95
Chapter 9
The Effects of Arctic Pollution on Population Health
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
Eric Dewailly and Pแl Weihe
guidelines for polychlorinated biphenyls (PCBs) in blood,
Summary
followed by west coast Greenland Inuit populations and
Very few major epidemiological studies on effects due to
Inuit populations from the Baffin and Nunavik regions
exposure to environmental contaminants have been con-
of eastern Canada. A similar pattern occurs for ex-
ducted in the Arctic. Arctic studies are extremely diffi-
ceedance of the blood Hg guidelines, but the data are
cult to undertake due to the remoteness of communities,
more limited. If the new U.S. EPA mercury guidelines are
the cultural context, climatic factors, small populations,
applied, then it is evident that most Inuit populations
and confounding social and behavioral factors, etc. The
and a significant proportion of several other populations
special characteristics of the Arctic raises the question as
exceed these guidelines.
to how far results and conclusions from epidemiological
studies conducted outside the Arctic can apply to this re-
9.1. General considerations
gion. Human exposure to contaminants generally in-
volves exposure to a mixture of many different sub-
There is considerably more information on the effects of
stances simultaneously, it is therefore not appropriate,
environmental factors on the health of Arctic popula-
and currently not even possible, in epidemiological stud-
tions than existed at the time of the previous AMAP As-
ies to establish the risk associated with individual sub-
sessment Report (AMAP, 1998), five years ago. This
stances. Nevertheless, serious consideration should be
chapter updates that information.
given to results from cohort studies into neurological dis-
In addition, this chapter examines the risks to Arctic
orders associated with prenatal methylmercury (MeHg)
peoples associated with exposure to the major contami-
exposure (in the Faroe Islands) and immune dysfunction
nants, also taking into account some lifestyle factors.
in children exposed prenatally to persistent organic pol-
These factors, when integrated, can result in risks that are
lutants (POPs) (in Nunavik). It is likely that the negative
greater than or different to those resulting from any single
effects, although small, can be detected at sites outside
effect alone. This combined effects risk assessment is prob-
the Arctic under similar exposure conditions. In adopt-
lematic. While it is intuitively obvious that exposure to
ing dietary recommendations intended to reduce the ex-
multiple risk factors should be of more concern than ex-
posure of Arctic populations, public health authorities
posure to one or two defined health risks, it is extremely
should give consideration to the possibility of negative
difficult to combine these risks in a quantitative or even
effects on public health caused by changes in lifestyle.
qualitative manner. It is possible to investigate the esti-
Inuit from the east coast of Greenland, who consume
mated risks to populations based on their exceedance of
large numbers of marine mammals, have the highest
health guidelines; however these are few in number, and
proportion of the population exceeding the (Canadian)
change as scientific knowledge improves (Table 9ท1).
Table 9ท1. Guideline values for levels of environmental contaminants in human tissues.
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
Contaminant
Media
Guideline Value (ตg/L)
Source
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
DDT and DDE
Plasma or serum
200 (total DDT)
WHO
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
PCBs1
Plasma or serum
For women of reproductive age
Canada
< 5: Tolerable
5ญ100: Concern
>100: Action
For men and post-menopausal women
Canada
< 20: Tolerable
20ญ100: Concern
>100: Action
-------------------------------------------------------------------------------------------------------------------------------------------
PCBs1
Breast milk
50: For protection of infants
Canada
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
Hg (total)
Whole blood
20: Normal acceptable range
Canada
20ญ100: Increasing risk
>100: At risk
> 4.42
United States
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
Cd
Whole blood
5: For occupational exposure
Canada
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
Pb
Whole blood
100: Action Level
Canada
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
Pb
Whole blood
100
United States
ญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญญ
1 PCBs measured as Aroclor 1260;
2 Based on the US EPA's 1999 re-evaluation of mercury: a BMDL of 44 ตg/L (58 ตg/L in cord blood), and
applying a 10-fold safety factor.
96
AMAP Assessment 2002: Human Health in the Arctic
prenatal exposure to low doses of MeHg; in New
9.2. Epidemiological studies on mercury
and persistent organic pollutants
Zealand, the Faroe Islands, and the Seychelles (Davidson
et al., 1998; Grandjean et al., 1997; Kjellstrom et al.,
The following discussion on epidemiology is restricted
1986; Myers et al., 1995a). High dietary MeHg expo-
to studies on MeHg and POPs, specifically PCBs. Other
sure in the Faroe Islands population results from con-
contaminants such as cadmium (Cd) and lead (Pb) are
sumption of pilot whale (Globicephala melaena) (Grand-
now considered of lower priority as the main sources
jean et al., 1992), while consumption of deep-sea and
giving rise to human exposure (smoking for Cd and lead
reef fish is the source of exposure for the Seychelles
shot or industrial emissions for Pb) are not Arctic-wide
(Myers et al., 1995b) and New Zealand populations
environmental contamination and information on their
(Kjellstrom et al., 1986). All these prospective MeHg
toxicity is in most cases already available, validated and
studies involved children without overt clinical symp-
widely used. Furthermore, most of the health concerns
toms of MeHg poisoning. Cord blood Hg was the main
associated with the presence of contaminants in the Arc-
indicator of prenatal exposure in the Faroe Islands
tic food chain are related to MeHg and POPs.
study, although maternal hair Hg concentration during
Over the past decades, epidemiological studies on
pregnancy was also determined. Maternal hair Hg con-
human health effects related to exposure to PCBs and
centration during pregnancy was the indicator of prena-
mercury (Hg) have been oriented toward prenatal expo-
tal MeHg exposure in the Seychelles and New Zealand
sure and children's health. Although, these have largely
studies. Hair Hg is approximately 90% MeHg and has
focused on neurological systems, hormonal effects (re-
the advantage of providing an historical record of MeHg
lated to reproduction and cancer), immune deficiency
exposure, whereas the MeHg half-life in human blood is
and cardiovascular effects have recently gained consider-
approximately 50 days (Cox et al., 1989; Kershaw et al.,
able attention.
1980; Sherlock et al., 1984). Average maternal hair Hg
Very few epidemiological studies have been conducted
concentrations in these studies varied from 4.3 to 8.8
in the Arctic. This is because knowledge about levels and
ตg/g, and a significant number of infants studied had
exposure of northern residents to lipophilic pollutants is
hair Hg concentrations higher than 10 ตg/g.
relatively recent (dating from the mid 1980s). In addi-
Two other studies investigated the effects of prenatal
tion, communities in this vast territory are small and
Hg exposure resulting from fish consumption, the first
scattered, hence travel and fieldwork are extremely ex-
in Canada ญ more specifically in the James Bay Cree pop-
pensive and the power of epidemiological studies is
ulation ญ and the second in Peru. However, these did not
weakened by small population size. However, problems
consider neurobehavioral outcome (Marsh et al., 1995;
associated with Hg contamination have been known for
McKeown-Eyssen et al., 1983).
many years; and recognized in the Arctic for nearly 25
The Faroe Islands study reported associations be-
years. As a consequence, effect studies have been con-
tween maternal hair Hg concentration corresponding to
ducted all over the world, including the Arctic.
the pregnancy period and children's performance in neu-
robehavioral tests, particularly concerning fine motor
function, concentration, language, visualญspatial abili-
9.2.1. Neurobehavioral effects
ties and verbal memory (Grandjean et al., 1997). How-
9.2.1.1. Mercury
ever, neuropsychological dysfunction was the parameter
Mercury is a heavy metal that enters the environment
most closely associated with cord blood Hg concentra-
from both natural and anthropogenic sources. The main
tion (Grandjean et al., 1999a).
anthropogenic sources include combustion of fossil
The New Zealand study also found adverse effects of
fuels, and waste incineration. Inorganic Hg is converted
prenatal MeHg exposure (Kjellstrom et al., 1986). More
to MeHg by bacteria in aquatic systems and this then
specifically, higher hair Hg levels were associated with
bioaccumulates in marine and freshwater food chains.
poorer neurodevelopmental test scores for similar do-
MeHg is highly fetotoxic. The developmental neurotox-
mains to those observed in the Faroe Islands study.
icity of MeHg first became evident during the 1950s
In the Seychelles study, in which the exposure and re-
when Minamata Bay, Japan, was heavily contaminated
search design were similar to that of the New Zealand
by industrial effluents discharged to the bay. Infants
study, however, prenatal MeHg exposure was not found
born to women who had eaten fish from the bay exhib-
to be related to neurobehavioral effects (Davidson et al.,
ited a range of impairments to the central nervous sys-
1995, 1998; Myers et al., 1995b).
tem, including mental retardation, primitive reflexes,
Thus, prenatal MeHg exposure from marine food
cerebral ataxia, and seizures (Harada, 1995). A second
consumption was associated with neurobehavioral
well-documented MeHg poisoning occurred in Iraq in
deficits in the Faroe Islands study but not the Seychelles
the 1970s when seed grain treated with a MeHg fungi-
study. Less consideration has been given to the outcome
cide was used in households to make bread (Amin-Zaki
of the New Zealand study because, until recently, it had
et al., 1976). The affected Iraqi population was exposed
not been subject to peer review. Differences in the neu-
to higher levels than that in Japan, and the exposure
ropsychological test battery, Hg exposure, age at testing,
took place over a much shorter period of time, yet the
and the source of Hg exposure are suggested to account
neurodevelopmental effects were similar, including se-
for the differences between the findings of the Faroe Is-
vere sensory impairment, general paralysis, hyperactive
lands and Seychelles studies. When the New Zealand
reflexes, cerebral palsy, and impaired mental develop-
data are considered, these factors no longer seem deter-
ment (Marsh et al., 1987).
minative, because the exposure and research design of
Three well-designed, prospective, longitudinal stud-
the New Zealand study were similar to those of the Sey-
ies have been performed which examine the effects of
chelles study, yet adverse neurobehavioral effects were
Chapter 9 ท The Effects of Arctic Pollution on Population Health
97
observed (as they were in the pilot study conducted
flexes, tonicity and activity levels) in three of the four
in the Seychelles) (NRC, 2000). One limitation of the
studies (Huisman et al., 1995a; Rogan et al., 1986a,
Faroe Islands study was that, owing to the confounding
Stewart et al., 2000). Adverse neurological effects of ex-
effects of prenatal exposure to both Hg and PCB (r =
posure to PCBs were found up to 18 months of age in
0.41 to 0.49) (Grandjean et al., 1997, 1999a), it was dif-
the Netherlands study (Huisman et al., 1995b). In Mich-
ficult to determine whether several of the neurodevelop-
igan and the Netherlands, higher cord serum PCB con-
mental deficits observed at age 7 years, especially those
centrations were associated with lower birth weight and
in language and memory function, were due to prenatal
slower growth rate (Fein et al., 1984; Jacobson et al.,
Hg exposure, to prenatal PCB exposure, or to both ex-
1990b; Patandin et al., 1998). In Michigan, prenatal
posures. However, patterns of neurobehavioral damage
PCB exposure was associated with poorer visual recog-
produced by developmental Hg exposure in animals re-
nition memory in infancy (Jacobson et al., 1985, 1990a,
semble those found in humans and include sensory sys-
1992), an effect recently confirmed in the Oswego study
tem effects, motor or sensorimotor system effects, and
(Darvill et al., 2000), and was linked to poorer intellec-
cognitive effects.
tual function at 4 and 11 years (Jacobson et al., 1990a;
The relationship between prenatal exposure to
Jacobson and Jacobson, 1996), a finding recently con-
MeHg and neurological and developmental abnormali-
firmed in the Netherlands study at 42 months (Patandin
ties was ascertained among 234 Cree Indian children
et al., 1999). In North Carolina, deficits in psychomotor
aged 12 to 30 months from four northern Quebec com-
development at up to 24 months were seen in the most
munities (McKeown-Eyssen et al., 1983). A pediatric
highly exposed children (Gladen et al., 1988; Rogan and
neurologist, `blinded' to the children's level of exposure,
Gladen, 1991). Although much larger quantities of PCBs
assessed neurological, physical, mental, and psychoso-
are transferred to nursing infants by breast feeding than
cial development. MeHg exposure was estimated from
prenatally across the placenta, virtually all the adverse
maternal hair segments representing the period of preg-
neurobehavioral effects reported to date are linked
nancy. Abnormality of the tendon reflexes, observed in
specifically to prenatal exposure, indicating that the em-
13 boys (11%) and 14 girls (12%), was positively asso-
bryo and fetus are particularly vulnerable to PCBs.
ciated with MeHg exposure in boys only and there was
A German cohort of 171 children showed average
no consistent doseญresponse relationship. Other neuro-
PCB concentrations (based on CB138, CB153 and CB180)
logical disorders were less prevalent and none were pos-
in cord serum and maternal milk that were similar to
itively associated with exposure; indeed, incoordination
those reported in the Netherlands study (Winneke et al.,
was positively associated with exposure in girls.
1998). Negative associations between maternal milk PCB
Cord blood samples (n=42) from Qaanaaq (north-
levels and mental/motor development were reported at
west Greenland) were collected and analyzed in 1982
all ages, becoming significant from 30 months onwards.
and the children examined at 7 to 12 years of age. Clini-
Over 30 months of age, for a PCB increase from 173 to
cal neurological examination did not reveal any obvious
679 ng/g lipids in milk there was a decrease of 8.3 points
deficiencies. However, neurophysiological tests (audi-
in the Bayley Scales of Infant Development mental scores,
tory evoked potentials) showed possible Hg exposure-
and a 9.1 point decrease in the Bayley Scales of Infant
associated deficiencies, although in only a few cases
Development motor scores. There was also a negative
reaching levels of statistical significance (Weihe et al.,
effect of postnatal PCB exposure via breast feeding, at
2002).
42 months. Home environment had a positive effect
from 30 months onwards (Walkowiak, 2001).
A prospective, longitudinal study set up to examine
9.2.1.2. Polychlorinated biphenyls
the effects of prenatal exposure to low doses of MeHg
The developmental toxicity of heat-degraded PCBs was
resulting from fish and pilot whale consumption was
first recognized in Japan in the late 1960s and in Taiwan
performed in the Faroe Islands (Grandjean et al., 1992,
in the late 1970s. In similar industrial accidents in both
1997). Because pilot whale tissues contain other neuro-
countries, infants born to women who had consumed
toxins this cohort was also exposed to PCBs. This is the
rice oil contaminated with mixtures of PCBs and poly-
only cohort studied to date where the main source of
chlorinated dibenzofurans (PCDFs) had skin rashes and
PCB exposure was the consumption of marine mam-
exhibited poorer intellectual functioning during infancy
mals, as is the case for the Inuit. However, it was diffi-
and childhood (Chen et al., 1992; Yu et al., 1991).
cult to determine whether several of the neurobehavioral
Effects of prenatal exposure to background levels of
deficits observed at age 7 years, especially on language
PCBs and other organochlorine compounds (OCs) from
and memory function (Budtz-Jorgensen et al., 1999),
environmental sources have been studied since the 1980s
were due to prenatal MeHg exposure, to PCB exposure
in prospective longitudinal studies in Michigan, North
or to both. Nevertheless, after determining PCB concen-
Carolina, the Netherlands and Oswego (New York).
trations in cord tissue for 50% of the cohort, detailed
The source of PCB exposure was fish consumption from
statistical analyses showed that confounding factors
the Great Lakes in both the Michigan (Schwartz et al.,
were limited, and interaction with MeHg-associated ef-
1983) and Oswego (Stewart et al., 1999) studies, and con-
fects was unlikely. A complete review of epidemiological
sumption of dairy products in the Netherlands (Koop-
studies related to PCB exposure and neurodevelopmen-
man-Esseboom et al., 1994a). Newborn infants from the
tal effects in newborn infants was published recently
North Carolina cohort were exposed to background lev-
(Ribas-Fito, 2001).
els of PCBs, and there was no specific source of exposure
New findings on the effects of both MeHg and PCB
(Rogan et al., 1986b). PCB exposure was associated
exposure have been made since the previous AMAP As-
with less optimal newborn behavioral function (e.g., re-
sessment Report, increasing significantly the body of sci-
98
AMAP Assessment 2002: Human Health in the Arctic
entific evidence linking neurodevelopmental effects dur-
of delivering women and their outcome within the gen-
ing infancy and childhood with prenatal exposure from
eral population of the area, including an assessment of
maternal consumption of fish and marine mammals.
essential and toxic trace elements. Over the course of the
One prospective longitudinal study on this topic involv-
study, the health of an occupationally exposed popula-
ing Nunavik and Greenland mothers and infants has
tion of female Ni industry workers and children living
been ongoing since 1997 (Dewailly and Bjerrgard, pers.
on the Kola Peninsula also became an important focus.
comm., 2001). The data are still being analyzed and no
Between April and June 1994, maternal information, de-
results have yet been published. In the Faroe Islands, the
livery information, and maternal and neonatal blood
cohort established in 1986/87 (Grandjean et al., 1992)
and urine samples were collected for approximately 50
was re-examined at age 14 years. The examination was
deliveries in each of the Russian cities Arkhangelsk,
basically the same as that undertaken when the children
Nikel, and Monchegorsk; the three Norwegian study
were 7 years of age. However, at 14 years special atten-
areas were Kirkenes, Hammerfest, and Bergen. Urinary
tion was given to the maturation of the body. In 2001
Ni concentrations were significantly higher in the Russ-
and 2002 a cohort of 182 7-year old children was exam-
ian study groups. Sources of Ni exposure for the Russian
ined in same manner as for the 1986/87 Hg cohort at 7
population that could account for this remain unidenti-
years. A third cohort of 650 children will be examined
fied. Environmental Ni exposure, as measured by uri-
neuropediatrically during 2002 to 2004 at 54 months
nary Ni excretion, was not shown to be a predictor vari-
(4.5 years), however the main emphasis of this study will
able of either low birth weight or newborn BMI. The
be on immunological deficits related to POPs.
mineral status of delivering women in Arctic and subarc-
tic areas of Norway and western Russia was adequate,
with the exception of zinc (Zn). The variation in serum
9.2.2. Reproduction
Zn concentrations demonstrated differences between
Typical OC mixtures found in highly exposed human
Arctic and subarctic areas across national borders. Ma-
populations contain a large variety of OCs, including
ternal serum Zn was a positive predictor variable for
substances with estrogenic, anti-estrogenic or anti-an-
birth weight. Maternal blood Pb was a negative predic-
drogenic properties. Arnold et al. (1996) showed that
tor of birth weight, even at the relatively low concentra-
compounds such as endosulfan, dieldrin, toxaphene and
tions in this study. The inclusion of maternal Pb in a
chlordane, which are practically devoid of estrogenic ac-
multivariate model caused the non-specific country dif-
tivity on their own, caused synergistic activation of the
ference to lose statistical significance. Blood Pb concen-
estrogen receptor when tested in combination in a yeast
trations in school children living in remote areas of the
estrogen system. It may thus be anticipated that complex
Kola Peninsula were up to a level of medical concern.
real life mixtures, composed of numerous compounds,
Mean birth weight and BMI were significantly lower in
which can interact with different receptors, may result in
the Russian study groups, suggesting possible nutritional
impaired male fertility in adulthood.
deficiencies during pregnancy in Russia. Congenital mal-
formations were too scarce to be assessed.
With the exception of the negative effects of mater-
9.2.2.1. Hormones
nal Pb concentrations, the other pollutants studied had
No studies on hormones have been conducted in the
no observed effects on pregnancy outcome, birth weight
Arctic to date. However, a pilot study was recently un-
or BMI. The statistically significant difference between
dertaken in Greenland (n=48 males) and the following
the Norwegian and Russian mean BMIs suggests that
male hormones measured: DHEA, 5-diol, 4, testos-
malnutrition during pregnancy may occur in the Russian
terone, DHT, E1, and E2. There were no correlations be-
study groups. Blood Pb concentrations up to a level of
tween hormone levels and POP levels adjusted for age,
medical concern were observed in children living in
body mass index (BMI) and smoking (Dewailly, pers.
remote areas of the Kola Peninsula (Odland et al.,
comm., 2001). A multi-center fertility study involving
1999a,c, 2001).
600 males and 200 pregnant women began in Greenland
in 2002. Male fertility parameters and time-to-preg-
9.2.3. Cancer
nancy are the most important endpoints of the study.
POPs are part of the exposure assessment (Pedersen,
Results from early human studies generally supported
pers comm., 2001).
the existence of a relationship (Dewailly et al., 1994a;
Falck et al., 1992; Mussalo-Rauhamaa et al., 1990;
Wolff et al., 1993) or suggested a possible link (Krieger
9.2.2.2. Pregnancy outcome
et al., 1994) between breast cancer risk and OC expo-
From 1989 to 1991 Dewailly et al. (1993) undertook a
sure, more specifically exposure to p,p'-DDE, the main
cohort study on the health effects of Inuit newborn pre-
metabolite of DDT. In contrast, recent studies involving
natally exposed to OCs. A statistically significant
larger sample sizes yielded negative results (H๘yer et al.,
association was found between male newborn length
1998; Hunter et al., 1997; L๓pez-Carillo et al., 1997;
and PCBs and PCDD/Fs, even after adjustment for po-
Moysich et al., 1998; van't Veer et al., 1997). In particu-
tential confounding factors.
lar, Hunter et al. (1997) and H๘yer et al. (1998) using a
A study was initiated in 1994 because of public con-
nested case control study design, failed to observe a rela-
cern about adverse reproductive health effects and preg-
tionship between p,p'-DDE or PCB plasma concentra-
nancy outcome in the RussianญNorwegian border zone
tions and breast cancer risk. However, H๘yer et al.
near the nickel (Ni) producing industry on the Kola
(1998) reported that high plasma concentrations of diel-
Peninsula. The original objective was to assess the health
drin were associated with breast cancer risk. Previous
Chapter 9 ท The Effects of Arctic Pollution on Population Health
99
studies have focused solely on the risk of developing a
at 3, 7 and 12 months of age. None of the immuno-
new breast cancer. However, hormonally active OCs
logical parameters were associated with prenatal OC
might also modulate cancer growth and prognosis (De-
exposure.
mers et al., 2000).
Otitis media was the most frequent disease with
Rubin et al. (2001) reported on levels of DDT, DDE,
80.0% of breast-fed and 81.3% of bottle-fed infants ex-
other chlorinated pesticides, and PCBs in serum samples
periencing at least one episode during the first year of
collected from 131 Alaska Native women between 1980
life. During the second follow-up period, the risk of oti-
and 1987 to enable a comparison with other published
tis media increased with prenatal exposure to p,p'-DDE,
studies of DDE and PCB exposure in U.S. women. The
hexachlorobenzene (HCB) and dieldrin. The relative risk
data collected during this case-control study of the rela-
(RR) for 4- to 7-month old infants in the highest tertile
tionship between OCs and breast cancer showed mini-
of p,p'-DDE exposure as compared to infants in the low-
mal differences between cases and controls. Data for
est was 1.87 (95% confidence interval (CI) 1.07ญ3.26).
case and control women were consequently pooled in
The relative risk of otitis media over the entire first year
the statistical analysis. More than 99% of the women
of life also increased with prenatal exposure to p,p'-
had detectable levels of p,p'-DDE (mean 9.1 ตg/L). The
DDE (RR 1.52; 95% CI 1.05ญ2.22) and HCB (RR 1.49;
mean total PCB level was 7.6 ตg/L. Levels of exposure
95% CI 1.10ญ2.03). Furthermore, the relative risk of re-
varied according to geographical location and ethnic
current otitis media (3 episodes) was augmented by
identification, and may reflect dietary differences. Five
prenatal exposure to these compounds. No clinically rel-
of the OCs were detected in at least half of the study
evant differences were noted between breast-fed and
population. Alaska women had levels similar to those
bottle-fed infants with regard to biomarkers of immune
reported for the United States.
function and prenatal OC exposure was not associated
with these biomarkers. It was concluded that prenatal
OC exposure could be a risk factor for acute otitis media
9.2.4. Immune system effects
in Inuit infants (Dewailly et al., 2000a).
Several OCs display immunotoxic properties in both
In another cohort, the risk of experiencing frequent
laboratory animals and humans. In children and young
infectious disease episodes was assessed in 89 children
adults accidentally exposed to PCBs and PCDFs in Tai-
exposed to PCBs and DDT in northern Quebec during
wan (`Yu-Cheng disease'), serum IgA and IgM concen-
their first year of life. The risks were related to maternal
trations as well as percentages of total T-cells, active T-
PCB and DDT blood levels during pregnancy. Ratios
cells and suppressor T-cells, were decreased compared to
were estimated using logistic regression and the results
values of age- and sex-matched controls (Chang, 1981).
were adjusted for maternal smoking during pregnancy,
An investigation of delayed type hypersensitivity re-
the number of smokers in the house, crowding, breast
sponses further indicated that cell-mediated immune sys-
feeding duration, and gender. This study supports the
tem dysfunction was more frequent among patients than
hypothesis that the high incidence of infections observed
controls. Infants born to Yu-Cheng mothers had more
in Inuit children (mostly respiratory infections) is due in
episodes of bronchitis and/or pneumonia during their
part to high prenatal exposure to POPs (Dewailly et al.,
first six months of life than unexposed infants from the
2001b).
same neighborhoods (Rogan et al., 1988). The authors
speculated that the increased frequency of pulmonary
9.2.5. Cardiovascular effects
disease could result from a generalized immune disorder
induced by transplacental or breast milk exposure to
While there are no studies that report an association be-
dioxin-like compounds, probably PCDFs (Rogan et al.,
tween cardiovascular disease and POPs, Salonen et al.
1988). Eight- to 14-year old children born to Yu-Cheng
(1995) suggest that the high mortality from cardiovascu-
mothers were shown to be more prone to middle-ear
lar disease observed among fish eaters from Finland
diseases than matched controls (Chao et al., 1997).
could be explained by the high Hg content in fish
Organic and inorganic Hg possess cytotoxic activi-
(mainly non-fatty freshwater species). They noted a sig-
ties for cellular components of immune systems in sev-
nificant association between Hg concentration in the
eral species of rodent. Moreover, it has been demon-
hair of eastern Finnish men and the risk of coronary
strated that MeHg can affect the functions of B-cells and
heart disease. Mercury can promote the peroxidation of
therefore reduce the humoral-mediated response (Daum,
lipids, resulting in more oxidized low-density lipopro-
1993). Exposure to inorganic Hg induces allergies and
tein, which has been implicated as an initiator of arte-
auto-immune problems in hypersensitive individuals.
riosclerosis. An enhanced risk of death from coronary
There is limited evidence from epidemiological studies.
heart disease in subjects with low serum selenium (Se)
In Nunavik, an epidemiological study investigated
concentrations, an antioxidant that can block the Hg-in-
whether OC exposure was associated with the incidence
duced lipid peroxidation, was previously observed in
of infectious diseases in Inuit infants and with immune
this population (Salonen et al., 1982).
dysfunction (Dewailly et al., 2000a). The number of in-
That both Hg and Se can modulate the risk of coro-
fectious disease episodes in 98 breast-fed and 73 bottle-
nary heart disease is also suggested by observations of
fed infants was compiled during the first year of life.
fish-eating coastal populations such as Inuit in Arctic re-
Organochlorine concentrations were measured in early
gions. Inuit consume large amounts of fish and marine
breast milk samples and used as surrogates for prenatal
mammals and consequently receive large doses of Hg.
exposure levels. Biomarkers of immune system function
However, contrary to the situation in eastern Finland,
(lymphocyte subsets, plasma immunoglobulins) were de-
the mortality rate from coronary heart disease in Inuit is
termined in venous blood samples collected from infants
extremely low (Dewailly et al., 2001a). This could be
100
AMAP Assessment 2002: Human Health in the Arctic
due to the high intake of Se in this population, through
originating from consumption of marine mammals.
the consumption of traditional food items rich in Se,
Given that neurobehavioral function varies considerably
such as muktuk (beluga and narwhal skin) and marine
within a population, even similar exposure circum-
mammal liver or polyunsaturated fatty acids.
stances may be associated with widely different perform-
Blood pressure in childhood is an important determi-
ance results in a group of exposed subjects. Also, despite
nant of hypertension risk later in life and MeHg expo-
results of functional tests remaining well within the
sure is a potential environmental risk factor. A birth
expected interval, differences can still be considerable
cohort of 1000 children from the Faroe Islands was ex-
between groups of individuals with different levels of ex-
amined for prenatal exposure to MeHg, and blood pres-
posure. Furthermore, although the individual may not
sure, heart rate, and heart rate variability were deter-
be aware of any dysfunction, even minimal changes can,
mined at seven years of age (Sorensen et al., 1999). After
in some cases, have severe implications for daily life.
adjustment for body weight, diastolic and systolic blood
These considerations are important when interpreting
pressure were shown to increase by 13.9 mmHg (95%
neurobehavioral data.
CI 7.4ญ20.4) and 14.6 mmHg (95% CI 8.3ญ20.8) re-
spectively, when cord blood Hg concentrations increased
9.3.1. Methylmercury
from 1 to 10 ตg/L. Above this level, which corresponds
to the current exposure limit, no further increase was
By spring 2002 there was still no international agree-
seen. Birth weight acted as a modifier, with the Hg effect
ment on the risk assessment of MeHg. According to the
being stronger in children with lower birth weights. In
World Health Organization (WHO, 1990) a "prudent
boys, heart rate variability decreased with increasing Hg
interpretation of the Iraqi data implies that a 5% risk
exposure, particularly from 1 to 10 ตg/L cord blood, at
may be associated with a peak mercury level of 10ญ20
which point the variability was reduced by 47% (95%
ตg/g in maternal hair." The WHO expert committee that
CI 14ญ68%). These findings suggest that prenatal expo-
drew up Environmental Health Criteria 101 on MeHg,
sure to MeHg might affect the development of cardio-
stated: "There is a need for epidemiological studies on
vascular homeostasis.
children exposed in utero to levels of methylmercury
that result in peak maternal hair mercury levels below
20 ตg/g, in order to screen for those effects only de-
9.3. Risk assessment
tectable by available psychological and behavioral
of exposure at levels found in the Arctic
tests."
Epidemiological research should be used, if available, in
Several major epidemiological studies have been per-
the risk assessment of exposure to contaminants. How-
formed since this recommendation. Scientific commit-
ever, few epidemiological studies exist and those that do
tees under the WHO and the U.S. National Research
cover few substances and only some of the possible clin-
Council (NRC) recently reviewed the outcome of these
ical endpoints. In most cases risk assessment is based
studies. However, their conclusions differ, as can be seen
on animal studies. Of the contaminants in the Arctic,
in the following extracts from the conclusions.
MeHg is the best researched using epidemiological
In 2000, the fifty-third meeting of the Joint FAO/
methods; and of the several relevant endpoints, neuro-
WHO Expert Committee on Food Additives concluded
toxic effects are the best described by epidemiological
on limit values to Hg exposure (WHO, 2000) as follows:
methods.
"The studies in the Faeroe Islands and the Seychelles
The neurobehavioral performance of an individual in
that were evaluated by the Committee did not provide
response to contaminant exposure is affected by several
consistent evidence of neurodevelopmental effects in chil-
factors. The outcome may depend on specific character-
dren of mothers whose intake of methylmercury yielded
istics of the exposure, i.e., severity and chronicity, as
hair burdens of 20 ตg/g or less. The Committee could
well as the possibility of simultaneous exposure to other
not evaluate the risks for the complex and subtle neuro-
contaminants. Also, the effect depends on the vulnera-
logical end-points used in these studies that would be as-
bility of the subject, as indicated by age, lifestyle, gender
sociated with lower intakes. In the absence of any clear
and premorbid status, etc. With prenatal exposure in
indication of a consistent risk in these recent studies, the
particular, the time of the neurobehavioral assessment is
Committee decided to maintain the PTWI of 3.3 ตg/kg
of importance, as the effects may not become apparent
and recommended that methylmercury be re-evaluated
until the nervous system has matured sufficiently to ex-
in 2002, when the 96-month evaluation of the Seychelles
press the dysfunction. Epidemiological data are not a
cohort and other relevant data that may become avail-
prerequisite for risk assessment. In fact, opportunities
able can be considered. The Committee noted that fish
for epidemiological studies of neurotoxicity may arise
contribute importantly to nutrition, especially in certain
only when prevention has failed, whether or not a risk
regional and ethnic diets, and recommended that nutri-
assessment has been carried out, and whether or not
tional benefits be weighed against the possibility of harm
the origin of the exposure is natural or anthropogenic.
when limits on the methylmercury concentrations in fish
Given that neurotoxic exposures continue to occur, the
or on fish consumption are being considered."
best possible epidemiological studies should be under-
In 2000, the Board on Environmental Studies and
taken to ensure that the unfortunate incidents will at
Toxicology of the U.S. NRC Committee on the Toxico-
least result in useful information that can provide a bet-
logical Effects of Methylmercury concluded as follows
ter basis for intervention. Frequently, several exposure
(NRC, 2000):
factors must be determined, as the exposure under study
"On the basis of its evaluation, the committee's con-
is associated with other chemical exposures originating
sensus is that the value of US-EPA's current Reference
from the same source, e.g., in the Arctic, Hg and PCBs
Dose (RfD) for MeHg, 0.1 ตg/kg per day, is a scientifi-
Chapter 9 ท The Effects of Arctic Pollution on Population Health
101
cally justifiable level for the protection of public health.
els of around 1.2 ตg/g and a maternal blood concentra-
However, the committee recommends that the Iraqi
tion of 4.4 ตg/L). From a food consumption survey in
study no longer be used as the scientific basis of the RfD.
New Jersey it was estimated that 21% of women of re-
The RfD should still be based on the developmental neu-
productive age exceed the Reference Dose. In general, the
rotoxic effects of MeHg, but the Faeroe Islands study
concentration in hair is 250 to 300 times the simultane-
should be used as the critical study for the derivation of
ous concentration in blood. Methylmercury is around
the RfD. Based on cord blood analyses from the Faeroe
25% higher in cord blood than maternal blood. Hence
Islands study, the lowest BMD (Benchmark Dose) for a
4.4 ตg/L in maternal blood corresponds to 5.8 ตg/L in
neurobehavioral end point the committee considered to
cord blood, which is the limit, when the lowest Bench-
be sufficiently reliable is for the Boston Naming Test.
mark Dose found by the U.S. NRC is divided by an un-
For that end point, dose-response data based on Hg con-
certainty factor of 10.
centrations in cord blood should be modeled using the
In the 1970s, Health Canada developed blood guide-
K-power model (K 1). This approach estimates a BMDL
lines for Hg and indicated that <20 ตg/L was the normal
(BMD lower confidence limit) of 58 ตg/L of Hg in cord
range, 20 to 100 ตg/L the increasing-risk range, and
blood (corresponding to a BMDL of 12ตg/g of Hg in hair)
>100 ตg/L the at-risk range. Figure 9ท1 also compares
as a reasonable point of departure for deriving a RfD. To
Hg levels reported in the Arctic with the new U.S. EPA
calculate the RfD, the BMDL should be divided by un-
level of 4.4 ตg/L for maternal blood (comparable to a
certainty factors that take into consideration biological
cord blood level of 5.8 ตg/L). Figure 9ท1 is in fact based
variability when estimating dose and MeHg database in-
on exceedance of 5.8 mg/L, which results in slightly
sufficiencies. As stated earlier, given those considera-
lower levels of exceedance than would have been the
tions, an uncertainty factor of at least 10 is supported by
case using the 4.4 mg/L value.
the committee."
In Greenland, only 3% of mothers in the Nuuk Re-
It is noted that the `reasonable point of departure' for
gion exceeded the 20 ตg/L blood Hg guideline, com-
deriving a reference dose is 12 ตg/g in hair, which is
pared with 31% of mothers in Ilullissat and 45% of
within the range of the 1990 WHO recommendations
non-pregnant women in Ittoqqortoormiit (the corre-
(10ญ20 ตg/g in hair). It is the chosen uncertainty factor
sponding value for pregnant women from this region
of 10 that leads to a limit value in hair of only 1.2 ตg/g.
was 13%). Non-pregnant Ittoqqortoormiit women also
Not only will some of the Arctic populations exceed this
had the highest percentage exceedance of the 5.8 ตg/L
value, e.g., in Greenland and the Faroe Islands, but even
blood Hg guideline ญ 93% ญ followed by mothers from
some non-Arctic populations in the United States (NRC,
Ilullissat (80%) and Ittoqqortoormiit women (both preg-
2000).
nant and non-pregnant) (68%). Although Nuuk women
From food consumption surveys, the U.S. EPA has
showed the lowest percentage exceedance of the 5.8
estimated that 7% of women nationally exceeded the
ตg/L guideline ญ 27% ญ this is still a substantial propor-
U.S. EPA Reference Dose (corresponding to hair Hg lev-
tion (Figure 9ท1).
Mercury in blood, % exceedance
Faroe
Alaska
Canada
Greenland
Finland
Siberian Russia
Islands
100
90
80
70
60
50
40
30
20
10
0
NWT
NWT
NWT
Inuit
Inuit
Inuit
Inuit
Inuit
NuukInuit
Other
BaffinInuit
BethelYup'ik
/M้tis
I๑upiat
Inuuvik
Kivalliq
Norilsk
Taymir
Yamal
Nunavik
Ilullissat
Lapland
Dudinka
Kitikmeot
Caucasian Dene
Salekhard
Indigenous Indigenous
Barrow 2001
Ittoqqortoormiit
Ittoqqortoormiit
Non-indigenousNon-indigenousNon-indigenous
Pregnant Inuit women
% of samples
5.8 ug /L
Non-pregnant Inuit women
Based on USEPA 1999 re-evaluation
of mercury; BMDL-derived
Figure 9ท1. Mercury levels in blood of indigenous and non-indige-
% of samples
20 ug /L
Based on Health Canada Guideline
nous women of reproductive age. Percentage of samples exceeding
U.S. EPA and Health Canada guidelines for increasing risk range.
102
AMAP Assessment 2002: Human Health in the Arctic
In Qaanaaq, in the Thule district of northern Green-
Risk management in the Faroe Islands community
land, 43 children were examined in 1995. Their age
has now become a matter of finding an appropriate un-
ranged from 6.2 to 12.0 years, with a median of 8.4. The
certainty factor. There is no consistent approach in the
children's hair Hg concentrations varied up to 18.4 ตg/g
application of uncertainty factors. The adoption of an
(geometric mean, 5.5 ตg/g). Maternal hair samples (n=31)
uncertainty factor represents a scientific policy judgment
showed a maximum Hg concentration of 32.9 ตg/g and
that has a major influence on what is acceptable to eat,
geometric mean of 15.5 ตg/g (Weihe et al., 2002).
e.g., what types of marine food. An uncertainty factor of
The Canadian data show that among the Inuit
10 is considered adequate by most risk assessment agen-
women from NWT/Nunavut, 3% exceeded 20 ตg/L, the
cies to address the variation in response between differ-
Canadian Level of Concern for MeHg, and 34% ex-
ent population groups.
ceeded the lower 5.8 ตg/L U.S.-based guideline; Nunavik
The responsible public health authorities must find
and Baffin Inuit women had the highest percentage ex-
the magnitude of this factor appropriate for the commu-
ceedance (16 and 9.7%, respectively). The percentage
nities in the Arctic. They must evaluate whether the pop-
exceedance of the 5.8 ตg/L guideline among Canadian
ulations they are responsible for are more or less sensi-
Inuit women overall was 34%, and ranged from 16% in
tive than the populations in the benchmark epidemiolog-
Inuvik to 68% in Baffin. Among non-Inuit women (i.e.,
ical studies. The uncertainty comprises many contribu-
Caucasian, Dene/M้tis, Others), none exceeded the 20
tors: nutritional status, levels of beneficial nutrients
ตg/L guideline and only 1%, all Dene/M้tis, exceeded
(such as Se and fatty acids from seafood), general health,
the 5.8 ตg/L guideline.
genetic background, inter-individual variability in sensi-
None of the blood Hg levels for women from the five
tivity, exposure to other neurotoxins found in the Arctic,
regions of Siberian Russia exceeded the 20 ตg/L guide-
e.g., PCBs, and peak exposures in critical periods of sus-
line, but 11% and 17% of those from Taymir and Ya-
ceptibility for the developing fetal brain.
mal, respectively ญ both indigenous regions ญ exceeded
However, other health endpoints more sensitive than
the 5.8 ตg/L guideline.
neurotoxicity could be of greater relevance in some
In Alaska, 48% of women from Bethel had blood Hg
areas. For example, the low incidence of coronary heart
levels greater than or equal to the 5.8 ตg/L guideline,
disease in Greenland Inuit, possibly due to the fatty acid
while all those from Barrow were below. Neither group
composition of their diet, could be attenuated by high
exceeded the 20 ตg/L guideline.
Hg exposure, since recent studies indicate that Hg can
In Finland, none of the maternal blood Hg levels ex-
have a negative effect on the cardiovascular system (Ris-
ceeded the 20 ตg/L guideline (no information was avail-
sanen et al., 2000). The reason for this is still unknown,
able on exceedance of the 5.8 ตg/L guideline, but a pop-
but Hg could inhibit important antioxidative mecha-
ulation mean and range of 1.4 and 0. to 6.0 respectively,
nisms in humans. Mercury could promote the peroxida-
suggests few individuals would exceed this level).
tion of unsaturated fatty acids such as docosahexaenoic
In the Faroe Islands the geometric mean hair Hg con-
acid (DHA) and docosapentaenoic acid (DPA). Regard-
centration at parturition in 1986/87 was 4.27 ตg/g; in
ing cardiovascular toxicity at low level MeHg exposures,
1994, 4.0 ตg/g; and in 1998/99, 2.1 ตg/g. Public health
the first Faroe Islands cohort showed that blood pres-
authority warnings about the consumption of pilot whale
sure tended to increase and heart rate variability tended
meat by pregnant women have reduced hair Hg levels sig-
to decrease when prenatal Hg exposure increased in the
nificantly, but the concentration for the majority of the
low-dose range (Sorensen et al., 1999). Alkylmercury poi-
Faroe Islands population is still above the 1.2 ตg/g limit.
soning is associated with increased blood pressure (H๖๖k
In 1986/87, 13% exceeded 10 ตg/g, but by 1998/99 this
et al., 1954) and children with Hg poisoning often have
had reduced to 3%. In other words, less than 3% now
increased heart rates and blood pressure (Warkany and
exceed the Benchmark Dose Limit of 12 ตg/g in hair.
Hubbard, 1953). Experimental evidence shows that MeHg
In order to understand and be able to mediate the
toxicity results in irreversible hypertension that remains
evaluation of risk, the implications of the results from
many months after the cessation of exposure (Wakita et
the research described in section 9.2.1.1. must be ex-
al., 1987). Although insufficient for risk assessment pur-
plained in generally understandable terms using the study
poses, this evidence suggests that the cardiovascular sys-
from the Faroe Islands as an example (Grandjean et al.,
tem should be considered a potential target for MeHg.
1997). In this respect, it is important to note that the
Even a slight negative impact on the cardiovascular sys-
children clinically do not differ from that expected. There
tem could be of greater public health relevance than a
are no children showing any physical signs of Hg poi-
slight impact on the central nervous system.
soning. Nevertheless, the Hg exposure is not negligible,
as illustrated by the calculations of delayed development
9.3.2. Persistent organic pollutants
using regression analyses. A more differentiated interpre-
tation could be obtained by including the regression coef-
Over the years, a number of biological guidelines have
ficient for known factors, e.g., age, which also influence
been issued by Health Canada through its Medical Ser-
the result. When the regression coefficients are compared,
vices Branch. In 1979, Health Canada issued a `Level of
it seems that a doubling of the Hg exposure corresponds
Concern' for PCBs in blood of 5 ตg/L for pregnant
to a delay in development of 1 to 2 months for most test
women, 20 ตg/L for the general population, and 100
results. Even if the regression coefficients are small, the in-
ตg/L as an `Action Level'. These guidelines are expressed
fluence is nevertheless considerable, when it is related to
as Aroclor 1254 concentrations, rather than Aroclor
the importance of age in a period where there is very fast
1260 concentrations. Because PCB concentrations de-
development. A supplementary interpretation can be based
rived from the AMAP datasets are expressed as Aroclor
on comparing intelligence quotients (IQ).
1260 equivalents (as calculated from CB138 and CB153
Chapter 9 ท The Effects of Arctic Pollution on Population Health
103
PCBs in blood, measured as Aroclor, % exceedance
100 Canada
Greenland
Iceland
Siberian Russia
Faroe
Norway
Sweden
Islands
90
80
70
60
50
40
30
20
10
0
y
๖
g
Inuit
Inuit
Inuit
Inuit
Inuit
Inuit
Inuit
NWT
NWT
NWTOther
NuukInuit
Baffin
Total
ๅ
Inuuvik
Kivalliq
Taymir Yamal
tis/Dene
Norilsk
Nunavik
Ilullissat
้
Olafsvik
Dudinka
Caucasian
Kitikmeot
M
Disko Bay
Reykjavik
Vestv
Salekhard
Indigenous Indigenous
Kiruna area
Arkhangelsk
Saudarkrokur
Non-indigenous
Non-indigenous
Non-indigenous
Non-indigenous
Ittoqqortoormiit
Ittoqqortoormiit
Vestmannaeyjar
% of samples 100 ug /L
Figure 9ท2. PCB levels in blood of indigenous and non-indigenous
Pregnant Inuit women
women of reproductive age. Percentage of samples exceeding public
Non-pregnant Inuit women
% of samples
5.0 ug /L and 100 ug /L
health guidelines for PCBs as Aroclor 1260 for Level of Concern
(LOC, >5 and < 100 ตg/L plasma) and Action Level (> 100 ตg/L
plasma).
levels), the Health Canada guidelines may not be di-
method, corresponds to concentrations of 2.5 mg/kg and
rectly applicable, however, they are adequate for the
2 mg/kg (or 10 ตg/L) respectively, using the capillary col-
purposes of the general comparisons made below.
umn method to determine PCBs (for an explanation of
The guideline proposed by Health Canada for PCBs
the calculation of the correction factor used to compare
in breast milk is 50 ตg/kg (whole milk) (Information let-
the two methods, see Rhainds et al., 1999). For interpre-
ter DD24 - March 1978). Assuming a 3.5% fat concen-
tation purposes the Health Canada guidelines are used.
tration in whole milk and an average lipid content of 10
In the Canadian Arctic, 43% of blood samples from
g/L in plasma, this guideline corresponds to a concentra-
Inuit women from NWT and Nunavut had blood PCBs
tion of 1.43 mg/kg in milk fat and a plasma concentra-
at a `Level of Concern' (i.e., above 5 ตg/L); of these,
tion of 14.3 ตg/L in pregnant women.
87% were less than 20 ตg/L, and none exceeded 100 ตg/L
It is also possible to use a guideline based on the
(Figure 9ท1). The extent of concentrations above the
dose/response assessment of Jacobson's data from the
5 ตg/L blood guideline varied widely, with higher per-
Michigan cohort (see section 9.2.1.2.) developed by
centages in Baffin (73%), Kivalliq (59%), and Nunavik
Tilson et al. (1990). The proposed NOAEL (no observed
(59%), where higher levels of PCBs were observed. The
adverse effects level) for visual recognition memory was
corresponding values for Dene/M้tis and Caucasians were
estimated at 1 mg/kg of PCBs in milk fat. By extension
3.2% and 0.7%, respectively (Van Oostdam et al., 1999).
this 1 mg/kg threshold concentration, expressed on a
Among women of reproductive age in the Greenland
lipid basis, can be applied to all tissues and biological
Disko Bay, Ilullissat, Nuuk, and Ittoqqortoormiit dis-
fluids, including plasma.
tricts, the >5 ตg/L `Level of Concern' for PCBs as Aro-
The analytical and quantification methods are cru-
clor 1254 was exceeded by 95%, 52%, 81% and 81%
cial in elaborating sound biological guidelines. For ex-
of the women, respectively (Deutch and Hansen, 2000).
ample, the 1 mg/kg lipid threshold for neurobehavioral
In Ittoqqortoormiit, 12% of pregnant women exceeded
effects identified in Jacobson's studies (Jacobson et al.,
the `Action Level' (100 ตg/L) for PCBs as Aroclor 1254,
1985, 1990a, 1992) corresponds to a 2 mg/kg lipid
relative to 52% of non-pregnant women. These marked-
NOAEL, using the quantification scheme developed in
ly higher proportions of the populations exceeding the
laboratories involved in the AMAP database. This
`Level of Concern' reflect the considerably higher PCB
threshold corresponds to the following plasma concen-
levels in Greenland Inuit.
trations (whole weight): umbilical cord, 5 ตg/L; preg-
The four Icelandic regions for which maternal blood
nant women, 20 ตg/L.
samples were analyzed for PCBs as Aroclor 1260 showed
Jacobson and Jacobson (1996) followed-up their as-
percentage exceedance of the `Level of Concern' (5 ตg/L)
sessment of the Michigan cohort at 11 years of age.
ranging from 22% to 50%, with Reykjavik having the
They reported a 1.25 mg/kg threshold PCB concentra-
highest value. The corresponding values for the Vestvๅg๘y
tion in lipids for full scale IQ, based on the prenatal ex-
region of Norway, Sweden and Finland were 70%, 68%,
posure assessment, and rounded this to 1 mg/kg for gen-
and 7.7% respectively. The higher percentages among the
eral use. This threshold concentration, derived using
Icelandic, Norwegian and Swedish women may be due
PCBs as measured by the packed column/Webb-McCall
to higher fish intakes and thus higher PCB levels.
104
AMAP Assessment 2002: Human Health in the Arctic
Lead in blood, % exceedance
20
Faroe
Alaska
Canada
Greenland
Finland
Siberian Russia
Islands
10
0
'ik
tis
upiat
NWT
NWT้
NWT
Inuit
Inuit
Inuit
Inuit
Inuit
NuukInuit
Other
BethelYup
/M
BaffinInuit
I๑
Inuuvik
Kivalliq
Norilsk
Taymir
Yamal
Nunavik
Ilullissat
Lapland
Dudinka
Kitikmeot
Caucasian Dene
Salekhard
Indigenous Indigenous
Barrow 2001
Ittoqqortoormiit
Ittoqqortoormiit
Non-indigenousNon-indigenousNon-indigenous
Pregnant Inuit women
Figure 9ท3. Lead levels in blood of indigenous and non-indigenous
Non-pregnant Inuit women
women of reproductive age. Percentage of samples exceeding the
blood guideline Action Level of 100 ตg/L.
saturated fatty acids especially DHA are essential for
9.3.3. Lead
brain development (Crawford et al., 1976). DHA defi-
Blood guidelines have also been developed for Pb. Figure
ciency impairs learning and memory in rats (Greiner et
9ท3 shows that levels of exceedance of the guideline
al., 1999). Studies have shown that n-3 polyunsaturated
value of 100 ตg/L are low throughout the Arctic. The low
fatty acid supplements can enhance visual acuity and
levels of exceedance found in some populations in Can-
brain development in preterm infants (Bjerve et al.,
ada and Greenland are thought to be related to local Pb
1992; Uauy et al., 1990), but it is not clear whether in-
contamination from the lead shot used by hunters, rather
creased levels during the fetal period can protect full-
than environmental Pb from long-range transport.
term infants against neurotoxicity associated with pre-
natal exposure to environmental contaminants.
Many health endpoints are multi-factorial and envi-
9.4. Combined effects
ronmental stressors contribute to a various extent to the
9.4.1. Mixtures
etiology of these diseases. Compared to the importance
Arctic residents are exposed to a variety of contaminants
of lifestyle and genetic factors in the etiology of most
present in the food chain. POPs are composed of numer-
diseases, contaminants are likely to play a modest role.
ous compounds; most of which are capable of accumu-
However, exposure to many persistent organic contami-
lating both in the food chain and in humans. Similarly,
nants is preventable, and their presence in the remote
concomitant exposure to MeHg and POPs is often ob-
Arctic is unethical.
served. It is thus difficult to determine which compound
In risk assessment based on epidemiological studies
is responsible for any particular observed effect. Any
there are many confounding factors that should be con-
single substance risk assessment is therefore of limited
sidered. Epidemiological studies on children should in-
relevance for regulatory purposes.
clude information on, at least: maternal and paternal
In the Faroe Islands, prenatal exposure to PCBs was
age at childbirth, parity, smoking, education, mothers
examined by analyses of cord tissue from 435 children
and if possible fathers intelligence, employment status,
from a birth cohort established in 1986/87 (Grandjean
migration, risk factors in past medical history, weight,
et al., 1992, 1997, 2001). Among 17 neuropsychological
height, breast feeding, age at examination, gestational
outcomes determined at age 7 years, the cord PCB con-
age, number and age of siblings, day care, home envi-
centration was associated with deficits on the Boston
ronment, and type of delivery.
Naming Test, the Continuous Performance Test reaction
time, and possibly, on long-term recall on the California
9.5. Conclusions
Verbal Learning Test. While no PCB effects were appar-
ent in children with low Hg exposure, PCB-associated
Except for Hg- and OC-induced neurodevelopmental ef-
deficits within the highest tertile of Hg exposure indicated
fects in the Faroe Islands, POPs and links with the im-
a possible interaction between the two neurotoxins.
mune system in Nunavik (Canada), and links between
PCB-associated increased thresholds were seen at two of
pregnancy outcome and metals on the Kola Peninsula,
eight frequencies in audiometric tests. No deficits oc-
very few major environmental epidemiological studies
curred on evoked potentials or contrast sensitivity. The
have been conducted in the Arctic. This is because Arctic
limited PCB-related neurotoxicity group in this cohort
studies are extremely difficult to undertake, owing to the
appears to be affected by concomitant MeHg exposure.
remoteness of communities, the cultural context, cli-
The neurotoxic effects of MeHg may be attenuated
matic factors, small population sizes, and confounding
by protective effects of Se and n-3 polyunsaturated fatty
social and behavioral factors etc. The special character-
acids. Increased intake of Se and n-3 fatty acids would
istics of the Arctic raises the question as to how far re-
be expected in peoples such as the Inuit who consume
sults and conclusions from epidemiological studies con-
relatively large quantities of fish and marine mammals.
ducted outside the Arctic can apply to this region. Mix-
Although the protective effects of Se on MeHg toxicity
tures of contaminants are also different. Due to the
have not been adequately documented in humans (NRC,
properties of contaminants that lead to their distribution
2000), there is strong evidence from animal studies that
throughout the Arctic, and their concentration in certain
Se can influence the deposition of MeHg in the body,
food webs, the exposure profiles for the Arctic are likely
and some evidence that Se can protect against Hg toxic-
to differ from those reported at mid-latitudes where
ity (Ganther et al., 1972; Whanger, 1992). n-3 polyun-
local sources are more likely to contribute to the mix-
Chapter 9 ท The Effects of Arctic Pollution on Population Health
105
ture. Patterns of exposure in the Artic can be influenced
and outcome, as well as in statistical power, must be
by hunting and fishing seasons, and resulting occasional
taken into account. For marine food-mediated exposure,
high exposures, as opposed to constant exposure, may
confounding influences of, for example, n-3 polyunsatu-
have different toxic consequences. Arctic residents con-
rated fatty acids must be considered, because these nu-
sume wild animals and plants. This country food con-
trients are also essential for the development of the nerv-
tains specific components, which may influence or coun-
ous system. Thus, birth weight and fatty acid status are
teract the toxicity of contaminants. For example, Inuit
important cofactors for consideration.
are exposed to similar amounts of Hg as people from the
Whenever possible, risk assessment of contaminants
Faroe Islands, but their Se intake is much higher. Finally,
should be based on epidemiological evidence, however
indigenous peoples may have specific genetic backgrounds
Arctic epidemiological studies are few in number, and
that influence their susceptibility to toxic agents.
are difficult to perform. Serious consideration should be
The highest proportions of exceedance of blood
given to the cohort study on neurological disorders asso-
guidelines for contaminants parallel the patterns for the
ciated with prenatal MeHg (in the Faroe Islands) and the
concentrations of contaminants in blood. The Inuit from
study of immune dysfunction in children exposed pre-
the east coast of Greenland, who consume large num-
natally to POPs (in Nunavik). As human exposure to
bers of marine mammals, have the highest proportion of
contaminants is to a mixture of many different sub-
blood concentrations exceeding the Canadian PCB guide-
stances simultaneously, as indicated by the exposure data
lines, thereafter followed by west coast Greenland Inuit
in this report, it is not appropriate and currently not
populations and Inuit populations from the Baffin and
even possible to establish the risk from single substances
Nunavik regions of eastern Canada. A similar pattern
using epidemiological studies. The Faroe Islands study
occurs for exceedance of the blood Hg guidelines (used
shows there are negative effects related to Hg and PCBs
by Canada and the United States), but the data are more
and perhaps DDT, DDE and other OCs. Similar expo-
limited. When the new U.S. EPA mercury guidelines are
sure levels have been found in other areas of the Arctic,
applied it is evident that most Inuit populations and a
e.g., Greenland. It is likely that the negative effects, al-
significant proportion of several other populations ex-
though small in the Faroe Islands, can be found at other
ceed these guidelines. Lead levels are also elevated among
places with similar exposures. It is the future responsi-
some Inuit groups in Arctic Canada and Greenland and
bility of the public health authorities to decide on suit-
these are also reflected in the increased proportions ex-
able undertakings to reduce the human exposure levels.
ceeding the guidelines for Pb concentration in blood.
Consideration should also be given to the possible nega-
Most of the Pb contamination in these communities is a
tive effects on public health that could be caused by
local problem associated with lead shot used for hunting
changes in lifestyle.
game, as opposed to Pb from long-range atmospheric
In the risk assessment of exposure at levels presently
transport. Northern Europeans from Norway, Sweden,
found in the Arctic, it is reasonable to conclude that the
the Faroe Islands, and Iceland have higher levels of PCBs
traditional diet in the Arctic contains xenobiotic sub-
and markedly higher proportions exceeding the PCB
stances which have a negative influence on health.
blood guideline than Caucasians from Arctic Canada.
Exposures in many risk assessment studies are not
9.6. Recommendations
`pure' and most include more than one neurotoxin. The
Faroe Islands studies offer some potential for separating
The following recommendations are made in order to
the effects of different contaminants because PCBs and
improve the understanding of health effects associated
Hg showed only moderate association and because Pb
with contaminant exposure in the Arctic.
exposures are very low (Grandjean et al., 1992). How-
1. Circumpolar epidemiological studies should be im-
ever, the most serious problem in this regard is that envi-
plemented.
ronmental PCBs do not comprise one well-defined chem-
2. MeHg- and POPs-related effects are still the key is-
ical but consist of 209 congeners. Several of these are
sues. However the role of new contaminants (poly-
considered neurotoxic (Sauer et al., 1994), but few are
brominated diphenylethers, polychlorinated naph-
included in routine analyses, and as a result are not in-
thalenes, etc.) should be investigated.
cluded in risk assessments of persistent congeners. Fur-
3. For exposure assessment, epidemiological studies
thermore, PCBs occur in conjunction with other OCs,
should consider mixtures and confounding factors
such as p,p'-DDE, which may contribute to their com-
such as the interaction of nutritional components.
bined toxicity. The PCB exposure estimate may not ad-
4. Epidemiological studies on the nutritional benefits of
dress differences in PCB profiles and other contaminant
traditional food should be included in the risk assess-
profiles in different settings, and comparisons between
ment profile.
epidemiological studies must therefore be performed
5. Tissue banking for samples collected in the course of
with caution.
epidemiological studies should be carefully organized
Among the reasons for different study outcomes are
to allow subsequent assessment of new contaminants
differences in concomitant exposures and nutritional
and time trend studies.
factors. In addition, imprecision in exposure assessment