1. ¹Department of Psychiatry, One Gustave Levy Place, Mount Sinai School of Medicine, New York, NY, USA
2. ²Department of Psychological Medicine, King's College London, Institute of Psychiatry, De Crespigny Park, London, UK

Correspondence: JM Silverman, E-mail: jeremy.silverman@mssm.edu

In a historical population-based cohort study, we reported a nearly sixfold increased risk of autism in the offspring of Israeli fathers aged 40 years and older compared to those under 30 years,¹ a finding observed in several other samples.^{2, 3} While we suggested that accumulating de novo mutations in the spermatogonia of older fathers might account for this finding,¹ an alternative explanation is that deferred paternity is a function of genetic traits present in the fathers of autistic children. Personality traits such as aloofness, rigidity, anxiousness and hypersensitivity, collectively called the broader autism phenotype (BAP), are overexpressed in unaffected relatives of autism probands.^{4, 5} Delayed paternity may be a social effect of these phenotypes in familial autism. To test this hypothesis, we assessed the relationship between these BAP traits and age at first procreation, regardless of affected status, in a sample of 131 fathers and mothers of autistic children.

As described elsewhere,⁶ families with at least two children meeting autism or autism spectrum disorder (ASD) criteria specified below were recruited as part of an ongoing family study. We used the Autism Diagnostic Interview-Revised⁷ to assess and diagnose Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) autism as well as prespecified criteria for other ASDs.⁶ The Modified Personality Assessment Schedule-Revised^{4, 5} was administered independently to parents and informants (usually the other spouse) to measure rigidity, aloofness, anxiousness and hypersensitivity. Parent and informant reports were combined to form best estimate scores, reflecting the trait as absent, present but mild or severe (scores of ‘0’, ‘1’ and ‘2’).

BAP characteristics were assessed in the parents of 131 families with two or more cases of ASD (n=275). While the majority of these cases presented with DSM-IV autism (n=228), other ASDs were also present. The male to female ratio among the ASD cases was 5.11:1 and the mean age of cases was 7.35±5.98 years. The mean age at first paternity was 32.1±5.1 years. The mean age at maternity was 29.8±4.56 years. On average, parents were assessed 10.58±6.34 (fathers) and 10.31±5.76 (mothers) years after the birth of their first child.

Table 1 shows the zero-order Pearson correlations between age at paternity and BAP characteristics in fathers and, separately, in mothers. These correlations ranged from −0.124 to 0.156, and there were no significant positive associations with increasing paternal age. Years of education was not associated with age at first paternity (r=−0.13, not significant) and made no discernable difference when included as a covariate in a partial correlation analysis.

Table 1 - Zero-order Pearson correlations between the age at paternity and the BAP characteristics in fathers and, separately, in mothers.

Full table

This null result suggests that genetic autism risk is unrelated to paternal age in familial ASD. While this finding does not preclude a different result in a different sample, if our negative findings reflect the general reality for these families, it would be unexpected to observe a positive result in a sporadic sample and, furthermore, difficult to explain through the genetic mechanism examined here. Rather, these results support our original de novo mutation hypothesis regarding autism risk and paternal age. According to mutagenesis, de novo mutations spontaneously arise and accumulate in successive generations of spermatogonia (sperm-producing cells), increasing the likelihood that men will pass on a point mutation or structural chromosomal abnormality as they age.^{8, 9} As higher rates of de novo mutations have been implicated in autism cases as compared to controls,¹⁰ it is likely that such a mechanism may influence our paternal age finding. BAP traits were no less apparent in the mothers and fathers as age at paternity increased (that is, genetic liability was equivalent); thus it seems probable that the interaction of these mutations with pre-existing autism susceptibility genes may have a cumulative effect, resulting in a genetically distinct and perhaps more easily expressed form of the disorder.

Our sample was primarily composed of parents of affected sibling pairs, and thus may over-represent cases in which autism is, in part, genetically determined. The increased genetic loading likely in these families should make associations between BAP traits and paternal age particularly apparent. That no association was nevertheless found emphasizes the need for further exploration of de novo events and their potential role in paternal age–autism risk association.