Clinical Chemistry 45:8
1214 –1223 (1999)
Lipids and
Lipoproteins
Estrogen Replacement Therapy, Serum Lipids, and
Polymorphism of the Apolipoprotein E Gene
Philip J. Garry,1* Richard N. Baumgartner,2 Steven G. Brodie,3 George D. Montoya,4
Hwa Chi Liang,4 Robert D. Lindeman,2 and Thomas M. Williams1
Background: Pharmacogenomics, the study of genetic
loci that modulate drug responsiveness, may help to
explain why estrogen replacement therapy (ERT) has
differential effects on serum lipid and lipoprotein concentrations in postmenopausal women who inherit distinct alleles of the apolipoprotein E gene (APOE).
Methods: We compared total-cholesterol, triglyceride,
and lipoprotein (LDL and HDL) concentrations in 66
postmenopausal women receiving ERT ([1]ERT) with
174 postmenopausal women not receiving ERT
([2]ERT), controlling for three APOE genotypes divided
into three groups: E2 (e2/e3, n 5 31), E3 (e3/e3, n 5 160),
and E4 (e3/e4 1 e4/e4, n 5 49).
Results: Mean total-cholesterol concentrations were
lower in all three [1]ERT groups compared with their
[2]ERT counterparts but were statistically significant
only for women in group E4 (P 5 0.014). The mean
LDL-cholesterol concentrations were significantly lower
in all three [1]ERT groups compared with their [2]ERT
counterparts (P <0.005). Although all three groups of
[1]ERT women tended to have higher mean HDLcholesterol concentrations compared with their [2]ERT
counterparts, the differences were not statistically significant. [1]ERT women in groups E2 and E3 had
significantly higher (P <0.05) triglyceride concentrations than their [2]ERT counterparts. In [1]ERT
women, the ratios of total and LDL-cholesterol to HDLcholesterol were significantly higher in group E3 and E4
women compared with E2 women (P <0.006). Group E4
[1]ERT women had ratios of total and LDL-cholesterol
Departments of 1 Pathology and 2 Medicine, and 4 Clinical Nutrition
Program, University of New Mexico School of Medicine, Albuquerque, NM
87131.
3
Genetics of Development and Disease Branch, National Institute of
Diabetes, Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892.
*Address correspondence to this author at: Department of Pathology,
Room 215, Surge Bldg., University of New Mexico Health Sciences Center,
2701 Frontier Place NE, Albuquerque, NM 87131. Fax 505-272-9135; e-mail
[email protected].
Received November 10, 1998; accepted May 6, 1999.
to HDL-cholesterol that were comparable to group E2
[2]ERT women.
Conclusions: Triglyceride concentrations in group E2
[1]ERT women may need to be monitored more closely
than those in E3 or E4 [1]ERT women. Group E4 women
should probably be targeted for ERT. Results suggest
that APOE genotypes have a differential effect on serum
lipids and lipoproteins in [1]ERT postmenopausal
women.
© 1999 American Association for Clinical Chemistry
Coronary heart disease (CHD)5 accounts for 250 000
deaths per year in American women, with most deaths
occurring after menopause (1 ). A major risk factor for
CHD in older women is the lack of ovarian estrogen.
Clinical trials and epidemiological studies have shown
that estrogen replacement therapy (ERT) reduces the
incidence of CHD in postmenopausal women (2–5 ). One
of the beneficial effects of ERT has been attributed to
known estrogen effects on lowering LDL-cholesterol
(LDL-C) and increasing HDL-cholesterol (HDL-C) concentrations (6 ).
Recognition of apolipoprotein E (apoE) polymorphism
as one determinant of serum lipid and lipoprotein concentrations has led to research on the relationship between apoE phenotypes and CHD risk (7 ). apoE is a
34-kDa protein that functions in the redistribution of
lipids among cells of various organs, based on its ability to
bind to two lipoprotein receptors, namely the LDL receptor and the non-LDL receptor, sometimes referred to as
the remnant receptor or LDL-related receptor protein
(LRP) (8 ). apoE plays central roles in mammalian cholesterol transport by serving as a ligand for the cell surface
5
Nonstandard abbreviations: CHD, coronary heart disease; ERT, estrogen
replacement therapy; LDL-C, LDL-cholesterol; HDL-C, HDL-cholesterol; apo,
apolipoprotein; LRP, LDL-related receptor protein; HRT, hormone replacement therapy; BMI, body mass index; NMAPS, New Mexico Aging Process
Study; NMMRL, New Mexico Medical Reference Laboratory; CNL, Clinical
Nutrition Laboratory; LPL, lipoprotein lipase; HTGL, hepatic lipase; and IDL,
intermediate-density lipoprotein.
1214
1215
Clinical Chemistry 45, No. 8, 1999
LDL and LRP receptors that mediate the endocytosis of
apoB- and apoE-containing lipoproteins. apoE differentially modulates blood lipid concentrations via three
different isoforms, apoE2, E3, and E4 (9 ). Charge and
sulfhydryl differences among the isoforms are thought to
be responsible for differences in binding affinity for the
LRP and LDL receptor in the liver (9 ). The three APOE
alleles, e2, e3, and e4, determine six genotypes: three
homozygous, identified as e2/e2, e3/e3, and e4/e4; and
three heterozygous, identified as e2/e3, e3/e4, and e2/e4.
Several studies have examined the relationship between apoE isoforms and blood lipids in postmenopausal
women not receiving ERT and those receiving ERT. In
1994, Schaefer et al. (10 ) examined the association of
APOE genotypes with plasma lipoproteins in pre- and
postmenopausal women in the Framingham Offspring
Study who were not receiving any type of hormone
replacement therapy (HRT). Schaefer et al. (10 ) found that
total-cholesterol and LDL-C concentrations were 12.7%
and 19.7% higher, respectively, in postmenopausal
women with an e3/e4 genotype compared with postmenopausal women with an e2/e3 genotype. The fact that
premenopausal women had similar total-cholesterol and
LDL-C values across all three genotypes and that postmenopausal women had variable increases in their total
cholesterol and LDL-C would suggest that estrogens
modulate total-cholesterol and LDL-C concentrations, depending on APOE genotype. Somekawa and Wakabayashi (11 ) were the first to examine the association of
apoE polymorphism and lipid profiles in three groups of
postmenopausal women based on their genotype, E2
(e2/e2 1 e2/e3, n 5 14), E3 (e3/e3, n 5 170), and E4 (e3/e4
1 e4/e4, n 5 52). One hundred fifty-two women (age
range, 42– 60 years) completed the study. These women
received HRT (0.625 mg/day conjugated equine estrogen
and 2.5 mg/day medroxyprogesterone acetate) continuously for 6 months. Serum lipids and lipoproteins were
compared before and after HRT. Total cholesterol and
LDL-C, but not HDL-C, were significantly reduced by
HRT in all three groups. The LDL-C/HDL-C ratio was
calculated to determine the risk of atherosclerosis before
and after the initiation of HRT in the three groups. The
ratio improved from 2.18 before HRT to 1.58 after HRT in
group E2, from 2.26 to 1.92 in group E3, and from 2.57 to
2.10 in group E4 women. Of interest was the finding that
even before HRT, the LDL-C/HDL-C ratio in group E2
women was similar to those in group E4 after HRT. Serum
triglycerides were minimally affected by HRT in these
Japanese women; however, unlike unopposed estrogen
therapy, combination therapy tends to blunt the increase
in triglycerides, as was the case in this study. Somekawa
and Wakabayashi (11 ) concluded from this study that
group E4 women had the highest risk of CHD and that
group E2 women the lowest risk. In addition, they suggested that oral HRT (conjugated equine estrogen and
medroxyprogesterone) be recommended for those Japanese women with an APOE e4 allele.
The present study, unlike the clinical trial of Somekawa
and Wakabayashi (11 ), reports observed differences in
lipid and lipoprotein concentrations in postmenopausal
women receiving ERT by APOE genotype after controlling for age and body mass index (BMI). The women’s
personal physicians prescribed unopposed estrogen use
or combination therapy (estrogen plus progestin) completely independently of the study protocol. Our data
suggest that women who inherit distinct alleles of the
APOE gene may require individualized ERT or combination therapy.
Materials and Methods
subjects
Subjects for this study were from the Albuquerque, NM
vicinity and enrolled in the New Mexico Aging Process
Study (NMAPS). Participation in the NMAPS is entirely
voluntary and limited to noninstitutionalized men and
women over 60 years of age who are basically healthy for
their age. For example, individuals with diagnosed CHD
and insulin-dependent diabetes and those with uncontrolled hypertension were not enrolled in this study. In
1996, the median (6SD) age was 76.2 6 7.1 years, and all
participants were Caucasian, with 3% being of Spanish/
Hispanic descent. The data in the present report were
obtained from all women who were enrolled in the
NMAPS and examined in 1996.
Although a more complete description of the NMAPS
population can be found in previous publications (12–14 ),
it is important to point out that this population is not a
random sample of elderly living in Albuquerque, NM.
This is a select group of elderly individuals who are, for
the most part, health-conscious, well-educated, financially
secure, and highly motivated participants of this longitudinal study. All inferences from these data must be made
with this in mind. Women receiving ERT or combination
therapy began their therapy after consulting with their
personal physicians and were not influenced to do so by
investigators in the NMAPS. None of the women were
taking any type of cholesterol-lowering drugs. None of
the six women with an APOE e2/e3 genotype receiving
ERT were taking progestins. Eleven of the 44 women
(25%) with an APOE e3/e3 genotype and receiving ERT
were taking progestins, and 4 of the 16 women (25%) with
an APOE e3/e4 or e4/e4 genotype were also taking
progestins. Fifty-six of the 66 women (85%) were taking
PremarinTM (Wyeth-Ayerst), 5 were taking EstratabTM
(Solvay Pharmaceuticals), 2 were taking EstraceTM (Bristol-Myers Squibb), and 2 were taking other forms of
estrogen. Of the 240 women in this study, 5 had had
a hysterectomy, 3 of whom were receiving ERT and 2
were not.
All volunteers were seen as outpatients each year in the
Clinical Nutrition Program laboratory at the University of
New Mexico Health Sciences Center. Outpatient visits
were distributed throughout the year. The Human Research Review Committee of the University of New
1216
Garry et al.: ERT, Lipids, and APOE
Mexico School of Medicine approved the study. Informed
consent was obtained from each participant.
laboratory measurements
Blood chemistries. After an overnight fast, ;50 mL of blood
was obtained from each person between 0800 and 0930 for
various biochemical measurements. Blood was drawn at
the annual outpatient visit. Analytical measurements
were performed in two laboratories: the New Mexico
Medical Reference Laboratory (NMMRL), located in Albuquerque, NM, and the Clinical Nutrition Laboratory
(CNL) at the University of New Mexico School of Medicine. Tests performed by the NMMRL were serum cholesterol and triglycerides. Cholesterol and triglyceride
assays were conducted on a Vitros 950 Analyzer, using
dry-slide technology (J & J Orthodiagnostic Products).
The NMMRL analyzed all blood samples on the day they
were drawn. Quality-control samples were included with
each batch of test specimens for monitoring accuracy and
precision of biochemical tests. Commercial preassayed
controls were obtained from the College of American
Pathologists. HDL-C assays were performed in the CNL
with an ABA-100 Biochromatic Analyzer (Abbott Laboratories), using a Sigma HDL-C kit. A modified heparinmanganese method was used for the precipitation of LDLand VLDL-cholesterol (15 ). We calculated the LDL-C
concentration by subtracting the HDL-C and VLDL-cholesterol concentrations from the total serum cholesterol
concentration. We estimated the VLDL-cholesterol concentration using the formula of Friedewald et al. (16 ),
which assumes that the concentration of VLDL-cholesterol approximates one-fifth of the plasma triglyceride
concentration when triglyceride concentrations are ,4
g/L (,400 mg/dL). Control samples provided by the
CDC Lipid Laboratory, Atlanta, GA, were included to
monitor accuracy and precision of the HDL-C assays.
Serum estrone assays were conducted in the CNL using
an ICN Pharmaceuticals RIA kit. The minimum detection
limit of this kit is 1.2 ng/L (1.2 pg/mL), and the interassay
CV, as determined by the manufacturer, is 10.2% at an
estrone concentration of 90 ng/L (90 pg/mL). We measured serum estrone because some women, when questioned about their use of exogenous estrogens, equate
ERT with occasional use of preparations such as Estrace
or Premarin or using transdermal estrogen patches and
vaginal creams, which do not have the same effect as oral
estrogens in raising serum estrone concentrations (17 ).
When women in the NMAPS were asked whether they
used exogenous estrogens, 31% (n 5 74) responded yes to
this inquiry. For those women who answered no to this
question (n 5 166), the mean (6 SE) serum estrone
concentration was 8.8 6 0.47 ng/L (8.8 6 0.47 pg/mL),
with a range of 0.90 –29.7 ng/L (0.90 –29.7 pg/mL). Thus,
we chose an estrone value of .30.0 ng/L (.30.0 pg/mL)
as the concentrations required to put women in the ERT
category. This reduced the number of women from 74 to
66 whom we felt assured were regular ERT users. All of
these 66 women had received ERT for at least 1 year.
apoE isotyping. Restriction fragment length polymorphism
analysis of PCR products was used to assess APOE
genotypes according to the procedure of Hixson and
Vernier (18 ). This assay was conducted in the CNL.
Statistical analyses. The balanced-gene estimates of the
APOE allele frequencies were calculated as follows. For
example, APOE e4 (frequency) 5 APOE e4/e4 1½(APOE
e2/e4 1 APOE e3/e4). The summary of descriptive statistics was reported. We applied the ANOVA model with
age and BMI adjusted to assess the effect of ERT and
genotype as well as their interaction on each lipid fraction.
To understand the change in magnitude with each lipid
fraction that could be attributed to ERT and genotype, we
fit a multiple regression model with predictors including
age, BMI, and each combination of ERT and genotype,
with the reference group being women with the APOE
e3/e3 genotype not receiving ERT. Loge transformations
were performed on serum triglyceride and estrone concentrations to achieve normality of distributions. All analyses were performed using SAS (release 6.12).
Results
The number of the various APOE genotypes and the allele
frequencies for women (n 5 248) examined in the NMAPS
are shown in Table 1. There were no significant differences in relative allele frequencies between women not
taking exogenous estrogens ([2]ERT) and women receiving ERT ([1]ERT), and the relative frequencies were all in
Hardy-Weinberg equilibrium. In addition, allele frequencies were similar to those in other healthy Caucasian
Table 1. APOE genotypes and allele frequencies of NMAPS women.
Allele frequencies
Genotype
«2/«2
«3/«3
«4/«4
«2/«3
«2/«4
«3/«4
No. of subjects
% of total
Allele
All women
(n 5 248)
[2]ERT women
(n 5 180)
[1]ERT women
(n 5 68)
0
160
4
31
8
45
0
64.52
1.61
12.50
3.23
18.15
«2
«3
«4
7.86%
79.84%
12.30%
8.61%
80.00%
11.39%
5.88%
80.15%
13.97%
1217
Clinical Chemistry 45, No. 8, 1999
Table 2. Summary of ages, BMI, serum estrone and lipids, and ratios of total cholesterol and LDL-C to HDL-C of women in
the NMAPS, according to ERT status: receiving ([1]ERT) or not receiving ([2]ERT) ERT.
[2]ERT
n
Age, years
BMI, kg/m2
Estrone, ng/L
Total cholesterol, mmol/Lb
LDL-C, mmol/L
HDL-C, mmol/L
Loge triglycerides, mmol/L
Triglycerides, mmol/Ld
Total cholesterol/HDL-C ratio
LDL-C/HDL-C ratio
174
174
174
174
171
174
174
174
174
171
[1]ERT
Mean (SE)
77.4
25.1
8.8
5.72
3.55
1.46
0.351
1.42
4.13
2.55
n
(0.55)
(0.32)
(0.47)
(0.07)
(0.06)
(0.03)
(0.035)
(1.03)
(0.08)
(0.07)
Pa
Mean (SE)
66
66
66
66
66
66
66
66
66
66
73.2
25.8
124
5.39
2.95
1.59
0.501
1.65
3.54
1.96
(0.82)
(0.53)
(11.5)
(0.11)
(0.10)
(0.04)
(0.057)
(1.06)
(0.14)
(0.10)
,0.0001
0.759
,0.0001
0.0138c
0.0001c
0.0173c
0.0283c
0.0003c
0.0001c
a
Represents differences between those receiving and those not receiving ERT.
To convert mmol/L values for cholesterol to mg/dL divide by 0.02586.
c
Adjusted for age and BMI.
d
To convert mmol/L values for triglycerides to mg/dL divide by 0.01129.
b
women (19 ). In subsequent analyses, the women were
divided into three groups. Group E2 included 31 women
with an APOE e2/e3 genotype, group E3 included 160
women with an APOE e3/e3 genotype, and group E4
included 36 women with an APOE e3/e4 genotype and 4
women with an APOE e4/e4 genotype. Eight women with
an APOE e2/e4 genotype were not included in subsequent analyses because we chose not to assign these
women to either group E2 or E4. Three women with
fasting triglyceride concentrations .4.92 mmol/L (400
mg/dL; one in group E4 and two in group E3) were
excluded from the LDL-C calculation. Table 2 shows the
mean age, BMI, serum lipid concentrations, and estrone
concentrations for the women enrolled in the NMAPS,
separated into women receiving ERT ([1]ERT) and
women not receiving ERT ([2]ERT), and excluding the
eight women with an APOE e2/e4 genotype. Women
receiving ERT ([1]ERT) were significantly younger than
[2]ERT women. After adjustments for age and BMI, the
mean total-cholesterol and LDL-C concentrations were
significantly lower, whereas HDL-C concentrations were
significantly higher, in [1]ERT women compared with
[2]ERT women, P 5 0.0138, 0.0001, and 0.0.0173, respectively. Loge-transformed triglyceride values were significantly higher in [1]ERT women compared with [2]ERT
women, P 5 0.0283.
The results for multiple regression analyses used to test
whether serum lipids and lipoprotein concentrations were
modified by the following variables (APOE, ERT, age, and
BMI) and to test for interaction effects between APOE and
ERT on lipid fractions are shown in Table 3. Group E3
(e3/e3) women not receiving ERT ([2]ERT) were used as
the reference group (see Table 3). When other covariates
were held constant in the multiple regression model, age
had a negative association only with total cholesterol
(b-coefficient 5 20.017 6 008, P 5 0.045). BMI had a
negative association with HDL-C but a positive association with triglyceride concentrations (b-coefficients 5
20.025 6 0.005, P 5 0.0001; and 0.025 6 0.007, P 5 0.0004,
respectively). Women receiving ERT, regardless of genotype, showed a significant negative association with
LDL-C, P ,0.02. The negative association between ERT
Table 3. Multiple regression of total cholesterol, LDL-C, and HDL-C and loge triglycerides for each separate APOE genotype,
controlling independently for ERT, age, and BMI.a
Total cholesterol, mmol/L
Independent
variable
b-Coefficient
Intercept
Group E2, [1]ERT
Group E3, [1]ERT
Group E4, [1]ERT
Group E2, [2]ERT
Group E4, [2]ERT
Age, years
BMI, kg/m2
7.17 6 0.81
20.67 6 0.24
20.22 6 0.16
20.48 6 0.24
20.23 6 0.20
0.20 6 0.18
20.017 6 0.008
20.006 6 0.014
a
b
b
LDL-C, mmol/L
P
b-Coefficient
0.075
0.172
0.048
0.247
0.260
0.045
0.663
4.63 6 0.72
21.51 6 0.33
20.51 6 0.14
20.65 6 0.22
20.43 6 0.17
0.13 6 0.16
20.013 6 0.007
20.002 6 0.012
The reference group included women in group E3 («3/«3) not receiving ERT.
Estimate 6 stranded error of the estimate.
Loge triglycerides, mmol/L
HDL-C, mmol/L
P
b-Coefficient
0.0001
0.0004
0.0032
0.0134
0.4134
0.0796
0.8875
2.16 6 0.31
0.37 6 0.14
0.10 6 0.06
0.11 6 0.09
0.15 6 0.07
20.06 6 0.07
20.001 6 0.003
20.025 6 0.005
P
b-Coefficient
P
0.0110
0.0861
0.2306
0.0533
0.3533
0.7617
0.0001
20.030
0.51 6 0.19
0.22 6 0.08
0.03 6 0.12
0.10 6 0.10
0.19 6 0.09
20.004 6 0.004
0.025 6 0.007
0.0082
0.0067
0.8068
0.3420
0.0377
0.3482
0.0004
1218
Garry et al.: ERT, Lipids, and APOE
and LDL-C in each of the genotypes paralleled the
changes noted for total cholesterol. ERT had variable
associations with HDL-C in groups E2, E3, and E4. Of
interest was the finding that ERT had significant positive
associations with triglyceride concentrations in group E2
(P 5 0.0082) and E3 women (P 5 0.0067), but not in group
E4 women (P 5 0.8068), that were independent of age and
BMI. We next examined for interactive effects between
APOE and ERT on serum lipids and lipoproteins by fitting
an ANOVA model, controlling for age and BMI. The only
significant interaction was for triglycerides, P 5 0.026.
This is demonstrated in Fig. 4.
The observed differences between [2]ERT and [1]ERT
women for mean total-cholesterol, LDL-C, HDL-C, and
triglyceride concentrations within each apoE group, adjusted for age and BMI, are shown in Fig. 1. Women
receiving ERT tended to have lower total cholesterol (Fig.
1A) and higher HDL-C (Fig. 1C) than [2]ERT women, but
the only statistically significant finding was that group E4
[1]ERT women had significantly lower total cholesterol
than group E4 [2]ERT women (P 5 0.014). [2]ERT
women in groups E3 and E4 had significantly higher
LDL-C than [2]ERT women in group E2, P 5 0.0079 and
0.0031, respectively (see Fig. 1B). All three groups of
[1]ERT women had significantly lower LDL-C than their
[2]ERT counterparts (P #0.005). HDL-C concentrations
were higher, but not significantly higher, in all three
groups of [1]ERT women compared with their [2]ERT
counterparts. We also observed that group E2 [2]ERT
women had significantly higher mean HDL-C than either
group E3 or E4 [2]ERT women, P 5 0.0533 and 0.0219,
respectively. The observed effect of ERT on triglyceride
concentrations is shown in Fig. 1D. Women in groups E2
and E3 receiving ERT had significantly higher triglycerides than group E4 [1]ERT women, P 5 0.0489 and
0.0067, respectively. There were no differences in triglyceride concentrations between E4 [2]ERT and E4 [1]ERT
women.
A summary of the percentage of difference (1 or 2) in
the serum lipid fractions in group E2, E3, and E4 women
receiving ERT compared with their counterparts not receiving ERT is provided in Fig. 2. Women in group E2
who were receiving ERT showed the greatest decrease in
LDL-C (;34%) and also the greatest increase in triglyceride concentrations (;50%).
Ratios of total cholesterol or LDL-C to HDL-C have
been proposed as important determinants of CHD relative risk (20, 21 ). We observed that each group of women
receiving ERT had significantly reduced ratios with the
exception of total/HDL-cholesterol in group E2 women
(Fig. 3).
The correlations between the natural log (loge) of
serum estrone and loge serum triglyceride concentrations
for each of the apoE groups are shown in Fig. 4. One
woman in group E2 [triglyceride concentration, 3.61
mmol/L; estrone concentration, 1.1 ng/L (1.1 pg/mL)]
and one in group E4 [triglyceride concentration, 7.86
mmol/L; estrone concentration, 2.2 ng/L (2.2 pg/mL)]
were determined to be outliers and were not included in
the regression analyses. In group E2 women, 31.1% of the
variance in triglyceride concentrations could be explained
by serum estrone concentrations. When we eliminated 11
group E3 and 4 group E4 women who were receiving
combination therapy (estrogen plus progesterone) from
the regression analyses shown in Fig. 4, the slope relationships between the three apoE groups remained the same,
but slightly lower triglyceride values were measured at a
given estrone concentration. For example, at an estrone
concentration of 100 ng/L (100 pg/mL; loge 4.6), triglyceride concentrations were lower in group E3 and E4
women receiving combination therapy compared with
those receiving only ERT [0.12 mmol/L (10 mg/dL) and
0.07 mmol/L (6 mg/dL), respectively]. No E2 women
were receiving combination therapy. Serum estrone also
showed a significant negative correlation with LDL-C in
group E2 women (r 5 20.50, P ,0.05), but not in group E3
women (r 5 20.161) or E4 women (r 5 20.308; data not
shown).
Discussion
This study supports the concept that genes or loci can
influence response to a given drug. In the present study,
the “drug” of interest was ERT in postmenopausal
women, primarily conjugated estrogen, Premarin (WyethAyerst) or estradiol tablets, e.g., Estrace (Mead Johnson).
The majority of women (85%) were taking Premarin, and
the remainder took nonconjugated estrogen; ERT was
verified by measuring serum estrone concentrations. The
genetic polymorphism of interest involved alleles of the
APOE gene. The response variables were serum lipid
fractions, i.e., total, LDL-, and HDL-cholesterol and triglycerides.
Of the three groups of women studied not receiving
ERT, group E2 women have the lowest risk for CHD
because they have the lowest total cholesterol and LDL-C
and the highest HDL-C compared with either group E3 or
E4 women (see Fig. 1, B and C). Because ERT produced a
significant increase in triglyceride concentrations in group
E2 women compared with E2 women not receiving ERT,
their triglycerides may need to be monitored more closely
than either group E3 or E4 women. As noted in Fig. 2,
mean increases in triglyceride concentrations ranged from
;50% for group E2 women to ;25% in group E3 women,
with a decrease of ;17% in group E4 women receiving
ERT compared with their counterparts not receiving ERT.
One limitation of the present study was that only six
women in group E2 were receiving ERT; however, four of
these six women had hypertriglyceridemia [triglyceride
concentrations $2.46 mmol/L (200 mg/dL)].
Group E4 women should probably be targeted for ERT
because they have the highest total cholesterol and LDL-C
and the lowest HDL-C when compared with group E2 or
E3 women not receiving ERT. We also observed that
group E4 women receiving ERT had lower total choles-
Clinical Chemistry 45, No. 8, 1999
1219
Fig. 1. Summary of mean serum total cholesterol (A), LDL-C (B), HDL-C (C), and loge triglyceride (D) concentrations in women receiving (Y) or not
receiving (N) ERT.
(A), the within-group differences between women receiving ERT ([1]ERT) and women not receiving ERT ([2]ERT) were not significant for groups E2 and E3 but were
significant (P 5 0.014) for group E4. The between-group differences for [2]ERT women were not significant for a vs b or b vs c but were marginally significant (P 5 0.071)
for a vs c. (B), the within-group differences between women receiving ERT ([1]ERT) and women not receiving ERT ([2]ERT) were significant for all groups: E2 (P 5
0.0031) E3 (P 5 0.0004), and E4 (P 5 0.0018). The between-group differences for [2]ERT women were significant for a vs b (P 5 0.0079) and a vs c (P 5 0.0031)
but not significant for b vs c. (C), the within-group differences between women receiving ERT ([1]ERT) and women not receiving ERT ([2]ERT) were not significant for
any group. The between-group differences for [2]ERT women were significant for a vs b (P 5 0.0533) and a vs c (P 5 0.0219) but not significant for b vs c. (D), the
within-group differences between women receiving ERT ([1]ERT) and women not receiving ERT ([2]ERT) were significant for groups E2 (P 5 0.0489) and E3 (P 5
0.0067) but not group E4. Bars, SEM.
terol, LDL-C, and triglycerides and higher HDL-C than
group E4 women not receiving ERT.
Observational data suggest that the total-cholesterol/
HDL-C and the LDL-C/HDL-C ratios are better predictors of subsequent CHD than are individual lipid
concentrations, especially in the presence of increased
triglycerides (20, 21 ). We examined the ratio of totalcholesterol/HDL-C and LDL-C/HDL-C by APOE genotype and the effects of ERT. The effects noted for the
LDL-C/HDL-C ratio between those receiving and not
1220
Garry et al.: ERT, Lipids, and APOE
Fig. 2. Percentage of difference (1 or 2) in serum lipid fractions in
women receiving ERT compared with women not receiving ERT.
N, number of women not receiving ERT; Y, number of women receiving ERT. p, P
,0.05; pp, P ,0.004; ppp, P ,0.007; pppp, P ,0.0005.
receiving ERT were greater than noted for the total/HDLcholesterol ratio (see Fig. 3). These results show that ERT
may help group E4 women lower their ratios, either total
cholesterol or LDL-C to HDL-C, to concentrations equivalent to those in group E2 women not receiving ERT.
Wakatsuki and Sagara (22 ) reported that plasma
LDL-C concentrations showed a significant negative correlation with plasma estrone concentrations (r 5 20.64,
P , 0.001) in 20 premenopausal, 10 postmenopausal, and
10 bilaterally oophorectomized women, but did not examine for any APOE genotypic differences. Our results
depended on the various genotypes, e.g., loge serum
estrone concentrations showed a significant negative correlation with LDL-C in group E2 women but not in group
E4 women. This information may be useful for group E2
women in that a lower dose of exogenous estrogen intake
could minimize any increase in their triglycerides, and at
the same time be beneficial in lowering their LDL-C.
This observational study suggests that individualized
ERT or combination HRT, based on APOE genotype and
dose of estrogen used, may be more beneficial than
standard ERT or combination therapy, e.g., 0.625 or 1.25
mg/day conjugated estrogen with or without progestogens. Whether individualized HRT will have an increased
cardioprotective effect over standard therapy in women
with and without established CHD awaits the outcome of
well-designed case-control and cohort studies. It should
be noted that Hulley et al. (23 ) conducted a randomized,
blinded, placebo-controlled secondary prevention trial
with daily oral conjugated equine estrogen (0.625 mg)
plus medroxyprogesterone acetate (2.5 mg) or a placebo
in .2700 women with established coronary disease. Dur-
Fig. 3. Ratio of total cholesterol to HDL-C (A) and LDL-C to HDL-C (B) in
women receiving and not receiving ERT.
Bars, SEM; N, number of women not receiving ERT; Y, number of women
receiving ERT. (A), effect of ERT within groups: P 5 0.073 for a vs d; P 5 0.017
for b vs e; P 5 0.002 for c vs f. Differences between groups: P 5 0.005 for d vs
e; P 5 0.002 for d vs f; and P is not significant for e vs f. (B), effect of ERT within
groups: P 5 0.008 for a vs d; P 5 0.003 for b vs e; P 5 0.004 for c vs f.
Differences between groups: P 5 0.004 for d vs e; P 5 0.002 for d vs f; and P
is not significant for e vs f.
ing a mean follow-up of 4.1 years, HRT did not reduce the
overall rate of CHD events in postmenopausal women
with established coronary disease.
Our current knowledge of possible mechanisms that
help explain the various lipid responses to ERT or combination therapy in women with different APOE genotypes comes from human and animal studies examining
the effects of estrogens, progestogens, lipoprotein lipase
(LPL), hepatic lipase (HTGL), APOE genotypes, and
known interactions between these variables on lipoprotein metabolisms. For example, Walsh et al. (17 ) conducted a double-blind crossover study in which healthy
postmenopausal women were given only conjugated estrogen (Premarin) at two different doses, either 0.625 or
Clinical Chemistry 45, No. 8, 1999
Fig. 4. Linear regression relationship between loge serum triglyceride
concentrations and loge serum estrone concentrations for women in
groups E2 (e2/e3), E3 (e3/e3), and E4 (e3/e4 1 e4/e4).
Line 1, E2/E3 phenotype (n 5 29, r2 5 0.311); line 2, E3/E3 phenotype (n 5
160, r2 5 0.071); line 3, E3/E4 phenotype (n 5 48, r2 5 0.011).
1.25 mg/day. The results, after 3 months of treatment,
were given in terms of percentage of change (1 or 2) plus
95% confidence intervals. Walsh et al. (17 ) found that the
higher dose of Premarin, 1.25 mg/day, did not have a
substantially greater effect than the lower dose of Premarin on lowering total cholesterol or LDL-C or increasing
HDL-C. However, Walsh et al. found that Premarin doses
of 1.25 mg/day produced a mean increase in the totaltriglyceride concentration of 38% (range, 29 – 47%), which
was significantly higher (P ,0.05) than in women taking
0.625 mg/day, who had a mean increase of 24% (range,
15–33%). Our results would suggest that the variation in
triglycerides at two levels of Premarin intake noted by
Walsh et al. could be explained in part by differences in
APOE genotype. We found a strong positive correlation
between loge serum estrone and loge triglyceride concentrations in group E2 women (r 5 0.56, P ,0.01), whereas
there was no association between estrone and triglyceride
concentrations in group E4 women (Fig. 4).
The significant increase in triglyceride and decrease in
LDL-C concentrations attributable to ERT in group E2
women mimic those in transgenic mice expressing human
apoE2 (24 ). A study by Huang et al. (24 ) showed that
human-E2-induced hypertriglyceridemia in mice could be
caused by a pathway unrelated to the current hypothesis
that ERT causes increased triglycerides and low LDL-C
concentrations because of differences in binding affinity
between the apoE isoforms for the hepatic lipoprotein
receptor. Huang et al. found that increasing concentra-
1221
tions of apoE2 significantly lowered LDL-C (r 5 20.92, P
,0.001) in transgenic female mice lacking LDL receptors
by impairing the lipolysis of triglyceride-rich lipoproteins
by LPL. Huang et al. showed that apoE2-containing
VLDLs were poor substrates for lipolysis and hypothesized that apoE2 could possibly displace or mask apoC-II,
which is an absolute cofactor for LPL activity. Huang et al.
also noted that all three apoE isoforms inhibit LPL activity
to a similar extent; thus, there needs to be an explanation
for the increased triglyceride concentrations in women
receiving ERT with an APOE e2 allele and not in women
with an e4 allele. Several factors could explain this difference. Somekawa and Wakabayashi (25 ) showed that ERT
in postmenopausal women produced significantly higher
(P ,0.05) serum apoE concentrations in women with an
e2 allele than in women with an e4 allele. This could be the
result of reduced clearance of the apoE2 isoform brought
about by its lower binding affinity for the hepatic lipoprotein receptors. Although we did not have any homozygous APOE e2 women, our results show that women with
an APOE e2 allele receiving ERT have significantly higher
triglycerides and lower LDL-C concentrations than
women with an APOE e4 allele. These results indicate
that, as postulated by Huang et al. (24 ), the increased
lowering of LDL-C in women expressing an APOE e2 vs
an APOE e4 allele could involve two different mechanisms: (a) The VLDL 3 IDL 3 LDL lipolytic cascade in
women with an APOE e2 allele is slowed by increased
circulating apoE concentrations. The partial inhibition of
LPL activity attributable to ERT may be further enhanced
and produces higher triglyceride and lower LDL-C concentrations. (b) The lower binding capacity of apoE2 for
the hepatic lipoprotein receptors lowers the transport of
cholesterol-rich remnant particles into the liver, causing
an up-regulation of hepatic LDL receptors and thereby
accelerating the clearance of any LDL-C produced in the
VLDL 3 IDL 3 LDL lipolytic cascade. In women with an
APOE e4 allele, the opposing conditions would prevail,
i.e., less inhibition of LPL activity as a result of normal
catabolism of apoE and down-regulation of the LDL
receptor as a result of increased hepatic uptake of cholesterol remnants containing apoE4 by hepatic lipoprotein
receptors. Thus, women receiving ERT who have an
APOE e4 allele would tend to have an increased VLDL 3
IDL 3 LDL lipolytic cascade, with lower triglyceride and
higher LDL-C concentrations than women with an e2
allele.
ERT has also been shown to increase VLDL production
(26 ) and lower LPL and HTGL activity (22 ). The suppression of HTGL activity by ERT may decrease conversion of
intermediate-density lipoprotein (IDL) to LDL and thus
lower LDL-C (27 ). In addition, the decrease in HTGL
activity, attributable to ERT, has been shown to increase
HDL2-C concentrations in postmenopausal women (28 );
this could account for the increased HDL-C concentration
noted in the present study. Progestogens have been
shown to inhibit HTGL activity; however, progestogens
1222
Garry et al.: ERT, Lipids, and APOE
can partially restore LPL activity (29 ). Thus, combination
therapy (estrogen 1 progestogen) should reduce triglyceride concentrations compared with estrogen-only treated
women as noted in the present study.
In studies of populations not selected for health, the
presence of an APOE e2 allele is usually associated with
higher triglyceride concentrations compared with the
APOE e4 allele (9 ). Our findings, and those of Xhignesse
et al. (19 ), show an opposite trend for women not receiving ERT. For example, women in the present study with
an APOE e2 allele had a geometric mean triglyceride
concentration of 1.62 mmol/L compared with 1.77
mmol/L in those women with an e4 allele (Fig. 1D). In
contrast to what might be observed in the general population, both the women in the present study and those in
the Xhignesse et al. (19 ) study were selected on the basis
of their being healthy and had few cardiovascular risk
factors such as obesity and diabetes. Thus, caution should
be taken when comparing APOE results between a random sampling and a select group of elderly. For example,
the APOE e4 allele, which promotes premature atherosclerosis, is associated with decreased longevity (30 ). Thus,
the frequency of the APOE e4 allele would be expected to
be different in a random vs an elderly population selected
on the basis of good health.
The significant interaction between APOE and ERT in
determining triglyceride concentration is demonstrated in
Fig. 4. Although there was a trend for interactions between APOE and ERT in affecting total-cholesterol and
LDL-C concentrations, the results were not statistically
significant. A larger sample would be required to show
significant interactions. Further studies are needed to
uncover the exact mechanisms responsible for the interactions between APOE polymorphism and ERT or combination therapy in modifying triglyceride and other lipid
concentrations. There are other potential regulatory
mechanisms that could explain the differences in mean
triglyceride and lipoprotein concentrations noted in the
women in the present study receiving ERT with different
APOE genotypes.
There are several issues that need to be considered in
assessing these results. First, the present study was not a
randomized trial of ERT in elderly women, and this
should be kept in mind when assessing the results of this
study. Second is the limited sample size. As noted previously, there were few women with an APOE e2 allele
(group E2, n 5 6) who were receiving ERT. Third, there is
no way to determine what the lipid concentrations would
be in the [1]ERT women if they were not taking exogenous estrogens. However, the serum lipid concentrations
that we observed between women receiving ERT and
those not receiving ERT were similar to those reported by
Walsh et al. (17 ) in their carefully controlled study.
Fourth, there is evidence that progestogens, in combination with oral estrogens, may prevent the increase in
triglyceride concentrations (31 ). None of the women in
the NMAPS with an APOE e2/e3 genotype (group E2)
used progestogens, whereas 11 of 44 (25%) of APOE e3/e3
(group E3) and 4 of 16 (25%) of APOE e3/e4 1 e4/e4
(group E4) women used a combination of oral estrogens
and progestogens. Our data indicate that the use of
estrogens plus progestogens blunts the rise in triglycerides because women using progestogens had significantly
lower mean loge serum triglyceride values than women
not using progestogens: 0.351 6 0.497 (n 5 15) and
0.688 6 0.508 mmol/L (n 5 51), respectively (P 5 0.027).
These loge values, when transformed, equate to geometric
mean triglyceride concentrations of 1.42 mmol/L (116
mg/dL) and 1.99 mmol/L (162 mg/dL), respectively.
Lastly, population admixture, unknown differences in
CHD risk factors, and other environmental factors could
undoubtedly explain some of the between-subject lipid
variances noted within each APOE genotype.
In summary, we have presented data that support considering APOE genotypes in treating postmenopausal
women with ERT to potentially reduce their risk of CHD.
Our data indicate that of the three groups not receiving
ERT, women with an APOE e2/e3 genotype have the
lowest risk of CHD. If these women were placed on oral
estrogens, then a combination therapy would seem to be
warranted to avoid hypertriglyceridemia. The same recommendation would hold for women with an APOE
e3/e3 genotype. Women with an APOE e3/e4 or e4/e4
genotype could possibly benefit the most from ERT, and
the use of progestogens in these women would not seem
to be warranted unless it is deemed necessary to reduce
rates of endometrial hyperplasia.
This work supported by grants from the National Institute
on Aging (Grants R37AG02049 and RO1AG10149) and the
University of New Mexico General Clinical Research
Center (NCRR-GCRC Grant MO1RR00977).
References
1. Newman KP, Sullivan JM. Coronary heart disease in women:
epidemiology, clinical syndromes, and management. Menopause
1996;3:51–9.
2. Grady D, Rubin SM, Petitti DB, Fox CS, Black D, Ettinger B, et al.
Hormone therapy to prevent disease and prolong life in postmenopausal women. Ann Intern Med 1992;117:1016 –37.
3. Colditz GA, Willet WC, Stampfer MJ, Rosner B, Speizer FE,
Hennekens CH. Menopause and the risk of coronary heart disease
in women. N Engl J Med 1987;316:1105–10.
4. Barrett-Conner E, Bush TL. Estrogen and coronary heart disease in
women. JAMA 1991;265:1861–7.
5. Sullivan JM, Van derzwaag R, Hughes JP, Maddock V, Kroetz FW,
Ramanathan KB, Mirris DM. Estrogen replacement and coronary
artery disease. Effect on survival in postmenopausal women. Arch
Intern Med 1990;150:2557– 62.
6. Bush TL, Barrett-Connor E, Cowan LD, Criqui MH, Wallace RB,
Suchindran CM, et al. Cardiovascular mortality and noncontraceptive use of estrogen in women: results from the Lipid Research
Clinics Program Follow-up Study. Circulation 1987;75:1102–9.
7. Stengard JH, Weiss KM, Sing CF. An ecological study of associa-
Clinical Chemistry 45, No. 8, 1999
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
tion between coronary heart disease mortality rates in men and
the relative frequencies of common allelic variation in the gene
coding for apolipoprotein E. Hum Genet 1998;103:234 – 41.
Beisieger U, Weber W, Ihrke G, Herz J, Stanley KK. The LDLreceptor-related protein, LRP, is an apolipoprotein E-binding protein. Nature 1989;34:162– 4.
Davignon J, Gregg RE, Sing CF. Apolipoprotein E polymorphism and
atherosclerosis. Arteriosclerosis 1988;8:1–21.
Schaefer EJ, Lamon-Fava S, Johnson S, Ordovas JM, Schaefer
MM, Castelli WP, Wilson PWF. Effects of gender and menopausal
status on the association of apolipoprotein E phenotype with
plasma lipoprotein levels. Results from the Framingham Offspring
Study. Arterioscler Thromb 1994;14:1105–13.
Somekawa Y, Wakabayashi A. Relationship between apolipoprotein E polymorphism, menopausal symptoms and serum-lipids
during hormone replacement therapy. Eur J Obstet Gynecol Reprod Biol 1998;79:185–91.
Garry PJ, Goodwin JS, Hunt WC, Hooper EM, Leonard AG. Nutritional status in a healthy elderly population: dietary and supplemental intakes. Am J Clin Nutr 1982;36:319 –31.
Garry PJ, Goodwin JS, Hunt WC. Nutritional status in a healthy
elderly population: folate and vitamin B12. J Am Geriatr Soc
1984;32:719 –26.
Garry PJ, Hunt WC, Koehler KM, Van derJagt DJ, Vellas BJ.
Longitudinal study of dietary intakes and plasma lipids in healthy
elderly men and women. Am J Clin Nutr 1992;55:682– 8.
Warnick GR, Nguyen T, Albers JJ. Comparison of improved precipitation methods for quantification of high-density lipoprotein cholesterol. Clin Chem 1985;31:217–22.
Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without
use of the preparation ultracentrifuge. Clin Chem 1972;18:499 –
502.
Walsh BW, Schiff I, Rosner B, Greenberg L, Ravnikar V, Sacks FM.
Effects of postmenopausal estrogen replacement on the concentrations and metabolism of plasma lipoproteins. N Engl J Med
1991;325:1196 –204.
Hixson JE, Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res
1990;31:545– 8.
Xhignesse M, Lussier-Cacan S, Sing CF, Kessling AM, Davignon
J. Influence of common variants of apolipoprotein E on measures
of lipid metabolism in a sample selected for health. Arterioscler
Thromb 1991;11:1100 –10.
Assman G, Schulte H. Relation of high density lipoprotein choles-
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
1223
terol and triglycerides to the incidence of atherosclerotic coronary
heart disease (the PROCAM experience). Am J Cardiol 1992;70:
733–7.
Castelli WP, Abbott RD, McNamara PM. Summary estimates of
cholesterol used to predict coronary heart disease. Circulation
1983;67:730 – 4.
Wakatsuki A, Sagara Y. Lipoprotein metabolism in postmenopausal and oophorectomized women. Obstet Gynecol 1995;85:
523– 8.
Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B,
Vittinghoff E. Randomized trial of estrogen plus progestin for
secondary prevention of coronary heart disease in postmenopausal women. JAMA 1998;280:605–13.
Huang YD, Liu XQ, Rall SC, Mahley RW. Apolipoprotein E2 reduces
the low-density-lipoprotein level in transgenic mice by impairing
lipoprotein lipase-mediated lipolysis of triglyceride-rich lipoproteins. J Biol Chem 1998;273:17483–90.
Somekawa Y, Wakabayashi A. Relationship between apolipoprotein-E polymorphism, menopausal symptoms, and serum-lipids
during hormone replacement therapy. Eur J Obstet Gynecol Reprod Biol 1998;79:185–91.
Campos H, Walsh BW, Judge H, Sacks FM. Effect of estrogen on
very low density lipoprotein and low density lipoprotein subclass
metabolism in postmenopausal women. J Clin Endocrinol Metab
1997;82:3955– 63.
Urabe M, Yamamoto T, Kashiwagi T, Okubo T, Tsuchiya H, Iwasa
K, et al. Effect of estrogen replacement therapy on hepatic
triglyceride lipase, lipoprotein lipase and lipids including apolipoprotein E in climacteric and elderly women. Endocr J 1996;43:
737– 42.
Kuusi T, Saarinen P, Hikkila EA. Evidence for the role of hepatic
endothelial lipase in the metabolism of plasma high density
lipoprotein in man. Atherosclerosis 1980;36:589 –93.
Kushwaha RS, Lewis DS, Carey KD, McGill HC. Effects of estrogen
and progesterone on plasma lipoproteins and experimental atherosclerosis in the baboon (Papio SP). Arterioscler Thromb 1991;
11:23–31.
Schachter F, Faure-Delanef L, Guenot F, Rouger H, Froguel P,
Lesueur-Ginot L, Cohen D. Genetic association with human longevity at the APOE and ACE loci. Nat Genet 1994;6:29 –32.
Saure A, Hirvonen E, Tikkanen MJ, Viinikka L, Ylikorkala O. A novel
oestradiol-desogestrel preparation for hormone replacement therapy: effects on hormones, lipids, bone, climacteric symptoms and
endometrium. Maturitus 1993;16:1–12.