Uncovering Sex-Steroid Related Conditions in Women with Autism: A Latent Class Analysis

Background:   Elevated levels of prenatal androgens, such as testosterone (T), during a critical period are hypothesized to contribute to the etiology of autism spectrum conditions (ASC), as fetal T shapes neurological development. Prenatal androgens also contribute to the programming of the hypothalamic-pituitary-gonadal (HPG) axis. In females, elevated levels of prenatal androgens also contribute to the etiology of polycystic ovary syndrome (PCOS). Following an earlier study in 2007, the present study further explores the link between ASC, HPG programming, and PCOS in a larger sample of women.

Objectives:   (1) To test if elevated rates of sex steroid-related symptoms and conditions in women with ASC replicate in a larger sample; and (2) to assess sex steroid-related symptom patterns within women with ASC and controls.

Methods:   We tested 2 groups of women, screened using the Autism Spectrum Quotient (AQ): Group 1: n=415 women with ASC (mean age (standard deviation) 36.39 (±11.98) years); Group 2: n=415 controls (mean age (standard deviation) 39.96 (±11.92) years). All participants completed the Testosterone-related Medical Questionnaire (TMQ) online, which was designed to measure steroid-related symptoms and conditions. Fisher’s Exact Test was used to assess group differences in reported rates of steroid-related symptoms. We assumed that response patterns on the TMQ reflect underlying steroid circulation and sensitivity, and a multiple-group latent class analysis (LCA) was used to identify differences in latent class structure between women with ASC and controls.

Results:   We confirmed significantly higher frequencies in females with ASC of (1) sex steroid-related medical conditions: epilepsy (p=0.016), amenorrhea (p=0.0002), dysmenorrhea (p<0.0001), severe acne [contraceptive pill users (p=0.053), non-users (p=0.002)] and anorexia (p<0.0001); (2) non-heterotypical gender identity and sexual orientation, related to fetal testosterone: gender dysphoria (p=0.0004), transsexualism (p=0.030), and differences in sexual preference (Χ²; p<0.001); and (3) sex steroid-related developmental changes: precocious puberty (p=0.003), and early growth spurt (p=0.002). We conducted a multi-group LCA on 11 items linked to sex steroids, comparing women with ASC to controls. In both groups, two latent classes (typical and ‘steroidopathic’) emerged. A significantly higher percentage of women with ASC than controls fell into the steroidopathic class characterized by higher posterior probabilities for each steroid-related condition (ΔG²=15, df=1, p=0.0001).

Conclusions:   Increased exposure to prenatal androgens atypically programs the HPG axis, resulting in conditions such as amenorrhea, early or late menarche, and hirsutism. We found that adult women with autism reported higher rates of such sex steroid-related conditions than controls.  Increased reports of sex-steroid related conditions in women with ASC are consistent with the hypothesis of elevated fetal testosterone in the development of ASC.