D.W. Lachenmeier, M. Uebelacker / Regulatory Toxicology and Pharmacology 58 (2010) 437–443
female, respectively) at doses of 0, 12.5, 15.0 and 50.0 mg/kg/day
for 13 weeks. Doses were given in ﬁve daily increments as a suspension in aqueous agar. Five rats (four males and one female) died
during acclimatisation and three others (one male from each of the
low and middle dose groups; one female control) died from a viral
infection during treatment. Post-treatment convulsions were frequently observed. No effects were observed on body weight gain,
or haematology, and histopathological examination at termination
did not reveal any dose-related lesions. The NOEL for males was
12.5 mg/kg/day; a NOEL could not be established for female rats
since one rat in the lowest dose group displayed convulsions on
two occasions. The results from Margaria (1963) and Surber
(1962) are summarised in Table 1.
Recently, the US National Toxicology Program (NTP) conducted
several animal experiments with thujone (a mixture of a- and bthujone), including short-term toxicity studies (2 weeks and
13 weeks) in rats and mice (10 animals/sex/species), as well as a
long-term carcinogenicity study (50 animals/sex/species). In this
evaluation, we considered only the long-term study, which is most
signiﬁcant for regulatory toxicology (NTP, 2009). Modelling of the
short-term studies showed BMD10 values of the same order of
magnitude as the long-term study (data not shown). The NTP provided results for several endpoints (including incidence of neoplastic and non-neoplastic lesions, as well as clinical observations such
as excitability, clonic and tonic seizures, eye abnormality, diarrhoea, head tilt, lethargic, different masses, nasal/eye discharge,
rufﬂed fur, thinness, ulcers/abscesses of different areas). No significant dose–response relationship was detected for any endpoint
besides mortality and the clonic and tonic seizures during clinical
observation (see summary in Tables 2 and 3).
The results of our BMD-modelling for seizures and mortality are
shown in Table 4. Only the result of the best-ﬁtting model (selected
according to p-value and Akaike’s information criterion) are presented. For almost all studies and selected endpoints, a signiﬁcant
dose–response was proven; most cases showed an excellent ﬁt
with p-values above 0.9. In Figs. 1 and 2, the modelling of tonic seizures in the NTP rat study is shown as an example. Some of the
data were problematic to model, as only one dose group (the highest) showed a response; nevertheless, we decided to show these
results in brackets for comparative purposes. The only modelled
endpoint with non-signiﬁcant dose–response was the tonic seizures in female rats from the NTP study. In that case, the incidence
was lower (4/50) in the highest dose group than in the dose group
Overall, the BMDL10 values fell within a range between 7 and
26 mg/kg bw/day. In some endpoints (convulsions), females ap-
peared to be slightly more sensitive than males, but under consideration of the BMD-modelling uncertainties, no clear difference
between the sexes was obvious.
As indicated anecdotally from intoxication cases, seizures are
probably the major adverse effect in humans (e.g., see Weisbord
et al., 1997). For this reason, we decided to use the lowest BMDL10
value from the long-term NTP studies, at 11 mg/kg bw/day from
clonic seizures in male rats, as a departure point for our regulatory
considerations. Although dose–response modelling of the mortality data shows results in the same order of magnitude as those
for the seizures, we did not consider these further in our evaluation
because of the high mortality, even at the control and low doses.
This may have been caused by viral infection, as in the study of
Surber (1962); however, no further explanations were currently
found on the NTP website (NTP, 2009).
To calculate an ADI, the traditional uncertainty factor (UF) of the
Joint FAO/WHO Expert Committee on Food Additives (JECFA) was
chosen (100). This assumes that the human being is 10 times more
sensitive than the test animal and that the difference in sensitivity
within the human population lies within a 10-fold range (IPCS,
1987). According to IPCS (2009), the ADI based on the composite
UF of 100 has been accepted by international institutions and countries as a health-based guidance value. The Committee of Experts on
Flavouring Substances (2005) used an increased UF of 500 because of
the poor data quality at the time, but we feel that the quality of the
new NTP data justiﬁes the use of the standard UF of 100.
With a BMDL10 of 11 mg/kg bw/day, the ADI would therefore
be 0.11 mg/kg bw/day, corresponding to an ADI of 6.6 mg/day for
a 60-kg human. To reach this ADI, 189 ml of an absinthe containing
the maximum-allowed amount of 35 mg/l of thujone would have
to be consumed. Given that the average thujone content in absinthe manufactured before the ban in 1915 was 25 mg/l (Lachenmeier et al., 2008), a volume of 264 ml would have to be
consumed to exceed the ADI.
Population-based intake estimates for thujone in food and beverages were provided by the Scientiﬁc Committee on Food (SCF,
2003) for France and the UK. The major dietary contribution appeared to derive from sage and sage-ﬂavoured products, as well
as alcoholic beverages. In France, the mean and 97.5th percentile
daily intakes were estimated to be 15.6 and 44.3 lg/kg bw/day,
respectively. The intakes in the UK were estimated at 3.9 and
14.2 lg/kg bw/day, respectively. The Committee of Experts on Flavouring Substances (2005) conﬁrmed these intakes for the UK, and
commented that the most important single source of intake is
sweets spiced with sage. Further contributors to total thujone intake are sage-ﬂavoured sausages and other meat products, sage
Studies on short-term thujone administration to rats.
Administration to rats by gavage on 6 days per week for
14 weeks (Margaria, 1963)a
Administration to weanling rats by gavage in ﬁve increments
daily for 13 weeks (Surber, 1962)b
Thujone dose (mg/kg
Thujone dose (mg/kg
The Margaria (1963) study is unpublished. The data was taken from a summarisation in WHO Food Additives Series 16 (WHO, 1981).
The Surber (1962) study was not available in full text. The data was taken from a summarisation in WHO Food Additives Series 16 (WHO,
1981), which without speciﬁc citation apparently includes the raw data table from Surber (1962), this was veriﬁed in the thujone monograph of
the Committee of Experts on Flavouring Substances (2005).
Data for month 3 are shown. A no-effect level cannot be established for female rats since one rat in the lowest dose group displayed
convulsions on two occasions at month 2 (WHO, 1981).