| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
CLINICAL REVIEW |
Knowledge and Encounter Research Unit (M.C., B.A.S., D.N.F., D.M.K., M.L.L., R.J.M., M.B.E., P.J.E., V.M.M.) and Division of Endocrinology, Diabetes, Metabolism, and Nutrition (M.C., B.A.S., V.M.M.), Department of Medicine, and Mayo Clinic Libraries (P.J.E.), Mayo Clinic College of Medicine, Rochester, Minnesota 55905; and State University of New York at Geneseo (M.L.L.), Geneseo, New York 14454
Address all correspondence and requests for reprints to: Victor M. Montori, M.D., M.Sc., Mayo Clinic, W18A, 200 First Street SW, Rochester, Minnesota 55905. E-mail: montori.victor{at}mayo.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Objective: Our objective was to conduct a systematic review and metaanalyses of randomized controlled trials of metformin or TZDs for the treatment of hirsutism.
Data Sources: We searched the following databases: MEDLINE, EMBASE, and Cochrane CENTRAL (up to May 2006). Review of reference lists and contact with hirsutism experts further identified candidate trials.
Study Selection: Reviewers working independently and in duplicate, with acceptable chance-adjusted agreement (
= 0.72), determined trial eligibility. Eligible trials randomly assigned women with hirsutism to at least 6 months of insulin sensitizers or control and measured hirsutism outcomes.
Data Extraction: Reviewers working independently and in duplicate determined the methodological quality of trials and collected data on patient characteristics, interventions, and outcomes.
Data Synthesis: Of 348 candidate studies, 16 trials (22 comparisons) were eligible. The methodological quality of these trials was low. Random-effects metaanalyses showed a small decrease in Ferriman-Gallwey scores in women treated with insulin sensitizers compared with placebo [pooled weighted mean difference (WMD) of –1.5; 95% confidence interval (CI), –2.8 to –0.2; inconsistency (I2) = 75%]. There was no significant difference between insulin sensitizers and oral contraceptives (WMD of –0.5; CI, –5.0, 3.9; I2 = 79%). Metformin was inferior to both spironolactone (WMD of 1.3; CI, 0.03, 2.6) and flutamide (WMD of 5.0; CI, 3.0, 7.0; I2 = 0%).
Conclusions: Imprecise and inconsistent evidence of low to very low quality suggests that insulin sensitizers provide limited or no important benefit for women with hirsutism.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Some effects of insulin sensitizers may favorably impact patients with PCOS and hyperandrogenism. Metformin decreases hepatic glucose production (3) and lowers insulin levels with improved insulin sensitivity. TZDs improve the action of insulin in the liver, skeletal muscle, and adipose tissue (4). By reducing hyperinsulinemia, both metformin and TZDs may also reduce adrenal and ovarian androgen biosynthesis, raise levels of SHBG, and improve gonadotropin secretion (5). The net effect could plausibly cause improvement of hirsutism.
Systematic reviews can enhance the ability of clinicians and patients to make sense of disparate and small clinical trials. These trials tend to yield imprecise answers and apply to only a narrow set of patients. Pooling such trials together, if the trials are consistent, can yield precise estimates that are more broadly applicable (6).
To clarify the role that insulin sensitizers play in the treatment of hirsutism, we conducted a systematic review and metaanalyses of randomized trials of metformin or TZDs, alone or in combination with other therapies [oral contraceptive pills (OCPs) or antiandrogens], for the treatment of hirsutism. The Endocrine Society Task Force on Hirsutism, assembled to produce clinical practice guidelines, commissioned these metaanalyses to support the formulation of evidence-based recommendations.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Eligibility criteria
Eligible studies were fully published randomized trials that enrolled women of age 12 yr and older with varying degrees of hirsutism and treated them for at least 6 months with metformin or TZDs, alone or in combination with OCPs or antiandrogens, or with placebo or active control (OCPs or antiandrogens). For this review, OCPs included single pill preparations, some of which contained a low dose (2 mg) of cyproterone acetate. We considered higher doses of cyproterone acetate separately as antiandrogens.
Eligible trials had to measure hirsutism as an outcome. Patients, clinicians, or researchers could use subjective methods [the Ferriman-Gallwey score (FGS), typically], or researchers could use objective methods (e.g. measurement of hair diameter, length, or rate of growth) to measure hirsutism.
We excluded studies enrolling patients with hirsutism due to etiologies other than PCOS, idiopathic hirsutism, or presumed late-onset congenital adrenal hyperplasia. Trials in which patients received GnRH, clomiphene, or glucocorticoids were also ineligible. Language of publication was not an eligibility criterion.
Study identification
An expert reference librarian (P.J.E.) designed and conducted the electronic search strategy with input from an endocrinologist (V.M.M.) with expertise in conducting systematic reviews. The systematic search included MEDLINE, EMBASE, and Cochrane CENTRAL electronic databases from their inception through May 2006. The detailed strategy is available upon request. To identify additional candidate studies, we reviewed the reference lists of the eligible primary studies, narrative reviews, and systematic reviews, and we queried the expert members of the commissioning task force.
Study selection
Working independently and in duplicate, reviewers (M.C., B.A.S., D.M.K., and V.M.M.) screened all abstracts and titles. After obtaining all potentially eligible studies in full text, these reviewers, again working independently and in duplicate, determined eligibility with acceptable chance-adjusted interobserver agreement ( statistic = 0.72). Disagreements were resolved by consensus or arbitration (by V.M.M.).
Data collection
Using a standardized data extraction form and working in duplicate, M.C., B.A.S., and V.M.M. abstracted the following descriptive data from every study: description of trial participants (age and body mass index or body habitus, e.g. lean or obese), etiology of hirsutism, presence of hirsutism as an explicit criterion for enrollment, degree of hirsutism at baseline, prior treatment that may have affected hirsutism, and characteristics of treatment and control interventions (medication type, dose, frequency, route, and duration).
We also abstracted specific hirsutism outcomes from every study. We collected end-of-study hirsutism scores or, when these were not reported, change-from-baseline scores. Outcomes collected were at the longest point of complete follow-up while patients were still exposed to the interventions. When authors seemed to have collected data but failed to report these, or reported data only with a statement of change or significance (e.g."no change" or "not significant"), we contacted the authors to obtain full details.
Quality assessment
To ascertain the reported methodological quality of eligible trials, pairs of reviewers (M.C., B.A.S., D.M.K., M.L.L., R.J.M., and V.M.M.), working independently and with adequate reliability, determined the adequacy of allocation concealment ( = 0.56) and blinding of patients ( = 0.69), healthcare providers ( = 0.63), and outcome assessors ( = 0.88). The proportion of participants randomized for whom the trial authors did not report hirsutism outcomes (i.e. the extent of loss to follow-up) was also noted.
Statistical analyses
Metaanalyses Due to a small number of randomized trials reporting patient self-assessment and objective evaluation of hirsutism, we focused our quantitative analyses on subjective assessment by a healthcare professional. We determined the pooled weighted mean difference between treatment and control interventions and the associated 95% confidence interval (CI) using a DerSimonian and Laird (8) random-effects metaanalysis as implemented in RevMan 4.2 (Cochrane Collaboration) (9), and in StatsDirect statistical software (10). We quantified inconsistency using the I2 statistic, which describes the proportion of variance across studies not due to chance, thus describing the extent of true inconsistency in results across trials: I2 less than 25% and I2 more than 50% reflect small and large inconsistency, respectively (11).
Subgroup analyses The following were a priori hypotheses to explain potential heterogeneity across eligible trials: study quality (blinding status and loss to follow-up), patient population characteristics (age, hirsutism etiology, and degree of hirsutism at baseline), and treatment and control interventions (type, dose, and length of treatment). To explore these hypotheses, we estimated the difference in treatment effects between subgroups or treatment-subgroup interactions (12).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
The search identified 348 candidate studies for review, of which 16 trials were confirmed eligible (Fig. 1
) contributing to 22 comparisons. One trial reported the results separately by patient weight groups (lean, overweight, or obese) (13), and two trials evaluated multiple interventions (14, 15); we abstracted each intervention of interest as a separate comparison while avoiding duplicate counting of the control group within the same metaanalysis.
|
Table 1 summarizes the methodological quality of the eligible trials. Several trials lacked details regarding the use of methodological features that protect against bias, whereas in other trials, it was clear that the techniques were not used. Four trials (16, 17, 18, 19) reported procedures consistent with allocation concealment that protects the randomization, and only two (16, 20) clearly reported blinding of the outcome assessors (i.e. the researchers estimating each patient’s degree of hirsutism). Loss to follow-up was, in some cases, substantial with seven trials (13, 17, 18, 20, 21, 22, 25) losing more than 20% of the randomized participants.
|
Table 2 describes the characteristics of the eligible trials. Enrolled participants were typically young women with median (calculated as median of the mean age of the patients in each trial) age 27 yr with PCOS (defined either by National Institutes of Health or Rotterdam criteria) and hirsutism quantified using FGS in all studies. Baseline mean hirsutism scores varied widely across the trials (FGS range, 6.9–22.8) with four trials (15, 16, 18, 20) requiring a minimal degree of hirsutism for enrollment. Nine studies (13, 14, 16, 17, 19, 22, 23, 24, 25) contributing to 12 comparisons included obese and overweight women (body mass index > 25 kg/m2), four studies (13, 15, 26, 27) contributing to six comparisons enrolled lean women (body mass index < 25 kg/m2), and four studies (18, 20, 21, 28) contributing to four comparisons did not have a weight eligibility criterion. Included trials evaluated the efficacy of either 1000–2000 mg metformin daily or TZDs (600 mg troglitazone daily or 4 mg rosiglitazone daily) for 6–12 months. The eligible trials involved two antiandrogens (50 mg spironolactone daily or 62.5–500 mg flutamide daily) and three OCPs (35 µg ethinyl estradiol with 250 µg norgestimate daily, 35 µg ethinyl estradiol with 2 mg cyproterone acetate daily, or 30 µg ethinyl estradiol with 300 µg drosperinone daily).
|
Figure 2 summarizes the results of the metaanalyses. The supplemental data (published on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org) include each of the metaanalytic plots for the analyses listed.
|
Subgroup analyses showed that one 11-month trial found troglitazone was better than placebo (21, 29) (–2.4; CI, –3.8 to –1.0) whereas metaanalysis of the eight comparisons of metformin vs. placebo (13, 14, 16, 17, 20, 23) found no significant effect (–1.4; CI, –2.8, 0.1; I2 = 77%). However, we found no significant interaction between sensitizer type (troglitazone vs. metformin) and the magnitude of the treatment effect (Pinteraction = 0.31). Further planned analyses found no subgroup-treatment interactions across subgroups defined by blinding status, extent of loss to follow-up (less or more than 20%), degree of hirsutism (less or more than the median FGS of 13.5), length of treatment (
6 months vs. >6 months), or type of reported outcome (i.e. end-of-study FGS or change in FGS from baseline). A post hoc analysis revealed that studies enrolling participants described as overweight or obese (or who had a mean body mass index that was 2 SD above 25 kg/m2) or enrolling a mixed population of lean and obese women (WMD, –2.3; CI, –3.4 to –1.2) had a significantly different result (Pinteraction = 0.02) than the study enrolling lean women exclusively (1.1; CI, –1.6, –3.8; supplemental Fig. S1).
Insulin sensitizers vs. OCPs Five trials compared insulin sensitizers with OCPs. Metaanalysis of these comparisons found no significant difference in FGS between these treatments (–0.5; CI, –5.0, 3.9) but with large inconsistency across these studies (I2 = 79%; supplemental Fig. S2).
One subgroup analysis, length of therapy, yielded a significant subgroup-treatment interaction. Trials assessing interventions delivered for 6 months or less (1.7; CI, –0.6, 4.0; I2 = 0%) had a significantly lower overall treatment effect (Pinteraction = 0.01) than the 12-month trial of metformin (–8.5; CI, –12.6 to –4.4) (18). All other planned subgroup analyses were noncontributory.
Metformin vs. antiandrogens Metaanalysis of the three trials of metformin vs. antiandrogens (there were no trials of TZDs vs. antiandrogens) found a significant difference in favor of antiandrogens (–3.7; CI, –6.8 to –0.6) but with large inconsistency across trials (I2 = 80%). A significant type-treatment interaction partly explains this inconsistency (Pinteraction = 0.002): metformin was much less effective compared with flutamide (14, 15) (5.0; CI, 3.0, 7.0; I2 = 0%;) than to spironolactone (28) (1.3; CI, 0.03, 2.6; supplemental Fig. S3. Other planned subgroup analyses were noncontributory.
Metformin in combination with other treatments One 9-month trial of metformin and flutamide vs. OCPs (26) found no significant difference in hirsutism between the two groups (–0.5; CI, –2.7, 1.7; supplemental Fig. S4a). Metaanalysis of two comparisons of metformin and flutamide vs. flutamide alone (14, 15) also found no difference between groups (0.9; CI, –0.4, 2.2; I2 = 0%; supplemental Fig. S4b). Metaanalysis of two comparisons of metformin and flutamide vs. metformin alone (14, 15) found that the combination led to significantly lower hirsutism scores than metformin alone (–4.6; CI, –7.9 to –1.3; I2 = 40%; supplemental Fig. S4c). We found no significant subgroup-treatment interactions to account for the moderate inconsistency.
Summary of studies reporting patient self-assessments of hirsutism
Three randomized controlled trials (RCTs) reported results of patient self-assessment of hirsutism along with FGS and laboratory measurements (18, 20, 21, 29).
Metformin vs. placebo In a crossover trial, Kelly and Gordon (20) asked 10 women to assess their hirsutism scores before and after 6 months of metformin or placebo. Compared with placebo, metformin significantly improved self-assessed hirsutism scores (P = 0.014), matching the reported improvements in clinician-ascertained FGS and laboratory measurements (e.g. hair growth velocity).
TZDs vs. placebo Guyatt et al. (29) validated a self-administered health-related quality of life questionnaire (PCOSQ) in women with PCOS. The PCOSQ included 26 questions addressing five areas of concern including body hair in women treated with troglitazone or placebo for 44 wk. The changes in the self-reported hair growth domain of the PCOSQ significantly correlated (r = –0.22, P < 0.01) with the clinician-assessed FGS [reported in the study by Azziz et al. (21)].
Metformin vs. OCPs In the study by Harborne et al. (18), investigators asked patients to assess their hirsutism status using a visual analog scale and with a questionnaire about the change in hair quality and need for cosmetic removal of excess hair. After 12 months, the patients taking metformin scored their hirsutism lower than the OCP group (P = 0.01) matching the improvement in clinician-assessed FGS in the same group; there was no difference in hair quality between the two groups at 12 months.
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Our systematic review and metaanalyses suggest that overall insulin sensitizers provide a small benefit for women with hirsutism when compared against placebo and no significant benefit when compared with OCPs, and antiandrogens. Subgroup analyses showed that insulin sensitizers appear to be more effective than placebo when used in women who are overweight or obese, and metformin was more effective than OCPs in one small study that treated women for more than 6 months. The addition of metformin to either antiandrogens or OCPs provided no significant advantage.
Weaknesses and strengths
Readers should interpret the results of these metaanalyses with caution for several reasons. We could not conduct metaanalyses of patient-reported hirsutism outcomes, the outcome of most relevance, because only three studies measured these outcomes. When measured and reported, however, these outcomes appear to correlate with clinician-assessed FGS. Publication bias could affect our review despite our extensive search procedure, and we did not contact drug manufacturers. We found only a small number of patients studied and trials published. Our metaanalyses include published RCTs only; potential unpublished studies may have been missed. Therefore, overestimation of treatment effect due to publication bias cannot be excluded.
Most studies include patients with a wide range of hirsutism severity at baseline, in some studies hirsutism was not even an explicit patient inclusion criterion, and several trials had baseline hirsutism scores that were not equivalent across intervention and control groups. Differences in how clinicians would score hirsutism in patients of different ethnic backgrounds and other limitations of the hirsutism scores need to be acknowledged (1). Although these limitations do not bias individual randomized trials (because they are likely to affect both arms) or their summary (i.e. this review), they argue for using patient-reported outcomes rather than subjective physician assessments of hirsutism in future clinical trials.
Also, the majority of eligible RCTs had methodological limitations (a lack of, or no reporting of, adequate allocation concealment and appropriate blinding and significant loss to follow-up). Furthermore, the often important between-study inconsistency in results remained largely unexplained. Overall, the quality of evidence informing the use of insulin sensitizers for the treatment of hirsutism is low to very low (30), and this therefore weakens clinical inferences and the corresponding clinical practice recommendations.
Our work has several strengths. We have summarized the best available evidence about the efficacy of insulin sensitizers for the treatment of hirsutism. Almost all eligible trials reported their findings in a way that allowed their inclusion in our metaanalyses, limiting reporting bias. The thorough systematic literature search, explicit eligibility criteria, extent of agreement between abstractors, and focused analyses with subgroup explorations strengthen the validity of this systematic review and the metaanalyses.
Since the completion of our review, we know that three pertinent trials have been published (31, 32, 33). In general, these trials found that insulin sensitizers provided no significant benefits on hirsutism when compared with antiandrogens or OCPs, results that support the findings and conclusions of our review.
Comparison with previous studies
There are two metaanalyses reporting on the effect of insulin sensitizers on hirsutism in patients with PCOS, both published in the Cochrane Database of Systematic Reviews. Lord et al. summarized trials (up to March 2004) of insulin sensitizers for the treatment of PCOS (34) against placebo or no treatment. This review included two trials with a total of 24 participants, both reporting no significant effect. The combined effect of these two trials, one using metformin alone (23) and one including metformin combined with clomiphene (35) favors metformin, largely due to the latter study and only after excluding patients that did not fulfill the National Institutes of Health consensus criteria for PCOS diagnosis (WMD, –5.12; P = 0.006; 95% CI, –8.77 to –1.47). Costello et al. (36) included a total of four trials comparing metformin vs. OCPs and two trials of metformin combined with OCPs vs. OCPs alone; they concluded that limited data demonstrated no difference in effect between metformin and OCPs, alone or in combinations. Our analyses included a larger number of studies; nevertheless, our results are consistent with the findings of these earlier and smaller reviews.
Implications for clinical policy and research
The overall quality of the evidence informing the use of insulin sensitizers for the treatment of hirsutism is low to very low using the GRADE system (30), and the current evidence suggests that insulin sensitizers are not effective agents for hirsute women.
Furthermore, there is limited reporting of adverse effects in these trials. Prolonged use of TZDs has led to heart failure, edema, and bone fractures in otherwise normal women at risk of, or with newly diagnosed, diabetes (37, 38). The uncertain safety of these agents during the first trimester of pregnancy may require the study of these agents alongside effective contraception. Given the positive findings associated with troglitazone use vs. placebo, further research on TZDs and hirsutism will be warranted as long as these potential harms associated with TZDs are acceptable to women with severe hirsutism who are unable to use or benefit from hair removal procedures, OCPs, and antiandrogens.
The accompanying clinical practice guidelines, based on the evidence from this review along with the values, preferences, and expertise of the Task Force members, provide clinicians and patients with current recommendations.
Unanswered questions
The poor methodological quality of the evidence summarized here highlights the research needs in this field. Investigators with nonprofit funding need to conduct large and rigorous randomized trials in women with different etiologies of hirsutism, using the various types and doses of insulin sensitizers, alone and in combination with other interventions (mainly OCPs in fertile women not seeking pregnancy) for longer periods of follow-up and measuring blinded patient self-assessments of hirsutism as the primary outcome. Randomized trials to clarify fully both the efficacy and safety of insulin sensitizers may have a greater likelihood of finding an important treatment effect in overweight women, a post hoc finding of our review that requires confirmation in new studies. Also needed are trials comparing insulin sensitizers with biological (e.g. eflornithine) and mechanical (e.g. shaving, depilation, electrolysis, and laser epilation) modifiers of hair growth.
Conclusion
Imprecise and inconsistent evidence of low to very low quality suggests that insulin sensitizers provide limited or no important benefit for women with hirsutism.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Disclosure Statement: M.C., B.A.S., D.N.F., D.M.K., M.L.L., R.J.M., M.B.E., P.J.E., and V.M.M. have nothing to declare.
First Published Online February 5, 2008
Abbreviations: CI, Confidence interval; FGS, Ferriman-Gallwey score; OCP, oral contraceptive pill; PCOS, polycystic ovary syndrome; RCT, randomized controlled trial; TZD, thiazolidinedione; WMD, weighted mean difference.
Received November 2, 2007.
Accepted January 25, 2008.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
nez-Bermejo E, Lasuncion MA, Escobar-Morreale HF 2007 Comparison of ethinyl-estradiol plus cyproterone acetate versus metformin effects on classic metabolic cardiovascular risk factors in women with the polycystic ovary syndrome. J Clin Endocrinol Metab 92:2453–2461This article has been cited by other articles:
 |
|
 |
|
  O. Koulouri and G. S Conway Management of hirsutism BMJ, March 27, 2009; 338(mar27_1): b847 - b847. [Full Text]
|
|
|
|
 |
|
 S. Palomba, A. Falbo, F. Zullo, and F. Orio Jr. Evidence-Based and Potential Benefits of Metformin in the Polycystic Ovary Syndrome: A Comprehensive Review Endocr. Rev., February 1, 2009; 30(1): 1 - 50. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
