TL;DR: There is some limited evidence of efficacy - especially if the patient also suffers heavy-metal poisoning - but it is rather weak.
[Note: This answer has had a substantial re-write, as I have now read the entire paper. Big props to Konrad Rudolph for finding me a copy.]
The following paper provides a systematic review for off-label and complementary and alternative medicine (CAM) treatments for Autism Spectrum Disorders (ASD):
Ann Clin Psychiatry. 2009 Oct-Dec;21(4):213-36.
Novel and emerging treatments for autism spectrum disorders: a systematic review.
Rossignol DA.
The author, Dr Daniel Rossignol, notes that there is only one FDA-approved treatment for autism (a treatment to lower irritability) and that, perhaps as a consequence, there is a high level of unapproved treatments being given to ASD children (as high as 74%), often without the knowledge of the child's medical doctors.
The story with chelation as a treatment is complex.
By some measures it is very successful:
In Table 1, titled "The top 20 complementary and alternative medicine treatments for children with an ASD, excluding diets", chelation ranks #1, with parents claiming an improvement in the child in 74% of 803 cases (based on tracking by the Autism
Research Institute).
But... there are some big caveats.
I note that the ranking system - based on a survey of parental opinion - isn't very objective or reliable.
The author ranks the quality of the recommendation into grades, and gives the evidence for chelation a grade of C.
Here is how he chose the ranking:
Grade of recommendation
- A At least one level 1a study or two level 1b studies
- B At least one level 1b, 2a, or 3a study, or two level 2b or 3b studies
- C At least one level 2b or 3b study, or two level 4
studies
- D Level 5 evidence, or troublingly inconsistent or
inconclusive studies of any level, or studies reporting
no improvements
- N No studies identified
To understand this, you need to understand the ranking of the underying studies:
Levels of Evidence
- 1a Systematic Review (SR) or meta-analysis of RCTs with homogeneity or Cochrane review with favorable findings
- 1b Prospective high-quality RCT
- 2a SR of cohort (prospective, nonrandomized) studies with
homogeneity
- 2b Individual cohort (prospective, nonrandomized) study or low-quality RCT
- 3a SR of case-control (retrospective) studies with
homogeneity
- 3b Individual case-control (retrospective) study
- 4 Case series or reports
- 5 Expert opinion without critical appraisal or based on physiology or bench research
[I've changed the table formatting to fit the StackExchange mark-up.]
So, we can see that while Level C is not to be simple dismissed, the best possible evidence it could include is an "Individual cohort (prospective, nonrandomized) study or low-quality RCT", which would not normally be considered enough evidence to recommend treatment under an Evidence-Based Medicine approach.
Delving deeper, the paper reviews well over a dozen different papers on chelation and autism.
Confusingly, several of them were on children who had both autism and heavy metal poisoning.
The studies sometimes gave apparently contradictory results, but given the low evidentiary quality of the studies (e.g. no control, small sample), this is to be expected.
The risks of chelation were briefly covered:
The reviewed studies of chelation in children
with an ASD suggest that when properly administered,
side effects of chelation are rare, idiosyncratic, and
reversible.
However, the author noted two children in the USA (one with autism) have died from medication errors while receiving chelation.
The author concludes:
Overall, the collective strength of these studies
investigating the use of chelation in individuals with
an ASD is limited, as some are case reports and none
included a control group. In some of these studies, it is
unclear if the children had elevated heavy metal levels
prior to initiating chelation or if this treatment directly
caused the clinical improvements described. It is also
possible that any improvements observed with chelation
in these children could have been due to some
chelator effect other than the removal of heavy metals
because some chelators also remove pesticides, raise
glutathione levels, and reduce oxidative stress.
It should be noted that placebo-controlled
studies of chelation in individuals with an ASD have not
been performed; a large planned placebo-controlled
study of chelation in children with autism was recently
cancelled by the National Institute of Mental Health.
However, despite the obvious limitations of the reviewed
studies, their cumulative findings suggest that chelation
might be a viable form of treatment in some individuals
with an ASD who have elevated heavy metal burden,
or as suggested by several studies, biochemical
changes suggestive of metal toxicity. Further research
investigating this possibility should thus be considered,
including carefully designed controlled studies of chelation
that include appropriate clinical monitoring as well
as objective prescreening to identify children with ASD
who have concomitant elevated heavy metal burden.