Jennifer Bi*, Puneet Gupta, Katherine Miclau, Sarah Ryan, Yong Shen, Katherine Venturo-Conerly, Akash Wasil

Harvard College, Cambridge, MA


Autism Spectrum Disorder (ASD) has increasingly become an important topic of research due to its status as a major public health concern. It exerts immense suffering on children and families and imposes an economic burden on society. As suggested by its name, ASD does not manifest in one form or present only a single set of symptoms, but instead contains heterogeneous sets of symptoms, leading to obstacles that obstruct the path to ASD etiology discovery. Under the DSM-5, a diagnosis of ASD requires deficits in social skills, communication, and repetitive behaviors (McPartland, Reichow, & Volkmar, 2012). Traditionally, researchers have focused on identifying genes that underlie most cases of ASD and have discovered several genes that correspond to a wide range of ASD cases. This approach, however, has limitations as ASD is now understood to have a much more complex set of causes than the known set of genes. In light of evidences suggesting the complexity of ASD etiology, some researchers have shifted focus from finding a universal cause for all symptoms of ASD to systematically grouping symptoms and comorbidities, and performing genetics studies to discover a set of genetic factors contributing to these groups of symptoms. Other researchers tackle the genotypic heterogeneity by attempting to identify genetic factors underlying phenotypically homogenous subgroups of ASD. Moreover, some researchers have broadened the scope of ASD etiology investigation to include the investigation of comorbidity etiology. This review elucidates the current strategies in discovering the biological basis of ASD research and highlights some of the most recent advancements in the understanding of ASD etiology.



Autism Spectrum Disorder; Mental Health; Co-morbidity; ASD Etiology; Biomarkers



Awareness and research relating to mental disorders has increased drastically in the past couple decades. Today, there are a multitude of mental health disorders that have gained importance in the clinic. One of the more prominent, though not completely understood, mental health disorders is autism spectrum disorder (ASD). The term autism spectrum disorders (ASD), also known as pervasive developmental disorders (PDD), was developed in 2013 when the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders was released.  ASD functions as an umbrella term for developmental disorders involving primarily communication, behavioral, and intellectual issues, such as autistic disorder and Asperger syndrome.1 The Center for Disease Control and Prevention (CDC) estimated in 2016 that the prevalence of ASD was 1 in 68 children aged eight years, and more specifically, 1 in 42 for boys and 1 in 189 for girls.2 These numbers have significantly increased ever since initial studies on ASD prevalence were done in the 1990s due, in part, to increased awareness and improved screening. Currently, ASD is often difficult to diagnose in clinics due to heterogeneity in symptom profiles within ASD, comorbidities between ASD and other diagnoses, and symptom overlap between ASD and intellectual disability.  Early signs may start to appear as early as infancy, but many parents are unsure of whether these signs reflect ASD or non-pathological personality development.3 However, as a child ages, these early signs may develop into noticeable differences in social behavior and communication.4


ASD has been linked to many families and groups of genes, and new discoveries are continually being made. Though the attempt to find a single underlying gene for ASD has not been determined, research focused on specific DNA segments and proteins has been initiated. Today, researchers have begun subgrouping ASD based on genotypic uniformity and are trying to understand the etiological heterogeneity on ASD and its comorbidity.



Comorbidity and Etiology

ASD is commonly comorbid with other neurodevelopmental or behavioral disorders. Thus, a focus on one of these disorders may cause a physician to neglect the consideration or diagnosis of another. More specifically, it may result in a failure to diagnose autism.


In particular, efforts to subdivide autism and do away with the confounding phenotypic heterogeneity observed in the disorder have become increasingly focused on the presence of ASD comorbidities.  Earlier research has shown that ASD has a large variety of associated comorbidities, including gastrointestinal disorders, sleep disorders, epilepsy, psychiatric illness, and immune disorders. Recently, an attempt to stratify ASD using comorbidity clusters has been made, leading to the idea that ASD with one group of comorbidities should be viewed somewhat differently from ASD with another group of comorbidities.5

Figure 1. Comorbidities and possible factors contributing to the development of ASD.

In order to find the optimal groups, researchers analyzed the clinical trajectories of a large group of ASD patients. They examined clusters of forty-five common comorbidities in around 5,000 ASD patients, checking in with the ASD patients every six months from birth to age fifteen.  The clusters they identified showed distinct medical trajectories, i.e. they followed similar developmental paths towards similar phenotypes. The three focal clusters identified were mainly characterized by seizure, psychiatric disorders, and gastrointestinal disorders, along with other minor comorbidities. The implication is that there might be distinct etiological roots that lead to ASD with seizures, roots that lead to ASD with psychiatric illnesses, and roots that lead to ASD with gastrointestinal disorders. Therefore, these clusters can be seen as subgroups of autism with distinct comorbidities and possibly distinct causes. This approach to subgrouping autism does not rely on previously identified factors, yet it still manages to create homogenous subgroups. Certainly, this approach has great potential in contributing to the search for causative factors of ASD.

Moreover, ASD is also commonly associated with delays in language development. However, many of the language development issues that people with ASD experience are not specific to ASD, and also often occur in people with developmental disabilities and/or other types of disorders. Genetic studies show overlap between specific areas of the chromosome that seem to be associated with both the expression of Specific Language Impairment (SLI) and the communication problems that arise in ASD patients. ADHD is another example of a disorder that is commonly comorbid with ASD. One study in particular showed that the ASD + ADHD group had lower marks on the Vineland Adaptive Behavioral Scales (VABS-II) and the Pediatric Quality of Life Inventory (PedsQL). These results suggest that the presence of both ADHD and ASD contributes to a lesser quality of life due to a hindered ability to execute skills needed in everyday life.7 Moreover, in areas of social interaction, the presence of both ADHD and ASD results in language impairment, executive functioning difficulties, and learning disabilities. Thus, children who are on the spectrum are also commonly suggested to screen their children for both ASD and ADHD.

Research suggests that the co-occurrence may be the result of two separate disorders with a common etiology. That is, the two disorders may share a common genetic basis.7 To further support this theory, both twin and family studies were undertaken, and the results showed that ASD is highly genetic. Linkage studies and Genome Wide Association Studies (GWAS) have pointed to pleiotropic genes, loci, and single nucleotide polymorphisms (SNPs). Pleiotropic genes are defined as one gene affecting multiple phenotypic traits, and SNPs are defined as a variation in a single nucleotide occurring at a specific position on the genome.8 Thus, the gene has more than one allele that can cause variations in the amino acid sequence. Various brain imaging scans have also been conducted, but specialists agree that solely looking at the brain is a simplification of the problem.

Studies and experts alike have shown the co-occurrence of ASD and other disorders result in more symptoms and a more complex prognosis. However, oftentimes only one diagnosis is given. As a result, ASD can go undiagnosed or untreated. Therefore, a better genetic and biological understanding of how ASD and other disorders interact is needed in order to give patients a proper diagnosis of ASD.


Etiological heterogeneity in ASD

Many research studies have explored the roles of rare mutations and genetic imbalances in Autism Spectrum Disorder (ASD). A series of various studies of cytogenetics and whole-genome linkage of exome sequencing has shown that the etiological component of autism is extremely complex and interwoven, presenting a high degree of pleiotropy and locus heterogeneity.9


ASD has a strong genetic basis revealed by the recurrence risk in families, twin studies, and the co-occurrence with chromosomal disorders and rare genetic syndromes. The information involving biological origins autism and mechanisms continues to be updated and become more extensive with the progression of genetics and more conducted animal model systems. Currently, ASDs are diagnosed in about 1% of children, four times more common in males than in females.10


Though scientific evidence is relatively scarce, some evolutionary psychologists have identified traits in autistic individuals that may have been beneficial. These traits, such as restricted interests, could lead to innovation and have been linked to the maintenance of autism alleles in the gene pool. Recent studies have speculated that as many as 1000 genes are implicated, including both rare mutations (occurring in less than 5% of the general population) with significant effect sizes as well as common variants (occurring in more than 5%) with smaller effect sizes.9 Rare mutations with minor allele frequency are often identified in the form of chromosomal abnormalities (present in 5% of autistic individuals), Mendelian genetic syndromes (i.e. syndromic autism in 5%), rare copy number (5-10%), and de novo mutations (5-10%). Common variants, on the other hand, could also contribute to the emergence of autism: it has been estimated that as many as 40% of simplex families and 60% of multiplex families could have increased autistic traits due to single nucleotide polymorphisms.9 De novo mutations, which are microdeletions, microduplication and single nucleotide variants, occur in particularly high rates in the paternal germline copy number mutations. Many concurrent genomic mutations have been proved to lead to expression of autism, however each of these individual factors alone is only minutely responsible for ASD in and of itself. For example, each copy number variation is only present in at most 1% of individuals with autism, revealing the very significant role of genetic heterogeneity.


Both past and current research has shown that there are a multitude of genetic implications involved in ASD. Due to the complex nature of the etiology of autism, a total of 103 disease genes and 44 genomic loci have been linked with ASD or autistic behavior. Only 10%–20% of autistic individuals have an identifiable genetic etiology, with 5% of cases presenting visible chromosomal alterations, the most frequent of which include 15q11–q13 duplications, and 2q37, 22q11.2 and 22q13.3 deletions. ASD can also be due to mutations of single genes involved in X-linked disorders, autosomal dominant and recessive, the most frequent being fragile X syndrome in 2% of cases.11 Mutation in the mitochondrial DNA (mtDNA) constitutes another portion of the possible contributing factors of ASD. Recent research has shown the positive correlation between mitochondrial dysfunction and autism.11 The incidence of mitochondrial disease occurring in ASD population is about 5%, whereas in the general population that number is roughly 0.01%.12 A recent analysis of mtDNA sequences from ASD children and their unaffected siblings found a pattern of heteroplasmic mtDNA mutations related to increased risks of developing ASD.13


Finally, the etiological heterogeneity of ASD is further complicated by the plethora of associated comorbidities and recent evidence that suggests some comorbidity might actually be important factors in ASD development. Recent research showed that interferon-𝛾, a chemokine secreted by immune cells, is necessary for proper mouse social behavior.14 A single injection of interferon-𝛾 was able to restore social preferences in the mouse, and decrease hyperconnectivity in the prefrontal cortex, one of the most common characteristics of ASD.14 Disorders in the immune system, one of the comorbidities of ASD, could actually influence the development of ASD. This is not completely surprising as the immune system has been shown to influence memory formation and learning as well.15 For instance, inflammation has been shown to lead to changes in microglia. These changes can overprune synapses in patients with autism. 16 Furthermore, peripheral nervous system (PNS) disorder is found to play a role in the development of ASD. A common comorbidity of ASD is abnormal tactile sensitivity. Recent research shows that the mutation of genes Mecp2, Gabrb3, Shank3, and Fmr1 in mouse cause altered tactile sensitivity.17 Deletion of Mecp2 or Gabrb3 during early development specifically in somatosensory neurons causes social interaction deficits and anxiety-like behavior, while disruption of the same genes during adulthood leads to less severe ASD phenotypes. Rescue of Mecp2 null mutant mice specifically in somatosensory neurons with functional Mecp2 during early development restores social interaction deficits and reduces anxiety-like behaviors.17 The research suggests that tactile dysfunction, a comorbidity of ASD, could directly contribute to ASD development.


One Gene Utopia

There is currently no known biomarker or specific genomic sequence that can be linked to ASD as the main causative factor. However, in a genetic utopia, clinical and laboratory researchers would be able to find a genomic sequence or biomarker common to all ASD patients. For this purpose, researchers have recently explored four particular genes and proteins that are deemed as high risk factors for ASD, although these mutations are not present in all ASD patients.


One gene that has been investigated recently is the TSHZ3 gene that encodes a zinc finger transcription factor, which consists of a zinc finger-binding domain that allows it to bind DNA, RNA, and other proteins. The TSHZ3 gene has been shown to be one of the genes most highly expressed in the developing human neocortex, however, its exact function is not well understood. In recent research, it was found that mice heterozygous for the TSHZ3 gene were affected and displayed a change in function of the synapses that are found between cerebral cortical projection neurons.17 Not only was synapse function affected, but also the mice displayed abnormalities in behavior similar to ASD patients.18


Chromodomain helicase DNA binding protein 8 (CHD8), which encodes a chromatin remodeling protein, is another gene that is commonly affected by mutations in patients with ASD. The protein encoded has been shown to be involved in regulating the p53 pathway and CTNNB1 gene, and it interacts with CHD7 protein, which is involved in many human abnormalities and defects.19 Recently, a study showed that mice that were heterozygous for CHD8 mutations demonstrated characteristics similar to those of ASD patients, such as increased anxiety, altered social behavior, and repetitive behavior.20 Furthermore, they showed that neurodevelopment was delayed in the mice with haploinsufficiency mutation of CHD8. Moreover, this haploinsufficiency was found to be associated with the activation of Re-1 silencing transcription factor (REST), in both mice and humans. Activation of REST leads to suppression of the transcription of many neuronal genes, thereby delaying development.

Figure 2. CHD8 protein structure in humans. Diagram of the structure of the CHD8 protein with indications of different domains and motifs. Numbers positioned above and below indicate the amino acid positions.


Another protein that was recently studied is the histone acetyltransferase CREB binding protein. Mutations in the protein were found to be associated with a large group of ASD human patients after a genomic study. In particular, this protein is involved in chromatin regulation and DNA function, and controls gene expression by its role as an epigenetic regulator that modifies chromatin. In mice, it was found that those with a deletion mutation in the CH1 domain of the protein displayed behaviors similar to that of ASD patients, including hyperactivity, social interaction deficits, motor dysfunction, abnormal synaptic plasticity, impaired recognition memory, and repetitive behavior, indicating that this CBP protein is likely a high risk factor in autism.21


The fourth gene of interest is the GABRA5 gene, which encodes α5 GABAA (gamma-aminobutyric acid type A) receptors. Upon deletion of this gene, autism-associated behaviors were exhibited in mice. Under limited social contact in certain tasks, mice also exhibited memory deficits.  Moreover, they also showed excess repetitive behaviors and impaired cognitive function/problem solving abilities, which were measured through self-grooming and their reactions to a puzzle box, respectively. In analysis of 396 human cases of ASD, two rare missense variants were found in the GABRA5 gene.22 These receptors were also found to be previously downregulated and show low mRNA and protein levels in ASD patients.23,24


Given our understanding of the research advances made in these four different genes and proteins indicated as risk factors in ASD, it is clear that the underlying cause of ASD cannot yet be implicated in a specific biomarker or gene. Many of the genes and proteins that have been found to be involved in ASD cases are usually found to be involved in other issues and diseases as well, making ASD a difficult disorder to categorize and treat.
Subgrouping ASD based on genotypic uniformity

One approach that researchers have taken to tackle this apparent genotypic heterogeneity is subgrouping ASD in such a way that each genetic problem corresponds to a phenotypically homogenous subgroup of ASD. These studies generally attempt to correlate specific genotypes to homogenous phenotypic expressions. One study assumed phenotypic heterogeneity arises from genotypic hetereogeneity of ASD and homogenous genotype causes phenotypic homogeneity.25 In accordance with this hypothesis, the group isolated a group of ASD patients with a specific genetic mutation implicated in ASD, 22q11 deletion syndrome, and compared the phenotypic homogeneity within that group to the phenotypic homogeneity in a mixed group of ASD patients. They found that the group with the same genotype had a much greater phenotypic homogeneity than that of the mixed group. These results imply that clinical heterogeneity of ASD can be reduced by subgrouping ASD patients based on specific genotypes.


However, as studies discover more genes implicated in ASD behaviors, it is important to recognize the limitations of such findings. In another study, the authors highlight how ASD is not a single clinical entity but instead a behavioral manifestation of anywhere from tens to hundreds of genetic and genomic disorders.9 Thus, the behavioral presentation of the individual should be taken into consideration when focusing on genetic mutations due to the complexity of ASD as an individually-specific disorder with various degrees of intellectual disability (ID) severity. Furthermore, autism is often associated with other neuropsychiatric disorders such as attention deficit-hyperactivity disorder, bipolar disorder, and obsessive compulsive disorder. Indeed, many of these studies indicating genetic mutations leading to ASD were conducted without indication of how formal diagnostic evaluation was performed.


The specific studies on the similar genetic causes of ID and ASD (70% comorbidity), and of ASD and epilepsy (25% comorbidity) indicate these neurodevelopmental disorders share common genetic bases. These studies redefine autism as the final common pathway for often co-occurring genetic brain disorders: well-recognized ID-genes, which do not always result in symptomatic ID, as well as some gene mutations characteristic of epilepsy can also correlate to ASD.9 Instead of considering ASD as one distinct disease, these genetic findings portray ASD as a continuum of neurodevelopmental disorders manifesting in diverse manners based on other genetic, environmental or stochastic factors. In fact, many disorders are well known to have a very high comorbidity with ASD: the most frequent are SHANK3 mutations, tuberous sclerosis (TSC1, TSC2), Rett syndrome (MECP2), fragile X syndrome (FMR1), Smith–Lemli–Opitz syndrome (DHCR7), cortical dysplasia-focal epilepsy syndrome (CNTNAP2), and adenylosuccinate lyase deficiency (ADSL).9



In order to better understand the underlying biological basis of ASD and to continue the advancement of treatment of future ASD individuals, it is essential to further emphasize the role of ASD in conjunction with other mental, neurodevelopmental or behavioral disorders. Because autism is caused by such a large variety of genetic abnormalities that cannot be linked back to a singular mutation or a uniform change in the genomic sequence consistent for all ASD individuals, it is very important to consider the nonhomogeneous nature of ASD and embrace its genomic complexities. As comorbidity is so common and widely varied for every affected individual, identifying the common genotypic base of ASD in relation to other disorders has been shown to strongly correlate to the prominent phenotypic groups in autistic patients. Focusing on subgrouping autism into more specific, phenotypically homogenous groups of ASD individuals is a promising approach to treating individuals with common symptoms and predicting future trajectories; particularly, the three most indicative clusters of recurrent comorbidities can be characterized by seizure, psychiatric disorders and gastrointestinal disorders. Continuing to identify various comorbidities into more individualized and specific subgroups will aid in furthering our understanding of the complexities of autism and exactly what characterizes it.


Research into Autism Spectrum Disorder (ASD) presents a discourse of questions and analysis concerning the biological basis and optimal treatment methods for the disorder. Although this type of research is necessary, it is also imperative to discuss and consider the sociocultural factors that play a role in ASD. These factors have significant implications for both diagnosis and treatment. When striving to understand the social and cultural basis of ASD, the hope is not to bring attention away from studying the underlying biology; rather, this research seeks to provide further insight into understanding the biology and provide a bridge between the multifaceted components and complexities of this disease.

An overwhelming amount of research makes clear one simple fact: diagnosis of autism requires recognition of autism. While this seems intuitive, a failed or delayed diagnosis is often a result of one of three circumstances: an inability to construct a universal, concrete framework for signs of autism, the failure to acknowledge cultural differences in defining the disease, and the tendency to attribute signs of autism to external or coexisting conditions. Any combination of these factors influences when and how someone is diagnosed with autism; whether it be by a family member, community member, or physician. Ultimately, there needs to be more emphasis on and responsibility for recognizing the disease in its various yet different forms. Autism is not a disease that can be generalized. Therefore, it is best to assess and diagnose each person individually.



  1. Rybakowski F., Bialek, A., … and Dunajska, A. (2014) Autism spectrum disorders—epidemiology, symptoms, comorbidity and diagnosis, Psychiatr Pol, 48(4), 653–65.
  2. Christensen D., Baio J., Braun K., … and Yeargin-Allsopp, M. (2016) Prevalence and Characteristics of Autism Spectrum Disorder Among Children Aged 8 Years – Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, MMWR Surveill Summ, 65(No. SS-3)(No. SS-3), 1–23.
  3. Tarbox, J., Matson, Johnny L., Sturmey, P., and Dixon, D. (2014) Handbook of Early Intervention for Autism Spectrum Disorders : Research, Policy, and Practice (Autism and Child Psychopathology Series), New York, NY, Springer New York, Imprint: Springer, pg 1.
  4. Lord, C., Cook, E., Leventhal, B., Amaral, D. (2000) Autism spectrum disorders, Neuron 28, (2), 355-363.
  5. Doshi-Velez, F., Ge, Y., and Kohane, I. (2014) Comorbidity clusters in autism spectrum disorders: An electronic health record time-series analysis, Pediatrics, 133(1), E54-63.
  6. Leyfer, O., Tager-Flusberg, H., Dowd, M., Tomblin, J., and Folstein, S. (2008) Overlap Between Autism and Specific Language Impairment: Comparison of Autism Diagnostic Interview and Autism Diagnostic Observation Schedule Scores, Autism Research 1, 284-296.
  7. Leitner, Y. (2014) The co-occurrence of autism and attention deficit hyperactivity disorder in children – what do we know?, Frontiers In Human Neuroscience, 8.
  8. Solovieff, N., Cotsapas, C., Lee, P., Purcell, S., and Smoller, J. (2013) Pleiotropy in complex traits: Challenges and strategies, Nature Reviews Genetics, 14(7), 483.
  9. Betancur, C. (2011) Etiological heterogeneity in autism spectrum disorders: More than 100 genetic and genomic disorders and still counting, Brain Research, 1380, 42-77.
  10. Lai, W., Goh, T., Oei, T., and Sung, M. (2015) Coping and Well-Being in Parents of Children with Autism Spectrum Disorders (ASD), Journal of Autism and Developmental Disorders, 45(8), 2582-2593.
  11. Giulivi, C., Zhang, Y., Omanska-Klusek, A., and Pessah, P. (2010) Mitochondrial Dysfunction in Autism Cecilia, Jama, 304, 2389–2396.
  12. Rossignol, D., and Frye, R. (2012) Mitochondrial dysfunction in autism spectrum disorders: a systematic review and meta-analysis, Molecular Psychiatry, 17(3), 290–314.
  13. Wang, Y., Picard, M., and Gu, Z. (2016) Genetic Evidence for Elevated Pathogenicity of Mitochondrial DNA Heteroplasmy in Autism Spectrum Disorder, PLOS Genetics, 12(10), e1006391.
  14. Filiano, A., Xu, Y., Tustison, N., Marsh, R. Baker, W., Smirnov, I., … and Kipnis, J. (2016) Unexpected role of interferon-γ in regulating neuronal connectivity and social behaviour, Nature, 1–18.
  15. Derecki, N., Cardani, A., Yang, C., Quinnies, K., Crihfield, A., Lynch, K., & Kipnis, J. (2010) Regulation of learning and memory by meningeal immunity: a key role for IL-4, J Exp Med, 207(5), 1067–1080.
  16. Orefice, L., Zimmerman, A., Chirila, A., Sleboda, S., Head, J., & Ginty, D. (2016) Peripheral Mechanosensory Neuron Dysfunction Underlies Tactile and Behavioral Deficits in Mouse Models of ASDs, Cell, 166(2), 299–314.
  17. Pardo, C. A., Vargas, D. L., & Zimmerman, A. W. (2005). Immunity, neuroglia and neuroinflammation in autism. International Review of Psychiatry, 2005, Vol.17(6), p.485-495, 17(6), 485–495.
  18. Caubit, X., Gubellini, P., Andrieux, J., Roubertoux, P. L., Metwaly, M., Jacq, B., … and Fasano, L. (2016) TSHZ3 deletion causes an autism syndrome and defects in cortical projection neurons, Nature Genetics, 48(11), 1359–1369.
  19. Krumm, N., O’Roak, B., Shendure, J., and Eichler, E. E. (2014) A de novo convergence of autism genetics and molecular neuroscience Trends in Neurosciences, 37(2), 95–105.
  20. Katayama, Y., Nishiyama, M., Shoji, H., Ohkawa, Y., Kawamura, A., Sato, T., … and Nakayama, K. (2016) CHD8 haploinsufficiency results in autistic-like phenotypes in mice, Nature, 537(7622), 675-679.
  21. Zheng, F., Kasper, L., Bedford, D., Lerach, S., Teubner, B., and Brindle, P. K. (2016) Mutation of the CH1 Domain in the Histone Acetyltransferase CREBBP Results in Autism-Relevant Behaviors in Mice, PLoS ONE, 11(1), e0146366.
  22. Zurek, A., Kemp, S., Aga, Z., Walker, S., Milenkovic, M., Ramsey, A. J., Sibille, E., Scherer, S., and Orser, B. (2016) α5GABAA receptor deficiency causes autism-like behaviors, Ann Clin Transl Neurol, 3: 392–398.
  23. Fatemi, S., Reutiman, T., Folsom, T., et al. (2010) mRNA and protein levels for GABAA α4, α5, β1 and GABABR1 receptors are altered in brains from subjects with autism, J Autism Dev Disord, 2010;40:743–750.
  24. Fatemi, S., Reutiman, T., Folsom, T., Thuras, P. GABAA receptor downregulation in brains of subjects with autism. J Autism Dev Disord 2009;39:223–230.
  25. Bruining, H., De Sonneville, L., Swaab, H., De Jonge, M., Kas, M., Van Engeland, H., … and Aziz, S. (2010) Dissecting the Clinical Heterogeneity of Autism Spectrum Disorders through Defined Genotypes (Genetic Subphenotypes in ASDs), PLoS ONE, 5(5), E10887.