January 26, 1998
To: Dr. Harry Conacher, Director
Bureau of Chemical Safety, Food Directorate
Health Protection Branch, Health Canada
Tunney's Pasture, Ottawa, Canada
From: Warren Bell, M.D., president, Canadian Association of Physicians for the Environment,
............... Salmon Arm, B.C.
Richard Clapp, D.Sc., associate professor, Department of Environmental
Health, Boston University
Devra Davis, Ph.D., director, Health, Environment, and Development
Program, World Resources Institute, Washington
Samuel Epstein, M.D., professor of environmental medicine, School of Public
Health, University of Illinois Medical Center, Chicago
Emmanuel Farber, M.D., Ph.D., professor of pathology, Jefferson Medical
College; chairman of the National Academy of Sciences' 1978 Panel I: Saccharin and its
Impurities; former professor, University of Toronto
Donald A. Fox, Ph.D., professor of biochemical and biophysical sciences,
University of Houston
Bruce Holub, Ph.D., professor of nutritional sciences, Department of Human
Biology and Nutritional Sciences, University of Guelph
Michael F. Jacobson, Ph.D., executive director, Center for Science in the
Public Interest, Washington and Toronto
William Lijinsky, Ph.D., former director, chemical carcinogenesis program,
Frederick Cancer Research Center
Erik Millstone, Ph.D., senior lecturer, Science Policy Research Unit, University
of Sussex, England; author of Additives: A Guide for Everyone (Penguin,
Melvin D. Reuber, M.D., consultant in human and experimental oncology and
pathology; former staff pathologist, National Cancer Institute; former chief, pathology laboratory,
Chemical Carcinogenesis Program, Frederick Cancer Research Center
David Suzuki, Ph.D., Vancouver, British Columbia
Norman J. Temple, Ph.D., associate professor, Centre for Natural and Human
Science, Athabasca University, Athabasca, Alberta
We appreciate this opportunity to provide input to the Health Protection Branch's (HPB) review
of the artificial sweetener saccharin. Concerns with regard to the safety of saccharin are of great
public health significance and of great interest to the public because saccharin is consumed by tens
of millions of people, including children and fetuses. Any evidence of carcinogenesis -- and there
is ample such evidence -- of such a widely used chemical should spur health officials to minimize
human exposure to it. It is worth noting that on October 31, 1997, the Board of Scientific
Counselors of the National Toxicology Program, a unit of the National Institute of Environmental
Health Sciences (NIEHS), voted not to delist saccharin from its Report on Carcinogens.
Carcinogenicity of Saccharin in Laboratory Animals
I. Limitations of Dose-Response and Mechanistic Studies
High dietary doses of sodium saccharin cause urinary bladder tumors in rats. Many chemicals are
known to cause bladder cancer, even in humans, including certain aromatic amines and azo dyes,
as well as cigarettes. Recently, however, the suggestion has been made that saccharin may be
different from other bladder carcinogens. It has been argued that a large dose-response study
shows that a "threshold" exists between 1% and 3% saccharin in the diets of male rats and that
levels below that are not carcinogenic. Hence, the argument goes, saccharin is a non-genotoxic
carcinogen that can have a threshold response below which human exposure should not be a
health concern. The proponents of that argument theorize that high levels of sodium saccharin
increase urinary sodium levels and pH, which lead to the formation of precipitates in the bladder
(containing calcium phosphate, silicate, alpha 2u-globulin protein, saccharin, and other
substances), which in turn leads to irritation, hyperplasia, and ultimately tumors.
While those studies may be relevant to evaluating saccharin carcinogenicity, they do not exculpate
saccharin as a bladder carcinogen in male rats. The dose-response study was not large enough to
evaluate the effects of the lowest tested doses of dietary sodium saccharin in Charles River CD
male rats. The authors of the study found that the incidence of bladder tumors in controls was
0/324 compared to 5/658 in the 1% dose group. The authors state that that difference was not
statistically significant. Nevertheless, the lack of significance could have been due to the small
number of animals. The proposal that saccharin is only a non-genotoxic carcinogen is, however,
only a theory, which is not supported by available animal and human data. The exact shape of the
dose-response curve at low doses of saccharin in this one strain and sex of rat, let alone in the
genetically diverse human population, is completely unknown.
Other research also indicates flaws in the theoretical exoneration of saccharin. Some studies have
shown that exposure to saccharin does not increase the urinary pH and osmolality. It has been
noted that a 7.5% sodium saccharin diet "scarcely represents a large increase from the usual daily
dietary intake of sodium ion." Furthermore, saccharin causes bladder cancer not only in male rats
but also female rats, whose urine has lower levels of protein and a higher pH. The mechanism by
which bladder tumors develop in females exposed to saccharin has not been well investigated.
Besides critically evaluating the rodent bladder-cancer studies, we urge the HPB to evaluate
carefully several lines of evidence -- tumors in rodents at sites other than the bladder,
co-carcinogenicity, genotoxicity, epidemiology -- that raise additional serious questions about
II. Rodent chronic feeding studies
Several studies on rats and mice found that saccharin causes tumors not just in the urinary
bladder, but at additional sites. Other studies show that saccharin promotes tumors initiated by
known chemical carcinogens. Some of that research is reviewed below.
Some rodent studies did not find increases in tumor rates following exposure to large doses of
sodium saccharin. Some of those studies focused on the urinary bladder without systematic
histopathologic examination of other organs, so tumors at sites other than the bladder could have
been overlooked. Also, the strains of rodents used varied among studies (for instance, the only
studies using Osborne-Mendel rats found tumors at sites other than the bladder). Thus, the
absence of reported tumors in some studies may mean only that affected organs were not
examined or that the strain was not susceptible. Human variability strongly suggests that any
finding of carcinogenesis in any strain of animal should signal great concern and caution.
A. Induction of tumors at sites other than the urinary bladder
- The two-generation WARF study found a dose-dependent increased incidence of ovarian plus
uterine tumors (controls: 1/17; 0.05%: 1/17; 0.5%: 3/15; 5%: 6/20, p = 0.07).
- The National Academy of Sciences' 1978 review concluded that, "An increase in benign uterine
tumors and ovarian lesions in saccharin-treated rats was suggested in a few studies." The NAS
stated further, "[A]s a result of detailed analyses of the pathology data, the committee
concludes that the experimental evidence suggests that ingestion of saccharin at the 5% or
7.5% dietary level may have contributed to an increase in benign uterine tumors and ovarian
lesions in female rats." The NAS's meta-analysis of four studies found a significant increase in
uterine tumors (p = 0.041) for the F0 generation and an increase that almost reached
significance (p = 0.063) for the F0+F 1 generations.
- Bio-Research Consultants' one-generation study found small numbers of forestomach and skin
tumors in treated male rats but not in controls (0%: 0/16; 1%: 2/28; 5%: 3/26).
- Several studies found an increased rate of tumors in "all organs." (The lack of monotonic
dose-response increases does not necessarily negate the finding of a higher rate in the
-- Chowaniec and Hicks found that saccharin increased the
incidence of non-bladder tumors in male rats (controls: 1/52; 2 g/kg/day: 11/71; 4 g/kg/day:
-- Bio-Research Consultants found increased rates of
non-bladder tumors (controls: 3/16 [19%]; 1%: 15/28 [54%]; 5%: 7/26 [27%]) in male rats.
Mice have been less well studied than rats. Positive findings include:
- Bio-Research Consultants found:
Saccharin caused an increased incidence of vascular tumors (males: controls:
1/19, 1%: 2/29, 5%: 10/34; females: controls: 1/17, 1%: 5/28, 5%: 7/36). OTA said those data
support "an association between saccharin and an increase in total and vascular tumors in males;
furthermore, the number of vascular tumors was increased in saccharin-fed female mice."
Saccharin caused an increased rate of lung tumors in males (controls: 11%;
1%: 48%; p = 0.007; 5%: 27%). OTA dismissed the higher rate in the 1% group, because an
additional increase did not occur at 5 percent, but the lack of an increase at 5% does not
necessarily disprove the finding at the lower dosage.
Male mice developed a higher incidence of squamous epithelium tumors (skin
or forestomach) in the 1% group (4/29) compared to none in the controls or the 5%
Male mice exposed to saccharin had higher incidences of tumors at all sites
examined (controls: 4/19 [21%]; 1%: 20/29 [69%], p = .002; 5%: 21/34 [62%], p =
An analysis by the U.S. National Cancer Institute of this study concluded that
"for tumors of the uterus among female mice the life table analysis reveals a significant effect in
the high dose group for females."
- The National Institute of Hygienic Sciences (Tokyo) found a higher incidence of tumors (all
sites examined) in female mice at doses of 0.2%, 1%, and 5% saccharin in a 21-month study.
Incidences of uterine cancer also increased at all three dosages (p = 0.00347 for comparison of
controls to all doses combined). OTA said, "If the data for uterine cancers at 21 months are
considered alone, cancer incidence appears to increase with saccharin dose. However, ... the
data do not convincingly show that saccharin caused or did not cause cancer.... [T]he incidence
in the treated animals is higher."
- Thyroid tumors occurred in 5/10 male and 3/10 female mice fed 1.5 g/kg/day saccharin for one
year (tumor incidence in controls was not reported).
B. Co-carcinogenicity studies
In addition to being tested by itself, saccharin has been tested in rodents following their exposure
to a known carcinogen. Some of those studies demonstrated that saccharin is a cancer promoter.
- A co-carcinogenicity study done by the U.S. National Center for Toxicological Research
(NCTR) found that 5% acid saccharin in the diets of female rats exposed to
N-Methyl-N-nitrosourea (MNU) increased the incidence of bladder tumors in dead and
moribund animals from 3% to 17%. The researchers considered dead and moribund animals --
about 60% of all the animals used -- especially important because they reflect both a dose
response and an indication of time to tumor. That study indicates that it is not just the sodium
salt of saccharin that increases the risk of cancer.
- Another leg of the NCTR co-carcinogenicity study found that dosages of sodium saccharin as
low as 0.1% increased tumor rates in dead and moribund animals: controls: 3%; 0.1%
saccharin: 8%; 0.5%: 17% (p < 0.018); 1%: 17% (p < 0.011); 2.5%: 37% (p < 0.002). (The
5% group had a tumor rate of only 3%, presumably reflecting a competing toxicity.) There
was no apparent threshhold for saccharin's effect. For the study as a whole, the tumor rate was
increased at 2.5% saccharin. The authors stated, "[S]odium saccharin induced a decrease in
time to tumor as observed in dead and moribund animals." They further noted that no changes
in urinary composition "were associated with increasing levels of sodium saccharin in the diet.
Thus, no threshold dose of saccharin was observed which could be related to gross changes in
urine composition." They concluded, "saccharin served as a tumor promoter in this two-stage
carcinogenesis model system by decreasing the latency period of the lesions."
- Male and female rats were given N-Butyl-N-(4-hydroxybutyl)nitrosamine (BBN) for four
weeks, followed by various levels of sodium saccharin for 32 weeks. There were
dose-dependence increases in the numbers of rats with simple hyperplasia or papillary or
nodular hyperplasia. In male rats, increases were statistically significant only when saccharin
constituted 5% of the diet, but smaller increases were seen at lower levels, including 0.4%
saccharin. In females, significant increases in simple, papillary, or nodular hyperplasia were
seen at 1% saccharin, with smaller increases at 0.04% and 0.2% saccharin. This study was
limited by high levels of hyperplasia in rats exposed only to BBN, the small numbers of rats in
each group (about 30), and by the short duration of the study. The authors stated,
"Apparently, doses up to 400 ppm saccharin in the diet are 'no effect' or threshold levels for
hyperplasia induction in both sexes, and doses from 2,000 [0.2%] to 50,000 ppm [5%] are
- Female rats (Sprague-Dawley) were given varying levels of sodium saccharin for four weeks
prior, overlapping, or after exposure to MNU. Despite the brief exposure to saccharin, tumor
rates were higher (double in some cases) in several test groups, though the differences did not
reach statistical significance at the 0.05 level.
- Several studies demonstrated that saccharin, when implanted in bladders in cholesterol pellets,
increased by several-fold the incidence of urinary-bladder tumors. (Saccharin, which is
absorbed quickly from the pellet, may have acted as an initiator, the pellet as a promoter.)
As in all chronic-toxicity studies, the association in rodents between saccharin
consumption and tumors in the bladder, uterus, ovary, skin, forestomach, lungs, and vascular
system -- with some tumor types occurred in two or more studies or both species -- does not
prove causality in every case, but the occurrence of such tumors suggests a significant public
health issue and that further research is needed.
In addition, saccharin caused bladder cancer when female rats were pre-treated
with MNU or BBN and when saccharin-cholesterol pellets were implanted in the bladders of
mice. Co-carcinogenicity is of particular concern because humans are exposed to a wide variety
of toxic agents in their food, water, air, drugs (e.g., tobacco smoke), and workplace. It is relevant
that the International Agency for Research on Cancer (IARC) has found that the urinary bladder
is the main target organ for 11 known human carcinogens. Obviously, there are many
carcinogens whose effects saccharin might promote.
The dosages of saccharin that appear to have elicited an effect in
laboratory-animal studies are not much greater than the amounts that some North Americans have
consumed. In one study, 0.5%, and possibly 0.1%, dietary saccharin (following exposure to
MNU) appeared to increase the incidence of bladder tumors in female rats; another
co-carcinogenicity study (using BBN as the initiator) found increased hyperplasia at 1% saccharin
and possibly levels as low as 0.04%. Other studies on rats and mice found an increased risk at 1%
dietary saccharin (Bio-Research Consultants). If, arguendo, a dose of 0.1% saccharin were
considered the no-effect level, that is equivalent to just 50 mg/kg bw/day, or just four times higher
than a 90th percentile adult consumer in the United States in 1977-1978 and just twice as high as
a child in the 90th percentile of consumption. That hardly gives one much confidence that
saccharin is safe for human consumption.
Several studies have shown that saccharin causes dominant lethal mutations. In one study, 1.72%
sodium saccharin in the drinking water of male mice (CBA strain) led to a six-fold increase at four
weeks in the percentage of intrauterine deaths of offsprings of females to which the males had
been mated. When injected intraperitoneally in male mice (ICR strain), several different dosing
regimens of sodium saccharin led to statistically significant (p<0.01 and <0.001) increases in
dominant lethal mutations, usually at 2-4 weeks after one or several injections. In a third study,
sodium saccharin administered five times intraperitoneally and orally to male mice (ICR strain)
caused signficant (p<0.001) decreases in implantations and live fetuses per corpora lutea. A
single intraperitoneal dose caused a significant (p<0.01) decreases of the same two measures. A
fourth study found that subcutaneous injection of saccharin in two inbred strains of mice caused
dominant lethal mutations (p<0.01).
Considering that the dominant-lethal test is rather insensitive, the fact that several studies found
that saccharin causes such mutations should be of great concern. Chemicals that cause
dominant-lethal mutations are generally found to be carcinogens. Thus, these genotoxicity studies
provide strong evidence that supports the other laboratory-animal research indicating that
saccharin is a genotoxic cancer initiator.
IV. Epidemiological studies
The question of whether saccharin consumption increases the risk of bladder cancer in humans is
significant considering that bladder cancer is the fourth most common cancer in Canadian males
(the rate is lower in females). In 1969, the incidence of bladder cancer in males was 23.8 per
100,000. That incidence rose sharply over the next decade and remained at 30 to 32 cases per
100,000 between 1978 and 1988. At that point, the incidence declined to an estimated 23.6 per
100,000 in 1996. It is conceivable that the increase and subsequent decrease were related partly
to the increased use of saccharin in the 1960s and early 1970s, followed by the ban on using
saccharin in processed foods.
Numerous case-control studies have sought to evaluate the relationship between
artificial-sweetener consumption (saccharin and cyclamate were generally used together) and the
incidence of bladder cancer. Several studies, including some of the largest ones, found significant
increases in rates of bladder cancer.
- National Cancer Institute (3,010 total cases) found relative risks (RR) for bladder cancer of
between 1.6 and 3.0 in several subgroups of Americans, including low-risk white females and
heavy-smoking males. Furthermore, with all males and females combined, this study found
relative risks of 1.53 to 1.64 among people who consumed two or more diet drinks and six or
more servings of artificial sweeteners per day (confidence intervals were not given).
- Sturgeon et al.'s analysis (1,860 cases) of the NCI data described above found that heavy use
of artificial sweeteners was associated (RR = 2.2) with higher-grade, poorly differentiated
- Howe et al. (632 cases) found an increased risk of bladder cancer in Canadian males (RR =
1.6); men who consumed more artificial sweeteners or consumed artificial sweeteners for a
longer period of time had relatively high risks.
- Cartwright (622 existing cases; 219 new) found an increased risk (RR = 2.2) in British
nonsmoking males, but not females.
- Morrison and Buring (592 male and female patients with lower-urinary-tract cancer -- 94% of
whom had bladder cancer) found increased risks in women who consumed dietetic beverages
(RR = 1.8 [1.0-3.3]) and who consumed sugar substitutes (RR = 1.9 [1.0-3.6]) (stratified for
age and smoking history). Women who consumed dietetic beverages for five years or more
had a relative risk of 3.7.
- Morrison (555 British cases) found an increased risk (RR = 2.3) in British females (but not
males or Japanese cases) who consumed more than 10 tablets of sugar substitutes (primarily
saccharin) a day.
Thus, in numerous studies, artificial-sweetener consumption was associated with significant
increased risks of bladder cancer in humans, though there were inconsistencies in risks to men and
women. Some (mostly smaller) studies did not find an association. The NTP acknowledges that
"a small increased risk in some subgroups, such as heavy users of artificial sweeteners, cannot be
unequivocally excluded." (That is an understatement that could have been expressed equally
accurately as: Several studies found an increased risk in some subgroups, and it is the subgroup
of heavy consumers about whom we should be especially concerned.)
- A small study (47 female cases) in Denmark found increased risks in all women (RR = 6.7) and
in nonsmoking women (RR = 3.3).
That some studies did not detect an increased risk could be real or due to the limited duration of
subjects' exposure to artificial sweeteners -- particularly in light of the long latency period for
cancer and the limited consumption of saccharin in the U.S. before the mid 1960s (many subjects
were exposed for under 15 years in the North American studies) -- lack of exposure in utero,
small numbers of cases and limited power to detect small risks, existence of such compounding
factors as smoking and occupational exposure to bladder carcinogens, and loss of sensitivity due
to lumping occasional users of artificial sweeteners in with heavy users. Similarly, studies of
diabetics are limited by the facts that diabetics smoke less than non-diabetics and often die
prematurely due to heart disease and other causes. Those and other limitations reduce the
likelihood that saccharin's link to a higher rate of bladder cancer could be detected by
Furthermore, no epidemiologic research has evaluated whether saccharin might cause tumors at
sites other than the urinary bladder, despite known differences in organ specificity between
species in the case of most carcinogens. In light of several rodent studies documenting higher
rates of cancer in other organs, that absence of information is troubling and suggests the need for
more research. New research would also benefit from the increased duration of exposure to
V. Summary and Recommendation
The proposal that sodium saccharin might be a non-genotoxic carcinogen is not supported by a
wide range of animal and human data. Some have argued that bladder tumors in male rats fed
saccharin are irrelevant to humans, but such arguments are flawed. While it cannot be proved that
sodium saccharin's causation of bladder tumors in male rats is relevant to humans, neither can it be
assumed to be irrelevant. Sodium saccharin also causes bladder tumors in female rats, which
differ physiologically in significant respects from males, but mechanistic studies on females have
not been conducted. In several studies in rats and mice, 1% saccharin (and possibly lower
amounts) enhanced the carcinogenicity of known bladder carcinogens. Furthermore, saccharin
also has caused tumors in the urinary bladder in mice and in other organs in various strains of rats
and mice. Positive findings in dominant-lethal tests in mice demonstrate that saccharin can cause
genetic damage consistent with a genotoxic carcinogen. Finally, in several case-control studies,
the consumption of artificial sweeteners has been associated with increased incidence of bladder
cancer in humans.
It would be highly imprudent for Health Canada to deny the carcinogenicity of saccharin on the
basis of current evidence. Doing so would give the public a false sense of security, remove any
incentive for further testing, and result in greater exposure to this probable carcinogen by millions
of Canadians, including children (indeed, fetuses). If saccharin is even a weak carcinogen, this
unnecessary additive would pose an intolerable risk to the public.
Thus, we urge the HPB on the basis of currently available data to conclude that saccharin is
reasonably anticipated to be a human carcinogen, because there is sufficient evidence of
carcinogenicity in animals (multiple sites in rats and mice) and limited or sufficient evidence of
carcinogenicity in humans (bladder cancer).
Comments on Acceptable Daily Intake
In 1993, the WHO/FAO's Joint Expert Committee on Food Additives (JECFA) reviewed the
safety of saccharin. That committee concluded that saccharin did not pose a cancer risk and
established an Acceptable Daily Intake (ADI) of 5 mg/kg bw/day. While we disagree with the
committee's conclusion about saccharin's carcinogenicity, we recognize that Health Canada or
other agencies, following JECFA's lead, might establish an ADI. We believe that JECFA set the
ADI at far too high a level, a level that could endanger the public health.
JECFA's ADI was based on a highest-no-observable-effect level (HNOEL) of 1% saccharin in
rodent studies. In fact, however, 1% has not been established as the HNOEL. As noted above, in
several studies tumor rates were increased in rodents that consumed 1% saccharin or less:
- Ovarian and uterine tumors at 0.5% saccharin in the two-generation rat study (increase was
not statistically significant, but significance might have been reached in a larger study; WARF);
- Total non-bladder tumors at 1% saccharin in male rats (Bio-Research Consultants);
- Lung, squamous epithelium (skin/forestomach), and all-sites tumors at 1% saccharin in male
mice (Bio-Research Consultants);
- Urinary bladder tumors in female rats ingesting 0.5% saccharin following one-time exposure to
MNU. There was also an increase at 0.1% that did not reach statistical significance, though
significance might have been reached in a larger study.
Such data do not support the finding of a threshold for saccharin's carcinogenic action. However,
for the sake of argument, one might consider the HNOEL to be 0.1% saccharin (equivalent to 50
mg/kg bw/day) or 0.05% -- tenfold lower than the 0.5% effect level in the West, Sheldon, et al.
1986 co-carcinogenicity study -- (equivalent to 25 mg/kg bw/day). Applying a 100-fold safety
factor to those levels would yield an ADI of 0.5 mg/kg bw/day or 0.25 mg/kg bw/day.
- Increased hyperplasia in the urinary bladder of female rats was caused by 1% dietary saccharin
for 32 weeks following exposure to BBN; non-significant increases occurred at levels of 0.2%
and 0.04% saccharin.
In 1977-1978 in the U.S., 3- to 5-year-old children in the 90th percentile of saccharin
consumption were consuming 19.67 mg/kg bw/day saccharin (NTP, p. 2-7). U.S. men and
women between the ages of 19 and 34 in the 90th percentile were consuming 10.19-10.48 mg/kg
bw/day. Those levels of consumption are significantly higher than what JECFA considered
acceptable and far higher than an ADI based on an HNOEL of 0.1% or 0.05% dietary saccharin.
Dr. Robert Maronpot, chief of the laboratory of experimental pathology at the U.S. National
Institute of Environmental Health Sciences, provided his comparisons of rat and human exposure,
as shown in the Attachment. He assumed that the No Observed Effect Level (NOEL) for urinary
bladder cancer in male rats was 1% dietary saccharin. Using the U.S. Environmental Protection
Agency's default assumption for rodent-to-human comparisons, (body weight)0.75, he concluded
that 1% saccharin in a rat's diet is only 10 times greater than a child in the 90th percentile of
consumption (U.S. Dept. Agr., 1977 survey of consumption) and 13 times greater than an adult in
the 90th percentile of consumption. Clearly, if saccharin did cause bladder cancer in rats, it could
pose a significant risk to humans who consume large amounts of saccharin.