June 29, 2008
Mitochondrial diseases are a group of heterogeneous disorders resulting from inborn, or sometimes acquired, defects that impair the function of mitochondria, tiny organelles inside cells that are essential for energy production. Although mitochondrial diseases can affect virtually any organ in the body, neurological manifestations are common since the high energy demands of the brain and nervous system make them particularly vulnerable to compromised energy production. A number of recent reports have described evidence of mitochondrial disease in children with features of autism or an autism spectrum disorder. The U.S. Department of Health and Human Services (HHS), the National Institute of Neurological Disorders and Stroke (NINDS), the National Institute of Mental Health (NIMH), the Centers for Disease Control and Prevention (CDC), and the Food and Drug Administration (FDA) convened this workshop with experts in either mitochondrial disease or autism to discuss how to advance research into the potential relationship between the two groups of disorders. Registration for the workshop was open for members of the public to attend as observers. The proceedings below summarize the overall content of the workshop's presentations and discussions.
Mitochondrial genetics and biology
Mitochondria allow cells to transfer chemical energy in the oxygen molecule into adenosine triphosphate (ATP), the major energy currency within cells. Mitochondrial dysfunction can lead to disease because cells may malfunction or die when they are unable to produce enough energy to fuel their essential processes. In addition, as a price for using oxygen as an energy source, mitochondria produce a low level of reactive oxygen species or "free radicals". If not contained inside mitochondria or by the cell's scavenging system, free radicals can also have destructive consequences for cells and tissues.
Mitochondria originate from ancient bacteria that were internalized by more complex cells very early in the evolution of plants and animals, and they continue to house their own mitochondrial DNA (mtDNA). Over time, human mtDNA has evolved into several variations, or haplotypes, thought to be the result of functional adaptations that coincided with early human migration patterns. Most genes necessary for mitochondrial function have come to be encoded by DNA in cell nuclei: of an estimated 1500 mitochondrial genes, mtDNA encodes only 37. Thirteen of these 37 genes code for proteins essential to the respiratory chain, a series of protein complexes involved in "oxidative phosphorylation," the process of taking electrons away from oxygen and converting their energy into ATP. Genetic mutations leading to mitochondrial disease can affect either mtDNA or nuclear DNA (nDNA), but they result in different patterns of inheritance. Mitochondrial genes encoded by nDNA are inherited from both parents with Mendelian patterns of inheritance (x-linked, autosomal dominant, or autosomal recessive), but mtDNA is passed down exclusively from the mother through maternal inheritance. Deletions or duplications in mitochondrial or nuclear DNA can also occur, and these are often sporadic, appearing for the first time in an individual whose parents did not carry the same genetic defect.
Each cell in the body contains hundreds to thousands of mitochondria, and every mitochondrion contains multiple copies of mtDNA. Deleterious mutations in mtDNA may affect all copies of mtDNA (homoplasmy), but frequently they only affect some copies (heteroplasmy). Since the many copies of mtDNA are distributed randomly between daughter cells during cell division, heteroplasmy can lead to significant variation in the proportion of mutated mtDNA over time and across different organs or tissues. This variation in mutation load can influence the clinical expression of mitochondrial disease. Heteroplasmy may also complicate the diagnosis of mtDNA diseases because the causative mutation may be present in only some tissues, such as specific brain regions or specific muscles, and not in others, such as blood or hair. In addition, an individual's mtDNA haplotype can modify the effect of pathogenic mutations in mitochondrial genes. More broadly, mtDNA haplotypes may also modify susceptibility for diseases in which mitochondrial dysfunction may not be a primary cause, including diabetes, multiple sclerosis, and some cancers and neurodegenerative diseases.
An overview of mitochondrial diseases
Researchers have identified more than 200 pathogenic mutations in mtDNA and over 2000 in nDNA, and the combined incidence of mitochondrial diseases is currently estimated to range between 1 and 5 in 10,000. Though much is known about some specific mitochondrial diseases, our understanding of how mitochondrial function and dysfunction contribute to the broader scope of human disease is incomplete and evolving, and the workshop panelists emphasized a growing recognition of the many ways that mitochondrial function can be disrupted. They suggested that mitochondrial disease may be more common than currently thought and will likely be implicated in an increasing number of medical conditions, including aging and cancer.
Known mitochondrial diseases show both genotypic and phenotypic diversity in that different disease presentations can result from the same underlying genetic defect, and conversely, many different genetic defects can lead to similar clinical manifestations. For example, lesions in the basal ganglia are a neuropathological signature of Leigh Syndrome, which has been associated with several different mtDNA and nDNA mutations. In MELAS (Mitochondrial Encephalomyopathy, Lactic Acidosis, and Stroke-like episodes), 80% of cases are caused by the same mtDNA mutation (A3243G); but several other mutations have been associated with the remaining 20% of cases, and the prevalence of the mutation in asymptomatic carriers may exceed 2 per 1000 (ref. 8). The onset and time course of mitochondrial diseases are also variable, occurring early and progressing rapidly in some but only after many months, years or even decades of life in others.
The diversity of mitochondrial diseases is influenced by their complex genetics, by cell-type specific expression of interacting genes, and by environmental influences. Despite their diversity as a group, some general rules do apply to the known mitochondrial diseases. In particular, they are typically progressive and multisystemic, most often affecting organs that have high energy demands such as the brain and nerves, skeletal and cardiac muscle, and the liver and kidneys. The workshop panelists presented a number of clinical features as red flags that should raise suspicion for mitochondrial disease4-7. Of particular interest for this workshop, central nervous system manifestations of mitochondrial disease can include hypotonia, seizures, encephalopathy, ataxia, intellectual disabilities, global delay, and brain malformations. Sensory and peripheral nerves can also be affected, leading to deafness, blindness, or neuropathy.
Diagnosing mitochondrial disease
The diagnosis of mitochondrial diseases is complicated by their heterogeneous presentations and by the lack of screening procedures or diagnostic biomarkers that are both sensitive and specific. The workshop panelists explained that diagnosis is often a lengthy process beginning with a general clinical evaluation followed by metabolic screening and imaging and finally by genetic tests and more invasive biochemical and histological analyses. A number of challenges exist at each of these stages, and even after extensive evaluation it is often not possible to be certain that a condition is caused by a mitochondrial disorder.
In terms of metabolic screening, a number of biochemical tests are used, but they have imperfect sensitivity and specificity in identifying mitochondrial disorders. Lactate is produced by "glycolysis," a biochemical pathway in cells that produces ATP quickly without using oxygen, and an increased level of blood lactate in a person at rest can signal mitochondrial dysfunction. Lactate is produced normally by muscle during exercise, or when brain regions are activated, but elevated lactate levels at rest may indicate that glycoysis is compensating for impaired energy production by mitochondria. However, spurious elevations in blood lactate can also occur with tourniquet use to draw blood or in a struggling child. Other biochemical abnormalities which can raise suspicion of a mitochondrial disorder include elevated levels of the amino acid alanine in fasting blood or spinal fluid samples, an increased ratio of lactate to pyruvate in blood and spinal fluid, elevated organic acids in urine, or an elevated ratio of acyl-carnitine to free carnitine in blood. It is important to note that biochemical abnormalities may not be present during periods when the mitochondrial disease is quiescent.
Magnetic resonance imaging (MRI) techniques and biopsy specimen analysis allow for more in-depth evaluation of mitochondrial disorders. For example, magnetic resonance (MR) spectroscopy can detect elevated lactate levels and other metabolic markers in brain and muscle. Such measures may be helpful in building evidence for a mitochondrial brain disorder, but unfortunately, no MR spectroscopy signature is specific for mitochondrial diseases. Moreover, elevated brain lactate also occurs in patients with other conditions including stroke and inflammation. The effects of sedating agents, which are often needed to acquire MR spectroscopy data from children, are also unknown. (Some mitochondrial diseases do have distinctive neuroimaging signatures detectable with structural MRI, such as basal ganglia and brainstem lesions in Leigh Syndrome or the waxing and waning stroke-like lesions in MELAS.)
Although more invasive, direct tests of mitochondrial enzyme function in muscle or skin biopsy tissue can also be helpful in making diagnoses, but they are very dependent on tissue collection, storage and handling procedures. The panelists noted a concerning lack of standardization and poor consistency in results across testing laboratories, such that in complex cases, it is not unusual that the results of mitochondrial function assays are abnormal in one laboratory yet normal in another. Classically, mitochondrial diseases are identified by the presence of "abnormal mitochondria" in biopsied cells. Electron microscopy provides detailed views of mitochondrial structure, and biochemical stains can demonstrate loss of mitochondrial enzyme function or proliferation of sick mitochondria (called "ragged red fibers" in muscle biopsy).
In the modern era, the identification of known mitochondrial mutations in tissue has greatly aided diagnosis. However, even when clinical features and family history strongly suggest mitochondrial disease, the underlying genetic mutation can elude detection, and there is no current screening procedure that would be practical for all cases of suspected mitochondrial disease. Because of heteroplasmy, mtDNA mutations may be present in some but not all tissues; multiple tissues may need to be sampled (muscle, blood, hair, uroepithelial cells, skin, etc.); and it is extremely difficult to "rule out" a mitochondrial contribution in some complicated cases. Although screening in any tissue could potentially detect nDNA mutations, with many possible mutations, testing must be guided by the specific features of each individual case. It is also important to distinguish between changes in mtDNA that cause disease and benign genetic variations such as those that define mtDNA haplotypes.
In summary, the diagnosis of a mitochondrial disorder is a lengthy and complicated process, with no single type of test providing a definitive answer. In clinical practice, the results of tests on blood, spinal fluid and urine, muscle and other tissue biopsies, as well as brain imaging and genetic studies are considered together with clinical symptoms to diagnose a patient. The combined evidence from all the tests is used to qualify the condition according to consensus guidelines as a "possible", "probable" or "definite" mitochondrial disorder.
Known triggers of mitochondrial disease
Mitochondrial diseases can go undetected for many years, and many cases display an episodic course with relatively stable periods punctuated by abrupt degeneration that may coincide with an infection or other stress to mitochondrial function. One study in young children definitively diagnosed with mitochondrial disease found that 60% showed an episodic disease course. In 72% of those cases, deterioration was associated with a naturally acquired infection45. Of possible importance, mitochondria are the major generators of body heat and are therefore extremely active during fever. It is not known whether fever or other aspects of the inflammatory or immune response to a virus or bacteria trigger deterioration after infection. To reduce the risk presented by acquired infections, the workshop panelists strongly encourage vaccinations in the hundreds of children they treat for mitochondrial disease. Among thousands of patients they had collectively seen, very few had deteriorated following vaccination, and in those few cases, it is difficult to determine that other stressors besides the vaccine did not play a role in the neurologic deterioration. In addition to febrile illnesses, other potential precipitating factors noted by the panelists included dehydration, reduced caloric intake, and in some cases, exercise. The exact mechanisms that lead to deterioration after these triggers are not well understood, nor is it known why some individuals recover function after deterioration while others are irreversibly impaired.
A number of medications and environmental agents can also impair mitochondrial function. Some medications act as precipitating factors in genetically susceptible individuals, as in aminoglycoside antibiotic-induced sensorineural hearing loss in carriers of a mutation at mtDNA position 1555. Others can result in acquired mitochondrial impairment, such as the AIDS drug AZT and valproate, an anticonvulsant and mood stabilizer. Additional medications that can affect mitochondria include some chemotherapy drugs and statins. Environmental agents that impair mitochondrial function include natural neurotoxins such as some fungal toxins and chemicals such as MPTP and the insecticide rotenone, both of which cause Parkinson's-like symptoms. Cigarette smoke and alcohol can also damage mitochondria, and some studies suggest they increase the risk of vision loss in individuals carrying the mtDNA mutation that causes Leber Hereditary Optic Neuropathy (LHON).
Mitochondrial diseases and autism: clinical parallels and evidence for a relationship
Autism spectrum disorders (ASDs) are a diverse group of neurodevelopmental disorders characterized by the appearance by age three of a triad of primary symptoms: reduced social behavior, abnormal language, and restricted or repetitive behaviors. A considerable minority of cases show regression after a seemingly normal early development, and a number of comorbid medical conditions are also variably associated with ASDs including seizures, mental retardation, gastrointestinal complications, and dysmorphic appearance. The workshop panelists discussed reports from the published scientific literature as well as their own experiences with patients in terms of evidence for a relationship between mitochondrial disease and autism.
Workshop panelists who treat children with mitochondrial disease noted that some of these children have autistic features, and some children eventually found to have mitochondrial disease are initially diagnosed with an ASD. In addition, siblings of children with maternally inherited mitochondrial disorders sometimes present with autism. Presumably, they have inherited the same mitochondrial mutation from their mother, but the mutation may be difficult to find. Workshop panelists who mainly see individuals with a primary diagnosis of autism found parallels with clinical observations in mitochondrial disease such as developmental regression, seizures, and gastrointestinal complications. A number of published and anecdotal reports have described laboratory testing and brain imaging findings consistent with mitochondrial dysfunction in some ASD cases. However, few cases have been studied with in-depth analyses of mitochondrial function, and drawing conclusions from the current literature is difficult since tests that are used may be nonspecific on their own or dependent on test conditions and procedures. A carefully designed program of mitochondrial study in autism will be required to move forward.
Additional parallels between ASDs and mitochondrial disease noted by the workshop panelists were in family histories and patterns of inheritance. These included possible maternal inheritance in some ASDs, a similar higher prevalence in males for both ASDs and some mitochondrial diseases, and a high frequency of psychiatric conditions such as depression, delusions and attention deficit disorder in families with mitochondrial disease, including in relatives who seem otherwise unaffected. In a study highlighted in the workshop presentations42, one child in a family with a group of heterogeneous neurological disorders had features of autism while his sister developed signs of Leigh Syndrome. Both children had the same mutation in the mitochondrial transfer RNA for lysine (G8363A), but molecular genetic testing showed that the mutation load in blood and muscle from the child with autism was lower than in the child with Leigh Syndrome, suggesting that relatively mild mitochondrial defects may manifest in autism in the absence of other signs of mitochondrial disease.
Geneticists studying ASDs hypothesize that many cases may result from sporadic, or de novo, mutations or gene copy number variations (deletions and duplications) that are dispersed across the genome. The workshop panelists suggested that such sporadic events may have a high likelihood of affecting mitochondria since genes important for their function are found throughout the genome. Consistent with this possibility, some but not all genetic studies have identified genetic variations associated with autism in the gene SLC25A12, a mitochondrial aspartate/glutamate carrier important for respiratory chain function. In addition, mitochondrial dysfunction has been reported in individuals with a recently described chromosomal abnormality (15q11-q13 inverted duplication) associated with an estimated 3-5% of autism cases35.
Despite these lines of evidence suggesting that mitochondrial dysfunction may contribute to autism, the extent of such a contribution is not known, and a mechanistic relationship has not been explored. Certainly, many children who are extremely disabled by mitochondrial disease have no signs of ASD. The progressive and severe myopathy, neuropathy, blindness, focal neurologic signs, cardiac dysfunction, renal failure, and other clinical features seen in many forms of mitochondrial disease do not occur in ASDs. Whether a low mitochondrial mutation load could contribute to the risk of autism in a subset of persons is currently unknown and quite difficult to rule out. The ASDs themselves represent a complex set of symptoms, likely with many causes, and workshop panelists expressed a range of opinions about what proportion of ASDs might be accounted for by mitochondrial disease. The panelists' discussions also revealed important differences in perspective between researchers and clinicians focused on autism and those focused on mitochondrial disease in terms of the types of cases they see and the clinical features they emphasize. Autism experts on the panel stressed a need for more detailed descriptions of behavioral phenotypes in mitochondrial diseases, while mitochondrial disease experts expressed a need for more thorough clinical and biochemical evaluation in ASDs.
Advancing research on the relationship between mitochondrial disease and autism: needs, priorities and emerging tools
The afternoon sessions of the workshop focused on discussions of how to approach research to better understand the relationship between mitochondrial disease and autism: If a study were designed to look at mitochondrial function in autism, how might the study population be selected? What sorts of tests might be conducted? What challenges exist? What new research tools or resources are needed?
The workshop panelists discussed two general research strategies, either a targeted or an unbiased approach. The former would involve a thorough investigation of a relatively small ASD population selected for characteristics that indicate a greater likelihood of mitochondrial involvement. Such a strategy might involve more in-depth or invasive testing, including, for example, muscle biopsy and brain imaging with MR spectroscopy. One possibility would be to focus on ASD subpopulations with symptoms similar to those reported in mitochondrial diseases. These could include individuals who carry the chromosome 15q inverted duplication or those presenting with multisystem involvement, a regressive phenotype, epilepsy, hypotonia, or global developmental delay. Study populations might also be selected on the basis of family histories or patterns of inheritance suggestive of mitochondrial disease. Because biochemical evidence of mitochondrial dysfunction may only be present when the disease is active, the panelists noted a possible need to time tests around periods of regression or metabolic stress.
The results of a targeted study would be informative about the relationship of mitochondrial disease to autism in the selected study population and would be a reasonable first step. On the other hand, an unbiased approach would instead survey a larger, more diverse population and could inform questions about the extent to which mitochondrial disease contributes to ASDs more broadly. The workshop panelists agreed that establishing an estimate for this extent should be an initial research priority. A recent study in Portugal screened children with autism for elevated blood lactate and estimated that as many as 7.2% may have had an underlying mitochondrial disease36. One possible way to better estimate the extent of mitochondrial dysfunction in ASDs would be to mount a more controlled population-based study with more comprehensive biochemical testing.
Genetic analyses could help identify variations in the sequence or dosage of mitochondrial genes that create or modify susceptibility to autism. A number of large scale genetic studies in ASDs are currently underway, and it will be important to see if they identify associations with genes that affect mitochondrial function. The choice of methodology (genome-wide association, gene dosage, linkage analysis, or full gene sequencing) and whether to take an unbiased or targeted approach for future genetic studies will depend in part on the anticipated magnitude of a genetic influence. Workshop panelists stressed that new techniques are needed for analyzing the combined influence of multiple genetic variations, such as interactions between mtDNA haplotype and pathogenic or disease-modifying mutations. They also recommended studying the potential contribution of epigenetic modifications that affect mitochondrial function.
Due to variations across tissues in mtDNA mutation load and in the expression of mitochondrial impairment, biological specimens to collect and biomarkers to measure must be chosen carefully. The panelists' order of preference for tissue to collect for the highest mtDNA mutation load and detectability was muscle first, followed by uroepithelial cells, then saliva and lastly blood. Urine samples may be particularly promising for a research study: their collection is noninvasive, they would allow for metabolic marker assays, and they would also contain uroepithelial cells as a source of DNA for genetic analysis. Since metabolic screening and assays of mitochondrial enzyme function depend significantly on testing procedures and methods for tissue collection, handling and transport, another challenge for future research would be to improve consistency and standardization for such procedures across laboratories.
Beyond the more general absence of a definitive biomarker for mitochondrial diseases, the workshop panelists noted the lack of a marker in the brain as a challenge for assessing the contribution of mitochondrial disease to ASD. Cerebrospinal fluid (CSF) can be used for metabolic marker screening, but its collection requires an invasive procedure. Noninvasive brain imaging and MR spectroscopy can be useful for the diagnosis of mitochondrial dysfunction, particularly when combined with other diagnostic testing. In addition to further developing MR spectroscopy methodologies, workshop panelists also suggested that additional imaging approaches be applied to mitochondrial disease research, such as functional connectivity analyses, diffusion tensor imaging (DTI), functional MRI. In terms of emerging brain imaging technologies, hyperpolarized carbon-13 imaging will potentially allow direct observation of cellular metabolic processes in vivo, but the panelists agreed it will require significant further development for useful application to mitochondrial disease research.
The panelists emphasized that new research tools will need to be sensitive enough to detect mitochondrial impairments that may be too mild to manifest in a classical mitochondrial disease but that may nevertheless contribute to autism or other conditions. Some approaches under development include analysis of breath metabolites and diffuse optical spectroscopy, which would use specific wavelengths of light to detect abnormalities in mitochondrial enzyme subunits through the skin's surface. In addition to new research tools, other research needs identified by the workshop panelists included better animal models of mitochondrial diseases and infrastructural resources such as well-organized and sufficiently described patient registries.
References cited in presentations at the workshop
(NOTE: these references do not represent a comprehensive listing of the published literature on these topics.)
Deficits in Fatty Acid beta-Oxidation
Phenotypic and genotypic diversity in mitochondrial disease
Evidence from brain imaging
In Rett Syndrome
Medications that damage mitochondria
Role of mitochondria in early immune response
Thomas R. Insel, M.D.
Director, National Institute of Mental Health, National Institutes of Health
Walter Koroshetz, M.D.
Deputy Director, National Institute of Neurological Disorders and Stroke, National Institutes of Health
Daniel Salmon, Ph.D., M.P.H.
Vaccine Safety Specialist, U.S. Department of Health and Human Services
Ed Trevathan, M.D., M.P.H.
Director, National Center on Birth Defects and Developmental Disabilities, Centers for Disease Control and Prevention
Kim M. Cecil, Ph.D.
Research Associate Professor, Cincinnati Children's Hospital Medical Center
Bruce Cohen, M.D.
Department of Neurology, Cleveland Clinic Foundation
Stephen R. Dager, M.D.
Departments of Radiology, Psychiatry, and Bioengineering, University of Washington School of Medicine
Interim Director, University of Washington Autism Center,
Darryl DeVivo, M.D.
Sidney Carter Professor of Neurology and Pediatrics, Columbia University
Salvatore DiMauro, M.D.
Lucy G. Moses Professor of Neurology, Neurological Institute of New York, Columbia University Medical Center
Pauline Filipek, M.D.
Associate Professor of Pediatrics and Neurology, University of California, Irvine
James F. Gusella, Ph.D.
Director, Department of Genetics, Center for Human Genetic Research, Massachusetts General Hospital
Richard Haas, M.D.
Co-Director, Mitochondrial and Metabolic Disease Center, University of California, San Diego, School of Medicine
Robert K. Naviaux, M.D., Ph.D.
Co-Director, Mitochondrial and Metabolic Disease Center, University of California, San Diego, School of Medicine
Joseph Piven, M.D.
Director, Neurodevelopmental Disabilities Research Center, University of North Carolina, Chapel Hill
Roberto Tuchman, M.D.
Director, Autism Programs, Miami Children's Hospital Dan Marino Center
Douglas Wallace, Ph.D.
Director, Center for Molecular and Mitochondrial Medicine and Genetics, University of California, Irvine
Last updated September 30, 2008