Predicting gluten avoidance outside of coeliac disease

The suggestion that dietary intervention may be a useful tool in ameliorating some of the core and/or peripheral signs and symptoms associated with some cases of autism spectrum conditions has been banded around for quite a few years now.

Of course like every 'intervention' suggested for autism, there is no universal 'one size fits all' as per the heterogeneity and comorbidity which accompany the condition. Combined with a lack of information on the whys and hows of dietary intervention, one perhaps needs to be a little bit cautious on how data on the use of gluten- and casein-free diets, for example, are reported and used.

With this in mind, I stumbled upon an interesting article by Tanpowpong and colleagues* looking at the various symptoms and conditions which might be predictive of the use of a gluten-free diet outside of the more mainstream application in cases of coeliac (celiac) disease (CD). I apologise that I can't post the full-text article but do let me summarise the study and main findings:

• A database was created based on all patients who presented at Massachusetts General Hospital, Boston (USA) between January 2000 and December 2010 for initial evaluation of possible coeliac (UK spelling) disease (N=720, aged 1-19 years).
• Lots and lots of information were collected from patients including demographic data, information on signs and presenting symptoms and the medical stuff like serological markers including an old friend, tissue transglutaminase (tTG). Dietary gluten avoidance and physician-recommended / parent instigated use of gluten-free diet (GFD) data were also collected as patients were grouped into three age bands: toddlers / pre-school (aged 1-5 years), school-aged (6-11 yrs) and adolescents (12-19 yrs).
• Results: only 579 children made it to the final analysis based on things like getting a CD diagnosis elsewhere or issues with confirming use of a GFD. Most patients initially presented with abdominal complaints or issues with bowel movements. Forty-three of the 579 children (7.4%) had some history of dietary gluten avoidance without a confirmed diagnosis of CD. Interestingly, quite a few factors seemed to related to avoidance including: having another family member with CD, complaints of irritability and poor temper control, a history of food allergy and a diagnosis of pervasive developmental disorder (PDD). Just in case you aren't aware, PDD is a sort of wrap around term for autism, Asperger syndrome and the like according to current diagnostic schedules. I say 'current' because as everyone keeps saying, the times they are a changin'.
• Indeed when it came to the statistics of prediction regarding gluten avoidance, a PDD diagnosis seemed to be particularly predictive, with an unadjusted odds ratio, OR of 6.76 (CI: 2.97 - 15.37). With the application of multivariate logistic regression, the OR ratio for PDD came down slightly (OR = 5.34) but still remained highly significant as a predictor. 
• It continued. One hundred and forty three patients who were negative for either serological or endoscopic markers of CD were inspected further - 32 of which implemented a GFD without a diagnosis of CD.  Predicting physician-led or parent-initiated use of a GFD revealed again that PDD was again a significant factor; indeed the most significant factor (OR = 7.9) among the few other descriptions.

There are a few messages to take from this data. For some of them I am in agreement with the authors, for others perhaps not so much.

Primary among these messages is that there is a perception that a confirmed diagnosis of CD might not be the only condition where a GFD might be viewed as being useful. We've already seen hints of this in the past few years; starting with the study by Jessica Biesiekierski and colleagues on non-coeliac gluten intolerance early in 2011 and continuing with the 'game-changer' that was Anna Sapone and colleagues on the spectrum of gluten-related disorders published this year (2012). With no disrespect intended to anyone I've talked to down the years about gluten-free diets and autism, there was always a degree of 'snobbery' about gluten and CD as in, if it 'aint coeliac disease, don't bother with a GFD. How things change.

I assume from this study that information about the gluten- and casein-free diet was/is quite widely disseminated in PDD circles. I was very interested in why the word 'autism' does not appear in this paper outside of the references (including our own trial). Don't get me wrong, PDD is factually correct when it comes to the blanket description of autism - something I've used myself - and the various other conditions in diagnostic manuals such as DSM and ICD (bearing in mind that they seemed to have used ICD-9 for their diagnoses). That being said, I found it slightly strange that further analysis was not undertaken on the individual diagnoses included under the PDD umbrella. At the very least it would have been useful to know for example, whether those with an autism diagnosis or those with a diagnosis of Asperger syndrome were more likely to be in the gluten-free dieting group as a function of behaviour and/or communication as per the diagnostic differences. Indeed one could inquire whether all the participants were cases of Rett syndrome or childhood disintegrative disorder which are also included in the PDD diagnostic fold? That and the lack of information about how many participants with PDD had CD or issues with tTG for example, are important omissions in my view.

I would perhaps also take issue with one sentence in their text "Moreover, no clear positive effects from the use of a gluten-free, casein-free diet have been shown in children with PDD". No it hasn't, but (a) referencing our ScanBrit trial and our conclusions that stated "Our results suggest that dietary intervention may positively affect developmental outcome for some children diagnosed with ASD" is probably not the best example of this message and (b) the fact that most interventions for autism have no "clear positive effect" in a population-wide sense, makes me a little uneasy about the wording of the conclusions being reached. I've felt this uneasiness before as per the Mayo Clinic conclusions on dietary intervention for autism (see here); to quote from a favourite film saga: 'I have a bad feeling about this'.

Finally(!), there is a distinct lack of follow-up information included in this paper. OK, you could argue that was not what they set out to look at, and I'm fine with that. Having said that, covering the number of patients they did and for the number of years across the study information, you would perhaps expect to see some small detail if only to say that no-one in the non-CD group showed any benefit whatsoever from the use of a gluten-free diet. Or did they...?

I hope that I've not turned this post from science into a rant. This is still an informative paper and adds to the growing science base about adoption of a GFD in cases of non-CD. That also and reiterating a possible need to screen for CD in cases of PDD?

* Tanpowpong P. et al. Predictors of gluten avoidance and implementation of a gluten-free diet in children and adolescents without confirmed celiac disease. The Journal of Pediatrics. May 2012.
DOI: 10.1016/j.jpeds.2012.02.049