The truth however is that whilst this is an interesting paper, excitement eventually has to give way to cold, hard, unfeeling science as questions about whether diagnosing autism by biochemistry over observation might so easily be done.
The paper is full-text but a short summary follows - hold on to your hats:
- Blood samples from 32 Iranian children with autism (by clinical opinion only) and 31 asymptomatic controls, median age 5-6 years were initially drawn for analysis. In the end, samples of plasma from 28 children with autism and 30 controls were analysed via SELDI- and MALDI-ToF mass spectrometry. In English, these are both methods of ionising a sample using a laser - exploding it into its constituent chemical parts - which are then scrutinised based on the mass-to-charge ratio (m/z) to give a molecular weight measured in Daltons (Da). The clever bit is that by using these technologies the authors were able to fragment and fragment the compounds into their basic amino acid parts via MS/MS. If you think the physics is impressive, you ought to see these systems in action. A confirmatory method was also used based on FTICR mass spectrometry but lets not get too into this aside from saying that the results were about as accurate as you can get with current available technology.
- Peptides were the target of this study; in particular peptide profiles. With the aid of some nifty software which normally accompanies systems like those used, target compounds were isolated between the autism vs. control groups revealing three peptides to be most [significantly] discriminating between the groups showing a mass-to-charge ratio of 1864 (up-regulated in autism), 1978 (down-regulated) and 2020 Da (up-regulated). These mass-to-charge ratios have been simplified given that the sensitivity of the findings went to 3 decimal places.
- All these peptides, 16-17 amino acids long, matched up with known peptide sequences from the C3 complement protein. Think immune system, inflammation and autism (yet again).
- Some data on exactly how discriminatory these peptides were between autism and control samples is given confirming some interesting findings.
Still here? Good. I have tended to focus more on the technology and biomarker 'potential' over the actual results and what they might mean because it strikes me that this is the more important aspect to the study. As impressive as the technology and methodology are though, its already been indicated that (a) the participant numbers were pretty low, (b) no control group was looked at in terms of for example, either learning disability or speech and language disorder without autism (thanks Jon) and (c) the discriminatory power of the study (AUC) was not perfect. I can't argue these points because I agree with them. Added to the fact that unlike in the Yang study discussed a few months ago looking at similar metabolomic methods for diagnosing schizophrenia, there was no training and test sets to confirm the findings, we have to be very careful. I'm not even going to mention the uneven sex ratios between the groups, comorbidity and medication as possible confounders (oops, just did).
What I do like about this study though is that is serves as a template for how this area of work could be done if there was suitable interest in applying metabolomics to autism and related conditions. I know there is some interest because even some of the autism research royalty are starting to venture into these areas. I am sure that Momeni and colleagues created a very large bank of data from this study of which we are only seeing a very small amount. So for example, I am guessing that the analysis was all done in positive ion mode over negative ion mode. If this and other external datasets were to be put together, a potential treasure trove of information would be created.
I still remain excited about this study despite the flaws and potential sources of bias and confounds and look forward to follow-up research in this important field of study.
* Momeni N. et al. A novel blood-based biomarker for detection of autism spectrum disorders. Translational Psychiatry. March 2012.