- Email: [email protected]
Fat distribution and gender differences in serum lipids in men and women from four European communities
Atherosclerosis, 87 (1991) 203-210 p 1991 Elsevier Scientific Publishers ADONIS
203 Ireland,
Ltd. 0021-9150/91/$03.50
0021915091OGO988
ATHERO
04618
Fat distribution and gender differences in serum lipids in men and women from four European communities Jacob C. Seidell ‘, Massimo Cigolini 2, Jadviga Charzewska 3, Britt-Marie Ellsinger 4, Per Bjijmtorp 4, Joseph G.A.J. Hautvast ’ and Wictor Szostak 3 I Department of Human Nutrition, Agricultural University, Wageningen (The Netherlands), -’Institute of Clinical Medicine, University of Verona, Verona (Italy), ’ National Institute of Food Science, Warsaw (Poland), 4 Department of Medicine I, University of Gothenburg, Gothenburg (Sweden) (Received 17 August, 1990) (Revised, received 10 December, 1990) (Accepted 18 December, 1990)
Summary
We studied male/female differences in serum lipids in randomly selected 38-year-old men (n = 337) and women (n = 342) from various cities in The Netherlands, Sweden, Italy, and Poland. Overall, men had higher triglycerides and total cholesterol levels and lower HDL-levels compared to women (P -c0.001). Adjustment for smoking habits, city, and body mass index did not remove the gender difference. Further adjustments for waist circumference alone and waist/hip and waist/thigh circumference ratio removed the gender differences in serum triglycerides and total cholesterol. Only adjustment for waist/ thigh ratio removed the gender difference in HDL-cholesterol but linear relationships were different in men and women. The average male/female difference in serum lipids, particularly for total and LDL-cholesterol varied considerably among centers. In analyses of the data from the separate centers we found that sex differences in serum triglycerides and HDL-cholesterol in all 4 centers disappeared when adjusted for waist circumference alone and for waist/hip and waist/ thigh ratio. For total and LDL-cholesterol, however, adjustment for circumference ratios tended to increase the male/female difference in 2 of the 4 centers. It is concluded that, in European men and women, fat distribution may be responsible for male/female differences in serum triglycerides but that such conclusions are less clear for HDL-, totaland LDL-cholesterol.
Key words: Obesity; Fat distribution;
Lipids; Cholesterol: Triglycerides
Introduction Correspondence to; J.C. Seidell, Department Nutrition, P.O. Box 8129, 6700 EV Wageningen, lands. Tel.: (08370)82574; Fax: (08370)83342.
of Human The Nether-
Circumference ratios such as waist/hip ratio and waist/thigh ratios are important predictors of
cardiovascular disease, stroke, diabetes mellitus type II, and total mortality in men [l-3]. The risks associated with an abdominal fat distribution are thought to be largely mediated through risk factors such as blood pressure, serum lipids and lipoproteins and insulin resistance [4]. We have previously shown that indicators of body fat distribution in 3%year old women and men from various European countries were related to serum lipids [5,6]. In two recent publications, investigators reported that in American populations indicators of fat distribution are responsible for the sex differences in some serum lipids and lipoproteins [7,8]. Sex differences in serum lipids may vary considerably in different European communities and although in all populations women have a more favorable fat distribution compared to men they may have higher serum cholesterol levels [9]. We therefore decided to investigate this apparant paradox and to analyse whether or not sex differences in serum lipids can be explained by differences in fatness or fat distribution in populations from different parts of Europe. Material and methods Populations
In 4 cities of various European countries lists of all women born in 1948 (in 1986) and men born in 1950 (in 1988) were obtained by using voting lists, up-dated population registers, or birth registers. From these lists random selections of between 200 and 280 women and men were made who, after exclusion of immigrant names, were asked by telephone to participate in a study on cardiovascular risk factors in Europe. A number of subjects was unreachable (moved away, had died recently or travelled). Those who were asked to participate but refused to do so were asked to answer items on a non-responders questionnaire which included items such as height, weight, socioeconomic status, and chronic illness. Table 1 shows the number of participants. The response rate ranged from 64.4% in Dutch women to 94.0% in Polish men. In none of the centers were there significant differences between the body mass index and educational level of participants and non-participants who answered the telephone questionnaire. Subjects on medication known to influence serum lipids were
TABLE
1
PARTICIPATION RATE IN THE EUROPEAN TRIBUTION STUDY IN MEN AND WOMEN
Sweden (Gothenburg) Poland (Warsaw) Netherlands @de) Italy (Verona) a Participation jects invited
FAT
Study in women
Study in men
n
%a
n
S”
88 92 85 87
82.2 92.0 64.4 87.0
83 94 72 100
71.6 94.0 14.2 93.5
rate = (number for participation)
of participants/number * 100.
DIS-
of sub-
excluded from the analysis. Results of 342 women and 337 men were used in the analysis. Information on smoking habits was obtained from questionnaires and coded as follows: never smoked, stopped smoking, currently smoking less than 15 cigarettes/day, currently smoking more than 15 cig./day. Anthropometty
All anthropometric measurements were performed by one or two trained observers from each participating center. All observers received a training in The Netherlands and all worked according to a detailed working plan describing all the measurements. All subjects were measured while wearing only underwear. Height (to the nearest mm) and weight (to the nearest 0.5 kg) were recorded. Body mass index was calculated as weight/ height *. Circumferences were measured in duplicate on subjects in a standing position at the end of gentle aspiration. Waist circumference was measured midway between the lower rib margin and the iliac crest, the hip circumference at the level of the widest circumference over the great trochanters, and the thigh circumference as the horizontal circumference at the level of the gluteal fold on the right thigh. Blood sampling and analysis
Venous blood was sampled after an overnight fast. Serum was obtained by low-speed centrifugation within 1 h of venipuncture, stored at -80°C and transported by airmail in a frozen state to the lipid laboratory of the Department of Human Nutrition in Wageningen, The Netherlands. The
205 samples were analyzed enzymatically at the end of the study to determine the levels of total and HDL cholesterol and triglycerides [lo-121. The coefficient of variation percentage of control sera within one run was 0.9% for total cholesterol, 1.3% for HDL cholesterol, and 1.7% for triglycerides. Accuracy was checked by analysis of three serum pools of known value provided by the Centers for Disease Control (Atlanta, GA) and, for HDL cholesterol only of three pools produced by the North-West Lipid Research Clinic [13]. Mean bias with regard to target values of the Centers for Disease Control was 0.01 mmol/l for total cholesterol and 0.08 mmol/l for triglycerides. Mean bias with regard to the North-West Lipid Research Clinic target value for HDL cholesterol was -0.06 mmol/l. The LDL cholesterol concentration was calculated using the Friedewald equation [14]. Statistical methods Differences between men and women were evaluated by the t-test. The effects of covariates on the gender differences in serum lipids were performed in multiple regression analysis with serum lipids as dependent variables and gender and covariates as independent variables. Interaction was tested between gender and covariates in the regression by introducing their cross-product terms in the model. All P-values are two-sided and were considered significant at P < 0.05. Results
Table 2 shows the average levels of serum lipids in men and women as well as anthropometric variables. Men had, on the average, higher body mass index and circumference ratios. In addition, men had higher concentrations of triglycerides, total cholesterol, and LDL-cholesterol but lower levels of HDL-cholesterol. Among men there were more heavy smokers and fewer who had never smoked compared to women. Table 3 shows the gender differences in the pooled material adjusted for differences between centers. Adjustment for smoking habits and body mass index reduced the gender differences by about 22-30% but all gender differences were still highly significant. Further adjustment for waist
TABLE
2
DESCRIPTION OF VARIABLES IN 38-YEAR-OLD MEN AND WOMEN FROM 4 DIFFERENT EUROPEAN COUNTRIES Values are mean i SEM. Variable Anthropometry BMI (kg/m2) Waist (cm) Waist/hip ratio Waist/thigh ratio Serum lipids (mmol/l) triglycerides total cholesterol LDL-cholesterol HDL-cholesterol Smoking habits (%) never smoked ex-smoker smoking < 15 cigs./day smoking > 15 cigs./day
Women
Men
23.6 +0.2 78.1 &OS 0.79 f 0.003 1.33+0.005
25.0 90.3 0.93 1.57
0.93 5.48 3.62 1.44
+ f + f
0.02 0.05 0.05 0.02
t-Value
+0.2 +0.5 f 0.003 i 0.006
1.3450.04 5.88 + 0.06 3.99 + 0.05 1.25 + 0.02
5.4 17.8 34.3 29.6 8.4 5.1 5.2 8.6
44.2% 18.7%
27.4% 24.1%
5.7 1.7
26.0%
20.8%
1.6
11.1%
27.1%
5.6
circumference alone and waist/hip and waist/ thigh circumference ratio reduced the gender differences in serum triglycerides and total cholesterol levels much more and significant differences between men and women were no longer observed for these serum lipids. Gender differences in serum HDL levels were only reduced to non-significant levels after adjustment for waist/thigh ratio. Statistical interaction of gender and fat distribution measurements (waist, waist/ hip, and waist/ thigh) were not significant except for HDL-cholesterol indicating that, in the combined material, linear relationships between fat distribution and HDLcholesterol were different between men and women. Table 4 shows the regression coefficients for indicators of fat distribution in relation to serum lipids in men and women. A significant P-value for interaction indicates that the regression lines are significantly different in men compared to women. For triglycerides and total cholesterol no significant interaction was observed but for HDL-cholesterol the sex-specific regression equations clearly demonstrate that regression lines in men were different to those in women. Fig. 1 shows the linear relationship in men and women
206 TABLE 3 GENDER DIFFERENCES IN SOME SERUM LIPIDS IN 3%YEAR-OLD EUROPEAN MEN AND WOMEN (n = 670) AND THE EFFECTS OF ADJUSTMENT FOR SMOKING AND BMI AND MEASURES OF FAT DISTRIBUTION ON THESE GENDER DIFFERENCES Serum lipid (mmol/l)
Adjustment
Triglycerides (log)
unadjusted a adjusted b adjusted + waist adjusted + WHR adjusted + WTR
0.328 f 0.037 0.231+ 0.036 0.064 f0.050 0.030 * 0.059 0.035 + 0.052
unadjusted a adjusted b adjusted + waist adjusted + WHR adjusted + WTR unadjusted ’ adjusted b adjusted + waist adjusted + WHR adjusted + WTR
HDL
Total cholesterol
Sex difference f SEM (in mmol/l)
P-value
I%reduction of difference
< 0.0001 < 0.0001 0.20 0.61 0.50
29.6 80.5 90.9 89.3
-0.199kO.023 -0.155f0.023 - 0.073 + 0.032 -0.lOOf0.038 - 0.049 f 0.034
< 0.001 < 0.001 0.024 0.013 0.15
22.1 63.3 49.7 75.4
0.397 f 0.079 0.296 zt 0.082 0.120*0.117 -0.224kO.136 0.046f0.121
< O.O@Ol < 0.0001 0.30 0.10 0.71
25.4 69.8 43.6 88.4
’ Unadjusted means corrected for differences in centers only (by dummy variables for centers in the regression model). b Adjusted is corrected for differences in centers, smoking habits and body mass index. WHR = waist/hip ratio; WTR = waist/thigh ratio. TABLE 4 REGRESSION COEFFICIENTS FOR INDICATORS OF FAT DISTRIBUTION VERSUS SERUM LIPIDS IN 38-YEAR-OLD EUROPEAN MEN AND WOMEN (REGRESSION COEFFICIENTS ADJUSTED FOR BMI, SMOKING HABITS AND CITY) Dependent variable
lo&triglycerides) HDL cholesterol total cholesterol log(triglycerides) HDL cholesterol total cholesterol log(triglycerides) HDL cholesterol total cholesterol
Gender
Regression coefficients intercept
P
women men women men women men
- 1.45 - 1.85 -0.44 -0.37 4.00 3.66
0.016 Waist 0.022 Waist - 0.012 Waist - 0.008 Waist 0.017 waist 0.023 Waist
women men women men women men
- 2.07 - 2.23 2.67 1.53 2.55 0.64
1.48 WHR 1.70 WHR - 1.03 WHR 0.17 WHR 3.18 WHR 5.13 WHR
women men women men women men
- 2.21 - 2.12 2.86 2.01 3.03 3.03
0.91 WTR 0.88 WTR -0.78 WTR - 0.22 WTR 1.13 WTR 0.16 WTR
* P-value for interaction indicates the significance of the cross-product linear regression model.
Interaction P-value * 0.15 0.07 0.53 0.74 0.004 0.19 0.90 0.008 0.97
term of gender * fat distribution indicator in the multiple
207 for waist/ thigh ratio and serum triglycerides illustrating the identical regression equations in both sexes. Table 5 shows the analysis for total cholesterol in the four centers separately. The differences between men and women for total and LDLcholesterol varied considerably from country to country with practically no gender difference in The Netherlands and Poland whereas Italian men had considerably higher serum cholesterol levels compared to Italian women. Adjustment for smoking habits and body mass index reduced the gender differences in serum cholesterol in The Netherlands but hardly in any of the other centers. Further adjustment of gender differences in serum cholesterol for waist, waist/ thigh or waist/hip resulted in a tendency towards lower cholesterol levels in men compared to women in three of the 4 centers (significant after adjustment for waist/hip in The Netherlands). After adjustment for waist and waist/ thigh ratio, men still had higher serum cholesterol levels compared to women in the Italian population. Of all interactions tested only waist * gender and waist/hip * gender in the
TABLE
5
GENDER DIFFERENCES (MEN/WOMEN) IN SERUM TOTAL CHOLESTEROL LEVELS IN 4 EUROPEAN CENTERS AND EFFECTS OF ADJUSTMENT FOR ANTHROPOMETRIC VARIABLES Center
Adjustment for
Sex difference (SEM) (in mmol/l)
Pvalue *
Netherlands
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking + WHR smoking + WTR
0.25 0.05 - 0.35 - 0.65 - 0.36
(0.18) (0.18) (0.29) (0.31) (0.20)
0.16 0.78 0.23 0.04 0.20
Sweden
none BMI BMI BMI BMI
+ + + +
smoking smoking+ waist smoking + WHR smoking + WTR
0.32 0.35 0.27 - 0.00 - 0.16
(0.13) (0.13) (0.16) (0.20) (0.23)
0.02 0.01 0.10 1.00 0.48
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking + WHR smoking + WTR
0.92 0.81 0.79 0.42 0.67
(0.16) (0.17) (0.25) (0.36) (0.24)
i 0.001 < 0.001 0.002 0.25 0.005
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking+ WHR smoking + WTR
0.07 0.02 0.38 0.36 0.20
(0.15) (0.17) (0.28) (0.28) (0.23)
0.66 0.90 0.18 0.20 0.40
Italy
Poland
* P-value
-
for the test Ha: sex difference
= 0.
_.
125 4
.
.-
:
.
100 075
-125
”
0
102
109
116
123
130
137
144
151
158
165
172 179 walstlthigh
Fig. 1. Plot of serum triglycerides versus waist/thigh ratio in 38-year-old European women (0) and men (0). The plotted lines are the regression equations. A logarithmic transformation of serum triglycerides was used to improve the linearity of the associations. Because of the large number of subjects studied not all values could be shown (215 values hidden).
Swedish population reached statistical significance. Table 6 shows a similar center-specific analysis for HDL-cholesterol. The sex differences in HDL-cholesterol concentrations were quite similar in all centers and after adjustments for smoking habits and body mass index these gender differences remained significant. Further adjustment for waist or circumference ratios reduced the sex differences considerably and to non-significant levels in all centers. In addition, the center specific analysis for serum triglycerides, shown in Table 7, was very similar for all centers. In Poland, adjustment for smoking habits and body mass index reduced the gender difference in serum triglycerides considerably. Adjustment for waist and circumference ratios led to a similarly large reduction in the gender difference in all 4 centers.
208 Discussion
TABLE
The main purpose of this paper was to verify recent observations in American populations [7,8] which suggested that male/female differences in serum lipids (HDL-cholesterol and serum triglycerides in particular) can be explained by male/female differences in regional fat distribution. In our multicenter study we could test this hypothesis in four populations with a large variation in male/female differences in serum lipids. Men and women differ considerably in fat distribution with only a minor degree of overlap [7,8] and statistical manipulations should be viewed with caution. Statistical adjustment for other differences between men and women, such as breast volume or facial hair growth could also potentially remove the male/female differences in serum lipids in statistical analysis without any biological
TABLE
Center
Netherlands
Sweden
Italy
Poland
* P-value
Center
Adjustment for
Sex difference
P-
(SEM) (in mmol/l)
value *
none BMI BMI BMI BMI
smoking smoking+ waist smoking+ WHR smoking + WTR
- 0.24 - 0.22 - 0.13 - 0.09 -0.11
(0.04) (0.05) (0.08) (0.08) (0.08)
< 0.001
+ + + +
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking+ WHR smoking + WTR
-
0.21 0.19 0.12 0.12 0.09
(0.05) (0.05) (0.06) (0.07) (0.08)
< 0.001 < 0.001 0.04 0.10 0.29
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking + WHR smoking + WTR
-0.14 -0.10 0.04 0.00 0.08
(0.04) (0.05) (0.07) (0.10) (0.06)
< 0.001 0.04 0.56 0.98 0.17
none BMI BMI BMI BMI
+ + + +
smoking smoking+ waist smoking+ WHR smoking + WTR
-
(0.04) (0.05) (0.08) (0.08) (0.06)
< 0.001 0.01 0.43 0.20 0.17
0.22 0.12 0.06 0.10 0.09
for the test He: sex difference
= 0.
< 0.001 0.08 0.26 0.14
Adjustment for
Sex difference (SEM) (in log mmol/l)
Pvalue *
Netherlands
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking+ WHR smoking + WTR
0.46 0.40 0.20 0.13 0.14
(0.07) (0.07) (0.11) (0.12) (0.20)
< 0.001 < 0.001 0.08 0.29 0.21
Sweden
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking + WHR smoking+ WTR
0.22 0.21 0.11 0.02 - 0.04
(0.07) (0.07) (0.08) (0.10) (0.12)
0.002 0.003 0.19 0.82 0.76
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking+ WHR smoking + WTR
0.33 0.22 - 0.02 - 0.08 - 0.05
(0.08) (0.08) (0.10) (0.16) (0.10)
< 0.001 0.004 0.84 0.50 0.62
none BMI BMI BMI BMI
+ + + +
smoking smoking+ waist smoking+ WHR smoking + WTR
0.31 0.12 0.01 0.04 0.08
(0.07) (0.08) (0.12) (0.12) (0.10)
< 0.001 0.11 0.94 0.74 0.45
Italy
6
GENDER DIFFERENCES (MEN/WOMEN) IN SERUM HDL-CHOLESTEROL LEVELS IN 4 EUROPEAN CENTERS AND EFFECTS OF ADJUSTMENT FOR ANTHROPOMETRIC VARIABLES
7
GENDER DIFFERENCES (MEN/WOMEN) IN SERUM TRIGLYCERIDE LEVELS IN 4 EUROPEAN CENTERS AND EFFECTS OF ADJUSTMENT FOR ANTHROPOMETRIC VARIABLES
Poland
* P-value
for the test Ha: sex difference
= 0.
meaning. In order to conclude that fat distribution or a highly correlated characteristic is a major biological determinant for male/ female differences in serum lipids we think that at least four criteria should be fulfilled: (a) Analysis of covariance or multiple regression should show that the gender differences in serum lipids disappear after adjustment for indicators of fat distribution (the effects of smoking and degree of obesity should be taken into account). (b) Linear relations should be similar for men and women (i.e., statistical interaction between gender and fat distribution should not be significant). (c) The results should be consistent in different populations. (d) There should be a plausible explanation for the disappearance of gender differences in serum lipids after adjustment for fat distribution.
209 In the present study gender differences generally disappeared after adjustment for smoking habits, body mass index plus one indicator of fat distribution. Waist alone as well as waist/hip and waist/ thigh were not clearly superior to each other in this respect. For HDL-cholesterol, however, there were considerable differences in regression coefficients for waist/hip and waist/ thigh ratio between men and women. Gender differences in total cholesterol were considerably different among centers and adjustment for indicators of fat distribution increased the gender difference in The Netherlands and in Poland while in Italy gender differences remained. The only serum lipid for which the first 3 criteria we mentioned above were fulfilled is serum triglycerides. The relationships between waist circumference and circumference ratios on the one hand and serum triglycerides on the other were very similar in men and women and consistent among the four centers. Our findings agree only partially with those observed by Freedman et al. [7] who also observed the largest reduction in male/female differences for serum triglycerides. In their population, men and women did not differ in serum total cholesterol levels and adjustement for fat distribution would have been likely to cause, as in our study, an increase in gender difference. Our findings do not agree with those of Ostlund et al. [8] who observed similar linear associations between waist/hip and HDL,cholesterol in elderly men and women. One explanation for the discrepancy between our studies and the study of Ostlund et al. is the difference between the two age of the populations studied. All women in our study were premenopausal while Ostlund et al. studied postmenopausal women. It is possible that hormonal changes associated with menopause could affect the linear relationships between fat distribution and serum lipids. There are several possibilities for a disappearance of gender differences in serum triglycerides after adjustment for fat distribution. One possibility is that with matching for fat distribution one matches for the absolute amount of visceral adipose tissue. Krotkiewski et al. were among the first to show that for a given degree of fatness, men have higher levels of risk factors compared to women and they suggested that the high waist/hip ratio in men could be responsible
for this observation [15]. It has been shown that, in both men and women, a high waist/hip and or waist/ thigh is associated with increased accumulation of visceral fat [16,17]. The total amount of visceral fat could be a major determinant of hepatic exposure to free fatty acids which are released by the portal adipose tissues [18]. It has not been shown, however, that adjustment for waist or circumference ratios would remove the gender difference in total visceral fat volume. Another possibility is that by adjusting for fat distribution one adjusts for differences in sex hormone levels. We have previously shown that in the women in this study, serum levels of free testosterone were weakly but significantly related to the indicators of fat distribution as well as to serum lipids [19]. We feel that it is unlikely that free testosterone levels explain the observed gender difference in serum lipids because levels in men are known to be lo-30-fold higher compared to levels observed in women [20]. The relationships between fat distribution and serum total or LDLcholesterol are less clear both from an epidemiologic point of view [5,6] as well as from a mechanistic point of view [21]. It has been proposed that there is a relationship between abdominal fat distribution and smaller and more dense LDL particles but which will not be reflected in the total or LDL-cholesterol concentration [21]. It should be noted that all analysis we presented for total cholesterol levels were very similar for LDLcholesterol. The negative relationship between fat distribution and HDL-cholesterol is consistently found in epidemiological studies [5,6] and is possibly related to hepatic triglyceride lipase activity [21]. No direct causal association between visceral fat accumulation and HDL-cholesterol levels has been demonstrated. The relationship between fat distribution and serum triglycerides (or VLDL-cholesterol levels) is thus, among the relationships studied, among the biologically most plausible relationships. It is likely that with adjustment for indicators of fat distribution one reduces the gender difference in visceral fat accumulation and the exposure of the liver to free fatty acids. In conclusion, we propose that the suggested major effect of fat distribution on male/female
210 differences in serum lipids and lipoproteins seems consistent for serum triglycerides. The reduction in gender differences in serum total cholesterol and HDL-cholesterol is clearly present but either the findings are not consistent in men and women or not consistent among centers. Acknowledgements
The authors thank all participants for their cooperation and the laboratory personnel of the Department of Human Nutrition, Wageningen University, for expert technical assistance. This study is part of the concerted action project on Nutrition and Health (EURO-NUT) of the Comae-Epidemiology Group within the Medical Research Council of the European Community. The guidance and support of the Project Management Group of EURO-NUT is gratefully acknowledged. Dr. Jacob C. Seidell is a research fellow of the Royal Netherlands Academy of Science. References Larsson, B., Svardsudd, K., WeIin, L., Wilhelmsen, L., Bjijrntorp, P. and Tibblin, G., Abdominal adipose tissue distribution, obesity, and risk of cardiovascular disease and death: a 13 year follow-up of participants in the study of men born in 1913, Br. Med. J., 288 (1984) 1401. Ohlsson, L.O., Larsson, B., Svlrdsudd, K., Welin, L., Eriksson, H., Wilhelmsen, L., Bjomtorp, P. and Tibblin, G., The influence of body fat distribution on the incidence of diabetes mellitus: 13.5 years of follow-up of the participants in the study of men born in 1913, Diabetes, 34 (1985) 1055. Welin, L., Svardsudd, K., Wilhelmsen, L., Larsson, B. and Tibblin, G., Family history and other risk factors for stroke. The study of men born in 1913, N. EngI. J. Med., 317 (1987) 521. Bjomtorp, P., The associations between obesity, adipose tissue distribution and disease, Acta Med. Stand., Suppl.. 723 (1988) 121. Seidell, J.C., Cigolini, M., Charzewska, J., Ellsinger, B.-M., Dibiase, G., Bjomtorp, P., Hautvast, J.G.A.J., Contaldo, F., Szostak ,V. and Scuro, L.A., Indicators of fat distribution, serum lipids, and blood pressure in European women born in 1948 - the European fat distribution study, Am. J. Epidemiol., 30 (1989) 53. Seidell, J.C., Cigolini, M., Deslypere, J.-P., Charzewska, J., Ellsinger, B.-M. and Cruz, A., Body fat distribution in relation to serum lipids and blood pressure in 38-year old men - the European fat distribution study, Atherosclerosis 86 (1991) 251-260.
7 Freedman, D.S., Jacobsen, S.J., Barboriak, J.J., Sobocinski, K.A., Anderson, A.J., Kissebah, A.H., Sasse, E.A. and Gruchow, H.W., Body fat distribution and male/female differences in lipids and lipoproteins, Circulation, 81 (1990) 1498. 8 Ostlund, R.E., Staten, M., Kohrt, W.M., Schultz, J. and Malley, M., The ratio of waist-to-hip circumference, plasma insulin level, and glucose tolerance as independent predictors of the HDL, cholesterol level in older adults, N. EngI. J. Med., 332 (1990) 229. 9 WHO Monica Project, Risk factors,. Int. J. Epidemiol., 18 (1989) 546. 10 Roschlau, P., Bemt, E. and Gruber, W., Enzymatische Bestimmung des Gesamt-Cholesterins im Serum, Z. Klin. Chem. Khn. B&hem., 12 (1974) 403. 11 Wamick, G.R., Benderson, J. and Albers, J.J., Dextran sulfate Mg2+ precipitation procedure for quantification of high-density-lipoprotein cholesterol, Clin. Chem., 28 (1982) 1379. 12 Sullivan, D.R., Kruijswijk, K., West, C.E., Kohlmeier, M. and Katan, M.B., Determination of serum triglycerides by an accurate enzymatic method not affected by free glycerol, Clin. Chem., 31 (1985) 1227. 13 Wamick, G.R. and Clapshaw, P., Availability of plasma with target values for certain lipids, Clin. Chem., 33 (1987) 2323. 14 Friedewald, W.T., Levy, RI. and Frederickson, D.S., Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge, Clin. Chem., 18 (1972) 499. 15 Krotkiewski, M., Bjomtorp, P., Sjostriim, L. and Smith, U., Impact of obesity on metabolism in men and women, J. Clin. Invest., 71 (1983) 1150. 16 Ashwell, M., Cole, T.J. and Dixon, A.K., Obesity: new insight into the anthropometric classification of fat distribution shown by computed tomography, Br. Med. J., 290 (1985) 1692. 17 Seidell, J.C., Oosterlee, A., Thijssen, M.A.O., Burema, J., Deurenberg, P. and Hautvast, J.G.A.J., Assessment of intraabdominal and subcutaneous abdominal fat - relation between anthropometry and computed tomography, Am. J. Clin. Nutr., 45 (1987) 7. 18 Bjdmtorp, P., “Portal” adipose tissue as a generator of risk factors for cardiovascular disease and diabetes, Arteriosclerosis, 10 (1990) 493. 19 Seidell, J.C., Cigolini, M., Charzewska, J., Ellsinger, B.-M., Dibiase, G., Bjiirntorp, P., Hautvast, J.G.A.J., ContaIdo, F., Szostak, V. and Scuro, L.A. Androgenicity in relation to body fat distribution and metabolism in 38-year old women - the European fat distribution study, J. Clin. Epidemiol., 43 (1990) 21. 20 Nestler, J.E., Clore, J.N. and Blackard, W.G., The central role of obesity (hyperinsulinemia) in the pathogenesis of the polycystic ovary syndrome, Am. J. Obstet. Gynecol., 16 (1989) 1095. 21 Despres, J.-P., Mootjani, S., Lupien, P.J., Tremblay, A., Nadeau, A. and Bouchard, C., Regional distribution of body fat, plasma lipoproteins, and cardiovascular disease, Arteriosclerosis, 10 (1990) 497.
203 Ireland,
Ltd. 0021-9150/91/$03.50
0021915091OGO988
ATHERO
04618
Fat distribution and gender differences in serum lipids in men and women from four European communities Jacob C. Seidell ‘, Massimo Cigolini 2, Jadviga Charzewska 3, Britt-Marie Ellsinger 4, Per Bjijmtorp 4, Joseph G.A.J. Hautvast ’ and Wictor Szostak 3 I Department of Human Nutrition, Agricultural University, Wageningen (The Netherlands), -’Institute of Clinical Medicine, University of Verona, Verona (Italy), ’ National Institute of Food Science, Warsaw (Poland), 4 Department of Medicine I, University of Gothenburg, Gothenburg (Sweden) (Received 17 August, 1990) (Revised, received 10 December, 1990) (Accepted 18 December, 1990)
Summary
We studied male/female differences in serum lipids in randomly selected 38-year-old men (n = 337) and women (n = 342) from various cities in The Netherlands, Sweden, Italy, and Poland. Overall, men had higher triglycerides and total cholesterol levels and lower HDL-levels compared to women (P -c0.001). Adjustment for smoking habits, city, and body mass index did not remove the gender difference. Further adjustments for waist circumference alone and waist/hip and waist/thigh circumference ratio removed the gender differences in serum triglycerides and total cholesterol. Only adjustment for waist/ thigh ratio removed the gender difference in HDL-cholesterol but linear relationships were different in men and women. The average male/female difference in serum lipids, particularly for total and LDL-cholesterol varied considerably among centers. In analyses of the data from the separate centers we found that sex differences in serum triglycerides and HDL-cholesterol in all 4 centers disappeared when adjusted for waist circumference alone and for waist/hip and waist/ thigh ratio. For total and LDL-cholesterol, however, adjustment for circumference ratios tended to increase the male/female difference in 2 of the 4 centers. It is concluded that, in European men and women, fat distribution may be responsible for male/female differences in serum triglycerides but that such conclusions are less clear for HDL-, totaland LDL-cholesterol.
Key words: Obesity; Fat distribution;
Lipids; Cholesterol: Triglycerides
Introduction Correspondence to; J.C. Seidell, Department Nutrition, P.O. Box 8129, 6700 EV Wageningen, lands. Tel.: (08370)82574; Fax: (08370)83342.
of Human The Nether-
Circumference ratios such as waist/hip ratio and waist/thigh ratios are important predictors of
cardiovascular disease, stroke, diabetes mellitus type II, and total mortality in men [l-3]. The risks associated with an abdominal fat distribution are thought to be largely mediated through risk factors such as blood pressure, serum lipids and lipoproteins and insulin resistance [4]. We have previously shown that indicators of body fat distribution in 3%year old women and men from various European countries were related to serum lipids [5,6]. In two recent publications, investigators reported that in American populations indicators of fat distribution are responsible for the sex differences in some serum lipids and lipoproteins [7,8]. Sex differences in serum lipids may vary considerably in different European communities and although in all populations women have a more favorable fat distribution compared to men they may have higher serum cholesterol levels [9]. We therefore decided to investigate this apparant paradox and to analyse whether or not sex differences in serum lipids can be explained by differences in fatness or fat distribution in populations from different parts of Europe. Material and methods Populations
In 4 cities of various European countries lists of all women born in 1948 (in 1986) and men born in 1950 (in 1988) were obtained by using voting lists, up-dated population registers, or birth registers. From these lists random selections of between 200 and 280 women and men were made who, after exclusion of immigrant names, were asked by telephone to participate in a study on cardiovascular risk factors in Europe. A number of subjects was unreachable (moved away, had died recently or travelled). Those who were asked to participate but refused to do so were asked to answer items on a non-responders questionnaire which included items such as height, weight, socioeconomic status, and chronic illness. Table 1 shows the number of participants. The response rate ranged from 64.4% in Dutch women to 94.0% in Polish men. In none of the centers were there significant differences between the body mass index and educational level of participants and non-participants who answered the telephone questionnaire. Subjects on medication known to influence serum lipids were
TABLE
1
PARTICIPATION RATE IN THE EUROPEAN TRIBUTION STUDY IN MEN AND WOMEN
Sweden (Gothenburg) Poland (Warsaw) Netherlands @de) Italy (Verona) a Participation jects invited
FAT
Study in women
Study in men
n
%a
n
S”
88 92 85 87
82.2 92.0 64.4 87.0
83 94 72 100
71.6 94.0 14.2 93.5
rate = (number for participation)
of participants/number * 100.
DIS-
of sub-
excluded from the analysis. Results of 342 women and 337 men were used in the analysis. Information on smoking habits was obtained from questionnaires and coded as follows: never smoked, stopped smoking, currently smoking less than 15 cigarettes/day, currently smoking more than 15 cig./day. Anthropometty
All anthropometric measurements were performed by one or two trained observers from each participating center. All observers received a training in The Netherlands and all worked according to a detailed working plan describing all the measurements. All subjects were measured while wearing only underwear. Height (to the nearest mm) and weight (to the nearest 0.5 kg) were recorded. Body mass index was calculated as weight/ height *. Circumferences were measured in duplicate on subjects in a standing position at the end of gentle aspiration. Waist circumference was measured midway between the lower rib margin and the iliac crest, the hip circumference at the level of the widest circumference over the great trochanters, and the thigh circumference as the horizontal circumference at the level of the gluteal fold on the right thigh. Blood sampling and analysis
Venous blood was sampled after an overnight fast. Serum was obtained by low-speed centrifugation within 1 h of venipuncture, stored at -80°C and transported by airmail in a frozen state to the lipid laboratory of the Department of Human Nutrition in Wageningen, The Netherlands. The
205 samples were analyzed enzymatically at the end of the study to determine the levels of total and HDL cholesterol and triglycerides [lo-121. The coefficient of variation percentage of control sera within one run was 0.9% for total cholesterol, 1.3% for HDL cholesterol, and 1.7% for triglycerides. Accuracy was checked by analysis of three serum pools of known value provided by the Centers for Disease Control (Atlanta, GA) and, for HDL cholesterol only of three pools produced by the North-West Lipid Research Clinic [13]. Mean bias with regard to target values of the Centers for Disease Control was 0.01 mmol/l for total cholesterol and 0.08 mmol/l for triglycerides. Mean bias with regard to the North-West Lipid Research Clinic target value for HDL cholesterol was -0.06 mmol/l. The LDL cholesterol concentration was calculated using the Friedewald equation [14]. Statistical methods Differences between men and women were evaluated by the t-test. The effects of covariates on the gender differences in serum lipids were performed in multiple regression analysis with serum lipids as dependent variables and gender and covariates as independent variables. Interaction was tested between gender and covariates in the regression by introducing their cross-product terms in the model. All P-values are two-sided and were considered significant at P < 0.05. Results
Table 2 shows the average levels of serum lipids in men and women as well as anthropometric variables. Men had, on the average, higher body mass index and circumference ratios. In addition, men had higher concentrations of triglycerides, total cholesterol, and LDL-cholesterol but lower levels of HDL-cholesterol. Among men there were more heavy smokers and fewer who had never smoked compared to women. Table 3 shows the gender differences in the pooled material adjusted for differences between centers. Adjustment for smoking habits and body mass index reduced the gender differences by about 22-30% but all gender differences were still highly significant. Further adjustment for waist
TABLE
2
DESCRIPTION OF VARIABLES IN 38-YEAR-OLD MEN AND WOMEN FROM 4 DIFFERENT EUROPEAN COUNTRIES Values are mean i SEM. Variable Anthropometry BMI (kg/m2) Waist (cm) Waist/hip ratio Waist/thigh ratio Serum lipids (mmol/l) triglycerides total cholesterol LDL-cholesterol HDL-cholesterol Smoking habits (%) never smoked ex-smoker smoking < 15 cigs./day smoking > 15 cigs./day
Women
Men
23.6 +0.2 78.1 &OS 0.79 f 0.003 1.33+0.005
25.0 90.3 0.93 1.57
0.93 5.48 3.62 1.44
+ f + f
0.02 0.05 0.05 0.02
t-Value
+0.2 +0.5 f 0.003 i 0.006
1.3450.04 5.88 + 0.06 3.99 + 0.05 1.25 + 0.02
5.4 17.8 34.3 29.6 8.4 5.1 5.2 8.6
44.2% 18.7%
27.4% 24.1%
5.7 1.7
26.0%
20.8%
1.6
11.1%
27.1%
5.6
circumference alone and waist/hip and waist/ thigh circumference ratio reduced the gender differences in serum triglycerides and total cholesterol levels much more and significant differences between men and women were no longer observed for these serum lipids. Gender differences in serum HDL levels were only reduced to non-significant levels after adjustment for waist/thigh ratio. Statistical interaction of gender and fat distribution measurements (waist, waist/ hip, and waist/ thigh) were not significant except for HDL-cholesterol indicating that, in the combined material, linear relationships between fat distribution and HDLcholesterol were different between men and women. Table 4 shows the regression coefficients for indicators of fat distribution in relation to serum lipids in men and women. A significant P-value for interaction indicates that the regression lines are significantly different in men compared to women. For triglycerides and total cholesterol no significant interaction was observed but for HDL-cholesterol the sex-specific regression equations clearly demonstrate that regression lines in men were different to those in women. Fig. 1 shows the linear relationship in men and women
206 TABLE 3 GENDER DIFFERENCES IN SOME SERUM LIPIDS IN 3%YEAR-OLD EUROPEAN MEN AND WOMEN (n = 670) AND THE EFFECTS OF ADJUSTMENT FOR SMOKING AND BMI AND MEASURES OF FAT DISTRIBUTION ON THESE GENDER DIFFERENCES Serum lipid (mmol/l)
Adjustment
Triglycerides (log)
unadjusted a adjusted b adjusted + waist adjusted + WHR adjusted + WTR
0.328 f 0.037 0.231+ 0.036 0.064 f0.050 0.030 * 0.059 0.035 + 0.052
unadjusted a adjusted b adjusted + waist adjusted + WHR adjusted + WTR unadjusted ’ adjusted b adjusted + waist adjusted + WHR adjusted + WTR
HDL
Total cholesterol
Sex difference f SEM (in mmol/l)
P-value
I%reduction of difference
< 0.0001 < 0.0001 0.20 0.61 0.50
29.6 80.5 90.9 89.3
-0.199kO.023 -0.155f0.023 - 0.073 + 0.032 -0.lOOf0.038 - 0.049 f 0.034
< 0.001 < 0.001 0.024 0.013 0.15
22.1 63.3 49.7 75.4
0.397 f 0.079 0.296 zt 0.082 0.120*0.117 -0.224kO.136 0.046f0.121
< O.O@Ol < 0.0001 0.30 0.10 0.71
25.4 69.8 43.6 88.4
’ Unadjusted means corrected for differences in centers only (by dummy variables for centers in the regression model). b Adjusted is corrected for differences in centers, smoking habits and body mass index. WHR = waist/hip ratio; WTR = waist/thigh ratio. TABLE 4 REGRESSION COEFFICIENTS FOR INDICATORS OF FAT DISTRIBUTION VERSUS SERUM LIPIDS IN 38-YEAR-OLD EUROPEAN MEN AND WOMEN (REGRESSION COEFFICIENTS ADJUSTED FOR BMI, SMOKING HABITS AND CITY) Dependent variable
lo&triglycerides) HDL cholesterol total cholesterol log(triglycerides) HDL cholesterol total cholesterol log(triglycerides) HDL cholesterol total cholesterol
Gender
Regression coefficients intercept
P
women men women men women men
- 1.45 - 1.85 -0.44 -0.37 4.00 3.66
0.016 Waist 0.022 Waist - 0.012 Waist - 0.008 Waist 0.017 waist 0.023 Waist
women men women men women men
- 2.07 - 2.23 2.67 1.53 2.55 0.64
1.48 WHR 1.70 WHR - 1.03 WHR 0.17 WHR 3.18 WHR 5.13 WHR
women men women men women men
- 2.21 - 2.12 2.86 2.01 3.03 3.03
0.91 WTR 0.88 WTR -0.78 WTR - 0.22 WTR 1.13 WTR 0.16 WTR
* P-value for interaction indicates the significance of the cross-product linear regression model.
Interaction P-value * 0.15 0.07 0.53 0.74 0.004 0.19 0.90 0.008 0.97
term of gender * fat distribution indicator in the multiple
207 for waist/ thigh ratio and serum triglycerides illustrating the identical regression equations in both sexes. Table 5 shows the analysis for total cholesterol in the four centers separately. The differences between men and women for total and LDLcholesterol varied considerably from country to country with practically no gender difference in The Netherlands and Poland whereas Italian men had considerably higher serum cholesterol levels compared to Italian women. Adjustment for smoking habits and body mass index reduced the gender differences in serum cholesterol in The Netherlands but hardly in any of the other centers. Further adjustment of gender differences in serum cholesterol for waist, waist/ thigh or waist/hip resulted in a tendency towards lower cholesterol levels in men compared to women in three of the 4 centers (significant after adjustment for waist/hip in The Netherlands). After adjustment for waist and waist/ thigh ratio, men still had higher serum cholesterol levels compared to women in the Italian population. Of all interactions tested only waist * gender and waist/hip * gender in the
TABLE
5
GENDER DIFFERENCES (MEN/WOMEN) IN SERUM TOTAL CHOLESTEROL LEVELS IN 4 EUROPEAN CENTERS AND EFFECTS OF ADJUSTMENT FOR ANTHROPOMETRIC VARIABLES Center
Adjustment for
Sex difference (SEM) (in mmol/l)
Pvalue *
Netherlands
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking + WHR smoking + WTR
0.25 0.05 - 0.35 - 0.65 - 0.36
(0.18) (0.18) (0.29) (0.31) (0.20)
0.16 0.78 0.23 0.04 0.20
Sweden
none BMI BMI BMI BMI
+ + + +
smoking smoking+ waist smoking + WHR smoking + WTR
0.32 0.35 0.27 - 0.00 - 0.16
(0.13) (0.13) (0.16) (0.20) (0.23)
0.02 0.01 0.10 1.00 0.48
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking + WHR smoking + WTR
0.92 0.81 0.79 0.42 0.67
(0.16) (0.17) (0.25) (0.36) (0.24)
i 0.001 < 0.001 0.002 0.25 0.005
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking+ WHR smoking + WTR
0.07 0.02 0.38 0.36 0.20
(0.15) (0.17) (0.28) (0.28) (0.23)
0.66 0.90 0.18 0.20 0.40
Italy
Poland
* P-value
-
for the test Ha: sex difference
= 0.
_.
125 4
.
.-
:
.
100 075
-125
”
0
102
109
116
123
130
137
144
151
158
165
172 179 walstlthigh
Fig. 1. Plot of serum triglycerides versus waist/thigh ratio in 38-year-old European women (0) and men (0). The plotted lines are the regression equations. A logarithmic transformation of serum triglycerides was used to improve the linearity of the associations. Because of the large number of subjects studied not all values could be shown (215 values hidden).
Swedish population reached statistical significance. Table 6 shows a similar center-specific analysis for HDL-cholesterol. The sex differences in HDL-cholesterol concentrations were quite similar in all centers and after adjustments for smoking habits and body mass index these gender differences remained significant. Further adjustment for waist or circumference ratios reduced the sex differences considerably and to non-significant levels in all centers. In addition, the center specific analysis for serum triglycerides, shown in Table 7, was very similar for all centers. In Poland, adjustment for smoking habits and body mass index reduced the gender difference in serum triglycerides considerably. Adjustment for waist and circumference ratios led to a similarly large reduction in the gender difference in all 4 centers.
208 Discussion
TABLE
The main purpose of this paper was to verify recent observations in American populations [7,8] which suggested that male/female differences in serum lipids (HDL-cholesterol and serum triglycerides in particular) can be explained by male/female differences in regional fat distribution. In our multicenter study we could test this hypothesis in four populations with a large variation in male/female differences in serum lipids. Men and women differ considerably in fat distribution with only a minor degree of overlap [7,8] and statistical manipulations should be viewed with caution. Statistical adjustment for other differences between men and women, such as breast volume or facial hair growth could also potentially remove the male/female differences in serum lipids in statistical analysis without any biological
TABLE
Center
Netherlands
Sweden
Italy
Poland
* P-value
Center
Adjustment for
Sex difference
P-
(SEM) (in mmol/l)
value *
none BMI BMI BMI BMI
smoking smoking+ waist smoking+ WHR smoking + WTR
- 0.24 - 0.22 - 0.13 - 0.09 -0.11
(0.04) (0.05) (0.08) (0.08) (0.08)
< 0.001
+ + + +
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking+ WHR smoking + WTR
-
0.21 0.19 0.12 0.12 0.09
(0.05) (0.05) (0.06) (0.07) (0.08)
< 0.001 < 0.001 0.04 0.10 0.29
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking + WHR smoking + WTR
-0.14 -0.10 0.04 0.00 0.08
(0.04) (0.05) (0.07) (0.10) (0.06)
< 0.001 0.04 0.56 0.98 0.17
none BMI BMI BMI BMI
+ + + +
smoking smoking+ waist smoking+ WHR smoking + WTR
-
(0.04) (0.05) (0.08) (0.08) (0.06)
< 0.001 0.01 0.43 0.20 0.17
0.22 0.12 0.06 0.10 0.09
for the test He: sex difference
= 0.
< 0.001 0.08 0.26 0.14
Adjustment for
Sex difference (SEM) (in log mmol/l)
Pvalue *
Netherlands
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking+ WHR smoking + WTR
0.46 0.40 0.20 0.13 0.14
(0.07) (0.07) (0.11) (0.12) (0.20)
< 0.001 < 0.001 0.08 0.29 0.21
Sweden
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking + WHR smoking+ WTR
0.22 0.21 0.11 0.02 - 0.04
(0.07) (0.07) (0.08) (0.10) (0.12)
0.002 0.003 0.19 0.82 0.76
none BMI BMI BMI BMI
+ + + +
smoking smoking + waist smoking+ WHR smoking + WTR
0.33 0.22 - 0.02 - 0.08 - 0.05
(0.08) (0.08) (0.10) (0.16) (0.10)
< 0.001 0.004 0.84 0.50 0.62
none BMI BMI BMI BMI
+ + + +
smoking smoking+ waist smoking+ WHR smoking + WTR
0.31 0.12 0.01 0.04 0.08
(0.07) (0.08) (0.12) (0.12) (0.10)
< 0.001 0.11 0.94 0.74 0.45
Italy
6
GENDER DIFFERENCES (MEN/WOMEN) IN SERUM HDL-CHOLESTEROL LEVELS IN 4 EUROPEAN CENTERS AND EFFECTS OF ADJUSTMENT FOR ANTHROPOMETRIC VARIABLES
7
GENDER DIFFERENCES (MEN/WOMEN) IN SERUM TRIGLYCERIDE LEVELS IN 4 EUROPEAN CENTERS AND EFFECTS OF ADJUSTMENT FOR ANTHROPOMETRIC VARIABLES
Poland
* P-value
for the test Ha: sex difference
= 0.
meaning. In order to conclude that fat distribution or a highly correlated characteristic is a major biological determinant for male/ female differences in serum lipids we think that at least four criteria should be fulfilled: (a) Analysis of covariance or multiple regression should show that the gender differences in serum lipids disappear after adjustment for indicators of fat distribution (the effects of smoking and degree of obesity should be taken into account). (b) Linear relations should be similar for men and women (i.e., statistical interaction between gender and fat distribution should not be significant). (c) The results should be consistent in different populations. (d) There should be a plausible explanation for the disappearance of gender differences in serum lipids after adjustment for fat distribution.
209 In the present study gender differences generally disappeared after adjustment for smoking habits, body mass index plus one indicator of fat distribution. Waist alone as well as waist/hip and waist/ thigh were not clearly superior to each other in this respect. For HDL-cholesterol, however, there were considerable differences in regression coefficients for waist/hip and waist/ thigh ratio between men and women. Gender differences in total cholesterol were considerably different among centers and adjustment for indicators of fat distribution increased the gender difference in The Netherlands and in Poland while in Italy gender differences remained. The only serum lipid for which the first 3 criteria we mentioned above were fulfilled is serum triglycerides. The relationships between waist circumference and circumference ratios on the one hand and serum triglycerides on the other were very similar in men and women and consistent among the four centers. Our findings agree only partially with those observed by Freedman et al. [7] who also observed the largest reduction in male/female differences for serum triglycerides. In their population, men and women did not differ in serum total cholesterol levels and adjustement for fat distribution would have been likely to cause, as in our study, an increase in gender difference. Our findings do not agree with those of Ostlund et al. [8] who observed similar linear associations between waist/hip and HDL,cholesterol in elderly men and women. One explanation for the discrepancy between our studies and the study of Ostlund et al. is the difference between the two age of the populations studied. All women in our study were premenopausal while Ostlund et al. studied postmenopausal women. It is possible that hormonal changes associated with menopause could affect the linear relationships between fat distribution and serum lipids. There are several possibilities for a disappearance of gender differences in serum triglycerides after adjustment for fat distribution. One possibility is that with matching for fat distribution one matches for the absolute amount of visceral adipose tissue. Krotkiewski et al. were among the first to show that for a given degree of fatness, men have higher levels of risk factors compared to women and they suggested that the high waist/hip ratio in men could be responsible
for this observation [15]. It has been shown that, in both men and women, a high waist/hip and or waist/ thigh is associated with increased accumulation of visceral fat [16,17]. The total amount of visceral fat could be a major determinant of hepatic exposure to free fatty acids which are released by the portal adipose tissues [18]. It has not been shown, however, that adjustment for waist or circumference ratios would remove the gender difference in total visceral fat volume. Another possibility is that by adjusting for fat distribution one adjusts for differences in sex hormone levels. We have previously shown that in the women in this study, serum levels of free testosterone were weakly but significantly related to the indicators of fat distribution as well as to serum lipids [19]. We feel that it is unlikely that free testosterone levels explain the observed gender difference in serum lipids because levels in men are known to be lo-30-fold higher compared to levels observed in women [20]. The relationships between fat distribution and serum total or LDLcholesterol are less clear both from an epidemiologic point of view [5,6] as well as from a mechanistic point of view [21]. It has been proposed that there is a relationship between abdominal fat distribution and smaller and more dense LDL particles but which will not be reflected in the total or LDL-cholesterol concentration [21]. It should be noted that all analysis we presented for total cholesterol levels were very similar for LDLcholesterol. The negative relationship between fat distribution and HDL-cholesterol is consistently found in epidemiological studies [5,6] and is possibly related to hepatic triglyceride lipase activity [21]. No direct causal association between visceral fat accumulation and HDL-cholesterol levels has been demonstrated. The relationship between fat distribution and serum triglycerides (or VLDL-cholesterol levels) is thus, among the relationships studied, among the biologically most plausible relationships. It is likely that with adjustment for indicators of fat distribution one reduces the gender difference in visceral fat accumulation and the exposure of the liver to free fatty acids. In conclusion, we propose that the suggested major effect of fat distribution on male/female
210 differences in serum lipids and lipoproteins seems consistent for serum triglycerides. The reduction in gender differences in serum total cholesterol and HDL-cholesterol is clearly present but either the findings are not consistent in men and women or not consistent among centers. Acknowledgements
The authors thank all participants for their cooperation and the laboratory personnel of the Department of Human Nutrition, Wageningen University, for expert technical assistance. This study is part of the concerted action project on Nutrition and Health (EURO-NUT) of the Comae-Epidemiology Group within the Medical Research Council of the European Community. The guidance and support of the Project Management Group of EURO-NUT is gratefully acknowledged. Dr. Jacob C. Seidell is a research fellow of the Royal Netherlands Academy of Science. References Larsson, B., Svardsudd, K., WeIin, L., Wilhelmsen, L., Bjijrntorp, P. and Tibblin, G., Abdominal adipose tissue distribution, obesity, and risk of cardiovascular disease and death: a 13 year follow-up of participants in the study of men born in 1913, Br. Med. J., 288 (1984) 1401. Ohlsson, L.O., Larsson, B., Svlrdsudd, K., Welin, L., Eriksson, H., Wilhelmsen, L., Bjomtorp, P. and Tibblin, G., The influence of body fat distribution on the incidence of diabetes mellitus: 13.5 years of follow-up of the participants in the study of men born in 1913, Diabetes, 34 (1985) 1055. Welin, L., Svardsudd, K., Wilhelmsen, L., Larsson, B. and Tibblin, G., Family history and other risk factors for stroke. The study of men born in 1913, N. EngI. J. Med., 317 (1987) 521. Bjomtorp, P., The associations between obesity, adipose tissue distribution and disease, Acta Med. Stand., Suppl.. 723 (1988) 121. Seidell, J.C., Cigolini, M., Charzewska, J., Ellsinger, B.-M., Dibiase, G., Bjomtorp, P., Hautvast, J.G.A.J., Contaldo, F., Szostak ,V. and Scuro, L.A., Indicators of fat distribution, serum lipids, and blood pressure in European women born in 1948 - the European fat distribution study, Am. J. Epidemiol., 30 (1989) 53. Seidell, J.C., Cigolini, M., Deslypere, J.-P., Charzewska, J., Ellsinger, B.-M. and Cruz, A., Body fat distribution in relation to serum lipids and blood pressure in 38-year old men - the European fat distribution study, Atherosclerosis 86 (1991) 251-260.
7 Freedman, D.S., Jacobsen, S.J., Barboriak, J.J., Sobocinski, K.A., Anderson, A.J., Kissebah, A.H., Sasse, E.A. and Gruchow, H.W., Body fat distribution and male/female differences in lipids and lipoproteins, Circulation, 81 (1990) 1498. 8 Ostlund, R.E., Staten, M., Kohrt, W.M., Schultz, J. and Malley, M., The ratio of waist-to-hip circumference, plasma insulin level, and glucose tolerance as independent predictors of the HDL, cholesterol level in older adults, N. EngI. J. Med., 332 (1990) 229. 9 WHO Monica Project, Risk factors,. Int. J. Epidemiol., 18 (1989) 546. 10 Roschlau, P., Bemt, E. and Gruber, W., Enzymatische Bestimmung des Gesamt-Cholesterins im Serum, Z. Klin. Chem. Khn. B&hem., 12 (1974) 403. 11 Wamick, G.R., Benderson, J. and Albers, J.J., Dextran sulfate Mg2+ precipitation procedure for quantification of high-density-lipoprotein cholesterol, Clin. Chem., 28 (1982) 1379. 12 Sullivan, D.R., Kruijswijk, K., West, C.E., Kohlmeier, M. and Katan, M.B., Determination of serum triglycerides by an accurate enzymatic method not affected by free glycerol, Clin. Chem., 31 (1985) 1227. 13 Wamick, G.R. and Clapshaw, P., Availability of plasma with target values for certain lipids, Clin. Chem., 33 (1987) 2323. 14 Friedewald, W.T., Levy, RI. and Frederickson, D.S., Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge, Clin. Chem., 18 (1972) 499. 15 Krotkiewski, M., Bjomtorp, P., Sjostriim, L. and Smith, U., Impact of obesity on metabolism in men and women, J. Clin. Invest., 71 (1983) 1150. 16 Ashwell, M., Cole, T.J. and Dixon, A.K., Obesity: new insight into the anthropometric classification of fat distribution shown by computed tomography, Br. Med. J., 290 (1985) 1692. 17 Seidell, J.C., Oosterlee, A., Thijssen, M.A.O., Burema, J., Deurenberg, P. and Hautvast, J.G.A.J., Assessment of intraabdominal and subcutaneous abdominal fat - relation between anthropometry and computed tomography, Am. J. Clin. Nutr., 45 (1987) 7. 18 Bjdmtorp, P., “Portal” adipose tissue as a generator of risk factors for cardiovascular disease and diabetes, Arteriosclerosis, 10 (1990) 493. 19 Seidell, J.C., Cigolini, M., Charzewska, J., Ellsinger, B.-M., Dibiase, G., Bjiirntorp, P., Hautvast, J.G.A.J., ContaIdo, F., Szostak, V. and Scuro, L.A. Androgenicity in relation to body fat distribution and metabolism in 38-year old women - the European fat distribution study, J. Clin. Epidemiol., 43 (1990) 21. 20 Nestler, J.E., Clore, J.N. and Blackard, W.G., The central role of obesity (hyperinsulinemia) in the pathogenesis of the polycystic ovary syndrome, Am. J. Obstet. Gynecol., 16 (1989) 1095. 21 Despres, J.-P., Mootjani, S., Lupien, P.J., Tremblay, A., Nadeau, A. and Bouchard, C., Regional distribution of body fat, plasma lipoproteins, and cardiovascular disease, Arteriosclerosis, 10 (1990) 497.