3.1 Clinical Aspects
Only a limited number of studies have examined the clinical characteristics of children with ASD and MD. Weissman et al. [
15] was one of the first compilations of clinical symptoms of children with ASD diagnosed with MD (ASD/MD). In their case series of 25 patients, they noted a high rate of non-neurological medical problems, including gastrointestinal (GI) dysfunction and prenatal or perinatal complications, constitutional symptoms such as excessive fatigability and exercise intolerance, early gross motor delay and unusual patterns of regression, including multiple regression and regression after 3 years of age.
Rossignol and Frye [
3] reviewed the clinical characteristics of all the reported cases of children with ASD/MD and compared these to the clinical characteristics of the general ASD population (ASD/NoMD) as well as children with MD but not ASD (MD/NoASD). As compared to ASD/NoMD children, ASD/MD children had a higher rate of neurodevelopmental regression, seizures, gross motor delay and GI abnormalities. As compared to MD/NoASD, ASD/MD children demonstrated higher rates of fatigue and lethargy, ataxia, GI abnormalities and normal brain imaging and were less likely to have abnormal light microscopy. ASD/MD children were more likely to have elevated lactate than both MD/NoASD and ASD/NoMD groups.
Frye and Rossignol [
16] pointed out that many of the clinical symptoms outlined in the Morava criteria [
17], which set forth clinical diagnostic criteria for MD, overlap with characteristics of children with ASD, including developmental delays, seizures, neurodevelopmental regression, GI and endocrine abnormalities, familial recurrence and neuropathies.
In the evaluation of the etiology of a person with ASD, the presence of regression especially with recurrent episodes and multiple organ dysfunctions should prompt an extended evaluation for MD [
18,
19]. Indeed, Shoffner et al. [
20] demonstrated that most patients diagnosed by his center with ASD/MD experienced neurodevelopmental regression resulting in the development of ASD following an inflammatory event associated with a fever within the preceding 2 weeks.
Several case series have noted the association between mitochondrial dysfunction in ASD and epilepsy [
15,
20]. Indeed, one of the first descriptions of the overlap of mitochondrial dysfunction and ASD was the HEADD syndrome, characterized by hypotonia, epilepsy, autism, and developmental delay [
21]. These authors noted electron transport chain (ETC) complex deficiencies in subunits encoded by mitochondrial DNA (mtDNA) and that several patients demonstrated large-scale mtDNA deletions. Interestingly, these authors noted a high rate of ETC Complex III (C3) deficiency, which has been echoed in other case reports of children with ASD/MD and epilepsy [
22,
23]. Another interesting case report linked mitochondrial dysfunction with ASD in a patient with Dravet syndrome, suggesting that even clearly genetically based epilepsy with ASD could have a mitochondrial component [
24].
Lastly, one report suggests that children with ASD and classic MD are more likely to also have intellectual disabilities, suggesting early identification and treatment may be particularly useful for these children [
25].
3.3 Genetic Aspects of Mitochondrial Disorders in Autism Spectrum Disorder (ASD)
Although it has been estimated that the genetic etiology of ASD may account for up to 40% of cases [
83,
84] and whole exome sequencing (WES) and chromosomal microarray analysis (CMA) studies have reported yields up to 30% [
85] and 26% [
86], respectively, separate clinical studies have failed to confirm this high rate of genetic disorders in children with ASD. For example, a study from Canada found that 9.3% and 8.4% of children with ASD received a molecular diagnosis using WES and CMA, respectively, resulting in only 15.8% of children with ASD receiving a molecular diagnosis [
84].
The moderate rate of a molecular diagnosis is at odds with the high heritability rate associated with ASD; for example, there is a 70–90% concordance rate for monozygotic twins and up to 10% for dizygotic twins and a 25-fold increased prevalence of ASD in siblings of ASD children [
87]. Thus, it is likely that the etiology of ASD is multifactorial and influenced by a complex interplay between the inherited genome and environmental effects, some of which may be related to the maternal environment [
88,
89]. Epigenetic interactions which modulate the expression of nuclear [
90] and mitochondrial [
91] genes can be influenced by environmental factors and may also play a crucial role in ASD. These interactions must surely modulate mitochondrial function, influencing neurodevelopment. Indeed, alterations in multiple bioenergetic and metabolic genes required for mitochondrial function may lead to abnormalities in cerebral activity, resulting in cognitive and behavioral abnormalities characteristic of ASD [
92].
Mitochondrial genes can be affected by copy number variations (CNVs) or regions of homozygosity (ROH). Chromosomal regions affected by CNVs can contain genes associated with mitochondrial function and neurodevelopmental disorders. For instance, a 7q31.1 deletion/duplication disrupts the IMMP2L gene encoding an inner mitochondrial membrane protease-like protein required for processing of cytochromes inside mitochondria and is implicated in ASD [
93]. The IMMP2L–DOCK4 gene region on chromosome 7 plays a role in ASD susceptibility [
94], and IMMP2L deletions have been demonstrated to have an association with ASD [
95]. However, deleterious point mutations in IMMP2L were not identified in a significant number of ASD patients in other studies [
96]. Mitochondrial dysfunction was also demonstrated in ASD patients with other CNVs, including 15q11-q13 duplication [
22,
60], 5q14.3 deletion [
48], 22q13 duplication or Phelan–McDermid syndrome [
49,
97] as well as chromosomal disorders such as Down syndrome [
98] and X- and Y-chromosome loci rearrangements [
99]. An interesting example is Phelan-McDermid syndrome, which is typically caused by a microdeletion in the 22q13 region. Although much research has concentrated on SHANK3, which is important for synaptic function, six mitochondrial genes are also present in this region and may account for symptoms associated with mitochondrial dysfunction [
49].
ROH are regions that are identical in homologous regions of paired chromosomes. Small ROH reflect our common human inheritance; larger and more numerous ROH can occur because of chromosomal segments shared between consanguineous family members. ROH can include genes associated with mitochondrial function as well as cerebral synapses and neurotransmitters that are associated with ASD [
100].
Primary mitochondrial disease (PMD) involves a genetic defect that results in an impairment of mitochondrial oxidative phosphorylation [
101]. PMD is estimated to affect 5% of children with ASD [
3], based upon three large studies [
102‐
104]. Mutations in mtDNA are the most common genetic mutations associated with ASD/MD [
3]. Numerous mtDNA point mutations [
105,
106] and deletions [
107] show association with ASD; large-scale mtDNA deletions have been associated with epilepsy [
21]. Examples of mtDNA genes reported include MT-ATP6 [
85,
108,
109], MT-ND5 [
108], MT-CYB [
110], MT-TK [
111] and MT-TL1 [
112], and MT-CO1 and MT-CO2 [
113]. Mutations in mtDNA such as m.3260A > G that causes mitochondrial encephalomyopathy, lactic acidosis and stroke-like episodes (MELAS) have been associated with ASD [
104,
112], as has primary Leber hereditary optic neuropathy mtDNA mutations [
104]. ASD can be an early presentation of another MELAS mutation (m.3243A > G) [
114], sometimes with a prominent manifestation of mtDNA depletion syndrome [
115]. Kent et al. [
116] found m.3243A > G to be a rare cause of isolated ASD. mtDNA lineages or haplogroups significantly contribute to overall ASD risk in some studies [
117,
118], while other studies have not confirmed a major role for mtDNA variation in ASD susceptibility [
119,
120]. Increased mtDNA copy number in leukocytes from children with ASD has been reported in six studies [
121‐
126], while increased mtDNA damage and deletions in leukocytes from children with ASD have been reported in four studies [
121,
122,
126,
127]. Interestingly, Wong et al. also found an increase in microdeletions of p53, which is a regulator of mtDNA integrity [
126], leading others to suggest that changes in mtDNA may be an epiphenomenon of genetic abnormalities in nuclear DNA (nDNA) mutations [
128].
nDNA gene mutations can also cause PMD. Mutations in NDUFA5 [
129,
130], NDUFS4 [
108], POLG [
104] and SCO2 [
104] have also been found to be associated with ASD. SNPs in the SLC25A12 gene have been reported by multiple authors to be strongly associated with ASD [
131‐
143]. Mutations in SLC6A8, resulting in X-linked creatine transporter disorder including features of ASD, have been reported [
144]. Heterozygous mutations in adjoining components in a multimeric complex or metabolic pathway may combine to lead to impairment through synergistic heterozygosity. This is well-described, but not yet for the mitochondrial–ASD relationship. Since these types of relationships are complex, bioinformatics platforms used to interpret WES data may not be able to detect these complex genetic interactions. In addition, novel variants of unknown significance can be difficult to interpret, and in some situations, a mutation may cause gain of function, and act in a dominant manner. A recently developed method called transmission and de novo association (TADA) analysis considers both transmitted and de novo mutations together in the same statistical model by weighting more detrimental mutations such as de novo loss-of-function mutation more heavily than an inherited loss-of-function mutation [
145].
Genetic abnormalities have not been reported in most ASD/MD cases [
3]. In addition, the number of children with ASD who have abnormal nutritional and mitochondrial biomarkers greatly outweighs the number who can be diagnosed with a PMD. This raises the possibility that many children with ASD may have secondary mitochondrial dysfunction (SMD). There are examples of SMD related to hereditary defects in non-mitochondrial diseases. For example, a pair of siblings with ASD-associated gene mutations in WDR45 and DEPDC5 were found to have evidence of mitochondrial dysfunction [
50]. In another interesting case series of children with Dravet syndrome and SCN1A mutations, two cases were found to manifest mitochondrial dysfunction, and one of the cases had ASD [
24]. In another case, lymphoblastoid cell lines (LCLs) derived from a child with ASD and a mutation in RPL10 were found to have changes in redox-sensitive components of energy metabolism [
146]. Interestingly, candidate genes implicated in ASD have been found to be enriched in modules related to mitochondrial function [
147]. These cases demonstrate the importance of advanced genetic testing combined with metabolic/mitochondrial evaluation in the workup of children with ASD [
71].
Mutations in non-mitochondrial genes or environmental factors may be acting via epigenetic mechanisms in ASD [
101,
148]. Down-regulation of genes of mitochondrial oxidative phosphorylation and varying gene expression related to myelination, inflammation and purinergic signaling have been previously identified in ASD patients [
149]. Others have found upregulation of ribosomal, spliceosomal, and mitochondrial pathways and the down-regulation of neuroreceptor-ligand, immune response and calcium signaling pathways in gene expression profiles of ASD patients [
150]. One meta-analysis of over 1000 microarray samples across 12 independent studies demonstrated that genes highly ranked with consistent changes in expression in the brain suggested modulation of mitochondrial function [
151]. However, one study examining expression of the C1 75-kDa subunit in blood did not find consistent changes in children with ASD [
152].
Other studies have examined the expression of genes in brain tissue in ASD. Depressed expression in ETC genes in the occipital and cerebellar areas and ETC and non-ETC genes in the cingulate, thalamus and frontal areas [
153] has been reported. In addition, changes in genes that control mitochondrial dynamics have been noted in the temporal lobe in ASD [
154]. Still others have demonstrated brain region-specific expression alterations in mitochondrial nDNA genes such as CMYA3, MTX2, SLC25A27, DNAJC19, DNM1L, LRPPRC, SLC25A12, SLC25A14, SLC25A24 and TOMM20 in ASD patients [
131,
132].
3.5 Mitochondrial Dysfunction in the Brain in ASD
Since the brain has a high metabolic demand and is especially dependent on mitochondrial function, abnormalities in mitochondria function would be expected to cause brain dysfunction. Studies provide support for mitochondrial dysfunction in the brain of individuals with ASD. Magnetic resonance spectroscopy (MRS) using both
31P and
1H techniques have examined energy metabolites in the brain of individuals with ASD.
31P-MRS has reported abnormal energy metabolites in frontal cortex [
180,
181]. Some
1H-MRS studies have found a reduction in N-acetylaspartate (NAA) in the global white and gray matter and the parietal, anterior cingulate and cerebellum [
182] associated with ASD, while others have not been able to detect differences in NAA [
183]. Another study in adults with Asperger syndrome has shown an increase in NAA and choline in the prefrontal cortex, with concentrations of these metabolites correlating with obsessive behavior and social function, respectively [
184]. An interesting meta-analysis using meta-regression suggests that these discrepancies may be due to an age-related decline in brain NAA specific to those with ASD [
185].
Studies using
1H-MRS have found an increase in lactate in the cingulate gyrus, subcortical gray matter nuclei, corpus callosum, superior temporal gyrus, and pre- and post-central gyri, with this abnormality being more common in adults with ASD [
186], while others have not found any increase in lactate in individuals with ASD [
187]. MRS studies on lactate in the brain are not without controversy. Some have suggested that the fact that lactate was found more often found in adults with ASD may reflect the notion that lactate was associated with comorbidities such as anxiety rather than related to the true etiology of ASD [
188]. The original authors did not agree and suggested that perhaps worsening of mitochondrial function with age or ascertainment bias in the recruitment of adults with ASD might be a more likely explanations [
189]. Others have pointed out that early MRS techniques may not be powerful enough to determine lactate abnormalities, especially because such early techniques are not uncommonly negative even in individuals with known MD, thus potentially explaining the negative findings; the lack of contemporaneous controls with MD make the interpretation of negative findings even more problematic [
190]. The authors of the original study pointed out that their use of propofol anesthesia could increase the sensitivity of their study by provoking mitochondrial dysfunction in the patients with an underlying MD [
191].
ETC function has also been directly measured in post-mortem brain tissue from individuals with ASD. ETC function has been reported to be depressed in frontal [
192], temporal [
154,
192] and cerebellar [
192] areas from ASD-derived brain tissue. Other studies noted decreases in the activity of non-ETC mitochondrial enzymes (aconitase, pyruvate dehydrogenase) in frontal [
193], temporal [
173] and cerebellar [
173] tissue derived from children with ASD.
3.8 Treatments of Mitochondrial Dysfunction in ASD
There is only one study published to date that examined the behavioral effects of a customized mitochondrial supplement in ASD children with mitochondrial dysfunction. Legido et al. [
55] in an open-label study treated 11 children with ASD and abnormal C1 and/or C4 activity with a mitochondrial cocktail containing carnitine, coenzyme Q10 and α-lipoic acid. Three months of treatment reduced the C1-to-C4 ratio as well as improved several behavior scales, including lethargy and inappropriate speech subscales of the Aberrant Behavior Checklist. Three months after withdrawal of the treatment, the lethargy and inappropriate speech subscales significantly worsened.
As previously mentioned, in a relatively large study, we have examined whether various treatments for mitochondrial disorders influenced enzymatic activity in children with ASD using the buccal swab technique [
54]. Overall, the study demonstrated that folate, cobalamin, fatty acids and antioxidant supplementation increased mitochondrial enzymatic activity and folate and cobalamin influenced that relationship between enzyme activity, suggesting increased ETC coupling.
As we have previously reviewed, although most studies have not selected individuals with ASD and mitochondrial dysfunction for treatment with supplements that target the mitochondria, several studies have demonstrated that mitochondrial supplements may be helpful in children with ASD [
73]. For example, carnitine deficiency appears to be common in children with ASD [
58,
70]. Two double-blind, placebo-controlled studies demonstrated improvement in ASD symptoms with carnitine supplementation [
74], with some improvements directly related to the change in blood carnitine levels [
72]. In another study, reduced NAD and ribose appeared to improve metabolic biomarkers in children with ASD and symptoms of mitochondrial dysfunction [
222]. Two studies have reported behavioral improvements in children with ASD using ubiquinol [
223] and coenzyme Q10 [
224].