Role of adiponectin and its receptor in prediction of reproductive outcome of metformin treatment in patients with polycystic ovarian syndrome


  • Hossam O. Hamed

    Corresponding author
    1. Department of Obstetrics and Gynecology, Qassim University, Buraidah City, Saudi Arabia
    2. Women's Health Center, Assiut University, Assiut, Egypt
    • Reprint request to: Associate Professor Hossam Omar Hamed, Women Health center, Assiut University, Assiut city, 71515, Egypt. Email:

    Search for more papers by this author



The aim of this study was to examine the effect of metformin on serum adiponectin and adiponectin receptor-1 (AdipoR1) and evaluate their role in prediction of ovulation in patients with polycystic ovarian syndrome (PCOS).

Material and Methods

The study cohort included 68 PCOS patients with clomiphene citrate resistance (group 1) and 28 healthy women as controls (group 2). Baseline serum adiponectin, AdipoR1, total testosterone (T), and homeostasis model of insulin resistance (HOMA-IR) were measured in all participants. Group 1 received metformin (1500 mg/day) for 6 months followed by second blood sampling.


Group 1 had significantly lower baseline adiponectin and AdipoR1 (P = 0.001) compared to group 2. During treatment, metformin resulted in conception in 5/68 (7%), ovulation in 33/68 (48%) and regular cycles in 41/68 (60%) patients. Group 1 showed post-metformin higher adiponectin and AdipoR1 (P = 0.01) and lower HOMA-IR (P = 0.006) and T (P = 0.001) compared to pre-treatment levels. Post-metformin ovulatory patients had higher adiponectin and AdipoR1 and lower HOMA-IR and T compared to anovulatory patients. Multivariate regression analysis in group 1 showed that only T and HOMA-IR were significant independent factors for predicting ovulatory cycles during metformin treatment (P = 0.04 and P = 0.05, respectively).


Metformin treatment enhances both adiponectin activity and insulin sensitivity, resulting in a less hyperandrogenic state in patients with PCOS. Serum adiponectin and AdipoR1 are poor predictors of ovulatory outcome during treatment.


Adiponectin is an adipocytokine expressed mainly in adipose tissue and is considered the most abundant circulating adipose-specific protein in humans.[1] It acts through two main receptors, adiponectin receptor-1 (AdipoR1) and adiponectin receptor-2 (AdipoR2).[2] The metabolic actions of adiponectin include enhancement of insulin sensitivity,[3] reduction of circulating lipid levels[4] and a protective anti-atherogenic and anti-inflammatory effect.[5, 6] In contrast to what might be expected, the production of adiponectin is decreased in obesity[7] and its serum level correlates negatively with waist circumference[8] (WC) and body mass index (BMI).[1]

Abdominal obesity and insulin resistance (IR) play a central role in pathogenesis of polycystic ovary syndrome (PCOS), which is considered the most common endocrine disorder that causes anovulatory infertility in young women.[9] The studies addressing serum adiponectin in PCOS patients are conflicting. Although some have reported decreased adiponectin levels compared to BMI-matched controls,[10-13] others have shown no difference after controlling for obesity.[14, 15] Whether this decline is secondary to associated obesity or IR is not clear.[10] The evidence of hypoadiponectinemia in both obese and lean PCOS patients potentiates the role of IR.[11, 16] Metformin, as an insulin sensitizer, has been widely used in treatment of PCOS and showed improvement of metabolic and hormonal disturbances in those patients.[17, 18] These improvements result in increased ovulatory and pregnancy rates whether used alone or in combination with clomiphene citrate (CC).[18, 19]

In this study we hypothesized that IR associated with PCOS is involved in the downregulation of adiponectin activity and hence, treatment with insulin sensitizers could restore normal adiponectin levels. We studied the effect of metformin on serum adiponectin and AdipoR1 levels in patients with PCOS and evaluated their role in prediction of reproductive outcome during treatment.


This was a cohort comparative study conducted in the Obstetrics and Gynecology Department, College of Medicine, Qassim University, Saudi Arabia, starting in January 2011 through October 2012.

All PCOS patients with CC resistance, who attend the infertility clinic in the Maternity-Childhood Hospital in Burraidah, Qassim provenance, were counseled to participate in the study after informed consent for all research procedures. The study was approved by the research center in the college of medicine and the Deanship of scientific research in Qassim University.

The inclusion criteria included PCOS patients with CC resistance (group 1) aged between 20 and 38 years. All women met the criteria for diagnosis of PCOS according to Rotterdam Consensus Group.[20] CC resistance is defined as failure to ovulate with CC for at least 3 months at a dose of 150 mg/day for 5 days.[18] The exclusion criteria included women under 20 or over 38 years and patients who had received gonadotrophins or hormonal contraception in the 3 months prior to the study. Patients with hyperprolactinemia (morning plasma prolactin ≥ 25 ng/mL) or other endocrine, hepatic, or renal disorders were also excluded. Another group of healthy and non-pregnant women with cross-matched age and BMI were recruited as controls (group 2).

The sample size was calculated according to the difference in adiponectin level in normal women compared to PCOS patients and the suspected change in group 1 after metformin therapy. Multiple previous studies[16, 21] had reported serum adiponectin range of 18–33 ng/mL (average 21 ng/mL) in normal women and 10–12 ng/mL in PCOS patients. Assuming that a 25% difference in adiponectin level before and after metformin treatment would be of significant clinical and biochemical value, we needed 58 patients in the PCOS group to demonstrate this difference with α of 0.05 and β of 0.2. We expected that some of the patients might have become pregnant or dropped out during follow-up, so an extra 15% of the required sample size was enrolled.

All participants in both groups were examined physically with measurement of BMI and WC. A fasting venous blood sample for baseline biochemical assays was taken on day 2–7 of the menstrual cycle. Samples were separated and stored at −20°C until assayed. Blood glucose was measured by oxidase method and spectrophotometric quantitation.[22] Insulin was detected by enzyme-linked immunosorbent assay (ELISA) kit. Insulin resistance was assessed using the homeostasis model assessment of insulin resistance (HOMA-IR).[23] Adiponectin and AdipoR1 levels were measured by ELISA kits. The soluble C-terminal fragment of AdipoR1 in blood was used to represent tissue receptor expression to minimize invasive procedures.[24] The levels of total testosterone (T), luteinizing hormone (LH) and follicle-stimulating hormone (FSH) were assessed with specific ELISA kits.

Patients in group 1 had received a course of oral metformin for 6 months with initial dose of 500 mg twice daily that increased within 2 weeks to 500 mg three times a day. A second post-treatment blood sampling was taken for measurement of all previous parameters after excluding those who became pregnant before completion of the 6-month treatment. The primary measured reproductive outcome was ovulation rate and the secondary outcomes included cycle rhythm and conception. The reproductive outcomes were assessed only during metformin treatment. Ovulation was confirmed if day-21 serum progesterone was ≥5 ng/mL or pregnancy had occurred. The patient was counted as regularly menstruating if at least 50% of her cycles during follow-up were 28 ± 7 days. Conception was confirmed by visualization of intrauterine gestational sac by vaginal ultrasound.

Data were analyzed using spss version 16 and were expressed as mean ± standard deviation and/or percentages. Unpaired t-test was used to compare quantitative parameters between patients and controls at baseline, while the paired test was used to compare parameters before and after treatment in the same patients. The χ2-test or Fisher's exact test were used for comparison of categorical variables in both groups. The correlation between adiponectin, its receptor and different variables of interest was assessed using Pearson's correlation coefficient. Multivariate regression analysis was done to identify independent variables of successful reproductive outcome after metformin treatment. P ≤ 0.05 was considered statistically significant.


Out of 68 participants enrolled in group 1, only five women became pregnant, giving a 7% conception rate within the 6-month follow-up period. One patient discontinued treatment due to bad compliance with metformin. This left only 62 patients in group 1 available for comparative analysis of hormonal and biochemical parameters with controls (n = 28).

Table 1 shows that both groups had comparable ages and BMI. Group 1 had significantly higher WC, HOMA-IR, T, and LH compared to controls while serum adiponectin and AdipoR1 were significantly lower. Table 1 shows also that overall ovulatory and regularly menstruating patients during metformin treatment were 33/68 (48%) and 41/68 (60%), respectively. Metformin resulted in significant increase in adiponectin and AdipoR1 (P = 0.01) while WC, HOMA-IR, T, and LH/FSH ratios were significantly decreased compared to baseline levels.

Table 1. Clinical, biochemical and hormonal variables in PCOS patients (before and after metformin) in comparison to controls
VariablesPCOS patients (n = 62)Controls (n = 28)P-value§
Before metforminAfter metformin(Before vs after) OR (95%CI)Before vs control
  1. †Data is expressed as mean ± standard deviation unless otherwise indicated. ‡Out of a total of 68 enrolled, five women became pregnant and one had discontinued treatment. §Unpaired t-test is used unless otherwise indicated. ¶Fisher's exact test. AdipoR1, adiponectin receptor-1; BMI, body mass index; CI, confidence interval; FSH, follicle-stimulating hormone; HOMA- IR, homeostasis model assessment of insulin resistance; LH, luteinizing hormone; OR, odds ratio; PCOS, polycystic ovarian syndrome; WC, waist circumference.
Age (years)29.3 ± 4.230.9 ± 4.30.1
Patients with regular cycles n(%)11(18)41(60)



Patients with ovulatory cycles n(%)7(13)33 (47)



BMI(kg/m2)30.2 ± 4.829.3 ± 2.728.4 ± 3.30.1450.08
WC(cm)98.1 ± 6.095.3 ± 6.385.4 ±
HOMA-IR(mmol/L-μU/mL)3.1 ± 1.42.3 ± 0.91.0 ± 0.30.0060.000
Adiponectin(ng/mL)13.9 ± 2.615.2 ± 1.715.6 ±
AdipoR1 (ng/mL)17.0 ± 4.319.1 ± 1.519.7 ±
Testosterone(ng/DL)93.1 ± 40.069.5 ± 26.846.8 ± 20.30.0010.000
LH(mIU/mL)13.9 ± 3.910.2 ± 2.06.6 ± 1.50.0010.000
FSH(mIU/mL)7.2 ± 1.77.1 ± 1.47.4 ± 1.30.6060.6
LH/FSH ratio2.0 ± 0.81.5 ± 0.70.9 ± 0.30.0020.000

Table 2 shows that serum adiponectin in group 1 correlated negatively with WC, HOMA-IR and T. The highest correlation coefficient was found with HOMA-IR (r = −0.55, P < 0.001). Also, AdipoR1 had inverse correlations with WC and HOMA-IR while it had a positive correlation with adiponectin level.

Table 2. Correlations between adiponectin and adiponectin receptor-1 with clinical and biochemical parameters at baseline in PCOS patients
  1. Adipo-R1, adiponectin receptor-1; BMI, body mass index; FSH, follicle-stimulating hormone; HOMA- IR, homeostasis model assessment of insulin resistance; LH, luteinizing hormone; PCOS, polycystic ovarian syndrome; r, Pearson correlation coefficient.
BMI (kg/m2)−0.250.126−0.180.272
Waist circumference (cm)−0.320.048−0.340.039
HOMA-IR (mmol/L-μU/mL)−0.550.001−0.360.025
Adipo-R1 (ng/mL)0.450.005
Total testosterone (ng/mL)−0.390.013−0.180.286
LH (mIU/mL)−0.270.1040.0610.717
FSH (mIU/mL)0.130.4360.170.315
LH/FSH ratio−0.280.092−0.0390.813

Table 3 shows subgroup analysis of reproductive outcome of metformin in group 1. Ovulatory patients had higher adiponectin (P = 0.006) and AdipoR1 (P = 0.01) and lower HOMA-IR (P = 0.000), T (P = 0.000) and WC (P = 0.002) compared to anovulatory patients. Similarly, patients who achieved regular cycles had lower HOMA-IR (P = 0.001), T (P = 0.000) and WC (P = 0.001) and higher adiponectin (P = 0.04) and AdipoR-1 (P = 0.05) compared to those with persistent oligomenorrhea. There was no significant difference in pre-treatment levels of serum adiponectin (13.7 ± 2.2 vs 14.1 ± 2.6, P = 0.5) and AdipoR1 (16.5 ± 3.8 vs 17.4 ± 4.4, P = 0.4) in ovulatory compared to anovulatory patients (data not shown).

Table 3. Relation of clinical, hormonal, and biochemical variables to ovulation and cycle rhythm in PCOS patients after metformin treatment
 Post-metformin ovulation (n = 62)Post-metformin cycle rhythm (n = 62)
Ovulatory (n = 27) Mean ± SDAnovulatory (n = 35) Mean ± SDP-value*Regular (n = 36) Mean ± SDIrregular (n = 26) Mean ± SDP-value*
  1. *Significance calculated by unpaired t-test. Adipo-R1, adiponectin receptor-1; BMI, body mass index; FSH, follicle-stimulating hormone; HOMA- IR, homeostasis model assessment of insulin resistance; LH, luteinizing hormone; PCOS, polycystic ovarian syndrome; SD, standard deviation.
BMI (kg/m2)28.8 ± 2.929.5 ± 2.50.3129.0 ± 2.929.6 ± 2.404
Waist circumference (cm)93.3 ± 6.597.6 ± 3.90.00292.4 ± 7.297.6 ± 3.70.001
HOMA-IR (mmol/L-μU/mL)1.9 ± 0.52.5 ± 0.70.0002.0 ± 0.652.7 ± 1.00.001
Adiponectin (ng/mL)15.8 ± 0.714.6 ± 2.10.00615.6 ± 0.714.4 ± 3.30.04
AdipoR1 (ng/mL)19.6 ± 1.218.7 ± 1.50.0119.4 ± 1.318.7 ± 1.50.05
Testosterone (ng/mL)54.6 ± 25.081.4 ± 22.20.00059.8 ± 22.682.6 ± 24.60.000
LH (mIU/mL)9.5 ± 2.810.6 ± ± 3.010.9 ± 2.70.08
LH/FSH ratio1.4 ± 0.71.5 ± 0.60.561.4 ± 0.91.5 ± 0.5061

Table 4 shows data of regression analysis of various biochemical and hormonal parameters that affect reproductive outcome in subgroup analysis in group 1. All parameters were studied together by enter method in a multivariate model. Only T and HOMA-IR were significant independent factors for ovulation (P = 0.04 and 0.05) and regular cycles (P = 0.05 and 0.04). Neither serum adiponectin nor AdipoR1 in this model could predict significantly ovulatory or regular cycles.

Table 4. Multivariate regression analysis of biochemical and hormonal variables for prediction of reproductive outcomes of metformin in PCOS patients (group 1)
Biochemical and hormonal variablesGroup 1 (n = 62) Ovulatory cyclesGroup 1 (n = 62) Regular cycles
βExp (β) [95%CI]P-valueβExp (β) [95%CI]P-value
  1. CI, confidence interval; HOMA- IR, homeostasis model assessment of insulin resistance; PCOS, polycystic ovarian syndrome.
Adiponectin0.311.37 [0.56–3.31] [0.65–2.37]0.5
Adiponectin reeptor-10.561.75 [0.89–3.44] [0.50–1.60]0.7
HOMA-IR−1.420.24 [0.05–1.01]0.05−1.130.32 [0.10–0.97]0.04
Total testosterone−0.050.94 [0.89–0.99]0.04−0.040.95 [0.90–1.00]0.05


We found higher IR and T levels but lower serum adiponectin and AdipoR1 in PCOS patients at baseline compared to controls. Among these findings, IR is the key point for understanding the biochemical and hormonal alterations in PCOS.[7] The cause of IR in PCOS is multifactorial, including genetic and metabolic factors. In a previous study,[25] many post-insulin receptor molecular gene defects in muscular and adipose tissues had been reported. In addition those patients have increased lipolysis process especially in visceral fat due to abnormal post-receptor sensitivity to catecholamine resulting in a lipotoxicity state, which was proved to interfere with insulin receptor signaling.[26, 27] All these factors act together to develop a state of hyperinsulinemia which in turn stimulates ovarian theca cell androgen production by direct ovarian action through binding to insulin or IGF-1 receptor or by stimulating LH release.[27] Hyperinsulinemia could also enhance the bioavailability of androgen by decreasing the biosynthesis of sex-hormone-binding globulins with net result of hyperandrogenism.[27, 28]

Another important consequence for development of IR in PCOS patients is decreased serum level of adiponectin and diminished expression of its receptor compared to the healthy population,[13, 17, 18, 29-31] which has been confirmed in the current study. The current baseline serum adiponectin level in PCOS patients is slightly higher than that reported by others (10–12 ng/mL).[16, 21] Different characteristics of study populations, including BMI and race, could be possible explanations. The mean baseline BMI in current PCOS patients is 30.2 kg/m2, which is just in the obesity range. The observed low adiponectin level in obese patients contradicts the logical thinking that adiponectin should increase in those patients as they have much adiposity responsible for adiponectin synthesis.[1] Decreased intracellular insulin activity necessary for adiponectin production is a possible explanation.[10] Moreover, the resulting hyperandrogenism is thought to contribute further to low synthesis of adiponectin and its receptors in adipose tissues.[32] Our low adiponectin level in PCOS women is in agreement with others.[11-13] Toulis et al. in their systematic meta-analysis had reported lower level of adiponectin in PCOS, which was probably related to IR.[11] As adiponectin has a principal insulin-sensitizing action, patients with PCOS enter into a vicious circle of IR, hypoadiponectinemia, and hyperandrogenism.[30] This vicious circle could be proved in our data by the negative correlation between adiponectin level on one side and HOMA-IR and testosterone on the other side. In contrast to this pathogenesis, many investigators however, had reported that the relation between insulin sensitivity and low adiponectin level in PSOS is an association rather than a cause–effect relation.[33, 34] They endorsed the potential role of obesity per se in determining adiponectin level in those patients. In one study, the authors reported lower adiponectin levels in obese PCOS patients compared to those with normal BMI.[35] This may suggest that adiposity, and not IR, is the main determinant factor of adiponectin.[35] Nevertheless, BMI could not be used as a single indicator of adipose tissue content in those patients. This was confirmed in the current study by higher WC (as an indicator of central obesity) in PCOS patients compared to controls, despite comparable BMI.

In the current study we also observed low serum AdipoR1 level, which is in agreement with others, who found lower AdipoR1 mRNA in obese subjects than controls and reported upregulation of gene expression with weight loss.[36] In contrast to this finding, one study found that women with PCOS had upregulation of adiponectin receptors.[37]

The major pharmacological action of metformin is reduction of glucose production by the liver and improvement of IR. It is widely used for treatment of PCOS-related infertility symptoms, especially in those with CC resistance.[18] Our metformin data agrees with previous studies, which showed significant decrease in WC, IR and T levels.[17, 18] Our data are also consistent with others, who showed significant increase of adiponectin after a 6-month course of metformin at a daily dose of 1000 mg.[29, 38] Metformin could therefore exert its beneficial metabolic effects by breaking the postulated vicious circle of IR and hypoadeponectinemia.[30] This effect seems to depend on type and dose of the insulin sensitizer.[39, 40] Inconsistent with current findings, Trolle et al. reported that metformin had no effect on adiponectin in spite of significant decrease of BMI and IR.[31] These conflicting results could be explained by different study populations, including race, BMI and severity of IR. Similarly, we observed a rise in serum AdipoR1 after metformin treatment, which is consistent with others, who reported that metformin modifies AdipoR1 expression in mice that takes place mainly in glucose-utilizing tissues resulting in higher AdipoR1 level in muscle tissues.[41]

The current total number of ovulatory women (47%) is in agreement with others who reported a range of 29–70% depending on study population and treatment duration.[18, 42-44] The subgroup analysis in the current study has observed a higher effect of metformin on adiponectin and AdipoR1 levels and other biochemical variables on patients who were ovulatory and regularly menstruating compared to those with persistent anovulation and oligomenorrhea. This observation was not due to a difference in pretreatment levels of adiponectin and AdipoR1, which was statistically insignificant. However, the multivariate regression analysis showed that only declines of IR and T were significant independent factors for prediction of ovulatory and regular cycles during metformin treatment. Neither adiponectin nor AdipoR1 in this model of regression had significant value to predict ovulation. This could be explained by that reduction of IR and consequently hyperandrogenism by metformin is the key point and is more important than the increase in adiponectin activity in enhancing ovulation. Replacing an intra-follicular androgenic environment to an estrogenic one could correct disturbances of ovarian–pituitary feedback and reduce LH/FSH ratio, resulting in more follicular recruitment and cycle regulation.[18, 45-49] On the other hand, the literature has not yet proved a similar direct role of adiponectin in enhancing ovulation with the exception of a few animal reports suggesting its role in release of pituitary gonadotropins.[50] Another explanation why adiponectin and AdipoR1 were insignificant independent factors for predicting ovulation is the multivariate regression model used in analysis. In this model, all serum parameters were included together in the equation, which could mask the predictive role of individual parameters according to their importance. In addition, we could not exclude the possible effect of central obesity as indicated by high WC in current PCOS patients as independent factor could affect both ovulation and adiponectin level.

In conclusion PCOS patients had lower baseline serum adiponectin and AdipoR1 compared to healthy women after controlling for BMI. Metformin enhances both adiponectin activity and insulin sensitivity and decrease the hyperandrogenism state resulting in more ovulatory and regular cycles. Improvements of androgen level and IR, but not adiponectin activity, have significant predictive value for reproductive outcomes of metformin treatment.


The author acknowledges the Deanship of Scientific Research, Qassim University, Saudi Arabia for its financial support for this work. The author also acknowledges Dr Kamal MM and Dr Salama RH for their assistance in biochemical and hormonal assays.


Apart from the official financial support received from the Deanship of Scientific Research in Qassim University (Research File no. 595-1432-1433) this research had no relationships with any companies that may have a financial interest in the information contained in the manuscript.