When grandmothers smoked during pregnancy...

Please, do not smoke during pregnancy

ALSPAC - the Avon Longitudinal Study of Parents and Children - continues to give in research terms as today I want to mention the findings reported by Jean Golding and colleagues [1] (open-access) observing "an association between maternal grandmother smoking in pregnancy and grand daughters having adverse scores in Social Communication and Repetitive Behaviour measures that are independently predictive of diagnosed autism."

The study results have garnered quite a bit of media interest (see here for example) with calls for quite a bit more research to be carried out on this important topic. I might also give a hat-tip to Jill Escher and the Escher Fund for Autism who were instrumental in this study. Indeed, her analysis of the Golding results (see here) will no doubt trump my contribution.

The Golding paper is open-access but a few details are worth mentioning:

• As I noted, ALSPAC provided the source data for the study (as it has on quite a few autism research occasions). This time around researchers looked at autistic traits among some of their 14,000 strong cohort comprising "a social communication score, a speech coherence score, a sociability temperament scale, and a repetitive behaviour score." 
• Mothers and fathers of children involved in ALSPAC also provided information about pregnancy including answering questions on whether mothers themselves were/had smoked during pregnancy and also whether their mother (the child's grandmothers on both sides) smoked during pregnancy.
• The data obtained from the autistic traits measures (delivered at various times of the child's development) and the smoking histories were analysed. Data on some 170 participating children actually diagnosed with an autism spectrum disorder (ASD) were also thrown into the statistical mix.
• Results: "We found that two of the four autistic traits in the grandchild (F2) were increased in prevalence if the maternal grandmother (F0) smoked in pregnancy especially if the mother herself (F1) did not herself smoke." Researchers also noted a particular association between grandmother pregnancy smoking and grand-daughters over grandsons, and several potentially important confounding variables were also taken into account. They also concluded that "diagnosed autism was also associated with the maternal grandmother smoking in pregnancy" but express some caution in light of the smaller numbers included in this part of the analysis.

"These results are intriguing" say the authors. Indeed they are. Accepting the reliance on parental report, acceptance of the myriad of other factors that might affect any relationship and the fact that "sets of trait questions were not designed as measures of autistic traits but rather to identify the child’s performance in regard to a large number of attributes at different ages" as a consequence of researchers not originally expecting the prevalence of autism to be anything like it is today(!), further research is indicated. I say this in the context that research looking at any link between maternal pregnancy smoking and offspring autism risk has not consistently found any correlation (see here).

Then to the question of mechanism of effect. Well, researchers talk about "two plausible candidate mechanisms" to account for results. First is "transmission of damage to mitochondrial DNA (mtDNA)" where mitochondria DNA is subject to 'mutation' as a result of exposure to tobacco smoke. The researchers noted that: "Mitochondrial transmission across the generations is exclusively via the mother, so is compatible with our observed associations between maternal prenatal tobacco exposure and adverse scores on Social Communication and Repetitive Behaviour measures in her granddaughters." The second possible mechanism is that of "epigenetic inheritance from one generation to the next." This builds upon something of a continuing debate where structural DNA issues (i.e. mutations) are put to one side in favour of chemical alterations to genes (e.g. the addition of methyl groups) affecting gene expression. It's a topic that has been talked about quite a bit on this blog with reference to autism (see here for example) but still needs a lot more science done on it specifically on transgenerational epigenetic inheritance. Smoking is known to have effects on DNA methylation (see here) but the suggestion is that these or other epigenetic effects could continue down into future generations.

The potential effects of tobacco smoking during pregnancy in relation to second (or even third generation) autism or autistic traits risk requires quite a bit more study. Importantly, if such results are confirmed in future investigations, it opens up a whole myriad of possibilities in terms of how other previous generational exposures outside of just tobacco smoking during pregnancy might have affected future generations. This research area could get very, very complicated indeed.

----------

[1] Golding J. et al. Grand-maternal smoking in pregnancy and grandchild's autistic traits and diagnosed autism. Sci Rep. 2017 Apr 27;7:46179.

----------

Golding J, Ellis G, Gregory S, Birmingham K, Iles-Caven Y, Rai D, & Pembrey M (2017). Grand-maternal smoking in pregnancy and grandchild's autistic traits and diagnosed autism. Scientific reports, 7 PMID: 28448061

Advancing grandparental age and autism risk

The paper by Emma Frans and colleagues* looking at autism risk across the generations is the focus of this post. Published in the journal JAMA Psychiatry alongside a provocative article by Andrea Roberts and colleagues** on maternal exposure to child abuse being "associated" with elevated risk for offspring autism (see here and here), the theme is transgenerational effects and quote: "that your father's and grandfather's lifestyle choices can affect you" as per some of media on this paper.

Ο Κακός Εγγονός @ Wikipedia  

I'm not going to head too heavily into the Frans study because others have already discussed it far better than I ever could (see here and here). Indeed NHS Choices carries a particularly good run-down of the study which is well worth a read (see here).

The main details were that based on an analysis of nearly 6000 cases of autism spectrum disorder (ASD) in Sweden, grandfathers who had fathered their daughter when aged 50 or above were 1.79 times more likely to have a grandchild diagnosed with autism than younger fathering grandfathers.

If grandfathers fathered a son when aged 50 or above, they were 1.67 times more likely to have a grandchild diagnosed with autism (again compared to grandfathers having children when they were younger).

The magic word 'epigenetics' is also mentioned to potentially account for results alongside the mutation side of things. I should also point out that Frans has published on similar things before with schizophrenia in mind***.

I'm interested in studies like the current Frans one despite their reliance on association and relatively limited increased risk of autism. Interested because alongside the 'older dads and autism risk' research (see here), I have actually talked about grandparents and risk of autism and schizophrenia previously on this blog (see here) based partly on some interesting data derived from ALSPAC published by Jean Golding and colleagues**** (open-access) and also that 2011 Frans study. In particular was the emphasis on the Golding 3M - meiotic mismatch methylation - hypothesis used to account for their results on grandmother's age as potentially being relevant to grandchild autism risk (please read the Golding article for more information on 3M complete with nice diagram).

I know it might sound a little far-fetched that the lives of our grandparents might so profoundly be able to affect the lives of subsequent generations but before we put this down to mere coincidence, let me draw your attention to some work that was done on a dark period of quite recent history: the Hongerwinter. The basics: the Dutch famine of 1944, where a Nazi blockade led to the deaths of thousands. As per the often cruel twists of fate, science actually learned something from the suffering of the Dutch people in these dark days. Not only the confirmation that wheat was tied into coeliac (celiac) disease but also the suggestion that famine exposure during a critical period of gestation *might* potentially affect offspring physical and mental health. I've kinda talked about something similar before with 'thin-fat bodies' and David Barker in mind (see here).

Granted in the current Frans study we are heading back even further through the germline as potentially hosting some effect, but to all intents and purposes, the theory is the same as per the intergenerational effects noted in other conditions like depression*****. I suppose one could ask whether specific types of autism might be more related to this grandparental age hypothesis over others. So for example, older grandparents at time of fathering or mothering impacting on the genome of their offspring - themselves then expressing certain traits associated with autism or the broader phenotype (not necessarily hitting the diagnostic threshold) - which are then transmitted (amplified?) to the next generation. Perhaps even some link to things like assortative mating theory too? I'm not saying that this is the only scenario and such 'transmission' works on its own to elevate risk of an autism diagnosis but the theory is an interesting one; even more so if we assume for example, that the autism and schizophrenia spectrums might not necessarily be poles apart (see here).

By the same token one might also extend such a hypothesis to include other variables other than just parental age at offspring conception. The availability of and exposure to certain food in these 'olden days', exposure events to pollutants, pharmaceuticals (yes, we did have them then) or lifestyle factors such as smoking and drinking habits, various psychological and somatic stressors; the list is seemingly endless. In at least some of these factors, there are subtle clues to potential future directions for autism research and beyond already being examined (see here and here). Assuming also that epigenetics might be tied into all of this, we also open up the concept of epigenetic reprogramming of the germline as per the very interesting article by Petra Hajkova****** (open-access); something which I think might be/have been discussed at the recent Environmental Epigenetics symposium hosted by the MIND Institute.

But let's not get too carried away with this area of inquiry or where it potentially leads in terms of the autism 'blame game'. Although I've not been able to find specific figures, I assume the actual numbers of grandfathers who fathered their children aged 50+ years is probably not going to be all that frequent if more current rates, at least here in the UK, are anything to go by (see this paper by Bray and colleagues*******). And then we have to wonder whether other variables might come into play such as the effect of paternal age at conception on birth factors such as birth weight or time of gestation even fecundity itself and how that might relate to autism risk.

Such transgenerational effects whilst interesting, should also not detract research attention away from other more here-and-now possibilities which might affect autism risk as per the recent valproate work or indeed all that immune activation research currently on-going. Neither should it deflect attention from the fact that the autism numbers are really starting to get quite serious - 1 in 50 US kids (with caveats) - and what needs to be done (a) asking why there is such an increase in cases and (b) to ensure the relevant help and support is available to all who need it.

To finish a very catchy tune from Jake Bugg - Lightning Bolt.

----------

* Frans EM. et al. Autism risk across generations. A population-based study of advancing grandpaternal and paternal age. JAMA Psychiatry. March 2013.

** Roberts AL. et al. Association of maternal exposure to childhood abuse With elevated risk for autism in offspring. JAMA Psychiatry. March 2013.

*** Frans EM. et al. Advanced paternal and grandpaternal age and schizophrenia: a three-generation perspective. Schizophr Res. 2011; 133: 120-124.

**** Golding J. et al. Parental and grandparental ages in the autistic spectrum disorders: a birth cohort study. PLoS ONE. 2010; 5: e9939.

***** Warner V. et al. Grandparents, parents, and grandchildren at high risk for depression: a three-generation study. J Am Acad Child Adolesc Psychiatry. 1999; 38: 289-296.

****** Hajkova P. Epigenetic reprogramming in the germline: towards the ground state of the epigenome. Phil. Trans. R. Soc. B. 2011; 366: 2266-2273.

******* Bray I. et al. Advanced paternal age: How old is too old? J Epidemiol Community Health. 2006; 60: 851–853.

----------

Frans, E. (2013). Autism Risk Across GenerationsA Population-Based Study of Advancing Grandpaternal and Paternal AgeAutism Risk JAMA Psychiatry DOI: 10.1001/jamapsychiatry.2013.1180