Parental Age Effects and Autism Spectrum Disorder: Exploring the De Novo Mutation Hypothesis in Affected Families

Background:  Epidemiologic studies have consistently shown an increased risk for autism spectrum disorder (ASD) with advanced parental age, with many showing independent associations of maternal and paternal age.  Recent genetic studies have pointed towards de novo mutations as possibly associated with risk of ASD, and also associated with advanced paternal age.  The de novo mutation–advanced paternal age model of ASD would imply different expected patterns of paternal age in families with single (e.g. “de novo”) versus multiple (greater genetic susceptibility) children with ASD.

Objectives:  Examine whether paternal age is greater for affected children in simplex compared to multiplex families.

Methods: We examined this in a large collection of nuclear families containing two or more children, with at least one affected with ASD, derived from linking the electronic files of the California Department of Developmental Services to state birth certificates, and matching to siblings by parental information.  A total of 11,859 families were studied with children born 1990-2003, including 11,190 simplex and 669 multiplex families (with 1,369 affected children).  We compared mean paternal and maternal ages at the birth of affected children in simplex versus multiplex families, stratified by birth order and sib-ship size, also calculating strata-specific differences. An overall weighted average of differences was calculated by weighting each strata difference by the inverse of its standard error. We also categorized age and calculated odds ratios (OR) for the likelihood of affected children being in multiplex vs. simplex families, adjusting for parental race and education as well as birth order and sib-ship size.

Results: None of the mean paternal ages by birth order-sibship size strata were significantly increased in simplex vs. multiplex families.  Rather, paternal and maternal age tended to be older for those with affected children in multiplex compared to simplex families (overall: paternal, 32.5 vs. 31.9 years; maternal, 29.6 vs. 29.0 years, respectively, or strata-weighted average differences of 0.82 years for fathers and 0.88 years for mothers). Similarly, the proportion of cases from multiplex families (10.9% overall) generally increased with increasing parental age category. In adjusted logistic regression models, we found a pattern of higher risk of being from a multiplex vs. simplex family with older paternal age (OR=1.34; 95% CI 1.00-1.81 for  ≥ 45y). The pattern was attenuated, but did not change direction, when examining paternal age at birth of first affected child only. The risk also somewhat increased with maternal age (adjusted OR=1.27, 95% CI 0.89-1.81 at ≥ 40y); including both parents in the model did not change the direction of paternal age associations. The patterns did not vary greatly by gender of the affected child.

Conclusions: The results from this study, based on the largest set of simplex and multiplex families assembled for analysis, do not support the hypothesis that the observed paternal age effect for ASD is due to de novo mutation as it was not greater in simplex families, examined by two different analytic approaches.  An alternative model that explains both advanced maternal and paternal age should be sought.