In their comprehensive meta-analysis of the literature on prenatal risk factors for autism, Gardener et al ^{Reference Gardener, Spiegelman and Buka1} examined and summarised more than 50 such antecedents. Under prenatal factors associated with an increased risk of later autism in the child, Gardener et al listed advanced parental age, maternal use of medication, maternal birth place abroad, bleeding, gestational diabetes, and sibling rank. The authors were rightly cautious to draw strong conclusions from these meta-analytic findings, as the evidence for a role of any of these prenatal risk factors in the aetiology of autism is not sufficient, although on the whole this set of findings suggests that complications during pregnancy in general might contribute to an increased risk for autism.

Fetal sex-hormone profiles might be added to the above list of identified prenatal antecedents of autism. The sex difference in the lifetime prevalence of autism-spectrum disorders, wherein boys and men exceed girls and women by a large margin, is well-known and has partly been attributed to possible influences of early (i.e. organisational) sex-hormone action which contributes to gender differences in neurocircuitry and neuroanatomy. ^{Reference Baron-Cohen, Knickmeyer and Belmonte2}

A role of fetal androgens for autism is suggested by recent research on the second-to-fourth digit ratio (2D:4D), a currently widely studied biomarker. ^{Reference Voracek and Loibl3} Many researchers believe that 2D:4D might provide a useful retrospective window into the prenatal sex-hormonal milieu during critical neurodevelopmental phases of fetal life (i.e. the second trimester) and might be a biomarker for prenatal testosterone exposure and sensitivity specifically. ^{Reference McIntyre4} Human 2D:4D is sexually differentiated (lower in the male than in the female gender), and gender and individual differences in 2D:4D emerge prenatally and are preserved during the growth phases of postnatal life. ^{Reference McIntyre4} Among other supportive evidence for the validity of this anatomical marker, lower (i.e. more male-typical) 2D:4D has been found to be associated with higher sensitivity to testosterone (as effectuated through functional polymorphisms in the androgen receptor gene) and with a higher testosterone-to-oestradiol ratio, as assayed from the amniotic fluid. ^{Reference McIntyre4}

Consistent with the above reasoning and background, Manning et al ^{Reference Manning, Baron-Cohen, Wheelwright and Sanders5} found that children with autism or high-functioning autism (Asperger syndrome), as well as their unaffected first-degree relatives (i.e. siblings, mothers, and fathers), have conspicuously lower (i.e. hypermasculinised) 2D:4D than healthy general population controls. Since then, the gist of this interesting evidence has been independently replicated by some ten further studies (reviewed elsewhere). ^{Reference Voracek6} Inter alia, the evidence base now includes successful replications across ethnicity (East Asians and Caucasians) and similar findings of a low (masculinised) 2D:4D among children with various subtypes of attention-deficit/hyperactivity disorder, ^{Reference Voracek6} all in all indicating that the effect is robust.

Of note, the initial study in this line of research (Manning et al), ^{Reference Manning, Baron-Cohen, Wheelwright and Sanders5} as well as subsequent related research reports, are found in PubMed when using the search terms Gardener et al ^{Reference Gardener, Spiegelman and Buka1} used. So it may well be that Gardener et al did not include this literature in their meta-analysis on the grounds that they categorised it under ‘medical hypotheses’, one of their listed non-eligibility criteria. However, it is interesting that Gardener et al, in their discussion, also noted the following general limitations: (a) only few prenatal risk factors for autism have been examined in multiple studies; (b) generally, fewer than six studies for any of these factors could be included; and (c) when risk factors were examined across multiple studies, the evidence was, for the most part, inconsistent. A formal meta-analysis of the emerging literature on 2D:4D and autism is beyond the present scope, but it is evident from one review ^{Reference Voracek6} that the limitations noted by Gardener et al do not apply for this literature. All in all, the evidence points to a possible role of masculinised sex-hormone profiles, already arising in utero, as a further prenatal risk factor in the pathways leading to the neurodevelopmental disorder autism.