Obesity and hyperandrogenism are implicated with anxiety, depression and food cravings in women with polycystic ovary syndrome

PCOS is associated with mood/eating disorders. Negative body image due to obesity, acne, hirsutism seems to play significant role, but hormonal derangements are probably implicated. To investigate the relation between insulin resistance (IR), obesity and hyperandrogenism with mood and eating disorders in women with PCOS. Forty-nine (60.5%) PCOS women and 32(39.5%) age- and BMI-matched healthy controls were enrolled. Emotional/food disorders were evaluated by using self-administered questionnaires: Eating Attitudes Test (EAT)-26, Beck Depression Inventory-II (BDI-II), Hamilton anxiety scale (HAS) and Food Craving Questionnaire-Trait (FCQ-T). The two groups had no significant differences regarding age, BMI and HOMA2-IR. PCOS women had significantly higher DHEA-S (p < 0.0001), Δ4Α (p < 0.0001) and Testosterone (p < 0.0001). When the two groups were subclassified according to the BMI, in lean (BMI < 25 kg/m2) or overweight (BMI ≥ 25 kg/m2), no significant differences were found with respect to EAT-26 and HAS. BDI-II was associated with obesity (overweight vs lean PCOS: 20.5 ± 6.4 vs 9.8 ± 3.9; p = 0.037) and hyperandrogenism (overweight PCOS vs overweight controls: 20.5 ± 6.4 vs 14.8 ± 8.1; p < 0.0001; lean PCOS vs overweight controls: 16.7 ± 4.7 vs 14.8 ± 8.1; p = 0.01). Additionally, a significant correlation between BDI-II and DHEA-S (rho = 0.305; p = 0.006), Δ4Α (rho = 0.259; p = 0.02) and Testosterone (rho = 0.328; p = 0.003) was reported. FCQ-T was associated with obesity (overweight PCOS vs lean PCOS: 47.6 ± 9.9 vs 29.3 ± 8.9; p < 0.0001; overweight controls vs lean PCOS: 45.5 ± 15.7 vs 29.3 ± 8.9; p < 0.0001), whereas a correlation between FCQ-T and BMI (rho = 0.593; p = 0.0001), waist circumference (rho = 0.554; p = 0.0001) and HOMA2-IR (rho = 0.328; p = 0.003) was documented. Obesity and hyperandrogenism increase the risk of depression and food cravings in women with PCOS, leading to a vicious circle of further aggravation of obesity and metabolic syndrome.


Introduction
Polycystic ovary syndrome (PCOS) is the most common endocrine disorder of women during the reproductive period, with an estimated prevalence of 9-18% and the highest rates presented in Western countries [1][2][3]. According to the National Institute of Child Health and Development (Revised 2003 Consensus), PCOS is defined by clinical or biochemical evidence of hyperandrogenism combined with oligomenorrhea or amenorrhea, in the absence of other known endocrine disorders [4]. Obesity, insulin resistance (IR), hyperinsulinemia, metabolic syndrome are comorbidities commonly associated with PCOS [5][6][7]. Except from hormonal derangements, PCOS seems to affect the psychological status of patients. Studies have shown that obesity, acne, hirsutism, are associated with negative body image and low self-esteem, leading to excessive anxiety, depression and eating disorders, such as food cravings and binge eating [8]. Moreover, apart from body's alterations, IR and impaired ghrelin's and cholecystokinin's secretion may also predispose to the development of psychological and eating disorders in patients with PCOS [9,10]. Moreover, increased testosterone's levels have been found to aggravate a bulimic behavior, whereas anti-androgen treatment seems to mitigate it [11,12]. Consequently, eating disorders further aggravate obesity, IR and metabolic syndrome, leading to a permanent vicious circle.
Although the association between PCOS and mood/eating disorders has been recognized, the existing data are limited and the underlying pathogenetic mechanism remains unclear and debatable. The aim of this study was to investigate the possible correlation between obesity, IR, hyperandrogenism and anxiety, depression, eating disorders and food cravings in patients with PCOS.

Subjects
In this prospective study, consecutive women with PCOS referred to our Obstetrics and Gynecology department during a 2-year period (September 2021-August 2022), were recruited. Diagnosis of PCOS was based on the criteria of Rotterdam PCOS 2003 consensus group [4]. Oligomenorrhea defined by presence of eight or less menstrual cycles during the preceding year and amenorrhea by absence of menstrual cycles for six months or longer in a previously normally ovulating-cycling woman [13]. Clinical hyperandrogenism defined by presence of hirsutism as estimated by the modified Ferimman Galway (F-G) score (moderate to severe acne, androgenic alopecia, and seborrhea) [14]. The control group composed of healthy volunteers with regular menses (24-32 days) and absence of clinical and biochemical signs of hyperandrogenism, with a normal sonographic appearance of the ovaries. Exclusion criteria: Age <18 or >35 years old, time of menarche less than three years before the initiation of the study, state of pregnancy, Cushing syndrome, Diabetes Mellitus, hyperprolactinemia, thyroid dysfunction, non-classic congenital adrenal hyperplasia or other endocrine disorder, known or clinically apparent psychiatric or behavioral disease, acute or chronic ailment, rheumatologic diseases, liver disease, treatment with metformin or any hormonal medication including combined oral contraceptive pills in less than six months before study entry, treatment with antidepressants, anxiolytics or any other psychiatric medication, or alcohol consumption greater than 20 gr/week were excluded. In total, forty-nine women with PCOS were finally enrolled (Table 1). Based on the Propensity score of women with PCOS as calculated by using the IBM SPSS version 28.0, 32 age-and body mass index (BMI)-matched healthy subjects were selected to participate as controls. The sample size was calculated with G * Power Software, using the Wilcoxon-Mann-Whitney test, 85% power, alpha 0.05. Data from the study of Deeks et al. [15] were considered for this calculation. Thus, a sample size of minimum 50 individuals (25 per group, effect size d: 0.7993) was required for this study. The protocol was in accordance with the Helsinki Declaration for Human Studies

Study protocol
At the initial visit, medical and reproductive history of PCOS women and controls was noted. Clinical examination was performed by the same physician (K.S.). Anthropometric measurements were recorded (the mean value of two measurements for each subject) ( Table 1). For all women weight was measured to the nearest 0.1 kg using a beam balance (Beam Balance 710; Seca, Birmingham, UK), while height was measured to the nearest 0.1 cm using a stadiometer (Stadiometer 208; Seca, Hanover, MD). Maximum waist circumference (WC) was measured in duplicate using a flexible measuring tape. Abdominal obesity defined by WC > 88 cm [16]. BMI (in kg/m 2 ) was used as a marker of overall obesity. Values between 18 and 24.9 were characterized as normal, while values ≥ 25 were characterized as overweight respectively. Blood samples were drawn at 8:00 am after an overnight fast.

Statistical analysis
Statistical analysis was performed by using SPSS 28.0 (SPSS software; SPSS Inc, Chicago, IL, USA). Data are represented as number with proportions for categorical variables and means ± standard deviation (SD) for continuous variables. Quantitative variables were compared with Student's t test or Mann-Whitney U test for normally and non-normally distributed variables respectively. Comparisons of quantitative variables among 3 or more groups were performed by using the One-Way ANOVA or Kruskal-Wallis test for normally and non-normally distributed variables respectively. The post-hoc analysis was adjusted by the Bonferroni's correction for multiple pairwise comparisons. Qualitative variables compared with corrected Chi-squared test or two-sided Fisher's exact test, as appropriate. Correlations between parameters were estimated by using the Spearman's correlation coefficient. All tests were two sided and p values < 0.05 were considered to be significant.

Comparisons between controls and women with PCOS
No statistically significant differences were found between the two groups regarding age, BMI and HOMA2-IR. Patients with PCOS had significantly higher DHEA-S (p < 0.0001), Δ4Α (p < 0.0001), Testosterone (p < 0.0001) and significantly lower SHBG (p < 0.0001), as well as a tendency towards lower WC (p = 0.095) ( Table 1).

Differences between control group and PCOS patients with respect to the obesity's status (lean vs overweight)
When the analysis was performed between the four groups of patients [lean controls (n = 14,17.3%), overweight controls (n = 18, 22.2%), lean PCOS (n = 26, 32.1%) and overweight PCOS (n = 23, 28.4%)], significant differences were documented regarding BMI, WC, HOMA2-IR, DHEA-S, Δ4Α, Testosterone and SHBG (Table 2). Furthermore, significant differences were verified regarding the FCQ-T and the BDI-II, whereas a tendency towards a significant difference was noticed concerning the HAS score.
Among the 4 groups, no statistically significant difference was found with respect to age and EAT-26 ( Table 2). The post-hoc analysis of the pairwise comparisons are presented in Table 3 HAS had also a significant positive correlation with BDI-II (rho = 0.384, p < 0.0001) and FCQ-T (rho = 0.38, p < 0.0001).

Discussion
According to previous studies, women with PCOS seem to exhibit an increased risk for emotional distress, defined by excessive anxiety, depression and food disorders like  cravings and bulimia [8,24]. Several pathogenetic mechanisms have been proposed, but data are still controversial. Moreover, it has not been clarified whether different psychological manifestations share the same etiological factor. In our study, women with PCOS had significantly higher concentrations of the androgenetic hormones Testosterone, Δ4Α, DHEA-S and lower levels of SHBG compared to control subjects. Regarding eating disorders, EAT-26 score did not differ statistically between women with PCOS and the control group. Furthermore, the correlation between EAT-26 and WC or HOMA2-IR though significant, was weak in both cases. This test was originally developed and studied in adolescent females with clinically diagnosed eating disorders [25]. Subsequently, its use was expanded to include overweight and obese individuals as well [26][27][28][29]. Several studies introduced EAT-26 as a screening tool for identifying eating disorders in a pre-clinical stage [30][31][32] and showed good specificity, but moderate sensitivity [28]. It seems that in non-clinical settings, the low performance value of EAT-26 and particularly its high false negative rates reflect the inadequacy of using self-report tools for estimating abnormal behaviors in subjects who deny such abnormalities, or present a lack of awareness regarding them [33]. Since our cohort was exclusively composed of clinically healthy patients, this could explain the limited accuracy of EAT-26 in evaluating eating disorders in our PCOS women.
Considering the HAS anxiety score, no significant difference was observed between PCOS patients and control subjects, whereas a significant but weak correlation between HAS and BMI was noticed. Likewise, in a previous study of Sulaiman et al. no difference at the anxiety scores was documented between patients with or without PCOS [34]. Furthermore, Watrowski R et al. showed that the presence of anxiety was not associated either with physical alterations and hirsutism, or androgenic hormones [35]. The authors concluded that TSH was the only hormonal predictor of overall negative affectivity and anxiety, and loweconomic status overtrumped the impact of hormones on the psychological well-being. Our results confirm that the existence of PCOS per se is not associated with increased anxiety, as estimated by the HAS score.
Interestingly, it has been shown that women with PCOS have eight times higher prevalence of depression than women without PCOS [36]. Recently, a case-control study confirmed the increased risk of depression in PCOS women [34]. Moreover, a meta-analysis of 18 studies by Cooney et al. showed that women with PCOS had 3.78-fold higher risk of any depressive symptoms and 4.18-fold higher risk of moderate/severe depressive symptoms [36]. Nonetheless, a concrete etiology of PCOS-associated depression has not been postulated yet. Previous studies implicated IR as a causative factor. Greenwood et al. conducted a randomized control study with 738 PCOS women and strongly suggested that PCOS-associated depression derived from IR [37]. Specifically, the homeostatic model assessment of insulin resistance (HOMA-IR) index was associated with a 2.3-fold odds ratio for depression and this association remained significant after adjusting for age and BMI [37]. Apart from IR, increased Testosterone levels have also been considered to aggravate the risk of depression in PCOS [38], while treatment with contraceptives has been found to significantly decrease the depressive symptoms [39]. According to our study, obesity seems to be a predisposing factor of depression, as overweight PCOS had significantly  higher BDI-II in comparison to lean PCOS while the two groups had no differences in testosterone levels (Table 3).
Except from obesity, high concentrations of androgenetic hormones probably contribute to the development of depression, as BDI-II was found to positively correlate with DHEA-S, Δ4Α and Testosterone. Additionally, the higher BDI-II of lean PCOS compared to lean and overweight controls, indicates that the magnitude of hyperandrogenism on the development of depression potentially overcomes that's of obesity.
Previous studies have shown that young women with reproductive abnormalities have an increased risk for food cravings. In particular, hyperandrogenemia and menstrual irregularities have been significantly associated with greater food cravings, independently of age, BMI and PCOS status [8]. The relationship between higher androgens' levels and uncontrollable desire to eat has already been reported in women with bulimia nervosa, while treatment with antiandrogens or androgens' antagonists improves the bulimic behavior [11,12]. However, Huijgen NA et al. in a more recent study did not observe an increased total food intake between hyperandrogenic and non-hyperandrogenic PCOS patients [40].
The results from our study are in accordance with the latter, as no correlation between FCQ-T and androgenic hormones was documented. Furthermore, no significant difference was established regarding FCQ-T between lean PCOS and lean controls, as well as between overweight PCOS and overweight controls respectively. On the contrary, FCQ-T was found to be related with obesity and IR, as a positive correlation with BMI, WC and HOMA2-IR was confirmed. In addition, the higher FCQ-T in overweight PCOS and overweight controls against lean PCOS patients, verified the potential association of obesity with food cravings. Obesity as a binge eating disorder, has already been recognized by El-Haschimi K et al. [41].
To the best of our knowledge, it is the first time that not a single test, but a combination of them has been applied in order to comprehensively evaluate mood and eating disturbances in PCOS. Moreover, the classification into lean and overweight individuals, discriminated the impact of obesity from that's of androgens on the development of psychological and eating disorders in PCOS. Thus, women with PCOS were found with more severe depressive symptoms and food cravings according to the BDI-II and FCQ-T respectively. Depression was associated with obesity and hyperandrogenemia, with the latter being the prominent factor, whereas food craving was related with obesity and IR.
In conclusion, women with PCOS probably need a careful psychological assessment in order to identify any mood or eating disorder. Management of the emotional distress along with the treatment of obesity, IR and hyperandrogenism is necessary to break the vicious circle between hyperphagia and further aggravation of obesity and metabolic syndrome.

Compliance with ethical standards
Conflict of interest The authors declare no competing interests.
Consent to publish All authors contributed to the interpretation of the data and reviewed and approved the manuscript.
Ethical approval The study was performed according to the principles of the declaration of Helsinki as revised in 1983 and approval was obtained by the Ethics Committee of Alexandra General Hospital of Athens, Greece.
Informed consent Informed consent was obtained from all individual participants included in the study.