Journal of Trace Elements in Medicine and Biology 44 (2017) 241–246
Contents lists available at ScienceDirect
Journal of Trace Elements in Medicine and Biology journal homepage: www.elsevier.com/locate/jtemb
Epidemiology
Selenium status parameters in patients with polycystic ovary syndrome a,⁎
b
c
MARK
d
P. Zagrodzki , M. Krzyczkowska-Sendrakowska , F. Nicol , R. Wietecha-Posłuszny , T. Milewiczb, J. Kryczyk-Kozioła, Z. Chaykivskab, R. Jachb a
Department of Food Chemistry and Nutrition, Jagiellonian University, Krakow, Poland Department of Gynecological Endocrinology, Medical College Jagiellonian University, Krakow, Poland Vascular Health Division, Rowett Institute of Nutrition and Health, The University of Aberdeen, UK d Department of Analytical Chemistry, Faculty of Chemistry, Jagiellonian University, Krakow, Poland b c
A R T I C L E I N F O
A B S T R A C T
Keywords: Selenium Polycystic ovary syndrome Hashimoto disease Partial Least Squares model
Polycystic ovary syndrome (PCOS) is the most common endocrine disorder in women of reproductive age. To date, no systematic study of interactions between selenium status parameters (SSPs: serum selenium concentration, plasma glutathione peroxidase, GPX3, plasma selenoprotein P, SELENOP), sex hormones, thyroid function parameters, and other laboratory parameters in patients with PCOS has been undertaken. Therefore we aimed to compare such parameters in women with PCOS and in the control groups, and to investigate the multidimensional interactions between various parameters in PCOS patients and in controls. The subjects were diagnosed either with PCOS (n = 28, 25.4 ± 5.2 y) or with PCOS + Hashimoto disease (n = 13, 27.3 ± 5.6 y). Female patients having normal menses were recruited into the first control group (n = 70, 26.8 ± 7.3 y) or to the second control group comprising women only with Hashimoto disease (n = 10, 26.2 ± 6.9 y). No apparent differences in SSPs between control subjects and patients with PCOS, also complicated with Hashimoto disease, were identified, though such differences were noticeable for total testosterone (tT), sex hormone binding globulin, free androgen index, dehydroepiandrosterone sulfate (DHEAS), and insulin profile. The correlation between tT and DHEAS was found the strongest. The other group of mutually highly and positively correlated parameters consisted of GPX3, follicle stimulating hormone, free triiodothyronine and free thyroxine. All the latter parameters correlated negatively with vitamin D3. SSPs took part in interactions with thyroid hormones, sex hormones and some other parameters, but only for GPX3 such interactions were statistically significant. The significance of these findings remains open for further investigation, particularly in patients with PCOS and/or Hashimoto disease.
1. Introduction Polycystic ovary syndrome (PCOS) is considered to be the most common endocrinopathy in women of child-bearing age, affecting about 5–10% of them with serious reproductive, metabolic, and cardiovascular consequences [1,2]. The ovarian histopathologic, endocrine and biochemical alterations are observed in patients with PCOS. The major clinical features of the syndrome are menstrual abnormalities (oligomenorrhea or amenorrhea in 70–80% of patients, and infertility in 40% of them due to higher rates of spontaneous abortion and pregnancy complications), hyperandrogenism and polycystic ovary (PCO) appearance on ultrasound. The patients with PCOS also suffer from insulin resistance (50–60% of patients), impaired glucose tolerance, obesity (40–50% of patients) and lipid abnormalities. Consequently, they are at higher risk of type 2 diabetes mellitus, central adiposity, hypertension, and metabolic syndrome [3]. They are also ⁎
Corresponding author. E-mail address: [email protected] (P. Zagrodzki).
http://dx.doi.org/10.1016/j.jtemb.2017.08.012 Received 7 June 2017; Received in revised form 9 August 2017; Accepted 17 August 2017 0946-672X/ © 2017 Elsevier GmbH. All rights reserved.
characterized by higher prevalence of cardiovascular disease, endothelial dysfunction, endometrial cancer [4,5] and autoimmune thyroiditis [6]. The diagnostic criteria of PCOS are based on above mentioned features and include any two of the following: 1) oligo- or anovulation; 2) clinical and/or biochemical hyperandrogenemia; 3) polycystic ovaries on ultrasound (PCO-US), with exclusion of other etiologies [4]. However, owing to heterogeneity of phenotypes and metabolic symptoms, which may even change during a woman’s lifetime, there are still some controversies regarding the diagnosis of PCOS and the reasons for such variety of clinical manifestations [3]. The disorder has putative genetic background which alongside with pathogenesis, still escapes full understanding. It is assumed that hyperinsulinemia plays a central, but not sole role in pathogenesis of PCOS. On the other hand, it is known that selenium status also interacts with ovary function which has been shown in many reports and summarized in our former papers [7,8]. The set of substantial observations
Journal of Trace Elements in Medicine and Biology 44 (2017) 241–246
P. Zagrodzki et al.
thyroxine fT4 (Elecsys® fT4), free triiodothyronine fT3), sex hormones and other endocrine factors (total testosterone tT (Elecsys® Testosterone II), estradiol E2 (Elecsys® ESTRADIOL III), sex hormone binding globulin SHBG (Elecsys® SHBG), dehydroepiandrosterone sulfate DHEAS (Elecsys® DHEA-S), luteinizing hormone LH (Elecsys® LH), follicle stimulating hormone FSH (Elecsys® FSH), prolactin PRL (Elecsys® Prolactin II), metabolic parameters (insulin (Elecsys® Insulin), vit. 25(OH) D3 (Elecsys® Vitamin D total)). Glucose was determined using Cobas 8000. Some indices were calculated according to proper formulas: free androgen index FAI = [total testosterone (nmol/L)/SHBG (nmol/L)] x 100; homeostatic model assessment-insulin resistance HOMA-IR = [fasting glucose (mmol/L) x fasting insulin (mIU/mL)]/22.5; LH/FSH index = [LH (mIU/mL)/FSH (mIU/mL)].
encompass significant positive correlation between serum selenium concentrations and pubic hair development in adolescent girls [9], endometrial GPX1 stimulation by estradiol (E2) and suppression by progesterone (P4) [10], cycle dependent changes in endometrial cellular glutathione peroxidase (GPX1) activity and significant correlation of that parameter with serum E2 and luteinizing hormone, LH [11]. Moreover, many authors found higher serum Se in contraceptive-pill users than in non-users in premenopausal women [12–14]. Paszkowski et al. [15] revealed significantly decreased follicular Se in women with unexplained infertility as well as significantly higher glutathione peroxidase (GPX) activity in follicles yielding oocytes successfully fertilized than in follicles with non-fertilized oocytes. Even better have been defined the interactions between selenium status and thyroid function. They were reviewed by Schomburg [16]. However, systematic investigation devoted to interactions between SSPs, sex hormones, thyroid hormones and other relevant biochemical parameters in PCOS patients has not been published to date. Therefore, this study was aimed at evaluating relevant parameters in women with PCOS and to investigate the multidimensional interactions between them.
2.3. Statistical approach Descriptive statistics were calculated for all parameters. To compare different groups of subjects, either ANOVA or Kruskal-Wallis test with post-hoc Tukey or Dunn’s tests were applied, when appropriate. In order to describe the correlation structure between parameters Partial Least Squares model (PLS) was used. The parameters with large weights (> 0.3) on final PLS plot were assumed to be correlated. In order to express the strength of bivariate associations, for the pairs of correlated parameters the algebraic products of their corresponding weights and cosine of corresponding angle were calculated (these coefficients are called the correlation weights). The “corresponding angle” means the angle determined by two lines connecting the origin with coordinates of both parameters on the PLS final plot. The Pearson correlation coefficients for pairs of parameters were also calculated in order to compare results obtained by different statistical methodologies. Statistical analyses were carried out using packages: STATISTICA v. 12 (Statsoft, Tulsa, OK, USA), GraphPad InStat v. 3.05 (La Jolla, CA, USA) and SIMCA-P v.9 (Umetrics, Umeå, Sweden). Correlation weights were calculated using the software delivered by MP System Co. (Chrzanów, Poland).
2. Materials and methods 2.1. Patients The participants were recruited on a voluntary basis at Department of Gynecological Endocrinology of Jagiellonian University. PCOS was diagnosed according to Rotterdam Consensus Criteria [4], i.e. in case of chronic amenorrhea (no cycles in the past 180 days) or oligomenorrhea (cycles lasting longer than 35 days), clinical hyperandrogenism (Ferriman-Gallwey score of hirsutism > 8) or laboratory hyperandrogenism (testosterone above 2.0 nmol/L and/or free androgen index (FAI) > 10). The presence of cystic ovaries was established by transvaginal pelvic ultrasound examination. Cases were classified as: (1) women with PCOS (n = 28), (2) women with PCOS and Hashimoto disease (n = 13), (3) controls, i.e. women without PCOS related problems and with normal menses (n = 70), (4) controls with Hashimoto disease (n = 10). All participants originated from the Krakow region (southeastern Poland), which is known to be a region of mild selenium and iodine deficiency. Ethical Committee of Jagiellonian University approved the protocol of the study, and all patients and volunteers signed informed consent.
3. Results The descriptive statistics and a comparison of parameters studied for subjects categorized as different groups were detailed in Table 1 and Figs. 1–6. The four groups under study differed neither in age nor in BMI. We found statistically significant differences for total testosterone (Fig. 1), sex hormone binding globulin (Fig. 2), free androgen index (Fig. 3), dehydroepiandrosterone sulfate (Fig. 4), insulin (Fig. 5) and HOMA-IR (Fig. 6). Total testosterone was higher in PCOS and PCOS with Hashimoto disease as compared with controls (Fig. 1), while SHBG was lower in PCOS than in controls or controls with Hashimoto disease (Fig. 2). Accordingly, FAI index differed even stronger than tT, showing the difference not only between both groups with PCOS and controls, but also between those groups and controls with Hashimoto disease (Fig. 3). The differences for DHEAS followed the pattern for testosterone (Fig. 4). The only differences revealed for insulin and HOMA-IR were between PCOS and controls (Figs. 5 and 6). There were no differences for TSH, fT4, fT3, E2, LH, FSH, LH/FSH index, PRL, glucose, vit. 25(OH)D3 between groups of patients and controls (Table 1). Similarly, no apparent differences in SSPs between control subjects and patients with PCOS, also complicated with Hashimoto disease, were identified (Table 1). When all subject were divided into upper or lower quartiles of such parameters like LH/FSH index, FAI or HOMA-IR, there were no more differences for these parameters (data not shown). The levels of vit. 25(OH)D3 were below the recommended value of 20 ng/mL in 53% of all patients, while next 28% of patients was in the range of insufficiency (between 20 and 30 ng/mL). There was marked but not significant vit. 25(OH)D3 reduction in control group with Hashimoto disease in comparison with other groups. No correlations were found between vit. 25(OH)D3 and TSH, fT3 and fT4 in any group
2.2. Methods Serum selenium was determined using double-channel atomic fluorescence spectrometer AFS-230 (Beijing Haiguang Instrument Co., China), equipped with hydride generation system. Standard stock solution containing 1000 mg/L Se was prepared from Titrisol standard (Merck, Germany). The working solutions were made with the use of concentrated HNO3 (Merck, Germany), concentrated HCl (POCh, Poland) and 98% (m/v) NaBH4 (Sigma-Aldrich Chemise, Germany), 0.5% (m/v) NaOH (Sigma-Aldrich Chemise, Germany). All reagents were of analytical reagent grade. Human serum (MI01181, level 1, SERO AS, Billingstad, Norway) was used as reference material. The mean analyzed value was 78.6 ± 1.3 μg/g (certified value: 80 ± 2.5 μg/g). All other details of analytical procedure were described in our former paper [17]. Plasma glutathione peroxidase (GPX3) activity was measured by Paglia & Valentine method with modifications [18], while selenoprotein P (SELENOP) was determined by the method developed in Rowett Institute of Nutrition and Health [19]. All remaining parameters were determined on routine basis by means of electrochemiluminescence immunoassay (ECLIA) methods with usage of Roche Cobas 6000 with Cobas e 601 module using automated commercial immunoassays (Roche Diagnostics International Ltd., Switzerland). These determinations included thyroid function parameters (thyroid stimulating hormone TSH (Elecsys® TSH), free 242
Journal of Trace Elements in Medicine and Biology 44 (2017) 241–246
P. Zagrodzki et al.
Table 1 Summary of the parameters which did not differ significantly between groups. All data are shown as mean values ± standard deviation. For variables with skew distribution, data were transformed in logarithms and retransformed after calculations. These results are marked with asterisk. Parameter
Age (y) BMI TSH (μIU/mL) fT4 (ng/dL) fT3 (pg/mL) E2 (pg/mL) LH (mIU/mL) FSH (mIU/mL) LH/FSH index PRL (ng/mL) Glucose (mmol/L) vit. 25(OH)D3 (ng/mL) Se (μg/L) GPX3 (U/L) Sel P (μg/mL)
Groups of subjects PCOS
PCOS + Hashimoto
Controls
Controls + Hashimoto
25.4 ± 5.2 26.9 ± 8.2 1.64 ± 0.84 15.7 ± 2.6 5.34 ± 0.81 203.9 ± 151.9 5.84 (2.08; 16.35)* 5.31 ± 1.93 1.61 ± 1.39 267.1 ± 172.6 4.89 ± 1.04 22.2 ± 7.8 56.1 ± 8.6 425.7 ± 73.5 4.21 ± 0.74
27.3 ± 5.6 26.7 ± 7.5 2.27 ± 0.77 16.9 ± 2.7 5.36 ± 1.00 169.4 ± 76.8 7.60 ± 4.60 6.33 ± 1.05 1.19 ± 0.62 310.4 ± 223.1 4.82 ± 0.30 21.7 ± 8.6 57.7 ± 11.1 388.5 ± 81.8 4.83 ± 1.05
26.8 ± 7.3 22.0 ± 2.8 1.73 ± 1.20 15.6 ± 2.2 5.21 ± 1.04 177.4 (72.1; 436.9)* 5.26 (2.25; 12.30)* 5.86 (3.52; 9.77)* 1.09 ± 0.72 307.3 ± 214.5 4.67 ± 0.32 22.8 ± 12.1 57.0 ± 10.5 422.1 ± 74.9 4.42 ± 0.88
26.2 ± 6.9 20.7 ± 2.6 2.27 ± 1.96 16.8 ± 3.2 5.37 ± 0.98 304.7 ± 177.6 8.50 ± 3.44 5.80 ± 2.30 1.58 ± 0.65 294.5 ± 144.0 4.73 ± 0.31 16.4 ± 4.1 55.4 ± 13.3 452.0 ± 69.4 3.96 ± 0.70
Fig. 3. DHEAS in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.046; b: p = 0.031.
Fig. 1. Total testosterone concentrations in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.008; b: p = 0.001.
Fig. 4. FAI index in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.000; b: p = 0.004; c: p = 0.000; d: p = 0.001.
Fig. 2. SHBG concentrations in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.006; b: p = 0.023.
(Table 2). The statistically significant PLS model, having three significant components with corresponding eigenvalues equal 1.43, 1.54 and 1.16, was derived for subjects. These components accounted for 59.0% and 34.9% of the variance in the predictor and response parameters,
with PCOS and/or with Hashimoto disease. However, negative correlation between vit. 25(OH)D3 and fT3 was noted in healthy controls (R = −0.342, p < 0.05) and in all cohort of women with PCOS and controls vit. 25(OH)D3 correlated negatively with fT3, fT4 and FSH 243
Journal of Trace Elements in Medicine and Biology 44 (2017) 241–246
P. Zagrodzki et al.
Fig. 7. The weights of the first two components of PLS model. Meaning of abbreviations: DHEAS − serum dehydroepiandrosterone sulfate, FSH − serum follicle-stimulating hormone, fT3–serum free triiodothyronine, fT4–serum free thyroxine, GPX3–plasma glutathione peroxidase activity, Se − plasma selenium concentration, SELENOP − plasma selenoprotein P concentration, vit. D3–serum vitamin 25(OH)D3 concentration. The parameters on the diagram outside rectangle marked with dashed lines were considered to be correlated.
Fig. 5. Insulin concentrations in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.025.
but positive loading on second latent component. Apart from GPX3, two other SSPs (Se and SELENOP) also took part in interactions with thyroid hormones, sex hormones and other parameters, as they were included in PLS model, but these interactions were not statistically significant. The projection of subjects on the plane defined by the first two latent components of the PLS model did not reveal any homogenous clusters (results not shown). Using Pearson approach, we found significant correlations between Se and SELENOP (R = 0.477) and between Se and GPX3 (R = 0.247). The correlation between SELENOP and GPX3 was not significant. 4. Discussion Fig. 6. HOMA-IR in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.003.
The experiments on animals and observations in humans delivered several lines of evidence that the expression levels of some selenoproteins is tightly controlled and differs between sexes, even when the selenium levels are comparable [20,21]. This is further reflected by profound sex-specific differences in selenoenzyme activities in particular organs and tissues. Such observations implicate the modulatory effects of sex hormones and their likely involvement in either translational efficiency of certain selenoproteins or protein stability [20]. The exact mechanisms remain to be analyzed, but nevertheless it seems to be of particular importance in patients with PCOS to maintain homeostatic control of selenium status for efficient expression of important selenoproteins during disturbances of sex hormones metabolism. As expected, PCOS patients were characterized by significantly elevated tT, DHEAS, FAI, insulin, and HOMA-IR, and decreased SHBG compared to controls (Figs. 1–5). All these changes are typical for PCOS. For some parameters (tT, DHEAS, FAI) the differences were even more significant in the group of PCOS with Hashimoto disease. The lack of differences between various groups of subjects for all thyroid function parameters (TSH, fT4, fT3) suggests that despite some biochemical ties between thyroid and ovaries, there were not any apparent clinical effects of pathologic ovarian cysts on thyroid function. Big range from normal to high values, and skewed distribution, particularly in controls and in controls with Hashimoto disease, may be responsible for the lack of differences for estradiol between various groups of subjects. For similar reasons we did not reveal any differences for other hormones (LH, FSH, LH/FSH, PRL). Contrary to that, glucose concentrations were within relatively narrow ranges, but again without
Table 2 Correlation weights (based on PLS model), and Pearson correlation coefficients for the same pairs of parameters. Pairs of correlated parameters
Correlation weights (PLS)
Pearson correlation coefficients and level of significance
tT and DHEAS fT3 and fT4 GPX3 and fT3 fT3 and FSH GPX3 and fT4 fT4 and FSH GPX3 and FSH FSH and vit. D3 GPX3 and vit. D3 fT4 and vit. D3 fT3 and vit. D3
0.545 0.271 0.191 0.190 0.188 0.186 0.132 −0.112 −0.119 −0.170 −0.172
R = 0.696, p R = 0.450, p NS NS NS NS NS R = −0.203, R = −0.174, R = −0.251, R = −0.242,
< 0.001 < 0.001
p p p p
< < <
Contents lists available at ScienceDirect
Journal of Trace Elements in Medicine and Biology journal homepage: www.elsevier.com/locate/jtemb
Epidemiology
Selenium status parameters in patients with polycystic ovary syndrome a,⁎
b
c
MARK
d
P. Zagrodzki , M. Krzyczkowska-Sendrakowska , F. Nicol , R. Wietecha-Posłuszny , T. Milewiczb, J. Kryczyk-Kozioła, Z. Chaykivskab, R. Jachb a
Department of Food Chemistry and Nutrition, Jagiellonian University, Krakow, Poland Department of Gynecological Endocrinology, Medical College Jagiellonian University, Krakow, Poland Vascular Health Division, Rowett Institute of Nutrition and Health, The University of Aberdeen, UK d Department of Analytical Chemistry, Faculty of Chemistry, Jagiellonian University, Krakow, Poland b c
A R T I C L E I N F O
A B S T R A C T
Keywords: Selenium Polycystic ovary syndrome Hashimoto disease Partial Least Squares model
Polycystic ovary syndrome (PCOS) is the most common endocrine disorder in women of reproductive age. To date, no systematic study of interactions between selenium status parameters (SSPs: serum selenium concentration, plasma glutathione peroxidase, GPX3, plasma selenoprotein P, SELENOP), sex hormones, thyroid function parameters, and other laboratory parameters in patients with PCOS has been undertaken. Therefore we aimed to compare such parameters in women with PCOS and in the control groups, and to investigate the multidimensional interactions between various parameters in PCOS patients and in controls. The subjects were diagnosed either with PCOS (n = 28, 25.4 ± 5.2 y) or with PCOS + Hashimoto disease (n = 13, 27.3 ± 5.6 y). Female patients having normal menses were recruited into the first control group (n = 70, 26.8 ± 7.3 y) or to the second control group comprising women only with Hashimoto disease (n = 10, 26.2 ± 6.9 y). No apparent differences in SSPs between control subjects and patients with PCOS, also complicated with Hashimoto disease, were identified, though such differences were noticeable for total testosterone (tT), sex hormone binding globulin, free androgen index, dehydroepiandrosterone sulfate (DHEAS), and insulin profile. The correlation between tT and DHEAS was found the strongest. The other group of mutually highly and positively correlated parameters consisted of GPX3, follicle stimulating hormone, free triiodothyronine and free thyroxine. All the latter parameters correlated negatively with vitamin D3. SSPs took part in interactions with thyroid hormones, sex hormones and some other parameters, but only for GPX3 such interactions were statistically significant. The significance of these findings remains open for further investigation, particularly in patients with PCOS and/or Hashimoto disease.
1. Introduction Polycystic ovary syndrome (PCOS) is considered to be the most common endocrinopathy in women of child-bearing age, affecting about 5–10% of them with serious reproductive, metabolic, and cardiovascular consequences [1,2]. The ovarian histopathologic, endocrine and biochemical alterations are observed in patients with PCOS. The major clinical features of the syndrome are menstrual abnormalities (oligomenorrhea or amenorrhea in 70–80% of patients, and infertility in 40% of them due to higher rates of spontaneous abortion and pregnancy complications), hyperandrogenism and polycystic ovary (PCO) appearance on ultrasound. The patients with PCOS also suffer from insulin resistance (50–60% of patients), impaired glucose tolerance, obesity (40–50% of patients) and lipid abnormalities. Consequently, they are at higher risk of type 2 diabetes mellitus, central adiposity, hypertension, and metabolic syndrome [3]. They are also ⁎
Corresponding author. E-mail address: [email protected] (P. Zagrodzki).
http://dx.doi.org/10.1016/j.jtemb.2017.08.012 Received 7 June 2017; Received in revised form 9 August 2017; Accepted 17 August 2017 0946-672X/ © 2017 Elsevier GmbH. All rights reserved.
characterized by higher prevalence of cardiovascular disease, endothelial dysfunction, endometrial cancer [4,5] and autoimmune thyroiditis [6]. The diagnostic criteria of PCOS are based on above mentioned features and include any two of the following: 1) oligo- or anovulation; 2) clinical and/or biochemical hyperandrogenemia; 3) polycystic ovaries on ultrasound (PCO-US), with exclusion of other etiologies [4]. However, owing to heterogeneity of phenotypes and metabolic symptoms, which may even change during a woman’s lifetime, there are still some controversies regarding the diagnosis of PCOS and the reasons for such variety of clinical manifestations [3]. The disorder has putative genetic background which alongside with pathogenesis, still escapes full understanding. It is assumed that hyperinsulinemia plays a central, but not sole role in pathogenesis of PCOS. On the other hand, it is known that selenium status also interacts with ovary function which has been shown in many reports and summarized in our former papers [7,8]. The set of substantial observations
Journal of Trace Elements in Medicine and Biology 44 (2017) 241–246
P. Zagrodzki et al.
thyroxine fT4 (Elecsys® fT4), free triiodothyronine fT3), sex hormones and other endocrine factors (total testosterone tT (Elecsys® Testosterone II), estradiol E2 (Elecsys® ESTRADIOL III), sex hormone binding globulin SHBG (Elecsys® SHBG), dehydroepiandrosterone sulfate DHEAS (Elecsys® DHEA-S), luteinizing hormone LH (Elecsys® LH), follicle stimulating hormone FSH (Elecsys® FSH), prolactin PRL (Elecsys® Prolactin II), metabolic parameters (insulin (Elecsys® Insulin), vit. 25(OH) D3 (Elecsys® Vitamin D total)). Glucose was determined using Cobas 8000. Some indices were calculated according to proper formulas: free androgen index FAI = [total testosterone (nmol/L)/SHBG (nmol/L)] x 100; homeostatic model assessment-insulin resistance HOMA-IR = [fasting glucose (mmol/L) x fasting insulin (mIU/mL)]/22.5; LH/FSH index = [LH (mIU/mL)/FSH (mIU/mL)].
encompass significant positive correlation between serum selenium concentrations and pubic hair development in adolescent girls [9], endometrial GPX1 stimulation by estradiol (E2) and suppression by progesterone (P4) [10], cycle dependent changes in endometrial cellular glutathione peroxidase (GPX1) activity and significant correlation of that parameter with serum E2 and luteinizing hormone, LH [11]. Moreover, many authors found higher serum Se in contraceptive-pill users than in non-users in premenopausal women [12–14]. Paszkowski et al. [15] revealed significantly decreased follicular Se in women with unexplained infertility as well as significantly higher glutathione peroxidase (GPX) activity in follicles yielding oocytes successfully fertilized than in follicles with non-fertilized oocytes. Even better have been defined the interactions between selenium status and thyroid function. They were reviewed by Schomburg [16]. However, systematic investigation devoted to interactions between SSPs, sex hormones, thyroid hormones and other relevant biochemical parameters in PCOS patients has not been published to date. Therefore, this study was aimed at evaluating relevant parameters in women with PCOS and to investigate the multidimensional interactions between them.
2.3. Statistical approach Descriptive statistics were calculated for all parameters. To compare different groups of subjects, either ANOVA or Kruskal-Wallis test with post-hoc Tukey or Dunn’s tests were applied, when appropriate. In order to describe the correlation structure between parameters Partial Least Squares model (PLS) was used. The parameters with large weights (> 0.3) on final PLS plot were assumed to be correlated. In order to express the strength of bivariate associations, for the pairs of correlated parameters the algebraic products of their corresponding weights and cosine of corresponding angle were calculated (these coefficients are called the correlation weights). The “corresponding angle” means the angle determined by two lines connecting the origin with coordinates of both parameters on the PLS final plot. The Pearson correlation coefficients for pairs of parameters were also calculated in order to compare results obtained by different statistical methodologies. Statistical analyses were carried out using packages: STATISTICA v. 12 (Statsoft, Tulsa, OK, USA), GraphPad InStat v. 3.05 (La Jolla, CA, USA) and SIMCA-P v.9 (Umetrics, Umeå, Sweden). Correlation weights were calculated using the software delivered by MP System Co. (Chrzanów, Poland).
2. Materials and methods 2.1. Patients The participants were recruited on a voluntary basis at Department of Gynecological Endocrinology of Jagiellonian University. PCOS was diagnosed according to Rotterdam Consensus Criteria [4], i.e. in case of chronic amenorrhea (no cycles in the past 180 days) or oligomenorrhea (cycles lasting longer than 35 days), clinical hyperandrogenism (Ferriman-Gallwey score of hirsutism > 8) or laboratory hyperandrogenism (testosterone above 2.0 nmol/L and/or free androgen index (FAI) > 10). The presence of cystic ovaries was established by transvaginal pelvic ultrasound examination. Cases were classified as: (1) women with PCOS (n = 28), (2) women with PCOS and Hashimoto disease (n = 13), (3) controls, i.e. women without PCOS related problems and with normal menses (n = 70), (4) controls with Hashimoto disease (n = 10). All participants originated from the Krakow region (southeastern Poland), which is known to be a region of mild selenium and iodine deficiency. Ethical Committee of Jagiellonian University approved the protocol of the study, and all patients and volunteers signed informed consent.
3. Results The descriptive statistics and a comparison of parameters studied for subjects categorized as different groups were detailed in Table 1 and Figs. 1–6. The four groups under study differed neither in age nor in BMI. We found statistically significant differences for total testosterone (Fig. 1), sex hormone binding globulin (Fig. 2), free androgen index (Fig. 3), dehydroepiandrosterone sulfate (Fig. 4), insulin (Fig. 5) and HOMA-IR (Fig. 6). Total testosterone was higher in PCOS and PCOS with Hashimoto disease as compared with controls (Fig. 1), while SHBG was lower in PCOS than in controls or controls with Hashimoto disease (Fig. 2). Accordingly, FAI index differed even stronger than tT, showing the difference not only between both groups with PCOS and controls, but also between those groups and controls with Hashimoto disease (Fig. 3). The differences for DHEAS followed the pattern for testosterone (Fig. 4). The only differences revealed for insulin and HOMA-IR were between PCOS and controls (Figs. 5 and 6). There were no differences for TSH, fT4, fT3, E2, LH, FSH, LH/FSH index, PRL, glucose, vit. 25(OH)D3 between groups of patients and controls (Table 1). Similarly, no apparent differences in SSPs between control subjects and patients with PCOS, also complicated with Hashimoto disease, were identified (Table 1). When all subject were divided into upper or lower quartiles of such parameters like LH/FSH index, FAI or HOMA-IR, there were no more differences for these parameters (data not shown). The levels of vit. 25(OH)D3 were below the recommended value of 20 ng/mL in 53% of all patients, while next 28% of patients was in the range of insufficiency (between 20 and 30 ng/mL). There was marked but not significant vit. 25(OH)D3 reduction in control group with Hashimoto disease in comparison with other groups. No correlations were found between vit. 25(OH)D3 and TSH, fT3 and fT4 in any group
2.2. Methods Serum selenium was determined using double-channel atomic fluorescence spectrometer AFS-230 (Beijing Haiguang Instrument Co., China), equipped with hydride generation system. Standard stock solution containing 1000 mg/L Se was prepared from Titrisol standard (Merck, Germany). The working solutions were made with the use of concentrated HNO3 (Merck, Germany), concentrated HCl (POCh, Poland) and 98% (m/v) NaBH4 (Sigma-Aldrich Chemise, Germany), 0.5% (m/v) NaOH (Sigma-Aldrich Chemise, Germany). All reagents were of analytical reagent grade. Human serum (MI01181, level 1, SERO AS, Billingstad, Norway) was used as reference material. The mean analyzed value was 78.6 ± 1.3 μg/g (certified value: 80 ± 2.5 μg/g). All other details of analytical procedure were described in our former paper [17]. Plasma glutathione peroxidase (GPX3) activity was measured by Paglia & Valentine method with modifications [18], while selenoprotein P (SELENOP) was determined by the method developed in Rowett Institute of Nutrition and Health [19]. All remaining parameters were determined on routine basis by means of electrochemiluminescence immunoassay (ECLIA) methods with usage of Roche Cobas 6000 with Cobas e 601 module using automated commercial immunoassays (Roche Diagnostics International Ltd., Switzerland). These determinations included thyroid function parameters (thyroid stimulating hormone TSH (Elecsys® TSH), free 242
Journal of Trace Elements in Medicine and Biology 44 (2017) 241–246
P. Zagrodzki et al.
Table 1 Summary of the parameters which did not differ significantly between groups. All data are shown as mean values ± standard deviation. For variables with skew distribution, data were transformed in logarithms and retransformed after calculations. These results are marked with asterisk. Parameter
Age (y) BMI TSH (μIU/mL) fT4 (ng/dL) fT3 (pg/mL) E2 (pg/mL) LH (mIU/mL) FSH (mIU/mL) LH/FSH index PRL (ng/mL) Glucose (mmol/L) vit. 25(OH)D3 (ng/mL) Se (μg/L) GPX3 (U/L) Sel P (μg/mL)
Groups of subjects PCOS
PCOS + Hashimoto
Controls
Controls + Hashimoto
25.4 ± 5.2 26.9 ± 8.2 1.64 ± 0.84 15.7 ± 2.6 5.34 ± 0.81 203.9 ± 151.9 5.84 (2.08; 16.35)* 5.31 ± 1.93 1.61 ± 1.39 267.1 ± 172.6 4.89 ± 1.04 22.2 ± 7.8 56.1 ± 8.6 425.7 ± 73.5 4.21 ± 0.74
27.3 ± 5.6 26.7 ± 7.5 2.27 ± 0.77 16.9 ± 2.7 5.36 ± 1.00 169.4 ± 76.8 7.60 ± 4.60 6.33 ± 1.05 1.19 ± 0.62 310.4 ± 223.1 4.82 ± 0.30 21.7 ± 8.6 57.7 ± 11.1 388.5 ± 81.8 4.83 ± 1.05
26.8 ± 7.3 22.0 ± 2.8 1.73 ± 1.20 15.6 ± 2.2 5.21 ± 1.04 177.4 (72.1; 436.9)* 5.26 (2.25; 12.30)* 5.86 (3.52; 9.77)* 1.09 ± 0.72 307.3 ± 214.5 4.67 ± 0.32 22.8 ± 12.1 57.0 ± 10.5 422.1 ± 74.9 4.42 ± 0.88
26.2 ± 6.9 20.7 ± 2.6 2.27 ± 1.96 16.8 ± 3.2 5.37 ± 0.98 304.7 ± 177.6 8.50 ± 3.44 5.80 ± 2.30 1.58 ± 0.65 294.5 ± 144.0 4.73 ± 0.31 16.4 ± 4.1 55.4 ± 13.3 452.0 ± 69.4 3.96 ± 0.70
Fig. 3. DHEAS in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.046; b: p = 0.031.
Fig. 1. Total testosterone concentrations in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.008; b: p = 0.001.
Fig. 4. FAI index in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.000; b: p = 0.004; c: p = 0.000; d: p = 0.001.
Fig. 2. SHBG concentrations in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.006; b: p = 0.023.
(Table 2). The statistically significant PLS model, having three significant components with corresponding eigenvalues equal 1.43, 1.54 and 1.16, was derived for subjects. These components accounted for 59.0% and 34.9% of the variance in the predictor and response parameters,
with PCOS and/or with Hashimoto disease. However, negative correlation between vit. 25(OH)D3 and fT3 was noted in healthy controls (R = −0.342, p < 0.05) and in all cohort of women with PCOS and controls vit. 25(OH)D3 correlated negatively with fT3, fT4 and FSH 243
Journal of Trace Elements in Medicine and Biology 44 (2017) 241–246
P. Zagrodzki et al.
Fig. 7. The weights of the first two components of PLS model. Meaning of abbreviations: DHEAS − serum dehydroepiandrosterone sulfate, FSH − serum follicle-stimulating hormone, fT3–serum free triiodothyronine, fT4–serum free thyroxine, GPX3–plasma glutathione peroxidase activity, Se − plasma selenium concentration, SELENOP − plasma selenoprotein P concentration, vit. D3–serum vitamin 25(OH)D3 concentration. The parameters on the diagram outside rectangle marked with dashed lines were considered to be correlated.
Fig. 5. Insulin concentrations in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.025.
but positive loading on second latent component. Apart from GPX3, two other SSPs (Se and SELENOP) also took part in interactions with thyroid hormones, sex hormones and other parameters, as they were included in PLS model, but these interactions were not statistically significant. The projection of subjects on the plane defined by the first two latent components of the PLS model did not reveal any homogenous clusters (results not shown). Using Pearson approach, we found significant correlations between Se and SELENOP (R = 0.477) and between Se and GPX3 (R = 0.247). The correlation between SELENOP and GPX3 was not significant. 4. Discussion Fig. 6. HOMA-IR in various groups of subjects. The same letter indicates statistically significant difference, a: p = 0.003.
The experiments on animals and observations in humans delivered several lines of evidence that the expression levels of some selenoproteins is tightly controlled and differs between sexes, even when the selenium levels are comparable [20,21]. This is further reflected by profound sex-specific differences in selenoenzyme activities in particular organs and tissues. Such observations implicate the modulatory effects of sex hormones and their likely involvement in either translational efficiency of certain selenoproteins or protein stability [20]. The exact mechanisms remain to be analyzed, but nevertheless it seems to be of particular importance in patients with PCOS to maintain homeostatic control of selenium status for efficient expression of important selenoproteins during disturbances of sex hormones metabolism. As expected, PCOS patients were characterized by significantly elevated tT, DHEAS, FAI, insulin, and HOMA-IR, and decreased SHBG compared to controls (Figs. 1–5). All these changes are typical for PCOS. For some parameters (tT, DHEAS, FAI) the differences were even more significant in the group of PCOS with Hashimoto disease. The lack of differences between various groups of subjects for all thyroid function parameters (TSH, fT4, fT3) suggests that despite some biochemical ties between thyroid and ovaries, there were not any apparent clinical effects of pathologic ovarian cysts on thyroid function. Big range from normal to high values, and skewed distribution, particularly in controls and in controls with Hashimoto disease, may be responsible for the lack of differences for estradiol between various groups of subjects. For similar reasons we did not reveal any differences for other hormones (LH, FSH, LH/FSH, PRL). Contrary to that, glucose concentrations were within relatively narrow ranges, but again without
Table 2 Correlation weights (based on PLS model), and Pearson correlation coefficients for the same pairs of parameters. Pairs of correlated parameters
Correlation weights (PLS)
Pearson correlation coefficients and level of significance
tT and DHEAS fT3 and fT4 GPX3 and fT3 fT3 and FSH GPX3 and fT4 fT4 and FSH GPX3 and FSH FSH and vit. D3 GPX3 and vit. D3 fT4 and vit. D3 fT3 and vit. D3
0.545 0.271 0.191 0.190 0.188 0.186 0.132 −0.112 −0.119 −0.170 −0.172
R = 0.696, p R = 0.450, p NS NS NS NS NS R = −0.203, R = −0.174, R = −0.251, R = −0.242,
< 0.001 < 0.001
p p p p
< < <
