Serum and follicular fluid chemerin and chemerin mRNA expression in women with polycystic ovary syndrome: Systematic review and meta‐analysis

Abstract Introduction Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders among women of reproductive age. Chemerin, a novel adipokine, is involved in inflammation, energy metabolism, adipogenesis, angiogenesis and insulin secretion in the adipose cells and ovary. This systematic review with meta‐analysis aimed to compare serum and follicular fluid (FF) chemerin and ovarian chemerin mRNA expression among women with PCOS and non‐PCOS. Methods Electronic databases including Web of Science, PubMed, Google Scholar, Scopus, Cochrane and CINAHL were used for a comprehensive search through April 2021. Of the 174 articles initially identified, 22 studies met the eligibility criteria. A random‐effects model with a weighted mean difference (WMD) and 95% confidence interval (CI) was performed to compare the outcomes between groups. Subgroup and sensitivity analyses were performed to detect the sources of heterogeneity. Results Women with PCOS compared to without PCOS showed significantly higher serum chemerin [WMD: 12.02 pg/ml (95% CI: [10.92, 13.13]), p < .001], chemerin mRNA expression [WMD: 0.38% (95% CI [0.25, 0.52]), p = .001] and FF chemerin [(WMD): 41.7 pg/ml (95% CI [17.89, 65.5]) p < .001]. Further, serum chemerin remained high in PCOS women even with subgroup analysis based on body mass index (BMI) or sample size (p < .001). Serum chemerin was higher in women with PCOS and higher BMI [(WMD): 3.29 pg/ml (95% CI: [2.73, 3.384]), p < .001]. The expression of chemerin mRNA was significantly higher in the PCOS group compared to the control group [WMD: 0.38% (95% CI [0.25, 0.52]), p < .001]. Conclusion Serum and FF chemerin and mRNA expression were higher in the PCOS group compared to the controls. Further, serum chemerin was higher in PCOS women with higher BMI compared to lower BMI. The present findings illustrate that chemerin may be associated with PCOS status and BMI, independently.


| INTRODUC TI ON
Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders among women of reproductive age 1 with a prevalence of 6%-10%. 2 The clinical and metabolic disturbances of PCOS include acne, polycystic ovaries, hyperandrogenism, hirsutism, anovulation, abnormal adipokine levels, inflammation, insulin resistance (IR), glucose intolerance, obesity, type 2 diabetes mellitus, metabolic syndrome and cardiovascular risk factors, infertility and eating disorders. [3][4][5] Chemerin, a novel adipokine, is encoded by gene retinoic acid receptor responder 2 6 and synthesized as a 163-amino acid precursor (pro-chemerin163S) 7 in a wide variety of cells and tissues such as placenta, lungs, skeletal muscles, kidneys, ovarian cells (granulosa cells), hepatocyte and adipocyte (brown/white). 8 The follicular fluid (FF) chemerin levels are higher than in plasma. 9 Chemerin with autocrine, paracrine or endocrine roles 10 is involved in inflammation, energy metabolism, adipogenesis, 11,12 angiogenesis, 13 failed assisted reproductive technology, 14 and insulin secretion, 15 IR and fat content in the adipose cells and ovary. 10 Some studies reported that serum, 16,17 and FF 18 chemerin levels were increased in women with PCOS.
One of the primary therapeutic strategies of PCOS is focusing on complications and metabolic disorders related to PCOS. 19 The modification of chemerin levels in patients with PCOS may be one of the therapeutic goals of PCOS in future, due to the role of chemerin in metabolic dysfunction. 18 A meta-analysis recently evaluated only serum chemerin levels in women with PCOS compared to non-PCOS (control) (both groups with normal weight). Moreover, only five studies were analysed in this regard and did not survey women with different BMI (obese or thin women). 20 Moreover, the findings of FF chemerin levels and ovarian mRNA expression of chemerin were not pooled in the previous meta-analyses.
Therefore, given the inconsistent and insufficient evidence (the various population (obese, normal weight, lean, with or without IR), small sample size and lack of the previous comprehensive metaanalysis), this systematic review with meta-analysis was aimed to evaluate serum or FF chemerin levels and also the ovarian mRNA expression of chemerin in women with PCOS in comparison with non-PCOS.

| Search strategy
This literature was designed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). 21 The registration number of our systematic review in PROSPERO is CRD42020218793. A comprehensive literature search was performed to find the relevant articles published from inception to April 2021. Two independent investigators (G.A. and M.A.) searched the electronic databases including ISI Web of Science, PubMed, Google Scholar databases, Scopus, Cochrane, Scientific Information Database (SID) and CINAHL using the combination of the following keywords: • Polycystic ovary syndrome, polycystic ovary disease, ovary syndrome, polycystic syndrome, polycystic ovary, stein-leventhal syndrome, stein-leventhal, sclerocystic ovarian degeneration, ovarian degeneration, sclerocystic ovary syndrome, polycystic ovarian syndrome, ovarian syndrome, sclerocystic ovaries, sclerocystic ovary, PCOS.
The search query of this review was formulated using the PICO The details of the search strategy for one of the electronic databases (PubMed) were shown in Table S1. In addition, the references cited in all selected articles were retrieved to identify further relevant publications. There was not any restriction according to the conference paper, publication year and language.

| Inclusion and exclusion criteria
The inclusion criteria for the included studies for the present metaanalysis were as follows: (1) observational studies (cross-sectional, cohort and case-control) and clinical trials (the presence of at least one of the studied variables at the baseline), (2) human studies with no restrictions on study parameters (study duration, design or sample size) and race, (3) women of reproductive age (no restriction for body mass index (BMI)) whom a diagnosis of PCOS was made according to the criteria defined by National Institutes of Health, 23 Rotterdam, 24 Androgen Excess Society, 25 laparoscopic 26 or International Classification of Diseases, 27 (4) the studies assessed serum chemerin levels, FF chemerin levels or ovarian chemerin mRNA expression in both groups and (5) the studies in which the form of reported variables was presented in mean ± standard deviation (SD) or standard error of the mean (SEM), and median with interquartile range (IQR) or (minimum-maximum). In the case of multiple reports for the same studied participants, the most complete data set was analysed.
Exclusion criteria included the studies which (1) compared the differences between women and men, (2) did not present reliable and sufficient data of serum chemerin, FF chemerin or ovarian chemerin mRNA expression and (3) did not have a control group.

| Study selection
After excluding duplicate studies, two researchers (G.A. and M.A.) independently reviewed the titles and abstracts to find the relevant articles. Then, they read the full text of the included studies for selecting the eligible studies, according to the predefined criteria. Any disagreement was resolved by the discussion with a third investigator (M.Z.).

| Data extraction
The data of the eligible studies were independently extracted by two researchers (G.A. and M.A.) in close consultation with a third researcher (M.Z.) as follows: First author's name, year of publication, PCOS diagnostic criteria, country, type of the study, the total number of the participants, number of the participants in each group, age, serum or/and FFchemerin levels, and ovarian chemerin mRNA expression and assay approach for all the studied variables. The final data used in the meta-analysis were rechecked to minimize the extraction errors by three researchers.

| Quality assessment of studies
The quality of the selected studies was appraised using the Newcastle- The NOS tool includes eight items in three domains with scores ranging from 0 to 9 stars. Three domains include selection, comparability and relating to the study type outcome (cohort studies) or exposure (case-control studies). A maximum of one star in each item is allocated to the highest quality studies with the exception of one item (comparability) that is awarded two stars. The quality of the studies was categorized as low (0-3 stars), moderate (4-6 stars) and high (7-9 stars) based on total stars. 30 The Cochrane Collaboration's tool includes six domains in seven items as follow: (1) selection bias ((two items): (a) random sequence generation and (b) allocation concealment), (2) performance bias (one item), (3) detection bias (one item), (4) attrition bias (one item), (5) reporting bias (one item) and (6) other sources of bias (one item). With respect to each item, the risk of bias was assessed as unclear (the details of the studies were not enough for the judgment) (score = 0), high (score = −1) or low (score = 1). The overall quality of each trial was categorized into low (score = −6 to 0), medium (score = 1-3) or high (score = 4-6). 29

| Outcome measures
The primary outcomes included the comparison of serum or FF chemerin levels and ovarian chemerin mRNA expression in women with and without PCOS.

| Statistical analysis
We calculated the pooled estimates of weighted mean differences (WMD) and 95% confidence interval (95% CI) to evaluate the differences in the outcomes between the case and control groups. A random-effects model was used to compare the outcomes between the groups. A chi-square test (Cochran Q test) and I 2 were used to evaluate between-study heterogeneity. Heterogeneity was considered low if I 2 < 30%, moderate if I 2 = 30-75% and high if I 2 > 75%. 31 Subgroup analysis was performed to detect the sources of heterogeneity due to insufficient available data based on BMI (BMI < or >25 kg/m 2 ) and the sample size of the studies (<30 or >30). Each study with a sample size <30 in the case or the control group was inserted in a group of sample size <30 and each study with a sample size of more than 30 (>30) in both case and control groups was inserted in a group of sample size >30. To estimate the impact of each study on the overall effect size by the leave-one-out method (when one study had been excluded in each turn and repeated the analysis), a sensitivity analysis was performed using influence analysis. Funnel plots were performed to indicate the presence of publication bias using Egger's regression asymmetry test. STATA 14.0 (Stata Corp.) was used for the analysis of data. Statistically, a significant difference was considered as p < .05.

| Search result
Primary identified records through the databases and other sources searching and also flow diagram of the study selection process were presented in Figure 1 and Table S2. In total, 175 studies were preliminarily identified by the search study and the reference lists of the studies. After removing duplicates, 77 articles remained. Based on the titles and abstracts screening, 46 articles were excluded. From 31 of the selected studies for fulltext screening, 22 studies met inclusion criteria after reading the full text. GetData Graph Digitizer software was used for the data extraction of some studies. 10,17,18,[32][33][34][35] Some data were received by email. 35,36 The reasons for the exclusion of nine articles after reading full text were illustrated in Table S3.

| Study characteristics
The characteristics of the 22 included studies for the present metaanalysis were summarized in Table S4.

| Qualitative synthesis
In total, serum and FF chemerin levels were measured in nineteen 16   The methods of assessing serum and FF chemerin levels were

| Quantitative synthesis
Forest plot and funnel plot diagrams and influence analysis (leaveone-out sensitivity analysis) on our outcomes are depicted in  Table S7. Meta-analysis and subgroup analysis for the comparison of the outcomes between groups were projected in Table 1.

| The comparison between the PCOS and non-PCOS groups
At first, the meta-analysis was performed for all studies. Significant ( Figure S1). Sensitivity analysis was performed using influence analysis through the leave-one-out method for all studies (Table S7); heterogeneity did not change by sensitivity analysis.  Figure 3).

| Meta-analysis for ovarian chemerin mRNA expression
The random-effect model of meta-analysis revealed a significantly higher expression of chemerin mRNA in the PCOS group compared to the control group (WMD: 0.38%, 95% CI: 0.25, 0.52, p < .001).

| Publication bias
The funnel plot for serum chemerin levels in all studies and in the studies regarding BMI >25 or <25 was depicted in Figures S1 and S2   PCOS compared to the controls with regardless of another metabolic or androgenic status (BMI, insulin, IR and lipid profiles). 14,18,33,35,39,47 The difference in the metabolic status may play a greater role in increased chemerin levels.
As the previous studies 14,18 and the present study emphasized, both serum and FF chemerin were higher in PCOS in compared to non-PCOS. In the present study, the difference between the metabolic status of women may have led to non-significant chemerin levels between women with PCOS and BMI <25 kg/m 2 compared to women with non-PCOS and BMI <25 kg/m 2 .
High heterogeneity was observed in all analyses which may be related to the different races, ages, different population backgrounds, severity of the disease, and diet of the studied population and type of kit was used to evaluate the outcomes. No single study contributed to between-study heterogeneity in subgroup and sensitivity analysis.
The function of chemerin on glucose and lipid metabolism  (Table S9)   the lack of enough data for chemerin mRNA expression was lead to its evaluation only in the ovary; (7) protein expression of chemerin was not assessed in the present study due to the lack of enough data. Therefore, it was impossible to evaluate whether an enhancement in ovarian chemerin mRNA expression leads to an increase in protein expression (protein expression was evaluated only in two studies (subcutaneous and omental adipose tissues in one study 17 and granulosa cells in another one) 18 ); (8) the quality of few studies were high (n = 5; (9), two studies that were found in the search, due to lack of the abstract, full text or every substantial detail, were excluded (Table S3). They might have been included some important data related to this review; and finally, of 22 studies used in the present review, 21 were observational studies (cross-sectional).
Hence, it seems to be necessary to interpret the results for wider applications, cautiously due to the methodological deficiencies of the included studies.
The strengths of the present systematic review include the following (1)

| CON CLUS ION
Serum chemerin levels were remarkably higher in women with PCOS compared to the controls even with subgroup analysis based on BMI or sample sizes. The serum chemerin levels were higher in women with PCOS and higher BMI compared to lower BMI. Moreover, FF chemerin levels and ovarian chemerin mRNA expression were higher in women with PCOS in comparison with women with non-PCOS. The present findings illustrated that chemerin may be associated with both PCOS status and BMI, independently. It seems to be necessary to design the prospective studies to control the potential confounders such as race, age, ethnicity and history of diseases.

ACK N OWLED G EM ENT
Not applicable.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest.

E TH I C S A PPROVA L A N D CO N S E NT TO PA RTI CI PATE
Not applicable.

CO N S E NT FO R PU B LI C ATI O N
Not applicable.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author.