The central issue? Visceral fat mass is a good marker of insulin resistance and metabolic disturbance in women with polycystic ovary syndrome

Authors


Dr J Lord, Department of Obstetrics and Gynaecology, Royal Cornwall Hospital, Truro, Cornwall TR1 3LJ, UK. Email jonathan.lord@pms.ac.uk

Abstract

Objective  To establish whether visceral fat mass is the most significant variable correlating with insulin resistance and other metabolic parameters in women with polycystic ovary syndrome (PCOS).

Design  Prospective cross-sectional trial.

Setting  Reproductive medicine clinic.

Population  Forty women with anovulatory PCOS.

Methods  Measurements were taken at recruitment, and analysis was performed to define correlations between the outcome measures and the explanatory variables.

Main outcome measures  Visceral and subcutaneous fat by computed tomography scan, insulin resistance, anthropometric measures, markers of the metabolic syndrome and androgens.

Results  Strong linear correlation of visceral fat to insulin resistance (r= 0.68, P < 0.001) was observed. There were also statistically significant correlations with fasting insulin (r= 0.73, P < 0.001), homeostasis model assessment β-cell function (r= 0.50, P= 0.007), triglycerides (r= 0.45, P= 0.003), high-density lipoprotein cholesterol (r=−0.42, P= 0.007), urate (r= 0.47, P= 0.002), Sex hormone binding globulin (r=−0.39, P= 0.01) and luteinising hormone (r=−0.32, P= 0.02). There were no significant correlations of testosterone with fat distribution or metabolic parameters. Insulin resistance showed closest correlation to visceral fat mass (r= 0.68, P < 0.001), then to waist circumference (r= 0.62, P < 0.001), with the weakest correlation being waist:hip ratio (r= 0.36, P= 0.01). The best regression model for predicting insulin resistance is with visceral fat mass and triglycerides as the explanatory variables (r= 0.72, P < 0.001).

Conclusions  Visceral fat is the most significant variable correlating with metabolic dysfunction in women with PCOS. Our data support the hypothesis that visceral fat either causes insulin resistance or is a very early effect of it. It also implies that reducing visceral fat should reduce insulin resistance which may account for the observations that exercise and weight loss appear to be more effective interventions than pharmacological treatments. The best anthropometric measure of insulin resistance is waist circumference.

Introduction

In the general population, visceral fat appears to be responsible for the adverse metabolic effects of obesity, probably as a result of its association with insulin resistance.1 Whether this relationship is similarly close in women with polycystic ovary syndrome (PCOS), where authorities hypothesise that there might be a primary genetic defect causing insulin resistance2 or increased androgen secretion in response to insulin,3 has not been well investigated. We have previously outlined why fat distribution may be a very early effect, or possibly even a cause, of PCOS.4 However, little is known about the relationships between different fat compartments and metabolic indices in women with PCOS (Figure 1).

Figure 1.

CT scan of abdomen at level of L2 showing visceral (V) and subcutaneous (SC) fat compartments.

Our primary objective was to establish whether visceral fat mass was the most significant variable correlating with insulin resistance in women with PCOS. As secondary objectives, we aimed to explore the relationships between fat distribution and other metabolic and endocrinological parameters in women with PCOS. We wanted to investigate whether visceral fat mass had a closer correlation with metabolic parameters than did subcutaneous fat or the visceral:subcutaneous fat ratio and whether these relationships could be detected using simple anthropometric measurements. We were also interested to examine whether the data were compatible with a state of primary insulin resistance in women with PCOS and to speculate whether the data supported the hypothesis that visceral fat could be a causative factor.

These issues are important clinically because if visceral fat mass is a significant variable in women with PCOS, then interventions that reduce it should be beneficial. It may also provide an explanation as to why lifestyle regimes can be more effective treatments than pharmacological ones.5

Methods

Participants

Women with anovulation and PCOS were recruited from those attending outpatients at the South West Centre for Reproductive Medicine, Derriford Hospital, Plymouth, UK. Anovulation was defined as oligomenorrhoea (<6 periods in preceding 12 months) or a luteal-phase progesterone level of <20 nmol/l. PCOS was defined as anovulation and a raised free androgen index (FAI) >5.0. Participants with diabetes mellitus, thyroid disease, raised prolactin and late-onset adrenal hyperplasia (i.e. raised 17α-hydroxyprogesterone) were excluded. Other exclusion criteria were the use of ovulation-inducing agents or drugs that could affect insulin metabolism within the previous 2 months, age outside the range of 18–40 years and pregnancy (excluded by high sensitivity urine β human chorionic gonadotrophin). Measures of weight and fat distribution did not form part of the recruitment criteria. All participants had a transvaginal ultrasound scan on recruitment. Measurements were taken on recruitment. Participants were recruited prospectively and represented a cohort that proceeded into a randomised controlled trial, with these data representing their recruitment characteristics.

Ethical approval was obtained from the local research ethics committee, and informed written consent was obtained from each participant.

Protocol

Measurements—fat distribution

Visceral and subcutaneous fat mass were measured by areal planimetry using a single 5-mm collimation slice through the midpoint of L2 at full inspiration, with the radiation dose reduced as much as possible. This was performed using a Siemens Somatom Plus 4 (Siemens PLC, Bracknell, Berkshire, UK), with a preliminary limited topogram of the lower abdomen to identify the vertebra. Postprocessing was performed on a Siemens Virtuoso Workstation (Siemens PLC) and all results were reported by one of two radiologists (B.F. and R.T.), with inter- and intra-observer variations recorded.

Anthropometric measurements were taken using standard techniques.6 Circumferences were measured to within 5 mm using a tape measure in the standing position. Waist circumference was taken at the midpoint between the lowest rib margin and the iliac crest at the end of normal expiration. Hip circumference was measured at the widest level of the greater trochanters. The body mass index (BMI) was calculated using the standard formula (kg/m2) using one set of scales and height measure which were regularly checked and calibrated by the hospital’s medical physics department. Inter- and intra-observer variations were recorded.

Measurements—serum

Serum was taken after an overnight fast of at least 10 hours on the same day as the measurements of fat distribution. Serum was centrifuged for 5 minutes and frozen to −80°C within 30 minutes of being taken. It was analysed in batches by the hospital’s clinical chemistry department using their standard operating procedures within 30 minutes of being thawed.

  • 1Insulin (Immulite®; DPC, Los Angeles, CA, USA): Insulin resistance was estimated by homeostasis model assessment (HOMA), a computerised model of the physiological loop which controls the concentration of blood glucose.7–9 It allows an estimation of insulin resistance from the fasting levels of insulin and blood glucose. The insulin assay cross reacts at <1% with proinsulin.
  • 2Glucose: COBAS® Integra glucose HK; Roche Diagnostics, Basel, Switzerland.
  • 3Testosterone: ACS Centaur®; Chiron Diagnostics Corporation, East Walpole, MA, USA.
  • 4Sex hormone binding globulin (SHBG): Immulite 2000®; DPC.
  • 5FAI: serum testosterone/SHBG × 100
  • 6Cholesterol (COBAS® Integra cholesterol; Roche Diagnostics), low-density lipoprotein (LDL) cholesterol and high-density lipoprotein (HDL) cholesterol (COBAS® Integra HDL-cholesterol direct; Roche Diagnostics).
  • 7Triglycerides: COBAS® Integra triglycerides; Roche Diagnostics.
  • 8Urate: COBAS® Integra uric acid; Roche Diagnostics.
  • 9Glycated haemoglobin (HbA1C): Biomen 8160; Menarini Diagnostics, Florence, Italy.
  • 10Progesterone (ACS Centaur®; Chiron Diagnostics): It was taken at recruitment in those with oligomenorrhoea in order to exclude ovulation or the possibility of the participant being in the luteal phase.

Statistical analysis

SPSS version 11.5 (SPSS Inc., Chicago, IL, USA) was used for analysis. Parameters were tested for normality. Pearson’s correlation coefficient was used for parametric data and Spearman’s ρ for nonparametric data (both two tailed). The primary outcome measure was visceral fat mass. Secondary dependent variables were subcutaneous fat mass, insulin resistance and testosterone. Explanatory variables are described in detail below and include anthropometric measurements and serum indices of the metabolic syndrome. Where linear regression models were used, multiple variables were entered on the basis of biological plausibility.

Trial validity

Intra- and inter-observer variability were checked for visceral fat measurement (r2= 0.97 and 0.99, respectively), subcutaneous fat measurements (r2= 0.99 and 0.99), waist circumference (r2= 0.98 and 0.98), hip circumference (r2= 0.98 and 0.94) and test/re-test variability for insulin (r2= 0.82).

Results

Baseline details

Forty-four participants were recruited, although four were subsequently excluded for not fulfilling the entry criteria at the time of recruitment: two had FAI <5.0 at trial entry and two were ovulating or were in the luteal phase. The cohort’s characteristics are summarised in Table 1. The mean age was 29 years (SD 5.0, range 21–38), mean testosterone level 2.7 nmol/l (SD 0.72, range 1.0–4.9) and mean FAI 10.4 (SD 4.4, range 5.0–21.9). The inclusion criteria did not include measures of weight or fat mass. As would be expected in a population with PCOS, while some participants were lean, 36 (89%) had BMI >25 kg/m2 and 33 (83%) had a waist circumference >88 cm and 23 (58%) >100 cm. The baseline characteristics were mean weight 95 kg (SD 16.4, range 50–127), mean BMI 35 kg/m2 (SD 7.1, range 20.7– 52.9), mean waist circumference 101.2 cm (SD 14.5, range 71–131) and mean waist:hip ratio (WHR) 0.8 (SD 0.07, range 0.7–1.0). All participants had features characteristic of PCOS on ultrasound scan (multiple small follicles, increased stromal echogenecity and increased volume), although this did not form part of the inclusion criteria (mean ovarian volume 11.2 cm3, SD 4.2).

Table 1.  Characteristics of participants
 MeanSD
  1. DHEAS, dehydroepiandrosterone sulphate.

Age, years29.135.02
Testosterone, nmol/l2.670.72
SHBG, nmol/l33.8232.39
FAI10.354.42
DHEAS, μmol/l5.482.87
FSH, iu5.321.84
LH, iu14.3310.67
Fasting insulin, miu/l20.2811.94
Glucose, mmol/l5.110.49
HOMA β-cell function270.59167.13
HOMA insulin resistance4.642.87
Cholesterol, mmol/l5.231.01
Triglycerides, mmol/l1.510.70
HDL cholesterol, mmol/l1.270.25
LDL cholesterol, mmol/l3.281.05
Cholesterol:HDL ratio4.261.12
HbA1C, %5.350.35
Urea, mmol/l4.200.79
Urate, mmol/l0.330.08
Weight, kg95.0016.36
Height, cm165.286.56
BMI, kg/m234.997.12
Waist, cm101.2114.47
Hip, cm120.8812.62
WHR0.840.07
Systolic blood pressure, mmHg127.3414.38
Diastolic blood pressure, mmHg76.398.57
Mean arterial blood pressure, mmHg93.388.80
Subcutaneous fat, mm235554.5715286.05
Visceral fat, mm211201.955029.23
Visceral:subcutaneous fat ratio0.330.14

Visceral fat

Visceral fat showed highly significant correlations with measures of glucose homeostasis (Table 2): insulin resistance (Figure 2) (r= 0.68, P < 0.001), fasting insulin (r= 0.73, P < 0.001) and HOMA β-cell function (r= 0.50, P= 0.001). The correlations for fasting glucose (r= 0.29, P= 0.07) and glycosylated haemoglobin HbA1C did not reach statistical significance (r= 0.24, P= 0.15).

Table 2.  Correlations with visceral fat mass as the dependent variable
VariablerP
  1. DHEAS, dehydroepiandrosterone sulphate; NS, not significant.

Testosterone, nmol/l−0.15NS
SHBG, nmol/l−0.390.01
FAI0.14NS
DHEAS, μmol/l−0.23NS
LH, iu−0.320.05
FSH, iu−0.27NS
Insulin, miu/l0.73<0.001
Glucose, mmol/l0.29NS
HOMA β-cell function0.500.001
HOMA insulin resistance0.68<0.001
Cholesterol, mmol/l−0.04NS
Triglycerides, mmol/l0.450.003
HDL cholesterol, mmol/l−0.420.007
LDL cholesterol, mmol/l−0.05NS
Cholesterol:HDL ratio0.23NS
Urea, mmol/l0.09NS
Urate, mmol/l0.470.002
HbA1C, %0.24NS
Weight, kg0.64<0.001
Height, cm−0.20NS
BMI, kg/m20.625<0.001
Waist, cm0.75<0.001
Hip, cm0.66<0.001
WHR0.460.002
Subcutaneous fat, cm20.57<0.001
Systolic blood pressure, mmHg0.26NS
Diastolic blood pressure, mmHg0.15NS
Mean arterial blood pressure, mmHg0.24NS
Figure 2.

Correlation of insulin resistance with visceral fat mass (r= 0.68, P < 0.001).

There were also statistically significant correlations with markers of the metabolic syndrome: triglycerides (r= 0.45, P= 0.003), HDL cholesterol (r=−0.42, P= 0.007) and urate (r= 0.47, P= 0.002). The correlation for cholesterol:HDL cholesterol ratio did not reach statistical significance (r= 0.23, P= 0.2). Correlations for blood pressure also failed to reach significance (r= 0.26, P= 0.12). There were no apparent correlations for total cholesterol (r=−0.04) or LDL cholesterol (r=−0.05).

SHBG showed a negative correlation with visceral fat (r=−0.39, P= 0.01), although other measures of androgens showed no correlations. Luteinising hormone (LH) showed a significant negative correlation with visceral fat mass (r=−0.32, P= 0.05), while follicle stimulating hormone (FSH) did not reach statistical significance (r=−0.27, P= 0.09).

Other measures of fat mass

Visceral fat showed a strong correlation with subcutaneous fat mass as measured by computed tomography (CT) scan (r= 0.57, P < 0.001). When secondary analysis was performed using subcutaneous fat as the dependent variable, similar correlations as for visceral fat mass were observed, although they were weaker and those for LH and triglycerides lost statistical significance.

The ratio of visceral:subcutaneous fat showed no significant correlations with any variables apart from FSH (r=−0.39, P= 0.01).

Insulin resistance

Secondary analysis using HOMA insulin resistance as the dependent variable showed similar correlations to those seen with visceral fat: glucose (r= 0.35, P= 0.03), SHBG (r=−0.36, P= 0.02), triglycerides (0.34, P= 0.03), HDL cholesterol (r=−0.48, P= 0.002), cholesterol:HDL cholesterol ratio (r= 0.30, P= 0.03) and urate (r= 0.50, P= 0.001). The correlations with LH and FSH were not statistically significant (r=−0.22, P= 0.09 and r=−0.29, P= 0.04, respectively), and neither was HbA1C (r= 0.28, P= 0.09).

Testosterone

Secondary analysis using testosterone as the dependent variable showed no correlations with other variables with the exception of dehydroepiandrosterone sulphate (r= 0.51, P= 0.001).

Correlation to anthropometric measurements

Visceral fat mass showed significant correlations to other anthropometric measures, with the strongest explanatory variable being waist circumference (Figure 3) (r= 0.75, P < 0.001), then weight (r= 0.64, P < 0.001) and BMI (r= 0.63, P < 0.001). Waist:hip ratio showed the weakest correlation (r= 0.46, P= 0.002). Subcutaneous fat showed similar trends, although the correlations were stronger (waist circumference r= 0.89, P < 0.001; WHR r= 0.49, P= 0.001).

Figure 3.

Correlation of waist circumference with visceral fat mass (r= 0.75, P < 0.001).

Insulin resistance showed closest correlation to visceral fat mass (r= 0.68, P < 0.001), then to waist circumference (r= 0.62, P < 0.001), BMI (r= 0.52, P < 0.001), weight (r= 0.51, P < 0.001), subcutaneous fat mass (r= 0.50, P= 0.001) and weakest correlation with WHR (r= 0.36, P= 0.01). Testosterone showed no correlation with any measure of fat mass.

Regression model to predict insulin resistance

The best model for predicting insulin resistance is with visceral fat mass and triglycerides as the explanatory variables (r= 0.72, P < 0.001). Additional variables that may plausibly affect insulin resistance, that is testosterone, cholesterol and subcutaneous fat mass, make no difference to the accuracy of the model. The model was also effective when waist circumference and triglycerides were used (r= 0.70, P < 0.001).

Discussion

Despite only accounting for about 6% of total fat mass in women, our findings suggest that visceral fat is the most significant variable correlating with metabolic dysfunction in women with PCOS. Our data confirm that visceral fat is closely related to insulin resistance in women with PCOS. It is already well described that women with PCOS have significant correlations between BMI and markers of insulin resistance,10–15 but our findings support the hypothesis that the main correlate of insulin resistance is that of visceral fat mass rather than overall body mass. Although we did find correlations with subcutaneous fat, weight and BMI, these were less marked. There was no evidence of effect from the visceral:subcutaneous fat mass ratio, even though others have found this measure useful in obese populations.16 Our data in women with PCOS are consistent with observations made in other populations, that it is visceral fat that appears to be the most important correlate with insulin resistance and that subcutaneous fat mass, overall fat mass and BMI are explanatory variables of this.1,17,18

Visceral fat mass and insulin resistance also had significant correlations with other aspects of the metabolic syndrome, notably triglycerides, urate, β-cell function and an inverse correlation with HDL cholesterol. These metabolic relationships have been well described in the general population19 suggesting that our population of women with PCOS have similar metabolic trends. However, our population was one which is relatively young and healthy compared with those who are usually investigated for metabolic dysfunction. Our participants presented because of anovulation and not for any other medical complaint, so it was perhaps surprising that despite the relatively small numbers, our study detected some metabolic trends that are more commonly associated with an older population or one specifically selected for being at high risk of diabetes.

There was no evidence for any significant effect of androgens on metabolic parameters or fat distribution. While androgens are known to have effects on fat mass, muscle distribution and carbohydrate metabolism, their effect in premenopausal women is less clear.20 Other authors have reported that it is unlikely that altered fat metabolism in PCOS is secondary to raised androgens.21,22 Our observations of both a lack of evidence for correlations with androgens and a significant correlation of triglycerides with insulin resistance and visceral fat are compatible with the hypothesis that insulin resistance is secondary to elevated portal fatty acids. It has been suggested that triglycerides may also diminish SHBG, thereby causing or exaggerating a hyperandrogenic state.22,23

The gonadotrophin LH showed an inverse relationship to visceral fat and insulin resistance. A similar inverse correlation between LH and BMI has been reported before,13,24 but remains unexplained.

This trial lacked the power to detect correlations where effects were subtle or where measurements were prone to error. A larger cohort may have yielded statistically significant effects for parameters we were unable to detect, such as for fasting glucose, cholesterol:HDL cholesterol ratio, HbA1C or blood pressure. Our aim was only to ascertain whether correlations existed, but the lack of a control group means that no inference can be made as to whether the relationships we found in women with PCOS are similar in magnitude to other populations. It would have been interesting to have been able to include a non-PCOS obese control group for comparison. Ethical permission would have required an alternative method of assessment that did not require ionising radiation, and most alternatives are less accurate,25 although the use of magnetic resonance imaging for visceral fat assessment is now becoming more established.26 In order to limit the radiation dose from the CT scan, we used a single-slice technique which introduces an estimated error of 4.6% when compared with multiple slices.27 We choose to measure L2 at full inspiration as this level gives the most accurate result from a single slice28 and approximates to the level at which waist circumference is measured.

In this cross-sectional analysis, no cause–effect relationship can be inferred. However, a longitudinal study in a population with similar metabolic characteristics to women with PCOS found that visceral fat accumulation preceded insulin resistance and is therefore presumed causative.29 The correlation, as shown in Figure 2, between visceral fat and insulin resistance is linear and extends throughout the whole range measured. If PCOS were a condition of primary insulin resistance, then it might be expected that the trend would be curved towards insulin resistance at the lean end of the spectrum. Our data therefore support the hypothesis that visceral fat either causes insulin resistance or is a very early effect of it.

It would be impractical to measure visceral fat in routine clinical practice, and consensus is that there is no role for testing insulin resistance in women presenting with PCOS.30 Our findings provide reassurance that one of the simplest anthropometric measurements, that of waist circumference, provides the best correlation with visceral fat mass, insulin resistance and metabolic disturbance. However, the more widely used measurements of BMI and WHR provide much weaker correlations and in our view are less relevant in clinical practice.

It is relevant to clinical practice that, in women with PCOS, visceral fat has such a close relationship to insulin resistance and metabolic dysfunction. It is known that there is a preferential loss of visceral fat with dieting31 and exercise.32 A small overall weight loss can lead to significant reductions in visceral fat33 and clinically significant effects in women with PCOS.34 Our findings would suggest that, in women with PCOS as with the general population,35 these changes should be most easily discernable by measuring the waist circumference and not the BMI or WHR. Our findings would support the view that women with PCOS should in the first instance receive lifestyle advice, particularly with regard to adequate exercise which has the greatest influence on body fat mass after genetic factors.36 The importance of visceral fat would imply that this should both help reduce short-term effects of metabolic dysfunction and also provide longer term health benefits,37,38 both of which metformin may fail to provide in younger women.39

Conclusion

In our population of women with PCOS, there is a strong linear relationship between visceral fat mass and insulin resistance, which includes the lean group. This supports the hypothesis that visceral fat deposition is either a very early effect of, or possibly cause of, insulin resistance in this population. It also implies that reducing visceral fat should reduce insulin resistance, which accounts for the observation that lifestyle modifications have beneficial effects on both the short-term and long-term sequelae of metabolic dysfunction in these women. It may also explain why interventions that do not reduce visceral fat may have only limited success and why exercise and weight loss appear to be more effective than pharmacological treatments. The findings also give rise to two predictions: first, that as obesity becomes more common, both the incidence and severity of PCOS will rise, which might not be expected if PCOS is a condition of primary insulin resistance. Second, a simple genetic explanation for PCOS is unlikely to be fruitful as glucose homeostasis and fat deposition are intimately related and the subject of many different gene–environment interactions.

Ancillary