determined by PCR amplification and subsequent diges- tion with the restriction enzyme HhaI as described by
Hixson and Vernier [19]. Genotyping was performed in a single site in Vandoeuvre-le`s-Nancy France for
Northern Ireland, France and Greece. The exactly same methods were transferred to Lisbon which carried out
their own determinations. Phenotyping was done in Hamburg Pr. U Beisiegel by isoelectric focusing [20] for
Spain and Finland.
2
.
5
. Effect of storage on serum apo E concentration The stability of apo E concentration during storage for
up to 3 months at − 80°C was assessed by comparing apo E concentration values measured in ten fresh serum
samples. In addition, we verified the effect of storage at −
196°C for up to 4 years on 34 samples from individuals from the Stanislas Cohort Study. Two measurements
were performed after 2 and 4 years of storage. At − 80°C no significant change in serum apo E concentration
occurred during for up to 3 months of storage. Similar results were obtained on EDTA plasma after storage at
−
80°C. Storage at − 196°C for up to 4 years did not significantly affect apo E serum concentration. The
regression equation obtained between two measurements made after 2 years and 4 years of storage at − 196°C,
was y = 0.986x + 0.383 with a coefficient of determina- tion of 0.9945.
2
.
6
. Statistical methods The distribution of the apo E polymorphism was tested
for Hardy – Weinberg equilibrium using x
2
testing. The independent association between the occurrence of cer-
tain apo E genotypes and centre, age, and gender, was expressed as multivariately adjusted odds ratios ORs
calculated using multiple logistic regression modelling. The statistical significance of the independent contribu-
tion of the latter three variables was evaluated using Wald x
2
testing. Individual ORs are accompanied by 95 confidence intervals CI in order to statistically
judge their difference from unity. Association between the same set of explanatory
variables and the apo E concentration was studied using multiple regression analysis after natural logarithmic
transformation. From these analyses, it was concluded that the effect of age on the apo E concentration in all
centres was significantly quadratic in men and linear in women. Hence, further age-adjustment was done
quadratically in men and linearly in women.
The distribution of apo E concentrations by centre, age, and gender, was characterised by the mean, standard
deviation SD, median, interquartile range IR, percen- tiles 25.0 – 75.0, percentiles 2.5, 5.0, 95.0 and 97.5,
the geometrical mean accompanied by an asymmetrical 95 CI, the mean of ln-transformed values [ln apo E]
and their SDs. A level of a = 0.05 was used to indicate statistical significance.
3. Results
3
.
1
. Population description The population sample stratified by age, sex, and
centre, is presented in Table 1. The contribution of each centre to the total population sample was 7.6 in Greece,
about 9 – 10 in Portugal and Northern Ireland, and 15 in Spain. It reached 28 in France and 30 in Finland.
Forty-seven point five percent of all young men, 25 – 34 years old, came from the Belfast cohort; a majority of
males and females 35 – 44 years old, came from the French cohort representing 50.5 and 54.3; 52 – 60 of
the subgroups of males and females aged 45 – 64 years came from the Finnish cohort. The designs of recruit-
ment, blood sampling and storage conditions used in the different centres are summarised in Table 2. Among the
six participating centres, subjects were fasting overnight
Table 1 Number of subjects of the population sample in each centre stratified by age and sex
All subjects Spain
France Portugal
Greece Crete Finland
Age groups years Northern Ireland
Men 208
60 45
71 –
25–34 54
438 330
129 70
678 –
35–44 1342
67 63
143 1142
136 99
534 45–54
235 86
470 –
40 171
784 17
55–64 975
315 674
241 497
25–64 3706
1004 Women
25–34 115
– 68
– 135
44 362
1064 150
97 –
35–44 63
686 –
45–54 568
172 38
– 103
165 978
5 515
824 55–64
199 61
– 44
558 393
– 213
978 1083
3228 25–64
Table 2 Design of recruitment, blood sampling and storage conditions used in the six centres
Northern Ireland France
Centres Spain
Finland Portugal
Greece Area of Belfast,
East and south North-east
Geographic areas Area of
Crete Area of Lisbon
Barcelona Northern Ireland
Industrial workforce Stanislas cohort
Monica-Cataloni Specific
Third Finmonica Specific
Recruitment a
survey recruitment
recruitment 111994–031995
Period 091993–081995
011992–031992 1993–1994
1996–1997 1997
Sampling Fasting
] 6 h
] 4 h
12 h 12 h
12 h 12 h
Sitting Supine
Sitting Sitting
Supine Posture
Supine Storage
− 80°C
− 196°C
− 80°C
–70°C −
70°C Temperature
− 80°C
2.6 years 1.5 year
Duration up to 3–4 years
5.5 years 6 months
4 months Table 3
Prevalence of apo E polymorphism by participating centres France
Finland Greece
Northern Ireland Portugal
Spain Obser6ed frequencies
12 2
o 2o2
1 2
9 271
o 3o2
38 196
60 69
117 1211
304 395
1092 439
o 3o3
611 372
61 172
o 4o3
99 524
190 41
1 7
38 6
o 4o2
10 33
4 11
8 9
o 4o4
60 1.08
2.99 x
2
statistic
a
6.42 7.57
1.41 7.00
0.78 0.39
0.09 0.70
P-value 0.07
0.056 Relati6e allele frequencies
0.0866 o
2 0.0524
0.0617 0.0519
0.0634 0.0766
o 3
0.7899 0.7598
0.8622 0.7922
0.8395 0.8081
0.1235 0.0854
0.1558 0.1784
0.0971 o
4 0.1152
a
x
2
statistic: between observed and expected values under Hardy–Weinberg equilibrium.
in four, and blood was drawn in the supine position in three. The storage temperature of serum samples until
analysis was 5 − 70°C in all centres.
3
.
2
. Distribution of Apo E polymorphism The distribution of apo E genotypes was in Hardy –
Weinberg equilibrium for all the participating centres in the ApoEurope Project Table 3. In Finland, the sig-
nificance of the deviation from the Hardy – Weinberg equilibrium was borderline due to the o2o2 polymor-
phism being underrepresented. However, regarding the number of x
2
tests performed globally, the possibility that this was a chance-finding cannot be excluded.
There is a clear North – South decreasing gradient in the prevalence of the o4o3 genotype. The mirror image of
this gradient is present for the o3o3 genotype and to a lesser extent, for the o4o4 genotype.
In order to study the independent effect of age, sex, and centre, on the prevalence of the apo E genotypes, a
multivariate analysis was done and the results are pre- sented in Table 4. The ORs and their 95 CI for
occurrence of genotypes o2o2 + o3o2 versus o3o3 and of o4o4 + o4o3 versus o3o3 are given by centre, age,
and sex. From these data, it seems that the o2o2 + o3
o 2 genotypes occur more frequently in France and in
Spain compared to Greece, independently of age and sex. Age by itself is not related to genotype, but in
women, the o2o2 + o3o2 genotypes are found more frequently, even when the data were adjusted for centre
and age. Similar results were obtained when ORs were calculated for occurrence of o2o2 or o3o2 genotypes
versus other genotypes data not shown. To consider more closely the gender difference observed previously,
age-adjusted ORs women versus men were performed after stratification by centre. The gender difference was
present only in Finland P = 0.008, but not in other countries P = 0.12 to P = 0.79. The prevalence of the
o 4o4 + o4o3 genotypes differs greatly between centres
compared to o3o3 genotype or to other genotypes P B 0.0001. The OR-values in Finland versus Greece
were, respectively 2.42 and 2.74, independently of age
and sex. These two variables were unassociated with the occurrence
of the
o 4o4 + o4o3
genotypes, independently of the centre.
3
.
3
. Apo E concentration by age group, sex, centre In Table 5, the distribution of the apo E concentra-
tion is given by centre, age and sex; the descriptive statistics are the number of observations, the mean
value, the SD, the median, the IR, the percentiles 2.5, 5.0, 95.0 and 97.5, the geometrical mean with an
asymmetrical 95 CI for the original apo E concentra- tions, and the mean and SD of the ln apo E concen-
tration on which this geometrical mean and the 95 CI are based.
Fig. 1 shows the geometrical mean of the apo E concentrations obtained in men and women by age,
adjusted for centre because of the large differences
Table 4 Multivariate adjusted odds ratios OR and 95 CI for occurrence of phenotypes o2o2 or o3o2 versus o3o3 and of o4o4 or o4o3 versus o3o3
by centre, age group and sex
o 4o4 or o4o3 versus o3o3
o 2o2 or o3o2 versus o3o3
OR 95CI OR 95CI
Centre Centre
1 1
Greece Greece
1.29 0.83–2.00 Northern Ireland
Portugal 1.14 0.81–1.61
1.48 1.08–2.03 Spain
Portugal 1.18 0.78–1.79
1.28 0.88–1.85 Finland
France 1.52 1.13–2.06
Northern Ireland 2.24 1.61–3.12
Spain 1.50 1.02–2.20
1.78 1.23–2.58 France
Finland 2.42 1.80–3.24
Significance P = 0.006
Significance PB0.0001
Age years Age years
1 25–34
25–34 1
35–44 1.01 0.77–1.34
35–44 1.14 0.91–1.43
1.10 0.83–1.47 45–54
45–54 1.10 0.87–1.39
1.12 0.81–1.55 1.16 0.90–1.50
55–64 55–64
Significance P = 0.83
Significance P = 0.61
Gender Gender
1 Men
Men 1
Women Women
1.23 1.04–1.44 1.07 0.94–1.21
P = 0.32 Significance
Significance P = 0.01
Table 5 Distribution of the mean SD, median IR, percentiles, geometrical mean CI of apo E concentration
a
by centre, age, and sex Median IR
Percentiles GM 95 CI
ln apo E mean SD N
Mean SD 97.5
2.5 5.0
95.0 Centre
Finland 38.2 21.8–67.2
2087 3.64 0.288
33.8 11.64 38.4 32.2–45.9
21.4 24.0
59.6 65.1
67.1 37.7 21.2–66.9
3.63 0.293 39.4 12.82
France 37.5 31.3–44.9
1953 21.6
23.8 61.2
3.91 0.311 83.1
93.2 49.7 27.0–91.5
Greece 30.7
528 52.4 20.50
48.8 40.6–59.4 27.9
41.5 22.7–76.0 674
3.73 0.308 43.7 15.21
41.0 34.3–49.9 24.1
26.1 69.9
79.7 Northern Ireland
77.9 637
41.7 23.6–73.7 3.73 0.291
43.6 13.74 41.9 34.3–49.7
25.2 26.9
67.0 Portugal
67.0 75.1
45.0 27.2–74.4 3.81 0.257
Spain 46.5 13.00
1055 44.9 38.4–52.4
27.2 29.6
Age years 25–34
3.65 0.311 38.6 21.0–71.1
74.8 64.5
24.2 20.9
38.4 31.5–46.8 40.6 13.67
800 3.68 0.298
64.1 25.0
39.6 22.1–71.0 22.7
74.5 39.4 32.9–47.2
41.5 14.41 2270
35–44 66.7
73.9 40.6 22.4–73.5
3.70 0.303 42.5 14.38
40.4 33.7–49.0 22.5
45–54 25.2
2256 66.7
74.4 42.5 24.0–75.3
3.75 0.292 55–64
1608 23.8
44.3 13.35 42.7 35.6–50.8
26.9 Sex
3706 41.2 22.5–75.5
3.72 0.308 43.3 15.44
40.8 34.0–49.4 23.3
25.7 68.6
Men 78.2
39.6 22.4–70.1 3.68 0.292
Women 3228
41.3 12.35 39.9 33.1–47.8
22.1 24.6
63.1 69.1
a
SD, standard deviation; IR, interquartile range; percentiles 25.0–75.0; percentiles: 2.5, 5.0, 95.0 and 97.5; GM, geometrical mean; CI, 95 confidence interval; ln apo E, log-transformed apo E concentration on which the geometrical mean and the 95 CI are based.
Fig. 1. Geometrical means of the apo E concentration by age and sex adjusted for centre.
existing between centres. The effect of age seems to be different by sex: in women, the effect seems to be positive
and linear; in men, the apo E concentration increases linearly with age up to the age 40 – 44 years, but this
increase seems to tail off thereafter. In order to study the curvilinear effect observed by age in men, regression
analysis was performed, modelling the association be- tween the apo E concentration after logarithmic trans-
formation and age for each of the centres separately. This analysis revealed that the age effect in men was
indeed curvilinear with a quadratic term which was highly significant P B 0.0001 in those centres which
covered the age range 30 – 64 years France, Greece, Portugal and Spain. Following an identical procedure
for modelling, the age-effect in women revealed that was no deviation from linearity as the quadratic term was not
significant.
For that reason, the sex-related difference in apo E concentrations varies in two opposite directions: the
mean of serum apo E concentration is higher in men than in women in the younger individuals, and lower in older
ones, with equivalent concentration in middle aged men and women. Because of this differential age-effect by sex,
we studied the gender difference in apo E concentration on the subgroup of 25 – 44 year old subjects for which the
age-effect on apo E is comparable between the sexes. As there were no Northern-Irish women in the sample
population and no Finnish men or women younger than 45 years, this subgroup analysis does not involve North-
ern Ireland or Finland. In France, Portugal, and Spain the concentration of apo E is significantly higher in men
than in women: respectively 39.1 versus 35.8 mgl P B 0.0001 in France; 40.9 versus 37.8 mgl P = 0.006 in
Portugal; and 43.5 versus 40.8 mgl P = 0.03 in Spain. The non-significant result in Greece 48.5 vs 46.3 mgl;
P = 0.24 may be due to insufficient power.
Knowing the estimated functional relationship be- tween age and the apo E concentration in men and
women, allows us to study possible differences between gender and centres adjusting for age as appropriate. Fig.
2 presents the apo E concentration by centre and gender, including all ages, and linearly adjusted for age in women
and quadratically in men. The differences between cen- tres persist and are highly significant.
3
.
4
. Association between the apo E concentration and apo E genotypes
Table 6 shows the distribution of the apo E concentra- tion by apo E genotype for men and women of all
participating centres in the ApoEurope Project. The concentration is significantly higher in the presence of the
o 2 allele and this is true for both genders and is
independent of centre. After adjustment for centre and age the apo E concentration stratified by apo E genotype
and by gender is 20 – 25 higher in men and women with the
o 2o2
or o
3o2 genotype
than those
with genotype o3o3 or o4o4 or o4o3 i.e. 51 versus
42.0 mgl and 41.2 mgl in men and 50 versus 41.8 mgl and 39 mgl in women.
Mean apo E concentrations by centre and sex ad- justed for apo E genotypes and age are shown in Fig. 3
for each of the participating centres. Large and signifi- cant differences in apo E concentration persist between
countries in both genders.
3
.
5
. Apo E serum concentrations by age, sex, and apo E genotype in the different European regions
We calculated apo E serum concentrations after exclusion of subjects with serum total cholesterol con-
centrations ] 11.0 mmoll or triglyceride concentra- tions
] 10.0 mmoll according to the following
Fig. 2. Mean apo E concentration by centre and sex after adjustment for age. Table 6
Distribution of the mean SD, median IR, percentiles, geometrical mean CI of apo E concentration
a
by apo E polymorphism and sex ln apo E mean SD
Mean SD Median IR
Percentiles N
GM 95 CI 2.5
5.0 95.0
97.5 Men
34.7 229.6
229.6 o
2o2 85.7 26.1–280.8
13 4.45 0.606
100.9 59.05 78.2 52.9–148.3
34.7 367
50.3 17.93 3.87 0.302
o 3o2
47.9 26.5–86.6 47.0 40.2–56.8
94.9 78.6
31.1 28.5
2208 73.8
o 3o3
40.5 23.1–71.1 3.70 0.287
65.0 26.1
40.2 33.8–48.3 42.3 13.56
24.1 24.5
70.6 79.2
39.3 20.9–74.2 o
4o3 3.67 0.324
776 41.6 15.23
38.6 31.8–47.4 22.0
28.6 48
30.5 67.7
80.4 46.9 28.6–76.9
3.85 0.253 48.4 12.65
45.0 41.2–55.5 o
4o2 37.6 30.4–45.6
17.1 21.8
63.6 79.9
36.9 18.5–73.9 3.61 0.353
60 o
4o4 39.3 14.95
Women 35.6
112.2 112.2
60.6 31.6–116.1 4.10 0.332
35.6 63.7 21.28
13 59.7 52.0–76.4
o 2o2
30.0 66.3
73.3 46.1 27.8–76.2
o 3o2
3.83 0.257 384
47.6 12.00 46.4 39.4–54.8
26.8 24.9
60.7 66.6
39.0 22.7–67.3 o
3o3 3.66 0.277
1844 40.6 11.42
39.4 32.8–46.7 22.3
3.60 0.293 36.5 20.6–64.8
64.8 58.2
23.1 o
4o3 21.1
37.0 30.1–44.0 38.1 11.68
642 46.0 39.9–53.8
29.7 31.4
74.5 95.3
46.9 27.4–80.5 3.85 0.275
55 o
4o2 48.9 15.93
32.3 26.3–38.1 34.1 11.29
3.48 0.316 32.4 17.4–60.3
67.8 55.9
21.0 65
16.5 o
4o4
a
SD, standard deviation; IR, interquartile range; percentiles, 25.0–75.0; percentiles, 2.5, 5.0, 95.0 and 97.5; GM, geometrical mean; CI, 95 confidence interval; ln apo E, log-transformed apo E concentration on which the geometrical mean and the 95 CI are based.
Fig. 3. Mean apo E concentration by centre after adjustment for apo E genotype and sex.
criteria: region, age, sex, and apo E polymorphism. The countries were grouped in four regions from North to
South: North; Middle; South and South-East. Males and females were divided in two age classes: B 45 years
old and ] 45 years old, and were classified in three groups according to their apo E genotypes: subjects
with o3o3 genotype; carriers of the o2 allele o2o3 and
o 2o2 and carriers of the o4 allele o4o4, and o4o3.
Subjects with the o2o4 genotype were excluded. The resulting data are presented in Table 7 for percentiles
2.5, 5.0, 50.0, 95.0 and 97.5. Serum apo E concentration is highest, irrespective of sex, age, and geographical
area in o2 carriers; subjects with the o3o3 genotype having intermediate values to o2 and o4 carriers. How-
ever, in men aged ] 45 years, values for apo E were very similar between o3 and o4 carriers from the South
and South-East. Moreover, a clear North – South gradi- ent in serum apo E concentration is observed in each
genotype, sex, and age groups. Subjects from the North exhibited the lowest values and subjects from South-
East the highest values. In subjects with the o3o3 genotype, the increase in apo E concentration at the
50th percentile was up to 20 for men B 45 years and 32 for women in the same age range according to
region. This North – South gradient leads to a shift of apo E distribution towards higher values in the South
and South-East.
4. Discussion