IGF-II, IGFBP-4, -6, and -7, and HMGB1 show changes in follicular fluid in PCOS.

Polycystic ovary syndrome (PCOS) is a multisystem endocrine and metabolic disorder which is reported to be the most common endocrine disease in women of reproductive age ( 1 , 2 ), affecting 5%-18% depending on the criteria used for diagnosis ( 3 ). Its etiology is still unclear; however, genetic predisposition, epigenetic factors, gestational environment, and lifestyle play an important role ( 4 ). PCOS typically presents during adolescence and is more frequent in overweight or obese women ( 5 ). In adults, the diagnosis of PCOS is based on the presence of at least two of the following characteristics: (I) clinical/biochemical signs of hyperandrogenism, (II) ovulatory dysfunction, and (III) polycystic ovarian morphology on ultrasound or elevated anti-mullerian hormone (AMH) levels, after the exclusion of other causes ( 6 ). Moreover, PCOS is associated with systemic low-grade inflammation and often with insulin-resistance (70%-75% of PCOS women) ( 7 ) which contribute to PCOS comorbidities such as infertility, hyperinsulinemia, impaired glucose tolerance, type 2 diabetes, dyslipidemia, obesity, and increased cardiovascular risk ( 8 – 11 ). In PCOS, an increase in circulating pro-inflammatory cytokines has been reported to contribute to ovarian dysfunction ( 12 – 17 ) and to hamper insulin sensitivity ( 18 , 19 ). We previously described increased high-mobility group box 1 (HMGB1) content in follicular fluid (FF) of PCOS patients with respect to healthy control women ( 20 ). HMGB1 is involved in inflammatory diseases ( 21 ) and different insulin resistance/hyperinsulinemia-related disorders such as obesity and type 2 diabetes mellitus ( 22 ). HMGB1 is linked also to glucose metabolism, a sensitive pathway in PCOS ( 23 ). The insulin-like growth factor (IGF) system is a complex network of proteins comprising insulin, IGF-I, IGF-II, and IGF-binding proteins (IGFBP1-7), the latter being involved in the regulation of bio-availability and activity of IGFs and IGFBP-2, -4, and -5 having mainly an inhibitory action on IGFs ( 24 – 26 ). The IGF system is involved in the regulation of glucose metabolism ( 27 ) and is altered in chronic inflammation where both circulating IGF-I and -II can be reduced ( 25 , 28 ). An interesting study showed that both low and high IGF-I levels can be associated with insulin resistance ( 29 ). IGFBP-1 and IGFBP-2 are negatively regulated by insulin and glucose. Moreover, IGFBP-2 has been reported to increase under conditions of increased/chronic inflammation ( 25 , 30 – 32 ). The IGF system has a relevant role also in obesity ( 33 ). Indeed, IGF-1 circulating levels have been described to be low in obese subjects, particularly in those with increased adiposity, inflammation, dyslipidemia, and hyperuricemia ( 34 ). Moreover, low serum levels of IGF-II and IGFBP-2 have been reported in both obese children and adults ( 31 , 35 , 36 ), and lower serum IGFBP-1 levels have been associated with an increased risk of metabolic syndrome ( 37 ). The IGF system is of importance also for the regulation of ovarian function ( 38 ). Both IGF-I and IGF-II promote follicular maturation in mammals ( 39 – 41 ), and their effects depend on follicular development stage and on the regulation by IGFBPs ( 42 ). IGF-I and IGF-II levels have been described to be reduced in PCOS ovaries ( 43 ). Conversely, free IGF-I circulating levels have been described to be increased and to correlate negatively with the total follicle number and with IGFBP-1 levels ( 44 ). IGFBP-1 has been found to play different roles in the female reproductive tract; in the ovary it would impact ovulation, follicular growth and steroidogenesis ( 45 ). IGFBP-2 was reported to be higher in PCOS and atretic follicles from normally cycling women compared with normal healthy estrogenic follicles suggesting a role in normal folliculogenesis ( 46 ). IGFBP-3 levels would be of importance for oocyte maturation and embryo development ( 41 ). Both IGFBP-4 and -5 would contribute to follicle development. In normal physiology, intact IGFBP-4 has been observed to increase in small antral follicles and to be lower in preovulatory follicles, whereas IGFBP-5 would show an opposite trend ( 40 ). There are no data to our knowledge on intrafollicular IGFBP-6 levels in humans. However, IGFBP-6 expression levels were reported to be increased in bovine theca cells in the final stages of follicular maturation ( 47 ); conversely, in cycling ewes, they were decreased in large atretic follicles compared with normal follicles ( 48 ). Recently, a bulk RNA-seq analysis of PCOS granulosa cells evidenced that IGFBP-7 mRNA levels were increased in granulosa cells from patients with PCOS and the knockdown of the IGFBP-7 gene in mice with PCOS led to decreased ovarian cell apoptosis, increased proliferation, and androgen synthesis ( 49 ). A recent proteomic analysis evidenced that the IGF system was among the most enriched pathways in FF in PCOS ( 50 ), suggesting that it is worth to be investigated to better understand this complex condition. The aim of this study was to investigate changes in the IGF system in FF from PCOS women compared with controls and to evaluate any relationships with body mass index (BMI) and HMGB1 as a marker of both inflammation and altered glucose metabolism. In particular, we hypothesized that the content and bioavailability of some IGF system proteins could change in PCOS and that these changes could be dependent on BMI.

IGF system peptide and HMGB-1 levels in FF from PCOS and control women are reported in Figure 1A . IGF-I (102.1 ± 46.1 ng/ml vs. 98.5 ± 58.7 ng/ml, n.s.), IGFBP-1 (9.0 ± 3.4 ng/ml vs. 9.4 ± 4.6 ng/ml, n.s.), IGFBP-2 (671.2 ± 204.7 ng/ml vs. 638.0 ± 228.0 ng/ml, n.s.), and IGFBP-3 (1,592.0 ± 597.3 ng/ml vs. 1,770.0 ± 835.6 ng/ml, n.s.) levels were similar in PCOS and control subjects. IGFBP-5 was undetectable in most FF samples. IGF-II (387.2 ± 210.5 ng/ml vs. 495.7 ± 157.9 ng/ml; p=0.0007) and IGFBP-4 (16.5 ± 7.7 ng/ml vs. 20.7 ± 9.5 ng/ml; p=0.005) were lower in FF from PCOS women compared with controls, whereas IGFBP-6 (55,570 ± 27,793 pg/ml vs. 46,417 ± 22,110 pg/ml; p=0.03) and IGFBP-7 (405.9 ng/ml ± 211.6 ng/ml vs. 324.1 ng/ml ± 145.8 ng/ml; p=0.009) were higher. IGFs, IGFBPs, and HMGB1 levels in FF of women with PCOS and controls. (A) IGF system protein levels in FF of women with PCOS compared with controls PCOS n=70; controls (C) n=70. (B) HMGB1 levels in FF of women with PCOS compared with controls. Data are reported as mean ± SD. The differences between the two groups were analyzed using unpaired Student’s T-test, and p-values< 0.05 were considered significant. HMGB1 levels in FF were higher in PCOS compared with controls (41.5 ± 22.1 ng/ml vs. 29.5 ± 20.4 ng/ml; p=0.003) ( Figure 1B ). In order to clarify whether pooling the different PCOS phenotypes in the PCOS group may obscure or dilute phenotype-specific effects, we performed a phenotype-stratified subanalysis for IGF-II, IGFBP-4, IGFBP-6, and IGFBP-7 considering the following comparisons: hirsute vs. not hirsute PCOS and oligo/amenorrhoeic PCOS vs. regularly cycling PCOS. As reported in Supplementary Figure 1 , we did not find any differences among groups. The molar ratios of IGFs to IGFBPs in both PCOS and controls are reported in Figure 2 . No differences were found for IGF-I molar ratios between PCOS and controls ( Figure 2A ), whereas IGF-II/IGFBP-2 (p=0.0007), IGF-II/IGFBP-3 (p=0.03), IGF-II/IGFBP-6 (p=0.003), and IGF-II/IGFBP-7 (p=0.03) molar ratios were significantly lower in the PCOS group with respect to controls ( Figure 2B ). IGF to IGFBP molar ratios in FF from women with PCOS with respect to controls. (A) IGF-I and (B) IGF-II to IGFBP molar ratios in FF from PCOS n=70; controls (C) n=70. Data are reported as mean ± SD. p-values <0.05 were considered significant. In under/normal weight women, IGF-II (380.7 ± 201.8 ng/ml vs. 514.1 ± 165.3 ng/ml; p=0.005) and IGFBP-4 (15.2 ± 6.9 ng/ml vs. 20.0 ± 9.7 ng/ml; p=0.04) FF levels were lower in PCOS women with respect to controls, whereas IGFBP-7 levels (444.7 ± 251.5 ng/ml vs. 300.6 ± 103.6 ng/ml; p=0.01) were higher ( Figure 3A ). IGF system and HMGB1 protein levels in FF of women with PCOS compared with controls according to BMI (A) IGF system protein levels in FF of women with PCOS compared with controls according to BMI: underweight/normal weight PCOS (n=37), overweight/obese PCOS (n=32), overweight/normal weight C (n=49), and overweight/obese C (n=20). (B) HMGB1 levels in FF of women with PCOS compared with controls stratified based on BMI: underweight/normal weight PCOS (n=32), overweight/obese PCOS (n=29), overweight/normal weight C (n=49), and overweight/obese C (n=15). (C) IGF-I and (D) IGF-II to IGFBPs molar ratios in FF from women with PCOS with respect to controls according to BMI: underweight/normal weight PCOS (n=37), overweight/obese PCOS (n=32), overweight/normal weight C (n=49), and overweight/obese C (n=20). Data are reported as mean ± SD. p-values <0.05 were considered significant. In overweight/obese women, IGFPB-2 FF levels were lower in controls with respect to their under/normal weight counterpart (525.2 ± 271.3 ng/ml vs. 690.9 ± 187.2 ng/ml; p=0.01) ( Figure 3A ). No differences in IGF-I, IGFBP-1, IGFBP-3, and IGFBP-6 levels were observed in PCOS versus controls according to BMI. HMGB1 FF levels were increased in the overweight/obese PCOS women compared with the overweight/obese controls (40.7 ± 26.3 ng/ml vs. 22.1 ± 10.4 ng/ml; p=0.03, Figure 3B ). The IGF-I/IGFBP-2 molar ratio was increased in the overweight/obese controls compared with under/normal weight control women (p=0.0001) but was also significantly increased compared with the overweight/obese PCOS women (p=0.002) ( Figure 3C ). The IGF-II/IGFBP-2 molar ratio was decreased in the overweight/obese PCOS women compared with the overweight/obese control women (p=0.009) ( Figure 3D ). The significant correlations between IGF system peptides in FF and BMI are reported in Figure 4A . Correlations among IGF system peptides and BMI in FF in PCOS and control women. (A) Significant correlations among IGF system peptides and BMI in PCOS (n=70) and controls (C; n=70). (B) Correlations among IGF system peptides in PCOS (n=70) and controls (C; n=70). Only correlations which reached statistical significance (p 0.30 are shown. In PCOS, BMI was negatively associated with IGFBP-2 (r= −0.352, p=0.003). In controls, BMI was associated with IGFBP-2 (r= −0.527, p<0.0001) and IGFBP-7 (r= 0.308, p=0.011). Moreover, BMI was negatively associated with IGFBP-2 in both under/normal weight controls (r=−0.332, p=0.02 ) and overweight/obese controls (r=−0.620, p=0.004). Correlations were performed separately in controls and PCOS subjects. Significant correlations are reported for both groups in Figure 4B . In PCOS women, IGFBP-1 correlated with IGFBP-2 (r=0.303, p=0.03) in FF. In control subjects, IGFBP-3 correlated with IGFBP-4 (r=0.453, p<0.0001). In women with PCOS, HMGB1 was positively associated with IGFBP-2 (r=0.341, p=0.007) whereas in controls, HMGB1 correlated with IGF-I (r= 0.303; p=0.02) ( Supplementary Figure 2 ).

The findings of this study showed changes in IGF system peptide content in FF from PCOS women and confirmed higher levels of HMGB1 in PCOS. Specifically, we observed a reduction in IGF-II and IGFBP-4 and an increase in IGFBP-6 and IGFBP-7. Overall IGF-II bioavailability was found to be reduced in FF in PCOS. When considering BMI, under/normal-weight PCOS women had lower IGF-II, IGFBP-2, and IGFBP-4 contents and increased IGFBP-7 in FF compared with controls, whereas in obese PCOS women, such changes were not so evident suggesting that PCOS is associated with changes that are further modified by obesity. This is consistent with increased IGF-I and IGF-II to IGFBP-2 molar ratios in FF being observed in obese healthy women, but a further reduction of these ratios was observed in obese PCOS women. Furthermore, BMI was negatively associated with IGFBP-2 content in FF in both PCOS and controls. In these latter, BMI was positively associated with IGFBP-7 FF content. In control women, IGFBP-3 and IGFBP-4 FF contents were associated, whereas in PCOS, IGFBP-1 and IGFBP-2 were the peptides to be correlated. HMGB1 was confirmed to be increased in FF in PCOS and obesity increased further its levels and was associated with IGFBP-2 FF content. In controls, HMGB-1 was found to be negatively and weakly associated with IGF-I FF content. The finding of a reduction in IGF-II levels in FF from PCOS patients is consistent with few previous findings ( 44 ). As these changes were observed in the under/normal-weight PCOS subjects, this supports the hypothesis that these changes are specific to PCOS. The lower IGF-II bioavailability in PCOS, confirmed by the molar ratios of IGF-II to IGFBPs, in consideration of the reported functions of IGF-II in the ovary, would suggest a negative effect on follicular development that is a distinctive feature of PCOS ( 39 , 40 , 42 , 53 ). IGF-II bioavailability was reduced in the FF from obese PCOS women compared with controls, but we did not find a reduced IGF-I content at variance with Barreca et al. ( 43 ) possibly due to the fact that these authors used a different detection method (RIA), and the BMI of the patients was not taken into account. IGF-I mRNA levels were previously reported to be reduced in bovine cystic follicles ( 54 ); however, differences in gene expression do not always reflect changes in protein levels, and findings in bovines may not reflect exactly findings in humans. We did not find any differences in IGFBP-1 levels between PCOS and controls. The data available to date refer to blood concentrations, where a decrease in IGFBP-1 levels has been reported in insulin-resistant PCOS women ( 55 ), and insulin and IGFBP-1 serum concentrations are well known to be negatively associated ( 56 ). Serum probably cannot be considered as a proxy of follicular fluid, as previously supported by other studies; these previous findings were likely related to body metabolic changes ( 57 ), with an increased risk of metabolic syndrome being a well-known comorbidity in women with PCOS ( 37 ). Although IGFBP-2 levels in FF were not significantly different between the PCOS and control groups, we observed some differences, in particular relative to the IGF-I and IGF-II to IGFBP-2 molar ratios as these were lower in the obese PCOS women compared with the obese control women, with BMI being associated with IGFBP-2 in both groups, and with IGFBP-2 being associated with IGFBP-1 in FF in PCOS women. Taken together, these data suggest that metabolic changes associated with an increased BMI have a stronger effect than PCOS per se . As previously mentioned, IGFBP-2 serum concentrations have been reported to be reduced in obesity ( 31 , 34 – 36 ), fatty liver disease, and type 2 diabetes mellitus ( 58 , 59 ) that represent all possible comorbidities of PCOS, and all are associated with reduced insulin sensitivity. In our cohort, IGFBP-3 follicular fluid levels were similar in PCOS and control women and were independent of BMI. These data are in line with a previous study in serum from PCOS women which reported similar serum IGFBP-3 levels in PCOS and controls ( 60 ) but at variance with one proteomic study reporting lower serum IGFPB-3 levels in PCOS compared with controls ( 61 ). Moreover, our findings are different from previous findings reporting reduced IGFBP-3 in PCOS FF ( 62 ). IGFBP-4 levels were lower in PCOS with respect to controls and particularly in under/normal-weight PCOS women compared with under/normal-weight controls, suggesting that these changes are specific to PCOS. IGFBP-4 is of special interest as it has been reported to be involved with follicle development; particularly, intact IGFBP-4 FF levels have been found to be higher in small antral follicles and to decrease in preovulatory follicles in patients with ovarian cancer and no endocrine abnormalities ( 40 ). Our findings reflect the perturbation of PCOS ovulatory function characterized by reduced follicular development and is in line with the findings of a previous report on serum IGFBP-4 levels in PCOS ( 61 ). IGFBP-5 was undetectable in most of our samples, possibly depending on the detection method used, and not allowing for any conclusion on possible roles related with ovarian function. We cannot exclude a matrix effect either. IGFBP-5 has been poorly investigated in human FF. To the best of our knowledge, only one paper described the IGF system including IGFBP-5 gene expression in follicles at different developmental stages, reporting changes from preovulatory follicles to small antral follicles with an opposite direction compared with IGFBP-4 ( 40 ). However, gene expression does not reflect necessarily protein content, and this could account for the extremely low amounts detected. The increased IGFBP-6 levels in FF in PCOS is a novel finding. IGFBP-6 expression levels have been found to be increased in bovine theca cells in the final stages of follicular maturation ( 47 ), and to be decreased in large atretic follicles in cycling ewes ( 48 ), suggesting that IGFBP-6 is involved in the regulation of follicular maturation. This suggests that IGFBP-6 should be further studied in human ovarian function and in PCOS. IGFBP-7 content in FF was also increased in PCOS compared with control subjects, in particular in under/normal-weight women, suggesting that these changes are likely specific to PCOS. This would be confirmed by a recent bulk RNA-seq analysis of PCOS granulosa cells where IGFBP-7 mRNA levels were found to be increased, and the knockdown of the IGFBP-7 gene, in a PCOS mouse model, led to increased granulosa cell proliferation, reduced apoptosis, and decreased testosterone secretion correcting PCOS-like features ( 49 ). This underscores the importance of IGFBP-7 in human ovarian function too and the need for further research. Finally, the weak positive association of BMI with IGFBP-7 FF content supports a worse outcome in obese PCOS women compared with the under/normal weight. Finally, HMGB1 FF levels were increased in PCOS patients with respect to control subjects confirming previous findings obtained also by our research team ( 20 , 23 , 63 – 66 ). This stood true also for HMGB1 levels being higher in overweight/obese PCOS. The positive association of HMGB1 with IGFBP-2 is compatible with the increased inflammatory status in PCOS patients and confirms previous findings in chronic inflammatory diseases characterized by an increase in circulating IGFBP-2 ( 25 , 66 ). This study does present some limitations. Indeed, we do not have information on IGFBP proteolysis as this has been shown in a few studies and would further impact IGF bioactivity ( 67 , 68 ). Furthermore, we are aware that the stimulation protocol for IVF could have had an effect on protein concentrations in FF, but both PCOS and control subjects underwent the same stimulation protocol; thus, the two groups can be compared as in previous studies ( 20 , 43 ), and for ethical reasons, FF could not have been obtained differently. It should be considered, however, that the study offers a picture of changes in PCOS versus controls at ovulation. The difference in age between PCOS patients and control subjects (mean age difference: 2.7 years) is unlikely to have partially impacted the between-group comparisons because of the physiological decline in circulating and intra-ovarian IGF system components with age. Furthermore, we acknowledge a relatively small number of patients in the overweight/obese control group and cannot entirely exclude that this could have had an effect on the differences between overweight/obese PCOS and overweight/obese controls, especially with regard to IGF-II and IGFBP-4. The means and SD of IGFBP-7 between overweight/obese PCOS and overweight/obese controls are almost identical so it is even less probable that a slight increase in the sample size of the overweight/obese control group could affect the strength of the observation. Moreover, we are aware that tubal factor infertility and unexplained infertility in control subjects may involve alterations in the intrafollicular IGF and inflammatory milieu. However, control subjects were screened to exclude other ovulatory dysfunctions or mild forms of PCOS to avoid any potential confounding factors. Overall, this study provides an overview of the IGF system in FF obtained from women with PCOS, showing significant changes, revealing and underscoring a role of IGFBP-6 and BP-7, currently poorly studied in PCOS and in humans. In the future, further research should evaluate whether these specific IGFBPs could be used as biomarkers of oocyte quality by evaluating considering also any correlations between IGFBP levels in FF and the rate of pregnancies. The study also highlights reduced IGF-II bioactivity in FF in PCOS. These data provide some useful information on the derangements in PCOS and on possible mechanisms related to comorbidities. Moreover, as the IGF system is related with the development of some cancers ( 69 ), somehow the changes described could be possibly related with the increased risk of developing ovarian cancer in these patients ( 70 ) that should be addressed by further research.

At the time of oocyte retrieval for in vitro fertilization (IVF) procedures, 70 patients with PCOS [CA (years):34.1 ± 4.7 years; BMI (kg/m 2 ):25.6 ± 5.6; hirsute (n):16; amenorrhoic (n):6; oligomenorrhoic (n):31; regularly cycling (n):33] and 70 regularly cycling women (control subjects) [CA (years):36.8 ± 3.8 years; BMI (kg/m 2 ):24.0 ± 5,0; none hirsute; all regularly cycling (n):70] were enrolled in the study by the Fertility Center, Department of Mother and Child, Arcispedale Santa Maria Nuova, AUSL - IRCCS of Reggio Emilia. The main clinical features and findings at ultrasound are reported in Supplementary Table 1 . PCOS patients were diagnosed according to the Rotterdam criteria ( 51 ). The presence of at least two symptoms among oligo/amenorrhea, clinical or biochemical signs of hyperandrogenism, and polycystic ovarian morphology at ultrasound were considered, and additional criteria whenever possible ( 6 ). Control subjects were women undergoing IVF because of tubal or idiopathic infertility causes, with normal endocrine exams and regular menstrual cycles. Control subjects were screened to exclude ovulatory dysfunctions or mild forms of PCOS. Exclusion criteria were the presence of tumors, endometriosis, celiac disease, genetic or chronic diseases, Cushing syndrome, hyperprolactinemia, and dimorphisms. All participants did not receive any additional treatment for at least 2 months before IVF except for the ovarian stimulation therapy. All subjects enrolled in this study underwent the same stimulation protocol. In detail, ovarian hormonal stimulation was performed according to a long luteal gonadotropin-releasing hormone (GnRH) agonist depot protocol. After the confirmation of complete ovarian downregulation by means of biochemical and instrumental analyses, recombinant FSH (rFSH) was administered in the first 5 days using a starting dose tailored according to the patient’s age and antral follicle count. From day 6 of ovarian stimulation, the dose was modified according to the ovarian response. In case of appearance of dominant follicles, ovulation was stimulated by the injection of 10.000 IU hCG in the 24 h after the last injection of rFSH. Then, 36 h after hCG administration, oocyte retrieval was performed by ultrasonography-guided transvaginal aspiration. Estradiol (E2) serum concentrations at oocyte retrieval are reported in Supplementary Table 1 . FF was aspirated from follicles with a diameter of 14–22 mm during oocyte retrieval and was processed immediately. FF was centrifuged for 10 min at 1,500×g, and the supernatant was transferred into fresh tubes and further centrifuged for 10 min at 2,000×g at room temperature for complete removal of red blood cells or debris. Samples were aliquoted and stored at −80 °C until assayed. The following analytes were quantified in FF using specific ELISA kits: IGF-I (DG100B, R&D, Minneapolis, USA), IGF-II (DG200, R&D Minneapolis, USA), IGFBP-1 (EHIGFBP1, Invitrogen, Waltham, USA), IGFBP-2 (E05, Mediagnost, Reutlingen, Germany), IGFBP-3 (DGB300, R&D Minneapolis, USA), IGFBP-4 (EHIGFBP4, Invitrogen, Waltham, USA), IGFBP-5 (EHIGFBP5, Invitrogen, Waltham, USA), IGFBP-6 (EHIGFBP6, Invitrogen, Waltham, USA), IGFBP-7 (EH252RB, Invitrogen, Waltham, USA), HMGB1 (HMGB1 ELISA, IBL, Hamburg, Germany). Technical characteristics of the assays are reported in Supplementary Table 2 . 17β-Estradiol (E2) serum concentrations were assayed the day of oocyte retrieval by means of the ADVIA Centaur Enhanced Estradiol Assay (IVD, Siemens AG, Munich Germany). IGFBP-5 in follicular fluid was always below the limit of detection of the kit and was not considered further (sensitivity: 8 ng/ml). IGF-I, IGF-II, IGFBP-1, IGFBP-2, IGFBP-3, IGFBP-4, IGFBP-6, and IGFBP-7 concentrations were considered also as nmol/L (nM) by multiplying the single values by 0.131, 0.134, 0.033, 0.035, 0.035, 0.038, 0.034, and 0.034, respectively ( 25 , 52 ). Subsequently, the molar ratios of the single IGFs to each IGFBP were calculated to obtain an estimate of bioactive IGF-I and IGF-II in follicular fluid. Statistical analysis was performed using the statistical package GraphPad Prism 10 (GraphPad Software, Boston, Massachusetts USA). Patients with PCOS and controls were stratified based on BMI <25 kg/m 2 (PCOS: 17.6-24.9 kg/m 2 , n=37; controls: 17.2-24.9 kg/m 2 , n=49) (under/normal weight) or ≥25 kg/m 2 (PCOS: 25.0-39.3 kg/m 2 , n=32; controls: 25.0-38.7 kg/m 2 , n=20) (overweight/obese) in order to evaluate differences in IGF and IGFBP content in FF related with BMI. Unpaired Student’s T-test was used to evaluate the differences between PCOS and control groups. One-way ANOVA analysis was performed to study the differences among the following subgroups: underweight/normal weight (BMI<25) control subjects, overweight/obese (BMI≥25) control subjects, underweight/normal weight (BMI<25) PCOS, overweight/obese (BMI≥25) PCOS. Šídák’s multiple comparisons test was used to assess differences between pairs of group means. Pearson’s correlation analysis was performed to investigate the correlation between the analyzed parameters. Only significant correlations are reported. The study was approved by the Ethical Committee of Reggio Emilia (project ID: PCOS2_15_17). Written informed consent was obtained from all subjects as appropriate.