A new study was released today in JAMA which looked at, in part, mitochondrial DNA overreplication in a sample of ten autistic children between the ages of 2 and 5 and ten matched controls.  Giulivi et al. found that 5 of the 10 autistic children and 2 of the control children had mitchondrial DNA overreplication.

In the comments section of a previous post, the question was raised of how autism is diagnosed in the samples being studied. This new study allows an opportunity to look at how carefully the sample is selected and controls matched to the sample.

For example, "The recruited children were aged 2 to 5 years and resided with a biological parent in a well-defined catchment area of more than 22 counties in northern California and parts of Los Angeles County, California." The controls were "sampled from birth files with frequency matching to the projected distribution of sex, age, and geographic area among cases of autism. Environmental, lifestyle, reproductive, maternal medical, and detailed demographic information was collected through an extensive telephone interview with the primary caregiver. Participants’ mothers classified parents into race and ethnicity categories identical to those used in the US Census and these same categories were used to define child race and ethnicity."

What about the autism diagnosis and making sure the recruited children really were on the spectrum and what was known about their functional levels? The authors stipulate precisely the means through which diagnoses were made and the scorings used: "Diagnoses were confirmed through clinical examinations using the Autism Diagnostic Inventory-Revised (ADI-R)9 and the Autism Diagnostic Observation Schedule (ADOS)." The authors "selected 10 individuals who met criteria for full syndrome autism on both the ADI-R and ADOS and 10 typically developing control children." The authors even took into account the diversity of the spectrum: "Because ASD represents a fairly diverse phenotype, we studied children meeting criteria for full syndrome autism and neurotypically developing children (controls) without a clinical diagnosis of full syndrome autism, ASD, or developmental delays." The authors continue, "We also attempted to achieve comparable age, sex, and race/ethnicity across groups to minimize confounding sociodemographic factors that may be surrogates for genetic, epigenetic, or cultural background." In other words, the researchers took care to make the best selection they could for the controls so that their results would be as accurate as possible.

In part, the study revealed that "Defective or abnormal lymphocytic mitochondria in children with autism were observed in this exploratory study as determined by the following parameters:
(1) low PDHC activity accompanied by low lactate-to-pyruvate ratios,
(2) impaired complex I alone or in combination with other complexes (mainly complex V), 
(3) enhanced mitochondrial rate of hydrogen peroxide production, and
(4) mtDNA overreplication and/or deletions."

The authors (despite focus on press releases that don't always present an accurate portrait of a study) were careful to note the limitations of the study, as well: small sample size, potential of Type I errors, "cases in this substudy were somewhat higher functioning than those from the original study, lessening the difference with controls," "the differences observed between our cases and controls could represent confounding effects due to possible associations between mitochondrial dysfunction and race, which also differed by case-control status," and "none of the children in this study had been previously diagnosed with a genetic syndrome or had any indications of genetic syndromes as determined by developmental pediatricians at the Medical Investigations of Neurodevelopmental Disorders Institute."

Perhaps the most relevant limitation that the study authors themselves note is this: "inferences about a cause and effect association between mitochondrial dysfunction and typical autism cannot be made in a cross-sectional study. Several factors influence expression of mitochondrial respiratory insufficiencies in both the affected and general populations (ie, nuclear genetic backgrounds,39 mtDNA heteroplasmy in different tissues,40 different energy thresholds within a given tissue or organ,41 and environmental factors42)."

People in the autism community have rejected studies with hundreds of thousands of children, crying foul. They have embraced case series with a dozen kids as proof of the MMR causing autism, all while insisting that the study doesn't say that at all. They have gone gaga over monkey studies (how many 14, 13, 12?). People in the autism community go beyond what the researchers themselves note are limitations in the study, ignore the caution, ignore the calls for more research ("More research is needed to understand the molecular causes of the mitochondrial dysfunction and how this and other neurometabolic defects may contribute to autism or related phenotypes.") and insist that answers are at hand.

We must be cautious. This is not the magic bullet. Half the sample did not have "mtDNA overreplication." Two of the controls did. Perhaps like epilepsy this is more likely to occur in the autistic population. With a sample of 10 and a control of 10, it would be inexcusable to speculate beyond this study and extrapolate its findings to all autistic individuals. Much more work remains to be done. The best that we can do, as members of this community, is to work to get the facts correct and not speculate wildly. We should recognize our limitations in understanding the details of the study, and we should never, ever leap to conclusions based upon a press release and an abstract.


Cecilia Giulivi; Yi-Fan Zhang; Alicja Omanska-Klusek; Catherine Ross-Inta; Sarah Wong; Irva Hertz-Picciotto; Flora Tassone; Isaac N. Pessah
Mitochondrial Dysfunction in Autism
JAMA. 2010;304(21):2389-2396.