| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
krha,
M. Hill,
M. Fanta,
L. Haaková,
J. VrbÍková and
J. 
ivn
Department of Obstetrics and Gynecology (D.C., M.F., L.H., J.Z.), Charles University, Praha 2, 120 00, Czech Republic; Department of Internal Medicine (J.S.), Charles University, Praha 2, 120 00, Czech Republic; and Institute of Endocrinology (M.H., J.V.), Prague, Praha 1, 100 00, Czech Republic
Address all correspondence and requests for reprints to: David Cibula, M.D., Ph.D., Department of Obstetrics and Gynecology, Charles University, Apoliná
ská 18, Praha 2, 120 00, Czech Republic. E-mail: david.cibula{at}iol.cz.
Abstract
Insulin resistance is a frequent (although not constant) abnormality in both obese and nonobese women with polycystic ovary syndrome (PCOS). It plays a key role in the predisposition to type 2 diabetes, which is the most important health consequence of the syndrome. Identification of patients with insulin resistance is significant both for follow-up and for therapeutic reasons. The aim of the study was to evaluate the relationships between insulin sensitivity, measured by euglycemic clamp, and both endocrine and metabolic indices and to identify the best model for predicting insulin sensitivity. A total of 41 nonobese women fulfilling the diagnostic criteria for PCOS were enrolled in the study. None of the androgens correlated with the insulin sensitivity index. All clamp parameters correlated with SHBG, triglycerides, and body mass index, although no correlation was found with waist to hip ratio or waist circumference. The close relationship between insulin sensitivity and SHBG was documented by factor analysis and by its presence in all prediction models as the most significant (or even the single) predictor of the insulin sensitivity index. In conclusion: 1) a decreased level of SHBG can be used as a single reliable parameter in the prediction of insulin sensitivity in nonobese women with PCOS; and 2) waist to hip ratio, waist circumference, and androgen concentrations have no predictive value.
DECREASED INSULIN SENSITIVITY has been documented in both obese and nonobese women with polycystic ovary syndrome (PCOS), but the prevalence of insulin resistance is not known (1, 2). Although there is no prospective study, it is assumed that it is the subgroup of women with PCOS and concurrent insulin resistance that is at increased risk of diabetes and possibly cardiovascular disease (3, 4). Identification of patients with impaired insulin sensitivity is significant, both for follow-up and for long-term therapy.
In our study, we investigated a large group of nonobese women fulfilling the generally accepted diagnostic criteria of PCOS, using the euglycemic clamp. The aim of the study was to evaluate the relationships between insulin sensitivity and endocrine and metabolic indices, and possibly to identify a parameter or its combination, that would be applicable in clinical practice to predict insulin sensitivity.
Subjects and Methods
Subjects
A total of 41 patients were enrolled in the study, based on the following diagnostic criteria of PCOS: 1) oligomenorrhea from menarche; 2) an increased concentration of testosterone (0.5–2.63 nM/liter), androstenedione (1.57–5.4 nM/liter), or dehydroepiandrosterone (DHEA) (0.8–10.5 nM/liter); and 3) clinical manifestation of hyperandrogenism (acne, hirsutism, or both). Only nonobese women with a body mass index (BMI) under 30 kg/m2, above 18 yr of age, who had not used hormonal therapy during the previous 6 months, were included. All patients were informed about the study protocol and signed an informed consent. The Local Ethics Committee approved the study protocol.
Euglycemic hyperinsulinemic clamp
The hyperinsulinemic euglycemic clamp was performed as described previously (5). Briefly, one cannule, to obtain blood for biochemical analyses during the clamp, was inserted into the wrist vein. A double-lumen catheter for continuous blood glucose determination was inserted into the cubital vein of the ipsilateral arm. A third cannule for insulin and glucose administration by Biostator (GCIIS, Elkhart, IN) was inserted into the contralateral forearm vein. After a 30-min washout period, hyperinsulinemic euglycemic state was reached during the next 45 min, and the clamp was then performed using a constant insulin infusion rate (1 mU/kg·min) over 120 min. The glucose solution (40% wt/vol) was sampled by Biostator (mode 7:1) to maintain the blood glucose level at baseline value. During the clamp, blood glucose levels were repeatedly determined by glucose analyzer (ESAT 6660–2, PGV, Freital, Germany). Two blood samples for insulin determination were collected in the last 20 min of the clamp.
The following characteristics of insulin action were calculated: glucose disposal rate (M), defined as the amount of glucose supplied by the Biostator to maintain blood glucose levels during the last 20 min of the clamps (M, mM/kg·min); the insulin sensitivity index (ISI), defined as the ratio of M to insulin concentration at the end of the clamps (ISI, mM/kg·min per mU/liter·100); metabolic clearance rate of glucose (MCRg), expressed as the ratio of M to blood glucose concentration (MCRg, ml/kg·min).
Assays
All analytical determinations were performed at the National Reference Laboratory. Serum LH, FSH, and testosterone concentrations were measured by chemiluminescent assay using an ACS:180 autoanalyzer (Bayer Corp. Diagnostics GmbH, Leverkusen, Germany). The concentrations of DHEA, dehydroepiandrosterone-sulfate (DHEAS) and androstenedione were determined by RIA methods (Immunotech, Marseille, France). SHBG was measured using immunoradiometric assay kits (Orion Turku, Finland). The free androgen index (FAI) was calculated according to the following formula: FAI = 100 x testosterone (nM/liter) ÷ SHBG (nM/liter). Plasma glucose concentration was determined by a glucose oxidase method (Olympus Corp. Diagnostica GmbH, Hamburg, Germany). Plasma insulin concentrations were measured by RIA kits (CIS-Bio International, Gif-sur-Yvette, France); normal range, 4–20 mIU/ml; interassay CV < 5%; intraassay CV < 8.5%. The homeostasis model assessment index (HOMA) was calculated using the following formula: fasting serum insulin (mIU/ml) x fasting plasma glucose (mM/liter) ÷ 22.5 (6). Serum cholesterol and triglycerides (TGD) were analyzed using CHOD-PAP and GPO-PAP-based kits, respectively (Oxochrome Lachema a.s., Czech Republic); HDL-cholesterol was determined by an immunoinhibition method (HDL-C Direct, Wako Pure Chemical Industries Ltd. GmbH, Neuss, Germany). Low-density lipoprotein (LDL)-cholesterol was calculated using the Friedewald formula [LDL-cholesterol = (total cholesterol - HDL-cholesterol - TGD) ÷ 2.19 mM/liter].
Statistical analyses
For evaluation of the relationships between indices of insulin resistance, several methods were applied, such as correlation analysis followed by factor analysis and multiple regression analysis. To approximate the Gaussian data distribution and to straighten the relationships between variables, the data were transformed by power transformation to minimum skewness in individual dimensions. To avoid the influence of univariate outliers, all data with absolute studentized values greater than 2 were excluded. The multivariate outliers were searched using F-distributed Mahalanobis distance (statistical software NCSS 2000). The data treated as described above underwent correlation analysis followed by factor analysis (the principal factor method) followed by VARIMAX factor rotation (statistical software Stagraphics Plus 3.1). Besides factor analysis, multiple regression models were used to find the best predictors of insulin sensitivity represented by the indices of the euglycemic clamp. Respecting the limited number of experiments, the correlation matrix of transformed variables was used for choosing the preliminary set of predictors. These were further evaluated using backward stepwise multiple regression, with F-statistics less than 4 as the exclusion criterion.
Results
The characteristics of the entire group are demonstrated in Table 1
. Table 2 summarizes the results of the correlation analysis. There was a strong correlation of statistical significance among ISI and fasting insulin (Io), the ratio of fasting glucose (Go) to Io (Go/Io), and HOMA. None of the androgens (testosterone, androstenedione, DHEA, or DHEAS) correlated with ISI; only a correlation of borderline significance was found among M, MCRg, and androstenedione. All parameters calculated on the basis of the euglycemic clamp correlated significantly with SHBG. No significant correlation of M or ISI with Go was found. From the anthropometric parameters, we found a correlation of clamp parameters with BMI, whereas the waist to hip ratio (WHR) and waist circumference did not correlate.
|
|
and Fig. 1). Factor loadings representing the correlation coefficients between the factor and the individual variables were used as a measure of the informative value in individual markers. Two factors fulfilled Kaiser’s criterion (which is commonly used for determination of factor number, i.e. only the factors with eigenvalue more than 1 should be considered). The first factor demonstrates the similarity among Io, Go/Io, and HOMA; whereas the second expresses the close relationship between clamp parameters and SHBG. The close alliance between insulin resistance and SHBG is documented by its location in the cluster of clamp indices close to the 2nd factor axis.
|
|
Discussion
The present study demonstrates SHBG as a single reliable parameter in the prediction model of insulin sensitivity in nonobese women with PCOS.
Women with PCOS are at significantly increased risk for type 2 diabetes and impaired glucose tolerance (7, 8). Because of its high prevalence, which ranges between 4–7% in unselected population groups, PCOS is one of the greatest risk factors at a young age (9, 10). Insulin resistance plays a key role in the predisposition of women with PCOS to diabetes mellitus (4, 11). Identification of the subgroup of women with PCOS and concurrent insulin resistance is very significant, for their follow-up but also for possible therapeutic measures using insulin sensitizers.
The authors are aware that the study does not include a control group. However, it was not the aim of the study to determine the normal values for the ISI. The objective was to evaluate the relationships between the studied parameters and the insulin sensitivity measured by the euglycemic clamp. For this purpose, it is of no importance whether ISI values lie within the normal range or not. Furthermore, we included only nonobese women with a BMI lower than 30. The obesity itself is a factor that influences the phenotype of the syndrome and worsens endocrine and metabolic parameters, including insulin action (12, 13, 14). It is apparent that in obese women, obesity is one of the most significant predictive parameters of insulin sensitivity (15). We assumed that the model of nonobese women would enable us to identify other relationships than those with anthropometric parameters.
In 2000, Gennarelli et al. (15) suggested three prediction models for insulin resistance in PCOS. The diagnosis of PCOS was made on the basis of ovarian morphology rather than endocrine criteria. The study included both nonobese and obese women. In all suggested models, waist circumference was represented as an independent predictor in combination with Io, serum TGD, or subscapularis skin fold. In nonobese women, we confirmed a close relationship between insulin sensitivity and BMI; on the other hand, we did not find any relationship between insulin sensitivity and visceral fat. Our results are in accord with other studies that did not find an association between WHR and insulin sensitivity in healthy nonobese women (16).
Although the euglycemic insulin clamp and a minimal model method applied to a frequently sampled iv glucose tolerance test are considered to be the gold standards in evaluating insulin sensitivity, several other indices based on oral glucose tolerance test values or on Go and insulin levels are discussed. In our study, besides the clamp parameters, we also evaluated Go, Io, Go/Io, and the homeostasis model assessment. Although a very strong correlation was found among all of the above-mentioned parameters with the clamp results, factor analysis demonstrated weaker relationships. This result is confirmed by the prediction models, in which none of the parameters calculated from the fasting values are present.
The most significant result of our study is the finding of a very close relationship between SHBG and insulin sensitivity. Insulin is an important regulator of hepatic SHBG production. In physiological concentrations, insulin decreases SHBG production by cultured hepatoma cells (17). OK as edited? In spite of stimulating hepatic production, the acute effect of insulin on SHBG is possibly attributable to changing of binding affinity (18). Low SHBG concentration is considered a risk factor for type 2 diabetes in women (19, 20). A direct correlation between insulin sensitivity and SHBG has been demonstrated in men with type 2 diabetes (21). In patients with PCOS, one of the characteristic features is low SHBG concentration (11). A direct relationship between insulin and SHBG in obese PCOS patients was documented by Nestler (22). Suppression of insulin by diazoxide in combination with a GnRH agonist was followed by an increase in SHBG, although no change was found after the administration of a GnRH agonist alone. In our study, SHBG significantly correlated with all clamp parameters. The results of the factor analysis show a clustering of clamp parameters with SHBG. It is apparent that SHBG is a much better marker of insulin sensitivity, in comparison with Go/Io or the HOMA index. SHBG is present in all prediction models for ISI, M, or MCRg as the most significant or even the single predictor for ISI. Based on our results, we may conclude that SHBG is a single reliable predictor of insulin sensitivity in lean PCOS women. Considering its simple evaluation, it could be used in clinical practice as an indicator for further tests of insulin sensitivity (preferably the euglycemic clamp or iv glucose tolerance test) and potentially for treatment with insulin sensitizers.
In summary, in nonobese women with PCOS, we did not find a correlation between insulin sensitivity and WHR, waist circumference, or androgen concentrations. Parameters of insulin action that were calculated based on fasting values correlated poorly with insulin sensitivity measured by the euglycemic clamp. On the other hand, a very strong relationship was found between insulin sensitivity and SHBG. Although our study unables us to indicate the cutoff value for SHBG, we may conclude that a decreased level of SHBG can be used as a single predictor to define the subgroup of patients with an increased risk of insulin resistance, among nonobese women with PCOS.
Footnotes
This work was supported by Grant NH/6558-3 from the Internal Grant Agency of the Ministry of Health of the Czech Republic.
Abbreviations: BMI, Body mass index; DHEA, dehydroepiandrosterone; DHEAS, dehydroepiandrosterone-sulfate; Go, fasting glucose; Go/Io, Go-to-Io ratio; HDL, high-density lipoprotein; HOMA, homeostasis model assessment index; ISI, insulin sensitivity index; Io, fasting insulin; LDL, low-density lipoprotein; M, glucose disposal rate; MCRg, metabolic clearance rate of glucose; PCOS, polycystic ovary syndrome; TGD, triglycerides; WHR, waist to hip ratio.
Received April 15, 2002.
Accepted September 4, 2002.
References
This article has been cited by other articles:
 |
|
 |
|
  N. Kajaia, H. Binder, R. Dittrich, P. G Oppelt, B. Flor, S. Cupisti, M. W Beckmann, and A. Mueller Low sex hormone-binding globulin as a predictive marker for insulin resistance in women with hyperandrogenic syndrome Eur. J. Endocrinol., October 1, 2007; 157(4): 499 - 507. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Miles and M. D. Jensen Counterpoint: Visceral Adiposity Is Not Causally Related to Insulin Resistance Diabetes Care, September 1, 2005; 28(9): 2326 - 2328. [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Adams, A. E. Taylor, W. F. Crowley Jr., and J. E. Hall Polycystic Ovarian Morphology with Regular Ovulatory Cycles: Insights into the Pathophysiology of Polycystic Ovarian Syndrome J. Clin. Endocrinol. Metab., September 1, 2004; 89(9): 4343 - 4350. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Ganie, M. L. Khurana, M. Eunice, M. Gulati, S. N. Dwivedi, and A. C. Ammini Comparison of Efficacy of Spironolactone with Metformin in the Management of Polycystic Ovary Syndrome: An Open-Labeled Study J. Clin. Endocrinol. Metab., June 1, 2004; 89(6): 2756 - 2762. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Skrha, T. Haas, G. Sindelka, M. Prazny, J. Widimsky, D. Cibula, and S. Svacina Comparison of the Insulin Action Parameters from Hyperinsulinemic Clamps with Homeostasis Model Assessment and QUICKI Indexes in Subjects with Different Endocrine Disorders J. Clin. Endocrinol. Metab., January 1, 2004; 89(1): 135 - 141. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P.-H. Ducluzeau, P. Cousin, E. Malvoisin, H. Bornet, H. Vidal, M. Laville, and M. Pugeat Glucose-to-Insulin Ratio Rather than Sex Hormone-Binding Globulin and Adiponectin Levels Is the Best Predictor of Insulin Resistance in Nonobese Women with Polycystic Ovary Syndrome J. Clin. Endocrinol. Metab., August 1, 2003; 88(8): 3626 - 3631. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
