Serum Chemerin Levels Correlate With Severity of Dysglycemia in Young Adult Women With Polycystic Ovary Syndrome

Abstract

Context

A subset of polycystic ovary syndrome (PCOS) individuals also have type 2 diabetes (T2D); an unmet need to identify this subgroup exists.

Objective

We looked at the potential role of serum chemerin, a proinflammatory adipokine, in identifying dysglycemic PCOS.

Methods

A total of 93 PCOS and 33 healthy controls were classified, based on fasting and 2-hour plasma glucose levels (2hPGPG) and glycated hemoglobin A_1c (HbA_1c) (%) into normoglycemic (n = 34), dysglycemic (n = 33), and T2D (n = 26). Serum chemerin were measured by enzyme-linked immunosorbent assay. Homeostatic model 2 assessment of insulin resistance (HOMA-2IR) and homeostatic model 2 assessment of β-cell function (HOMA-2β) were computed using serum C-peptide.

Results

Metabolic syndrome was present in 9.7% (National Cholesterol Education Program) of PCOS. Waist circumference, body fat (%), 2hPGPG, and HbA_1c levels were significantly higher in T2D group. Serum triglycerides/high-density lipoprotein cholesterol (TGs/HDL-c) ratio was increased in PCOS individuals with T2D; no significant changes in total cholesterol and LDL-c levels were seen. Serum chemerin levels were significantly higher (P < .001) in the PCOS group. Total body fat (%), 2hPGPG, HbA_1c, and TG/HDL-c ratio correlated positively with chemerin levels. Serum chemerin levels correlated positively with HOMA2IR and negatively with HOMA-2β. On receiver operating characteristic curve analysis, a serum chemerin cutoff level of greater than 309.3 ng/mL differentiated PCOS individuals with dysglycemia from those without (sensitivity 85.71%, specificity 89.47%). The Cohen kappa test revealed a substantial agreement (P < .001) between chemerin cutoff and 2hPGPG levels greater than 200 mg/dL. The present study is arguably the first ever to define a serum chemerin cutoff to distinguish PCOS individuals with T2D from those without.

Conclusion

Elevated serum chemerin levels reliably identify PCOS individuals with dysglycemia. Further, longitudinal studies with larger samples are required to confirm this association.

Polycystic ovary syndrome (PCOS), is a common endometabolic disorder of women of reproductive age, and is characterized by ovulatory dysfunction, menstrual irregularities, and clinical and/or biochemical hyperandrogenism, with a worldwide prevalence ranging from 10-15% [1, 2]. PCOS is also termed the ovarian manifestation of metabolic syndrome (MetS) [2] and gives rise to type 2 diabetes mellitus (T2D) [3-5] in around 25% of cases [6]. PCOS and T2D share insulin resistance (IR) as the common denominator [5, 7-9]. However, there is no reliable biomarker available to distinguish PCOS individuals having prediabetes and T2D from individuals without.

 

Adipose Tissue Secretes Many Adipokines That, Through Paracrine and Endocrine Effects, Regulate Systemic Insulin Resistance.

Chemerin, a proinflammatory adipokine, is a critical regulator of adipogenesis, glucose homeostasis, food intake, and body weight [10, 11]. Increased chemerin messenger RNA expression has been reported in adipocyte cell lines as well as in adipose tissue isolated from diabetic animal models [12-15]. In vivo and in vitro studies on the effects of chemerin on glucose metabolism have yielded contradicting results. However, in humans, circulating chemerin concentration shows a consistently positive association with the components of MetS [12, 16-19]. Recently, elevated chemerin levels have also been reported in obese or lean PCOS women [12, 16-19]. However, whether circulating chemerin levels discriminate PCOS individuals with dysglycemia from those without is unknown. In this study, we compared serum chemerin levels along the glycemic spectrum of PCOS individuals. We also looked at the correlation of chemerin level with IR and pancreatic β-cell dysfunction.

Materials and Methods

Study Participants

Ninety-three participants fulfilling the inclusion/exclusion criteria as per Rotterdam Criteria, 2003(Supplementary Fig. S1) [20] were recruited from outpatient services. Written informed consent was obtained from each patient in their own vernacular. Sample size was calculated (33 each group) using the method as referred by Sami et al in 2013, anticipating the prevalence of PCOS is 10% and T2D is 8% [21-23]. Power of the study was calculated using G* Power software.

Studies were approved by the institutional ethics committee of the IPGME & R and SSKM Hospital, Kolkata, India (letter No. IPGME&R/IEC/2018/046) in accordance with the code of ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans.

Blood Collection and Processing of Samples:

A fast of 10 to 12 hours was recommended to the study participants before blood samples were collected. A total sample of 6 mL of venous blood was then collected in clot vials. The study participants were then asked to drink 75-g anhydrous glucose, dissolved in 300 mL of water over 5 minutes. Plasma glucose level was estimated in blood samples drawn 2 hours from the beginning of the glucose drink. For the estimation of glycated hemoglobin A_1c (HbA_1c) (%), a 50-µL whole-blood sample was used. To separate serum and cellular components in the remaining blood samples, centrifuges were run at 825g for 10 minutes at 4 °C. Serum samples were collected in cryovials for further biochemical and molecular assays.

Serum glucose level was measured for both FPG (fasting plasma glucose) and 2-hour post glucose plasma glucose (2hPGPG) test by the colorimetric method using a commercially available kit. HbA_1c % was measured by high-performance liquid chromatography method (Bio-Rad, full-line automated HbA_1c testing systems). Fasting plasma lipid profile components—total cholesterol (TC), triglycerides (TGs), high-density lipoprotein cholesterol (HDL-c), and low-density lipoprotein cholesterol (LDL-c)—were estimated spectrophotometrically. Serum C-peptide levels were measured by enzyme-linked immunosorbent assay (ELISA) (catalog No. E0009Hu, Bioassay Technology Laboratory, RRID: AB_3083012). Homeostatic model 2 assessment of insulin resistance (HOMA-2IR) and homeostatic model 2 assessment of β-cell function (HOMA-2β) were calculated from the C-peptide values using HOMA calculator version 2.2.3 from the Radcliffe Department of Medicine, University of Oxford. Serum levels of chemerin (sensitivity 7.8 ng/mL, detection range: 31.2-2000 ng/mL), tumor necrosis factor α (TNF-α) (sensitivity 1.52 ng/L, detection range: 3-900 ng/L), and interleukin-1β (IL-1β) (sensitivity 10.07 pg/mL, detection range: 20-6000 pg/mL) were measured by ELISA using commercial kits (CusaBio for chemerin, RRID: AB_3083462 and BT, Bioassay, for TNF-α, RRID: AB_3083463 and IL-1β, RRID: AB_3083464) as per the manufacturers’ protocol. Intra-assay precision of chemerin and TNF-α is coefficient of variation (CV) (%) less than 8% and interassay precision CV (%) less than 10.

Anthropometric measurements, that is, height, weight, body mass index (BMI), and waist circumference were captured for all participants. Body fat level of the participants was measured by the TANITA UM-075 bioelectrical impedance method.

Blood pressure (systolic and diastolic) were recorded in the left arm in sitting posture after 15 minutes rest for all participants with a manual aneroid instrument. An average of 3 readings were taken to arrive at the final value.

Statistical Analysis

To evaluate the normality of continuous variables, the D’Agostino and Pearson test were performed. Data were represented using mean ±SD. More than 2 normally distributed groups were compared by ordinary 1-way analysis of variance/Brown-Forsythe and Welch analysis of variance test, depending on the equal or different SD value, respectively. For nonnormally distributed groups, the Kruskal-Wallis test was used to compare. Unpaired t test was performed for comparison between 2 normally distributed groups and Mann-Whitney U test for nonnormally distributed groups. Depending on the distribution, the correlation between 2 variables was represented using the Pearson or Spearman correlation coefficient. Binary logistic regression was performed between PCOS (normal glucose tolerance [NGT]) and PCOS (impaired glucose tolerance [IGT]/T2D) to distinguish the contribution of serum chemerin as a predictor of dysglycemia in PCOS. An analysis of receiver operating characteristics (ROC) was performed to determine the area under the curve (AUC), the cutoff value, and the sensitivity and specificity of serum chemerin level for discriminating PCOS (T2D) individuals from PCOS (NGT) individuals. The Cohen kappa test was performed to identify the agreement between chemerin-detected T2D with 2hPGPG. A P value less than .05 was considered to be statistically significant. GraphPad Prism (version 9) and SPSS 26 (trial version) were used to perform statistical analyses.

Results

Of 93 PCOS individuals, 34 had NGT, 33 had IGT, and 26 had T2D (see Supplementary Fig S1) [20]. A family history of T2D was present in 41.8% (paternal: 20.4%, maternal: 13.1%, biparental: 8.1%). MetS (defined by National Cholesterol Education Program, 2004 criteria) was present in 9.7% of participants with PCOS but was absent in the control group, although 3 controls matched 2 out of 3 criteria. Waist circumference, body fat (%), 2hPGPG, and HbA_1c levels were significantly higher in the T2D group compared to those with IGT and NGT; no significant difference in BMI and FPG was seen (Table 1). Serum TGs/HDL-c ratio level was significantly increased in PCOS individuals with T2D compared to those with IGT and NGT; no significant changes in TC and LDL-c levels were seen (see Table 1). Among the PCOS participants, 45.16% of individuals had an FG score greater than 8; only 1 healthy individual had an FG score greater than 8 (Supplementary Table S1) [20]. Serum testosterone level was significantly higher in the PCOS group (37.52 ± 29.65 ng/dL) compared to controls (22.09 ± 3.51 ng/dL; P < .05) (see Supplementary Table S1) [20]. Although serum testosterone levels were higher in the PCOS groups, the levels did not correlate with serum chemerin. In the post hoc analysis, chemerin level was found to be significantly higher in the overall PCOS group compared to controls (P < .001) (Fig 1A), and chemerin level significantly correlated with the metabolic and anthropometric parameters. A highly significant positive correlation was seen between chemerin and total body fat (r = 0.4492; P = .0047), 2hPGPG (r = 0.5755; P = .0021), HbA_1c (r = 0.4187; P = .0058), TGs/HDL-c level (r = 0.5491; P = .0001), and a partially significant correlation was seen with waist circumference (r = 0.2863; P = .0733) (Fig 2 and Supplementary Table S2) [20]. In this context, serum TG level was significantly correlated with body fat percentage (r = 0.3898; P = .0039) but did not correlate strongly with waist circumference (P value not significant). Additionally, we further classified the PCOS individuals based on their body fat percentage as lean PCOS and obese PCOS (obese PCOS ≥ 30). Our results suggest that mean serum chemerin level is significantly higher in the obese PCOS (381.6 ± 268.6 ng/mL) group compared to the lean PCOS (199.2 ± 151 ng/mL) group. But there was no significant change in serum chemerin levels between healthy individuals and the lean PCOS group (Supplementary Fig S2) [20]. Also, serum chemerin level had a significantly negative correlation with HOMA-2β (r = −0.5326; P = .0004) and a partial significantly positive correlation with HOMA 2IR (r = 0.3499; P = .0798) (see Fig 2 and Supplementary Table S2) [20]. HOMA 2IR and HOMA-2β were calculated from C-peptide measured by Ethe LISA method with a CV of less than 8% for intra-assay and less than 10% for interassay. Serum chemerin level did not correlate with BMI or FPG (see Supplementary Table S2) [20]. Serum chemerin level was significantly higher (P < .001) in PCOS participants with T2D (479.3 ± 138.9 ng/mL) and IGT (360.9 ± 166.5 ng/mL) compared to NGT (181.4 ± 61.53 ng/mL) and controls (154.3 ± 58.17 ng/mL) (see Table 1). No significant change was observed in the chemerin level between the control and PCOS group without dysglycemia (P value nonsignificant) (Fig 1B). ROC curve analysis showed that the cutoff value of serum chemerin level to predict dysglycemia in PCOS individuals is greater than 309.3 ng/mL with 85.71% sensitivity and 89.47% specificity (Fig 3). Chemerin level in control vs T2D individuals had an AUC of 0.9643 (P < .0001) with a 95% CI of 0.9112 to 1 and a likelihood ratio of 8.143. Using the aforementioned cutoff, 89.5% of the T2D participants and 14.3% of individuals with NGT had a higher value of chemerin. Similarly, serum chemerin at a cutoff value greater than 222.5 ng/mL had an AUC of 0.8421 with a sensitivity of 82.35% and specificity of 78.95% for IGT (see Fig 3). Using this cutoff, 82.35% of the IGT participants and 21.05% of NGT had a higher value of chemerin. Intra-assay precision of chemerin is CV (%) less than 8% and interassay precision is CV (%) less than 10. We further ran binary logistic regression analysis in which the Nagelkerke R Square value for differentiating PCOS individuals having T2D from the women having only PCOS was 0.793 and for differentiating PCOS individuals having dysglycemia (either T2D or prediabetes) from individuals having only PCOS was 0.589 (with insignificant Hosmer and Lemeshow test). This indicates that the model is good for testing; chemerin detects 79.8% of the PCOS individuals having T2D correctly and 58.9% of any dysglycemic individuals correctly with sufficient power (>0.95).

Table 1.

Characteristics of the study participants: measurement of anthropometric, metabolic parameters in control and polycystic ovary syndrome subgroups

Parameters	Control (n = −33)	PCOS (NGT) (n = 34)	PCOS (IGT) (n = 33)	PCOS (T2D) (n = 26)	P
BMI	24.90 ± 3.50	24.89 ± 4.28	24.24 ± 2.56	27.71 ± 5.91	.0681
Body fat, %	31.67 ± 3.38	32.91 ± 6.42	33.06 ± 4.04	38.97 ± 4.46	.0008
Waist circumference, cm	83.16 ± 9.85	86.47 ± 8.85	83.98 ± 10.15	94.89 ± 11.68	.0028
FPG, mg/dL	88.67 ± 10.59	89.13 ± 9.71	91.28 ± 10.64	96.95 ± 14.37	.0856
2hPGPG, mg/dL	119.8 ± 15.63	115.8 ± 5.61	155.2 ± 19.52	270.3 ± 47.11	<.0001
HbA1c, % (mmol/mol)	5.37 ± 0.22 (35 ± 2.4)	5.38 ± 0.33 (35 ± 3.6)	5.53 ± 0.54 (37 ± 5.9)	6.29 ± 0.91 (45 ± 9.9)	.0002
Triglycerides, mg/dL	117.2 ± 19.58	128.7 ± 25.50	133.1 ± 38.34	176.7 ± 60.53	.0023
Cholesterol, mg/dL	152.2 ± 24.37	173.1 ± 41.27	163.6 ± 27.94	174.4 ± 42.92	.2332
LDL, mg/dL	79.94 ± 18.02	93.93 ± 28.87	96.33 ± 23.66	91.13 ± 33.89	.3317
HDL, mg/dL	53.70 ± 7.14	45.64 ± 10.27	43.97 ± 8.78	43.54 ± 6.49	.0010
TNF-α, ng/L	106.7 ± 25.68	92.84 ± 32.01	106.3 ± 31.11	125.8 ± 49.42	.2557
IL-1β, pg/L	757.1 ± 115.6	721.6 ± 120.1	807.9 ± 187.3	881.4 ± 197.3	.0474
Chemerin, ng/mL	154.3 ± 58.17	181.4 ± 61.53	360.9 ± 166.5	479.3 ± 138.9	<.0001

Data are expressed as mean ± SD. P less than .05 is considered statistically significant. Statistically significant values were represented in bold. “n” represents the number of participants in respective groups.

Abbreviations: 2hPGPG, postglucose plasma glucose level; BMI, body mass index; FPG, fasting plasma glucose; HbA_1c, glycated hemoglobin A_1c; HDL, high-density lipoprotein; IGT, impaired glucose tolerance; IL-1β, insulin-1β; LDL, low-density lipoprotein; NGT, normal glucose tolerance; PCOS, polycystic ovary syndrome; T2D, type 2 diabetes; TNF-α, tumor necrosis factor α.

Figure 1.

Measurement of circulating chemerin level. A, Comparison of serum chemerin level between the control and polycystic ovary syndrome (PCOS) group. Mann-Whitney U test was performed to compute the statistical significance. P value less than .05 is considered significant. B, Measurement of serum chemerin level in control and subgroups of PCOS. One-way analysis of variance was performed to compute the P value among the groups. P value less than .05 is considered significant. All data are represented as mean ± SD.

Figure 2.

Correlation of serum chemerin level with anthropometric and metabolic parameters. A, Graphical representation of serum chemerin level positively correlated with body fat (%); chemerin as independent variable (x-axis). B, Graphical representation of serum chemerin level positively correlated with triglycerides/high-density lipoprotein cholesterol (TG/HDL-c). C, Graphical representation of serum chemerin level negatively correlated with homeostatic model 2 assessment of β-cell function (HOMA-2β).

Figure 3.

Receiver operating characteristic (ROC) curve of chemerin. ROC curve analysis of serum chemerin for the diagnosis of type 2 diabetes (T2D) and impaired glucose tolerance (IGT).

Discussion

PCOS, often considered as the ovarian manifestation of MetS, increases the risk of T2D. The identification of biomarkers that predict dysglycemia in PCOS could play an important role in preventing T2D in a young population with PCOS. The gold-standard protocol for detecting IGT, the 2hPGPG, has its limitations, including high cost, being labor-intensive, and variability issues. Therefore, identifying a biomarker for population studies remains an unmet need.

Chemerin, a recently identified adipocytokine, regulating adipocyte differentiation and gene expression, has been reported to regulate glucose and lipid homeostasis [13].

While studies have shown serum chemerin levels to be significantly increased in PCOS individuals [24], the present study shows increased serum chemerin levels only in the subset of PCOS participants having dysglycemia. To find out whether the higher chemerin levels in the PCOS individuals with dysglycemia could be attributed to higher body fat percentage alone, we classified the study population based on their body fat content as follows: control, lean PCOS, and obese PCOS. The obese PCOS group had significantly higher mean serum chemerin levels compared to the lean PCOS group. But there was no significant difference in mean serum chemerin level between the control and lean PCOS groups. These findings also strengthen our observation that elevated serum chemerin level has no association with PCOS. Previous studies have reported increased chemerin levels in MetS, nonalcoholic fatty liver disease, and T2D, but no diagnostic cutoffs of chemerin for dysglycemia in PCOS have been reported. In our study, ROC curve analysis showed that the cutoff value of serum chemerin level greater than 309.3 ng/mL predicts dysglycemia in PCOS individuals with 85.71% sensitivity and 89.47% specificity and an AUC of 0.9643. Similarly, serum chemerin at a cutoff value greater than 222.5 ng/mL had an AUC of 0.8421 with sensitivity of 82.35% and specificity of 78.95% for IGT. The Cohen kappa showed a substantial agreement (0.767; P < .001) between the chemerin cutoff greater than 309.3 ng/mL and postglucose plasma glucose levels greater than 200 mg/dL. The present study is arguably the first ever to define a serum chemerin cutoff to distinguish PCOS individuals with T2D from those without and raises the possibility of it being used as a biomarker for dysglycemia in PCOS patients.

A study by Fatima et al in 2015 [25] found that 73% of newly diagnosed T2D individuals (n = 23) can be discriminated by chemerin value greater than 13.7 ng/mL (91% sensitivity and 96% specificity, respectively). Difference may be attributed either to smaller sample size or ethnicity differences of the population.

Another study by Hamza and colleagues in 2016 [26] addressed elevated chemerin level as a noninvasive, reproducible diagnostic marker for fatty liver in obese children and adolescent patients. They predicted that the serum chemerin level at a cutoff value greater than 350 ng/mL discriminates cases from controls with 93.3% of sensitivity and 88% specificity .

We also looked at the correlation between biochemical hyperandrogenism, a defining feature of PCOS, and serum chemerin levels, but did not find any, further indicating that increased serum chemerin levels correlate only with the presence and severity of dysglycemia, rather than the level of hyperandrogenism.

In contrast to Yang et al [24], we have not found any significant correlation of chemerin level with BMI. This difference in findings may happen because in India, multiple studies have shown BMI to be a poor predictor of obesity and IR. Rather, it is total body fat that correlates best with IR [27, 28]. In context, we have shown significantly higher positive correlation of chemerin levels with body fat percentage but not with BMI. Chemerin level was measured in women with PCOS as well as women without PCOS in a study by Guvenc et al [29]. According to that study, no significant difference was observed in serum chemerin level between individuals with and without PCOS. However, circulating serum chemerin level was elevated in overweight PCOS women regardless of serum glucose levels.

Strong positive correlations between chemerin levels on the one hand and 2hPGPG and HbA_1c on the other suggest chemerin levels to be a biomarker for IR. A negative correlation between serum chemerin levels and HOMA-2β is also indicative of a predictive role of chemerin in identifying pancreatic β-cell dysfunction. A partial correlation between serum chemerin levels and C-peptide HOMA2IR is in agreement with Kort et al [30], who suggested that chemerin might play a role in IR pathogenesis.

Our study suggests that chemerin may be strongly associated with dysglycemia in PCOS patients. Individuals having chemerin levels greater than 309.3 ng/mL have a higher likelihood of having oral glucose tolerance test–confirmed T2D. A correlation study of chemerin with anthropometric and metabolic parameters confirms the involvement of chemerin in the pathogenesis of IR in PCOS individuals; however, the molecular mechanism of chemerin signaling needs to be further studied.

Limitations of the Study

We acknowledge that our study has some limitations. First, it is a cross-sectional study and all study participants belong to the same geographical area and were recruited by the clinic-based method. Although the power of the study is sufficient enough to conclude, still the sample size is small. However, further longitudinal studies with larger samples are required to confirm the association of serum chemerin levels with the severity of glycemia in adult women with PCOS.

In our study testosterone was measured by automated methos not by liquid chromatography–mass spectrometry, and free androgen index was not performed. IR was measured only by C-peptide hOMA-IR and body fat was measured by the bioelectrical impedance method, not by dual-energy x-ray absorptiometry.

Abbreviations

2hPGPG

2-hour post glucose plasma glucose level

AUC

area under the curve

BMI

body mass index

ELISA

enzyme-linked immunosorbent assay

FG Score

Ferriman-Gallwey Score

FPG

fasting plasma glucose

HbA_1c

glycated hemoglobin A_1c

HDL

high-density lipoprotein

HOMA-2IR

homeostatic model 2 assessment of insulin resistance

HOMA-2β

homeostatic model 2 assessment of β-cell function

IGT

impaired glucose tolerance

IL-1β

interleukin-1β

IR

insulin resistance

LDL

low-density lipoprotein

MetS

metabolic syndrome

NGT

normal glucose tolerance

PCOS

polycystic ovary syndrome

ROC

receiver operating characteristic

T2D

type 2 diabetes

TC

total cholesterol

TGs

triglycerides

TNF-α

tumor necrosis factor α

Funding

This work was supported by the Endocrine Society of India & Bengal (ESI & ESB). CB (award No. IF180004) is the recipient of the DST Inspire fellowship. B.M. (award No. 3/1/2(23)/OBS/2022-NCD-II) is the recipient of the ICMR senior research fellowship.

Author Contributions

S.M. conceptualized the study; S.M. and C.B. designed the study; C.B. and B.M. investigated and performed analysis and data validation; B.M. wrote the manuscript; A.Mu. assisted in the recruitment of study participants; S.P. and C.S. assisted in statistical analysis; A.Mo. assisted in performing a few of the experiments; and S.M. supervised the work and critically reviewed and edited the manuscript.

Disclosures

The authors declare no potential conflict of interest.

Data Availability

Data are available on request from the corresponding author. Supplementary data are available in the data repositories mentioned in “References.”