Regression and autism: then and now

The words 'regression and autism' mean different things to different people. To some, regression relates to diagnosis and in particular that little-known diagnostic box, childhood disintegrative disorder or Heller syndrome. To some, regression is a debate of definition and characterisation in the course of autism symptoms. To some, regression relates to events and sudden loss of functions and onset of symptoms.

In this post I am specifically focusing on the definition and characterisation of 'regression' relating to autism, and in particular how views and opinions on regression might have changed over the years. I would stress that I am not talking about regression occurring as a consequence of an intervention or some other man-made change i.e. an adverse drug reaction or a change of school / environment; although I do realise how difficult it might be to separate out such factors from symptom course.

Looking into the early years of autism research, there were reports and reports which detailed regression-type presentation of autistic symptoms - that is, a loss of previously-acquired skills. Interestingly, many of these reports whilst mentioning regression, tended to characterise behavioural regression as being due to things like epilepsy / seizure-type disorders or diagnosed as Heller syndrome or in some cases 'blamed' on co-morbidity such as tuberous sclerosis. Puberty was also a potential explanator in later childhood (I will perhaps come back to this at a later date). I may be wrong but it seemed like autism was somehow not seen as being able to be a regressive condition in its earliest years.

Sometime during the early to mid-1990's a change seemed to be apparent. First, researchers started talking more about regression in autism. Papers such as this one by Fred Volkmar (1992) described how uneven patterns of symptoms and developmental regressions were actually quite prevalent in autism and, furthermore, seemingly more prevalent in the 'lower-functioning' ability range.

At about this time, Cathy Lord and colleagues also published their revisions to the ADI (the ADI-R) (1994). Those in the know will know that ADI-R includes items discussing the issue of regression - or more accurately 'loss of language/other skills'. I am not saying that this was a significant addition to the ADI-R, merely that one of the gold-standard assessment instruments now included 'regression' as part of its schedule.

The floodgates seemed to be opened after this point and regression was something that appeared in various papers, texts and books as being, quote: "a typical event in the natural course of autism". So where has the concept of regression in autism gone since?

Well first of all some authors have suggested that regression might be hard-wired as part of the symptom presentation of some people with autism - a sort of genetic predisposition to regressive presentation of some symptoms. The same authors do not rule out regression occurring as a result of environmental events as a 'second-hit'. Specific assessment tools have also been developed to more accurately quantify regression in autism; one example is the Regression Supplement Form - a sort of add-on for the ADI-R.

We don't have a precise figure on the percentage of autism cases associated with regression partly because regression is such a nebulous term covering all kinds of different regressive symptoms. Estimates of varying degrees of language regression in autism range from 15% in this study to 20% in this study.  Regression in social interactive behaviours seems to be following suit in terms of percentages, bearing in mind that language and social behaviours seem to be linked.

Various other co-morbid conditions have also been tied into regression in autism. A family history of coeliac disease, inflammatory bowel conditions and rheumatoid arthritis were associated with language regression in this study - oh, also specific functional bowel symptoms too.

The debate has moved on from asking if regression is present in autism, to things like ascertaining sub-groups within the regression camp. I remember being very interested in seeing the work from Sally Ozonoff and colleagues on their description of the 'delays-plus-regression' phenotype for example.

A possible association between regression in autism and comorbid mitochondrial disease has been the source of quite a lot of speculation quite recently and perhaps represents regression in autism turning full circle and going back to 'caused by other things' (assuming that mitochondrial disorder is comorbid and not a 'cause' of autism). I can't really add too much more to this debate because I don't know enough about mitochondrial disorders aside from what I read in places like this.

So there you have it. Regression has come along way in autism over the past 20 years or so. Having said that, can anyone tell me if DSM or ICD criteria for autism mention regression, or propose to mention it in future editions?