroxycoumarin excretion (08 h) was decreased by 37%
race; in this population, the frequency of poor meta- compared with the values obtained from healthy volunteers
bolizers is close to zero [37]. In contrast, the two [43]. However, the patients of the German coumarin study
systematic placebo-controlled studies with recurrent did not suffer from severe or acute liver disease, and

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
Mol. Nutr. Food Res. 2010, 54, 228239 233

moderately decreased CYP2A6 activity levels are not expected Cassia cinnamon) [46]. The chemical composition of the two
to result in a strong increase of o-HPA liver concentrations cinnamon species is different, particularly with respect to
(modeling data [40]). Therefore, the risk factors of previous their coumarin levels. Coumarin concentrations were
hepatitis and elevated baseline g-GT are probably not related detected from below the detection limit to 190 mg/kg in
to impaired coumarin detoxification. Ceylon cinnamon (n 5 12) and from 700 to 12 230 mg/kg
Since evidence for a metabolic cause of the high in cassia cinnamon (n 5 12) [47]. Due to these high
susceptibility to hepatotoxicity in a human subgroup is concentrations in cassia cinnamon (compared with other
missing, other possible causes have to be considered as well. foods), it seems obvious that despite the relatively low
Of the patients with hepatotoxic response described by Cox amounts of the consumption of spices coumarin exposure
et al., about half were re-treated with coumarin (after from food consumption is mainly due to cassia cinnamon.
cessation of medication and return of elevated liver enzymes This applies to both the direct addition of cinnamon to foods
levels to normal) [16]. The authors observed that the time to but also the use of cinnamon oils by the food industry.
onset of a rise in the liver enzymes was much shorter than Cinnamon is available in the grocery store in dried form
in the first treatment period. In addition, they observed a as cinnamon powder or as sticks; for the latter only, a visual
favorable response (lower liver enzyme levels) to immuno- differentiation between cassia and Ceylon cinnamon is
suppressive therapy in three patients while on coumarin. possible. On the German retail market, mostly cassia
These observations as well as the absence of any clear dose- cinnamon is available [48]; in the majority of cases, however,
dependency of the severity of response within the subgroup the botanical species is not indicated on the packaging.
suggest an immune mechanism may be involved in the Analyses of cinnamon samples and of the most
coumarin-induced hepatotoxicity in humans. To test for a important cinnamon-containing foods done by the Federal
drug-mediated allergic response, lymphocytes from the States in Germany in 2006 and reported to the Federal
patients of the German coumarin study with elevated liver Office of Consumer Protection and Food Safety up
parameters were incubated with coumarin and 7-hydro- to March 2007 revealed the coumarin levels listed in Table 1.
xycoumarin; however, none of them showed a positive Some analyses done by the Chemisches und Veter-
lymphocyte proliferation test response with a stimulation in.aruntersuchungsamt Karlsruhe were published in more
index of more than three [22]. In general, hepatotoxic detail [6]. With a median coumarin level of 2920 mg/kg
responses observed during treatment are classified as (maximum 8790 mg/kg) the analyses confirmed the predo-
adverse drug reactions which do not occur in most minance of cassia cinnamon in the retail market. Coumarin
patients at any readily achieved dose of a drug and do not levels in cinnamon-containing foods were found in the
involve the known pharmacological effects of the drug [44]. range expected from the recipes. For example, cinnamon
star cookies (Zimtsterne) with the highest cinnamon
content (about 1%, see also Section 3.3) were found to have
3 Human exposure the highest coumarin content (median 39.4 mg/kg, maxi-
mum 113.3 mg/kg). Most of these coumarin levels consid-
3.1 Coumarin in cinnamon species erably exceeded the maximum permitted level (2 mg/kg) of
the EU Flavorings Directive 88/388/EEC. Following the
There has been no large-scale analysis of coumarin in foods. public discussion in Germany, most of the manufacturers
The substance is contained in various plants (for example took measures to reduce the coumarin content in 2007 and
sweet clover, tonka beans, lavender); however, given the 2008.
general eating habits in most countries, these are probably
not relevant for nutritional coumarin exposure. Recently,
coumarin levels in specific plants or their essential oils were 3.2 Estimate of oral coumarin exposure
analyzed using a very sensitive method. Significant
concentrations in plants relevant for food consumption were In its opinion of 2004, the EFSA calculated a theoretical
only found in cassia cinnamon (see below) and in woodruff added maximum daily intake (TAMDI) of 1.5 mg coumarin
(Asperula odorata, 203 mg/kg) [45]. The latter is used in for an adult with a default body weight of 60 kg (0.025 mg/
Germany to flavor May punch (Maibowle) and has a long kg body weight per day) [4]. The calculation was performed
history of regulations due to its coumarin content. However, considering the concomitant consumption of 324 g general
the alcoholic beverage is consumed by very few people and beverages, 133.4 g general solid food, 27 g sweets, 2 g
for only a short period each year. chewing gum and 20 g alcoholic beverages each with the
Cinnamon bark is the dried inner bark of the shoots maximum level of coumarin permitted by the Flavorings
grown on cut stock of Cinnamomum verum J.S. Presl (Syn. Directive 88/388/EEC (2 mg/kg for foodstuffs and beverages
Cinnamomum zeylanicum Nees, true cinnamon, Ceylon in general). Lake calculated a TAMDI of 4.1 mg coumarin
cinnamon) or of the trunk bark, freed of cork, of Cinnamo- from general consumption data, which he rated as unrea-
mum cassia Blume (Syn. Cinnamomum aromaticum Nees, listically high. A value of 1.2 mg coumarin (0.02 mg/kg body
Chinese cinnamon, Saigon cinnamon, Vietnam cinnamon, weight for 60 kg weight) seemed to him to be more realistic

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
234 K. Abraham et al. Mol. Nutr. Food Res. 2010, 54, 228239

Table 1. Coumarin levels (mean, median and maximum) in cinnamon and cinnamon-containing foods from the German market measured
in 2006

n5 Mean (mg/kg) Median (mg/kg) Maximum (mg/kg)

Cinnamona) 170 2680 2920 8790

Tea with cinnamon 16 231.3 105.0 918.0
Cinnamon star cookies (Zimtsterne) 218 37.7 39.4 113.3
Cereals with cinnamon 28 25.5 23.9 60.0
Almond cookies (Spekulatius) 40 16.2 17.0 30.2
Gingerbread cake (Lebkuchen) 80 10.3 7.8 46.0
Desserts with cinnamon 29 10.2 10.8 19.0
Chocolate with cinnamon 25 9.4 5.6 32.9
Mulled wine 48 0.2 0.1 4.3

Analyses were done by the Federal states in Germany in 2006 and reported to the Federal Office of Consumer Protection and Food Safety
(BVL) until March 2007. For all foods, minimum values were below the limit of detection.
a) Ground to powder in most cases (no differentiation between cassia and Ceylon cinnamon).

(assumption: maximum 5% of solid food flavored with kilogram body weight (exposure on single a day, normally by
cinnamon) [3]. These two worst-case estimates were based eating rice pudding with cinnamon and sugar). Exposure
on high amounts of flavored foods consumed daily and on lasting for longer periods was estimated by a worst-case
the maximum level of coumarin permitted by the Flavorings approach assuming two of these meals per week. This
Directive 88/388/EEC. However, due to the predominant results in a daily exposure of 0.063 g cinnamon per kilogram
use of cassia, the majority of the cinnamon-containing foods body weight per day. Based on a coumarin level of 3000 mg/
in Germany were found to be far above this level, as shown kg cinnamon this would indicate a coumarin exposure of
by the data in Table 1. Therefore, the TAMDI approaches 0.19 mg/kg body weight as the worst case for oral exposure
mentioned [3, 4] cannot be considered as reliable estimates of the 25 year-old children [5].
at least for Germany. A simple estimate of coumarin exposure in young chil-
Even with the improvements in knowledge on the dren is possible using the levels measured in Christmas
coumarin levels in food (Table 1), it is still not easy to cookies (Table 1). For a 4-year-old child weighing 15 kg, a
estimate the maximum daily intake of coumarin from daily consumption of three cinnamon star cookies (Zimt-
cinnamon and cinnamon-containing foods. Epidemiological sterne weighing about 6 g each) with the maximum level
data on consumption of spices is inadequate. This is not measured already results in a coumarin exposure of
only due to missing consumption data on particular spiced 0.13 mg/kg body weight daily higher than the TDI
foods, but also due to a lack of estimates of the amount of (0.046 mg/kg body weight daily for the cookies with the
spice used by the consumer at home (e.g. the use of cinna- median level). This calculation allows an estimation of the
mon and sugar to spice rice pudding). In addition, a coumarin intake from single foods; however, other
simultaneous dermal coumarin exposure may also need to coumarin-containing foods may be consumed simulta-
be considered (Section 3.4). As an approach to estimate the neously. In order to get an estimation of the total coumarin
seasonally higher consumption of cinnamon-containing intake during Christmas time (in adults), a telephone survey
foods during Christmas time, BfR has directed a telephone has been performed.
survey of adults (Section 3.3).
When estimating the consumption of cinnamon, young
children require a separate consideration as a group with 3.3 Telephone survey on cinnamon in Christmas-
possibly high exposure on a body weight basis (higher food related foods
consumption due to their higher energy requirements and
special eating habits). In this case, exposure data from the Christmas time is the period of highest consumption of
German VELS study is helpful (food consumption study to cinnamon-containing foods. Typical Christmas treats like
determine the dietary intake of infants and toddlers in order almond cookies (Spekulatius), gingerbread cakes (Lebku-
to estimate the acute toxicity risk from pesticide residues). chen) and cinnamon star cookies (Zimtsterne) have a long
For two non-consecutive 3-day periods, parents kept food tradition in Germany as well as other European countries. In
records [49]. The evaluation of toddlers aged between 2 and addition, a variety of teas, chocolates and desserts containing
5 years (n 5 475) showed that 140 children ate cinnamon or cinnamon have been launched in recent years to have a
cinnamon-containing products at least on one of the six days typical taste and flavor of winter and Christmas. In order to
recorded; 47% of the consumption days were between estimate the total amount of cinnamon consumed with
September and December. For these consumers the 97.5 Christmas-related foods by German adults, the BfR
percentile showed consumption of 0.22 g cinnamon per performed a questionnaire survey by telephone.

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
Mol. Nutr. Food Res. 2010, 54, 228239 235

Table 2. Christmas-related foods in Germany asked for in the telephone interviews: estimated average cinnamon content, size of one portion,
resulting content of cinnamon in one portion, and relative contribution of each food to the total cinnamon consumption of the
interview sample.

Average cinnamon Size of one Cinnamon content in Proportion of the total cinnamon
content (g/kg) portion (g) one portion (g) consumption (%)

Teaa) 77.1b) 200 0.10c) 22.2

Mulled wine 0.07b) 200 0.02 0.9
a)
Dessert 3.4b) 150 0.52 4.4
Chocolatea) 2.8b) 100 0.28 10.2
Almond cookies (Spekulatius) 5.8b) 7.9 0.05 19.3
Domino cookies (Dominosteine) 0.5d) 12.4 0.01 1.1
Cinnamon star cookies (Zimtsterne) 12.6b) 5.8 0.07 5.2
Small gingerbread cake 3.2b) 10.0 0.03 7.5
Large gingerbread cake 3.2b) 50.0 0.16 9.8
Home-baked cookies 5.0e) 17.0 0.09 19.4

a) Marketed as a winter/Christmas specialty.

b) Estimated from mean content of coumarin in Table 1, assuming a coumarin content in cinnamon of 3000 mg/kg.
c) Assuming the use of 2 g tea for a cup containing 200 mL and a coumarin transfer rate of 50% (average rate of transfer experiments
performed by the BfR).
d) A Dominostein is a sweet primarily sold during Christmas season in Germany. On average one weighs 12.4 g, the base consists of
15% of soft gingerbread (manufacturer information). Next are a layer of jelly and a layer of either marzipan or persipan, it is covered
with a thin icing of dark chocolate. Since there was no coumarin data available, cinnamon contents had to be estimated based on
measurements of gingerbread cakes.
e) Basing on a list of cinnamon containing Christmas cookie recipes (n 5 30), on average one cookie weighs about 17 g. Regarding these
recipes cinnamon containing homemade cookie dough consists on average of 0.5% of cinnamon. The first question was about the
number of consumed homemade cookies, and then the interviewee had to estimate how many of these cookies did contain cinnamon
(34% of homemade cookies did reputedly contain cinnamon). If cinnamon proportion was not specified, this value was taken as
computation base.

The survey was carried out by USUMA GmbH, an inde- mate. In case of winter/Christmas specialty teas, the use of
pendent research institute for questionnaire surveys in 2 g tea for a cup containing 200 mL and a coumarin extraction
Berlin. During three days of the third week of December in rate of 50% was assumed. Possible co-exposure to other
2006, 1012 persons aged 14 years or older living in German- cinnamon-containing meals eaten throughout the year (for
speaking private households were interviewed using a stan- example: rice pudding with home-used cinnamon or break-
dardized questionnaire. The interviews lasted 12 min on fast cereals with cinnamon) was not considered in the survey.
average. The amount of selected foods was assessed by asking The total coumarin intake was estimated for each of the
for the frequency of consumption within a typical Christmas interview partners by adding up the intake data from the ten
week and typical portion sizes. To validate the estimates different Christmas-related foods. Figure 2 shows the
given by consumers for 1 week, a 24 h recall question was distribution of all of the estimated individual consumptions
included for each of the selected food items. A plausibility per week, ordered by total amounts. As expected for spice-
check excluded two interviews because of implausibly high related consumptions, a high variation was observed:
consumption of cinnamon-containing foods. The ques- 42 people (4.2%) reported that they did not consume any of
tionnaire also included socio-demographic parameters like the 10 foods, whereas 154 people (15.2%) consumed half of
age, gender and education. The sample was randomly the total coumarin amount. Mean coumarin intake was
generated and population-proportionally weighted to adjust estimated to be 5.0 mg coumarin per week; median, 95th
for unavoidable biases. Hence the results are representative and 97.5th percentile were estimated to be 3.2, 15.9 and
for the German population aged 14 years and older. 21.6 mg coumarin per week, respectively. The coumarin
Interview partners were asked about their average intake of the heaviest consumers (six subjects) was esti-
consumption of ten different cinnamon-containing foods mated to be higher than 35 mg per week (maximally 47.5 mg
presented in Table 2 (which also shows the estimated average per week). These consumers more or less reach an estimated
cinnamon content and portion size of each food item). To daily coumarin intake of 0.1 mg/kg body weight (assumed
assess the individual coumarin intake, consumption data was weight 60 kg). These results confirm the often-used worst-
multiplied by the mean coumarin levels of the foods in case rule of thumb that the heaviest consumer may eat up to
Table 1 (food monitoring data of 2006). In case of Domi- ten times more than the average consumer.
nosteine and home-baked Christmas-cookies, only data on Regarding subgroups of the population, average
their cinnamon content was available; therefore, the coumarin intake was lower in women (5.0 mg per week)
coumarin content was estimated assuming an average than in men (5.8 mg per week); this may be due to their
coumarin content of 3000 mg/kg cinnamon as a rough esti- lower general food and calorie intake. The age distribution

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
236 K. Abraham et al. Mol. Nutr. Food Res. 2010, 54, 228239

estimated weekly intake of coumarin (mg)

50

45

40

35

30

25

20

15

10 Figure 2. Individual weekly coumarin intake

from Christmas-related foods in Germany
5
in 2006, estimated from the telephone
0 survey of 1012 consumers. The estimated
0 100 200 300 400 500 600 700 800 900 1000 intake of each individual subject is presen-
ted; subjects were sorted by their intake
individuals (sorted and numbered by coumarin intake) value.

showed a distinct trend towards a lower consumption at an average consumption of 17 g cinnamon or roughly half of
higher ages. Whereas the average intake of the group of 14- the total annual cinnamon consumption during the
to 24-year-old persons was 8.4 mg coumarin per week, it was Christmas period. This is considered plausible.
5.0 mg per week in the group of 35- to 54-year-old persons,
and 3.1 mg coumarin per week in the group of the more
than 75-year-old persons. 3.4 Dermal exposure
As expected, the contribution of the ten different foods to
the total weekly intake of coumarin in the interview sample In addition to oral exposure, a simultaneous dermal expo-
was different (Table 2). The main contributions were from sure has to be taken into account as well. This is an
winter/Christmas specialty teas (22.2%), almond cookies uncommon case in risk assessment of food ingredients. In
(Spekulatius, 19.3%), home-baked cookies with cinnamon contrast to food production, coumarin is used without
(19.4%) and gingerbread cakes (small and large size, 17.3%). constraints as a fragrance in cosmetics, leading to a dermal
Despite their relatively high coumarin content, cinnamon exposure of consumers, which is by no means insignificant.
star cookies (Zimtsterne) accounted for only 5.2% of the The annual world production for the use as a fragrance in
total coumarin intake, as they only were eaten by a few cosmetics today amounts to about 2000 tonnes [50]. Based
people: nearly 80% of the interview partners reported not on production data from the USA, an average daily
eating any of these cookies. This may have been influenced coumarin amount of 1.2 mg per US American was calcu-
by the public discussion on coumarin in cinnamon and lated [51]. As the formulations of cosmetics are not normally
especially on cinnamon star cookies which took place in made public, only insufficient data is available on coumarin
Germany in late 2006. Sixty-seven percent of the interview levels. Lake referred to a compilation of the International
partners said they had noticed warnings in the media Fragrance Association in Geneva, which indicates a
against a high consumption of cinnamon and cinnamon- coumarin level of 6.4% in a few thousand fragrance
containing foods. mixtures as the 97.5th percentile [3]. Based on this value he
The average consumption of cinnamon obtained from estimated a daily dermal exposure of 9.8 mg coumarin for
the telephone interviews was checked against market data adults (0.16 mg/kg body weight for a person weighing 60 kg)
from Germany. A total consumption of 2887 tonnes as a worst case. However, this amount appeared to be
(amount imported minus amount exported) corresponding unrealistically high for him and he re-estimated a daily
to an average consumption of 35.2 g cinnamon per year in intake of 2.3 mg coumarin (0.04 mg/kg body weight for an
Germany (population: 82 million) and a weekly average of adult weighing 60 kg) as reasonable worst-case scenario.
0.68 g cinnamon has been reported [48]. This number is This calculation reflected the fact that coumarin is absorbed
lower than that of 1.7 g per week estimated from our study quickly and efficiently via the skin (absorption rate
(assuming a coumarin content of 3000 mg/kg cinnamon). approximately 60%: [3, 51]). Currently it is under discussion
However, an important proportion of the yearly cinnamon whether the risk of a hepatotoxic effect of coumarin from
consumption is expected to take place during Christmas dermal uptake is comparable to that of from oral intake. Due
time. If we assume that Christmas-related foods are offered to slower absorption and the fact that the first pass
intensively over a 10-week period, this would correspond to phenomenon does not apply, hepatic peak concentrations of

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
Mol. Nutr. Food Res. 2010, 54, 228239 237

coumarin are expected to be much lower after dermal weeks. In autumn 2006, the patient became aware of the
compared with oral exposure to the same dose. Accordingly, public discussions on cinnamon and hepatotoxicity. She
coumarin would be much less hepatotoxic after remembered that as a big fan of cinnamon, during the 12
dermal compared to oral exposure, if hepatotoxicity months before the onset of her hepatitis, she had consumed
is a threshold effect depending on the peak concentration, even higher amounts than before (about 12 g cinnamon
but not if hepatotoxicity is related to the area under the every day, used to spice different foods). It is therefore
curve. The question of which dose metric is relevant, suspected that this case was caused by high consumption of
especially for the sensitive human subpopulation, is cassia cinnamon.
currently open.

5 Summary and concluding remarks

4 Risk assessment
Clinical data from patients treated with coumarin (case
As is evident from the estimation of exposure, heavy reports and controlled studies) revealed the existence of a
consumers may reach a daily coumarin intake of 0.1 mg/kg relevant subgroup making up a single-digit percentage of
body weight corresponding to the TDI just from the the human population, which is sensitive for the hepatotoxic
consumption of Christmas-related foods during autumn and effect. In contrast to coumarin-treated patients, even slightly
winter. With an additional consumption of cinnamon, these elevated liver enzyme levels, as a sign of liver damage, are
consumers may exceed the TDI. Similarly, a heavy home-use not acceptable (even if reversible) in case of consumers
of cinnamon as spice itself (for example rice pudding with eating food with a high coumarin content.
sugar and cinnamon) may lead to an exposure in excess of The underlying mechanism of coumarin-related hepato-
the TDI during a relevant period of time, especially if the toxicity in a human subgroup has not yet been elucidated.
particular product is at the upper end of the range of Considering all the data available, evidence for a genetic
coumarin content (possibly higher than 8000 mg/kg). polymorphism of CYP2A6 with deficient 7-hydroxylation of
Therefore, no risk of hepatotoxicity of coumarin is coumarin as the cause of high sensitivity is missing. Espe-
expected for the vast majority of the population consuming cially investigations of coumarin metabolism pathways in
low or moderate amounts of (cassia) cinnamon. However, patients with hepatotoxic response following the treatment
consumers who really eat a lot may have a risk if they belong with coumarin and in subjects with deficient 7-hydroxyla-
to the sensitive subgroup. tion would be helpful to clarify the toxic mechanism. As
During the public discussions on coumarin and cinnamon long as this data is not available the question of a possibly
in autumn 2006 in Germany, the question was raised whether metabolic cause for the hepatotoxicity in the human
coumarin in the food matrix of cinnamon has the same subgroup remains open, and other possible mechanisms
bioavailability as the pure compound used in animal experi- have to be considered as well.
ments and human medicine. This issue was addressed by the Evidence of coumarin hepatotoxicity in a subgroup of the
German Senatskommission zur Beurteilung der gesundhei- human population is striking and has to be considered for
tlichen Unbedenklichkeit von Lebensmitteln (SKLM) http:// risk assessment of coumarin, whatever the underlying
www.dfg.de/aktuelles_presse/reden_stellungnahmen/2006/ mechanism may be. Considering the lowest dose (25 mg
download/sklm_natinh_en_05092006.pdf. [52]. It was coumarin daily) with documentation of a hepatotoxic
concluded that in the absence of data on the influence of the response in patients, a TDI of 0.1 mg/kg body weight daily
respective constituent and the respective matrix in a food- was derived identical to the TDI derived by EFSA in 2004
stuff, toxicological data on the pure compound should be using animal data [4]. For the latter derivation, an inter-
used as a basis for risk assessments. Clarification of the species factor of 10 was used, as confirmed by EFSA in 2008
issue can only be achieved by further investigations on a [27]. A reduction of this factor [53] from 10 to 2.5 (no kinetic
case-by-case basis. Currently, the BfR is performing a sub-factor) was suggested by Felter et al. [8], based on the
human study to investigate the bioavailability of pure fact that in contrast to humans the CYP2A6 mediated
coumarin compared to coumarin in different foods detoxification to 7-hydroxycoumarin is only a minor pathway
containing cassia cinnamon. in many animal species including rodents and dogs. Their
Evidence for the possibility of a hepatotoxic response TDI of 0.64 mg/kg body weight daily, starting from of a
during high consumption of cassia cinnamon arises from a NOAEL in rats of 0.16 mg/kg body weight daily [41], would
case reported to the BfR. In summer 2006, a 23-year-old be equivalent to an absolute dose of about 38 mg coumarin
woman was hospitalized with acute hepatitis. Laboratory (assuming a body weight of 60 kg); for these doses, however,
tests produced no evidence of a viral infection. Liver biopsy elevated liver enzyme levels and/or hepatitis is documented
showed distinct signs of inflammation, acute cholangitis in sensitive people. Therefore, a reduction of the inter-
and canalicular cholestasis. A toxic hepatitis was presumed, species factor, based on kinetic data only, may be misleading
but anamnesis did not turn up any hints as to its cause. A in risk assessment, because a high susceptibility in a rele-
slow but continuous recovery was observed in the following vant human subgroup may be due to dynamic causes not

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
238 K. Abraham et al. Mol. Nutr. Food Res. 2010, 54, 228239

covered by an interspecies factor of 2.5 and an intraspecies Hull, NC 27599, USA. April 2003, Submitted by European
factor of 10. Flavour and Flagrance Association (EFFA), Square Marie-
Comparison of the toxicological data on coumarin avail- Louise, 49, B-1000, Brussels.
able from animals and humans on the other hand demon- [12] Hagan, E. C., Hansen, W. H., Fitzhugh, O. G., Jenner, P. M.
strates that toxicological mechanisms may not be identical, et al., Food flavourings and compounds of related structure.
and that in humans an important subgroup may be much II. Subacute and chronic toxicity. Food Cosmet. Toxicol.
more susceptible than the majority in the population. If 1967, 5, 141157.
available, data on toxicological effects in humans should be [13] Pearce, R., Greenway, D., Parkinson, A., Species differences
considered in risk assessment with high preference. Often and interindividual variation in liver microsomal cyto-
such data is not available and risk assessment has to use chrome P450 2A enzymes: effects on coumarin, dicumarol,
and testosterone oxidation. Arch. Biochem. Biophys. 1992,
animal data. In this case, a reduction of safety factors during
298, 211225.
extrapolation from animals to humans should only be made
if this is securely justified. [14] Vassallo, J. D., Hicks, S. M., Daston, G. P., Lehman-
McKeeman, L. D., Metabolic detoxification determines
species differences in coumarin-induced hepatotoxicity.
The authors have declared no conflict of interest.
Toxicol. Sci. 2004, 80, 249257
[15] Marshall, M. E., Mohler, J. L., Edmonds, K., Williams, B.
et al., An updated review of the clinical development of
6 References coumarin (1,2-benzopyrone) and 7-hydroxycoumarin.
J. Cancer Res. Clin. Oncol. 1994, 120, S3942.
[1] Egan, D., OKennedy, R., Moran, E., Cox, D. et al., The [16] Cox, D., OKennedy, R., Thornes, R. D., The rarity of liver
pharmacology, metabolism, analysis, and applications of toxicity in patients treated with coumarin (1,2-benzopyr-
coumarin and coumarin-related compounds. Drug Metab. one). Hum. Toxicol. 1989, 8, 501506.
Rev. 1990; 22, 503529.
[17] WHO, Coumarin: a strong association with hepatotoxicity.
[2] Hazleton, L. W., Tusing, T. W., Zeitlin, B. R., Thiessen, R., WHO Drug Information 1995, 9, 159.
Murer, H. K., Toxicity of coumarin. J. Pharmacol. Exp. Ther.
[18] Andrejak, M., Gersberg, M., Sgro, C., Decocq, G. et al.,
1956, 118, 348358.
French pharmacovigilance survey evaluating the hepatic
[3] Lake, B. G., Coumarin metabolism, toxicity and carcino- toxicity of coumarin. Pharmacoepidemiol. Drug. Saf. 1998,
genicity: relevance for human risk assessment. Food Chem. 7, S45S50.
Toxicol. 1999, 37, 423453.
[19] Mohler, J. L., Gomella, L. G., Crawford, E. D., Glode, L. M.
[4] EFSA, Opinion of the Scientific Panel on Food Additives, et al., Phase II evaluation of coumarin (1,2-benzopyrone)
Flavourings, Processing Aids and Materials in Contacts with in metastatic prostatic carcinoma. Prostate 1992, 20, 123131.
Food (AFC) on a request from the Commission related to
[20] Loprinzi, C. L., Kugler, J. W., Sloan, J. A., Rooke, T. W. et al.,
Coumarin; adopted on 6 October 2004, EFSA J. 2004, 104,
Lack of effect of coumarin in women with lymphedema
136. Includes in the Appendix also the previous opinions of
after treatment for breast cancer. N. Engl. J. Med. 1999, 340,
the Scientific Committee on Food (SCF) of 1994 and 1999.
346350.
[5] BfR, High daily intakes of cinnamon: Health risk cannot be
[21] Burian, M., Freudenstein, J., Tegtmeier, M., Naser-Hijazi, B.
ruled out. Health Assessment No. 044/2006, 2006.
et al., Single copy of variant CYP2A6 alleles does not
[6] Sproll, C., Ruge, W., Andlauer, C., Godelmann, R., Lachen- confer susceptibility to liver dysfunction in patients treated
meier, D. W., HPLC analysis and safety assessment of with coumarin. Int. J. Clin. Pharmacol. Ther. 2003, 41,
coumarin in foods. Food Chem. 2008, 109, 462469. 141147.
[7] EU Food Law, Cinnamon flavoured cereal recalled over [22] Schmeck-Lindenau, H. J., Naser-Hijazi, B., Becker, E. W.,
coumarin levels. EU Food Law 2006, November 10, 1011. Henneicke-von Zepelin, H. H., Schnitker, J., Safety aspects
[8] Felter, S. P., Vassallo, J. D., Carlton, B. D., Daston, G. P., of a coumarin-troxerutin combination regarding liver func-
A safety assessment of coumarin taking into account tion in a double-blind placebo-controlled study. Int. J. Clin.
species-specificity of toxicokinetics. Food Chem. Toxicol. Pharmacol. Ther. 2003, 41, 193199.
2006, 44, 462475. [23] Vanscheidt, W., Rabe, E., Naser-Hijazi, B., Ramelet, A. A.
[9] Edwards, A. J., Price, R. J., Renwick, A. B., Lake, B. G., Lack et al., The efficacy and safety of a coumarin-/troxerutin-
of effect of coumarin on unscheduled DNA synthesis in the combination (SB-LOT) in patients with chronic venous
in vivo rat hepatocyte DNA Repair Assay. Food Chem. insufficiency: a double blind placebo-controlled rando-
Toxicol. 2000, 38, 403409. mised study. Vasa, 2002, 31, 185190.
[10] Api, A. M., Lack of effect of coumarin on the formation of [24] Adam, B. S., Pentz, R., Siegers, C. P., Strubelt, O., Tegt-
micronuclei in an in vivo mouse micronucleus assay. Food meier, M., Troxerutin protects the isolated perfused rat liver
Chem. Toxicol. 2001, 39, 837841. from a possible lipid peroxidation by coumarin. Phytome-
[11] Swenberg, J. A., Covalent binding index study on dicine 2005, 12, 5261.
coumarin, Report of Laboratory of Molecular Carcinogen- .
[25] Bergmann, K., Sachverstandigengutachten zur Beurteilung
esis and Mutagenesis, University of North Carolina, Chapel von Cumarin in Arzneimitteln in Bezug auf lebertoxische

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com
Mol. Nutr. Food Res. 2010, 54, 228239 239

Wirkung beim Menschen, Rheinische Friedrich-Wilhelms- coumarin metabolites in rats and different human
. Bonn. 1999, Original report written in German
Universitat populations evaluated through physiologically based
available from the BfArM, Bonn. biokinetic (PBBK) modeling. Toxicol. In Vitro 2008, 22,
[26] Abraham, K., Zimt und Cumarin: eine Klarstellung aus 18901901.
wissenschaftlich-behordlicher Sicht. Dtsch. Lebensmittel- [41] Carlton, B. D., Aubrun, J.-C., Simon, G. S., Effects of
Rundschau 2007, 103, 480487. coumarin following perinatal and chronic exposure in
[27] EFSA, Scientific Opinion of the Panel on Food Additives, SpragueDawley rats and CD-1 rats. Fundam. Appl. Toxicol.
Flavourings, Processing Aids and Materials in Contact with 1996, 30, 145151.
Food on a request from the European Commission on .
[42] Sotaniemi, E. A., Rautio, A., Backstrom, M., Arvela, P.,
Coumarin in flavourings and other food ingredients with Pelkonen, O., CYP3A4 and CYP2A6 activities marked by the
flavouring properties. EFSA J. 2008, 793, 115. metabolism of lignocaine and coumarin in patients with
[28] Pelkonen, O., Rautio, A., Raunio, H., Pasanen, M., CYP2A6: a liver and kidney diseases and epileptic patients. Br. J. Clin.
human coumarin 7-hydroxylase. Toxicology. 2000, 144, Pharmacol. 1995, 39, 7176.
139147. [43] Pasanen, M., Rannala, Z., Tooming, A., Sotaniemi, E. A.
[29] Zhang, Q. Y., Gu, J., Su, T., Cui, H. et al., Generation and et al., Hepatitis A impairs the function of human hepatic
characterization of a transgenic mouse model with hepatic CYP2A6 in vivo. Toxicology 1997, 123, 177184.
expression of human CYP2A6. Biochem. Biophys. Res. [44] Uetrecht, J., Idiosyncratic Drug Reactions: Past, Present,
Commun. 2005, 338, 318324. and Future. Chem. Res. Toxicol. 2008, 21, 8492.
[30] Ritschel, W. A., Brady, M. E., Tan, H. S., First-pass effect of [45] Rychlik, M., Quantification of free coumarin and its libera-
coumarin in man. Int. J. Clin. Pharmacol. Biopharm. 1979, tion from glucosylated precursors by stable isotope dilution
17, 99103. assays based on liquid chromatography-tandem mass
[31] Rautio, A., Kraul, H., Kojo, A., Salmela, E., Pelkonen, O., spectrometric detection. J. Agric. Food. Chem. 2008, 56,
Interindividual variability of coumarin 7-hydroxylation in 796801.
healthy volunteers. Pharmacogenetics 1992, 2, 227233. [46] WHO (World Health Organization) 1999, Cortex Cinnamomi.
[32] Sharifi, S., Pharmakokinetische Untersuchungen von WHO monographs on selected medicinal plants, Volume 1,
Cumarin und seinen Metaboliten nach oraler und intrave- Geneva, 1999, pp. 96104.
noser Applikation bei gesunden Probanden. PhD thesis, [47] Miller, K. G., Poole, C. F., Chichila, T. M. P., Solvent-
University of Gottingen 1990. assisted supercritical fluid extraction for the isolation of
[33] Sharifi, S., Lotterer, E., Michaelis, H. C., Bircher, J., Phar- semivolatile flavor compounds from cinnamons of
macokinetics of coumarin and its metabolites. Preliminary commerce and their separation by series-coupled column
results in three volunteers. J. Ir. Coll. Physicians Surg. 1993, gas chromatography. J. High. Resol. Chromatogr. 1995, 18,
22, 2932. 461471.

[34] Meineke, I., Desel, H., Kahl, R., Kahl, G. F., Gundert-Remy, [48] Weber, G., ZimtWarenkunde und Markt. Dtsch. Lebens-
U., Determination of 2-hydroxyphenylacetic acid (2HPAA) in mittel-Rundschau 2007, 103, 492493.
urine after oral and parenteral administration of coumarin [49] Banasiak, U., Heseker, H., Sieke, C., Sommerfeld, C.,
by gas-liquid chromatography with flame-ionization detec- .
Vohmann, C., Abschatzung der Aufnahme von Pflan-
tion. J. Pharm. Biomed. Anal. 1998, 17, 487492. .
zenschutzmittel-Ruckst .
anden in der Nahrung mit neuen
[35] Shilling, W. H., Crampton, R. F., Longland, R. C., Metabo- Verzehrsmengen .
fur Kinder. Bundesgesundheitsbl
lism of coumarin in man. Nature 1969, 221, 664665. Gesundheitsforsch Gesundheitsschutz 2005, 1, 8498.

[36] Hadidi, H., Zahlsen, K., Idlem, J. R., Cholerton, S. A., Single [50] Floch, F., Mauger, F., Desmus, J. R., Gard, A. et al.,
amino acid substitution (Leu160His) in cytochrome P450 Coumarin in plants and fruits: implication in perfumery.
CYP2A6 causes switching from 7-hydroxylation to 3-hydro- Perfumer Flavorist 2002, 27, 332336.
xylation of coumarin. Food Chem. Toxicol. 1997, 35, 903907. [51] Yourick, J. J., Bronaugh, R. L., Percutaneous absorption and
[37] Mizutani, T., PM frequencies of major CYPs in Asians and metabolism of Coumarin in human and rat skin. J. Appl.
Caucasians. Drug Metab. Rev. 2003, 35, 99106. Toxicol. 1997, 17, 153158.

[38] Rautio, A., Polymorphic CYP2A6 and its clinical and toxi- [52] SKLM (Senatskommission zur Beurteilung der gesundhei-
cological significance. Pharmacogenomics J. 2003, 3, 57. tlichen Unbedenklichkeit von Lebensmitteln), Natural
Constituents of Foods: Assessing the toxicity of substances
[39] Peamkrasatam, S., Sriwatanakul, K., Kiyotani, K., Fujieda,
administered as pure compounds compared to their
M. et al., In vivo evaluation of coumarin and nicotine as
ingestion as inherent food components. Statement of the
probe drugs to predict the metabolic capacity of CYP2A6
DFG-Senate commission on Food Safety. 2006.
due to genetic polymorphism in Thais. Drug Metab. Phar-
macokinet. 2006, 21, 475484. [53] Renwick, A. G., Data-derived safety factors for the evalua-
tion of food additives and environmental contaminants.
[40] Rietjens, I. M., Boersma, M. G., Zaleska, M., Punt, A.,
Food Addit. Contam. 1993, 10, 275305.
Differences in simulated liver concentrations of toxic

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com