Académique Documents
Professionnel Documents
Culture Documents
Accepted Article
Article type: 5 Unsolicited Review
Mina Amiri 1, 2, Fahimeh Ramezani Tehrani *2, Fatemeh Nahidi 3, Razieh Bidhendi
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/cen.13389
This article is protected by copyright. All rights reserved.
*Correspondence Author: Fahimeh Ramezani Tehrani, MD, PhD,
Abstract
Background and Objective: A limited number of studies have evaluated the relationship
between clinical and biochemical hyperandrogenism (HA). This study aimed to evaluate the
Sciences databases (2000 to 2015) to identify studies investigating clinical and biochemical
parameters of HA in PCOS patients. In this meta-analysis, both fixed and random effect
models were applied to estimate pooled effect size. To assess the relationships between BHPs
regression analysis of pooled data from all eligible studies showed significant positive
Accepted Article
relationships of FG score with androstenedione (A4) (p= 0.034) and dehydroepiandrosterone
sulfate (DHEAS) (p=0.012), whereas it showed no association with total testosterone (tT),
free testosterone (fT), sex hormone bonding globulin (SHBG) and free androgen index (FAI).
The results did not change after adjusting for quality assessment or method of assay. Nor did
the associations between A4 and FG score remain after adjusting for age and BMI, diagnostic
criteria for PCOS and study design or the association between DHEAS and FG score remain
Conclusion: This meta-analysis confirmed the associations of some BHPs, including A4 and
DHEAS with FG score, indicating that measurement of these parameters can be useful for
Ferriman-Gallwey score.
Introduction
Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders in
(HA) and chronic anovulation. The criteria of the Androgen Excess Society (AES) for
diagnosis of PCOS require clinical or biochemical evidence of HA, the cornerstone of PCOS.
2
Excess synthesis of ovarian androgens plays an essential role in the clinical and biochemical
Although the most frequently measured laboratory parameters are fT and FAI, increased
levels of A4 or other androgen precursors are the only findings in many patients. 6, 7
Hirsutism, a common complaint in women with PCOS, refers to abnormal growth of the
terminal hair in a male-like pattern. 2, 5 The modified Ferriman-Gallwey (FG) score has now
become the gold standard for determining the density of terminal hairs at 9 different body
sites, (i.e., upper lip, chin, chest, upper back, lower back, upper abdomen, lower abdomen,
arm and thigh); a total score of 8 is considered as hirsutism. 8, 9 Many hirsute women are
found to have increased circulating androgen levels, although in some of them, androgen
levels are in normal ranges. 3, 9, 10 Some studies show that the severity of hirsutism correlates
poorly with the severity of androgen excess; it has not yet been clarified, which of the
androgenic parameters have the most significant relationship with FG score. 9, 11 This obvious
paradox raises the question of the role of other factors including peripheral metabolism of
androgens, sensitivity of target tissues to androgens, key enzymes for biosynthesis and
Although androgens modulate the biological mechanism regulating the hair cycle, limited
studies have evaluated the relationship between clinical and biochemical HA and these have
clinical characteristics in hirsute patients with PCOS have important implications for follow-
PCOS women; this information will potentially be helpful for selection of treatment options.
Accepted Article
Methods
We conducted this systematic review and meta-analysis based on the Preferred Reporting
Items for Systematic Reviews and Meta-Analyses (PRISMA) 15 to investigate the association
between BHPs and FG score. The ethics committee of the Research Institute for Endocrine
Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran approved this study.
Eligibility criteria
prospective and retrospective cohort studies, and case-control studies restricted to papers
published in English between 2000 and 2015. Studies were eligible if they had: 1. a study
(NIH) or the Androgen Excess Society (AES) criteria, with or without hirsutism, 2. assessed
FG score and at least one androgen marker, including tT, fT, A4, DHEAS, FAI and SHBG,
and 3. used a modified FG scoring system for diagnosis of hirsutism. We also excluded non-
human studies, reviews, commentaries, editorials, letters, meeting abstracts, case reports, and
studies that did not measure FG score or any hormonal markers, or those that had a study
population with idiopathic hirsutism or the other types of HA, or did not provide accurate and
PubMed, Scopus, Google Scholar, and ScienceDirect and Web of Sciences were searched for
Accepted Article
studies documented from 2000 to 2015 investigating androgen levels and FG scores in
patients with PCOS, using MeSH terms including the following keyword combinations:
OR "androstenedione" OR "dehydroepiandrosterone".
Study selection
conducted a primary selection using the titles, abstracts, and keywords to exclude studies that
did not fulfill the inclusion criteria. They independently read the full text of remaining papers
to identify eligible articles. Disagreements between reviewers were resolved by discussion or,
if necessary, by a third reviewer (FRT). Journals and authors were not blinded during study
selection.
Data extraction
The following data were collected from each study: First author and publication year, country
of study, study design and settings, total number of patients, diagnostic criteria for PCOS,
patients characteristics including age and BMI, outcome measurements including FG score,
tT, fT, A4, DHEAS and SHBG. Data were extracted from full text articles by one reviewer
(MA) in close consultation with the second (FRT). Data of all studies were extracted with
mean and standard deviation values. 16 To prevent extraction errors, a control check between
the final data used in the meta-analysis and the original publications was performed by all
authors. Baseline data of all studies such as clinical trials were extracted.
All studies included for the meta-analysis were appraised for the quality of their
Accepted Article
methodological and results presentation. Two reviewers (MA and SB) who were blinded to
study author, institution and journal name, assessed the quality of the studies separately.
validated quality assessment checklist for clinical trials. 17 All clinical trial papers were
categorized in four groups including high, moderate, low and very low quality; studies with
scores 70 % of the highest level of the CONSORT checklist were considered as high, 40 to
Moreover, the quality assessment for case control studies was based on the modified
highest level of STROBE score was considered to be 25; if a paper obtained 70 % of the
highest level of STROBE score, it was considered as high quality, 40 to 70% as moderate, 20
In addition, cohort studies were appraised according to the NewcastleOttawa scale for
observational cohort studies. 19 In this respect, three domains were scored concerning
selection and comparability of study cohorts, and outcome of interest was ascertained using a
scale, it was considered as high quality, 40-70% as moderate, 20 to 40% as moderate and <
This meta-analysis was conducted to obtain pooled effect size of FG score and BHPs
Accepted Article
including tT, SHBG, fT, A4, DHEAS and FAI. Chi-squared and I-squared statistics were
estimated as the measures of heterogeneity. Random effect meta-analysis was applied for
significant results (I2 greater than 50% or significant Chi2 test), otherwise fixed-effect model.
Forest plots summarize the effect sizes, by the Mantel-Haenszel method of pooled estimation,
which used variance of between studies to weight each study and the estimated pooled mean
of FG score means (mean of means). Hence, the Forest plot of FG score presents pooled
effect size (95% CI) of FG score (Figure 2) and the other forest plots show the pooled effect
size (CI) of BHPs of included studies (Figures S1-S3). To illustrate associations, we depicted
Bubble plots, which show associations between androgens and FG score by a linear
regression model (Figure 3); each study has its own contribution to estimate this linear
association, which is illustrated by bubbles, where the size of bubble is proportional to the
Publication bias was found to be significant for p-values < 0.05, which indicated significant
asymmetry in funnel plot, analogously. For significant results or an asymmetric Funnel plot,
the trim and fill method was used to identify and correct for publication bias by adding some
associations between BHPs and mean FG score across studies. First, the associations between
BHPs and FG score were estimated separately without adjusting for confounders. Then,
meta- regression models were fitted to investigate the associations between BHPs and FG
score adjusted for age and BMI, diagnostic criteria for PCOS (Rotterdam, NIH and AES),
quality assessment (moderate and high), ethnic groups (Western, Middle East, Eastern,
between BHPs and FG score in PCOS patients with clinical and/or HA signs, in which
studies with unknown phenotypes of Rotterdam criteria were excluded (Supplementary Data
5). A subgroup analysis was also done to evaluate the association between BHPs and FG
score based on the different methods of hormonal assay (Supplementary Data 6).
An overall test of all the covariates in the random effects model was assessed by t statistics
and p-values as well as 95% CIs for the regression coefficients (). For all results, p-values
<0.05 were considered significant. Statistical analysis was done using STATA software
Results
The selection process and reasons for exclusions are depicted in Figure 1. Thirty-five studies
were included in the meta-analysis, of which 19 had no comparative groups, 13 had two
comparative groups, two had three comparative groups and one study had four comparative
groups. Of the 35 studies selected, 5 had cross sectional design, 5 case-control, 6 cohort, 7
Quality assessment
Regarding quality, 15 studies were classified as high and 20 as moderate, while none were
identified as being of low or very low quality (Table 1). Therefore, all studies included in the
Characteristics of the included studies are reported in Table 1. Sixteen study groups were
Accepted Article
performed in the West region, 11 in the Middle East, 7 in the East and one in Africa. All
studies were published between 2000 and 2015, and assessed a total number of 6593
4.1, 95% CI: 1-7.3; p = 0.012) and FG scores across studies. The associations of FG score
with tT, fT, SHBG and FAI were not significant (as model 1).
Model 2 represented the relationships between BHPs and FG score adjusted for age and BMI;
there were significant positive observational relationships between DHEAS and FG score (
= 3.8, 95% CI: 1.1-6.45; p = 0.008). Other associations of FG score with tT, fT, SHBG, A4
Model 3 detected the associations adjusted for diagnostic criteria for PCOS; there were
significant positive associations between DHEAS and FG score ( = 4.1, 95% CI: 1.1-7.1; p
= 0.009). We found no significant associations for FG score with tT, fT, SHBG, A4, FAI
Model 4 fitted meta-regression models were adjusted for quality assessment; we detected
associations of FG score with A4 ( = 2.5, 95% CI: 0.5-4.4; p = 0.017) and DHEAS ( = 3.5,
95% CI: 0.3-6.7; p = 0.035). The association between tT, fT, SHBG, FAI and FG score were
not significant.
similar to results found for model 1. We observed significant positive relationships between
Accepted Article
DHEAS and FG score ( = 4.0, 95% CI: 0.8-7.2; p = 0.016). Other relationships between
Model 6, fitted BHPs and FG score were adjusted for ethnic groups; we detected a positive
significant association only between A4 and FG score ( = 2.6, 95% CI: 0.4-4.8; p = 0.023).
Finally, in model 7, associations of FG score with BHPs adjusted for the hormonal assay
score with A4 ( = 2.3, 95% CI: 0.2-4.5; p = 0.032), DHEAS ( = 4.8, 95% CI: 1.1-8.4; p =
0.012). There were no significant associations between other BHPs and FG score after
Figure 3 shows the association between BHPs and FG score, which fitted meta-regression
line for FG score by BHPs using bubble plot. All included studies assessed the FG score; as
shown in the forest plot, there was significant heterogeneity in mean FG score among studies
(I-squared= 99.4%, p< 0.001) (Figure 2). In addition, we observed significant heterogeneity
for BHPs including tT (I-squared= 99.4%, p< 0.001), fT (I-squared= 99.0 %, p< 0.001),
SHBG (I-squared= 95.6%, p< 0.001), A4 (I-squared= 98.4%, p< 0.001), DHEAS (I-squared=
95.8%, p< 0.001) and FAI (I-squared= 98.2%, p< 0.001) (see forest plots in the supplemental
figures S1-S3).
patients with clinical and/or biochemical HA showed changes in regression models including
Accepted Article
a positive significant association between SHBG and FG score in model 1 ( = 0.17, 95% CI:
model 2 ( = 3.75, 95% CI: 0.44-7.07; p = 0.029), no significant association between BHPs
and FG score in models 3 and 5, a reverse significant association between fT and FG score (
= -0.71, 95% CI: -1.29- -0.13; p = 0.022) and a positive significant association between
SHBG and FG score ( = 0.17, 95% CI: 0.01-0.34; p = 0.041) in model 4, a reverse
significant association between FG score FAI ( = -1.22, 95% CI: -2.03- -0.41; p = 0.009) in
model 6 and a positive significant association between FG score and A4 ( = 1.90, 95% CI:
0.33-3.48; p = 0.025) in model 7 (see detailed results in the supplemental Data 5).
There was no association for FG score with tT, fT, SHBG and FAI in any of the subgroups of
enzyme). However subgroup analysis based on the methods of hormonal assay showed that
DHEAS were significant (see detailed results in the supplemental Data 6).
Publication bias
We found statistically significant publication bias only for fT; the results of Egger and
Beggs tests were not concordant in this case (Table 2); therefore, trim and fill correction
were applied, and showed that non-significant publication bias was reliable (Figure S4).
Discussion
and DHEAS, whereas associations of FG score with tT, fT, SHBG and FAI were not
Accepted Article
significant. Adjustment for potential confounders including diagnostic criteria for PCOS or
Although the pathogenesis of hirsutism is complex and has not been clarified 9, it is well
known that the intensity and extent of hirsutism is influenced by several factors including the
androgen receptor activity, defects in 5- reductase production and other key enzymes. 26 In
addition, the severity of hirsutism can differ by age 27, race, ethnicity and familial
background. 9, 26 Although the main source of androgen synthesis is the ovary, increased
adrenal androgen production has been documented in 25% of PCOS women 28, as result of a
The major androgens in the serum of normal women are DHEAS, dehydroepiandrosterone
(DHEA), A4, T and dihydrotestosterone (DHT) 30; T and DHT bind to the androgen receptor
to promote changes in gene transcription, whereas DHEA, DHEAS and A4 do not bind to the
receptor gene also influence the activity of the receptor. 31 Circulating T in women is a result
of peripheral conversion of other steroids, produced in the ovaries and adrenals in equal parts.
DHT is the most active androgen, and is localised in androgen target tissues where the
androgens for growth of human sebaceous glands and hair follicles. 33 Therefore, circulating
androgen levels may not reflect local androgen concentrations at the pilosebaceous unit. In
pilosebaceous unit. 34
Accepted Article
Based on the mechanisms mentioned, hirsutism is considered to be the result of the
interaction between circulating androgen levels, local androgen levels and the sensitivity of
hair follicles to androgens. Hence, the severity of hirsutism does not exactly correlate with
circulating androgen levels. Moreover, the hair follicle response to circulating androgens
varies significantly within and between individuals; therefore some women with elevated
androgen levels do not show hirsutism, or they may have seborrhea, acne or alopecia without
clinical hirsutism. 9
In this meta-analysis, we did not observe significant associations of FG scores with tT, fT,
FAI and SHBG. Some investigators have reported that FG score is not correlated with serum
tT; they also suggest that tT may not be helpful in reflecting biochemical HA. 5, 10 Most
physicians consider testosterone and FAI in the diagnosis of biochemical HA, whereas
increased levels of DHEAS, A4 or other androgen precursors are the only findings in many
patients. 6, 35
Some previous studies have shown that compared to testosterone, pro-hormones have more
correlation with hirsutism. Rittmaster et al. (1990), in a study of PCOS patients, found that
the change in hair growth correlated with changes in A4, but not significantly with changes in
testosterone. They concluded that hair growth rates correlated more significantly with
changes in serum A4 levels. 36 Carmina et al. (1993) showed evidence for increased
As shown in the results, FG score had positive relationships with A4 and DHEAS. A4, a
are a frequent finding among women with PCOS, it is not often considered a main diagnostic
Accepted Article
criterion of biochemical HA. DHEAS, DHEA, and A4 are all precursors of testosterone. 38 In
marker to assess biochemical HA 39; A4 and DHEAS are of both adrenal and ovarian origin;
therefore, the existence any positive associations between FG score and these parameters can
represent an adrenal disorder. Indeed, an adrenal contribution has been reported in about 50%
Since hirsutism and / or BHPs are influenced by several factors such as age, BMI, methods of
hormonal assay, diagnosis criteria, study design and method of assay, we adjusted for these
factors. The prevalence of hirsutism may decline with age, which might be due to the
progressive decline in circulating androgens in women with PCOS during their reproductive
years. 27 In addition, hormonal profiles, especially DHEAS, may change with age.7 However,
this study showed that after adjusting for age, FG score was only associated with DHEAS.
Obesity, theoretically, can exert an additive or synergistic impact on the clinical and / or
effects 43. Obesity can increase aromatase enzyme action, which converts T to E2. 44 In our
study, after adjusting for BMI, FG score was only associated with DHEAS. However, the
relationship of BMI with FG score seems weak and thus remains unclarified.
The prevalence of hirsutism differs with ethnicity 45, and after adjustment for this, only A4
can influence androgen levels. As data were grouped by assay method, the between-study
heterogeneity in the subgroups was lower compared to the main analysis. However,
adjustment for methods of assay did not significantly affect associations between BHPs and
FG score. We also performed a subgroup analysis based on assay method to investigate the
association between androgens and FG score. The findings of subgroup analysis confirmed
After adjustment for quality assessment, our results did not change; the associations of FG
score with A4 and DHEAS remained, indicating that this confounder had no significant effect
on results. The association between A4 and FG score did not remain after adjusting for
The majority of studies included in this analysis used the Rotterdam criteria for diagnosing
PCOS and they may have included PCOS patients with oligomenorrhoea and cystic ovaries
relationship between BHPs and FG score in patients with HA phenotype. Although the results
association of FG score with SHBG, fT and FAI, after adjustment for important confounders
such as BMI and assay method, significant associations were observed only for FG score
and clinical hyperanrogenism; there is no reliable tool for precise assessment of these
Accepted Article
parameters, there is no single scoring system, and their assessment is prone to major inter-
observer bias. In addition, various surveys have noted a relatively high prevalence of acne in
the general population, particularly among younger women. Consequently, the degree to
which PCOS increases the risk of acne to above the prevalence in the general population is
unclear. Androgen sensitivity of the pilosebaceous unit varies, and there is poor correlation
androgen excess, other potential aetiologies of alopecia may be genetic, environmental, and
nutritional. Besides, limited studies have evaluated acne and alopecia as clinical
manifestations of HA and had different diagnostic methods. 7, 47, 48 Furthermore it has been
shown that women with hirsutism often have higher levels of A4, DHEAS and testosterone;
and lower levels of sex hormone-binding globulin (SHBG) than patients with alopecia and
acne. 12, 14, 49 Hence, in this meta-analysis, we considered hirsutism as the main clinical
indicator of HA.
The strengths of this meta-analysis include adjusting for potential confounders and
performing subgroup analysis based on assay methods and diagnostic criteria for PCOS. The
study, of course has its limitations that should be considered when interpreting the findings.
The majority of studies did not report the ethnicity of patients; however, we divided countries
into ethnicity groups (Western, Middle East, Eastern, and African) and adjusted for ethnicity
as a confounder. Most studies in this meta-analysis recruited participants from settings such
as hospitals and clinics, which may represent a population with more severe manifestations of
PCOS. In this study, we could not investigate the associations of FG score with BHPs,
studies.
analysis, which may have resulted from clinical heterogeneity related to variability in PCOS
Accepted Article
diagnosis, race, ethnicity and/or interpreting hirsutism. Additionally, FG score is a visual and
significant heterogeneity in mean BHPs between studies, which was predictable because of
variability in age, BMI, SHBG concentration and assay methods. 9 Moreover, we used
random effect analysis to deal with the problem. To minimize selection bias, we selected
studies based on the eligibility criteria that had been set just before the study. We also
performed a hand search on the references lists of all selected studies to prevent missing
studies.
In conclusion, the current study demonstrated positive associations of some BHPs such as A4
and DHEAS with FG score. In most analyses, DHEAS was associated with FG score. These
relationships were not causal; thus our findings confirmed a relationship between clinical and
biochemical HA. Since DHEAS and A4 levels were strongly associated with hirsutism, these
markers should be considered valuable to assess HA in patients with PCOS. The data suggest
that peripheral androgen metabolism is one of the major determinants of hirsutism in PCOS
patients.
Declaration of interest
The authors would like to thank Dr. Maryam Tohidi for valuable suggestions on hormonal
Accepted Article
assay methods and Ms. Niloofar Shiva for critical editing of English grammar and syntax of
the manuscript. The authors also thank the Research Institute for Endocrine Sciences, Shahid
Beheshti University of Medical Sciences for approval of this project and its funding as the
research project.
Supporting Information
S4 Figure. Funnel plots of publication bias and its corrections (jpg, 89.4 KB)
between BHP and FG score in patients with clinical and/or HA findings (docx, 76.1 KB).
Supplementary Data 6. Subgroup analysis for assessing the association between BHP and
FIGURE CAPTIONS
Fig 3. Bubble plots fitted meta-regression line for associations between biochemical
hyperandrogenism parameters and FG score. Bubble plots present the following data in order
from the top and left to right: association between SHBG and FG score, association between
1 Azziz, R., Woods, K.S., Reyna, R. et al. (2004) The prevalence and features of the
2 Azziz, R., Carmina, E., Dewailly, D. et al. (2006) Criteria for defining polycystic
3 Coskun, A., Ercan, O., Arikan, D.C. et al. (2011) Modified FerrimanGallwey
hirsutism score and androgen levels in Turkish women. European Journal of Obstetrics &
4 Buyalos, R.P., Pekonen, F., Halme, J.K. et al. (1995) The relationship between
binding protein-1 in the polycystic ovarian syndrome. American journal of obstetrics and
5 Panidis, D., Tziomalos, K., Papadakis, E. et al. (2013) The clinical significance and
6 Cibula, D., Hill, M. & Starka, L. (2000) The best correlation of the new index of
hyperandrogenism with the grade of increased body hair. European Journal of Endocrinology
143, 405-408.
Androgen Excess and PCOS Society criteria for the polycystic ovary syndrome: the complete
Accepted Article
task force report. Fertility and sterility 91, 456-488.
8 Goodman, N., Bledsoe, M., Cobin, R. et al. (2001) American Association of Clinical
Endocrinologists medical guidelines for the clinical practice for the diagnosis and treatment
diagnosis and management of hirsutism: a consensus statement by the Androgen Excess and
differentiates hirsute from non-hirsute Sicilian women with polycystic ovary syndrome.
11 Legro, R.S., Schlaff, W.D., Diamond, M.P. et al. (2010) Total testosterone assays in
women with polycystic ovary syndrome: precision and correlation with hirsutism. The
12 Guzel, A.I., Kuyumcuolu, U. & elik, Y. (2012) Factors affecting the degree of
hirsutism in patients with polycystic ovary syndrome. Archives of gynecology and obstetrics
285, 767-770.
13 Sales, M.F., Ster, M.O., Cndido, A.L. et al. (2015) Ferriman-Gallwey Score
correlates with obesity and insulin levels in Polycystic Ovary Syndrome: an observational
biochemical markers and clinical signs of hyperandrogenism in women with polycystic ovary
syndrome (PCOS) and women with non-classic congenital adrenal hyperplasia (NCAH).
15 Moher, D., Liberati, A., Tetzlaff, J. et al. (2009) Preferred reporting items for
Accepted Article
systematic reviews and meta-analyses: the PRISMA statement. Annals of internal medicine
151, 264-269.
16 Welt, C., Gudmundsson, J., Arason, G. et al. (2006) Characterizing discrete subsets of
polycystic ovary syndrome as defined by the Rotterdam criteria: the impact of weight on
phenotype and metabolic features. The Journal of Clinical Endocrinology & Metabolism 91,
4842-4848.
17 Moher, D., Hopewell, S., Schulz, K.F. et al. (2010) CONSORT 2010 explanation and
elaboration: updated guidelines for reporting parallel group randomised trials. Journal of
18 Von Elm, E., Altman, D.G., Egger, M. et al. (2007) The Strengthening the Reporting
19 Quigley, J., Thompson, J., Halfpenny, N. et al. (2014) Critical appraisal of nonn-
assessment 4, 4.
20 Higgins, J.P., Thompson, S.G., Deeks, J.J. et al. (2003) Measuring inconsistency in
23 Egger, M., Smith, G.D., Schneider, M. et al. (1997) Bias in meta-analysis detected by
testing and adjusting for publication bias in meta-analyses. Fort Collins, CO: Colorado State
Accepted Article
University.
25 Duval, S. & Tweedie, R. (2000) Trim and fill: a simple funnelplotbased method of
testing and adjusting for publication bias in metaanalysis. Biometrics 56, 455-463.
26 Zhang, H.Y., Guo, C.X., Zhu, F.F. et al. (2013) Clinical characteristics, metabolic
features, and phenotype of Chinese women with polycystic ovary syndrome: a large-scale
27 Panidis, D., Tziomalos, K., Macut, D. et al. (2012) Cross-sectional analysis of the
effects of age on the hormonal, metabolic, and ultrasonographic features and the prevalence
of the different phenotypes of polycystic ovary syndrome. Fertility and sterility 97, 494-500.
29 Moran, C. & Azziz, R. (2001) The role of the adrenal cortex in polycystic ovary
30 Burger, H.G. (2002) Androgen production in women. Fertility and sterility 77, 3-5.
31 Calvo, R.M., Asuncin, M., Sancho, J. et al. (2000) The Role of the CAG Repeat
in the Pathogenesis of Hirsutism 1. The Journal of Clinical Endocrinology & Metabolism 85,
1735-1740.
33 Thiboutot, D., Gilliland, K., Cong, Z. et al. (2003) Human skin is a steroidogenic
tissue: steroidogenic enzymes and cofactors are expressed in epidermis, normal sebocytes,
and an immortalized sebocyte cell line (SEB-1). Journal of Investigative Dermatology 120,
34 Chen, W.-C. & Zouboulis, C.C. (2009) Hormones and the pilosebaceous unit.
Accepted Article
Dermato-endocrinology 1, 81-86.
35 Azziz, R., Sanchez, L., Knochenhauer, E. et al. (2004) Androgen excess in women:
experience with over 1000 consecutive patients. The Journal of Clinical Endocrinology &
36 Rittmaster, R.S. & Thompson, D.L. (1990) Effect of leuprolide and dexamethasone
on hair growth and hormone levels in hirsute women: the relative importance of the ovary
and the adrenal in the pathogenesis of hirsutism. The Journal of Clinical Endocrinology &
some normoandrogenic women with acne. The Journal of Clinical Endocrinology &
39 Georgopoulos, N.A., Papadakis, E., Armeni, A.K. et al. (2014) Elevated serum
androstenedione is associated with a more severe phenotype in women with polycystic ovary
Enzyme Defects and Exaggerated Adrenarche*. The Journal of Clinical Endocrinology &
presentations of polycystic ovary syndrome among obese and nonobese women. Fertility and
Accepted Article
sterility 92, 1960-1965.
production and clearance in hirsute and obese women. Journal of steroid biochemistry 19,
607-614.
43 Samojlik, E., Kirschner, M., Silber, D. et al. (1984) Elevated production and
metabolic clearance rates of androgens in morbidly obese women. The Journal of Clinical
45 Tehrani, F.R., Simbar, M., Tohidi, M. et al. (2011) The prevalence of polycystic
47 Rosenfield, R. & Lucky, A. (1993) Acne, hirsutism, and alopecia in adolescent girls.
48 Lucky, A.W., McGuire, J., Rosenfield, R.L., Lucky, P.A. et al. (1983) Plasma
androgens in women with acne vulgaris. Journal of Investigative Dermatology 81, 70-74.
dependence of hirsutism, acne, and alopecia in women: retrospective analysis of 228 patients
Accepted Article
investigated for hyperandrogenism. Medicine 88, 32-45.
50 Elter, K., Imir, G. & Durmusoglu, F. (2002) Clinical, endocrine and metabolic effects
polycystic ovarian syndrome: a randomized controlled study. Human Reproduction 17, 1729-
1737.
51 Mastorakos, G., Koliopoulos, C. & Creatsas, G. (2002) Androgen and lipid profiles in
adolescents with polycystic ovary syndrome who were treated with two forms of combined
52 Guido, M., Romualdi, D., Giuliani, M. et al. (2004) Drospirenone for the treatment of
hirsute women with polycystic ovary syndrome: a clinical, endocrinological, metabolic pilot
53 Hahn, S., Janssen, O.E., Tan, S. et al. (2005) Clinical and psychological correlates of
860.
54 Palep-Singh, M., Mook, K., Barth, J. et al. (2004) An observational study of Yasmin
in the management of women with polycystic ovary syndrome. Journal of Family Planning
drospirenone in the treatment of women with polycystic ovary syndrome. The European
biochemical characteristics and psychiatric distress in young women with polycystic ovary
58 Wu, J., Zhu, Y., Jiang, Y. & Cao, Y. (2008) Effects of metformin and ethinyl
59 zdemir, S., Grkemli, H., Gezgin, K. et al. (2008) Clinical and metabolic effects of
polycystic ovary syndrome. International Journal of Gynecology & Obstetrics 103, 44-49.
61 Mahmood, M., El-Kattan, E.A., EL-Aal, H.A. et al. (2009). Evaluation of the Clinical
and Biochemical Effects of Medical Therapy in Women with Polycystic Ovary Syndrome.
62 Kriplani, A., Periyasamy, A.J., Agarwal, N. et al. (2010) Effect of oral contraceptive
containing ethinyl estradiol combined with drospirenone vs. desogestrel on clinical and
biochemical parameters in patients with polycystic ovary syndrome. Contraception 82, 139-
146.
63 Taheripanah, R., Sepahvandi, M., Entezari, A. et al. (2010) Evaluation of serum PSA
after cyproterone compound treatment compared with oral contraceptive pill in hirsute
polycystic ovary syndrome patients. Middle East Fertility Society Journal 15, 159-162.
oestroprogestins in patients affected by acne and polycystic ovary syndrome: clinical and
Accepted Article
instrumental evaluation. Journal of the European Academy of Dermatology and Venereology
26, 1364-1371.
66 Liang, S.-J., Hsu, C.-S., Tzeng, C.-R. et al. (2011) Clinical and biochemical
presentation of polycystic ovary syndrome in women between the ages of 20 and 40. Human
67 Misichronis, G., Georgopoulos, N., Marioli, D. et al. (2012) The influence of obesity
68 Alanbay, I., Ercan, C.M., Sakinci, M. et al. (2012) A macrophage activation marker
chitotriosidase in women with PCOS: does low-grade chronic inflammation in PCOS relate
69 Ilie, I.R., Marian, I., Mocan, T. et al. (2012) Ethinylestradiol 30g-drospirenone and
cyproterone acetate treatment on metabolic syndrome, fat distribution and carotid intima
71 Bhattacharya, S.M. & Jha, A. (2012) Comparative study of the therapeutic effects of
patients with polycystic ovary syndrome. Fertility and sterility 98, 1053-1059.
73 Aydin, K., Cinar, N., Aksoy, D.Y. et al. (2013) Body composition in lean women with
74 Quinn, M., Shinkai, K., Pasch, L. et al. (2014) Prevalence of androgenic alopecia in
patients with polycystic ovary syndrome and characterization of associated clinical and
75 Kahraman, K., kr, Y.E., Atabekolu, C.S. et al. (2014) Comparison of two oral
patients with polycystic ovary syndrome: a randomized clinical trial. Archives of gynecology
76 Yildizhan, R., Gokce, A.I., Yildizhan, B. et al. (2015) Comparison of the effects of
PCOS
Elter, 2002 50 Turkey RCT 20 Rotterdam 23.456.07 21.831.40 12.065.25 tTa, SHBGa, fTb, A4b, High
DHEASa
Elter, 2002 50 Turkey RCT 20 Rotterdam 24.906.62 22.742.66 9.475.48 tTa, SHBGa, fTb, A4b, High
DHEASa
Mastorakos, 2002 Italy RCT 14 NIH 17.530.51 25.501.79 15.711.63 tTb, SHBGb, fTb, A4b, Moderate
51
DHEASb
Mastorakos, 2002 Italy RCT 14 NIH 17.460.43 24.841.09 16.781.15 tTb, SHBGb, fTb, A4b, Moderate
51
DHEASb
Guido, 2004 52 Italy Clinical trial 15 Rotterdam 25.303.51 24.685.95 16.205.81 tTb, SHBGb, A4b, DHEASb High
Hahn, 2005 53 Germany Case-control 120 NIH 24.005.40 31.009.30 9.20+5.80 tTa Moderate
Palep-singh, 2005 UK Clinical trial 17 Rotterdam 26.405.30 29.003.70 18.502.74 tTd, SHBGd Moderate
54
Amato, 2006 10 Italy R-Cohort 75 Rotterdam 24.725.83 31.197.36 17.949.27 tTc, fTc, A4a, DHEASa Moderate
Amato, 2006 10 Italy R-Cohort 55 Rotterdam 24.354.28 28.706.16 6.502.02 tTc, fTc, A4a, DHEASa Moderate
Pehlivanov, 2007 55 Bulgaria Clinical trial 20 Rotterdam 23.004.10 26.450.65 11.600.90 tTc, SHBGc, DHEASc Moderate
Chae, 2008 56 Korea Case-control 87 Rotterdam 25.405.70 22.805.10 8.703.10 tTb, SHBGb, fTb, DHEASb High
Chae, 2008 56 Korea Case-control 52 Rotterdam 25.704.90 20.904.40 3.103.00 tTb, SHBGb, fTb, DHEASb High
Chae, 2008 56 Korea Case-control 23 Rotterdam 23.406.90 23.104.80 10.503.00 tTb, SHBGb, fTb, DHEASb High
Adali, 2008 57 Turkey P-Cohort 42 Rotterdam 23.543.13 28.424.30 12.423.40 tTa, DHEASa High
Wu, 2008 58
China
China
RCT
RCT
7
12
Rotterdam
Rotterdam
25.004.30
26.104.60
25.300.80
21.401.60
8.301.00
7.801.30
tTa
tTa
High
High
Wu, 2008 58 China RCT 6 Rotterdam 24.502.40 25.201.00 8.201.40 tTa High
Wu, 2008 58 China RCT 10 Rotterdam 25.804.00 21.601.40 7.802.10 tTa High
Ozdemir, 2008 59 Turkey RCT 32 Rotterdam 22.703.80 22.703.80 10.403.60 tTa, SHBGa, fTa Moderate
Liou, 2009 41 Taiwan R-Cohort 115 Rotterdam 27.605.50 31.204.40 3.103.80 tTb High
Liou, 2009 41 Taiwan R-Cohort 180 Rotterdam 26.205.20 20.302.10 3.103.30 tTb High
Landay, 2009 60 USA Cross-sectional 749 NIH 27.507.40 33.609.30 8.234.92 tTb, fTb, DHEASb High
Mahmood, 2009 61 Egypt RCT 21 AES 26.052.44 29.422.02 9.711.85 fTa, DHEASa Moderate
Kriplani, 2010 62 India RCT 30 Rotterdam 22.504.70 22.702.30 12.604.50 tTa, SHBGa High
Taheripanah, 2010 Iran RCT 30 Rotterdam 22.900.50 21.102.06 10.782.40 fTc, DHEASc High
63
Taheripanah, 2010
63 Iran RCT 30 Rotterdam 23.900.6 21.702.76 11.502.30 High
fTc, DHEASc
1
Colonna, 2012 64 Italy RCT 32 Rotterdam 25.604.40 23.353.60 16.212.71 tTb, SHBGb, A4b, DHEASb Moderate
Colonna, 2012 64 Italy RCT 27 Rotterdam 25.403.40 23.273.63 16.092.54 tTb, SHBGb, A4b, DHEASb Moderate
Naka, 2011 65 Greece Clinical trial 13 Rotterdam 20.903.70 23.004.00 8.502.20 tTa, SHBGa Moderate
Guzel, 2011 12 Turkey Case-control 224 Rotterdam 24.194.44 26.674.22 12.523.28 tTa Moderate
Guzel, 2011 12 Turkey Case-control 80 Rotterdam 23.104.32 25.233.09 4.311.79 tTa Moderate
Coskun, 2011 3
Taiwan
Turkey
R-Cohort
P-Cohort
453
43
Rotterdam
Rotterdam
27.904.70
26.050.76
25.706.60
29.270.84
3.803.80
11.204.20
tTb, DHEASd
tTa, DHEASa
High
Moderate
Chanukvadze, 2012 Georgia P-Cohort 24 AES 17.303.50 26.306.80 16.405.60 tTc, SHBGc, fTc, DHEASc Moderate
14
Misichronis, 2012 Greece Case-control 1087 Rotterdam 23.405.60 26.707.00 8.504.70 tTb, SHBGb, A4b, DHEASb Moderate
67
Alanbay, 2012 68 Turkey Cross-sectional 25 Rotterdam 27.304.00 23.101.00 10.103.90 tTc, fTc High
Alanbay, 2012 68 Turkey Cross-sectional 54 Rotterdam 25.303.70 27.801.90 10.203.20 tTc, fTc High
Iiie, 2012 69 Romania Clinical trial 25 AES 22.760.83 28.441.24 9.721.06 tTc, SHBGc High
Karabulut, 2012 70 Turkey Clinical trial 6 Rotterdam 22.406.30 28.554.24 12.003.40 tTd, DHEASd Moderate
Bhattcharia, 2012 India RCT 58 Rotterdam 22.244.47 25.414.49 5.554.51 tTa, SHBGc Moderate
71
Bhattcharia, 2012 India RCT 56 Rotterdam 22.324.17 26.413.81 6.845.17 tTa, SHBGc Moderate
71
Bhattcharia, 2012 India RCT 57 Rotterdam 22.334.76 26.474.65 6.145.15 tTa, SHBGc Moderate
71
Romualdi, 2013 72 Italy RCT 13 Rotterdam 22.923.80 22.133.34 11.625.66 tTa, SHBGa, A4a, DHEASa Moderate
Romualdi, 2013 72 Italy RCT 13 Rotterdam 21.922.43 22.652.75 11.425.88 tTa, SHBGa, A4a, DHEASa Moderate
Aydin, 2013 73 Turkey Clinical trial 28 Rotterdam 21.404.20 21.803.40 9.258.03 tTd, SHBGd, A4d, DHEASd Moderate
Panidis, 2013 5 Greece Cross-sectional 780 Rotterdam 23.805.60 27.307.00 11.402.50 tTb, SHBGb, A4b, DHEASb High
Panidis, 2013 5 Greece Cross-sectional 517 Rotterdam 25.105.90 25.906.80 3.402.40 tTb, SHBGb, A4b, DHEASb High
Zhang, 2013 26 China Case-control 719 Rotterdam 25.543.28 23.673.57 7.304.10 tTa Moderate
Quinn, 2014 74
USA
USA
Cross-sectional
Cross-sectional
56
198
Rotterdam
Rotterdam
28.805.67
27.806.10
30.908.45
30.108.17
9.276.43
8.136.04
tTb, SHBGb, A4b, DHEASb
High
Kahraman, 2014 75 Turkey RCT 19 AES 22.303.20 23.203.30 15.807.30 tTb, SHBGb, fTb, A4b, High
DHEASd
Kahraman, 2014 75 Turkey RCT 20 AES 22.402.70 23.205.50 15.307.30 tTb, SHBGb, fTb, A4b, High
DHEASd
Yildizhan, 2015 76 Turkey RCT 60 Rotterdam 25.362.91 24.821.08 9.984.86 tTa, SHBGa, DHEASb Moderate
Yildizhan, 2015 76 Turkey RCT 60 Rotterdam 24.823.20 23.563.32 5.563.44 tTa, SHBGa, DHEASb Moderate
Sales, 2015 13 Brazil Cross-sectional 50 Rotterdam 31.005.00 28.905.30 11.007.00 tTa Moderate
Abbreviations: PCOS, polycystic ovary syndrome; SD, standard deviation; BMI, body mass index; RCT, randomised controlled trial; P-cohort, prospective cohort; R-Cohort, retrospective
cohort; NIH, National Institutes of Health; AES, Androgen Excess Society; BHP, biochemical hyperandrogenism parameter; tT, total testosterone; fT, free testosterone; SHBG, sex hormone
bonding globulin; A4, androstenedione; DHEAS, dehydroepiandrosterone; QA, quality assessment.
(p-value)
Abbreviations: FG score, Feriman-Gallwey; tT, total testosterone; fT, free testosterone; SHBG, sex hormone bonding
globulin; A4, androstenedione; DHEAS, dehydroepiandrosterone sulfate and FAI, free androgen index.
$
Degree of freedom for chi squared test = number of included studies - 1
**Statistical significance level P < 0.05
* Overall mean equal to zero z- test.
&
Egger test of publication bias to assess small-study effects.
$
Beggs test: an adjusted rank correlation test to detect the association between estimated effect size and its variance the
based on Kendalls tau.
Table 3. Meta-regression results for univariate and multiple (adjusted effect) models assessing the observational
relationships between BHPs and FG score.
(95% CI for ) (95% CI for ) (95% CI for ) (95% CI for ) (95% CI for ) (95% CI for ) (95% CI for )
Accepted Article
tT mean 1.2 -0.3 0.8 1 1.2 -0.5 1.4