Testosterone Not Increased in Autistic Children

In their 2005 Medical Hypothesis article, which inspired their questionable 2006 Hormone Research study (background here), Mark and David Geier cited a 1997 French case series (really a letter to the editor by Tordjman et al in the American Journal of Psychiatry) to support their measurement of serum testosterone in children with autism. Geier and Geier wrote...

In addition, Tordjman et al have reported on a case-series of 12 prepubertal autistic children (6-10 years old) in their inpatient child psychiatry department, four of whom the researchers observed to have precocious secondary sexual characteristics (growth of pubic hair, increase of testis volume) that suggest high androgenic activity in autistic disorders.

What Tordjman et al actually did, on the basis of their observation of 4 autistic children with precocious secondary sexual characteristics in their practice, was to measure plasma testosterone and adrenal androgen (presumably DHEA or DHEA-S) in 9 pre- or postpubertal inpatients with autism and 62 matched control children. Because of the possible positive correlation between testosterone and aggression, the investigators divided the 9 autistic children into 3 groups according to their aggressive behaviors. Notably, they observed that autistic children who displayed aggression against others were less likely to demonstrate the typical core symptoms of autism (withdrawal, stereotypy, language dysfunction)—which suggests, perhaps, that these children may actually have an alternative behavioral disorder.

Three of their 9 autistic subjects had abnormally high plasma testosterone levels (Table), given the study's matched reference values. These 3 children all showed aggression against others—meaning, according to the authors, they were less likely to demonstrate typical, core autistic symptoms.

Patient	Serum Testosterone, ng/mL
	Level	Ref Mean ± SD (range)	U Iowa Range
10-year-old boy	0.64	0.06 ± 0.03 (0.01-0.15)	0.18-1.50
17-year-old boy	8.8	5.51 ± 1.27 (0.27-7.50)	3.50-9.70
13-year-old girl	0.5	0.12 ± 0.09 (0.01-0.25)	0.15-0.35

The authors noted that the 10-year-old boy exhibited pubic-hair growth, which is probably not a sign of precocious puberty in boys aged 9 years or older. The 13-year-old girl, whose serum testosterone level exceeded the reference range in the study and that provided by the University of Iowa, also demonstrated a very high level of adrenal androgen, 4.40 ng/mL, at least according to the mean level in the study's control population (mean study reference, 0.88 ng/mL ± 0.39 [range, 0.36-1.70]; U Iowa reference range for DHEA, 1.5-5.7 ng/mL).

What Geier and Geier failed to note, however, in both their Medical Hypothesis and Hormone Research articles, is that a previous study by Tordjman et al (1995) did not find elevated testosterone levels in 31 prepubertal children with autism, when compared with 12 prepubertal subjects who had mental retardation/cognitive impairment* or 10 prepubertal control subjects. The mean levels of plasma testosterone and DHEA-S in 8 postpubertal autistic children were also similar to those in 11 postpubertal normal controls. Tordjman et al concluded from this study that "altered secretion of the androgens is not a common feature of autism."

A PubMed search by this blogger failed to disclose any other studies that assessed testosterone or other androgen levels in pre- or postpubertal children with autism.

DHEA = dehydroepiandrosterone; DHEA-S = dehydroepiandrosterone sulfate.

* The study did find increased plasma DHEA-S levels in prepubertal children with cerebral palsy (among those with mental retardation/cognitive impairment).