Is There Evidence of Intergenerational Influences on Autism?

Background: Mammalian experiments and human observations indicate that population variation in development and health is influenced by the early-life experience of parents and ancestors (Pembrey et al J Med Genet 2014; 51:563). For example, smoking in pregnancy by either grandmother is associated with the grandchildren’s growth patterns in a sex-specific manner (Golding et al Am J Hum Biol. 2014; 26:731).

Objectives: We aimed to determine whether similar intergenerational effects of prenatal smoking could be demonstrated for autism.

Methods: We have used grandmothers’ smoking behavior to test for associations with autism and autistic traits in their grandchildren. The Avon Longitudinal Study of Parents and Children (ALSPAC), which enrolled pregnant women in 1990-1992 has information on prenatal exposure of father (n=9677) and mother (n=12,707). Follow up of their offspring includes four traits independently predictive of autism, plus autism diagnosis itself. Trait measures used, identified in ALSPAC (Steer et al PLoS One 2010; 5(9):e12633), cover Social Communication, Speech Coherence, Repetitive Behaviour, and Sociability Temperament. The most extreme ~15% of each were used as outcomes in the analysis. Adjusted analyses took account of various features of the grandparents including years of birth, ethnic origin, ages, social categorisation and education level; separate analyses subdivided the data according to the sex of the child, and whether or not the study mother herself smoked prenatally.

Results: Three of the four traits tested showed an association with maternal grandmother smoking when pregnant. For two traits the effect sizes were greatest if the study child was a girl and the mother did not smoke e.g. for Social Communication the adjusted odds ratio (AOR) was 1.67 [95% CI 1.25, 2.25] (P= 0.001) and for Repetitive Behaviour 1.48 [1.12, 1.94] (P=0.005). The trait Speech Coherence showed a marginal effect with grandsons AOR 1.24 [1.02, 1.50] (P=0.030) but not granddaughters. 170 children monitored through pregnancy had a diagnosis of autism, again associated with maternal grandmother smoking in pregnancy when the mother did not smoke; AOR 1.53 [1.06, 2.20], (P=0.022). There were no associations with prenatal exposure of the study father.

Conclusions: There is evidence that some of the more extreme levels of behaviour traits predictive of a high risk of autism have an association with mothers (but not fathers) who have been exposed to their own mother smoking during pregnancy. Assuming that these findings are replicated in other studies, they raise the possibility that transgenerational exposures of other substances may influence the risk of traits leading to autism.