Description of the condition
According to the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5), autism spectrum disorders (ASD) are a set of pervasive neurodevelopmental conditions that are characterised by difficulties in social interaction and communication, and the presence of restricted, repetitive behaviours (i.e. stereotypies) (APA 2012). Historically, Asperger’s disorder, autistic disorder, atypical autism, and pervasive developmental disorder not otherwise specified (PDD NOS) were considered examples of conditions falling within the autism spectrum. However, the latest edition of the DSM (DSM-5) now uses the term ASD for diagnostic classification (APA 2012).
The severity of ASD varies considerably, and there is great variability in symptoms and manifestations. Individuals with ASD have difficulty in social-emotional reciprocity (e.g. such as participating in reciprocal conversations or maintaining eye gaze), communicating verbally and non-verbally, forming and maintaining relationships, and understanding the social behaviour of others (Shattuck 2007; APA 2012). Individuals with ASD also exhibit preoccupations and restricted, repetitive patterns of interest and behaviours, which may include strong adherence to routines and stereotyped speech and motor movements (Lecavalier 2006; APA 2012). Some may present with behavioural symptoms (e.g. such as irritability, aggression, anxiety, self-injury, and hyperactivity), but these features are not part of diagnosis, rather they are co-occurring symptoms. The specific causes of ASD are currently unknown. However, genetic factors and prenatal and perhaps postnatal environmental factors are believed to contribute to the onset of ASD, although the role of environmental triggers remains uncertain (Hallmayer 2011).
There are many interventions for ASD, and while some such as early intensive intervention are effective in improving communication, social interaction, and behaviours, none are capable of producing complete remission of all symptoms. Pharmacologic interventions are often prescribed to individuals with ASD, primarily to target specific associated symptoms or co-occurring features. Currently, however, no pharmacologic interventions target core symptoms of ASD. Individuals with ASD also frequently use complementary and alternative medicine (CAMs); approximately 75% of children with ASD use CAMs (Hanson 2007). Examples of CAMs that are used for ASD include exclusion diets, essential fatty acids, multi-vitamins, acupuncture, auditory integration training, and chelation therapy. To date, there is no consistent evidence that CAMs are effective interventions for core features and associated behaviours of ASD (Nye 2005; Gold 2006; Millward 2008; Cheuk 2011; Sinha 2011; James 2011).
Description of the intervention
Chelation therapy involves the administration of a chelating substance that binds to heavy metals, such as lead and mercury, and is then excreted in urine. Chelating substances are approved for treating cases of heavy metal poisoning. However, they have also been used for unapproved reasons and conditions such as Alzheimer’s disease, coronary heart disease, and ASD (Ernst 2000; Dans 2002; Sinha 2006; Hedge 2009). Types of chelating substances used to reduce heavy metal poisoning are outlined in Table 1 . The administration of chelation therapy for approved uses occurs in a very controlled environment, which is different to the process followed by practitioners administering chelation for unapproved uses, such as for ASD.
Cases of heavy metal poisoning, such as acute lead poisoning, require urgent hospitalisation and administration of chelating substances either intravenously or by deep intramuscular injection for four hours. During the treatment of acute lead poisoning, blood and urine are monitored constantly, as significant shifts of the heavy metal can occur between the blood and central nervous system with dire consequences. Moreover, because minerals and metal ions are essential elements that serve important functions in multiple biological processes, excessive removal can result in deleterious consequences; for example, recently a child with ASD experienced fatal myocardial necrosis resulting from hypocalcaemia after receiving chelation therapy (Brown 2006). Additionally, the local health department is notified, as the source of the lead needs to be identified and dealt with appropriately.
The unapproved use of chelation therapy, for example when used as an intervention for ASD, may involve practitioners using various chelating substances and unlicensed routes of administration (e.g. such as through the rectum or skin) to remove reported excess levels of mercury or other heavy metals or both (Semple 2011). Prior to treatment, individuals with ASD may undergo preliminary tests with the chelating substance to evoke a response, followed by timed urine collection to determine the levels of heavy metals in the body (Bradstreet 2003; Adams 2009). One of the more commonly used chelating substances, oral dimercaptosuccinic acid (DMSA; also called succimer), is being given on a cyclical basis at doses of 10 mg/kg/day every eight hours for three days, followed by 11 days with no DMSA (Bradstreet 2003; Adams 2009). These two-week cycles are then repeated up to six times, totaling approximately three months of treatment (Adams 2009).
Between 6% and 11% of families of children with ASD in various English speaking countries, including the United States, Canada, and Australia, have sought out and tried chelation therapy, and most of these families perceived chelation therapy to result in improvement of symptoms (Green 2006; Goin-Kochel 2009; Christon 2010).
How the intervention might work
The theoretical basis for mercury or other heavy metals causing ASD draws on a wide variety of hypotheses, none yet confirmed. One hypothesis is that mercury or other heavy metals are present in greater levels in children with ASD, compared to their peers, as a result of intrauterine exposure to maternal stores or intake, increased intake from immunisations (thimerosal), oral ingestion (fish or medication), inhalation (airborne pollution), increased absorption, altered metabolism, or decreased excretion (Bernard 2001; Goldman 2001; Holmes 2003; Levy 2003; Counter 2004; Kern 2007).
The excess of stored or circulating total body mercury or other heavy metals is thought to interfere with developmental processes implicated in ASD, and it has been suggested that symptoms of mercury poisoning and ASD share some characteristics (Bernard 2001). Mercury, in being able to cross the blood-brain barrier and the placental barrier, can affect the nervous system and disrupt normal development of the foetus (Aschner 1990; Liu 2008). Prenatal mercury poisoning may result in neurological impairments and global developmental delay and intellectual disability, while postnatal exposure can result in memory loss, irritability, fatigue, intention tremor, skin discolouration, and other organ involvement, including kidney dysfunction (e.g. nephrotic syndrome or tubular dysfunction or both) (Bakir 1973; Amin-Zaki 1974; Grandjean 1997; Goldman 2001; Counter 2004).
Research has produced contradictory findings with regard to levels of heavy metals in individuals with ASD. Although some studies found higher levels of heavy metals in individuals with ASD compared to individuals without ASD (Cohen 1976; Cohen 1982; Adams 2013), more emerging research found no association between ASD and higher levels of heavy metals (Hertz-Picciotto 2010; Albizzati 2012; Rahbar 2013).
An alternative hypothesis is that mercury or other heavy metals could cause ASD through altered cellular functioning that does not require increased body stores or circulating mercury or other heavy metals. In this context, it is thought that individuals with ASD have an impaired capacity to excrete heavy metals and that the severity of autism symptomatology is inversely correlated with excretion ability (Holmes 2003; Kern 2007). This hypothesis is currently being explored (Deth 2008; Zecavati 2009; Garrecht 2011), but it is unlikely that chelation therapy would be beneficial in this context.
Why it is important to do this review
Novel therapies are used frequently by individuals with ASD (Hanson 2007). Despite their increasing use, most CAMs for ASD lack a robust evidence base (Nye 2005; Gold 2006; Millward 2008; Cheuk 2011; Sinha 2011; James 2011). Chelation therapy is one CAM that continues to be used and promoted as efficacious, despite being discouraged by physicians (Golnik 2009) and reports of harm, including death (Brown 2006). A systematic review that examines potential beneficial and harmful effects of chelation for symptoms of ASD is urgently needed to provide best evidence to inform future decision-making about the use of this intervention. Results from this systematic review will help families with ASD make well-informed decisions about the use of chelation therapy and will assist relevant services and other organisations make decisions about best practice.
To assess the potential benefits and adverse effects of chelation for ASD symptomatology.
Criteria for considering studies for this review
Types of studies
Randomised controlled trials.
Types of participants
Individuals of any age diagnosed with ASD using established diagnostic criteria (e.g. DSM-IV) or standardised diagnostic instruments (e.g. the Childhood Autism Rating Scale, Autism Diagnostic Observation Scale, or Autism Diagnostic Interview - Revised) will be eligible for inclusion.
Types of interventions
Interventions will include chelating substances of any type and dose, regardless of administration frequency or method, compared to placebo. We will include trials if the chelating substances are provided alone or as an adjunctive treatment compared to placebo (e.g. chelation in combination with a behavioural intervention versus placebo in combination with a behavioural intervention).
Types of outcome measures
- Core symptoms of ASD, using any measure
- Social interaction*
- Adverse events*
- Non-core behaviours, using any measure
- Quality of life for individual or family*
- Heavy metal levels
Asterisked (*) outcomes will be included in a 'Summary of findings' table.
For our primary analysis, we will assess outcomes at immediate post-intervention (i.e. completion of treatment).
Search methods for identification of studies
We will search the following databases without imposing language or date restrictions:
- The Cochrane Central Register of Controlled Trials (CENTRAL), part of the Cochrane Library;
- Science Citation Index;
- Social Science Citation Index;
- Conference Proceedings Citation Index – Science;
- Conference Proceedings Citation Index – Social Science and Humanities;
- Autism Data (www.autism.org.uk/autismdata);
- WorldCat (limited to theses and dissertations) www.worldcat.org/;
- metaRegister of Controlled Trials metaRegister of Controlled Trials;
- ClinicalTrials.gov (clinicaltrials.gov/);
- ICTRP (apps.who.int/trialsearch/);
- TOXNET (toxnet.nlm.nih.gov/);
- Scieclo; and
- Google Scholar.
We will search Ovid MEDLINE using the following search strategy, which we will adapt as necessary for other databases:
1 exp child development disorders, pervasive/
2 Developmental Disabilities/
3 pervasive development$ disorder$.tw.
4 (pervasive adj3 child$).tw.
5 (PDD or PDDs or PDD-NOS or ASD or ASDs).tw.
9 childhood schizophrenia.tw.
12 Chelation Therapy/
13 Chelating Agents/
14 2,2'-dipyridyl/ or caseins/ or chitosan/ or citric acid/ or cuprizone/ or dimercaprol/ or dithizone/ or ditiocarb/ or edetic acid/ or egtazic acid/ or fura-2/ or humic substances/ or nitrilotriacetic acid/ or penicillamine/ or pentetic acid/ or phytochelatins/ or razoxane/ or succimer/ or technetium tc 99m pentetate/ or thenoyltrifluoroacetone/ or trientine/ or unithiol/
15 Iron Chelating Agents/
17 ferrozine/ or pentetic acid/ or deferoxamine/ or enterobactin/
18 Fursultiamin/ or (Fursultiamin$ or TTFD or thiamine tetrahydrofurfuryl disulfide).tw.
20 (chelation or chelating or chelator$).tw.
21 (metal adj3 antagonist$).tw.
22 ("2,2'-Dipyridyl" or alcaligin or antipyrylazo or arsenazo or bixalomer or cabiotraxetan or calcobutrol or caldiamide or calixarene or calteridol or caloxetic acid or "carboxymethyl beta cyclodextrincaseins" or catenane or chitson or choline tetrathiomolybdate or citric acid or clathrin$ or (crown adj (compound or ether)) or cyclodextrin or cyclophane or cuprizone).mp.
23 (dimercaprol or dimercaptosuccinic acid or DMSA or deferasirox or deferiprone or deferitrin or deferoxamine or deferriferrithiocin or diglycine or dimercaprol or dimethyldithiocarbamate or dithizone or ditiocarb).mp.
24 (edetate or Edetic Acid or Egtazic Acid or Fura-2 or Humic Substances or Nitrilotriacetic Acid or Penicillamine or Pentetic Acid or phytic acid or Phytochelatins or Razoxane or salicylhydroxamic acid or sugammadex or Succimer or "Technetium Tc 99m Pentetate " or Thenoyltrifluoroacetone or Trientine or triethylenetetramine$ or tropantiol or Unithiol or versetamide).mp.
25 (siderophores or ferrozine or pentetic acid or deferoxamine or enterobactin).mp.
27 11 and 26
28 exp animals/ not humans.sh.
29 27 not 28
Searching other resources
We will search reference lists from the retrieved articles for studies not already identified, and we will contact known experts in the field to enquire about other sources of information. We will also search relevant websites including Autism Speaks (http://www.autismspeaks.org/), Research Autism (http://researchautism.net/) and the United States Department of Health and Human Services (http://www.hhs.gov/).
Data collection and analysis
Selection of studies
Two authors (SJ and SS) will independently screen the titles and abstracts of the citations identified from the search. SJ and SS will then obtain the full text of studies that meet, or seem likely to meet, the inclusion criteria. The other members of the review team (NS and KW) will act as arbiters in the event of dispute.
Data extraction and management
Two authors (SJ and SS) will independently extract data from the included studies using a data extraction form designed and piloted for this review. The following information will be extracted:
- Study methods and setting;
- Participant details;
- Intervention details; and
We will resolve disagreements through consultation with the other authors (NS and KW).
SJ will enter data into Review Manager 5.2 (Review Manager 2012), which will be checked by SS.
Assessment of risk of bias in included studies
Two authors (SJ and SS) will independently assess the risk of bias in the included studies using the tool described in the Cochrane Handbook of Systematic Reviews of Interventions (Higgins 2011, 8.5). For each study we will judge the risk of bias to be either low, high, or unclear risk for each of the following domains.
- Sequence generation
- low risk: if a random component was used in the sequence generation process, such as coin-tossing, computer-generated random numbers, or a table of random numbers.
- high risk: if a non-random component was used in the sequence generation process.
- unclear: the sequence generation process was not described.
- Allocation concealment
- low risk: if participants and trial investigators had no foreknowledge (i.e. prior to eligibility decisions being made and informed consent being obtained) of intervention assignment through the use of, for example, central allocation or sequentially numbered envelopes that were opaque and sealed.
- high risk: if participants and trial investigators had foreknowledge of intervention assignment.
- unclear: the method of allocation concealment was not described.
- Blinding of participants and personnel
- low risk: if there was no blinding or incomplete blinding but review authors judge the outcome is unlikely to have been influenced by lack of blinding, or if blinding of study participants and personnel was ensured and it is unlikely that blinding could have been broken.
- high risk: if there was no blinding or incomplete blinding and the outcome was likely influenced by a lack of blinding, or if blinding of study participants and personnel was attempted but it is likely that blinding could have been broken and the outcome influenced by lack of blinding.
- unclear: if lack of information prohibits judgement of either low or high risk of bias, or if the study did not address this outcome.
- Blinding of outcome assessment
- low risk: if there was no blinding of outcome assessment but review authors judge the outcome measurement is unlikely to have been influenced by lack of blinding, or if blinding of outcome assessment was ensured and it is unlikely that blinding could have been broken.
- high risk: if there was no blinding of outcome assessment and the outcome measurement was likely influenced by a lack of blinding, or if there was blinding of outcome assessment but it is likely that blinding could have been broken and the outcome measurement influenced by lack of blinding.
- unclear: if lack of information prohibits judgement of either low or high risk of bias, or if the study did not address this outcome.
- Incomplete outcome data
- low risk: if no missing data were reported or if appropriate methods were used to impute missing data, or if the reason for missing data is unlikely to be related to the true outcome.
- high risk: if missing data were reported and no appropriate methods were used to impute missing data, or if the reason for missing data is likely to be related to the true outcome.
- Selective reporting
- low risk: if a study has a protocol and all pre-specified outcomes were reported in the pre-specified manner, or if a study has no protocol but all expected outcomes have been reported.
- high risk: if a study has a protocol and one or more pre-specified outcomes were either not reported or were reported in a manner that was not pre-specified, or if a study has no protocol and all expected outcomes have not been reported.
- unclear: if lack of information prohibits judgement of either low or high risk of bias.
- Other bias
- low risk: if other sources of bias (e.g. contamination or recruitment bias) do not appear to exist.
- high risk: if other sources of bias exist.
- unclear: if lack of information permits judgement of whether other sources of bias exist.
We will resolve disagreements by reaching consensus with the help of the other review authors (NS and KW).
Measures of treatment effect
We will analyse dichotomous outcomes by calculating the risk ratio (RR) and corresponding 95% confidence interval (CI).
For continuous outcomes we will calculate mean differences (MD) and corresponding 95% CI if studies use the same rating scales. As recommended by Higgins 2011, we will focus on final values unless change scores are used in some of the studies. We will combine in the same meta-analysis studies that report final values with studies that report only change scores, provided that the studies use the same rating scale (Higgins 2011). A potential problem of including change scores is that the standard deviation of changes may not be reported in the original study (Higgins 2011). We will contact trial authors and attempt to estimate the standard deviation of changes if it is not reported. We will calculate the standardised mean difference (SMD) with 95% confidence intervals if studies use different scales to measure the same outcomes.
If studies provide multiple, interchangeable measures of the same construct at the same point of time, we will calculate the average SMD across the outcomes and the average estimated variances (Higgins 2011).
Unit of analysis issues
We will incorporate cross-over trials into meta-analysis using results from paired analyses (Elbourne 2002; Higgins 2011). We do not anticipate finding cluster-randomised trials. We will create a single pair-wise comparison for each identified multi-arm study by combining all relevant experimental groups into a single group, and by combining all relevant control groups into a single group (Higgins 2011).
Dealing with missing data
We will attempt to contact trial investigators to request missing data. If missing data are provided by the trialists, we will conduct meta-analysis according to intention-to-treat principles using all data and keeping participants in the treatment group to which they were originally randomised, regardless of the treatment they actually received (Higgins 2011). If missing data are not provided, we will analyse only the available data, and we will not impute missing data given that symptoms of ASD vary greatly. We will document missing data and attrition in the 'Risk of bias' table, and we will discuss how missing data may affect the interpretation of the results.
Assessment of heterogeneity
We will assess clinical heterogeneity by comparing the between-trials distribution of participant characteristics (e.g. children versus adults), trial characteristics (e.g. cross-over versus parallel design), and intervention characteristics (e.g. treatment type and dose). We will evaluate statistical heterogeneity using the I
- 0% to 29% might not be important;
- 30% to 49% may represent moderate heterogeneity;
- 50% to 74% may represent substantial heterogeneity;
- 75% to 100% represents considerable heterogeneity (Higgins 2011).
Assessment of reporting biases
If 10 or more studies are found, we will use funnel plots to investigate the relationship between intervention effect and study size. Asymmetry of a funnel plot may indicate, among other things, publication bias or poor methodological quality (Egger 1997). We will explore possible reasons for any asymmetry found.
We will synthesize results in a meta-analysis using a fixed-effect model when studies are similar enough with regard to the intervention, population, and methods to assume that the same treatment effect is estimated. We will synthesize results in a meta-analysis using a random-effects model when statistical heterogeneity is found or when studies differ enough with regard to the intervention, population, and methods to assume that different yet related treatment effects are estimated, and when it is deemed to be clinically relevant (Higgins 2011).
Subgroup analysis and investigation of heterogeneity
We will conduct the following subgroup analyses:
- Type of ASD (e.g. autistic disorder versus Asperger's disorder);
- Participant age (e.g. adults versus children, preschool versus school-age);
- Treatment type (e.g. DMSA versus other agents);
- Treatment dosage (e.g. DMSA dose of 10 mg/kg bodyweight administered three times per day versus higher doses); and
- Length of follow-up (e.g. ≤ 3 months versus > 3 months).
We will conduct sensitivity analyses to investigate the effect on the overall results of excluding trials where:
- Allocation concealment or sequence generation was inadequate (selection bias);
- Blinding was not done (performance bias); and
- Outcome data were incomplete (attrition bias).
We would like to thank Laura MacDonald and Geraldine Macdonald for their feedback and guidance. We would also like to thank Margaret Anderson for her assistance with the development of our search strategy.
Contributions of authors
All authors contributed to the authorship of the protocol.
Declarations of interest
Stephen James - none known.
Katrina Williams - I gave a talk about treatments for autism at a symposium organised by Janssen-Cilag Pty Ltd. Janssen-Cilag had no control over the contents of the talk and the speaker's fee was paid to the University that employs me. I do not have an ongoing relationship with Janssen-Cilag.
Natalie Silove - Children's Hospital Westmead is enrolling up to eight participants in a phase two drug trial in adolescence with fragile x syndrome. It is not related to autism or to chelation. I receive no personal funds at all.
Shawn Stevenson - At the time of this review, I was employed by Autism Victoria in the information and training services team delivering professional development workshops and training to people who work with people affected by an autism spectrum disorder. I was also an unpaid Board Member of Support and Advocacy Autism Individuals and Families.
Sources of support
- University of Melbourne, Australia.Shawn Stevenson received a salary through support of the Williams Collie Trust, University of Melbourne, for autism research during the time of this protocol. Katrina Williams received a salary through support of the APEX Australia Chair of Developmental Medicine, University of Melbourne. The University provided IT, office and library facilities during the time of this protocol.
- NSW Health Service, Australia.Natalie Silove's salary is paid for by NSW Health Service - The Sydney Children's Hospital Network
- No sources of support supplied