Clinical
Chemistry
42:11
1817-1823 (1996)
Frequency of phenotype-genotype discrepancies
at the apolipoprotein E locus in a large
population study
CARLOS
LAHOZ,*
DOREEN
OSGOOD,
JOSE
PETER
W.F.
M.
ORDOVAS
genetic
mn1S:
polymerase
epidemiology.
chain reaction
#{149}
genetics
ERNST
J.
SCHAEFER,
and
low-density
lipoprotein
(LDL) receptor and the LDL receptorrelated protein [1]. The gene is located in chromosome
19 in a
cluster with the apoC-I and C-I! genes. The apoE gene is
polymorphic
with three common codominant
alleles (#{128}2,
#{128}3,
and
#{128}4)
encoding
three plasma apoE isoforms (E2, E3, and E4,
respectively).
These isoforms differ by amno acid substitutions
at one or both of two sites (residues 112 and 158). The apoE3
isoform has cysteine at residue 112 and argimine at residue 158,
whereas arginmne in apoE4 and cysteime in apoE2 are present at
Apolipoprotemn
E (apoE) genotypes
were determined
in a
random subset of 1041 subjects
enrolled
in the Framingham Offspring Study by using DNA amplification
followed
by restriction
isotyping.
The results were compared
with
the apoE phenotypes
previously
assessed
by isoelectric
focusing.
Discrepancies
in apoE allele assignment
were
found in 98 subjects (9.4%). Both genotype
and phenotype
were reassessed
in these subjects.
Genotype
misclassification was observed
in 20 subjects,
whereas the initial phenotype assignment
was modified
in 46 subjects.
No concordance between
apoE phenotype and genotype
remained
in
32 subjects
(3.07%). Both methods resulted in similar apoE
allele frequencies.
Furthermore,
no differences
were observed regarding
the average allelic effect on total cholesterol, LDL cholesterol,
or HDL cholesterol
concentrations;
however,
a significant
difference
was noted on triglyceride
concentrations.
Our results indicate that most discrepancies between
genotype
and phenotype
assessment
of apoE
polymorphism
were due to sample mishandling,
data entry,
and technical
difficulties
rather than true discordances.
INDEXING
WILsoN,
both sites [2].
The study of the genetic variability at the apoE gene locus
has aroused considerable
interest in recent years. First, apoE
isoforms have been shown to have a significant contribution
to
the variability in LDL cholesterol
(LDL-C)
concentrations
in
populations
[3]. Second, subjects with the apoE4 allele appear to
be at an increased risk of coronary heart disease (CHD) after
adjusting for LDL-C concentrations,
and this may also be the
case for some apoE2 carriers [4]. Third, a significant association
has been shown between the apoE4 allele, Alzheimer disease [5],
and dementia.
Moreover,
apoE genetic variation
may also
explain some of the interindividual
variability
in plasma lipid
response to dietary and drug therapies [6, 7].
Traditionally,
the characterization
of the different
apoE
isoforms has been carried out by isoelectric focusing (JEF) with
the VLDL fraction [8] or by IEF and immunoblotting
of plasma
or serum [9]. More recently the availability of techniques
based
on the polymerase
chain reaction (PCR) permits the analysis of
apoE variability at the gene level [10, 11]. Intralaboratory
discrepancy between phenotype and genotype has been reported to
vary between 0.2% and 24% [12-20], the latter being found in
diabetic subjects [13, 14, 18]. The potential
clinical interest of
the determination
of apoE polymorphism,
due initially to its
association
with cardiovascular
disease, and more recently as a
risk factor for Alzheimer disease and dementia, underscores
the
need for accurate analysis of apoE genetic variants.
The aim of the present study is to assess the discrepancy
between protein phenotyping
and DNA genotyping
in a random
sample of 1041 subjects participating
in the Framingham
Off-
#{149}
lipids.
Apolipoprotemn
E (apoE) is a 299-amino
acid polypeptide
synthesized primarily in the liver.’ This protein is associated with
chylomicrons,
very-low-density
lipoproteins
(VLDL), and highdensity lipoproteins
(HDL)
and serves
as a ligand
for the
Lipid Metabolism
Laboratory,Jean
Mayer-USDA
Human Nutrition
Research
Center on Aging at Tufts University,
Boston, MA 02111, and The Framingham
Study, Framingham,
MA 01701.
*AUthor for correspondence
(Tufts University
address). Fax 617-556-3103;
e-mail [email protected].
‘Nonstandard
abbreviations:
apoE, apolipoprocein
E; LDL-C
(HDL-C,
VLDL-C),
LDL cholesterol;
CHD,
coronary
heart disease; IEF, isoelectric
focusing; and ASO, allele-specific
oligonucleotide.
Received January 26, 1996; accepted June 6, 1996.
1817
Apo E phenotype-
1818
spring Study and to estimate, by using the measured genotype
effect [21], the influence of apoE alleles, assigned by genotype or
phenotype,
on plasma total cholesterol,
LDL-C,
triglycerides,
and HDL-C
concentrations.
Subjects and Methods
SUBJECTS
In the present
study we included
a random subset of 1041
subjects (543 men and 498 women, age range 22 to 76 years,
mean age 48 ± 10 years) enrolled
in the Framngham
Offspring
Study. This cohort was initiated in 1971 with the intent to
evaluate the role of genetic factors in the etiology of CHD [22].
Informed
consent was obtained from each participantin the
study; all procedures followed were in accordance with the
Helsinki Declarationof 1975, as revised in 1983.
SAMPLE
COLLECTION
AND
APOE
PHENOTYPING
Blood samples were collected
at the Framimgham
Study into
EDTA-containing
tubes (final concentration
1 g/L EDTA)
from fasting (>12 h) participants
at the third examination
of the
Framingham
Offspring Cohort between 1983 and 1987. Plasma
was immediately obtained by using low-speed ultracentrifugation, and 5 mL was subjected
to ultracentrifugation
in a
Beckman 40.1 rotor at 115 000g for 18 h at4 #{176}C,
ata densityof
1.006 kg/L. VLDL samples were obtained after tube slicing
from the upper part of the centrifugetube, and the sample
volume was brought to 2.5 mL. VLDL fractions were maintained at 4 #{176}C
and transported
on ice to the Lipid Metabolism
laboratory
in Boston within 24-48
h after their isolation.
\TLDLs were dialyzed against bicarbonate,
lyophilized,
delipidated, and subjected to IEF within a pH range of 4.0 to 6.5. This
procedure
has been previously described in detail [8].
ApoE phenotypes
were reassessed in a subset of the population during 1995 by IEF of whole plasma followed by immunoblotting. The plasma samples (1-mL aliquots) utilized for this
purpose had been stored at -80 #{176}C
since the time of their
collection (1983-1987).
genotype
discrepancies
volume of 50 JLL. Each reaction mixture was heated at 94 #{176}C
for
2 mm and subjected to 35 cycles of amplification
(94 #{176}C
for 40 s,
62 #{176}C
for 30 s,and 72 #{176}C
for 1 mm). The PCR products were
digested with 5 U of HhaI and the fragments separated by
electrophoresis
on an 8% polyacrylamide
nondenaturing gel.
Afterelectrophoresis
the gelwas treated with ethidium bromide
for 30 mm and DNA fragments
were visualized by ultraviolet
illumination.
The results were recorded with a Polaroid MP-4
camera and Polaroid type 57 high-speed
film.
LIPID
AND
LIPOPROTEIN
STATISTICAL
ANALYSES
Differences
in apoE allele frequency were assessed by x2 analysis. A logarithmc
transformation
was applied to plasma triglyceridesbefore analysis because of its skewed distribution.
Mean plasma lipidsand lipoproteim concentrationswere compared across categories
of apoE genotypes
(2/2, 2/3, 3/3, 3/4,
2/4, 4/4) by using ANOVA. The average effect (a) of the #{128}2,
#{128}3,
and #{128}4
alleles on total cholesterol,
LDL-C,
HDL-C,
and
triglycerides
was determined
by using previously
described
equations [21]:
a,
522j.L22 + l/2f232s
ISOLATION
AND
APOE
iC2
f33p33
+ l/2523p3
+ l/2f34M34
-ILL
Id
f4ji#{247}+ l/2f,424
GENOTYPING
DNA was extracted from 5-mL blood aliquots collected during
the fourth examination period of the Framingham
Offspring
Study (1987-1991)
by using a salting-out
method [23]. Most
DNA extractions
were carried out within 2-4 days after blood
collection. When the procedure could not be carried out within
that period, the blood was frozen at -20 #{176}C
and the DNA was
extracted within the following 4-8 weeks.
ApoE genotyping
was performed
as described by Hixson and
Vernier [10]. A 244-bp sequence of the apoE gene including the
two polymorphic
sites was amplified by PCR in a DNA Thermal
Cycler (PTC-lOO;
MJ Research, Watertown,
MA) with oligonucleotide
primers F4 (5’-ACAGAATFFCGCCCCGGCCTGGTACAC-3’)
and F6 (5 ‘-TAAGCYFGGCACGGCTGTCCAAGGA-3’).
The reaction mixture contained
the following
reagents:
200 jkmol/L each deoxynucleoside
triphosphate,
20
pmol of each primer, 100 mLIL dimethyl sulfoxide, 1 mmol/L
MgCl2, 50 mmol/L KCI, 10 mmollL Tris (pH 8.4), 250 ng of
genomic DNA, and 0.5 j.L of Taq DNA polymerase
in a final
+ 1/2f2424
=
a4 =
DNA
ANALYSES
Plasma HDL-C
was measured
after precipitation
of apoBcontaininglipoproteins
with dextran sulfate-Mg2 + [24]. Plasma
total cholesterol,
1.006 kg/L infranatant
cholesterol,
HDL-C,
and triglyceride
concentrations
were measured
by automated
enzymatic
teclmiques
with an Abbott
Diagnostic
ABA-200
bichromatic
analyzer. Plasma LDL-C values were obtained by
subtracting
HDL-C
from 1.006 kg/L infranatant
cholesterol.
CVs between runs for all lipid assays were <5%.
+ 1/2f3434
Sd
where 122, 123, etc., are the expected genotype
or phenotype
frequencies
assuming
Hardy-Weinberg
equilibrium;
5C2’ J’3,
and
are the allele frequencies;
j.L22,
P23,
etc., are the means
for the genotype or phenotype, and .t is the grand mean of the
sample.
Results
ApoE genotype
analysis of 1041 participants
in the Framimgham
Offspring
Study was compared
with the results previously
obtained by protein phenotyping.
The genotype assignment was
carried out without knowledge
of the previous phenotyping.
Discrepancies
were noted in 98 cases, which represents a lack of
concordance
of 9.8%. To identify the sources for these misclassifications,
phenotype
and genotype
analyses were repeated,
with new plasma and DNA samples, on those subjects with
discordant
results. This analysis revealed
that the repeated
genotyping
identified 20 subjects in whom the initial genotypes
were misclassified (12 of them were apoE3/4 subjects, previously
Clinical Chemistry
typed as 3/3). Following
the second protein phenotyping,
apoE
typing was corrected in 46 subjects (21 E3/3 had been previously
classified as apoE4 carriers, and 15 E3/3 had been previously
classified as apoE2 carriers). Lack of concordance
between apoE
phenotype
and genotype assignments
remained unresolved in 32
subjects (3.07%).
Distribution
of apoE genotypes
and phenotypes
in this
population before and after reanalysis of nonconcordant
samples
is shown
in Table 1. No significant
differences
were noted
between apoE allele distribution
as determined
by genotype or
phenotype
(y, P >0.5). This was true for both the initial
assignments
and those determined after repeating the nonconcordant samples. In all cases, the phenotype and genotype
distributions
were in Hardy-Weinberg
equilibrium.
The data inTable 2 present a cross-tabulation
of prevalence
of genotype-phenotype
discordancein the 32 subjectsin which
the lack of concordance could not be resolved.The most
common
mismatches consisted
of subjectstyped as E3/3 by
phenotype but classified
as E2/3 (n = 7) or E3/4 (n = 7) by
genotype.
The average effect of #{128}2,
#{128}3,
and #{128}4
alleleson totalcholesterol, LDL-C,
triglycerides,
and HDL-C
concentrations
in the
population
and in the 32 subjects with genotype-phenotype
discrepancies
is shown in Tables 3 and 4, respectively.
In the
whole group, no significant
differences
were found in the
average effect of apoE alleles, obtained by genotype or phenotype, on plasma total cholesterol,
LDL-C,
and HDL-C.
However, a significant difference (P <0.01) was noted in the average
42, No.
1819
11, 1996
apoE allele effect on plasma triglycerides,
calculated
with the
genotype or the phenotype
assignments.
Similarfindings were
noted regarding
the effect of apoE locus-related
variability on
total cholesterol,
LDL-C,
triglycerides,
and HDL-C
in the 32
subjects with discrepancies.
The VLDL-C/triglycerides
ratio
was >0.3 in 14.4% of the subjects in the entire population.
This
percentage
was significantly
higher (28%) (P = 0.03) in the
group with genotype-phenotype
discrepancies.
Discussion
ApoE genetic variability
has been traditionally
determined
by
using IEF separation
of VLDL proteins or whole plasma [8, 9].
Most recently, PCR has simplified the directassessment of the
specific nucleotide
mutations responsiblefor the common apoE
isoforms[10]. Assuming that the most common mutations in the
population are those at the 112 and 158 positions, one would
expectcloseto complete concordance on the assessment of apoE
variants with the two methodological approaches; however,
several laboratories
have reported a wide range of discrepancies
between phenotype
and genotype
assessment
(0.2-24%)
[1220]. The high rate of misclassification
reported in some studies
constitutes
a major concern, given the large number of population studies reported in the literature and the important
findings
derived from their analyses.
The data obtained in the present study, the largest to date to
compare apoE genetic variability by using IEF and PCR, agreed
with previous reports on the presence of discrepancies
between
these two techniques.
Of 1041 subjects, 98 (9.4%) were appar-
Table 1. Frequency distribution of apoE genotypes and phenotypes and apoE allele frequencies, before and after correction
of phenotype-genotype discrepancies.
Genotype
Phenotype
Initial analysis
Cotrected
n
COrrected
n=1041
2/2
2/3
3/3
3/4
1
123
678
215
11.8
65.1
119
11.4
121
11.6
115
0.3
11.0
670
64.3
683
65.6
675
64.8
20.6
227
21.8
212
20.3
221
21.2
4/4
10
1.0
10
1.0
11
1.0
14
1.3
2/4
14
1.3
14
1.3
11
1.0
13
a
n
initial analysis
0.1
n
1
0.1
n
3
0.3
3
1.2
#{128}2
0.066
0.064
0.066
0.064
#{128}3
0.813
0.809
0.816
0.809
#{128}4
0.119
0.125
0.117
0.125
Includes the 32 subjects presenting lack of concordance between phenotyping and genotyping.
Table 2. Cross-tabulatlo n of the genotypes
and the phenotype s of our 32 subjects
with
discrepancies.
Phenotype
Genotype
n
2/3
10
3/3
10
3/4
10
4/4
1
2/4
1
2/2
2/3
3/3
3/4
4/4
2
6
15
4
5
7
1
2
3
1
1
5
1
7
1
1
2
1820
Apo E phenotype-genotype
Table 3. Average
effect
of #{128}2,
#{128}3,
and #{128}4
alleles, genotype and phenotype assigned,
on total cholesterol (TC), LDL-C,
triglycerides (TG), and HDL-C on the entire population (mmol/L).
LDL-C
IC
Genotype
Phenotype
#{128}2
-0.290
-0.189
#{128}3
#{128}4
-0.003
0.137
-0.005
0.114
a
discrepancies
Genotype
Phenotype
-0.341
0.016
0.147
HDL-C
TG
Genotype
Phenotype
Genotype
-0.300
0.098
0.168
0.016
-0.016
0.0
-0.023
-0.017
0.038
0.0
-0.003
-0.047
-0.036
0.145
0.103
Differencebetweengenotypeand phenotypeis statistically significant (P <0.05).
ently misclassified
by either of these techniques.
This initial
number decreased to 32 (3.07%) after repeating the nonconcordant samples by both methods.
Misclassification
of apoE isoforms was related to both the
PCR and the IEF techniques.
The most common
misassignment, after genotype analysis, was due to the absence of one E4
band in the first genotype, probably secondary to poor amplification. Conversely,
almost half of the 46 phenotype
modifications were due to faint E4 bands and almost one third had an
extra E2 band. Similar results were reported by Mailly et al. [15].
These authors reported a reduction of discordant
samples from
14 to 5, after reanalyzing
the phenotypes
and genotypes
of
nonconcordant
samples.
These
same authors
could further
reduce the number
of nonconcordant
samples by four, after
drawing new blood samples from those subjects, suggesting that
mislabeling
of the initial tubes may have played a role in their
results. This has also been the experience
reported
by other
authors doing genetic testing for Huntington
disease [25].
Nine studies have been reported that compared
apoE phenotypes and genotypes [12-20] (Table 5). It should be noted that
some of the studies with worse concordance
were those with a
high proportion
of diabetics [13, 14, 18], suggesting
that the
effect of posttranslational
modification,
mainly nonenzymatic
glycation of apoE [26], on isoform mobility may be of particular
significance
in diabetic patients; however, when Wenham
et al.
[14] compared
the ratio of mismatches
in two populations,
insulin-dependent
diabetics vs nondiabetic
subjects, a greater
discrepancy
was observed among nondiabetics
(17% vs 13%).
Moreover,
James et al. [19] only found one mismatch
in 151
non-insulin-dependent
diabetics (0.66%), suggesting
that the
differences in the IEF procedures
may have been responsible
for
the differences
within other studies in diabetics. The pretreatment of VLDL samples with neuraminidase
does not seem to
affect acidic isoforms of apoE in diabetics [13, 26]. The number
of diabetic patients in our study population was insufficient (n =
32, 3.07%) to demonstrate
whether
diabetic
status plays a
significant role in phenotype
interpretation.
Stavljenic-Rukavina
et al. [18] found a different number of
mismatches
depending
on the genotyping
method, being higher
with allele-specific
oligonucieotide
probe (ASO) than with
amplification
refractory
mutation
system (ARMS). However,
when PCR plus restriction
isotyping was compared with singlestrand conformation
polymorphism
(SSCP) or with ASO, the
agreement
was complete [27, 28].
In this large population
sample, no significant
differences
were found between allele distributions
before and after correction of phenotype-genotype
misassignments.
Our results also
suggest that apoE allele assignment
by phenotype
or genotype
will yield similar allele frequency distributions,
suggesting
that
the apoE allele frequencies
previously reported in large population studies [29] may represent
the true allele frequencies
in
the population,
despite the probable
occurrence
of a certain
number
of misidentifications.
More problematic
may be the
interpretation
of the data presented
in small population
studies
in which even a low number of misclassifications
may significantly affect the final allele distribution.
One of the most important
outcomes of previous population
studies examining
apoE isoforms
has been the association
between apoE alleles and LDL-C
concentrations.
Our results
indicate that similar lowering effects were associated with the
apoE2 allele as determined
by genotype or phenotype.
This was
also true regarding the raising effect associated with the apoE4
allele. Most intriguing
was the significant
difference
found in
the average effect of #{128}2
and #{128}4
alleles on triglyceride
concentrations, depending on genotype or phenotype
assignment.
This
feature could cast doubt on the effect of apoE alleles on
triglyceride
concentrations
as determined
only by phenotype,
as
previously reported [21, 30, 31].
The multiple steps involved during genotyping
or phenotyping may be the major source of misclassification.
Human errors
Table 4. Average effect of #{128}2,
#{128}3,
and #{128}4
alleles, genotype and phenotype assigned, on total cholesterol
triglycerides (TG), and HDL-C in 32 subjects with discrepancies after verification (mmol/L).
LDL.C
TC
Genotype
#{128}2
-0.592
#{128}3
0.0
#{128}4
0.566
a
Phenotype
Phenotype
0.272
-0.166
0.300
Genotype
-0.517
-0.008
0.597
ma
Phenotype
0.212
-0.186
0.390
Allele effects Statistically different between genotype and phenotype (P
<
0.05).
Genotype
(TC), LDL-C,
HDL.C
Phenotype
Genotype
-0.690
0.178
0.240
-0.075
-0.019
0.113
0.0
-0.031
0.353
-0.469
-0.194
Phenotype
0.124
Clinical
Chemistry42, No.
Table 5. Studies reported comparing apeE phenotypes
References
Kontulaet al.(12]
Snowden et al.[13)
Wenham et al.[14]
n
Phenotyping
method
1821
11, 1996
Genotyplng
method
40
S-IEF+IB
RI
95
52
V.IEF+CBS
P-IEF+IB
ASO
RI
195
104
47
V-IEF+IB
V-IEF+IB
V-IEF+CBS
ASO
RI
50
V-IEF+CBS
and genotypes.
Mlsmatchesa
58
Maillyet al.[15]
Tsukamoto et al.(16]
Tsaiet al.[17]
Stavljenic-Rukavina
et
aI.[18]
James et al.[19]
Hansen et al. (20]
Presentstudy
151
460
1041
2.5%
16%
13%
Comments
NIDDM
DID
17%
7.1%-*0.2%
4.8%
SScP
lO.6%*6.4%L
ARMS
ASO
lO.6%*6.4%t
ARMS
RI
S-IEF+IB
P-IEF-IB
RI
RI
V-lEE-CBS
24%
DID
16%
DID
NIDDM
0.6%
2%_*1.1%c
9.4%-’3.1%
P-IEF+IB
some studies
before and after
repeating phenotype and genotype.
b Only phenotype was repeated.
C After rereading the immunoblots.
a In
CBS, Coomassie Blue staining; lB. immunoblotting; RI, restriction
isotyping; SSCP, single-strand conformational polymorphism; ARMS, amplification refractory
mutation system; DID, diabetic insulin-dependent patients; NIDDM, non-insulin-dependent diabetic patients; S. serum;V, VLDL; P, plasma.
may occur at several points in this process and they apply to both
PCR and IEF. These include labeling of the tubes, interpretation of hand-written
numbers, transcription
of the numbers to
the assay tubes, placement
of assay samples into gels, interpretation of the results, and introduction
of the results in the
database.
These problems
have been previously
reported
in
testing for Huntington
disease [25], and they underscore
the
importance
of precise labeling, interpretation
of the results by
several investigators,
and verification
of data entry accuracy.
Moreover,
duplicate sampling should be carried out to reduce
the need to obtain new samples later. Additional sources of error
are associated specifically with each of these techniques.
Thus,
genotyping
assessment
may be affected by the quality of the
DNA,
especially when high-throughput
methods are being used
[32]. Poor amplification
may result in bands with very low
intensity or artifactual
bands. This is especially important
for
apoE restriction
isotyping [10], due to the low-molecular-mass
bands used to identify the apoE2/3 and especially the apoE2/2
genotypes; however, this can also affect the interpretation
of the
apoE4 allele. Several other PCR-based
methods
have been
successfully adapted to the study of apoE mutations [13, 17, 18].
In the case of phenotyping
by IEF, the technical problems
will be different for IEF of VLDL and protein staining or total
plasma and immunoblotting.
In the first case, VLDL should be
isolated as soon as possible from fresh plasma. In our experience,
if VLDL is not used within 2 days of its isolation, the IEF profile
presents an enrichment
of E2-like bands. Furthermore,
samples
with very low triglyceride
concentrations,
and consequently
low
VLDL concentrations,
did not provide enough material to carry
out successful staining, especially when minigels were used, due
to the limited loading capacity of these gels. Interpretation
of
the IEF gels may also be affected by the presence
of other
proteins within the same isoelectric point range that could be
misinterpreted as apoE isoforms.
Finally, a different distribution
apoE isoforms has been described between the VLDL and
HDL fractions. This property of the isoforms may alsoaffect
the phenotype interpretation
when VLDL is used.
Direct phenotype assessment with whole plasma and specific
antibodiesavoidsmany of these problems, and several variants
of this technique have been recentlypublished[33-35]. By using
whole plasma, no bias is introduced
because of the different
distribution
of apoE between lipoprotein
fractions,
and the
problem
associated
with low \TLDL
samples is eliminated.
Furthermore,
the specificity obtained by immunostaining avoids
the problem associatedwith other proteins focusing within the
of
same range.However, the antibody should be tested to ensure
similar reactivity
with all differentisoforms.
Some discrepancies
that remain after careful assessment
of
the technicalerrorsmay be due to true rare genetic mutations.
Over 20 rare apoE alleles have been detected by either genotyping or phenotyping [36, 37]. Although
phenotyping may
detect most of these mutations,only the careful selection
of
PCR primers, specific probes,or directsequencing
permits the
identification
of these mutations
with DNA analysis. As a first
step to investigate
this possibility, we examined the lipid profile
of the 32 subjects
for which the lack of discrepancy
remained
after verification
of phenotype
and genotype.
When the apoE
allele-associated
effect on plasma lipid concentrations
was examined in these subjects, we found that the genotype assignment
followed the normal trend; i.e., LDL-C
concentrations
were
lower in E2 subjects and greater in E4 subjects. However, when
the same calculations
were carried out on the basis of the
phenotype
assignment,
a different picture emerged. The apoE2
allelewas associated with higher LDL-C and triglyceride
concentrations.
In the case of the apoE4 allele a somewhat reduced
hypercholesterolemic effect was noted and a significant triglyc-
1822
Apo E phenotype-genotype
eride lowering effect was noted. Furthermore,
the VLDL-C/
triglyceride
ratio in nine of these subjects was >0.3, although
none of them was classified as an E2/2 by either phenotype
or
genotype.
This data suggests that a large proportion
of these
subjects (28%) had a lipoprotein
profile consistent with -dyslipoproteinemia.
This percentage
was significantly
higher than
that observed
for the entire population
(14.4%) (P = 0.03).
These findings suggest that other rare apoE mutations
may be
present in this population
and may be an important determinant
of lipid disorders,
population
[38].
as we have
recently
shown
in a different
discrepancies
7.
8.
9.
10.
In conclusion,
at the population
level, both apoE typing methods resulted in similar apoE allele frequencies
and allelic effects
on total cholesterol,
LDL-C,
and HDL-C
concentrations,
although
this was not true for triglyceride
concentrations.
Despite the similarities,
these results and those reported
by
other authors should raise some concern at the individual level.
Most of the errors may be attributable
to sample misidentification and handling, interpretation
of the results, and data entry.
The technical errors associated with PCR may be reduced if a
good-quality
DNA is used and good amplification
is achieved.
Phenotyping
should be carried out with whole-plasma
IEF
followed by immunoblotting
with polyclonal
antibodies
that
recognize
all apoE isoforms.
In our experience,
genotyping
should be the method of choice for population
studies and for
specific assessment
of the common
apoE alleles, especially to
detect risk of coronary
artery disease or Alzheimer
disease.
Phenotyping
should be reserved for those situations in which
DNA is not available or when the purpose is to identify the
presence
of other
rare alleles.
This work was supported
by contract
HV-83-03
from the
National Institutes of Health and contract 53-k06-5-l0
from the
US Department
of Agriculture
Research
Service. C.L. was
supported
by a fellowship
of the Fondo de Investigaciones
Sanitarias
(94/5629),
Spanish
Ministry of Health,
Madrid,
Spain.
References
1. Mahley RW. Apolipoprotein
2.
3.
4.
5.
6.
E: cholesterol transport protein with
expanding role in cell biology. Science 1988;240:622-30.
RaIl SC, Weisgraber KH, Mahley RW. Human apolipoprotein E: the
complete amino acid sequence. J Biol Chem 1982;257:4171-8.
Schaefer EJ, Lamon-Fava 5, Johnson 5, Ordovas JM, Schaefer
MM, Castelli WP, et al. Effects of gender and menopausal status
on the association of apolipoprotein E phenotype with plasma
lipoprotein levels: results from the Framingham Offspring Study.
Arterioscier Thromb 1994;14:1105-13.
Wilson PWF, Myers RH, Larson MG, OrdovasJM, Wolf PA, Schaefer
Li. Apolipoprotein E alleles, dyslipidemia, and coronary heart
disease:
the Framingham
Offspring Study.
JAMA 1994;
272:1666-71.
Saunders AM, Strittmatter Wi, Schmechel D, George-Hyslop 5,
Pericak-Vance MA, Joo SH, et al. Association of apolipoprotein E
allele #{128}4
with late-onset familial and sporadic Alzheimer’s disease. Neurology 1993;43:1467-72.
Lopez-Miranda J, Ordovas JM, Mata P, Lichtenstein AH, Clevidence
B, Judd JT, et al. Effect of apolipoprotein
E phenotype on dietinduced lowering of plasma low density lipoprotein cholesterol. J
Lipid Res 1994;35:1965-75.
Ordovas JM, Lopez-Miranda J, Perez-Jimenez F, Rodriguez C, Park
JS, Cole T, et al. Effect of apolipoprotein E and A-IV phenotypes on
the low density lipoprotein response to HMG C0A reductase
inhibitor therapy. Atherosclerosis 1995;113:157-66.
Ordovas JM, Litwack-Klein L, Wilson PWF, Schaefer MM, Schaefer
Li. Apolipoprotein E isoform phenotyping methodology and population frequency with identification of apoEl and apoE5 isoforms.
Lipid Res 1987;28:371-80.
Menzel HJ, Utermann G. Apolipoprotein E phenotyping from serum
by western blotting. Electrophoresis
1986;7:492-6.
Hixson JE, Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with Hhal. J Lipid Res
1990;31:545-8.
Walden CC, Hegele RA. Apolipoprotein E in hyperlipidemia. Ann
Intern Med 1994;120:1026-36.
12. Kontula K, Aalto-Setfllfl K, Kuusi 1, H#{228}m#{228}l#{228}inen
L, Syvflnen A-C.
Apolipoprotein E polymorphism determined by restriction enzyme
analysis of DNA amplified by polymerase chain reaction: convenient alternative to phenotyping by isoelectric focusing. Clin Chem
11.
1990;36:2087-92.
13. Snowden C, Houlston RS, Arif MH, Laker ME, Humphries SE,
Alberti KGMM. Disparity between apolipoprotein E phenotypes
and genotypes (as determined by polymerase chain reaction and
oligonucleotide probes) in patients with non-insulin-dependent
diabetes mellitus. Clin Chim Acta 1991;196:49-58.
14. Wenham PR, Sedky A, Spooner RJ. Apolipoprotein E phenotyping:
a word of caution. Ann Clin Biochem 1991;28:599-605.
15. Mailly F, MoIl P, Kottke BA, Kamboh Ml, Humphries SE, Ferrell RE.
Estimation of the frequency of isoform-genotype
discrepancies at
the apolipoprotein E locus in heterozygotes for the isoforms.
Genet Epidemiol 1992;9:239-48.
16. Tsukamoto K, Watanabe 1, Matsushima T, Kinoshita M, Kato H,
Hashimoto Y, et al. Determination by PCR-RftP of apo E genotype
in a Japanese population. i Lab Clin Med 1993;121:598-602.
17. Tsai MY, Suess P, Schwichtenberg
K, Eckfeldt JH, Yuan J,
Tuchman M, et al. Determination of apolipoprotein #{163}
genotypes by
single-strand conformational polymorphism. Clin Chem 1993;39:
2121-4.
18. Stavljenic-Rukavina A, Sertic J, Salzer B, Dumic M, Radica A,
Fumic K, et al. Apolipoprotein E phenotypes and genotypes as
determined by polymerase chain reaction using allele-specific
oligonucleotide probes and the amplification refractory mutation
system in children with insulin-dependent diabetes mellitus. Clin
Chim Acta 1993;216:191-8.
19. James RW, Ruiz J, Blanche H, Pometta D, Passa P. Froguel P.
Phenotyping is an accurate means of analysing the principal
apolipoprotein E isoforms. Clin Chim Acta 1994;225:77-82.
20. Hansen PS, Gerdes LU, Klausen IC, Gregersen N, Faergeman 0.
Genotyping compared with protein phenotyping of the common
apolipoprotein E polymorphism. Clin Chim Acta 1994;224:131-7.
21. Boerwinkle E, Visvikis 5, Welsh D, Steinmetz i, Hanash SM, Sing
CF. The use of measured genotype information in the analysis of
quantitative phenotypes in man. II. The role of the apolipoprotein
E polymorphism in determining levels, variability and covariability
of cholesterol, betalipoprotein and triglycerides in a sample of
unrelated individuals. Am J Med Genet 1987;27:567-82.
22. Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP.
An investigation of coronary heart disease in families: the Framingham Offspring Study. Am J Epidemiol 1979;11O:281-90.
23. Miller SA, Dykes DD, Polesky HF. A simple salting out procedure
for extracting DNA from human nucleated cells. Nucleic Acids Res
1989;16:1215.
Clinical Chemistry
24.
25.
26.
27.
28.
Warnick R, Benderson I, Albers JJ. Dextran sulfate-Mg2
precipitation procedure for quantitation of high-density-lipoprotein
cholesterol. Clin Chem 1982;28:1379-88.
Morrison PJ, Graham CA, Nevin NC. DNA storage and duplicate
sampling: lesson learnt from testing for Huntington’s disease.
J Med Genet 1993;30:1042.
Black SC, Hewett S, Kotubi Y, Brunt RV, Reckless JPD. lsoform
patterns of apolipoprotein E in diabetes mellitus. Diabet Med
1990;7:532-9.
Aozaki R, Kawaguchi R, Ogasu U, Hikiji K, Kubo N, Sakuraboyashi
I. Rapid identification of the common apoE isoforms genotype
using polymerase chain reaction-single strand conformation polymorphism (PCR-SSCP). Mol Cell Probes 1994;8:51-4.
Maekawa B, Cole TG, Seip RL, Bylund D. Apolipoprotein E
genotyping methods for the clinical laboratory. J Clin Lab Anal
42, No.
33.
1823
Golaz 0, Sanchez JC, James RW, Hochstrasser DF. Phenotypingof
apolipoprotein
E using immobilized pH gradient gels for onedimensional and two-dimensional
separations.
Electrophoresis
1995;16:1184-6.
Corbo RM, Scacchi R, Mureddu L, Mulas G, Alfano G. Apolipoprotein E polymorphism in Italy investigated in native plasma by a
simple polyacrilamide gel isoelectric focusing technique. Comparison with frequency data of other European populations. Ann Hum
Genet 1995;59:197-2O9.
35. Kolhmeier M, Drossel HJ, Sinha P, Kottgen E. Rapid and simple
method for the identification of apolipoprotein
E isomorphic
phenotypes from whole serum. Electrophoresis
1992;13:258-
34.
61.
36.
1995;9:63-9.
Davignon I, Gregg RE, Sing CF. Apolipoprotein E polymorphism
and atherosclerosis.
Arteriosclerosis 1988;8:1-21.
30. SingCF, Davignon J. Role of the apolipoprotein E polymorphism in
determining normal plasma lipid and lipoprotein variation. Am J
Hum Genet 1985;37:268-85.
31. Eto M, Watanabe K, lshii K. Reciprocal effects of apolipoprotein E
alleles (e2 and e4) on plasma lipid levels in normolipidemic
subjects. Clin Genet 1986;29:477-84.
32. BoIla MK, Haddad L, Humphries SE, Winder AF, Day IN. Highthroughput method for determination of apolipoprotein E genotypes with use of restriction digestion analysis by micropl ate array
diagonal gel electrophoresis.
Clin Chem 1995;41:1599-6O4.
11, 1996
29.
Richard P. Thomas G, de Zulueta MP, De gennes JL, Thomas M,
Cassaigne A, et al. Common and rare genotypes of human
apolipoprotein E determined by specific restriction profiles of
polymerase chain reaction-amplified
DNA. Clin Chem 1994;4O:
24-9.
37. Zannis VI, Kardassis D, Zanni EE. Genetic mutations affecting
human lipoproteins, their receptors, and their enzymes. In: Harris
H, Hirschhorn K, eds. Advances in human genetics, Vol. 21. New
York: Plenum Press, 1993:145-316.
38.
Pocovi M, Cenarro A, Civeira F, Myers RH, Casao E, Esteban M,
Ordovas JM. Incomplete dominance of type Ill hyperlipoproteinemia is associated with the rare apolipoprotein E2 (Arg136-*Ser)
variant in multigenerational
pedigree studies. Atherosclerosis
1996;122:33-46.