Some cases clearly present as a disease state, with autoimmune issues, neurological issues, etc. Whether these are comorbid conditions or really part of autism itself isn't clear at this point. The fact that children also commonly 'regress' following periods of normal development also argues in favor or some sort of disease state, for the same reason we call Parkinson's a disease.
I have a strong hunch that "Autism" as we know it will continue to be teased apart into a variety of more specific diagnosis. We've already seen this to some extent where other developmental disorders once lumped in with ASD have been specifically named following etiological discovery.
From my own experience: My son is autistic. I suppose he'd fall under the 'low functioning' side of the spectrum. He can't talk, doesn't understand language receptively very well, is incapable of controlling himself when excited, etc.
Anyway, I try to be involved (or at least aware) of every aspect of his education, and in doing so regularly visit his classroom. What strikes me most is that there is such large variability between 'autistic' individuals. There are some commonalities, unquestionably, but there are just as many differences. However, through regular observation I have noticed trends amongst the kids, or 'clusters' as I call them. I am sure social factors play a substantial role in the variability, but trying to see past that, it seems to me like there are different 'types' of autism.
Of course, I'm no autism researcher (though I've been involved in research at the MIND Institute), and don't pretend to be any sort of neuroscientist with only a BS in biochemistry and a few classes involving neurobiology years ago -- but I'm not the only one who notices this. It almost seems 'common knowledge' to the Occupational/Speech/Adaptive ...etc therapists and educators who deal with dozens of kids on a daily basis. That these subgroups of autism aren't more commonly researched/published really surprises me, and trust me, I look for them.
In time we'll uncover much more about neurological/genetic/epigenetic/microbial factors involved in these 'types' and probably result in a number of different formal diagnosis/treatments. I would not be surprised if those considered 'high functioning' are found to differ enough to warrant a separate diagnosis from those on the 'low functioning' side.
edit: a word
That regression seems to divide the Aspergers from the hard-core Autistics. Our son has Aspergers, as well, but is quite intelligent. Getting him to shut up about his own little world can be a challenge, though. He is learning to cope, and dealing with his issues is far preferable to low functioning autism. (unless of course one likes having ones house tore up most of the time, "potty breaks" in random spots around the house, and trying to keep a child from running out of the house who may or may not be dressed...)