Thursday, 25 April 2013

Prenatal valproate exposure and autism: reloaded

When does the the term 'correlation does not equal causation' become a moot point? It's a question I've often pondered, having discussed the issue quite a few times on this blog for all manner of correlations and associations linked to autism (sorry, the autisms).

The weight of the heart @ Wikipedia  
Is there, for example, a recognised tipping point where the weight of evidence correlating A with B might actually lead to the consensus that A causes B either wholly or partially?

Yes, I know that science deals with probabilities not absolutes (something which we are all guilty of forgetting from time to time) and that science is generally quite reserved about its findings. But surely as per the example of smoking and lung cancer, there must be a time when the likelihood that A causes B creeps over the 'chance' explanation to something a little more concrete and directional?

The reason for the question(s) follows the publication of a study by Jakob Christensen and colleagues* (open-access) which suggested that in large and pretty well-defined Danish cohort "maternal use of valproate during pregnancy was associated with a significantly increased risk of autism spectrum disorder and childhood autism in the offspring". Regular readers might remember that quite recently there was some similar chatter on this antiepileptic medication based on the Bromley paper (see here) but on an altogether smaller scale compared with the current dataset.

There has been some media attention paid to the recent trial (see here and here) which is perhaps not surprising given the suggestion that approximately 1 in 20 mothers who were using valproate during pregnancy to control epilepsy or seizure disorders subsequently had a child with autism or an autism spectrum disorder (ASD). The actual risk (absolute risk) was quoted as 2.5% and 4.4% respectively based on the 508 children exposed to valproate in-utero. Even the latest 'survey' figures of 1 in 50 children presenting with an ASD in the US are seemingly dwarfed by the Christensen findings.

There are obviously caveats to all this talk about risk, and how risk is risk, not certainty. That also valproate is actually quite effective in controlling cases of epilepsy** is a point which should not get lost in any discussions on risk. Indeed when one reads such studies linking drug A to condition B, it's all too easy to forget that drug A is being taken for a reason; often a very important reason. Physicians generally do not enter lightly into such clinical decisions, particularly in light of past scandals of medication and pregnancy (see here). Not forgetting too that epilepsy can, in extreme cases kill***.

Outside of the autism-valproate link (if I can call it that) the Christensen data also includes some other potentially interesting factoids, as per the suggestion that among children of mums with epilepsy who were not exposed to valproate during pregnancy (n=6152), the absolute risk of a diagnosis of autism and ASD were 1.02% and 2.44% respectively. I hasten to add that I'm not an expert on risk, absolute risk, but 2.44%, by my reckoning, equates as roundabout 1 in 40 with an ASD born to mums with epilepsy. I'm cautious not to read too much into this just in case I've got it completely wrong but if it is correctly interpreted, might imply some greater connection between offspring autism and a maternal history of epilepsy as per previous findings****.

I'm not going to go through all the possible weaknesses in the Christensen paper because the manuscript does that quite well enough itself including some discussion on that folate-autism link. Likewise my previous post on valproate and offspring autism talked about some of the possible mechanisms to account for any effect, so again no need to cover all that ground. There is one tidbit to pick up on: "Valproate is a fatty acid derivative" so the authors report. I've often wondered about this point and the suggested mechanism of seizure control in some cases by use of the ketogenic diet impacting on fatty acids (see the paper by Chang and colleagues*****). Assuming the Chang findings are accurate, does this place more emphasis on the HDAC inhibition side of things when it comes to valproate and offspring autism risk?

The question still remains about the 'correlation does not equal causation' mantra with prenatal valproate exposure and offspring autism in mind. The Christensen paper at the very least, makes a really strong case for a lot more detailed inspection of this potential association as once again the use of pharmacotherapy during pregnancy comes under the spotlight.

Oh, and just in case you thought I was singling out valproate for special attention in relation to autism, have a look at the recent paper by Dheeraj Raj and colleagues****** (open-access) on prenatal antidepressant exposure and offspring autism risk again adding to the previous literature. Indeed it makes me wonder if that environmental exposome fish experiment carried out a while back might well be a model, albeit with revisions, we need to revisit.

A song to close methinks. Something vintage and snazzy today.... Elvis and Viva Las Vegas.


* Christensen J. et al. Prenatal valproate exposure and risk of autism spectrum disorders and childhood autism. JAMA. 2013; 309: 1696-1703.

** Marson AG. et al. The SANAD study of effectiveness of valproate, lamotrigine, or topiramate for generalised and unclassifiable epilepsy: an unblinded randomised controlled trial. Lancet. 2007; 369: 1016–1026.

*** Berg A. Mortality in epilepsy. Epilepsy Curr. 2001; 1: 28.

**** Leonard H. et al. Maternal health in pregnancy and intellectual disability in the offspring: a population-based study. Ann Epidemiol. 2006; 16: 448-454.

***** Chang P. et al. Seizure control by ketogenic diet-associated medium chain fatty acids. Neuropharmacology. 2013; 69: 105-114.

****** Raj D. et al. Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum disorders: population based case-control study. BMJ. 2013; 346: f2059

---------- Jakob Christensen, Therese Koops Grønborg, Merete Juul Sørensen, Diana Schendel, Erik Thorlund Parner, Lars Henning Pedersen, & Mogens Vestergaard (2013). Prenatal Valproate Exposure and Risk of Autism Spectrum Disorders and Childhood Autism JAMA