Journal of Autism and Developmental Disorders

, Volume 42, Issue 4, pp 643–646

Further Commentary on Mitochondrial Dysfunction in Autism Spectrum Disorder: Assessment and Treatment Considerations


    • Department of RadiologyUniversity of Washington
    • University of Washington Autism Center
    • Department of BioengineeringUniversity of Washington
  • Neva M. Corrigan
    • Department of RadiologyUniversity of Washington
  • Annette Estes
    • University of Washington Autism Center
    • Department of Speech and Hearing SciencesUniversity of Washington
  • Dennis W. W. Shaw
    • Department of RadiologyUniversity of Washington
    • Seattle Children’s Hospital
Letter to the Editor

DOI: 10.1007/s10803-011-1352-4

Cite this article as:
Dager, S.R., Corrigan, N.M., Estes, A. et al. J Autism Dev Disord (2012) 42: 643. doi:10.1007/s10803-011-1352-4


The authors respond to a recent letter (Rossignol and Frye 2011) critical of their paper, “Proton magnetic resonance spectroscopy and MRI reveal no evidence for brain mitochondrial dysfunction in children with autism spectrum disorder” (Corrigan et al. 2011). Further considerations regarding the assessment of mitochondrial dysfunction in autism spectrum disorder, and related treatment considerations, are discussed.


AutismDevelopmental disordersMRSMRIMitochondrial disordersBrain lactate

Rossignol and Frye (2011) raise several criticisms regarding our recent paper, “Proton magnetic resonance spectroscopy and MRI reveal no evidence for brain mitochondrial dysfunction in children with autism spectrum disorder” (Corrigan et al. 2011), to which we welcome the opportunity to respond and expand upon. They raise concerns about the specificity and sensitivity of magnetic resonance spectroscopy (1H MRS) lactate measurements and whether the study methods may have missed a subset of children with subtle mitochondrial dysfunction. Furthermore, Rossignol and Frye maintain that a negative diagnostic measure, such as normal lactate levels, however measured, has no bearing on establishing or ruling out a diagnosis of mitochondrial disease. They also express concern that results from our paper will dissuade families from pursuing a mitochondrial diagnostic work up and, by implication, obtaining treatment for occult mitochondrial dysfunction. Each of these points will be addressed in turn.

In their letter, Rossignol and Frye suggest lactate may be an insensitive marker for mitochondrial dysfunction in children with ASD. While we agree that there is no single gold-standard laboratory test for the definitive diagnosis of mitochondrial dysfunction, prior work has demonstrated that 1H MRS measurement of brain lactate can be a sensitive biomarker for detecting brain bioenergetic alterations consistent with mitochondrial dysfunction (Dager et al. 2004). The original case reports suggesting a connection between mitochondrial dysfunction and autism spectrum disorder (ASD) were based on blood lactate elevations, and several more recent studies have used measurement of elevated blood lactate to identify children for follow-up muscle biopsy (Coleman and Blass 1985; Correia et al. 2006; Oliveira et al. 2005). Contrary to their current argument that lactate measures are not sensitive markers, Rossignol has in fact previously touted the utility of elevated blood lactate as a peripheral biomarker to identify mitochondrial dysfunction in children with ASD, even in the absence of other classic signs of mitochondrial disorder and for children with normal muscle biopsies (Rossignol and Bradstreet 2008). Our study evaluated brain lactate levels, instead of in blood, since brain lactate elevations would provide stronger evidence for brain mitochondrial dysfunction than peripheral blood lactate elevations. This is because of heteroplasmy, the variable loading and distribution of mitochondrial mutations between organs. The potential for heteroplasmy makes it important to evaluate directly the organ suspected to have alterations in mitochondrial function, especially when looking for links to a neurodevelopmental disorder such as ASD.

Our laboratory has substantial experience measuring brain lactate in clinical populations (Dager 2010; Dager et al. 1999; Dager et al. 2004; Dager and Steen, 1992; Friedman et al. 2000). Previously, using similar 1H MRS acquisition methods, but a less sensitive analytic technique, there was sufficient sensitivity to detect subtle brain lactate elevations in untreated individuals with bipolar disorder, which we interpreted to reflect brain mitochondrial dysfunction (Dager et al. 2004). This finding, parenthetically, has led to new treatment considerations in bipolar disorder (Lyoo et al. 2010). For our study (Corrigan et al. 2011), we employed a recently developed 1H MRS analytic approach for the improved detection of brain lactate that had been carefully validated for measurement specificity and sensitivity (Corrigan et al. 2010). In our report (Corrigan et al. 2011), we were quite clear that we applied this improved 1H MRS analytic method to previously reported 3–4 year old data (Friedman et al. 2003) and then also used this more sensitive 1H MRS analytic approach to assess new data collected at 6–7 and 9–10 years of age.

Rossignol and Frye misrepresent our study findings as being based solely on brain lactate levels. In fact, an important component of our study was the systematic evaluation of high-resolution MR images evaluating for brain structural abnormalities that could be related to mitochondrial disorders (Saneto et al. 2008). The combination of brain lactate levels measured by 1H MRS and brain structural changes detected by MRI, in conjunction with a review of medical history, are among the most widely used clinical criteria for the diagnostic evaluation of potential brain mitochondrial disease (Bianchi et al. 2007; Saneto et al. 2008). In addition to applying the above methods, we further assessed for more subtle evidence of mitochondrial dysfunction, seeking to identify an ASD subgroup having non-significant elevations of compartmental or global brain lactate that might exhibit specific symptom expression or atypical clinical course. We were unable to identify any such individuals within our large sample. We additionally evaluated for mitochondrial dysfunction longitudinally at three distinct age-points during childhood to increase our likelihood for detecting progressive or new onset mitochondrial dysfunction. Similar to blood lactate that can be spuriously elevated under common clinical situations such as a child struggling to resist a blood draw (Haas et al. 2008), brain lactate can be spuriously elevated by hyperventilation (Posse et al. 1997). In our study (Corrigan et al. 2011), this potential confound was obviated by the use of propofol anesthesia in the ASD and DD groups (Amundsen et al. 2005). Moreover, the use of propofol would tend to unmask occult mitochondrial dysfunction, if present, and thus, result in increased brain lactate generation due to its suppressant effects on respiratory chain complexes (Morgan 2007).

Rossignol and Frye are correct in pointing out that we have previously published findings of decreased brain NAA in children with ASD at 3–4 years of age (Friedman et al. 2003; Friedman et al. 2006), the same cohort evaluated for brain lactate at that age point in the current study. However, they misconstrue the absence of this evaluation in the current study as an attempt by us to conceal other 1H MRS markers that might indicate mitochondrial dysfunction. In fact, the pulse sequence parameters used to collect brain lactate data for the current study were optimized for measurement of lactate, and are not suitable for reliable quantification of other chemicals, such as NAA (Friedman et al. 2003). Longitudinal quantification of brain NAA and other metabolites that used different 1H MRS pulse sequence parameters to study this same cohort are currently being finalized for publication. In the context of this discussion, it is also important to point out that NAA decreases are present in many brain disorders having no known associations with mitochondrial disorder (Dager et al. 2011; Dager et al. 2008; Haas et al. 2008). There are also corresponding increases in brain lactate when NAA is decreased in association with brain mitochondrial dysfunction (Saneto et al. 2008). Another useful secondary 1H MRS marker of brain mitochondrial dysfunction, when observed in conjunction with elevated brain lactate, is elevated glutamate (Dager et al. 2004). Glutamate (GLX) elevations were not observed in the children with ASD at age 3–4 years of age (Friedman et al. 2003; Friedman et al. 2006).

Rossignol and Frye portray our exclusion of subjects with known genetic syndromes having some overlapping symptom expression as ASD as a major flaw. We agree that studying genetic syndromes that can be associated with autistic features in some cases, such as Angelman, Prader-Willi, Fragile X, and Rett syndromes, may provide insight into the mechanisms resulting in autism. However, the abnormalities in these disorders involve multiple pathways affecting factors controlling dendritic shape and density and synaptic density that are not related to mitochondrial function (Ramocki and Zoghbi 2008). Thus, the common thread of these disorders points to a failure of neuronal homeostasis and not mitochondrial dysfunction.

Rossignol and Frye express their concerns that by publishing these negative findings we will discourage further research into possible links between mitochondrial dysfunction and ASD. They further state that negative findings have no bearing on establishing or ruling out a diagnosis of mitochondrial disease. First, as we were careful to make clear in our recent report (Corrigan et al. 2011), although no evidence on the basis of 1H MRS or MRI was found linking ASD and mitochondrial dysfunction in the large sample studied, our findings alone do not rule out the possibility of a relationship nor a relatively infrequent co-occurrence of brain mitochondrial dysfunction and ASD. As indicated in our paper, we are continuing to pursue research to investigate the potential for a link between the two disorders at younger ages by studying infants at high genetic risk due to having an older sibling with ASD. More generally, we believe publication of negative findings from large samples is useful for informing scientific discourse. While Rossignol and Frye clearly disagree with this premise, it is important to note that there is no available objective data conclusively showing a causal link between mitochondrial dysfunction and ASD. Thus, negative findings need to be taken into account when considering the likelihood or plausibility of this unproven association.

We have several remaining concerns regarding Rossignol and Frye’s comments. First, it is our impression from prior disclosures that these individuals are not unbiased observers. In previous communications, but not in their letter, Rossignol has acknowledged receiving financial support from the International Hyperbarics Association and provides fee-for-service hyperbaric treatment. As such, the authors are undoubtedly aware of the substantial financial burden a course of hyperbaric oxygen treatment can place upon the families of an individual with ASD. Hyperbaric oxygen has been advocated to treat ASD on the basis of anecdotal case reports or small, uncontrolled studies, and one short-term, controlled treatment study of ASD participants (Rossignol et al. 2009). However, at least two subsequent controlled treatment studies of ASD have failed to demonstrate any consistent evidence for treatment efficacy (Granpeesheh et al. 2010; Jepson et al. 2011). Jepson et al. (2011) evaluated both for group treatment effects across multiple symptom domains and for individual participant treatment responses; those authors have critiqued results from the Rossignol et al. (2009) study, raising substantial questions about their interpretation of positive treatment findings. More apropos to the current discussion, there is no solid evidence to support that hyperbaric oxygen treatment is effective in treating known cases of mitochondrial disease or dysfunction. As stated in a recent review from the Mitochondrial Medicine Society concerning treatment approaches for mitochondrial disease, “There is no evidence that hyperbaric oxygen therapy is beneficial for patients with mitochondrial disease, and there is a theoretical concern about oxygen toxicity in these patients.” (Parikh et al. 2009). We believe strongly that advocacy of an unsubstantiated treatment based on an unproven association with mitochondrial dysfunction has the potential to substantially deplete often limited family resources at the expense of proven, evidence-based treatment options for ASD, such as early, intensive behavioral intervention (e.g., Dawson et al. 2010).


This study was supported by NIH grants 2P01 HD 35465 and 1P50 HD 55782.

Conflict of interest

None of the authors report any conflicts of interest with this work.

Copyright information

© Springer Science+Business Media, LLC 2011