1

Effects of gender, age and menopausal status on serum apolipoprotein

concentrations

Short title: Menopause, gender, age and apolipoproteins

Panagiotis Anagnostisa, John C Stevensonb, David Crookc, Desmond G Johnstona, Ian F

Godslanda

aDiabetes Endocrinology and Metabolic Medicine, Faculty of Medicine, Imperial College

London, St. Mary's Campus, London, UK

bNational Heart and Lung Institute, Imperial College London, Royal Brompton Campus, UK.

cClinical Investigations and Research Unit, Royal Sussex County Hospital, Sussex, UK.

Corresponding author:

Ian F Godsland, PhD, Division of Diabetes, Endocrinology and Metabolism, Department of

Medicine, Imperial College London, Room G1, Norfolk Place, London W2 1NH, UK.

email: i.godsland@imperial.ac.uk

Keywords: apolipoprotein A-I, apolipoprotein A-II, apolipoprotein B, male gender, ageing,

menopause

Declaration of interests: The authors have no competing interests to disclose

Funding: Panagiotis Anagnostis was supported in this work by the European Society of

Endocrinology and the Endocrine Society (USA), as a part of the International Endocrine

Scholars Programme (IESP). Ian Godsland is supported by the Higher Education Funding

Council for England. No funding source was involved in the analysis or in the writing of the

report.

Word count: 3928

2

Abstract

Objective: to undertake a comprehensive evaluation of apolipoprotein risk markers for

cardiovascular disease (CVD) according to gender, age and menopausal status.

Design: cross-sectional analysis of independent associations of gender, age and

menopause with serum apolipoproteins

Participants: apparently healthy Caucasian premenopausal (n=109) and postmenopausal

(n=252) women not taking oral contraceptives or hormone replacement, and Caucasian men

(n=307).

Measurements: Serum apolipoprotein (apo) B, A-I and A-II concentrations were measured,

plus serum total cholesterol, low-density and high-density lipoprotein cholesterol (LDL-C and

HDL-C respectively), triglycerides, cholesterol in HDL subfractions and the apoB/apoA-I,

LDL-C/apoB, HDL-C/apoA-I and HDL-C/apoA-II ratios. Analyses were undertaken with and

without standardization for confounding characteristics and in 5 year age ranges.

Results: Overall, apoB concentrations were highest in men but in women rose with age and

menopause to converge, in the age range 50-55 years, with concentrations in men. The LDL-

C/apoB ratio was generally higher in women than men. ApoA-I concentrations were highest

in postmenopausal women and lowest in men (standardized median (IQR) 144 (130,158) vs

119 (108, 132) g/L, respectively, p<0.001). ApoA-II concentrations were also highest in

postmenopausal women but were lowest in premenopausal women (40.3 (37.5, 44.5) vs

32.9 (30.5, 35.7) g/L, respectively, p<0.001). Nevertheless, postmenopausal women had

HDL-C/apoA-I and HDL-C/apoA-II ratios approaching the lowest ratios, which were seen in

men.

Conclusions: Consistent with adverse effects on CVD risk, male gender, ageing in women

and menopause were associated with increased apoB concentrations, and menopause and

male gender were associated with a decreased cholesterol content of HDL particles.

3

Introduction

Studies spanning almost 40 years suggest an increase in cardiovascular disease

(CVD) risk as a consequence of the transition to menopause1, 2, although the influence of

ageing might predominate3. Either way, an important contributor is likely to be a more

atherogenic lipid profile. For example, in our own studies, menopause was associated with

an increase in total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), and

triglycerides (TG), and a decrease in high-density lipoprotein cholesterol (HDL-C)4, 5.

Menopause or ageing in women may, therefore, be associated with a lipid profile that

converges on the more atherogenic profile reported for men. Other theoretically pro-

atherogenic changes with menopause include increases in lipoprotein(a) concentrations6,

central obesity1, endothelial dysfunction7 and systemic inflammation8.

Evaluation of serum concentrations of the protein components of lipoproteins, the

apolipoproteins, could provide further information9, 10. In particular, apolipoprotein B (apoB),

one molecule of which is contained in each LDL particle, can predict ischemic CVD events in

both genders to a greater extent than LDL-C9-12. Moreover, it has been shown that both low

concentrations of apolipoprotein A-I (apoA-I), the main apolipoprotein of HDL, is also

predictive of CVD risk in both sexes beyond the risk predicted by conventional CVD risk

factors9, 10 and other lipid-related measures13. Apolipoprotein A-II (apoA-II), also a component

of HDL, has been much less studied and might exert a protective effect against CVD14,

although we previously reported that apoA-II levels were positively associated with

coagulation factors VIIc and Xc,, suggesting a prothrombotic involvement15.

The few studies comparing apolipoproteins in pre- and postmenopausal women have

generally been of small size and have not included men of comparable age nor discriminated

between effects of menopause and effects of aging16, 17. In the present cross-sectional

analysis we have used subgroup information from a previously-described cohort of pre- and

postmenopausal women and men5 to evaluate effects of gender, menopause and age on

serum concentrations of apoA-I, apoA-II and apoB.

4

Subjects and Methods

Study design and subjects

Our previous analysis of effects of menopause, gender and age on lipids and lipoproteins

was undertaken in 1833 individuals5, of whom over a third also had apolipoproteins measured

and are the subject of the present cross-sectional analysis. Participants included healthy

Caucasian pre- and postmenopausal women, recruited mainly from family planning or

menopause clinics or general practices, who had participated in previously-reported studies

of the effects of oral contraceptives (OC) and hormone replacement therapies (HRT) but were

not taking sex steroids at the time of their measurements. Caucasian men participating in an

occupational cohort study were also included. Participants taking drugs likely to affect lipid

metabolism or taking glucose-lowering medications, those with a fasting plasma glucose

concentration of ≥7 mmol/L or with evidence of endocrine conditions or CVD, were excluded

from the present analysis. Apolipoprotein measurements were made on a routine basis, along

with lipids and lipoprotein cholesterol concentrations, in the same laboratory: for

premenopausal women from 1987 to 1991 in the course of studies that began in 1984; for

postmenopausal women from 1994 to 1998 in the course of studies that began in 1987; and

for men from 1992 to 1998 in the course of a study that began in 1971. The apolipoprotein

measurements, therefore, constitute a continuous sample taken during the latter stages of

three different research themes. Premenopausal women were studied between days 21 and

27 of the menstrual cycle. Menopause was defined as amenorrhea for at least 12 months and

elevated gonadotrophin levels (FSH>40 IU/L). The studies from which the data derive

received ethics committee approval and full informed consent was obtained from each

participant.

In the present analysis, an analytical scheme has been adopted identical to that in our

previous analysis of lipids and lipoproteins5. To ensure comparability between the two

analyses, it was important that, in the present analysis, lipid and lipoprotein concentrations for

5

the pre- and postmenopausal women and men were representative of those found in our

previous analysis. This was the case for TC, LDL-C and TG concentrations, but for HDL-C

significantly lower concentrations were found in the 190 premenopausal women who had

undergone apolipoprotein concentration measurements (median (IQ range) 1.54 (1.32-1.71)

vs 1.66 (1.42-1.89) mmol/L, p<0.001). Eighty-two premenopausal women were excluded to

achieve a distribution of HDL-C concentrations that would match that of premenopausal

women in our previous, larger study5. This had the effect of increasing the median HDL-

cholesterol concentration to 1.66 mmol/L, with accompanying changes in HDL subfractions

and apoAI and apoAII. However, median concentrations of total cholesterol, triglycerides, LDL

cholesterol and apoB were unaffected. The total study group then consisted of 668 individuals

meeting inclusion criteria.

Laboratory assessments

Serum TC, LDL-C, TG, HDL-C and HDL-C subfraction concentrations were

measured as described previously5. Serum apoA-I, apoA-II and apoB (predominantly apoB-

100, since the samples were obtained in the fasting state) concentrations were measured

using immunoturbidimetry-based commercial kits (Immuno AG, Vienna, Austria) on a Cobas

Mira discrete analyzer [Roche, Basel, Switzerland], as described previously15. Between-batch

coefficients of variation (CV) were 2-4%, 3-4% and 1-3% for apoB, apoA-I and apoA-II,

respectively. CVs were 1-2% for TC and TG, 2-4% for HDL-C, 11% for HDL2-C and 8% for

HDL3-C. Quality control of the assays was monitored according to both internal and external

quality assessment schemes throughout the study period. Measurements were made by a

historically continuous research group and attention was paid throughout to continuity of

standardization.

Statistical analysis

Statistical analyses were carried out in STATA (STATA 13.0, College Station, Tx,

U.S.A.). The apoB/apoA-I, LDL-C/apoB, HDL-C/apoA-I and HDL-C/apoA-II concentration

ratios were calculated. The analysis structure was identical to that adopted for our larger,

6

recently-published study of effects on lipids and lipoproteins5 and for our previous study of

effects of menopause on lipids and lipoproteins, published in 19934. Summary statistics for

each group were derived and significant variation between the three groups was explored

using ANOVA, Kruskal-Wallis ANOVA or chi-square test, as appropriate. Then, in an

analysis designed to illustrate the observed variation in apolipoprotein-related variables with

age, means and 95 percent confidence intervals for each variable in 5-year age ranges were

plotted against age range in each of the three groups. Multiple linear regression analyses

were undertaken in each group separately with each apolipoprotein-related variable

predicted by age, BMI, cigarette smoking, alcohol consumption and exercise habit categories

and number of pregnancies. Coefficients were compared between groups to identify

significant differences between groups in variation of apolipoprotein-related variables with

each demographic variable. Using coefficients from the multiple linear regression analyses in

each group, apolipoprotein-related variables were then standardized to a representative age

of menopause of 50 years and BMI of 25 kg/m2, and to zero smoking, alcohol, exercise and

pregnancies. Missing values for demographic variables were imputed (STATA command:

‘impute’ – utilizing best subset regression). Statistically significant differences in variables

between the three groups were then evaluated using the standardized data. An analysis of

standardized data was also undertaken without imputation of missing values for demographic

variables. Effects of menopause, gender and age were evaluated with reference both to raw

data, and to standardised data, with and without imputation.

7

Results

After exclusions, there were available for analysis data from 109 pre-menopausal

women (aged 23-44y, mean 32.9y), 252 post-menopausal women (aged 42-73y, mean

57.0y; mean time since menopause, 8.5y) and 307 men (aged 26-82y, mean 52.4y). Group

characteristics are shown in Table 1. As well as for age, there were significant differences

between groups in BMI, smoking, alcohol and exercise. In particular, men had higher BMI

than both pre- and postmenopausal women. Postmenopausal women had also higher BMI

compared with premenopausal women. The highest percentage of current smokers was

found among the premenopausal women. This percentage was also higher in men compared

with postmenopausal women. Premenopausal women had the highest moderate

consumption of alcohol but the men had the highest heavy consumption. Premenopausal

women had the highest percentage of non-aerobic exercisers, whereas men had the highest

percentage of aerobic exercisers (Table 1).

Between-group differences in apolipoproteins and lipid-related variables - unstandardized

analysis

Premenopausal women had lower median apoB, apoA-I and apoA-II concentrations

than postmenopausal women (Table 1). Compared with men, premenopausal women had

lower median apoB, higher apoA-I and lower apoA-II concentrations. Median apoB levels in

postmenopausal women were comparable to those of men and apoA-I and apoA-II

concentrations were higher. The apoB/apoA-I ratio was higher in men compared with both

female groups and higher in postmenopausal compared with premenopausal women. The

LDL-C/apoB ratio was similar in men and premenopausal women, but significantly higher in

postmenopausal women. The median HDL-C/apoA-I and HDL-C/apoA-II ratios were highest

in premenopausal women and lowest in men (Table 1). This pattern of differences was also

seen for the HDL2-C/apoA-I, HDL2-C/apoA-II and HDL3-C/apoA-II ratios (results not shown).

The HDL3-C/apoA-I ratio was lowest in postmenopausal women and highest in

premenopausal women.

8

Regarding the lipid and lipoprotein profile, premenopausal women had lower TC, TG,

LDL-C, non-HDL-C levels and TC/HDL-C ratio, but higher HDL3-C levels compared with

postmenopausal women (Table 1). HDL-C was higher in pre-menopausal women at

borderline significance (p=0.07). The same findings, plus significantly higher HDL-C and

HDL2-C, were apparent when premenopausal women were compared with the men.

Compared with the men, postmenopausal women had higher TC, HDL-C and HDL-C

subfraction concentrations, but lower TG concentrations and TC/HDL-C ratio. LDL-C and

non-HDL levels did not differ between men and postmenopausal women.

Variation in apolipoproteins and lipid-related variables with age

The age-related variation underlying the differences in apolipoprotein concentrations

between the groups is illustrated in Figures 1 and 2. ApoB concentrations rose steadily in the

women and levels converged with those in men in the age range 50-55 years. This pattern of

change was also apparent for the apoB/apoA-I ratio. ApoA-I levels were consistently lowest

in men, diverging most from those in women between 40 and 60 years of age. ApoA-II levels

were lowest in pre- and highest in postmenopausal women. The LDL-C/apoB ratio did not

differ between premenopausal women and men but tended to be higher in postmenopausal

women than men, though fully converging above age 65. The HDL-C/apoA-I ratio was lowest

in men and highest in premenopausal women. Similar differences were also seen for the

HDL-C/apoA-II ratio, though with more marked divergences between the groups, and for the

HDL2-C/apoA-I, HDL2-C/apoA-II and HDL3-C/apoA-II ratios (results not shown). The HDL3-

C/apoA-I ratio was lowest in postmenopausal women, highest in premenopausal women but

showed little gender difference over the age ranges studied.

Variation in apolipoproteins and lipid-related variables with individual characteristics

Multiple linear regression analysis models included apolipoprotein-related variables

as the dependent variable and, as independent variables: male gender, post-menopausal

status, age, BMI, smoking, alcohol, exercise and number of pregnancies. In multivariable

regression analysis, significant or borderline significant (p≤0.1) independent associations

9

consistent in all three groups included: higher TC and LDL-C with greater age and BMI and

higher TG with greater BMI. Higher TG and lower levels of HDL-C, HDL2-C, apoA-I, HDL2-

C/apoA-I, HDL-C/apoA-II and HDL2-C/apoA-II were significantly associated with greater BMI

among postmenopausal women and men. Premenopausal women and men showed higher

HDL3-C and apoA-II associated with heavy drinking (there were no postmenopausal heavy

drinkers), higher apo-B and apoB/apoA-I associated with greater BMI, lower HDL-C/apo-A-I

associated with light smoking and higher LDL-C/apoB with greater age.

Between-group differences in apolipoproteins and lipid-related variables - unstandardized

and standardised analysis

Differences between the three groups were evaluated in terms of the differences

seen in the raw apolipoprotein-related variable measures reported above (Table 1:

subsequently referred to as analysis 1, which reflects the overall characteristics of the

groups) and in terms of variables standardised to a nominal age at menopause of 50 years,

BMI 25kg/m2 and zero cigarette smoking, alcohol intake, exercise and, for the women, zero

pregnancies (theoretically, most accurately discriminating discrete effects of menopause and

gender). Standardisation was undertaken with imputation of missing values for the above

characteristics (analysis 2, n=668: Table 2) and without imputation (analysis 3, n=588:

results not shown). Numbers of missing values for age, BMI and smoking were 39, 23 and 78

respectively and these were concentrated among the postmenopausal women (39, 21 and

76, respectively).

Relative to postmenopausal women, consistent mean concentration differences in

premenopausal women were: lower apoA-I and apoA-II and higher HDL-C/apoA-II ratio.

ApoB tended to be lower in premenopausal women, though non-significantly in analysis 3

(p=0.2). The apoB/apoA-I ratio showed no consistent differences across the three analyses.

The LDL-C/apoB ratio, which was lower in premenopausal women in the unstandardized

analysis, tended to become higher with standardisation, though not significantly in analysis 2

10

(p=0.1). The HDL-C/apoA-I ratio was higher in analyses 1 and 2 but did not differ in analysis

3 (p=0.7).

Relative to men, consistent significant mean concentration differences in

premenopausal women were: lower apoB, higher apoA-I, lower apoA-II, lower apoB/apoA-1

ratio and higher HDL-C/apoA-I and HDL-C/apoA-II ratios. The LDL-C/apoB ratio, which did

not differ in the un-standardised comparison (p=0.1), became higher in premenopausal

women with standardisation (p<0.001, analyses 2 and 3).

Relative to men, consistent significant mean concentration differences in

postmenopausal women were: higher apoA-I and apoA-II, lower apoB/apoA-1 ratio, higher

LDL-C/apoB ratio and higher HDL-C/apoA-II ratio. ApoB did not differ in the unstandardized

analysis (p=0.5) but tended to become lower in postmenopausal women in the standardised

analyses, though not significantly in analysis 2 (p=0.1). The HDL-C/apoA-I ratio in

postmenopausal women was higher in analyses 1 and 3 (p<0.001) but did not differ in

analysis 2 (p=0.7).

11

Discussion

In this analysis we have endeavored to differentiate effects of gender, menopause

and aging. ApoB levels were generally higher in men but, in women, progressively higher

levels were seen with age and, independently, with menopause to levels comparable with

those of men. The LDL-C/apoB ratio was positively associated with female gender. Male

gender was consistently associated with a lower apoA-I concentration and menopause with

higher apoA-I and apoA-II. Effects of gender on apoA-II were, however, difficult to distinguish

because of the marked positive effect of menopause - levels began higher in men relative to

premenopausal women but were then overtaken by levels in postmenopausal women. A

lower HDL-C/apoA-I ratio and, more consistently, a lower HDL-C/apoA-II ratio were

associated with male gender and menopause.

In most previously-published studies7, 12, 18-23, and in accord with our current and

previously-published analyses4, 5, menopausal transition is accompanied by a more

atherogenic TC, LDL-C and TG profile, but with less consistency regarding HDL-C

concentrations. Only a few previous studies have included apolipoprotein measurements. In

one study23 postmenopausal women had 8.2% higher concentrations of apoB relative to

premenopausal women without a significant difference in apoA-I. Whereas we found no clear

effect of menopause on the apoB/apoA-I ratio, others have reported it elevated within two

years of menopause24. Few studies have assessed differences in atherogenic profile in terms

of both gender and menopausal status. In a sub-study of 285 individuals from the

Framingham Offspring Study12, premenopausal women had lower TC, TG, LDL-C and apoB

concentrations than men, while post-menopausal women did not differ in apoB

concentrations but had higher TC and LDL-C concentrations than men, which generally

accords with our findings. Both pre- and postmenopausal women had higher apoA-I and

HDL-C concentrations than men, again according with our findings. Further support comes

from a study of 409 individuals from Taiwan25 in which men had higher apoB and lower apoA-

I concentrations than women overall but, as in our study, apoB concentrations were similar

12

when men and postmenopausal women were compared. Moreover, postmenopausal women

had higher apoA-I, and apoB concentrations than premenopausal women. Regarding

differences in apoA-II, studies are few and findings conflict. A small study from Japan (n=15

in each group) found no difference between pre- and post-menopausal women with respect

to apoA-II (or apoA-I) concentrations16 but in the previously-mentioned study from Taiwan,

postmenopausal women had higher apoA-II concentrations25, which accords with our

findings. However, apoA-II levels in premenopausal women were not lower than those in

men.

The cholesterol content of LDL particles can be inferred from the LDL-C/apoB ratio

because each LDL particle contains a single apoB molecule. We found no distinct effect of

menopause on LDL cholesterol content but women, especially postmenopausal women,

generally had a more cholesterol-rich LDL particle than men. An evaluation of LDL-C and

apoB concentrations in men (n=1,533) and women (n=1,597) in the Framingham Offspring

Study reported that whereas LDL-C concentrations increased with age more markedly in

women than men after the age of 60, apoB levels increased at a similar rate in both

genders17. As with our findings, these imply a more cholesterol-rich LDL particle in

postmenopausal women relative to men.

Unlike LDL particles and apoB, HDL particles vary in their numbers of apoA-I and

apoA-II molecules. Different approaches have been used to characterize HDL

concentrations, the concentration of plasma cholesterol carried in HDL particles (i.e. HDL-C)

and in cholesterol-rich and cholesterol-poor HDL particles (i.e. HDL2-C and HDL3-C,

respectively) being the most familiar. An alternative has been to discriminate HDL particles

containing apoA-I but not apoA-II (referred to as Lp A-I ), which approximate to HDL2 , and

particles containing both apo A-I and apo A-II (referred to as Lp A-I:A-II), which approximate

to HDL3. In the previously-mentioned study from Taiwan25, postmenopausal women had

higher concentrations of Lp A-I and of Lp A-I:A-II compared with pre-menopausal women,

which accords with our findings. However, the Framingham Offspring study analysis found

13

no difference between men and women nor between pre- and postmenopausal women in Lp

A-I:A-II concentrations17.

Mechanisms underlying the effects we observed are likely to mainly reflect the

consequences for lipoprotein metabolism of the presence or absence of oestrogens3. In one

of the few studies there have been of the effects of sex steroids on apolipoprotein

metabolism, oral ethinyl estradiol increased apoA-I and apoA-II concentrations, by 27% and

17%, respectively, with no effect on catabolic rates but with a significant increase in

production rates only detected for apoA-I26. Nevertheless, support for this observation comes

from the in vitro finding that estradiol stimulates newly transcribed hepatic apoA-I mRNA

without any effect on apoA-II synthesis27. By contrast, the ability of oestrogens to lower LDL

apoB concentrations appears to reside in increased clearance rather than decreased

synthesis28. These observations are consistent with the relatively low apoB levels we

observed in pre- relative to postmenopausal women and men and the relatively high apoA-I

levels in premenopausal women relative to men, but the differences in apoA-II levels remain

unexplained.

The data we analysed were not generated on the basis of population sampling but

rather from pools of relatively healthy research volunteers. However, rather than aiming to

arrive at generalizable ranges for apolipoprotein concentrations in the three groups, our

intention was to distinguish discrete effects of gender, menopause and ageing. To that end,

we attempted to minimise as much as possible effects of potential confounding factors on

variation in apolipoprotein concentrations by studying individuals known to be relatively

healthy. We also undertook to further minimise confounding effects by statistical

standardisation, which we performed in each of the three groups separately so as to

minimise any confounding by variation in interactions of gender, menopause and age with

confounding variables (BMI, smoking, alcohol and exercise) in their effects on variation in

apolipoprotein concentrations. There will, nevertheless, have been aspects of population

variation in apolipoprotein concentrations that were not standardised. Abdominal adiposity in

14

particular, could be an important factor, particularly with regard to the effects of menopause

29,30. A further limitation of our study was the separation in time between the measurements

made in pre-menopausal women and those in post-menopausal women and men and it is

possible that changes in dietary or lifestyle habits could have affected the differences we

observed. However, the profiles in the premenopausal women were appreciably healthier

than those in the other groups, yet awareness of the benefits of healthy diet and lifestyle

would have been greater during the later sampling period for the men and premenopausal

women. It should also be noted that median apoB concentrations were somewhat lower than

the majority of reported population averages (e.g. Li et al 1996 17). Nevertheless, variation in

apoB concentrations showed the expected associations with other variables, in particular

with LDL cholesterol, with correlation coefficients of 0.88, 0.79 and 0.86 (all p<0.001) for pre-

and postmenopausal women and men, respectively.

With regard to effects of gender, age and menopause on lipid, lipoprotein and HDL

subfraction-cholesterol concentrations, our findings in the present analysis parallel those we

reported and discussed in the larger dataset, which was not restricted to those with

apolipoprotein measurements recorded 5. The present analysis then extends that analysis

with effects on apolipoprotein concentrations and the additional perspectives on lipoprotein

composition that apolipoprotein measurements can offer. Conventionally, LDL particles are

considered to be atherogenic and, since each LDL particle contains one molecule of apoB,

our findings imply postmenopausal women had an atherogenic LDL profile comparable to

that in men and both ageing and menopause may contribute. Effects of menopause on

cholesterol loading of the LDL particle were not consistent, but female gender was generally

associated with higher LDL particle cholesterol content than male, particularly for

postmenopausal women. ApoA-I concentrations were consistently highest in

postmenopausal women and lowest in men, suggesting anti-atherogenic changes in

postmenopausal women. However, our findings regarding cholesterol in HDL and its

subfractions, apoA-II concentrations and ratios between HDL-C and apoA-I and apoA-II

15

further suggest that menopause and, to an even greater extent, male gender are associated

with HDL particles that are markedly depleted in cholesterol. To the extent that cholesterol

loading of HDL particles may reflect efficient elimination of cholesterol from tissues, these

relative differences are consistent with pro-atherogenic effects of menopause and male

gender.

There has been controversy over mechanisms underlying the cardiovascular benefits

of therapeutic intervention with statins and possible cardiovascular benefits of

postmenopausal hormone replacement therapy (HRT). Interestingly, both these therapies

can decrease apoB and, therefore, LDL particle number and can increase HDL particle

cholesterol loading, i.e. they reverse the potentially adverse effects of female aging, male

gender and menopause that we distinguished. Detailed evaluations of lipoprotein

composition in the context of large, prospective studies of CVD, both observational and with

intervention, will be needed to clarify the clinical importance of these relationships.

Acknowledgements

We thank the many clinical, scientific, technical, nursing and administrative staff who

contributed to the studies from which the data for this analysis derive.

16

References

1. Dosi, R., Bhatt, N., Shah, P., Patell, R. (2014) Cardiovascular disease and

menopause. J Clin Diagn Res, 8, 62-4

2. Gordon, T., Kannel, W. B., Hjortland, M. C., McNamara, P. M. (1978) Menopause

and coronary heart disease. The Framingham Study. Ann Intern Med, 89, 157-61

3. Godsland, I. F., Wynn, V., Crook, D., Miller, N. E. (1987) Sex, plasma lipoproteins,

and atherosclerosis: prevailing assumptions and outstanding questions. Am Heart J, 114,

1467-503

4. Stevenson, J. C., Crook, D., Godsland, I. F. (1993) Influence of age and menopause

on serum lipids and lipoproteins in healthy women. Atherosclerosis, 98, 83-90

5. Anagnostis, P., Stevenson, J. C., Crook, D., Johnston, D. G., Godsland, I. F. (2015)

Effects of menopause, gender and age on lipids and high-density lipoprotein cholesterol

subfractions. Maturitas, 81, 62-8

6. Abbey, M., Owen, A., Suzakawa, M., Roach, P., Nestel, P. J. (1999) Effects of

menopause and hormone replacement therapy on plasma lipids, lipoproteins and LDL-

receptor activity. Maturitas, 33, 259-69

7. Salhotra, S., Arora, S., Anubhuti, Trivedi, S. S., Bhattacharjee, J. (2009) Influence of

menopause on biochemical markers of endothelial dysfunction-A case-control pilot study in

North Indian population. Maturitas, 62, 166-70

8. Muzzio, M. L., Miksztowicz, V., Repetto, E. M., Brites, F., Berg, G., Schreier, L.

(2012) Increased MMP-2 in healthy postmenopausal women. Ann Clin Biochem, 49, 75-9

9. Walldius, G., Jungner, I., Holme, I., Aastveit, A. H., Kolar, W., Steiner, E. (2001) High

apolipoprotein B, low apolipoprotein A-I, and improvement in the prediction of fatal

myocardial infarction (AMORIS study): a prospective study. Lancet, 358, 2026-33

10. St-Pierre, A. C., Cantin, B., Dagenais, G. R., Després, J. P., Lamarche, B. (2006)

Apolipoprotein-B, low-density lipoprotein cholesterol, and the long-term risk of coronary heart

disease in men. Am J Cardiol, 97, 997-1001

17

11. Shai, I., Rimm, E. B., Hankinson, S. E., et al. (2004) Multivariate assessment of lipid

parameters as predictors of coronary heart disease among postmenopausal women:

potential implications for clinical guidelines. Circulation, 110, 2824-30

12. Benn, M. (2009) Apolipoprotein B levels, APOB alleles, and risk of ischemic

cardiovascular disease in the general population, a review. Atherosclerosis, 206, 17-30

13. McQueen, M. J., Hawken, S., Wang, X., et al. (2008) Lipids, lipoproteins, and

apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART

study): a case-control study. Lancet, 372, 224-33

14. Maïga, S. F., Kalopissis, A. D., Chabert, M. (2014) Apolipoprotein A-II is a key

regulatory factor of HDL metabolism as appears from studies with transgenic animals and

clinical outcomes. Biochimie, 96, 56-66

15. Atta, M., Crook, D., Shafique, F., Johnston, D. G., Godsland, I. F. (2008)

Procoagulant activities of plasma factor VIIc and factor Xc are positively and independently

associated with concentrations of the high-density lipoprotein apolipoprotein, apo A-II. .

Thromb Haemost, 100, 391-6

16. Ikenoue, N., Wakatsuki, A., Okatani, Y. (1999) Small low-density lipoprotein particles

in women with natural or surgically induced menopause. Obstet Gynecol, 93, 566-70

17. Li, Z., McNamara, J. R., Fruchart, J. C., et al. (1996) Effects of gender and

menopausal status on plasma lipoprotein subspecies and particle sizes. J Lipid Res, 37,

1886-96

18. Matthews, K. A., Crawford, S. L., Chae, C. U., et al. (2009) Are changes in

cardiovascular disease risk factors in midlife women due to chronological aging or to the

menopausal transition? J Am Coll Cardiol, 54, 2366-73

19. Bayrak, A., Aldemir, D. A., Bayrak, T., Corakçi, A., Dursun, P. (2006) The effect of

hormone replacement therapy on the levels of serum lipids, apolipoprotein AI, apolipoprotein

B and lipoprotein (a) in Turkish postmenopausal women. Arch Gynecol Obstet, 274, 289-96

18

20. Bonithon-Kopp, C., Scarabin, P. Y., Darne, B., Malmejac, A., L, G. (1990)

Menopause-related changes in lipoproteins and some other cardiovascular risk factors. Int J

Epidemiol, 19, 42-8

21. Dallongeville, J., Marecaux, N., Isorez, D., Zylbergberg, G., Fruchart, J. C., Amouyel,

P. (1995) Multiple coronary heart disease risk factors are associated with menopause and

influenced by substitutive hormonal therapy in a cohort of French women. Atherosclerosis,

118, 123-33

22. Campos, H., McNamara, J. R., Wilson, P. W., Ordovas, J. M., Schaefer, E. J. (1988)

Differences in low density lipoprotein subfractions and apolipoproteins in premenopausal and

postmenopausal women. J Clin Endocrinol Metab, 67, 30-5

23. Peters, H., Westendorp, I., Hak, A. E., et al. (1999) Menopausal status and risk

factors for cardiovascular disease. J Intern Med, 246, 521-8

24. Murano, T., Izumi, S., Kika, G., et al. (2003) Impact of menopause on lipid and bone

metabolism and effect of hormone replacement therapy. Tokai J Exp Clin Med, 28, 109-19

25. Lyu, L. C., Yeh, C. Y., Lichtenstein, A. H., Li, Z., Ordovas, J. M., Schaefer, E. J.

(2001) Association of sex, adiposity, and diet with HDL subclasses in middle-aged Chinese.

Am J Clin Nutr, 74, 64-71

26. Brinton, E. A. (1996) Oral estrogen replacement therapy in postmenopausal women

selectively raises levels and production rates of lipoprotein A-I and lowers hepatic lipase

activity without lowering the fractional catabolic rate. Arterioscler Thromb Vasc Biol, 16, 431-

40

27. Jin, F., Kamanna, V., Kashyap, M. (1998) Estradiol stimulates apolipoprotein A-I- but

not A-II-containing particle synthesis and secretion by stimulating mRNA transcription rate in

Hep G2 cells. Arterioscler Thromb Vasc Biol, 18, 999-1006

28. Walsh, B. W., Schiff, I., Rosner, B., Greenberg, L., Ravnikar, V., Sacks, F. M. (1991)

Effects of postmenopausal estrogen replacement on the concentrations and metabolism of

plasma lipoproteins. N Engl J Med, 325, 1196-204

19

29. Ley, C., Lees, B., Stevenson, J. (1992) Sex- and menopause-associated changes in

body fat distribution. Am J Clin Nutr, 55, 950-54

30. Hodson, L., Banerjee, R., Rial, B., et al. (2015) Menopausal status and abdominal

obesity are significant determinants of hepatic lipid metabolism in women. J Am Heart Assoc,

epub Oct 2,

20

Table 1: Group characteristics

Group 11987-1991Premenopausal women (n=109)

Group 21994-1998Postmenopausal women (n=252)

Group 31992-1998Men (n=307)

pGroup 2 v Group 1

pGroup 3 v Group 1

pGroup 3 v Group 2

age (years) 32.9 (5.4) 57.0 (6.5) 52.4 (10.5) <0.001 <0.001 <0.001BMI (kg/m2) 21.6 (1.9) 24.3 (3.0) 26.2 (2.9) <0.001 <0.001 <0.001number of pregnancies 0.5 (1.0) 1.7 (1.6) - <0.001 - -smoking <0.001 0.01 0.04None 83 (76) 166 (93) 267 (88)<15/day 18 (17) 10 (6) 26 (8)≥ 15/day 8 (7) 1 (1) 12 (4)alcohol <0.001 <0.001 <0.001None 4 (4) 138 (55) 26 (8)≤28 units/week 101 (92) 115 (45) 226 (74)>28 units/week 4 (4) 0 (0) 55 (18)exercise <0.001 <0.001 0.001None 50 (46) 146 (58) 136 (44)non-aerobic 49 (45) 41 (16) 47 (16)Aerobic 10 (9) 66 (26) 124 (40)apolipoprotein B (mg/dL) 55 (47, 62) 77 (65, 88) 76 (66, 90) <0.001 <0.001 0.5apolipoprotein A-I (mg/dL) 133 (119, 142) 144 (126, 158) 121 (110, 139) <0.001 <0.001 <0.001apolipoprotein A-II (mg/dL) 36 (33, 38) 41 (37, 45) 39 (36, 43) <0.001 <0.001 <0.001apoB / apoA-I 0.41 (0.35, 0.49) 0.53 (0.43, 0.64) 0.63 (0.52, 0.75) <0.001 <0.001 <0.001LDL-C / apoB x 102 4.31 (4.00, 4.62) 4.59 (4.13, 4.99) 4.36 (4.00, 4.79) <0.001 0.1 <0.001HDL-C / apoA-I x 102 1.25 (1.17, 1.33) 1.10 (1.01, 1.24) 1.06 (0.97, 1.16) <0.001 <0.001 <0.001HDL-C / apoA-II x 102 4.78 (4.11, 5.31) 3.80 (3.29, 4.56) 3.24 (2.83, 3.82) <0.001 <0.001 <0.001total cholesterol (mmol/L) 4.35 (4.01, 4.77) 5.64 (5.07, 6.29) 5.41 (4.77, 6.07) <0.001 <0.001 0.002

21

triglycerides (mmol/L) 0.57 (0.43, 0.75) 1.06 (0.83, 1.41) 1.27 (0.86, 1.94) <0.001 <0.001 <0.001calculated LDL cholesterol (mmol/L) 2.35 (1.95, 2.69) 3.47 (2.94, 4.11) 3.34 (2.80, 3.94) <0.001 <0.001 0.1HDL cholesterol (mmol/L) 1.66 (1.53, 1.84) 1.60 (1.33, 1.87) 1.30 (1.11, 1.53) 0.07 <0.001 <0.001HDL2 cholesterol (mmol/L) 0.73 (0.57, 0.91) 0.64 (0.48, 0.89) 0.45 (0.31, 0.64) 0.1 <0.001 <0.001HDL3 cholesterol (mmol/L) 0.93 (0.88, 1.01) 0.91 (0.79, 1.03) 0.83 (0.72, 0.92) <0.001 <0.001 <0.001cholesterol/HDL cholesterol 2.56 (2.28, 2.97) 3.50 (2.91, 4.24) 4.13 (3.31, 5.06) <0.001 <0.001 <0.001non-HDL cholesterol (mmol/L) 2.64 (2.25, 3.06) 4.01 (3.40, 4.61) 3.99 (3.37, 4.79) <0.001 <0.001 0.7

For continuous, normally distributed variables, mean (SD) are shown. For continuous non-normally distributed variables, median (IQ range); categorical variables, n (%) are shown. Sampling periods are given for each group. Significances: ANOVA or t-test for normally distributed variables, Kruskall-Wallis ANOVA for continuous non-normally distributed variables and chi-square for categorical variables all returned p<0.001 for significant variation between the three groups. Post-hoc, between-group comparisons were made using t-test for normally distributed variables, Mann-Whitney U test for continuous non-normally distributed variables and chi-square for categorical variables. Abbreviations: BMI: body mass index, LDL: low-density lipoprotein, HDL: high density lipoprotein, apo: apolipoprotein

22

Table 2: Lipid–related variables - standardized analysis

Group 11987-1991Premenopausal women (n=109)

Group 21994-1998Postmenopausal women (n=252)

Group 31992-1998Men (n=307)

pGroup 2 v Group 1

pGroup 3 v Group 1

pGroup 3 v Group 2

apolipoprotein B (mg/dL) 67.0 (62.1, 74.2) 72.4 (63.2, 83.9) 74.6 (63.8, 87.0) 0.002 <0.001 0.1apolipoprotein A-I (mg/dL) 134 (124, 144) 144 (130, 158) 119 (108, 132) <0.001 <0.001 <0.001apolipoprotein A-II (mg/dL) 32.9 (30.5, 35.7) 40.3 (37.5, 44.5) 36.5 (33.0, 40.1) <0.001 <0.001 <0.001apoB / apoA-I 0.50 (0.45, 0.56) 0.50 (0.41, 0.61) 0.64 (0.53, 0.76) 0.9 <0.001 <0.001LDL-C / apoB x 102 4.88 (4.61, 5.13) 4.78 (4.40, 5.18) 4.46 (4.06, 4.79 0.1 <0.001 <0.001HDL-C / apoA-I x 102 1.26 (1.20, 1.34) 1.11 (1.02, 1.23) 1.12 (1.04, 1.22) <0.001 <0.001 0.7HDL-C / apoA-II x 102 5.37 (4.78, 5.80) 4.03 (3.43, 4.59) 3.63 (3.27, 4.13) <0.001 <0.001 <0.001total cholesterol (mmol/L) 5.16 (4.74, 5.46) 5.63 (5.11, 6.29) 5.29 (4.69, 5.94) <0.001 0.1 <0.001triglycerides (mmol/L) 0.41 (0.28, 0.56) 0.92 (0.72, 1.28) 1.12 (0.75, 1.67) <0.001 <0.001 <0.001calculated LDL cholesterol (mmol/L) 3.15 (2.93, 3.50) 3.51 (2.98, 4.07) 3.32 (2.74, 3.86) <0.001 0.1 0.006HDL cholesterol (mmol/L) 1.71 (1.53, 1.88) 1.61 (1.39, 1.88) 1.33 (1.17, 1.56) 0.01 <0.001 <0.001HDL2 cholesterol (mmol/L) 0.76 (0.64, 0.96) 0.73 (0.58, 0.96) 0.57 (0.45, 0.72) 0.3 <0.001 <0.001HDL3 cholesterol (mmol/L) 0.93 (0.85, 0.99) 0.87 (0.77, 0.99) 0.77 (0.67, 0.86) 0.001 <0.001 <0.001cholesterol/HDL cholesterol 3.01 (2.77, 3.29) 3.48 (3.01, 4.15) 3.92 (3.31, 4.79) <0.001 <0.001 <0.001non-HDL cholesterol (mmol/L) 3.40 (3.15, 3.69) 3.95 (3.45, 4.57) 3.91 (3.31, 4.57) <0.001 <0.001 0.3

Standardisation was to age 50 years, BMI 25 kg/m2 and zero smoking, alcohol, exercise and pregnancies: median (IQ range). Sampling periods are shown for each group. Significances: Kruskall-Wallis ANOVA all returned p<0.001 for significant variation between the three groups. Post-hoc, between-group comparisons were made using Mann-Whitney U test.Abbreviations: BMI: body mass index, LDL: low-density lipoprotein, HDL: high density lipoprotein, apo: apolipoprotein

23

Figure legends

Figure 1: Mean and 95 percent confidence intervals (derived as back transformed from transformed data, as appropriate ) for apolipoprotein B (Apo B), apolipoprotein A-I (Apo A-I) and apolipoprotein A-II (Apo A-II) concentrations by age: closed circles, continuous line: premenopausal women aged 40-45 or less, postmenopausal women aged 40-45 or more; open squares, dotted line: men. Significances * p<0.05, † p<0.01, § p<0.001

Figure 2: Mean and 95 percent confidence intervals (derived as back transformed from transformed data, as appropriate ) for apolipoprotein B / apolipoprotein A-I (Apo B/Apo A-I) ratio, LDL cholesterol / apolipoprotein B (LDL-C/Apo B) ratio and HDL cholesterol / apolipoprotein A-I (HDL-C/Apo A-I) ratio by age: closed circles, continuous line: premenopausal women aged 40-45 or less, postmenopausal women aged 40-45 or more; open squares, dotted line: men. Significances * p<0.05, † p<0.01, § p<0.001

24

25