Polycystic ovary syndrome (PCOS) is one of the most common infertility causes in women and is mainly characterized by oligoanovulation, ovarian polycystic morphology and hyperandrogenism (1). Some recent studies have demonstrated an increased risk of autism spectrum disorder (ASD) in the offspring of PCOS patients (2–5).

ASD is a neurodevelopmental disease with core symptoms, including social communication disorder, narrow interests, and repetitive behaviors (6). There are many factors in the etiology of ASD, such as genetics (7, 8), environment (9, 10), and some factors in research progress, including metabolism (11), gut microbiome (12), and endocrine (13, 14). Numerous researchers have reported that prenatal exposure to high levels of androgen may increase the risk of ASD in offspring (13–16). It seems that sex steroids might be the potential link between PCOS and ASD. However, previous studies have not explored the autistic traits of PCOS itself and the association with sex steroids. This study aimed to test the relationship between the PCOS family (themselves and their offspring) and autistic traits and to analyse the correlation between sex steroid and autistic trait scores in patients with PCOS.

There were no significant differences in RBS scores between groups. However, the PCOS group had a higher body mass index (BMI) and AQ score than the control group ( Table 1 ). Furthermore, the PCOS group had significantly higher abnormal scores in social ability, attention and communication ability but not in details and imagination factors in AQ ( Table 1 ). The age of the control group was higher than that of the PCOS group. We further analysed the influence of age on the AQ scores (positive data) of the two groups by using covariance and found that age had no significant influence on the AQ scores (P=0.330), but the difference in the AQ scores between the groups was still obvious (P=0.000). Women with PCOS had a significantly higher levels of LH (7.27 ± 4.96 vs 3.73 ± 1.64, P=0.000, Figure 1A ) and testosterone (1.94 ± 0.66 vs 1.54 ± 0.45, P=0.000, Figure 1E ) than women without PCOS. No difference was found in FSH (PCOS vs control: 7.08 ± 2.24 vs 7.83 ± 3.31, P=0.069, Figure 1B ), estradiol (PCOS vs control: 0.15 ± 0.08 vs 0.13 ± 0.09, P=0.150, Figure 1C ), or progesterone (PCOS vs control: 2.06 ± 1.12 vs 2.30± 1.40, P=0.226, Figure 1D ). Further correlation analysis revealed that there was a significant positive correlation between the total AQ score and LH concentration as well as between the RBS-r score and total testosterone and a significant negative correlation between the RBS-r score and progesterone. However, there were no significant correlations among the total AQ score, RBS-r score and BMI, FSH, E2, F-G score, or acne score ( Table 2 ).

For the offspring follow-up study ( Figure 2 ), 61 patients filled out the questionnaire (30 from the PCOS group vs 31 from the control group), 47 patients refused to fill out the questionnaire (PCOS vs control: 29% vs 18%, P=0.124), 46 patients failed in pregnancy (17 from the PCOS group vs 29 from the control group), and 46 patients could not be contacted with telephone number changes (24 from the PCOS group vs 22 from the control group). There was no significant difference between the two groups in delivery age, complications during pregnancy, special medication during pregnancy, negative events during pregnancy, delivery situation, or postpartum breastfeeding ( Table 3 ). Due to the existence of twins, 73 children were followed up (34 from the PCOS group vs 39 from the control group). There was no significant difference in the age, AQ and ABC scores of children between the two groups ( Table 4 ). Furthermore, after combining the two groups of data to analyse the correlation, it can be found that the AQ scores (P=0.036, r=0.248) and ABC scores (P< 0.001, r = 0.457) of offspring are positively correlated with the RBS-r scores of their mothers. Moreover, the ABC scores of offspring were positively correlated with the childbearing age of their mothers (P=0.008, r=0.306) ( Table 5 ).

In this study, we found that women with PCOS had more autistic traits than those in the control group. The total AQ scores of women in social skills, attention switching and communication were significantly higher than those of the control group. The phenotypes were similar to the clinical manifestations of ASD, for example, social and language communication disorders. There was no difference between the two groups in attention to detail, imagination, or RBS-r scores. During the follow-up, PCOS patients were more likely to refuse the follow-up (29% in the PCOS group vs 18% in the control group). From another perspective, it showed that PCOS patients are more likely to have social dysfunctions.

In 2015, a population-based nationwide study in Sweden involving 23 748 ASD cases and 208 796 controls revealed that children of women with PCOS appear to have a higher risk of developing ASD (3). In 2018, another study based on Clinic Practice Research Datalink in the UK between 1990 and 2014 determined that women with PCOS and their children had a greater risk of autism (24). In addition, several studies showed that women with PCOS experienced higher rates of psychiatric diseases, such as depression and anxiety (25–29). It seems that there are some connections between ASD and PCOS. However, these two studies with large sample sizes were retrospective tracking trials on offspring without showing the sex hormone levels of mothers. In comparison, our research was a prospective study and tried to determine the associations between hormone levels and autistic traits in infertile patients with PCOS and their offspring. Our study provided more evidence that women with PCOS had more autistic traits, especially in social skills, communication, and attention switching. This suggested that women with PCOS are not only prone to have offspring with ASD but also suffer from social impairment themselves.

Why do PCOS patients have more autistic traits? PCOS patients are characterized by a high level of androgen, and ASD is currently considered to be associated with intrauterine hyperandrogen (30–32). Our study showed that the levels of LH and total testosterone in PCOS patients were significantly higher than those in the control group, which was consistent with known conclusions (1). Further correlation analysis revealed that autistic traits were associated with elevated total testosterone and LH.

Some studies have shown that testosterone easily binds to androgen receptors located in the cytoplasm through cell membranes and blood–brain barriers and then enters the nucleus, affecting the regulation of DNA transcription. It thus inhibits oxytocin (OXT), known as “social factors”, and oxytocin has been used to improve the clinical symptoms of ASD (33, 34). Testosterone can also affect the development of the nervous system and nerve contact by mediating programmed cell death, altering neurochemical mediators, and increasing the formation of dendritic synapses (35), and it might thereby affect social skills.

Regarding the relationship between LH and autistic traits, previous literature reported that the serum LH level in female patients with Asperger’s syndrome was higher than that in healthy women (36). LH stimulates ovarian follicular stromal cells to produce more androgen, which then affects the OXT system. This might be the mechanism by which LH affects social function.

In addition, we also found that progesterone was negatively correlated with the RBS-r score. In 2014, Whitaker-Azmitia et al. found that a decrease in progesterone in maternal serum was associated with an increase in the incidence of autism in offspring (37). He explained that progesterone could stimulate the production of more immune T cells, regulate autoimmune function, and combat neuroinflammatory effects. It thereby protects the brain development of offspring. Generally, progesterone rises after ovulation in women. The progesterone level in PCOS patients is low for a long time due to sparse ovulation. This may be one of the reasons why PCOS patients have more autistic traits.

AQ has been widely used in both clinical and research studies as a screen for diagnosis (38). In this study, the AQ scores, which represented sociability, did not show a significant difference between the two groups of children. This might be due to the sample size. However, our data showed that the social behaviors of children might be influenced by the rigid behaviors and delivery age of their mothers. Is age an independent high-risk factor affecting ASD in offspring? According to a literature review, compared with mothers aged 25 to 29, mothers aged ≥ 35 had an increased risk of autism in their offspring (RR=1.52), while mothers aged < 20 had a significantly reduced risk compared with mothers aged 25 to 29 (RR = 0.77). RR increases monotonically with increasing mother’s age. After considering the influence of father’s age and other potential confounding factors, this relationship still exists, which supports the independent relationship between mother’s older age and autism (39). According to another meta-analysis, when the age of mother and father increased by 10 years, the risk of autism of offspring increased by 18% and 21%, respectively (40). Our data showed that children’s ABC scores were positively correlated with their childbearing age (r=0.306, p=0.008), which was consistent with the existing literature results. At the same time, the children’s ABC score was positively correlated with the mother’s own rigid behavior (r=0.457, P < 0.001). This might be related to heredity, or it might be related to some behavioral influences of mothers in raising children. In addition, our study found that PCOS had a high rate of refusing follow-up (29% in the PCOS group vs 18% in the control group), which might be related to a low desire for social communication and the potential underestimation of offspring problems.

Finally, the limitations of our study must be considered: (1) PCOS is a heterogeneous disease. According to the Rotterdam criteria in 2003, the diagnostic evidence for PCOS is rare ovulation, hyperandrogenism, and polycystic ovarian changes. In other words, not all PCOS patients have hyperandrogenism, which leads to abnormal oxytocin and social behaviors. If it can be further subtyped, it will be more convincing. (2) People’s social behaviors are extremely complex. Whether social function can be fully evaluated by simply filling out the scale needs further confirmation. (3) This study mainly focused on infertile patients in the Reproductive Medicine Center, Peking University People’s Hospital. Infertility may be one factor causing social dysfunction. (4) Follow-up is in the form of an online questionnaire rather than an on-site examination of children, so the results may be biased.

In summary, our results demonstrated that PCOS patients have more autistic traits, which might be related to the elevation in testosterone concentration and luteinizing hormone levels, as well as the decline in progesterone levels. Moreover, the autistic traits in the offspring of PCOS patients might be related to the parental high delivery age and high tendency to autistic traits.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by Peking University Institutional Review Board (Approval No. IRB00001052-15031). The patients/participants provided their written informed consent to participate in this study.

LJ is responsible for filling out questionnaires and writing articles. LT is responsible for recruiting patients. JY, XX and FQ are responsible for statistic data. RZ is responsible for reviewing data and revising articles. JW is responsible for revising articles. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.