atherosclerosis in coronary arteries and aorta among greenlanders: an autopsy study
TRANSCRIPT
Atherosclerosis in coronary arteries and aorta among Greenlanders:an autopsy study
Henning Sloth Pedersen a,*, Gert Mulvad a, William P. Newman, III b,Donald A. Boudreau b
a The Primary Health Care Clinic, P.O. Box 1001, DK-3900 Nuuk, Greenlandb Department of Pathology, Louisiana State University Health Sciences Center, New Orleans, LA, USA
Received 21 December 2002; received in revised form 9 June 2003; accepted 13 June 2003
Atherosclerosis 170 (2003) 93�/103
www.elsevier.com/locate/atherosclerosis
Abstract
In a cross-sectional autopsy study of 107 Inuit in Greenland, the extent of arterial surface involvement with atherosclerosis was
evaluated in the presence of known or estimated environmental risk factors for coronary heart disease (CHD): age, gender, obesity,
serum lipids, smoking, and hypertension. Mean, median, and range values for all of the risk factor variables and for the extent of
atherosclerosis in the thoracic aorta, abdominal aorta, right coronary artery, and left anterior descending coronary artery are
reported by age strata, along with the results of covariant analysis of the dependence of the extent of atherosclerosis upon the risk
factors. No significant differences between females and males were found in either the risk factors or prevalence and extent of
atherosclerosis in the aorta and in the coronary arteries. It appears that the extent of advanced atherosclerotic lesions in
Greenlanders appears to be the same as that previously reported in a similar study in Alaska Natives.
# 2003 Elsevier Ireland Ltd. All rights reserved.
Keywords: Inuit; Greenland; Atherosclerosis; Risk factors; Coronary heart disease
1. Introduction
Low mortality from ischaemic heart disease (IHD)
has been reported among Inuit [1�/3]. The ratios
between age standardized mortality rates of IHD in
Greenland Inuit (1968�/1985) and Denmark (1980) was
0.5 (95% CI: 0.4�/0.6) for males and 0.9 (95% CI: 0.7�/
1.1) for females [2]. In Alaska, the corresponding ratios
for natives versus non-natives were 0.59 and 0.61 (1979�/
1988) [4]; however, more recent reports indicated that
the death rate of IHD among Alaska Eskimo may now
exceed that of non-native Alaskans [5]. Canadian Inuit
were also reported to have a substantially lower rate of
death from IHD than the rest of the population [6].
Caution has been recommended in the evaluation of
these findings since they are based upon reviews of death
certificates [7]; however, the underlying suggestion of
reduced risk of IHD mortality in these populations has
been intriguing.
Among the various possible explanations suggested
by investigators for low IHD mortality reported among
Inuit is that this population group had lower prevalence
and/or less extent of atherosclerotic lesions in the
coronary arteries. However, studies have suggested
that there is no difference in the degree of atherosclerosis
among Greenland natives and Danes, as interpreted by
ultrasonographic investigation of the carotid and fe-
moral arteries [8] and by X-ray examinations of the
lumbar spine to diagnose arteriosclerotic complications
of the abdominal aorta [9]. This is in accordance with a
recent study of cardiovascular risk factors in Inuit of
Greenland which showed the prevalence of heavy
smoking, obesity, blood pressure and total serum
cholesterol are similar in Greenland and Denmark,
although it was reported that the low density lipoprotein
cholesterol (LDL-C) and triglyceride concentrations
were lower in Greenlanders [10].
It is generally felt that for studies of diseases such as
IHD, autopsy results provide the only conclusive* Corresponding author. Tel.: �/299-344-424; fax: �/299-344-425.
E-mail address: [email protected] (H.S. Pedersen).
0021-9150/03/$ - see front matter # 2003 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/S0021-9150(03)00240-5
evidence of outcome. However, only three autopsy
studies among Inuit have been reported: two of these
[11,12] were inconclusive, but suggested that athero-
sclerosis among Inuit in Alaska was not uncommon; thethird was a systematic evaluation of the prevalence and
extent of atherosclerotic lesions in the coronary arteries
and aorta in Alaska natives compared with Alaska non-
natives. Indeed, findings in the latter were that of a
substantially lower extent of raised lesions in the native
population [13,14]. The purpose of this report is to
describe the findings of an autopsy study in the Green-
landic population in regard to environmental IHD riskfactors and the extent of atherosclerosis in the aorta and
coronary arteries.
2. Materials and methods
During the period of the study (1990�/1994) the
population of Greenland natives, (‘‘Greenlanders’’),
was approximately 55 000, of which most (82%) livedin Greenland and the rest mainly in Denmark. As in
other epidemiological studies, a Greenlander was de-
fined as a person born in Greenland. Over the years,
some genetic admixture with Caucasians has taken
place; however, until 1960, the non-Greenlander popu-
lation was less than 5%, increasing to a level of 18% in
the 1980s, and then declining to 12% in 1996 [15]. The
Greenlanders today constitute the largest geographicalconcentration of a total of 167 000 Circumpolar Inuit.
The study cases included Greenlanders ]/18 years of
age and were obtained primarily from the capital city,
Nuuk, and from Ilulissat, a small community northwest
of Nuuk. Since the central Greenlandic hospital is in
Nuuk, some cases from other areas of the country were
also included. During the data collection period (Nuuk:
1990�/1994; Ilulissat: 1991�/1994), a total of 342 deathsoccurred in the study areas, for which 107 autopsies
(31%) were performed and included in the study. This
sample constitutes 26.4% of the total deaths in Green-
land among adult Inuit during the study period. As
there were no trained pathologists in Greenland, two
clinicians in Nuuk and one in Ilulissat received special
training in standardized autopsy and specimen collec-
tion procedures at Gentofte Hospital in Denmark andLouisiana State University Health Science Center
(LSUHSC), New Orleans, LA, USA. To minimize
selection bias, cases were excluded only because: (a)
relatives refused or were unable to be contacted; (b) a
prosector was not available; or (c) contagion risk was
high.
Height, from the vertex of the cranium to the base of
the heel, and weight were both determined prior toautopsy. Body mass index (BMI) was calculated as the
weight in kg divided by the square of the height in
meters. Blood was collected by syringe and needle from
the vena cava, heart, or aorta; separated by centrifuga-
tion into serum and cells; and frozen. The aorta and the
coronary arteries were systematically collected accord-
ing to protocol. The aorta, removed 2 cm proximal tothe ligamentum arteriosum and 2 cm distal to the iliac
bifurcation, was opened longitudinally on the dorsal
surface midway between the openings of the intercostal
and lumbar arteries. The opened aorta was flattened
with the adventitial surface down and bisected along a
line on the ventral surface midway between the celiac,
superior mesenteric, and inferior mesenteric ostia. The
right half was frozen for chemical analyses and the lefthalf fixed in buffered neutral formalin for shipping and
subsequent staining. The coronary arteries were opened
longitudinally and fixed in the same manner as the aorta
(all segments remained in formalin for periods of days
until shipping). The right coronary artery was removed
from its point of origin in the right anterior sinus of
Valsalva to the point where it passed down the posterior
interventricular sulcus. The left anterior coronary arterywas removed from the origin at the left main coronary at
its bifurcation downward along the anterior interven-
tricular sulcus to the apex. The arteries were shipped to
the Department of Pathology, LSUHSC, where they
were stained with Sudan IV and sealed in flat, poly-
propylene pouch bags with 10% buffered formalin.
Specimens, identified only by code numbers, were
processed in batches containing specimens from otherpopulations so that graders had no knowledge of age,
sex, cause of death, or ethnic group. For lesion grading,
the aorta was divided in two sections: the thoracic
section, from a horizontal line through the first paired
intercostals to a horizontal line through the upper edge
of the orifice of the celiac artery; and the abdominal
section, extending from the same line through the upper
edge of the celiac orifice, to a horizontal line drawnthrough the vertex of the iliac bifurcation.
Visual grading of the lesions was done independently
by three pathologists at LSUHSC, specially trained in
using methods developed for the International Athero-
sclerosis Project [16]. Lesions were evaluated by visual
inspection and palpation of the specimens in the sealed
pouch bags [17]. The thoracic aorta and the abdominal
aorta segments were graded separately. Recorded foreach segment was the estimated total percentage of the
intimal surface involved with all types of atherosclerotic
lesions and the proportion contributed by each of the
four types of grossly distinguishable lesions: flat or
slightly elevated intimal lesions that stained distinctly
with Sudan IV were designated as fatty streaks; fibrous
plaques were firm, elevated, pale gray, glistening intimal
lesions that sometimes had sudanophilic deposits super-imposed on their intimal surfaces; complicated lesions
were those in which there was grossly visible hemor-
rhage, ulceration, or thrombus, with or without calcium
deposits; and calcified lesions were areas in which
H.S. Pedersen et al. / Atherosclerosis 170 (2003) 93�/10394
calcium was detectable either visually or by palpation,
without the above attributes of a complicated lesion.
The term raised lesion was defined to be the sum of the
extent of fibrous plaques, complicated lesions andcalcified lesions in a given arterial segment.
Total serum cholesterol levels were determined using
the cholesterol-oxidase method (Gilford Systems Ciba-
Corning, Oberlin, OH) and the Gilford SBA 300
spectrophotometer. The high density lipoprotein cho-
lesterol (HDL-C) concentration was determined after
precipitation of serum by heparin MnCl2 (Bio-Rad ECS
Division, Anaheim, CA). The cholesterol assay wasstandardized with reference material obtained from the
College of American Pathologists, Northfield, IL. Con-
centration of the very low density lipoprotein�/low
density lipoprotein cholesterol (VLDL�/LDL-C) was
calculated as the difference between total serum choles-
terol and HDL-C. Thiocyanate, a marker for smoking,
was analyzed using a method [18] adapted for the
Gilford SBA 300 that allowed detection of athiocyanate�/ferric nitrate complex by measuring the
amount of color produced when that complex is formed.
A threshold thiocyanate level of 90 mmol/l for classifying
a subject as a smoker was established using unpublished
observations collected during one of the studies by the
PDAY research group [19]. It should be noted that the
threshold was established in a US population study and,
therefore, may not be precisely applicable in othergroups.
Kidney samples were saved in 10% buffered formalin
and processed for examination by light microscopy.
Under the 10�/ objective, the outer diameters of the
media of the cortical renal arteries were measured.
Those with outer diameters of 80�/300 mm were eval-
uated. The thicknesses of the arteries’ intima were
measured under the 40�/ objective. These measure-ments were combined to calculate the renal measure-
ment of hypertension (RMI). The RMI is defined as
being equivalent to intima thickness/outer media dia-
meter. Using age and the RMI in a published equation,
an algorithm has been established that estimates mean
blood pressure (MBP: systolic blood pressure�/2�/
diastolic blood pressure, all divided by 3). MBP less
than 110 mmHg is considered normotensive and MBPequal to or greater than 110 mmHg is considered
hypertensive [20].
Statistical analyses, using SAS version 8.2 (SAS
Institute, Inc., Cary, NC), included general descriptive
statistics, the chi-square or Fisher’s exact test for
distribution comparisons, Student’s t-test for gender
differences, and the general linear models (GLM)
procedure for covariance. To minimize the error ofpossible false positives in grading minimal lesion pre-
sence, lesion prevalence estimates were based on cases
with 5% or more of extent of arterial surface involve-
ment [19]. Where appropriate to better approximate
normal distributions, the lesion data were log-trans-
formed after addition of 0.01 to each data point to avoid
zeros. It is recognized that pre- and post-mortem events,
especially in hospitalized patients, may affect serumanalytes; therefore, to minimize bias due to possible
hemodilution effects upon analyte concentrations, the
blood analytes (lipids and thiocyanate) in seven cases
(two female; five male) for which the value of total
serum cholesterol was less than 100 mg/dl were excluded
from analyses [21].
3. Results
Of the original 107 autopsy cases, 102 were available
for analyses after five were excluded: two because of
severity of burns that affected the organs of interest; two
because the bodies were found more that 3 months after
death; and one in which cardiac surgery prevented the
use of the arteries. The autopsies were all performed nolater than 48 h after death, and all the bodies except one
were without putrefication; in one case we observed
putrefication caused by the bacteria Clostridium perfrin-
gens . Also, in two cases the right coronary artery, and in
one case the left anterior descending artery, were lost in
storage, handling, or transport. Cases were classified
into one of three broad categories of cause-of-death:
coronary heart disease (CHD); CHD-related (knownhypertension, diabetes, and/or chronic renal disease); or
basal (non-CHD, non-CHD-related) and Table 1 shows
the distribution of cases among these categories. Among
the 61 males 20 had a violent death (nine accidents, eight
suicides, three homicides) and among the 46 females ten
had a violent death (two accidents, one suicide and
seven homicides). As reported in many of the past
studies and cited in the references, atherosclerotic lesionsmay begin in early youth as fatty streaks that potentially
progress to raised lesions in early adulthood, and may
then become complicated lesions in later years. Usually,
by age 60, the factors of influence in the earlier years of
life have set the stage for subsequent arterial pathology
during the remaining years. While it would be desirable
to stratify the data in this study by decade, the lack of a
sufficiently large number of cases prohibits doing this.Therefore, to best illustrate the progression of athero-
sclerosis in the Greenlanders, the cases were partitioned
into intervals of early youth (age B/24 years), early
adulthood (25�/34 years), mid-life (35�/59 years), and
late life (60�/89). Basal cases represent the category for
which there is the least potential for autopsy selection
bias due to prevalence of CHD, a concern whenever
selection includes hospitalized patients [22,23]. As canbe seen, most (72%) of the cases are basal and there was
no significant difference in the distribution of cause-of-
death by gender. An issue of concern, discussed later, is
H.S. Pedersen et al. / Atherosclerosis 170 (2003) 93�/103 95
that 56% (57/102) of the study cases were of age]/60
years.
Summary descriptive statistics for all cases and also
stratified by sex, are shown in Table 2a for age, blood
analytes (total cholesterol, LDL�/VLDL-C, HDL-C,
thiocyanate), BMI, and estimated MBP. None of the
variables were found to differ significantly between
females and males. Of note are the mean and median
values for the serum lipids which are relatively high in
comparison to the currently recommended guidelines
[24]. In Table 2b, the distribution of risk factor variables
in the sample are described in terms of their respective
Table 1
Greenland autopsy study: number of cases by broad cause-of-death classification and age
Cause of death category Age (years)
B/24 25�/34 35�/59 60�/89 Totals
M F M F M F M F
CHD 0 0 0 0 0 0 1 5 6 (6%)
CHD related 0 0 0 0 2 3 10 8 23 (23%)
Basal (non-CHD; non-CHD-related) 4 0 4 3 18 11 18 15 73 (72%)
Totals 4 0 4 3 20 14 29 28 102
M, males; F, females.
Table 2
Greenland autopsy study
Variable N Mean S.E.M. Median Minimum Maximum P
(a ) Descriptive statistics for risk factors associated with atherosclerosis and CHD
Age (years) 102 59.38 1.77 63.00 19.00 89.00
F 45 61.27 2.49 64.00 25.00 85.00
M 57 57.89 2.48 60.00 19.00 89.00 0.3466
BMI (kg/m2) 102 23.74 0.48 23.10 13.30 37.50
F 45 23.41 0.82 22.40 13.30 37.00
M 57 24.01 0.58 23.70 16.00 37.50 0.5446
Est. MBP (mmHg) 87 106.67 1.15 107.23 80.92 126.14
F 37 107.68 1.74 106.09 86.05 122.47
M 50 105.92 1.54 107.56 80.92 126.14 0.4546
Thiocyanate (mmol/l) 78 67.94 3.62 62.75 22.50 164.00
F 38 69.74 4.89 66.25 27.00 153.50
M 40 67.19 5.37 61.25 22.50 164.00 0.8322
Total serum cholesterol
(mg/dl)
83 223.37 8.30 217.00 100.50 398.50
F 39 224.28 11.41 217.00 115.50 391.00
M 44 222.57 12.07 217.75 100.50 398.50 0.9186
HDL-C (mg/dl) 54 51.29 4.41 41.25 9.50 146.00
F 26 55.10 6.28 52.25 9.50 146.00
M 28 47.75 6.21 39.75 13.50 140.00 0.4099
LDL�/VLDL-C (mg/dl) 54 167.95 9.76 162.50 37.50 329.50
F 26 161.12 13.03 162.75 43.50 315.00
M 28 174.30 14.55 162.50 37.50 329.50 0.5048
Risk factors All cases Females Males P
(b ) Comparison of CHD risk factors by gender
Obesity (BMI]/30) 13.7% (14/102) 20.0% (9/45) 8.8% (5/57) 0.1472
Smoking (thiocyanate]/90
g/l)
16.9% (14/83) 15.4% (6/39) 18.2% (8/44) 1.0000
High LDL�/VLDL-C (]/
130 mg/dl)
68.5% (37/54) 69.2% (18/26) 67.9% (19/28) 1.0000
Low HDL-C (B/40 mg/dl) 48.2% (26/54) 46.2% (12/26) 50.0% (14/28) 0.7926
Hypertension (estimated
MBP]/110)
39.1% (34/87) 40.5% (15/37) 38.0% (19/50) 0.8277
MBP is estimated from renal artery morphology and thiocyanate is used as a marker for smoking. Probabilities are for no difference between
females (F) and males (M). Values are percentages (ratios) of gender groups for each variable; for some, data were not available for all cases.
H.S. Pedersen et al. / Atherosclerosis 170 (2003) 93�/10396
high risk categories. The high serum lipid levels seen in
Table 2a are manifest here in the form of large
proportions of the group (68.5% with high LDL�/
VLDL-C and with 48.2% low HDL-C) having high
risk as marked by these; however, the risk measure of
obesity appears to be low (20% in females; 8.8% in
males). The estimated proportion of smokers
(thiocyanate]/90 mg/l) is 15.4% for females and 18.2%
for males, while the prevalence of estimated hyperten-
sion is 40.5% in females and 38.0% in males.
Prevalence of atherosclerotic fatty streaks and raised
lesions, assessed as the percent of cases with 5% or more
arterial surface involvement, is shown in Table 3 for all
cases and for basal cases. Ideally, these data should be
stratified not only by gender but also by appropriate age
ranges; however, as can be seen in Table 1 there are very
few (n�/11) cases below age 35 years, therefore, strata
cell counts would be too low for meaningful compar-
isons. Indeed, the fact that slightly more than half of the
cases are 60 years or older suggests that age may well be
a contributing factor in the observation that, with the
single exception of the difference for fatty streaks in the
abdominal aorta for all cases (which vanishes for the
basal cases), there are no significant differences in
prevalence between females and males. In an attempt
to determine whether age is influencing the lack of an
expected gender difference, comparisons were per-
formed within the younger and older age groupings of
ageB/40 years (n�/17) and age]/40 years (n�/85), the
generally accepted age-threshold of increased risk for
CHD in males. In each case the same conclusion of no
prevalence difference was found. For these reasons, the
prevalence data are presented as shown in Table 3
without age stratification.
The extent of arterial surface involved with athero-
sclerotic lesions in the four arterial segments evaluated
in the study is described by the summary statistics in
Table 4a, where results are presented for all cases, for
males, and for females. As a group, the distributions of
lesion extent among the segments appear to follow a
pattern similar to that reported by autopsy studies in
Blacks and Whites in New Orleans [25] and in Alaska
natives and non-natives [14], with the greatest extent
found in the abdominal aorta, followed in descending
order by the thoracic aorta, left anterior descending
coronary artery, and right coronary artery. Although
females in Greenland do seem to have greater extent of
raised lesion involvement than males in both the
abdominal (60.6 vs. 46.6%) and thoracic (39.6 vs.
28.5%) aorta, this difference is not statistically signifi-
cant and does not appear to follow a consistent pattern
among the other segments for either fatty streaks or
raised lesions. Since there was not a clear gender
difference in the extent of atherosclerosis, the cases
Table 3
Greenland autopsy study: estimated prevalence of atherosclerosis in the aorta and coronary arteries*/all cases and basal cases
Variable Basal cases All cases
n % Positive P n % Positive P
Abdominal aorta, FS 73 78.1 102 72.5
F 29 62.1 45 64.4
M 44 88.6 0.0100 57 79.0 0.1214
Thoracic aorta, FS 73 91.8 102 92.2
F 29 96.6 45 93.3
M 44 88.6 0.3920 57 91.2 1.0000
Left anterior descending coronary artery, FS 72 36.1 101 32.7
F 28 32.1 44 27.3
M 44 38.6 0.6229 57 36.8 0.3932
Right coronary artery, FS 72 48.6 100 44.0
F 29 48.3 44 45.5
M 43 48.8 1.0000 56 42.9 0.8410
Abdominal aorta, RL 73 80.2 102 86.3
F 29 86.2 45 91.1
M 44 77.3 0.3828 57 82.5 0.2554
Thoracic aorta, RL 73 69.9 102 77.5
F 29 69.0 45 77.8
M 44 70.5 1.0000 57 77.2 1.0000
Left anterior descending coronary artery, RL 72 65.3 101 73.3
F 28 67.9 44 75.0
M 44 63.6 0.8023 57 71.9 0.8222
Right coronary artery, RL 72 66.7 100 71.0
F 29 65.5 44 72.7
M 43 67.4 1.0000 56 69.6 0.8528
Values are case totals and percent of cases that are positive for lesions (]/5% extent involvement). Probabilities are for no difference between
females (F) and males (M). FS, fatty streaks; RL, raised lesions.
H.S. Pedersen et al. / Atherosclerosis 170 (2003) 93�/103 97
Table 4
Greenland autopsy study
Variable n Mean S.E.M. Median Minimum Maximum P
(a ) Descriptive statistics for extent of aortic and coronary artery atherosclerosis in all cases
Abdominal
aorta, FS
102 14.70 1.45 10.20 0.00 68.40
F 45 15.22 2.47 8.80 0.00 68.40
M 57 14.29 1.74 11.00 1.50 63.30 0.3232
Thoracic
aorta, FS
102 14.67 1.00 10.70 1.60 50.20
F 45 14.70 1.25 12.70 1.60 39.30
M 57 14.64 1.51 10.40 1.60 50.20 0.4516
Left anterior
descending
coronary ar-
tery, FS
101 4.96 0.65 2.70 0.00 37.80
F 44 3.70 0.64 2.20 0.00 19.80
M 57 5.94 1.03 2.90 0.00 37.80 0.2271
Right cor-
onary ar-
tery, FS
100 6.78 0.86 3.90 0.00 41.80
F 44 6.48 1.16 3.65 0.00 30.80
M 56 7.01 1.24 4.05 0.00 41.80 0.7307
Abdominal
aorta, RL
102 52.76 3.37 63.95 0.00 97.70
F 45 60.55 5.26 76.80 0.00 97.70
M 57 46.62 4.24 50.30 0.00 96.10 0.1698
Thoracic
aorta, RL
102 33.39 2.75 31.55 0.00 96.10
F 45 39.63 4.40 37.50 0.00 96.10
M 57 28.46 3.37 22.60 0.00 96.10 0.3883
Left anterior
descending
coronary ar-
tery, RL
101 34.83 3.36 19.50 0.00 97.70
F 44 33.87 5.24 17.65 0.00 97.70
M 57 35.58 4.41 32.30 0.00 94.80 0.8860
Right cor-
onary ar-
tery, RL
100 34.54 3.36 31.55 0.00 97.70
F 44 35.29 5.23 25.55 0.00 97.70
M 56 33.96 4.41 31.55 0.00 97.70 0.8580
BMI Estimated MBP HDL-C (mg/dl) LDL�/VLDL-C (mg/dl) Estimated smoker
Age n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
n % yes
(b ) CHD quantitative risk factors and extent of atherosclerosis
H.S
.P
edersen
eta
l./
Ath
erosclero
sis1
70
(2
00
3)
93�
/10
39
8
Table 4 (Continued )
Variable n Mean S.E.M. Median Minimum Maximum P
CHD quantitative risk factors
B/24 4 24.6 (0.7) 24.1
(23.6,26.7)
4 83.0 (1.0) 83.2
(80.9,84.8)
3 51.7 (6.4) 51.0
(41.0,63.0)
3 227.3 (3.6) 229.0
(220.5,232.-
5)
4 50.0
25�/34 7 23.3 (1.2) 22.6
(18.6,27.4)
6 90.4 (1.8) 89.2
(86.1,98.1)
3 38.7 (6.3) 44.5
(26.0,45.5) 3
3 163.3 (55.8) 126.5
(90.5,273.0)
5 40.0
35�/59 34 24.1 (0.7) 23.1
(16.5,35.8)
27 104.0 (1.3) 103.9 (92.1,
126.1)
19 62.5 (8.7) 59.0
(12.5,146.0)
19 166.1 (16.9) 163.0
(57.5,313.0)
28 25.0
60�/99 57 23.6 (0.7) 22.9
(13.3,37.5)
50 112.0 (1.1) 113.6
(94.5,123.3)
36 40.6 (4.9) 33.0
(2.0,140.0)
36 143.6 (12.8) 137.0
(31.5,329.5)
46 6.5
Abdominal aorta Thoracic aorta Left descending coronary artery Right coronary artery
Age n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
Atherosclerosis-fatty streaks
B/24 4 31.9 (13.4) 27.5
(9.3,63.3)
4 29.7 (5.1) 27.2
(20.3,44.3)
4 2.6 (1.5) 1.7 (0.0,7.0) 3 2.3 (2.3) 0.0 (0.0,6.8)
25�/34 7 30.4 (7.0) 31.7
(6.1,58.3)
7 14.0 (2.2) 12.0
(8.7,25.0)
7 1.2 (0.7) 0.3 (0.0,4.7) 7 0.5 (0.3) 0.0 (0.0,1.7)
35�/59 34 20.8 (2.8) 16.2
(2.0,68.4)
34 14.7 (1.9) 10.2
(2.7,44.3)
33 6.4 (1.3) 3.3
(0.0,29.8)
34 6.1 (1.4) 3.4
(0.0,41.8)
60�/99 57 8.0 (0.8) 6.8
(0.0,22.8)
57 13.7 (1.3) 10.7
(1.6,50.2)
57 4.8 (0.9) 2.2
(0.0,37.8)
56 8.3 (1.2) 5.7
(0.0,34.5)
Abdominal aorta Thoracic aorta Left descending coronary artery Right coronary artery
Age n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
n Mean
(S.E.M.)
Median
(Mn, Mx)
Atherosclerosis-raised lesions
B/24 4 0.0 (0.0) 0.0 (0.0,0.0) 4 0.4 (0.4) 0.0 (0.0,1.7) 4 0.9 (0.8) 0.2 (0.0,3.3) 3 0.8 (0.8) 0.0 (0.0,2.5)
25�/34 7 0.5 (0.5) 0.0 (0.0,3.5) 7 0.0 (0.0) 0.0 (0.0,0.0) 7 0.8 (0.5) 0.0 (0.0,3.3) 7 0.7 (0.5) 0.0 (0.0,3.3)
35�/59 34 32.2 (4.61) 22.4
(0.0,87.1)
34 13.7 (2.3) 8.7
(0.0,46.4)
33 22.6 (4.5) 8.5
(0.0,81.0)
34 20.0 (4.0) 12.1
(0.0,94.7)
60�/99 57 75.2 (2.5) 78.5
(22.2,97.7)
57 51.6 (2.9) 55.0
(8.2,96.1)
57 48.5 (4.5) 43.1
(0.0,97.7)
56 49.4 (4.5) 48.6
(0.0,97.7)
F, females; M, males. Lesions values are % total vessel surface involved; probabilities are for no differences between females and males. FS, fatty streaks; RL, raised lesions. Values, shown by age
level, are mean (S.E.M.) and median (minimum, maximum). Values for atherosclerotic lesion extent are percent of total surface area involved.
H.S
.P
edersen
eta
l./
Ath
erosclero
sis1
70
(2
00
3)
93�
/10
39
9
were pooled and analyzed by age levels ageB/24, 25�/34,
35�/59 and �/60 years. The age-stratified results are
shown in Table 4b for both environmental risk factor
variables and lesion extent. For the risk factors, no
apparent change with age is seen for the BMI or the
HDL-C, while the mean and median estimated MBP
appears to increase (83.0�/112.0 and 83.2�/113.6 mmHg,
respectively) and that of the LDL�/VLDL-C (227.3�/
143.6 and 229.0�/137.0 mg/dl, respectively) appears to
decrease. Both mean and median of the extent of
atherosclerotic lesions are seen to increase as age
increases, with the exception of fatty streaks in the
aorta. The estimated smoking risk (serum thiocyanate]/
90 mg/l) is seen to decrease by age level, possibly
suggesting that smoking may be increasing in younger
Greenlanders as compared with their elders; however,
this may instead be due to sampling bias as a result of
higher mortality among smokers.Table 5 presents a summary, by sex, of the analysis of
dependence of the extent of the atherosclerotic lesions in
the aorta and coronary arteries upon the risk factor
covariants of age (years), smoking (thiocyanate�/90 mg/
l), estimated hypertension (MBP�/110), high LDL�/
VLDL-C (]/130 mg/dl), low HDL-C (B/40 mg/dl),
and obesity (BMI]/30). Lesion values were log-trans-
formed and, to minimize potential sampling bias, only
the basal cases were included in this analysis. The model
results are the probabilities of random influence for each
variable after adjusting for the co-variables. No sig-
nificant interactions were found among the independent
variables. There is no obvious indication of a strong
dependent association of lesions upon risk factors, with
the exception of raised lesions upon age. In males, age is
clearly an influential variable for raised lesions in all
segments (P B/0.001), but less so for females, where the
only instance of raised lesion dependence upon age is inthe thoracic aorta (P�/0.0453). The only other risk
factor variables upon which significant lesion depen-
dence was found were in females where smoking (P�/
0.0310), low HDL-C (P�/0.0381), and obesity (P�/
0.0436) showed significant influence upon fatty streaks
in the abdominal aorta.
4. Discussion
The sample cases in this study were comprised
primarily of subjects from the capital city, Nuuk, and
from a rural community, Ilulissat. The hospital in Nuukis the central hospital for all of Greenland, so some cases
from other areas of the country were also included. For
the period of data collection, 1990�/1994, the study
sample represents 26.4% of all adult deaths in Greenland
and is, therefore, a reasonable sampling of the rural
versus urban population distribution, as well as a
substantial sampling of all deaths in the country. As
this was an autopsy study of atherosclerosis, the majorunderlying cause of CHD, it was necessary to determine
whether there was any substantial selection bias due to
CHD deaths alone. Since 72% of the cases were basal
(i.e. cause-of-death was non-CHD or non-CHD re-
lated), this would seem to provide a margin of safety
in this regard. Because of the small sample size due to
the difficulties and resource demands associated with
Table 5
Greenland autopsy study*/basal cases (n�/73): covariant analysis of the dependence of the extent of atherosclerosis in the aorta and coronary
arteries upon risk factors
Risk factors Aorta Coronary arteries
Abdominal Thoracic Left anteriror descending Right
FS RL FS RL FS RL FS RL
Females (n�/ 29)
Age (years) 0.7139 0.8612 0.3833 0.0453 0.8696 0.4377 0.7013 0.1438
Smoking (thiocyanate]/90 mg/l) 0.0310 0.6384 0.7713 0.3720 0.6443 0.4227 0.7157 0.8929
Hypertension (MBP]/110) 0.0965 0.4961 0.6738 0.9978 0.5038 0.1781 0.5598 0.8881
High LDL�/VLDL-C (]/130 mg/dl) 0.3731 0.5773 0.8220 0.0852 0.4313 0.6833 0.5527 0.3162
Low HDL-C (B/40 mg/dl) 0.0381 0.0708 0.5954 0.5849 0.9205 0.5523 0.4937 0.7342
Obesity (BMI]/30) 0.0436 0.8531 0.3009 0.1441 0.5675 0.6196 0.6190 0.2490
Males (n�/44)
Age (years) 0.2117 B/0.0001 0.0920 B/0.0001 0.2626 B/0.0001 0.0962 0.0007
Smoking (Thiocyanate]/90 mg/l) 0.4207 0.4938 0.7270 0.9516 0.6963 0.6289 0.7756 0.3901
Hypertension (MBP]/110) 0.7388 0.5483 0.7808 0.3436 0.2537 0.1852 0.2056 0.1881
High LDL�/VLDL-C (]/130 mg/dl) 0.6974 0.6178 0.6437 0.7267 0.2146 0.4371 0.2626 0.6762
Low HDL-C (B/40 mg/dl) 0.7031 0.7064 0.8644 0.2790 0.1492 0.2910 0.3813 0.5799
Obesity (BMI]/30) 0.9702 0.5533 0.2135 0.5604 0.4089 0.7294 0.4336 0.6047
Table values are probabilities of random influence after adjusting for covariates; significant associations are in bold font. FS, fatty streaks; RL,
raised lesions.
H.S. Pedersen et al. / Atherosclerosis 170 (2003) 93�/103100
collecting autopsy specimens, data from all available
cases were included in as many of the analyses as
possible, as indicated in table headers. Also, there is
always a concern when analyzing blood that wascollected post mortem, especially in the case of subjects
who expired while hospitalized, and as described earlier,
blood analyte data from seven cases in which hemodilu-
tion was suspected were excluded.
Greenlanders appear to have a high prevalence of
some of the CHD risk factors. The mean and median
values for the serum lipids (Table 2a) are relatively high
in comparison to the currently recommended guidelines[24]. Indeed, more than half (69.2% females; 67.9%
males) have high LDL�/VLDL-C and almost half
(46.2% females; 50.0% males) have low HDL-C as
seen in Table 2b. It may be argued that the attempt to
limit hemodilution bias by excluding low values of blood
analytes from calculations may have resulted in a
reverse-bias. However, a review of the calculations
without exclusions indicated otherwise: all of theexcluded HDL-C values except one (46 mg/dl for one
male) were less than the cut-point for CHD risk (B/40
mg/dl) and when all data were included, the re-calcu-
lated values for High LDL�/VLDL-C were found to be
Females: 64.3%, males: 57.6%; and females: 46.4%,
males: 57.63% for low HDL-C, neither being of
significant difference by gender. For hypertension, the
figures (40.5% females; 38.0% males) are likewise high incomparison to the reported 15.7�/29.8% in other popu-
lations world-wide [26,27]. On the other hand, the
prevalence of obesity (20.0% females; 8.8% males)
appears to be comparable to, or perhaps a bit lower
than that reported in Europe and North America [28].
Lastly, the estimated prevalence of smoking (15.4%
females; 18.2% males), compares favorably with that
reported for ‘‘heavy smokers’’ (14.5% females; 31.0%males) in an epidemiological study conducted in Green-
land during the same time period [10], but appears to be
low compared with the reported 30% or more for
northern Europe [29]. Since the thiocyanate levels in
the excluded cases (34.5�/83.0 mg/l) were all below the
smoking estimator cut-point of ]/90 mg/l, they could
not be implicated here. However, as also noted pre-
viously, the estimator cut-point for thiocyanate as amarker for smoking was established in a different
population with different dietary habits and may,
therefore, be an imprecise measure in this population
group. Furthermore, there is also the possibility that
some of the cases were patients that expired in the
course of lengthy hospitalization during which cessation
of smoking was required, leading to possible false
negative results of classification; however, the longhalf-life of serum thiocyanate [30] should reduce this
risk.
Prevalence of raised lesions, indicating advanced
atherosclerosis, ranges from 69.9 to 80.2% in the aorta
and 65.3�/66.7% in the coronary arteries of the Green-
landers in this study (Table 3). The one observed
significant difference in prevalence between females
and males, seen in the fatty streaks of the abdominalaorta, does not seem to argue strongly for a difference in
prevalence of atherosclerosis by gender, and all differ-
ences vanish for the basal cases. The absence of a gender
difference in prevalence is consistent with the observed
absence of any significant gender differences for the risk
factors in this study sample. Of course, the dispropor-
tionate number of older cases cannot be completely
dismissed as a possible biasing influence upon theseobservations, despite the fact that gender comparisons
within younger and older age-groupings (age B/40 and
]/40 years), as previously noted, also showed no
statistically significant differences.
In all four arterial segments in this study, the
minimum and maximum extent of atherosclerosis ran-
ged from 0 to 3.3% for cases in the 15�/24 years age
range and 0�/97% in the �/60 years age range for themore advanced raised lesions. Although the upper level
of the range for raised lesions increased with age, only in
the aorta did the minima rise above zero, beginning with
the cases above age 34 years, following the expected
pattern of lesion progression reported in other autopsy
studies. Also, only in the aorta was there found any
gender difference in atherosclerosis, with females having
a greater mean extent of raised lesions than males;however, this was not statistically significant. In the
abdominal aorta, females had 60.6% mean surface
involvement with raised lesions versus 46.6% in males,
and in the thoracic aorta, the difference was 39.6 versus
28.5%. This may be associated with the greater pre-
valence of obesity found in females.
Analysis of the association of aortic and coronary
atherosclerotic lesions with generally known risk factorsis summarized in Table 4b, Table 5. In Table 4b, as has
been reported in autopsy studies of other population
groups, the extent of the clinically significant raised
lesions in both the aorta and the coronary arteries
increases with age. On the other hand, fatty streaks seem
to increase with age in the right coronary artery and left
anterior descending coronary artery, but not in the
aorta. This may be a consequence of raised lesionprogression in lesion-prone areas initially occupied by
fatty streaks. In Table 5, there is the first suggestion of
significant gender differences. Males show a statistically
significant (P B/0.001) dependence of raised lesions
upon age in all segments, whereas only for raised lesions
in the thoracic aorta does age rise to significance as a
contributing factor for females. In females, smoking
(P�/0.0310), obesity (P�/0.0436), and low HDL-C(P�/0.0381) show strong influence on fatty streaks in
the abdominal aorta, but this is not seen in males. The
latter two may reflect the gender differences in pre-
valence seen in Table 2b for obesity (20.0% in females
H.S. Pedersen et al. / Atherosclerosis 170 (2003) 93�/103 101
vs. 8.8% in males) and Low HDL-C (46.2% in females
vs. 50.0% in males), despite the fact that these were not
statistically significant. The risk factor variables of
estimated hypertension and high LDL�/VLDL-C donot seem to contribute significantly to the model for
either gender.
A final observation is that the prevalence of advanced
atherosclerosis (raised lesions) among Greenlanders in
this study is very high as compared with that reported
for Alaska native Eskimo and Indians in a similar
autopsy study in those groups. Prevalence of raised
lesions in Alaska natives was found to be 30.0�/44.6% inthe aorta and 39.7�/45.2% in the coronary arteries [14]
compared with 77.5�/86.3 and 71.0�/73.3%, respectively,
reported here for Greenlanders. Although mean age is
not explicitly mentioned in the Alaska study, one can
estimate it based on the reported age-range counts. It is
evident that it is somewhat lower by approximately 30
years for the Alaska cases, so age difference may well
account for the large difference in prevalence. However,comparing extent of coronary lesions in the comparable
age ranges in this study to that reported for Alaska
shows that the mean extent of raised lesion involvement
in the Greenlanders appears to be the same as that of the
Alaska natives. In that study, the mean values (%)
reported for raised lesions in males/females for the 15�/
24 years age range were 0.9/0.0 in the left anterior
descending coronary (LAD) and 0.5/2.0 in the rightcoronary (RC), and in the 25�/34 years age range the
values were 4.7/0.9 and 3.1/1.0, respectively. In the
Greenlanders, the values for the combined genders in
the 15�/24 years range were 0.9 in the LAD, 0.8 in the
RC; and for the 25�/34 years range, 0.8 and 0.7,
respectively. Despite the mixed gender values for the
latter, it seems clear that no appreciable differences are
apparent between the extent of raised lesions in thecoronary arteries of the Greenland and Alaska natives
in those two age-matched groups. Furthermore, since
the Alaska study included non-native cases and reported
that extent of atherosclerosis was less in natives than in
non-natives, one can infer that the Greenlanders have
less atherosclerosis than Alaska non-natives.
There are two conclusions that can be made from
these data. First, no evident differences were observed,either in regard to the prevalence and extent of athero-
sclerosis or to the commonly associated risk factors,
between females and males in the Greenlanders of this
study. This possible lack of gender difference in athero-
sclerosis is of interest, especially if there is indeed less
IHD mortality in both females and males this popula-
tion, since pre-menopausal women below age 60 years
are reported to develop CHD at half the rate of men[31�/33]. The second conclusion is that the extent of
advanced atherosclerosis in Greenland natives in the age
range B/35 years, although based upon a small sample
size (n�/11), is the same as has been reported in Alaska
natives and less than that reported for Alaska non-
natives, supporting the suggestion that reduced extent of
atherosclerosis may be a factor in the low IHD mortality
in this population.
References
[1] Bang HO, Dyerberg J, Nielsen AB. Plasma lipid and lipoprotein
pattern in Greenlandic West-coast Eskimos. Lancet
1971;1(7710):1143�/5.
[2] Bjerregaard P, Dyerberg J. Mortality from ischaemic heart disease
and cerebrovascular disease in Greenland. Int J Epidemiol
1988;17:514�/9.
[3] Middaugh JP. Cardiovascular deaths among Alaskan natives
1980�/1986. Am J Public Health 1990;80:282�/5.
[4] Davidson M, Bulkow LR, Gellin BG. Cardiac mortality in
Alaska’s indigenous and non-native residents. Int J Epidemiol
1993;22:62�/71.
[5] Ebbesson SOE, Schraer C, Nobmann E, Van Citters RI, Boyko
EJ, Adler A. Cardiovascular disease on the increase in Alaskan
Eskimos. 2nd international heart health conference, May 28�/June
1, Barcelona, Spain, 1995 (Abstract).
[6] Young TK, Moffatt EK, O’Neil JD. Cardiovascular diseases in a
Canadian Arctic population. Am J Public Health 1993;83(6):881�/
7.
[7] Bjerregaard P. Validity of Greenlandic mortality statistics. Arct
Med Res 1986;42:18�/24.
[8] Hansen JPH, Hanche S, Møller-Petersen J. Atherosclerosis in
Native Greenlanders: an ultrasonographic investigation. Arct
Med Res 1990;49:151�/6.
[9] Ingeman-Nielsen MW. Arteriosclerosis in Greenlanders. Ugeskr
Læger 1990;152:2641�/3.
[10] Bjerregaard P, Mulvad G, Pedersen HS. Cardiovascular risk
factors in Inuit of Greenland. Int J Epidemiol 1997;26(6):1182�/
90.
[11] Gottman AW. A report of 103 autopsies on Alaskan natives.
AMA Arch Pathol 1960;70:117�/24.
[12] Arthaud B. Cause of death in 339 Alaskan natives as determined
by autopsy. Arch Pathol 1970;90:433�/8.
[13] Newman WP, III, Middaugh JP, Propst MT, Rogers DR.
Atherosclerosis in Alaska natives and non-natives. Lancet
1993;341:1056�/7.
[14] Newman WP, III, Middaugh JP, Guzman MA, Propst MT,
Rogers DR. Comparison of atherosclerosis in Alaska natives and
nonnatives. Arch Pathol Lab Med 1997;121:1069�/75.
[15] Greenland 1997-Statistical Yearbook (ISBN 87-90393-27-9).
Statistics Greenland, P.O. Box 1025, DK-3900 Nuuk, Greenland
(http://www.statgreen.gl/english/yearbook/index-yb.html).
[16] Guzman MA, McMahan CA, McGill HC, Jr, et al. Selected
methodology aspects of the International Atherosclerosis Project.
Lab Invest 1968;18:479�/97.
[17] Guzman MA, McMahan CA, Strong JP. Unaided visual estima-
tions of atherosclerotic lesions. Biological variability compared
with grading variability. Lab Invest 1974;31:398�/402.
[18] Bowler RG. The determination of thiocyanate in blood serum.
Biochem J 1944;38:385�/8.
[19] Pathobiological Determinants of Atherosclerosis in Youth
(PDAY) Research Group. Effects of serum lipoproteins and
smoking on atherosclerosis in young men and women, Arter-
ioscler Thromb Vasc Biol 1996;17(1):95�/106.
[20] Tracy RE, Mulvad G, Pederson HS, Jul E, Bjerregaard P,
Newman WP, III. Blood pressure in people in Greenland assessed
by measuring renovasculopathies of hypertension at autopsy. Am
J Hypertens 1996;9(6):560�/5.
H.S. Pedersen et al. / Atherosclerosis 170 (2003) 93�/103102
[21] McGill HC, Jr, McMahan CA, Herderick EE, et al. Effects of
coronary heart disease risk factors on atherosclerosis of selected
regions of the aorta and right coronary artery. Arterioscler
Thromb Vasc Biol 2000;20:836�/45.
[22] Strong JP, Restrepo C, Guzman M. Coronary and aortic
atherosclerosis in New Orleans I. Sampling bias due to source
of autopsy specimens. Lab Invest 1978;39(4):358�/63.
[23] McGill HC, Jr. The relationship of dietary cholesterol to serum
cholesterol concentration and to atherosclerosis in man. Am J
Nutr 1979;32:2664�/702.
[24] Executive summary of the third report of the National Choles-
terol Education Program (NCEP) expert panel on detection,
evaluation, and treatment of high blood cholesterol in adults
(adult treatment panel III), J Am Med Assoc 2001 May
16;285(19):2486�/97.
[25] Strong JP, Restrepo C. Coronary and aortic atherosclerosis in
New Orleans II. Comparison of lesions by age, sex, and race. Lab
Invest 1978;39(4):364�/9.
[26] Van den Hoogen PC, Feskens EJ, Nagelkerke NJD, et al. The
relation between blood pressure and mortality due to coronary
heart disease among men in different parts of the world. New Engl
J Med 2000;342:1�/8.
[27] Burt VL, Whelton P, Roccella EJ, Brown C, Cutler JA, Higgins
M, Horan MJ, Labarthe D. Prevalence of hypertension in the US
adult population. Results from the Third National Health and
Nutrition Examination Survey, 1988�/1991. Hypertension
1995;25(3):305�/13.
[28] Arterburn D, Hitchcock Noel P. Obesity. Br Med J
2001;322:1406�/9.
[29] Manninen J, editor. Smoking prevalence and tobacco policies in
the member states of the European union: a summary of the
overviews by country, National Public Health Institute, Depart-
ment of Epidemiology and Health Promotion, European Network
on Young People and Tobacco. Helsinki, Finland 1997. Julk-
aisija-Utgivare (http://www.ktl.fi/enypat/data/smokeprev_eu.
htm).
[30] Butts WC, Keuhneman M, Widdowson GM. Automated method
for determining serum thiocyanate to distinguish smokers from
nonsmokers. Clin Chem 1974;20(10):1344�/8.
[31] Kannel WB, Hjortland MC, McNamara PM, Gordon T.
Menopause and risk of cardiovascular disease: the Framingham
study. Ann Intern Med 1976;85(4):447�/52.
[32] Nathan L, Chaudhuri G. Estrogens and atherosclerosis. Annu
Rev Pharmacol Toxicol 1997;37:477�/515.
[33] Lusis JL. Atherosclerosis. Nature 2000;407:233�/41.
H.S. Pedersen et al. / Atherosclerosis 170 (2003) 93�/103 103