TOXICOLOGICAL SCIENCES 75, 715 (2003)

DOI: 10.1093/toxsci/kfg165

FORUM
Risk Assessment of Acrylamide in Foods
E. Dybing* ,1 and T. Sanner

Received April 30, 2003; accepted June 1, 2003

and a Category 2 carcinogen and Category 2 mutagen by the

European Union (http://ecb.jrc.it/classication-labelling/), this
nding has caused worldwide concern (WHO, 2002; European
Commission, 2002). Earlier studies (Tareke et al., 2000) had
documented that rats fed fried animal standard diet had generated an adduct of N-(2-carbamoylethyl)valine to the N-termini of hemoglobin, similar to the adduct formed from occupational exposures to acrylamide (Bergmark et al., 1993).
Recently, it was documented that the acrylamide generated
from food components during heat treatment was a result of the
Maillard reaction between primarily the amino acid asparagine
and reducing sugars (Mottram et al., 2002; Stadler et al.,
2002).
The Norwegian Food Agency has performed a limited estimate of the intake of acrylamide from food in the Norwegian
population and used this as inputs for its Scientic Committee
to perform a risk assessment of acrylamide from foods, with
special emphasis on cancer risk (Norwegian Food Agency,
2002). In the present communication, this risk assessment will
be described and discussed.

Daily mean intakes of acrylamide present in foods and coffee in

a limited Norwegian exposure assessment study have been estimated to be 0.49 and 0.46 g per kg body weight in males and
females, respectively. Testicular mesotheliomas and mammary
gland adenomas have consistently been found in 2-year drinking
water rat cancer studies with acrylamide. Acrylamide also shows
initiating activity in mouse skin after systemic administration.
Since acrylamide is converted to the mutagenic metabolite glycidamide and forms adducts to hemoglobin in rodents and humans,
the tumorigenic endpoints in rats were assumed to be an expression of acrylamide genotoxicity. Using the default linear extrapolation methods LED10 and T25, the lifetime cancer hazard after
lifelong exposure to 1 g acrylamide per kg body weight per day,
scaled to humans, was calculated to be, on average, 1.3 10 3.
Using this hazard level and correlating it with the exposure estimates, a lifetime cancer risk related to daily intake of acrylamide
in foods for 70 years in males was calculated to be 0.6 10 3,
implying that 6 out of 10,000 individuals may develop cancer due
to acrylamide. For females, the risk values were slightly lower. It
must be emphasized that this risk assessment is conservative. A
number of processes may result in nonlinearity of the dose-response relationships for acrylamide carcinogenicity in the lowdose region, including detoxication reactions, cell cycle arrest,
DNA repair, apoptosis, and immune surveillance. Thus, the true
risk levels related to acrylamide intake may be considerably lower.
Key Words: acrylamide; food exposure; risk assessment; cancer;
LED10, T25.

Exposure Assessment
The Norwegian Food Agency (2002) has conducted analysis
of 30 different food products plus eight brands of coffee (seven
lter coffee, one instant coffee) bought on the Norwegian
market. The sampling was chiey based on results from Sweden, such that priority was given to food items (potato crisps,
French fries) that were shown to have a high content of
acrylamide in the Swedish analysis. The analysis was performed on pooled samples of three identical products from the
same producer, but with different production dates. The analyses were carried out by the same accredited laboratory
(AnalyCen, Lidkoping, Sweden) that had conducted the Swedish analyses using LC-MS-MS methodology.
Data on intake in the adult population has been taken from
the national food survey NORKOST 1997 (Johansson et al.,
1997) based on a quantitative frequency questionnaire answered by 1291 males and 1381 females aged 16 79 years.

On April 24, 2002, the Swedish National Food Agency

presented data that, in part, showed high concentrations of
acrylamide in certain fried, baked, and deep-fried foods, and
later in coffee (Swedish National Food Agency, 2002). Since
acrylamide has been classied as a Group 2A carcinogen by
the International Agency for Research on Cancer (IARC, 1994)
1

To whom correspondence should be addressed. Fax: 00 47 22 04 26 86.

E-mail: erik.dybing@fhi.no
Toxicological Sciences 75(1), Society of Toxicology 2003; all rights reserved.
7

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013

*Division of Environmental Medicine, Norwegian Institute of Public Health, P.O. Box 4404 Nydalen, NO-0403 Oslo, Norway,
and Institute of Cancer Research, Norwegian Radium Hospital, Montebello, NO-0310 Oslo, Norway

FORUM

TABLE 1
Estimated Intake of Acrylamide

Parameter

Males
g/day

Females
g/day

Males
g/kg bw/day

Females
g/kg bw/day

Mean
Median
90-percentile
97.5-percentile

38
29
77
125

29
24
55
88

0.49
0.41
1.01
1.62

0.46
0.42
0.86
1.45

Hazard Identication
Acrylamide is oxidized to glycidamide, a reactive epoxide, and undergoes conjugation with glutathione (Calleman
et al., 1990; Sumner et al., 1992). DNA adducts from
glycidamide have been reported following administration of
acrylamide to rodents (Segerback et al., 1995). Hemoglobin
adducts from direct reaction of acrylamide and from reaction with glycidamide have been detected in rodents administered acrylamide, in exposed humans, and in cigarette
smokers (Bailey et al., 1986; Bergmark, 1997; Bergmark et
al., 1991, 1993; Calleman et al., 1990, 1994; Fennel et al.,
2003). The relationship between the acrylamide and glyci-

TABLE 2
Contribution of Various Food Items to Estimated
Intake of Acrylamide

Food item

Males
Mean intake 38 g/day %

Females
Mean intake 29 g/day %

Soft bread
Bread, other
Biscuits
Potatoes, fried
French fries
Potato crisps
Other snacks
Coffee
Other

13.0
7.7
5.0
7.5
8.0
17.6
5.0
28.0
8.2

11.9
12.2
5.9
6.3
6.3
17.4
4.6
28.6
6.8

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013

The mean daily intake of acrylamide was estimated to 38 g

per day for males and 29 g for females, corresponding to
daily doses of 0.49 g/kg body weight in males and 0.46 g/kg
body weight in females (Table 1). The 97.5-percentile showed
intakes that were approximately three times higher than the
mean intakes.
In Table 2 the contribution of the various food items to the
estimated intake of acrylamide is presented. Coffee (28%)
contributed the most to the to the total mean intake of acrylamide, whereas potato crisps were responsible for almost 20%
of the intake. From the analytical results, bread (potato cakes
and dark crisp-bread excepted) contains low amounts of acrylamide. But since bread is consumed daily in relatively large
amounts in Norway, this food group contributed to a considerable proportion of the acrylamide intake (2124% of mean
intake).
Food consumption data for children and adolescents were
from UNGKOST 2000 (unpublished), a 4-day food intake
registration in which portions were assigned according to an
illustrative book with different portion sizes. The estimated
intake of acrylamide in groups of children aged 9 and 13 years
is shown in Table 3.
The largest source for the intake of acrylamide both in 9- and
13-year-old children is potato crisps, but butter biscuits and
sweet biscuits are also important sources. These products contribute to approximately 55 65% of the total mean intake. It is
notable that the 97.5-percentile of 13-year-old boys and girls
have intakes that are about 4- to 5-fold higher than the mean
intakes.

damide adducts on the N-terminal valine of hemoglobin is

proportional in man and rat. Comparison of free acrylamide
in plasma, valine adducts on haemoglobin, and urinary
S-(2-carboxyethyl)cysteine indicates that the rate of elimination of acrylamide is at least ve times lower in man than
in rats (Calleman, 1996).
The metabolite glycidamide is mutagenic (Adler et al., 2000;
Barfknecht et al., 1988; Butterworth et al., 1992; Generoso et
al., 1996; Hashimoto and Tanii, 1985). Acrylamide is neurotoxic in animals and humans (Miller and Spencer, 1985; Spencer and Schaumberg, 1974a,b). Acrylamide has been shown to
be carcinogenic in F344 male and female rats after administration in drinking water for up to 2 years (American Cyanamid
Co., 1989; Friedman et al., 1995; Johnson et al., 1984, 1986)
and in skin painting and oral administration experiments with
mice (Bull et al., 1984a,b; Robinson et al., 1986).
Two long-term studies of possible carcinogenic effects of
acrylamide have been conducted in rats, in that F344 male and
female rats were given acrylamide in drinking water for up to
2 years. In the rst study (Johnson et al, 1984, 1986) the
animals received doses of acrylamide (98% purity) corresponding to 0, 0.01, 0.5, and 2.0 mg per kg body weight per
day (mg/kg bw/day) (60 rats of each sex at all doses). Maximum tolerated dose (MTD) appeared to have been reached
based on reduced body weight increase and toxic effects in the
high-dose group. There were temporary symptoms of a viral
infection (sialodacryoadenitis virus) in some rats from day 210
of the study. All groups appeared to be affected to a similar
degree.
In the second study, groups of 75204 F344 male rats
received doses of acrylamide (99.9% purity), corresponding to
0, 0.1, 0.5, and 2.0 mg/kg bw/day, and groups of 50 100 F344
female rats received doses corresponding to 0, 1.0, and 3.0
mg/kg bw/day (American Cyanamid Co., 1989; Friedman et
al., 1995). Acrylamide induced statistically signicant increases in the incidence of several tumor types in the experimental animals of both sexes, compared to the control animals
(Table 4 [males] and Table 5 [females]).

FORUM

TABLE 3
Estimated Intake of Acrylamide in Children Aged 9 and 13 Years

Parameter

Boys 9 years
(n 1299)
g/kg/day

Girls 9 years
(n 1658)
g/kg/day

Boys 13 years
(n 1711)
g/kg/day

Girls 13 years
(n 2068)
g/kg/day

Mean
Median
90-percentile
97.5-percentile

0.36
0.23
0.72
1.50

0.32
0.25
0.61
1.10

0.52
0.30
1.35
2.85

0.49
0.28
1.20
2.07

nervous system of glial origin or in glial proliferation, an

indication of initial tumor development. Astrocytomas were
found in the brain and spinal cord, both in male and female
rats. However, there were no clear dose-response relationships.
In a series of skin painting experiments with mice, it was
found that acrylamide initiates the formation of skin tumors,
both in SENCAR and Swiss-ICR mice, as well as lung
adenomas in the strains SENCAR, Swiss-ICR and A/J (Bull
et al., 1984a,b; Robinson et al., 1986). In an experiment
with Swiss-ICR mice, the animals were given 300 mg/kg
body weight acrylamide in drinking water divided in 6 doses
over a 2-week period. Thereafter, the animals were painted
on the skin with 0.2 ml acetone three times per week for 20
weeks. The animals were sacriced after 52 weeks. The
incidence of skin tumors was 10/40 (25%) (controls 0/40;
p 0.001) and of lung tumors 14/36 (39%) (controls 4/36;
p 0.06) (Bull et al., 1984b).
Sobel and coworkers (1986) have investigated mortality
among 371 employees in a factory in Michigan that produced
acrylamide monomer and who also worked in polymerization.
In the three plants investigated, the production of acrylamide

TABLE 4
Male Rats: Tumor Incidence after Exposure of 344 Male Rats to Various Doses of Acrylamide in the Drinking Water for 2 Years
Tumor type
Thyroid adenomas
Trend test p 0.001**
Trend test p 0.001
Testicular mesotheliomas
Trend test p 0.01
Trend test p 0.001
Adrenal pheochromocytomas
Trend test p 0.06
Oral focal hyperplasia
Trend test p 0.01

Control

0.01

0.1

0.5

2.0

1/60
(2%)
3/204
(1%)
3/60
(5%)
8/204
(4%)
3/60
(5%)
0/60
(0%)

0/58
(0%)

0/60
(0%)

7/59
(12%)
1/60
(2%)

2/59
(3%)
9/203
(4%)
7/60
(12%)
9/204
(4%)
7/60
(12%)
1/60
(2%)

1/59
(2%)
5/101
(5%)
11/60
(18%)*
8/102
(8%)
5/60
(8%)
4/60
(7%)

7/59
(12%)
12/75
(16)*
10/60
(17%)*
13/75
(17%)*
10/60
(17%)*
5/60
(8%)

Note. Acrylamide doses in mg/kg bw/day. The upper lines represent the study by Johnson et al. (1984, 1986), and the lower lines represent the study by
American Cyanamid Co. (1989) and Friedman et al. (1995).
*Signicant increase in tumor incidence (p 0.05), Fischers exact test; **Mantel-Haenszel trend test.

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013

Among the male rats, signicant increased incidences of

thyroid adenomas and testicular mesotheliomas were detected in both investigations. Adrenal pheochromocytomas
were signicantly increased in the high-dose group of the
rst study, but not in the second study. In addition, focal
hyperplasia in the oral cavity was seen in the rst study. An
increased incidence of astrocytomas was also detected
among the high-dose rats.
Among the female rats, there was a signicantly increased
incidence of thyroid adenomas in the second study. In the
rst study, the incidence of thyroid adenomas and adenocarcinomas combined was signicantly increased in the
high-dose group. A signicant positive trend was found in
both studies with respect to increases of thyroid adenomas
and adenocarcinomas. The incidence of broadenomas in
the mammary gland was signicantly increased in the rats of
both studies. In the rst study, an increased incidence of oral
papillomas was found among the rats receiving the high
dose. A positive trend was found for oral cavity tumors and
uterine adenocarcinomas in the rst study. Such tumors
were not found in the second study.
There was a signicant increase in tumors of the central

10

FORUM

TABLE 5
Female Rats: Tumor Incidence after Exposure of 344 Male Rats to Various Doses of Acrylamide in the Drinking Water for 2 Years
Tumor type
Thyroid adenomas
Trend test p 0.01**
Trend test p 0.001
Thyroid adenocarcinomas
Trend test p 0.01

Trend test p 0.001

Mammary gland adenocarcinomas
Trend test p 0.01
Trend test p 0.17
Uterine adenocarcinomas
Trend test p 0.01
Oral cavity papillomas
Trend test p 0.01

0.01

0.1

0.5

2.0

0/58
(0%)
0/100
(0%)
1/58
(2%)
2/100
(2%)
10/60
(17%)
9/96
(9%)
2/60
(3%)
2/96
(2%)
1/60
(2%)
0/60
(0%)

0/59
(0%)

0/59
(0%)

11/60
(18%)

1/60
(2%)

2/60
(3%)
3/60
(5%)

1/59
(2%)

0/59
(0%)

9/60
(15%)

1/60
(2%)

1/60
(2%)
2/60
(3%)

1/58
(2%)
7/100
(7%)*
0/58
(0%)
3/100
(3%)
19/58
(33%)*
20/94
(21%)*
2/58
(3%)
2/94
(2%)
0/59
(0%)
1/60
(2%)

3/60
(5%)
16/100
(16%)*
3/60
(5%)
7/100
(7%)
23/61
(38%)*
26/95
(27%)*
6/61
(10%)
4/95
(4%)
5/60
(8%)
7/61
(11%)*

Note. Acrylamide doses in mg/kg bw/day. The upper lines represent the study by Johnson et al. (1984, 1986) and the lower lines represent the study by
American Cyanamid Co. (1989) and Friedman et al. (1995).
*Signicant increase in tumor incidence (p 0.05), Fischers exact test; **Mantel-Haenszel trend test.

had occurred from the end of the 1950s and the beginning of
the 1960s. A total of 29 deaths were observed up to 1982 (38.0
expected). There were 11 cancer deaths, compared to 7.9
expected. The increased number of cancer cases was due to
cancer of the digestive tract and the respiratory organ in a
subgroup that had previous exposure to organic dyes. Among
the employees who had no exposure to organic dyes, four
deaths were observed compared to 6.5 expected.
Collins and coworkers (1989) investigated causes of death in
four factories, three in the United States and one in the Netherlands, in order to assess possible cancer risks after exposure
to acrylamide. The investigation included 8,854 males employed in the factories between 1925 and 1976, and exposure
to acrylamide was dened as cumulative exposure 0.001
mg/m 3-years. A total of 2,293 males were identied as being
exposed. Information on smoking habits was available from
approximately 1/3 of the group. Among the exposed workers,
a signicantly reduced mortality from all causes was found
(Standardised Mortality Ratio [SMR] 81), but a weak indication of increased cancer incidence was revealed for pancreatic cancer (eight cases; SMR 203) (95% CI 87 400) as
well as for Hodgkins disease (ve cases; SMR 129 (95%
CI 42300). No increased trend was found in cancer deaths
when higher-exposed workers were compared to lower-exposed workers.
Marsh and coworkers (1999) have presented an update for
the period 1984 1994 of the same cohort of Collins and
coworkers (1989). For the period 19251994 the following

results were found: tumors of the brain and central nervous

system, SMR 65 (95% CI 36 109); thyroid tumors,
SMR 211 (95% CI 44 612); tumors in the testes and
other sexual organs, SMR28 (95% CI1159); and respiratory cancers, SMR 110 (95% CI 99 122). As presented in the article, mean occupational duration was 1.7
years, and mean acrylamide exposure was 0.1 mg/m 3 . The
authors conclude that there is little evidence for a causal
relationship between exposure to acrylamide and cancer
mortality.
Recently, Mucci et al. (2003) reanalyzed a population-based
Swedish case-control study encompassing cases with cancer of
the large bowel (n 591), bladder (n 263) and kidney (n
133) and 538 healthy controls. They assessed dietary acrylamide by linking extensive food frequency data with the acrylamide levels reported by the Swedish National Food Agency
(2002). There was a consistent lack of an excess risk, or any
convincing trend of cancer of the bowel, bladder, or kidney in
high consumers of 14 different food items with a high or
moderate acrylamide content.
Hazard Characterization
The no-observed-adverse-effect level (NOAEL) of acrylamide neurotoxicity for the most sensitive effect (microscopic
nerve changes) in rats was 0.5 mg/kg body weight/day (Spencer and Schaumberg, 1974a). In reproductive toxicity studies
with rats, reduction in the number of sperm and decreased

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013

Trend test p 0.03

Mammary gland broadenomas
Trend test p 0.001

Control

11

FORUM

TABLE 6
Comparison of LED10 andT25 Values Estimated from the Two Acrylamide Drinking Water Studies

Experiment

LED10
Males
(testicular mesotheliomas)

Females
(mammary gland adenomas)

T25
Males
(testicular mesotheliomas)

Females
(mammary gland adenomas)

Johnson et al., 1986

Friedman et al., 1995

0.66
0.84

0.40
0.86

0.89
3.7

0.64
1.9

Note. All estimates are in mg/kg bw/day.

pared to testicular mesotheliomas. Therefore, mammary

gland adenomas in the Johnson et al. (1986) study were
chosen as the most sensitive outcome for the further calculations. Assuming that the tumor responses caused by acrylamide reect a genotoxic mechanism, the LED10 and T25
values were converted to values corresponding to the lowhazard level of 10 5 (World Health Organization, 1996) by
default linear extrapolation by dividing with 10,000 and
25,000, respectively. These gures were then scaled to the
human situation using the scaling factor W 0.75 (weight ex0.75
perimental animals /weight humans )
(70/0.25) 0.25 4.1 for rat to
human conversion (U.S. EPA, 1996). The outcomes of these
calculations are presented in Table 7.
The two calculations resulted in a lifetime hazard level that
was 1.6 times higher with the T25 compared to the LED10
method. For the risk characterization step, a mean value of
1.3 10 3 was used to represent the hazard corresponding to
a lifelong exposure to 1 g acrylamide per kg body weight per
day.
Risk Characterization
When the intake doses presented in Table 1 are correlated to
the hazard estimates given in Table 7 [risk (lifetime hazard
after lifelong exposure to 1 g acrylamide per kg body weight
per day) dose], a lifetime cancer risk related to daily intake
of acrylamide in foods for 70 years becomes, on average, 0.6
10 3 (corresponding to 6 cancer cases per 10,000 individuals)
(Table 8). For the 10% and 2.5% males with the highest intake,
lifetime cancer risks were estimated to 13 and 21 cancer cases

TABLE 7
Lifetime Cancer Hazards after Exposure to Acrylamide in Humans

Model
LED10 0.40 mg/kg bw/d
LED10 5 0.040 g/kg bw/d
HLED10 5 is estimated by dividing by 4.1
T25 0.64 mg/kg bw/d
T10 5 0.0256 g/kg bw/d
HT10 5 is estimated by dividing by 4.1

Lifetime exposure dose which

represents a cancer hazard of 10 5

Lifetime hazard after lifelong exposure to 1 g

acrylamide per kg body weight per day

0.0098 g/kgbw/day

1.0 10 3

0.0062 g/kgbw/day

1.6 10 3

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013

fertility has resulted, plus an increase in premature embryonic

deaths (Tyl et al., 2000). The NOAEL for reduced fertility was
5 mg/kg body weight/day and for premature embryonic death
2 mg/kg body weight/day.
For genotoxic carcinogens for which the most plausible
mode of carcinogenic action is a consequence of genotoxic
events, the prudent position has been to assume nonthreshold
dose-response relationships. The basic default is to assume
linearity if that it is suggested by the mode of action analysis
or if the mode of action is not understood (U.S. EPA, 1996,
1999; EU Technical Guidance Document, 2003). The U.S.
Food and Drug Administration has also advocated the use of a
linear-at-low-dose, no threshold model (Lorentzen, 1984).
The U.S. Environmental Protection Agency (1996) has proposed to use the parameter LED10 as the point of departure for
quantitative hazard characterisation of nonthreshold carcinogens. The LED10 is the lower 95% condence limit for the
dose giving the animals an increased tumor incidence of 10%,
calculated with the multistage model. In the European Union,
the tumor dose indicator T25 (Dybing et al., 1997; Sanner et
al., 2001) has also been used as a basis for quantitative hazard
characterization of carcinogens. The T25 is the dose that increases the tumor incidence by 25% under standard conditions.
In Table 6 are given LED10 and T25 values for the two most
sensitive tumor endpoints in the Johnson et al. (1986) and
Friedman et al. (1995) studies, namely testicular mesotheliomas and mammary gland adenomas, respectively.
The differences in the estimated tumorigenic dose indicators are small. Mostly, the LED10 and the T25 values
were somewhat lower for mammary gland adenomas com-

12

FORUM

TABLE 8
Lifetime Cancer Risks Related to Acrylamide in Food
Calculation basis

Lifetime risk

0.49
0.41
1.01
1.62

0.610 3
0.510 3
1.310 3
2.110 3

0.46
0.42
0.86
1.45

0.610 3
0.510 3
1.110- 3
1.910 3

per 10,000 individuals, respectively. For females, the lifetime

cancer risks were somewhat lower than for males.
Assessment of potential risks for neurotoxic and reproductive toxic effects assumes that such effects have a dose threshold, where the NOAEL is used as a surrogate for the dose
threshold. Since the NOAEL for the most sensitive effect
(neurotoxicity) was 0.5 mg/kg bw/day (or 500 g/kg bw/day)
and the exposure dose for males 16 30 year olds was 0.49
g/kg bw/day (Table 1), the margin of safety for these individuals becomes approximately 1,000 (500/0.49). For the 97.5percentile for 13-year-old boys with an estimated intake of
2.85 g/kg bw/day, the margin of safety becomes 175.
Uncertainties in the Risk Assessment
There is considerable uncertainty related to the analytic
results of acrylamide in the reported food products; this is
especially the case for the lowest acrylamide levels, as in
bread. The analyses of the Norwegian foods included in the
calculation are restricted to 30 different products, and an only
very limited pool of three samples. However, the results give
clear indications and present a basis to assume that the mean
values used in the calculation are of a correct order of magnitude. Both the Swedish and the Norwegian analytical data
showed that there are large differences between the different
brands of the same food item (e.g., potato crisps).
A basic premise for the present risk assessment is that the
tumor responses are a reection of a genotoxic mechanism of
action. Since acrylamide produced tumors in endocrine-sensitive organs (testis and mammary gland), this could point to an
endocrine, nongenotoxic mechanism. On the other hand, the
acrylamide metabolite glycidamide is clearly mutagenic (Adler
et al., 2000; Barfknecht et al., 1988; Butterworth et al., 1992;
Generoso et al., 1996; Hashimoto and Tanii, 1985) and acrylamide showed carcinogenic initiating activity in mouse skin
after systemic administration (Bull et al., 1984b). Given that
there is no strong evidence to support an endocrine mechanism
of action in the tumor responses, it seems prudent to assume
that the responses are related to the genotoxic actions of
acrylamide. This was also the reason for IARC (1994) to

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013

Males
mean
median
90-percentile
97.5-percentile
Females
mean
median
90-percentile
97.5-percentile

Daily dose (g/kg bw/day)

upgrade the classication of acrylamide from Group 2B to

Group 2A.
The present risk assessment has used a linear extrapolation
from the points of departure (LED10 and T25), to describe the
hazard at low levels of exposure. Such a default extrapolation
can also be questioned, since cellular protective mechanisms
may be operative at low doses, including detoxication processes, cell cycle arrest, DNA repair, apoptosis and the control
of neoplastically transformed cells by the immune system.
Thus, a linear extrapolation may overestimate the actual risk
for the exposure levels related to acrylamide in food. In support
of a linear extrapolation, Abramsson-Zetterberg (2003) has
demonstrated a linear dose-response relationship for acrylamide in inducing micronucleated erythrocytes in mice exposed to acrylamide in single doses of 1 to 30 mg/kg body
weight intraperitoneally.
There are several examples in the literature of genotoxic
carcinogens that show a linear dose-response relationship in
the low dose range, such as for liver tumors in rats induced by
aatoxin B 1 (Wogan et al., 1974) or N,N-diethylnitrosamine
(Peto et al., 1991), and liver tumors in mice induced by
2-acetylaminouorene (Littleeld et al., 1979). On the other
hand, in a review of carcinogenesis data for 315 chemicals
from the NCI-NTP bioassay programs, the tumor site data were
more often consistent with a quadratic response than with a
linear response, suggesting that the default use of linear extrapolations will often overestimate the risk (Hoel and Portier,
1994). In fact, genotoxic carcinogens were found to differ from
linearity more often than nongenotoxic compounds. On the
other hand, because of the small number of doses tested experimentally, i.e., usually only two or three, almost all data sets
t equally well various mathematical functions, and it is generally not possible to dene dose-response curves on the basis
of mathematical modeling.
It is further assumed that acrylamide shows 100% bioavailability from the gastrointestinal tract; this is supported by
animal experiments (Miller et al., 1982). However, it is possible that there are matrix effects when the reactive acrylamide
is generated from the various food products; this may reduce
the bioavailability and, thus, the risk estimates. On the other
hand, it may be envisioned that the present risk values are an
underestimation, since the tissue dose of glycidamide may be
higher in humans compared to rats (Calleman et al., 1996).
It may be questioned whether the epidemiological studies
should have identied an increased cancer risk in relation to
acrylamide exposure, should this risk be a reection of a
genotoxic response with a relative risk of 1.006. The study by
Marsh et al. (1999) is the only workplace study that can be
used for quantitative considerations. However, it can be calculated that, using the present risk estimates, the exposure levels
should have been about 5 to 10 times higher in the Marsh et al.
(1999) study, in order to observe a signicant increase in any
of the tumor types recorded. A possible reason for the difference in response between the test animals and the occupational

13

FORUM

Conclusion
In a limited exposure assessment of acrylamide in food
conducted in Norway, mean intakes were estimated to 38 and
29 g acrylamide per day in males and females, respectively,
corresponding to intake doses of 0.49 and 0.46 g per kg body
weight and day. The 97.5-percentiles of adults had intakes that
were approximately 3-fold higher than the mean intakes.
In 2-year drinking water cancer studies with acrylamide
(Friedman et al., 1995; Johnson et al., 1986), testicular mesotheliomas and mammary gland adenomas were consistently
found. Since acrylamide is converted to a mutagenic metabo-

lite (Adler et al., 2000; Barfknecht et al., 1988; Butterworth et

al., 1992; Generoso et al., 1996; Hashimoto and Tanii, 1985)
and shows initiating activity in mouse skin after systemic
administration (Bull et al., 1984b), it is assumed that the tumor
responses seen after acrylamide administration are a reection
of its genotoxic potential. Using the conservative default linear
extrapolation methods LED10 (U.S. EPA, 1996) and T25
(Dybing et al., 1997; Sanner et al., 2001) for genotoxic carcinogens, the lifetime cancer hazard after lifelong exposure to
1 g acrylamide per kg body weight per day scaled to humans
was on average calculated to be 1.3 10 3. Using this hazard
level and correlating it with the exposure estimates, a lifetime
cancer risk related to daily intake of acrylamide in foods for 70
years in males was calculated to 0.6 10 3, corresponding to
6 cancer cases per 10,000 individuals. For the 10% and 2.5%
males with the highest intakes, lifetime cancer risks were
estimated to 13 and 21 cancer cases per 10,000 individuals. For
females, the risk values were slightly lower.
It must be emphasized that the risk assessment presented
here is highly conservative, so that the true risk levels may be
lower. Firstly, the exposure estimates are based on a limited
number of measurements of a limited number of products and
samples. Also, the exposures are estimated from food intake
questionnaires, which may overestimate the exposure. Secondly, the assessment does not take into account the possibility
that bioavailability of acrylamide from the gastrointestinal tract
may be reduced due to food matrix effects. Thirdly, it is
assumed that the tumor responses seen in rats are an expression
of the genotoxic activity of acrylamide. Given that the critical
carcinogenic responses occur in endocrine-sensitive organs,
they could also reect a nongenotoxic mechanism. And nally,
the risk assessment presented here assumes that there is a direct
linear relationship between tumor responses seen in the rat
experiments and responses at the calculated exposure levels, an
assumption that may be questioned. Thus, the calculated risk
numbers are presumably very conservative but do give an
indication of the magnitude of a risk for cancer related to
acrylamide in foods.
Available epidemiological studies have not been sensitive
enough to detect any possible cancer risk based on the calculated risk levels from the experimental data, given that these
levels are a true reection of the actual risk.
Any risk of neurotoxic or reproductive toxic effects associated with acrylamide in foods is judged to be very small.
REFERENCES
Abramsson-Zetterberg, L. (2003). The dose-response relationship at very low
doses of acrylamide is linear in the ow cytometer-based mouse micronucleus assay. Mutat. Res. 535, 215222.
Adler, I. D., Baumgartner, A., Gonda, H., Friedman, M. A., and Skerhut, M.
(2000). 1-Aminobenzotriazole inhibits acrylamide-induced dominant lethal
effects in spermatids of male mice. Mutagenesis 15, 133136.
American Cyanamide Company (1989). Chemistry of Acrylamide. Process
Chemicals Department, American Cyanamide Company, Wayne, NJ.

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013

cohorts is that the animals were dosed at a much younger age.

In the study of Mucci et al. (2003) acrylamide intake through
the diet was estimated to approximately 0.4 g per kg body
weight and day for a person weighing 70 kg. Using the present
risk characterization it is unlikely that the difference in exposure between their groups with the lowest and highest exposure
levels would represent a lifetime cancer risk of more than
0.5 10 3. The Mucci et al. (2003) study is much too small to
signicantly detect such a low increase in risk, given that this
risk level is true.
Were the tumor responses revealed in the drinking water
studies with rats a reection of a nongenotoxic mechanism
with a threshold response, the NOAEL for testicular mesotheliomas and mammary gland broadenomas would be 0.1
mg/kg bw/day, although there was a non-signicant increase in
testicular mesotheliomas at this dose in one of the two experiments. Relating this NOAEL to the mean exposure of acrylamide from food, the corresponding margins of exposure
would be 204 and 217 in males and females, respectively.
The attempts to quantify cancer risks at low doses of acrylamide in food should be seen as an aid to the regulatory
agencies in their assignment of priorities and choice of actions.
Alternatively, instead of attempting to quantify the risk, the
regulators may apply the ALARA principle (As-Low-As-Reasonably-Achievable) for a genotoxic carcinogen, aiming at
reducing the acrylamide exposures as much as possible. This
approach does not run the chance of over-interpreting the
hazard characterization data and leaves it to the risk manager to
decide the level of reduction that may be technologically and
nancially practicable. A disadvantage of using the ALARA
principle is that it does not give a quantitative dimension on
which to prioritize the problem.
Even if there are considerable uncertainties in the exposure
assessment of acrylamide in food, the safety margins with
respect to neurotoxic effects of acrylamide are judged to be
large enough to conclude that the risks appear to be very low.
This also holds true for the 97.5-percentile of 13-year-old boys
who have the highest estimated acrylamide intake. Since the
NOAELs for reproductive toxicity are higher than that for
neurotoxicity, any risk for reproductive toxic effects related to
consumption of acrylamide in food should be extremely low.

14

FORUM

Bailey, E., Farmer, P. B., Bird, I., Lamb, J. H., and Peal, J. A. (1986).
Monitoring exposure to acrylamide by the determination of S-(2-carboxyethyl)cysteine in hydrolyzed hemoglobin by gas chromatography-mass
spectrometry. Anal. Biochem. 157, 241248.

Hoel, D. G., and Portier, C. J. (1994). Nonlinearity of dose-response functions

for carcinogenicity. Environ. Health Perspect. 102(Suppl. 1), 109 113.

Barfknecht, T. R., Mecca, D. J., and Naismith, R. W. (1988). The genotoxic

activity of acrylamide. Environ. Mol. Mutagen. 11(Suppl.), 11.

Johansson, L., Solvoll, K., Bjrneboe, G. E., and Drevon, C. A. (1997). Dietary
habits among Norwegian men and women. Scand. J. Nutr. 2, 6370.

Bergmark, E. (1997). Hemoglobin adducts of acrylamide and acrylonitrile in

laboratory workers, smokers, and nonsmokers. Chem. Res. Toxicol. 10,
78 84.

Johnson, K., Gorzinski, S., Bodner, K., and Campbell, R. (1984). Acrylamide:
A two-year drinking water chronic toxicity-oncogenicity study in Fischer
344 rats. Dow Chemical USA, Midland, MI.

Bergmark, E., Calleman, C. J., and Costa, L. G. (1991). Formation of hemoglobin adducts of acrylamide and its epoxide metabolite glycidamide in the
rat. Toxicol. Appl. Pharmacol. 111, 352363.

Johnson, K. A., Gorzinski, S. J., Bodner, K. M., Campbell, R., Wolf, C. H.,
Friedman, M. A., and Mast, R. W. (1986). Chronic toxicity and oncogenicity
study on acrylamide incorporated in the rinking water of Fischer 344 rats.
Toxicol. Appl. Pharmacol. 85, 154 168.

Bull, R. J., Robinson, M., Laurie, R. D., Stoner, G. D., Greisiger, E., Meier,
J. R., and Stober, J. (1984a). Carcinogenic effects of acrylamide in Sencar
and A/J mice. Cancer Res. 44, 107111.
Bull, R. J., Robinson, M., and Stober, J. A. (1984b). Carcinogenic activity of
acrylamide in the skin and lung of Swiss-ICR mice. Cancer Lett. 24,
209 212.
Butterworth, B. E., Eldrigde, S. R., Sprankle, C. S., Working, P. K., Bentley,
K. S., and Hurtt, M. E. (1992). Tissue-specic genotoxic effects of acrylamide and acrylonitrile. Environ. Mol. Mutagen. 20, 148 155.
Calleman, C. J. (1996). The metabolism and pharmacokinetics of acrylamide:
Implications for mechanisms of toxicity and human risk estimation. Drug
Metab. Rev. 28, 527590.
Calleman, C. J., Bergmark, E., and Costa, L. G. (1990). Acrylamide is
metabolized to glycidamide in the rat: Evidence from hemoglobin adduct
formation. Chem. Res. Toxicol. 3, 406 412.
Calleman, C. J., Wu, Y., He, F., Tian, G., Bergmark, E., Zhang, S., Deng, H.,
Wang, Y., Crofton, K. M., Fennell, T., and Costa, L. G. (1994). Relationships between biomarkers of exposure and neurological effects in a group of
workers exposed to acrylamide. Toxicol. Appl. Pharmacol. 126, 361371.
Collins, J. J., Swaen, G. M. H., Marsh, G. M., Utidjian, H. M. D., Caporossi,
J. C., and Lucas, L. J. (1989). Mortality patterns among workers exposed to
acrylamide. J. Occup. Med. 31, 614 617.
Dybing, E., Sanner, T., Roelfzema, H., Kroese, D., and Tennant, R. W. (1997).
T25: A simplied carcinogenic potency index: Description of the system
and study of correlations between carcinogenic potency and species/site
specicity and mutagenicity. Pharmacol. Toxicol. 80, 272279.
European Commission (2002). Opinion of the Scientic Committee on Food
on new ndings regarding the presence of acrylamide in food. http://
europa.eu.int/comm/food/fs/sc/scf/out131_en.pdf

Littleeld, N. A., Farmer, J. H., Gaylor, P. W., and Sheldon, W. G. (1979).

Effects of dose and time in a long-term, low-dose carcinogenesis study. J.
Environ. Pathol. Toxicol. 3, 1734.
Lorentzen, R. J. (1984). FDA procedures for carcinogenic risk assessment.
Food Technol. 38, 108 111.
Marsh, G. M., Lucasw, L. J., Youk, A. O., and Schall, L. C. (1999). Mortality
patterns among workers exposed to acrylamide: 1994 follow up. Occup.
Environ. Med. 56, 181190.
Miller, M. J., Carter, D. E., and Sipes, I. G. (1982). Pharmacokinetics of
acrylamide in Fischer-344 rats. Toxicol. Appl. Pharmacol. 63, 36 44.
Miller, M. S., and Spencer, P. S. (1985). The mechanisms of acrylamide
axonopathy. Annu. Rev. Pharmacol. Toxicol. 25, 643 666.
Mottram, D. S., Wedzicha, B. L., and Dodson, A. T. (2002). Acrylamide is
formed in the Maillard reaction. Nature 419, 448 449.
Mucci, L. A., Dickman, P. W., Steineck, G., Adami, H. O., and Augustsson, K.
(2003). Dietary acrylamide and cancer of the large bowel, kidney, and
bladder: Absence of an association in a population-based study in Sweden.
Br. J. Cancer 88, 84 89.
Norwegian Food Agency (2002). Risk assessment of acrylamide intake from
foods with special emphasis on cancer risk. Oslo, 6 June 2002. Available at
http://www.snt.no.
Peto, T., Gray, R., Brantom, P., and Grasso, P. (1991). Effects on 4080 rats of
chronic ingestion of N-nitrosodiethylamine or N-nitrosodimethylamnie: A
detailed dose-response study. Cancer Res. 51, 64156451.
Robinson, M., Bull, R. J., Knutsen, G. L, Shields, R. P., and Strober, J. A.
(1986). A combined bioassay utilizing both the lung adenoma and skin
papilloma protocols. Environ. Hlth. Perspect. 68, 141145.
Sanner, T., Dybing, E., Willems, M. I., and Kroese, E. D. (2001). A simple
method for quantitative risk assessment of non-threshold carcinogens based
on the dose descriptor T25. Pharmacol. Toxicol. 88, 331341.

EU TGD (2003). Technical Guidance Document on Risk Assessment. 2nd ed.

http://ecb.jrc.it/new-chemicals/.

Segerback, D., Calleman, C. J., Schroeder, J. L., Costa, L.G., and Faustman,
E. M. (1995). Formation of N-7-(2-carbamoyl-2-hy7droxyethyl)guanine in
DNA of the mouse and the rat following intraperitoneal administration of
[ 14C]acrylamide. Carcinogenesis 16, 11611165.

Fennel, T. R., Snuder, R. W., Krol, W. L., and Sumner, S. C. J. (2003).

Comparison of the hemoglobin adducts formed by administration of Nmethylolacrylamide and acrylamide to rats. Toxicol. Sci. 71, 164 175.

Spencer, P. S., and Schaumberg, H. H. (1974a). A review of acrylamide

neurotoxicity: I. Properties, uses and human exposure. Can. J. Neurol. Sci.
1, 143150.

Friedman, M. A., Dulak, L. H., and Stedham, M. A. (1995). A lifetime

oncogenicity study in rats with acrylamide. Fundam. Appl. Toxicol. 27,
95105.

Spencer, P. S., and Schaumberg, H. H. (1974b). A review of acrylamide

neurotoxicity. Part II. Experimental animal neurotoxicity and pathological
mechanisms. Can. J. Neurol. Sci. 1, 152169.

Generoso, W.M., Sega, G. A., Lockhart, A. M., Hughes, L. A., Cain, K. T.,
Cacheiro, N. L., and Shelby, M. D. (1996). Dominant lethal mutations,
heritable translocations, and unscheduled DNA synthesis induced in male
mouse germ cells by glycidamide, a metabolite of acrylamide. Mutat. Res.
371, 175183.

Sobel, W., Bond, G. G., Parsons, T. W., and Brenner, F. E. (1986). Acrylamide
cohort mortality study. Brit. J. Ind. Med. 43, 785788.

Hashimoto, K., and Tanii, H. (1985). Mutagenicity of acrylamide and its

analogues in Salmonella typhimurium. Mutat. Res. 158, 129 133.

Sumner, S. C., MacNeela, J. P., and Fennel, T. R. (1992). Characterization and

quantitation of urinary metabolites of [1,2,3 13C]acrylamide in rats and mice

Stadler, R. H., Blank, I., Varga, N., Robert, F., Hau, J., Guy, P. A., Robert,
M. C., and Riediker, S. (2002). Acrylamide from Maillard reaction products.
Nature 419, 449.

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013

Bergmark, E., Calleman, C. J., He, F., and Costa., L. G. (1993). Determination
of hemoglobin adducts in humans occupationally exposed to acrylamide.
Toxicol. Appl. Pharmacol. 120, 4554.

International Agency for Research on Cancer [IARC] (1994). Acrylamide.

IARC Monogr. Eval. Carcinog. Risks Hum. 60, 389 433.

FORUM
using 13C nuclear magnetic resonance spectroscopy. Chem. Res. Toxicol. 5,
81 89.
Swedish National Food Agency. Press conference. Uppsala, 24 April 2002.
Available at http://www.slv.se.
Tareke, E., Rydberg, P., Karlsson, P., Eriksson, S., and Tornqvist, M. (2000).
Acrylamide: A cooking carcinogen? Chem. Res. Toxicol. 13, 517522.
Tyl, R. W., Friedman, M. A., Losco, P.E., Fisher, L. C., Johnson, K. A.,
Strother, D. E., and Wolf, C. H. (2000). Rat two-generation reproduction
and dominant lethal study of acrylamide in drinking water. Reprod. Toxicol.
14, 385 401.

15

U.S. Environmental Protection Agency [U.S. EPA] (1996). Proposed guidelines for carcinogen risk assessment. Fed. Reg. 61, 17960 18011.
U.S. Environmental Protection Agency [U.S. EPA] (1999). Draft Revised
Guidelines for Carcinogen Risk Assessment. USEPA, Washington, DC.
Wogan, G. N., Paglialunga, S., and Newberne, P. M. N. (1974). Carcinogenic
effects of low dietary levels of aatoxin B 1 in rats. Food Cosmet. Toxicol.
12, 681 685.
World Health Organization [WHO] (1996). Guidelines for Drinking-Water
Quality. 2nd ed., Vol. 2. Health criteria and other supporting information.
International Programme on Chemical Safety, Geneva.

Downloaded from http://toxsci.oxfordjournals.org/ at Universitatea Dunarea de Jos, Galaţi on December 6, 2013