|Gone (data) fishing @ Wikipedia|
I'm a big believer in big data. In particular how, with the right sources, technology, techniques and people, big data might be able to open up some real insights into many important areas including mental health research* and very possibly autism research with a specific focus on the science of biomarkers to aid things like early diagnosis. Indeed, I'm not the only one talking about this (see here).
I've spoken before on this blog about biomarkers for autism and other conditions - the promises, the problems, the future - and how alongside the various autism research banks (genes, brains, etc.) and systems biology chatter, we are just starting to understand the value of those big data resources such as the archived bloodspot samples which so many neonates provide these days.
Indeed with the greatest appreciation for pioneers like Robert Guthrie, I offer a post on an interesting paper by Gerald Mizejewski and colleagues** discussing results suggestive of potential candidate biomarkers for autism based on archived bloodspot samples. I should point out that this is not the first time that Dr Mizejewski has talked about the feasability of biomarkers for autism as per this article*** (open-access) as part of quite a distinguished research career it has to be said (see here) with a specific focus on an interesting molecule called alpha-fetoprotein****.
The most recent paper is unfortunately not at the time of writing open-access, so I'll just go through a few summary points about the work:
- This was a retrospective study based on that tantalising resource of archived bloodspot cards which sit in many a hospital basement. Out of a total case group of 200 families with a child with autism, 40 families with children aged between 3-5 years old were initially contacted for participation. This was eventually whittled down to 16 participants (all diagnosed with autism by the same clinician with the same diagnostic manual) for whom archived neonatal bloodspot cards were available.
- Two age-matched control specimens located immediately before and after the dried bloodspot card in question in the filing system were also chosen.
- A small 3mm punch of the Guthrie cards was analysed by immunoassay which in this case, probed for 90 potential biomarkers covering everything from neurotrophins to cytokines, immunoglobulins to more direct inflammatory markers (including C-reactive protein).
- Some fancy statistical modelling was applied to the obtained results - including Bayesian information criterion (BIC) - which eventually resulted in three models of best-fit based on findings from the bloodspots of those who went to be diagnosed with an autism spectrum disorder (ASD).
- The 'best model' of five compounds included some familiar names to this blog: glutathione-S-transferase (GST), IL-7, IL-5, TNF-beta and something called Lp(a) (lipoprotein a). Most were increased in quantity in the autism samples aside from GST which was decreased.
- There is a very nice illustration in the paper (Figure 3) showing how the potential connections between the biomarkers identified and some of the more biomedical themes of autism research might fit. So we have methionine metabolism mentioned (see here and here), oxidative stress (see here), gastrointestinal comorbidity (see here) and immune activation (see here) to name a few. It's all very systems biology.
- The authors caution that their results are preliminary and that although said biomarkers were modelled as being related to autism they "have not been confirmed to be causative with autism".
Before I get too carried away with this research, there are a few issues worth mentioning. Yes, the sample size was small in this preliminary communication and indeed very little information is provided about participants outside of just them fulfilling the DSM-IV criteria for autism in terms of things like comorbidity. Also why out of 200 families such a small number of participants were eventually included for study.
Indeed there is also an assumption from this study that a biomarker for autism is present in the neonatal phase which for example, might not take into account the issue of behavioural regression that seems to cover quite a percentage of cases.
Whilst the identified best-fit biomarkers are of potentially real interest to autism research as per other similar studies (see here), it is the method and resources used in this paper which is the real 'big data' story allied to all those lovely -omics which reign supreme these days. Parents in many countries will be acquainted with that bloodspot taken during the earliest days of infancy to test for various inborn errors of metabolism such as phenylketonuria (PKU). Many people don't however give a second thought to what happens to those bloodspot cards, and how valuable a resource they might constitute. Although not usually in the business of crystal-ball gazing, I would hazard a guess that we are one day going to hear big news about the big data from those archived bloodspot cards; if not with autism in mind, then something else.
* Ayers JW. et al. Seasonality in seeking mental health information on Google. Am JPrev Med. April 2013.
** Mizejewski GJ. et al. Newborn screening for autism: in search of candidate biomarkers. Biomark Med. 2013; 7: 247-260.
*** Mizejewski GJ. Biomarker testing for suspected autism spectrum disorder in early childhood: is such testing now feasible? Biomark Med. 2012; 6: 503-506.
**** Mizejewski GJ. Biological roles of alpha-fetoprotein during pregnancy and perinatal development. Exp Biol Med (Maywood). 2004; 229: 439-463.
Mizejewski GJ, Lindau-Shepard B, & Pass KA (2013). Newborn screening for autism: in search of candidate biomarkers. Biomarkers in medicine, 7 (2), 247-60 PMID: 23547820