Prediction models for insulin resistance in the polycystic ovary syndrome

G. Gennarelli1,5, J. Holte1, L. Berglund2, C. Berne3, M. Massobrio4 and H. Lithell2

1 Departments of Obstetrics and Gynaecology, 2 Geriatrics, 3 Internal Medicine, Akademiska Hospital, Uppsala University, Uppsala, Sweden and 4 Department of Obstetrics and Gynecology S. Anna Hospital, Torino University, Torino, Italy


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Women with the polycystic ovary syndrome (PCOS) have a high prevalence of insulin resistance, with consequent increased risk of metabolic diseases later in life. An early metabolic screening would therefore be of clinical relevance. By using stepwise regression analysis on several variables obtained in 72 women with PCOS, we constructed simple and reliable mathematical models predicting insulin sensitivity, as measured by the euglycaemic hyperinsulinaemic clamp. The normal ranges of insulin sensitivity were calculated from 81 non-hirsute, normally menstruating women with normal ovaries, and similar body mass index (BMI) and age as the women with PCOS. Measured variables included BMI, waist and hip circumferences, truncal–abominal skin folds, circulating concentrations of gonadotrophins, androgens, sex hormone-binding globulin (SHBG), triglycerides, total cholesterol and cholesterol subfractions, fasting insulin, C-peptide and free fatty acids. The three best prediction models included waist circumference, together with insulin (model I: R2 = 0.77), serum triglycerides (model II: R2 = 0.65), and the subscapularis skin fold (model III: R2 = 0.64). Using reference limits for insulin sensitivity obtained in the 81 normal pre-menopausal women, the models identify insulin resistant women with PCOS. These simple and inexpensive models are potentially useful in clinical practice as an early screening in women with PCOS.

Key words: euglycaemic hyperinsulinaemic clamp/insulin resistance/PCOS/screening


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Insulin resistance, defined as subnormal insulin action on glucose uptake, represents a unifying pathogenetic link in a cluster of abnormalities grouped under the term of the insulin resistance syndrome (Reaven, 1995Go), strongly associated with the development of cardiovascular disease (De Fronzo and Ferranini, 1991Go; Moller and Flier, 1991Go; Bjorntorp, 1993Go). Insulin resistance is found in most overweight women with the polycystic ovary syndrome (PCOS) to a greater extent than can be expected from obesity per se (Dunaif et al., 1989Go; Holte et al., 1994aGo; Ciampelli and Lanzone, 1998Go; Holte, 1998Go). Women with PCOS also show an increased prevalence of abnormal ß-cell function (Dunaif et al., 1987Go; Holte et al., 1994aGo, 1995Go; Ciampelli et al., 1997Go), dyslipidaemia (Conway et al., 1992Go; Holte et al., 1994bGo), truncal–abdominal fatness (Holte et al., 1994aGo, 1995Go; Bouchard, 1997Go), blood pressure abnormalities (Holte et al., 1996Go), and progression to frank diabetes type II (Dunaif et al., 1987Go; Holte et al., 1994aGo; Legro et al., 1999Go; Ehrmann et al., 1999Go). A risk factor model for cardiovascular disease shows women with previously proven PCOS to be at great risk later in life (Dahlgren et al., 1992aGo,bGo).

Since most women with PCOS come to clinical attention when their glucose tolerance is still normal and the cardiovascular damage is presumably at an early stage, screening for insulin resistance would be crucial in order to identify those cases at greater risk, allowing an appropriate medical intervention.

The hyperinsulinaemic euglycaemic clamp (De Fronzo et al., 1979Go) is considered to be the gold standard for evaluation of insulin sensitivity. In our department, we have been using this technique in recent years to investigate women with PCOS. However, the procedure is expensive, invasive, and time-consuming. In the present study, we aimed at defining prediction models for insulin resistance developed from clinical, anthropometric, hormonal and metabolic variables.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Subjects
Seventy-two women with PCOS, consecutively recruited among patients seeking medical advice for menstrual irregularities with or without hirsutism and/or infertility, at the Department of Obstetrics and Gynaecology of Uppsala University Hospital, formed the basis for the prediction models. The diagnosis of PCOS was based on the ultrasonographic evidence of polycystic ovaries, as previously defined (Adams et al., 1986Go), in association with a history of menstrual irregularities, indicating chronic anovulation. The ultrasound examination was performed transvaginally with an Acuson machine (Acuson 128/10®, 5 MHz; Acuson Corp., Mt View, CA, USA). The ovarian volume was also calculated according to the formula {pi}/6xD1xD2xD3, where D1, D2, and D3 are the maximum ovarian diameters. Hirsutism was present in 37 (51%) women with PCOS, who had a score >=7 according to a modified version of the Ferriman and Gallwey protocol (Hatch et al., 1981Go).

Eighty-one non-hirsute, normally menstruating women (mean age 30, range 19–44 years) with normal ovaries according to ultrasound, covering a similarly wide range of body mass index (BMI kg/m2: range 17.3–40.9), formed the reference group for the calculation of the normal percentiles of insulin sensitivity.

All the women were in good physical condition, non-diabetic, with normal levels of prolactin, and did not suffer from any other metabolic disease. None of the subjects had been taking any drug known to affect carbohydrate metabolism, or any hormonal substance for at least 3 months prior to the metabolic and endocrine investigations. Seven women with PCOS were glucose intolerant, according to the results of a frequently sampling intra-venous glucose tolerance test (IVGTT) (k value <1) (Holte et al., 1994aGo).

A group of 20 elderly men, participating in a survey at the Department of Geriatrics of Uppsala University, were investigated twice 29 ± 7 days apart with the clamp technique in order to correct for within-subject variation in insulin sensitivity (see below).

Informed consent was obtained from all the women, and the study received the approval of the Human Ethics Committee of the Medical Faculty, Uppsala University.

Anthropometric variables
BMI was calculated as weight (kg) divided by squared height (m2). Measurements of the waist and hip circumferences and of the subscapularis, suprailiaca and umbilicalis skin folds were performed in duplicate as previously described in detail (Holte et al., 1994aGo).

Euglycaemic hyperinsulinaemic clamp and laboratory investigations
Insulin sensitivity was measured by the euglycaemic hyperinsulinaemic clamp as previously described (Pollare et al., 1990Go). Insulin (Actrapid Human®; Novo, Copenhagen, Denmark) was infused at a rate of 56 mU/(minxm2 body surface area). The amount of glucose infused to maintain the target plasma glucose concentration (5.1 mmol/l) during the second hour of the test was defined as glucose disposal (M: mg glucosexkg–1xmin–1). Adjusting M for the steady-state insulin concentration defined the insulin sensitivity index [M/I: mg glucosexkg–1xmin–1x (100 mU/l)–1x100]. The within-subject variability of insulin sensitivity, calculated from the reproducibility study on 20 elderly men, was 0.61 units (13.9% of the mean).

The methods for the hormone and lipid analyses have been described previously (Holte et al., 1994aGo,bGo).

The coefficients of variation for the variables chosen in the models (see below) were 15.7 and 14.8%, for plasma insulin and serum triglycerides respectively (Berglund and Lithell, 1996Go), and 8.5 and 2.9% for the subscapularis skin fold and the waist circumference respectively.

Statistics
All variables were examined for normality of distribution with Kolmogorov–Smirnov goodness-of-fit test and, where necessary, log transformation was performed. A forward stepwise-regression analysis was performed on a set of variables to find the best models for insulin sensitivity. The variables considered for the analyses were selected on the basis of a significant simple correlation coefficient with log M/I >=0.5. Criteria for choosing the models were: ß coefficients significantly (P < 0.05) different from zero and high multiple correlation coefficients. The within-subject variation in insulin sensitivity will cause an under-estimation of the multiple correlations and an over-estimation of the prediction errors. In order to remove those biases, according to a published method (Rosner and Willett, 1988Go), we used the data from the reproducibility study for insulin sensitivity.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The clinical, anthropometric, hormonal, and metabolic variables in the women with PCOS, together with the respective simple correlation coefficients to the insulin sensitivity index, are given in Table IGo. The highest correlation coefficients were observed for fasting insulin and for measures of truncal abdominal adiposity. The stepwise regression analysis resulted in several models. The three best models are reported here (Table IIGo, Figures 1–3GoGoGo). All models have the waist circumference as one of the predictors, together with the fasting plasma insulin (model I), serum triglycerides (model II), or subscapularis skin fold (model III). Models with more than two predictors gave only slight improvement of prediction ability.


View this table:
[in this window]
[in a new window]
 
Table I. Mean, range, and simple correlation coefficients with insulin sensitivity index M/I (log-transformed) of the clinical, anthropometric, hormonal, and metabolic variables in 72 women with PCOS. Those variables with correlation coefficients with M/I >= 0.5 are printed in bold type
 

View this table:
[in this window]
[in a new window]
 
Table II. Significance level for single variables and multiple correlation coefficients after correcting according to Rosner and Willett (1988), for the three best models
 


View larger version (29K):
[in this window]
[in a new window]
 
Figure 1. Nomogram indicating the relationship between waist girth and the insulin sensitivity index (M/I) obtained during a euglycaemic hyperinsulinaemic clamp, for different values of fasting insulin in women with PCOS. Left axis: predicted values of M/I. Right axis: prediction error at that level of M/I.

 


View larger version (27K):
[in this window]
[in a new window]
 
Figure 2. Nomogram indicating the relationship between waist girth and the insulin sensitivity index (M/I) obtained during a euglycaemic hyperinsulinaemic clamp, for different values of serum triglycerides in women with PCOS. Left axis: predicted values of M/I. Right axis: prediction error at that level of M/I.

 


View larger version (28K):
[in this window]
[in a new window]
 
Figure 3. Nomogram indicating the relationship between waist girth and the insulin sensitivity index (M/I) obtained during a euglycaemic hyperinsulinaemic clamp, for different values of subscapularis skinfold in women with PCOS. Left axis: predicted values of M/I. Right axis: prediction error at that level of M/I.

 
The final results are illustrated in nomograms (Figures 1–3GoGoGo) derived from the estimated equations for the prediction models. In the nomograms, specified values on the two predictors will give a predicted value of M/I together with a value of the prediction error. With a probability of 80%, the true value of M/I will lie within predicted value ± error. The nomograms also report the 5th percentile of the distribution of the insulin sensitivity index for the reference group of normal women (4.9 units). Taking into account the prediction errors, in model I predicted values of M/I less than 3.8 would indicate insulin resistance, whereas the M/I threshold for models II and III would be 3.5 (Figures 1–3GoGoGo).


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
In the present study we constructed prediction models for insulin sensitivity in women with PCOS, based on results obtained by the euglycaemic clamp in women with PCOS covering a wide range of BMI and controls with normal ovaries and similar BMI. After regression analyses, three statistical models based on simple measurements were obtained. These included the waist circumference in combination with either fasting plasma insulin (model I), or fasting serum triglycerides (model II), or the subscapularis skin fold (model III).

Waist circumference was an independent predictor of insulin resistance in all three models. This simple measure has proven to be a reliable predictor of visceral fat and related cardiovascular risk in both sexes (Hartz et al., 1984Go; Pouliot et al., 1994Go; Han et al., 1995Go), and it is correlated with indices of insulin resistance, especially in women (Bjorntorp, 1990Go; Pouliot et al., 1994Go). The presence of such a measure of visceral fat in all models is in line with the well-documented relationship between abdominal fat and insulin resistance in women with PCOS (Holte et al., 1994aGo, 1995Go; Bouchard, 1997Go).

In the most efficient model, fasting insulin was an independent predictor of insulin sensitivity. Indeed, fasting insulin showed the highest degree of simple correlation (inverse) to the insulin sensitivity index (Table IGo). This was expected and it is in line with results obtained in different non-diabetic populations (Bergman et al., 1985Go; Matthews et al., 1985Go; Ferranini et al., 1997Go). However, the fact that fasting insulin and waist girth had similar and independent statistical impacts in model I suggests that fasting insulin provides only a partial measurement of peripheral insulin resistance.

A recent study reported that the glucose to insulin ratio measured in the fasting state may provide a useful prediction measure of insulin resistance in women with PCOS (Legro et al., 1998Go). In that study measurements of body fat distribution were not included. In spite of a significant simple correlation with the insulin sensitivity index in our population (Table IGo), the glucose to insulin ratio was not better than fasting insulin alone, and the index was not selected in any of the best models, suggesting that a measurement of truncal fat increases the power of the model. In the above cited study, insulin sensitivity was measured with the frequently sampled i.v. glucose tolerance test (`FSIGT'; Legro et al., 1998). Although the two techniques of investigating insulin sensitivity usually give similar results, the clamp technique is regarded as the gold standard. A recent thorough study on subjects with varying degrees of glucose tolerance failed to find any close correlation between the glucose to insulin ratio and the insulin sensitivity index obtained with the euglycaemic clamp (Matsuda and DeFronzo, 1999Go).

In model II, fasting serum concentration of triglycerides was chosen as the second variable. Increased triglyceride concentrations are found in states of insulin resistance, most probably as a result of lower activation of the lipoprotein lipase and impaired triglyceride clearance (Frayn, 1993Go). In analogy with the findings for body fat distribution, the association between triglycerides and insulin resistance is common in states of insulin resistance (Laakso et al., 1990Go; Berglund and Lithell, 1996Go), suggesting that the metabolic derangements encountered in PCOS are not unique, but generally conform with the well established insulin resistance syndrome (De Fronzo and Ferranini, 1991Go; Reaven, 1995Go; Holte, 1996Go).

The subscapularis skin fold formed the basis for prediction model III, along with the waist girth. Interestingly, the two anthropometric measures predicted independently the degree of insulin resistance. Most probably, the waist girth and the subscapularis skin fold measure two different types of fat, the predominantly visceral and the subcutaneous truncal fat respectively (Bouchard et al., 1993Go), both types being independently associated with insulin resistance (Ross et al., 1996Go). Recent studies show that in healthy individuals subcutaneous truncal–abdominal fat is highly correlated with the level of insulin resistance even more so than is intraperitoneal fat (Abate et al., 1995Go), even when the insulin resistance is measured by DXA (dual X-ray absorptiometry) (Marcus et al., 1999Go). In addition, it has been suggested that the thickness of subscapularis skin fold could help to identify women at risk of non-insulin dependent diabetes mellitus (NIDDM; Peiris et al., 1989). The coefficient of variation for waist girth was 2.9%, whereas it was somehow higher for the subscapularis skin fold (8.5%). The latter value is not far from that of previous investigations on different populations (Durnin and Womersley, 1974Go; Peiris et al., 1989Go). It should be stressed that despite the obvious variability in anthropometric measurements, the predictive power of model III was strong, making it suitable for clinical use.

Interestingly, no measures of androgen excess qualified in the prediction models, in spite of fairly high simple correlation coefficients. These findings suggest that such associations are mainly indirect, and the results are in line with previous reports of similar degrees of metabolic derangements in women with normal androgen concentrations as in those with increased concentrations (Norman et al., 1995Go).

In total, the results support the strong relationship between decreased insulin sensitivity, truncal–abdominal fatness, and dyslipidaemia in women with PCOS, whereas measures of androgen excess are lost when tested in multiple regression analyses. In conclusion, the present study allowed the construction of three simple and inexpensive models with high predictive power, which thus would be potentially useful in clinical practice on an outpatient basis.


    Notes
 
5 To whom correspondence should be addressed at: Department of Obstetrics and Gynaecology, S. Anna Hospital, Torino University, via Ventimiglia 3, 10126 Torino, Italy.E-mail: gennarelligl{at}libero.it Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Abate, N., Garg, A., Peshock, R.M. et al. (1995) Relationships of generalized and regional adiposity to insulin sensitivity in men. J. Clin. Invest., 96, 88–98.[ISI][Medline]

Adams, J., Polson, D.W. and Franks, S. (1986) Prevalence of polycystic ovaries in women with anovulation and idiopathic hirsutism. Br. Med. J., 293, 355–359.[ISI][Medline]

Berglund, L. and Lithell, H. (1996) Prediction models for insulin resistance. Blood Pressure, 5, 274–277.[Medline]

Bergman, R.N., Finegood, D.T. and Ader, M. (1985) Assessment of insulin sensitivity in vivo. Endocr. Rev., 6, 45–86.[ISI][Medline]

Bjorntorp, P. (1990) Abdominal obesity and risk. Clin. Exp. Hypertens., 12, 783–794.

Bjorntorp, P. (1993) Visceral obesity. A `civilization syndrome'. Obes. Res., 1, 206–222.

Bouchard, C., Després, J.P. and Mauriége, P. (1993) Genetic and nongenetic determinants of regional fat distribution. Endocr. Rev., 14, 72–93.[Abstract]

Bouchard, C. (1997) Genetic determinants of regional fat distribution. Hum. Reprod., 12 (Suppl. 1), 1–5.[Free Full Text]

Ciampelli, M. and Lanzone, A. (1998) Insulin and polycystic ovary syndrome: a new look at an old subject. Gynecol. Endocrinol., 12, 277–292.[ISI][Medline]

Ciampelli, M., Fulghesu, A.M., Cucinelli, F. et al. (1997) Heterogeneity in beta-cell activity, hepatic insulin clearance and peripheral insulin sensitivity in women with polycystic ovary syndrome. Hum. Reprod., 12, 1897–1901.[Abstract]

Conway, G.S., Agrawal, R., Betteridge, D.J. and Jacobs, H.S. (1992) Risk factors for coronary artery disease in lean and obese women with the polycystic ovary syndrome. Clin. Endocrinol., 37, 119–125.[ISI][Medline]

Dahlgren, E., Janson, P.O., Johansson, S. et al. (1992a) Polycystic ovary syndrome and risk for myocardial infarction. Evaluated from a risk factor model based on a prospective population study of women. Acta Obstet. Gynecol. Scand., 71, 599–604.[ISI][Medline]

Dahlgren, E., Johansson, S., Lindstedt, G. et al. (1992b) Women with polycystic ovary syndrome wedge resected in 1956 to 1965: a long-term follow-up focusing on natural history and circulating hormones. Fertil. Steril., 57, 505–513.[ISI][Medline]

De Fronzo, R.A. and Ferranini, E. (1991) Insulin resistance: a multifaced syndrome responsible for NIDDM, obesity, hypertension, dyslipidemia, and atherosclerotic cardiovascular disease. Diabetes Care, 14, 173–194.[Abstract]

De Fronzo, R.A., Tobin, J.D. and Andres, R. (1979) Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am. J. Physiol., 237, E214–E223.[Abstract/Free Full Text]

Dunaif, A., Graf, M., Mandeli, J. et al. (1987) Characterization of groups of hyperandrogenic women with acanthosis nigricans, impaired glucose tolerance, and/or hyperinsulinemia. J. Clin. Endocrinol. Metab., 65, 499–507.[Abstract]

Dunaif, A., Segal, K.R., Futterweit, W. and Dobrjansky, A. (1989) Profound peripheral insulin resistance, independent of obesity, in polycystic ovary syndrome. Diabetes, 38, 1165–1174.[Abstract]

Durnin, J.V. and Womersley, J. (1974) Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years. Br. J. Nutr., 32, 77–97.[ISI][Medline]

Ehrmann, D.A., Barnes, R.B., Rosenfield, R.L. et al. (1999) Prevalence of impaired glucose tolerance and diabetes in women with polycystic ovary syndrome. Diabetes Care, 22, 141–149.[Abstract]

Ferranini, E., Natali, A., Bell, P. et al. (1997) Insulin resistance and hypersecretion in obesity. J. Clin. Invest., 100, 1166–1173.[Abstract/Free Full Text]

Frayn, K.N. (1993) Insulin resistance and lipid metabolism. Curr. Opin. Lipidol., 4, 197–204.

Han, T.S., McNeill, G., Baras, P. and Foster, M.A. (1995) Waist circumference predicts intra-abdominal fat better than waist:hip ratio in women. Proc. Nutr. Soc., 54, 152A.

Hartz, A.J., Rupley, D.C. and Rimm, A.A. (1984) The association of girth measurements with disease in 32 856 women. Am. J. Epidemiol., 119, 71–80.[Abstract]

Hatch, R., Rosenfield, R.L., Kim, M.H. and Tredway, D. (1981) Hirsutism: implications, etiology, and management. Am. J. Obstet. Gynecol., 140, 815–830.[ISI][Medline]

Holte, J. (1996) Disturbances in insulin secretion and sensitivity in women with the polycystic ovary syndrome. Baillière's Clin. Endocrinol. Metab., 10, 221–247.[ISI][Medline]

Holte, J. (1998) Polycystic ovary syndrome and insulin resistance: thrifty genes struggling with over-feeding and sedentary life style? J. Endocrinol. Invest., 21, 589–601.[ISI][Medline]

Holte, J., Bergh, T., Berne, C. et al. (1994a) Enhanced insulin response to glucose in relation to insulin resistance in women with polycystic ovary syndrome and normal glucose tolerance. J. Clin. Endocrinol. Metab., 78, 1052–1058.[Abstract]

Holte, J., Bergh, T., Berne, C. and Lithell, H. (1994b) Serum lipoprotein lipid profile in women with the polycystic ovary syndrome: relation to anthropometric, endocrine and metabolic variables. Clin. Endocrinol., 41, 463–471.[ISI][Medline]

Holte, J., Bergh, T., Berne, C. et al. (1995) Restored insulin sensitivity but persistently increased early insulin secretion after weight loss in obese women with polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 80, 2586–2593.[Abstract]

Holte, J., Gennarelli, G., Berne, C. et al. (1996) Elevated ambulatory day-time blood pressure in women with polycystic ovary syndrome: a sign of a pre-hypertensive state? Hum. Reprod., 11, 23–28.[Abstract]

Laakso, M., Sarlund, H. and Mykkanen, L. (1990) Insulin resistance is associated with lipid and lipoprotein abnormalities in subjects with varying degrees of glucose tolerance. Arteriosclerosis, 10, 223–231.[Abstract]

Legro, R.S., Finegood, D. and Dunaif, A. (1998) A fasting glucose to insulin ratio is a useful measure of insulin sensitivity in women with polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 83, 2694–2698.[Abstract/Free Full Text]

Legro, R.S., Kunselman, A.R., Dodson, W.C. and Dunaif, A. (1999) Prevalence and predictors of risk for type 2 diabetes mellitus and impaired glucose tolerance in polycystic ovary syndrome: a prospective, controlled study in 254 affected women. J. Clin. Endocrinol. Metab., 84, 165–169.[Abstract/Free Full Text]

Marcus, M.A., Murphy, L., Pi-Sunyer, F.X. and Albu, J.B. (1999) Insulin sensitivity and serum triglyceride level in obese white and black women: relationship to visceral and truncal subcutaneous fat. Metabolism, 48, 194–199.[ISI][Medline]

Matsuda, M. and DeFronzo, R.A. (1999) Insulin sensitivity indices obtained from oral glucose tolerance testing. Diabetes Care, 22, 1462–1470.[Abstract]

Matthews, D.R., Hosker, J.P., Rudenski, A.S. et al. (1985) Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia, 28, 412–419.[ISI][Medline]

Moller, D.E. and Flier, J.S. (1991) Insulin resistance – mechanisms, syndromes, and implications (see comments). N. Engl. J. Med., 325, 938–948.[ISI][Medline]

Norman, R.J., Hague, W.M., Master, S.C. and Wang, X.J. (1995) Subjects with polycystic ovaries without hyperandrogenaemia exhibit similar disturbances in insulin and lipid profiles as those with polycystic ovary syndrome. Hum. Reprod., 10, 2258–2261.[Abstract]

Peiris, A.N., Aiman, E.J., Drucker, W.D. and Kissebah, A.H. (1989) The relative contributions of hepatic and peripheral tissues to insulin resistance in hyperandrogenic women. J. Clin. Endocrinol. Metab., 68, 715–720.[Abstract]

Pollare, T., Lithell, H. and Berne, C. (1990) Insulin resistance is a characteristic feature of primary hypertension independent of obesity. Metabolism, 39, 167–174.[ISI][Medline]

Pouliot, M.C., Despres, J.P., Lemieux, S. et al. (1994) Waist circumference and abdominal sagittal diameter: best simple anthropometric indexes of abdominal visceral adipose tissue accumulation and related cardiovascular risk in men and women. Am. J. Cardiol., 73, 460–468.[ISI][Medline]

Reaven, G.M. (1995) Pathophysiology of insulin resistance in human disease. Physiol. Rev., 75, 473–486.[Abstract/Free Full Text]

Rosner, B. and Willett, C. (1988) Interval estimates for correlation coefficients for within-person variation: implication for study design and hypothesis testing. Am. J. Epidemiol., 127, 377–386.[Abstract]

Ross, R., Fortier, L. and Hudson, R. (1996) Separate associations between visceral and subcutaneous adipose tissue distribution, insulin and glucose levels in obese women. Diabetes Care, 19, 1404–1411.[Abstract]

Submitted on April 26, 2000; accepted on July 3, 2000.