The association between autistic traits and serum testosterone, oxytocin, and androstenedione levels in prepubertal male drug naive children with attention‐deficit/hyperactivity disorder

Children with attention‐deficit/hyperactivity disorder (ADHD) might have similar problems as in autism spectrum disorder (ASD) and show impairment in social behaviour. Also, there is a relationship between social relationship skills and ToM (theory of mind) skills of children with ADHD. Besides, ASD is associated with prenatal exposure to high levels of androgens, and oxytocin plays a role in the modulation of emotions, coping with stress, and social behaviour like ASD. In this study, the relationship between autistic traits and serum oxytocin, testosterone, and androstenedione levels in prepubertal male drug naive children with ADHD has been investigated.


| INTRODUCTION
Attention-deficit/hyperactivity disorder (ADHD) is one of the most common neurodevelopmental disorders in childhood and is seen in 3-7% of school-age children (American Psychiatric Association, 2013). ADHD leads to severe social, academic, and psychological impairments in all stages of child and adolescent development.
Some researchers consider that children with ADHD have inadequate social functionality due to its associated behavioural patterns (attention deficit, difficulty in family, and lack of social skills) (Gentschel & McLaughlin, 2000). Impaired social functionality is often seen as a rejection by peers and engaging in conflict with other children and adults. Social functionality is a concept, which is based on cognitive and social abilities and affected by individual characteristics and environmental factors (Happé & Frith, 1996). Children with ADHD have difficulties behaving emphatically in social situations, as they find it difficult to put themselves in place with another person. These children also have difficulties in interpreting social clues (Braaten & Rosén, 2000). Besides, it has been shown that children with ADHD might experience problems similar to autism spectrum disorders (ASD) involving communication and social reciprocity problems and repetitive behaviours (Hattori et al., 2006). It seems worthwhile to consider the presence of socio-communicative and/or repetitive traits in children with ADHD who do not meet diagnostic criteria for ASD. In a research, it has been reported that children with ADHD might display clinically significant symptoms of ASD in three core categories (social interaction, social communication, and repetitive behaviour) (Nijmeijer et al., 2009). On the other hand, some other studies have reported clinical symptoms only in social interaction (Clark et al., 1999) or only in repetitive behaviour (Polderman et al., 2014). In a study using the social responsiveness scale (SRS), it has been found that children with ADHD have higher mean scores on the SRS compared with the control group, and 54.7% of these children have experienced subthreshold or clinically significant social problems (Ayaz et al., 2013).
Social cognition is the ability of an individual to think about the thoughts, feelings, and behaviours of himself/herself and others (Flavell et al., 1990). It is known as a basic concept for the theory of mind (ToM). ToM, in other words having a mental theory or mentalization capacity, is being able to realize that people (others) other than himself/herself have a mind different from his/her own, understand mental states such as intent, belief, desire, and knowledge of himself/herself or others and represent them mentally (Baron-Cohen, 2000). In terms of psychopathology, the concept of ToM disorders was first used to explain the symptoms in children with ASD ( Baron-Cohen, 2000). In literature, there is a limited number of studies in which ADHD patients were evaluated through false belief tests for ToM skills. The disorders of the prefrontal cortex affect ToM skills; therefore, any functional or structural problem in the nerve pathways of this region can disrupt ToM skills, which is essential for social functionality (Koenigs et al., 2007). Clinical studies have shown that patients with ADHD might show impairment in social behaviour, although their cognitive abilities are normal (Koenigs et al., 2007). It has been indicated that ADHD, which is related to frontostriatal dysfunction, is associated with the poor performance in ToM tests (Mary et al., 2016).
Testosterone is a steroid hormone, and it is the end product of the hypothalamus-pituitary-gonadal (HPG) axis located in the reproductive and immune systems. Testosterone plays a vital role in the growth and differentiation of genital and extragenital organs (Baron-Cohen, 2002). One of the theories explaining ASD is the "extreme male brain" theory proposed by Simon Baron-Cohen (Baron-Cohen, 2002). This theory is based on the observation that ASD is predominantly seen among men, and it states that ASD is associated with prenatal exposure to high levels of testosterone (Manning et al., 2001). Manning et al. showed that the 2D:4D ratio, which is the ratio between the length of the second and fourth finger (2D/4D), is negatively correlated with autistic traits. Studies have also shown that the 2D:4D ratio reflects foetal testosterone levels (Manning et al., 2001). Another evidence of this theory is that some autistic traits are increased in patients with congenital adrenal hyperplasia . It has been shown that serum testosterone, free testosterone, dehydroepiandrosterone, and androstenedione levels have been significantly increased in patients with ASD (Geier & Geier, 2007). Androstenedione is a precursor of testosterone, and in hormone-sensitive tissues, uptake circulating androstenedione is converted to the active androgen metabolites. In a study, whether adrenal and gonadal sex hormones, especially testosterone, oestradiol, DHEA-S (dehydroepiandrosterone sulfate), and androstenedione, are higher in individuals with ASD compared with controls was evaluated (Ruta et al., 2011). It is found that serum androstenedione levels have been significantly higher in the ASD group. There is only one study in the literature evaluating the relationship between salivary testosterone levels and autistic traits in healthy controls (Takagishi et al., 2010). The results have determined that saliva testosterone levels have been positively associated with autistic traits in both men and women when the effect of age is controlled.
Oxytocin is an important hormone for attachment and breastfeeding. It is synthesized in magnocellular neurons in the paraventricular (PVN) and supraoptic nuclei (SON) of the hypothalamus, transported to the posterior pituitary and secreted from there (Lee et al., 2009). Oxytocin acts on central brain regions and peripheral tissues, and it plays a role in the modulation of emotions, coping with stress, and social behaviour (Lukas & Neumann, 2013). Besides, oxytocin has been thought to have an important role in the aetiology of ASD due to its effects on emotional and social behaviour (Stavropoulos & Carver, 2013). There are several studies evaluating serum oxytocin levels in ASD. In some of them, it has been reported that plasma oxytocin levels are lower among individuals with ASD compared with healthy ones (Modahl et al., 1998). However, in another study, it has been found that plasma oxytocin concentrations are higher among individuals with ASD compared with healthy ones (Jansen et al., 2006). The disrupted oxytocin system hypothesis in ASD has led to studies in which oxytocin application developed social skills among individuals with autism. In a study conducted on adults with ASD, it has been presented that the administration of intranasal oxytocin has improved social cognition scores (Anagnostou et al., 2012).
To date, few studies have investigated the role of oxytocin in ADHD. Taurines et al. have compared plasma oxytocin levels in ADHD, ASD, and healthy individuals. They have found out that plasma oxytocin concentrations are significantly lower in ADHD patients than healthy controls and individuals with ASD (Taurines et al., 2014). Park et al. have associated specific oxytocin receptor polymorphisms with social cognitive deficits in children with ADHD diagnosed based on the DSM-IV criteria (Park et al., 2010).). In another study, it has been determined that serum oxytocin levels are negatively correlated with aggression, but positively correlated with empathy in the ADHD group (Demirci et al., 2016).
Our knowledge of neurohormones in relation to ADHD is based on few studies that only a limited number of them focused on comorbidity or traits associated with ADHD. To the best of our knowledge, this is the first study investigating the relationship between autistic traits and serum oxytocin, testosterone, and androstenedione levels in prepubertal boys with ADHD. The primary aim of this study is to investigate the relationship between autistic traits and ToM skills with serum oxytocin, testosterone, and androstenedione levels in prepubertal male patients with ADHD. It is hypothesized that in prepubertal boys with ADHD associating higher autistic traits and false ToM tests would be correlated with low oxytocin, high testosterone, and high androstenedione levels.

| Procedure
The study was conducted from May 2017 to March 2018. Children, who were diagnosed with ADHD according to DSM-5 criteria for the first time in the Department of Child and Adolescent Mental Health and Diseases, Hacettepe University Faculty of Medicine, have been referred as the study group. All the patients with ADHD who were suitable for the inclusion criteria were asked to participate in the study. Totally, 400 participants that met the criteria were requested to join the study, and finally, 83 of them accepted the request. Eighty-three patients between the ages of 72 and 120 months who voluntarily accepted to participate in the study have been included in. Detailed information about the purpose and the method of the study was provided with the patients and their families. Besides, written consent was obtained from patients and their families. Psychiatric interviews were conducted with patients who volunteered to participate in the study and who gave consent and their families to confirm the diagnosis of ADHD and to evaluate concomitant psychiatric diseases. In the psychiatric interview, ADHD diagnosis and accompanying psychiatric disorders have been determined according to the "The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5)" diagnostic criteria by applying a semi-structured Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL). Wechsler Intelligence Scale for Children-IV (WISC-IV) has been applied to all participants by the clinical psychologist in order to determine the intelligence score. Also, a social responsiveness scale (SRS) has been given to the parents in order to evaluate the patients' autistic traits. In this study, WISC-IV has been used to determine the intelligence score and SRS has been used to evaluate the patients' autistic traits as it was Turkish normed. Autistic traits that are evaluated with SRS and ToM tests have been performed by a clinical psychologist. In this study, the theory of mind has been measured by storytelling. Visualization with toys has also been used in order to make the story be understood more easily. In this process, while conveying the story verbally and by playing the toys, the tester asks regular "Reminder questions" to the child simultaneously to make sure that the story is understood correctly. After making sure that the story is understood correctly, the test question stage is started. At this stage, to succeed in the test, it is expected from the child to observe the information that one of the story characters does not have and accordingly predicts the path that the character will follow correctly. After that, a series of questions are asked to confirm that the correct answer of the child is not a coincidence. When the correct answer is given in all stages including the Confirmation Question, Reality Question and Memory Question stages, the test is considered to be completed successfully. The patients included in this study are divided into three groups as high autistic traits, autistic traits, and no autistic traits according to SRS and ToM-true ToM and false ToM-tests. To evaluate the pubertal stage, Tanner staging has been performed at Child and Adolescent Mental Health Outpatient Clinic, and an assessment has been made about adolescence.
In this study, there have been several exclusion criteria including psychoactive drug use, female gender, adolescence, accompanying intellectual disability, specific learning disorder, developmental coordination disorder, behavioural disorder, tic disorder, autistic spectrum disorder, psychotic disorder, bipolar disorder, generalized anxiety disorder, separation anxiety disorder or obsessive-compulsive disorder, metabolic or endocrine disease, and history of head trauma or chronic neurological disease. The diagnoses for exclusion have been made by using K-SADS-PL and DSM-5 diagnostic criteria.
The research has been approved by Hacettepe University Non-Interventional Clinical Research Ethics Committee with the number GO 17/420 on 04.07.2017. Written consent has been obtained from the participants and their families.

| Schedule for Affective Disorders and
Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL) Gökler et al. (2004) adapted the scale into Turkish in 2004 as "ÇDSG-SY," which had been originally developed by Kaufman et al. (1997), and the updated version according to DSM-5 criteria were adapted by Ünal et al. (2019). ÇDSG-SY is able to evaluate 20 different psychiatric diagnoses.

| Wechsler Intelligence Scale for Children-IV (WISC-IV)
The standardization and norm studies of WISC-IV, which was reorganized in the United States in 2003 (Grizzle, 2011), introduced to use, and has been the latest version of the Wechsler Intelligence Scales, was conducted in our country between 2007 and 2011 by the Turkish Psychologists Association (Öktem et al., n.d.).

| Screening and Rating Scale for Behavioural Disorders in Children and Adolescents Based on Turgay-DSM-IV (T-DSM-IV-S)
This scale developed according to DSM-IV criteria consists of a total of 41 items, 9 questioning attention deficit, 6 questioning excessive mobility, 3 questioning impulsivity, 8 questioning oppositional defiant disorder, and 15 questioning behavioural disorder (Turgay, 2004).

| Social responsiveness scale (SRS)
The scale that was originally developed in 2000 by Constantino as Social Reciprocity Scale later named as the "SRS-Social Responsiveness Scale." In 2003, Constantino stated that the scale could be used to evaluate the OSBlike symptom cluster (Constantino et al., 2003). There are 65 items in total, including 6 items on the social use of language, 39 items on observable social behaviours, and 20 items on autistic behaviours. It is accepted that patients who score between 60 and 80 points on the scale show signs of autism and patients who get more than 80 points show severe autism symptoms.

| Theory of mind evaluation
Since ToM switched to psychopathology terms through developmental psychology theories, the first tests developed are "false belief" tests to understand the developmental process. It was originally developed by Wimmer and Perner (1983).

| Tanner staging
The pubertal stage has been evaluated by using the schematic drawing of secondary sex characteristics associated with five Tanner staging of pubertal development (Tanner, 1986).

| Hormonal evaluation
Participants were asked to apply to the outpatient unit in a relaxed state, from 08:00 AM to 10:00 AM, and by avoiding physical exercise and stress. In this manner, a 10 ml venous blood sample was taken to measure the serum oxytocin concentration. Blood samples were centrifuged for 15 min at 1000g within 30 min for the evaluation of serum oxytocin. Separated serum samples were stored at À80 C until analysis. Serum oxytocin concentrations were measured in the biochemistry laboratory by using the ELISA method.
For the measurement of plasma concentrations of testosterone and androstenedione, 10 ml venous blood samples were taken into purple capped tubes with EDTA. Blood samples were centrifuged for 15 min at 1000g within 30 min. The separated plasma samples were stored at À80 C until analysis. Plasma testosterone and androstenedione levels were measured in the biochemistry laboratory by using the liquid chromatographytandem-mass spectrometry (LC-MS-MS) method.

| Statistical analysis
The obtained data have been evaluated by using "SPSS (Statistical Package for Social Sciences) for Windows 22.0 (SPSS Inc, Chicago, IL)." Descriptive statistics are presented as mean AE standard deviation (minimummaximum), frequency distribution, and percentage. Pearson chi-square test has been used to evaluate the categorical variables. The suitability of variables to normal distribution has been evaluated by using visual (histogram and probability plots) and analytical methods (Kolmogorov-Smirnov test). Since serum testosterone value is normally distributed, Student's t test is used to compare it between false and true groups of ToM, and ANOVA (analysis of variance) is used to compare it among autistic traits status groups according to SRS. After ANOVA, if an overall significance is observed, pairwise post hoc test is performed by using Tamhane's T 2 test. As serum oxytocin and androstenedione measurements are not normally distributed, the Kruskal-Wallis tests are conducted to compare the parameters between the results of ToM and autistic trait according to SRS groups. The Mann-Whitney U test is performed to test the significance of pairwise differences by using Bonferroni correction to adjust for multiple comparisons. The relationship between the variables has been evaluated by using the Spearman correlation test. An overall p value of less than 0.05 is considered to show a statistically significant result. However, the significance level has been accepted as p < 0.017 after Bonferroni correction in binary comparison with Mann-Whitney U test.
While the total mean score of the SRS has been 64.5 AE 21.0 (28-127), the mean score of the "social behaviours" subscale has been 38. . In addition, the mean score of "social use of language" has been 7.1 AE 2.4 (1-12), and "pathognomonic autistic behaviour's" mean score has been 18.8 AE 7.6 (6-40). "Autistic symptom status" is determined according to the cut-off scores of the SRS. While 44 (53.0%) of the patients have no autistic symptoms, 19 (22.9%) have mild, and 20 (24.1%) have severe autistic symptoms. In the theory of mind, the ToM assessment of 48 (57.8%) of the patients is "true," whereas the remaining 35 (42.2%) are "false." The number and percentage of patients, who have the true ToM test and are without autism symptoms (SRS score < 60) (n = 44 and 100%), are found to be significantly higher than the patients with autistic symptoms (SRS score is 60-80) (n = 4 and 21%) and patients with severe autistic symptoms (SRS score > 80) (n = 0 and 0%). A statistically significant difference has been found between the autistic symptom levels of the patients evaluated with SRS and ToM (p < 0.001) ( Table 1). Mean score of SRS is significantly higher in the patients with the wrong ToM than the patients with the true ToM statistically (score of  vs. 49.9 AE 9.5 [28-68], p < 0.001) ( Table 2). In our study, no significant difference has been found between ADHD subtypes and SRS scores (p = 0.34) and the theory of mind tests (p = 0.56). A positive correlation has been found between ToM and SRS (r = 0.90, p < 0.0001). A positive correlation has also been found between androgens (testosterone and androstenedione) and SRS or ToM. There has been a negative correlation between oxytocin and ToM or SRS or testosterone. In addition, ToM has had a negative correlation with age (r = À0.24, p = 0.03) ( Table 3).
An increase in testosterone and androstenedione levels and a decrease in oxytocin level have been determined as independent predictors to develop an autistic trait in ADHD according to SRS ( (Table 4).
An increase in testosterone and androstenedione levels and a decrease in oxytocin level and age have been determined as independent predictors for poor perfor-  (Table 5).

| DISCUSSION
In this study, it has been aimed to investigate the serum testosterone, androstenedione, and oxytocin levels in an ADHD population, which was grouped according to the severity of the autistic traits. The whole sample was divided into three subgroups according to the social responsiveness scale (SRS) scores and two subgroups according to ToM test results. It has been found out that patients with more severe autistic traits according to both SRS scores and ToM test results have higher serum testosterone, androstenedione levels, and lower serum oxytocin levels. The difference between groups has been statistically significant. In our study, no significant difference has been found between ADHD subtypes and SRS scores and the theory of mind tests. In the literature, it has been shown that children with ADHD are likely to have autistic traits (Ayaz et al., 2013). To our knowledge, in the literature, there has been no study in which the social functionality and autistic traits of ADHD patients were evaluated together with both SRS and theory of mind tests. We have evaluated autistic traits with both SRS and ToM test and found higher SRS scores in impaired ToM groups.
There are several studies showing the relationship between increased androgen levels and autistic traits in the literature. Some psycho-endocrinology studies have shown that saliva testosterone level is strongly associated with social cognition (Van Honk et al., 1990). Also, Takagishi et al. investigated the relationship between salivary testosterone levels and autistic traits in adults. In that study, a positive correlation has been found between testosterone levels and autistic traits (Takagishi et al., 2010).
Studies investigating the association between testosterone and autistic traits are rooted in the extreme male brain theory (EMB). In this theory, Baron-Cohen proposed two basic cognitive styles as male brain type and female brain type. Male brain type defines individuals in whom systemizing is more developed than empathizing. On the other hand, female brain type defines individuals in whom empathizing is more developed than systemizing, and those are measured by empathizing quotient (EQ) and systemizing quotient (SQ) (Baron-Cohen, 2009). The EMB theory states that ASD is associated with reduced empathy and increased systemizing that is hypothesized to be related to prenatal testosterone exposure (Auyeung et al., 2006). The fact that ASD is approximately four times higher in men suggested that an effect such as prenatal androgen exposure might be significant in this disorder. In clinical studies, androgen levels were examined in patients diagnosed with ASD. For example, Tordjman et al. (1997) showed a significant increase in plasma testosterone levels in one out of every three children diagnosed with ASD compared with healthy children (Tordjman et al., 1997). Similarly, in patients diagnosed with ASD, DHEA, and serum testosterone levels were shown to increase significantly compared with age-and gender-specific reference intervals (Geier & Geier, 2006). All mentioned studies emphasize the association between prenatal androgen exposure and autistic traits and ASD, which are highly comorbid with ADHD. In our study, there has been a significant positive correlation between serum testosterone and androstenedione levels and higher autistic traits in ADHD patients. This study extends the findings of previous researches from ASD to autistic trait in ADHD and supports the idea that early foetal environment might play a role in the development of cognitive phenotype associated with autistic traits coexisting with ADHD.
Oxytocin plays a role mainly in the regulation of emotions, coping with stress, and social behaviour. It has been suggested that the dysfunction of social skills in ASD might be associated with the dysfunction of central oxytocin system (Lukas & Neumann, 2013). Some studies have reported lower oxytocin levels in patients with ASD compared with healthy individuals (Alabdali et al., 2014). Also, various other studies have shown that serum oxytocin concentrations had a negative correlation with autism symptoms in children with ASD, and individuals with low oxytocin levels show autism symptoms more frequently (Zhang et al., 2016). In this study, it has been determined that serum oxytocin levels have been significantly lower in patients with higher autistic traits, which is consistent with the results of previous studies. It is thought that the negative correlation between oxytocin and autistic traits in ADHD, the result obtained in this study, is accurate since we defined autistic traits by using two different measures (SRS and ToM tests), which are correlated with each other and neurohormone levels including oxytocin.
Although there are many studies about neuroendocrine that correlate autistic traits and ASD, their results seem to be contradictory. It is sensible that a number of limitations might have influenced the contradictory results obtained in the previous studies. In our study, it has been aimed to overcome some of the limitations such as puberty, gender, the presence of comorbid psychiatric disorders, and/or mental retardation. For this reason, only prepubertal boys were included in the study, considering that the hormone changes might be observed in adolescence. Since many psychiatric diseases affect hormone levels, patients with additional psychiatric diagnoses other than diagnoses frequently accompanied to ADHD like ODD and CD disorder and who had a score below 70 in the WISC-4 intelligence test were not included in this study. We are of the opinion that the method that has been used to define the autistic traits in our study is the methodological power. Autistic traits were defined by using SRS, and this definition was supported by ToM tests. To our knowledge, there is no study in the literature investigating the relationship between ToM skills, autistic characteristics, and serum oxytocin levels in ADHD patients. This study is the first investigating this relationship. There are three main limitations to our study. The first of them is the absence of a healthy control group. The second one is not to have another clinical condition frequently accompanied by autistic traits. Therefore, further studies are needed to understand the specificity of these results for autistic traits. Lastly, in this study, only boys and children diagnosed with prepubertal ADHD were included. Therefore, it is not possible to generalize the study results to the whole ADHD diagnosis group.
As a conclusion, it has been suggested that oxytocin, testosterone, and androstenedione could play roles in impaired social functions such as autistic traits, and thus, they could be related to social relationships and theory of mind skills in patients with ADHD. Further progress in understanding the neurobiological processes underlying autistic traits can help develop more targeted interventions and also enhance the early detection of children before the phenomenological manifestation of autismlike traits in ADHD.