ASDs are a highly variable spectrum of conditions emerging in early development, associated with many genes, various medical comorbidities, and levels of intelligence ranging from profoundly impaired to highly gifted. Historically they were defined behaviorally [1[1], and this remains the case today, given that no one genetic or physiological feature has yet been found consistently across all individuals who meet behavioral criteria for this spectrum disorder. In 2014, the Centers for Disease Control and Prevention (CDC) estimated that 1 in 68 children had ASDs as of 2010 [2[2]; this was a 30% increase over the figure announced in 2012. Given that these spectrum conditions clearly have major and apparently growing human, social, and economic impacts, a clearer understanding of their pathogenesis and mechanisms could have great public health significance.

The heterogeneity in ASDs is present between individuals across multiple levels, including nuances of behavioral manifestations as well as genetic, brain, physiological and medical features—and yet everyone with this diagnosis meets the same defining behavioral criteria, which include social impairments, communication difficulties, and restricted, repetitive, and stereotyped patterns of behavior. Cancers also involve great heterogeneity – not only are there many kinds of cancer, but within cancer types there is also heterogeneity. Tumor heterogeneity has been observed in many cancers including of breast, prostate, colon, brain, bladder and blood. Different tumor cells can show distinct morphological and phenotypic profiles, involving characteristic features at the levels of cellular morphology, gene expression, metabolic profiles, motility, proliferation, and metastatic potential.

A growing number of published studies from various disciplines and perspectives have identified ways that ASDs are also related to other conditions with very different phenotypes, such as other neurological diseases, cancer, metabolic conditions, and heart diseases [[3], [4], [5], [6]]. While this additional dimension may add further complexity to our attempts to understand ASDs, on the other hand looking at ASDs in a framework that also includes other diseases and conditions may allow us to discern patterns and features that might not be so easily perceived by looking at ASDs – or for that matter at any of the other apparently related conditions – alone.

Recently researchers have looked for overlapping genes and pathways between ASDs and cancers, and studies to date have consistently observed them [[7], [8]]. But it is not yet clear what common mechanism(s) between these two conditions and amongst the overlapping genes and pathways might exist. Here we discuss the possible relationships between ASDs and cancers, and propose a hypothesis of how they may overlap and what may contribute to their common features.

The identification of autism risk genes that were also related to cancer was the first indication of an autism-cancer link. This evidence emerged from examining findings from studies of copy number variants and epigenetic modifiers, and data from genome-wide sequencing of individuals with ASDs. A recent study, which assessed characteristics of ASD risk genes based on protein functions, found that 43 specific genes associated with autism susceptibility also have associations with cancer [9[9]. These genes are involved in varieties of biological activities in relation to “chromatin remodeling and genome maintenance, transcription factors, and signal transduction pathways leading to nuclear changes” [9[9], which are known to be associated with tumorigenesis.

As of December 2016, the SFARI (Simons Foundation Autism Research Initiative) Gene-Human Gene Module recorded 859 human genes implicated as relevant to ASDs (https://gene.sfari.org/autdb/HG_Home.do). The Cancer Gene Census, in its release v80 (February 13th, 2017) of a list of all cancer census genes from COSMIC (the Catalogue Of Somatic Mutations In Cancer) recorded 616 genes for which mutations have been causally implicated in cancer (https://cancer.sanger.ac.uk/census). We compared the list of the SFARI genes to the list from the Cancer Gene Census, and found these two gene lists shared 77 genes in common (Table S1Table S1). These overlapping genes include both oncogenes such as AR, BRAF, GNAS, and HRAS, and tumor suppressor genes such as APC, BRCA2, NF1, and TSC1. It is notable that ASD associated genes include both oncogenes and tumor suppressor genes (TSG). Table S1Table S1 lists the ASD and cancer overlapping genes, their molecular function, as well as their roles in ASDs and cancer, in reference to SFARI and Cancer Gene Census. Oncogenes are genes that have the potential to cause cancer; while tumor suppressor genes, also called antioncogenes, are genes that could protect cells from becoming tumor cells. In tumor cells, oncogenes are often over-expressed or mutated, while tumor suppressor genes may not work properly. While these genes may play non-cancer-related roles with regard to how they contribute to the features of Autism Spectrum Disorder, it is possible that they may also affect the risk in these individuals of developing cancer.

Although one might infer that if people with ASDs have genes in common with cancer, they may also have increased risk of developing cancer, whether the rates of cancer in ASDs are actually increased is rather unclear. A study conducted in 2010 reviewed 702 ASD cases and did not find correlations between ASD diagnosis and incidence of childhood cancer [10[10]. Later in the same year, an epidemiological study investigated the relationship between incidence of autism and rates of cancers based on state-wide prevalence of these two conditions, and found correlations between autism and in situ breast cancer. A database analysis published in 2015 that used data from the Taiwan National Health Insurance database compared the number of cancers in years 1997–2011 in patients with autism with a standardized incidence ratio (the expected number). The authors investigated over 8000 cases and found 20 cases with cancer, which “was significantly higher than a total number of expected cancers with a standardized incidence ratio (SIR) estimate of 1.94 (95% CI 1.18–2.99)”[11[11]. However, a recent study using data from the University of Iowa Hospitals and Clinics' electronic medical record reported that patients with ASDs had lower cancer rates [12[12]. More recently, another study using registry data from the Department of Health of Western Australia found that mothers of children with autism but without intellectual disability had an increased risk of cancer [13[13].

Thus, the evidence to date is inconsistent regarding whether people classified as having ASDs have different rates of cancers than the general population. It is possible that although ASDs have many genes in common, it is far from the case that all of these genes are carried in common across people with a ASD diagnosis; thus, as a group, people carrying this diagnosis may have neither higher nor lower cancer rates. This may be due to the opposing impacts of the involvement of both oncogenes and tumor suppress genes. It is also possible that subsets not yet identified within ASDs may have greater autism risk. For example, the above-cited study identifying greater risk of cancer in a subset of mothers of children with autism [13[13] suggests the value of designing studies to assess the risk of cancer outcomes in children of such mothers.

It is not novel to find that genes and pathways that are implicated in neurological diseases have also been associated with cancer [14[14]. Conversely, cancer genes have also been identified as playing roles in neural development. For example, BRCA1 (Breast cancer susceptibility gene 1), which is a breast and ovarian cancer tumor suppressor, has been found to be involved in brain development [15[15]. Chromatin regulators such as EZH2 and CHD8 have been found to play essential roles throughout neural development [16[16]. It would be of great interest to seek to identify common features in the contributions of these genes to neural development on the one hand and cancer on the other.

Both ASD and cancers involve altered cellular or system development. We examined the 77 ASD and cancer shared genes from Table S1Table S1 and found that they indeed were enriched for “Cell Development” under the category Biological Process (BP) of Gene Ontology (GO). To do this, we investigated the 77 genes in Molecular Signatures Database, MSigDB [17[17] v6.0 (http://www.broadinstitute.org/gsea/msigdb/index.jsp). Gene symbol was used as gene identifier to import the genes. We applied the “Compute Overlaps” tool from MSigDB website, under “Investigate gene sets” category, which uses the hypergeometric distribution to examine how the ASD and cancer common genes may overlap with GO-BP gene set. The result shown that the common genes were enriched for GO_CELL_DEVELOPMENT (p-value 2.27 e⁻²³, Table S2Table S2). These initial findings suggest that further investigations of the common features between ASDs and cancers regarding developmental processes might well be fruitful.

One implication of the fact that a remarkable number of ASD-associated genes are also associated with cancer, as well as the fact that cancer associated genes also function in neural development, is that the functions associated with these genes might be associated with processes whose disturbance can increase vulnerability to either cancer, ASDs or both, depending on circumstances. Such overlap might be due to pleiotropy, i.e. genes playing multiple roles and participating in diverse systems. It is possible that abnormalities or defects in the function of these gene in different systems—either during different developmental periods, or in the setting of differences in other pertinent genes—may potentially lead to different trajectories, culminating in some cases in neurological conditions and in other cases in cancer. There is also the possibility that identifying mechanisms operating in genes that are associated with both autism and cancer may allow us to penetrate more deeply into core pathophysiological mechanisms in both conditions. To explore such possibilities in more depth, we need further “3-Dimensional” systems pathophysiological studies that characterize 1) how these genes function together, 2) where these genes function and 3) when these genes exert their effects. With these “how-where-when” dimensions better specified, similarities and differences between conditions with overlapping genes can be better understood.

Since genes work in groups, when investigating genes and their potential contribution to diseases it is important to understand how genes function together. Taking this systems-oriented approach, a pathway network analysis focusing on gene-pathway relationships and pathway-pathway interactions in ASDs yielded major overlaps with cancer [8[8]. This study found that ASD-associated genes were enriched for KEGG (Kyoto Encyclopedia of Genes and Genomes) cancer-associated pathways (e.g. “Pathways in cancer”). It also identified a key pathway, “MAPK signaling pathway,” as likely playing an important role in ASDs, as inferred from its prominence in the network findings. It is notable that defects in MAPK signaling are also implicated in causing uncontrolled cell growth that could lead to cancer [18[18].

MAPK signaling has additionally been identified as associated with the development of neurological diseases including Alzheimer's disease [AD], Parkinson's disease (PD), and amyotrophic lateral sclerosis (ALS)[19[19]. This suggests that neurological conditions might share some common mechanisms at the functional level with cancer as well as with ASDs in relation to MAPK signaling.

Besides MAPK signaling, other pathways, especially signal transduction pathways, were identified as being involved in both autism and cancer. For example, PI3 K-Akt signaling, mTOR signaling and Ras signaling pathways have been discussed in recent years as likely related to ASDs [[8], [20], [21], [22], [23]]. It is interesting to note that all these pathways are also known to be involved in cancer and have been discussed extensively in this regard [[24], [25], [26], [18]].

Also of note, these signaling pathways profoundly influence cellular metabolism and are critical for metabolic processes such as glucose metabolism, anabolic metabolism, and biosynthesis of proteins and lipids. [[27], [28], [29], [30]].

We compared the genes that participate in the Ras, MAPK, PI3 K-Akt and mTOR signaling pathways (in reference to KEGG Pathway Database) to the 77 ASD and cancer common genes (Table S1Table S1) and found that some of the common genes indeed participate in these signaling pathways (Table 1Table 1). This supports the importance of more deeply investigating these overlaps regarding whether and how these pathways may be involved in both ASDs and cancer.

The signaling pathways regulate each other and co-regulate a variety of functions. Fig. 1Fig. 1 illustrates how these pathways interact with each other in a cell. The receptor tyrosine kinases (RTKs) on the cell membrane activate Ras/MAPK/PI3 K/Akt/mTOR signaling, by binding to extracellular growth factors;, and then intracellular signal molecules Grb2 (growth factor receptor-bound protein 2), Sos (son of sevenless) and IRS-1 (insulin receptor substrate 1). Activated Ras trigger a phosphorylation cascade Raf-MEK-MAPK. PI3 K activation can be accomplished in three different ways: 1) by binding to the phosphorylated intracellular domain of RTK, 2) by binding to IRS-1 which binds to RTK, or 3) by binding to activated Ras. The PI3 K activation triggers the activation of Akt by phosphorylating PIP3. Akt activates mTOR, which eventually leads to a series of reactions [31[31].

Fig. 1

Key components of Ras/MAPK/PI3K/Akt/mTOR signaling pathways, involving complex interactions and co-regulation in the cell.

View Large Image | View Hi-Res Image | Download PowerPoint Slide

There are common ASD and cancer associated genes involved in the pathways (Table 1Table 1). For example, HRAS encodes protein H-Ras which belongs to Ras family; BRAF encodes protein B-Raf which is activated by Ras and which then activate MAPK; MAPK1 encode MAP kinases 1 (also known as ERK2) which belong to MAP Kinase family; PTEN encodes phosphatase and tensin homolog (PTEN) which inhibits Akt activity by dephosphorylating PIP3; MTOR encodes mechanistic target of rapamycin (mTOR) which belongs to a family of phosphatidylinositol kinase-related kinases regulates phosphorylation. TSC1/2 encodes the tuberous sclerosis heterodimer TSC1/2, which negatively regulates mTORC1 through Rheb (Ras homologue enriched in brain), and which is inhibited by Akt.

Taken together, autism and cancer not only share genes in common, but also overlap in varieties of functional pathways, many of which are also involved in regulating metabolic activities— which are systemic. Further studies that focus on the involvement in metabolic regulation of these common genes and pathways of ASD and cancer may help us better understand the shared mechanisms of these two conditions.

Metabolic problems are indeed associated with both ASD and cancer. Obesity has been linked with cancer [[32], [33], [34]] and diabetes has been linked with cancer as well [[35], [36]]. Regarding ASDs, studies have shown that many people with autism are at risk for developing or having metabolic problems such as obesity as comorbidities. Although the comorbid obesity in individuals with ASDs could be caused by medications that are prescribed to control difficult behavioral features of autism, the overweight issue has been identified repeatedly in people on the spectrum [[37], [38]] and is likely not due exclusively to pharmaceutical exposure. Further, maternal metabolic conditions such as obesity, diabetes and metabolic syndrome have been recognized as increasing risk of autism in offspring [[39], [40], [41]].

Autism brain studies have also identified a variety of alterations in metabolic profiles. Abnormalities have been discerned using proton emission tomography (PET), single proton emission computed tomography (SPECT) and magnetic resonance spectroscopy (MRS). While the findings differ across these studies, as across many other kinds of studies of individuals on the autism spectrum (where differences in study groups regarding age and a variety of other factors undoubtedly contribute to inconsistencies), overall, metabolic differences from neurotypical individuals include reductions in N-acetylaspartate (NAA), glutamate and glutamine (Glx), γ-aminobutyric acid (GABA), creatine and choline; increases in glutamate; elevated glucose utilization, and abnormalities in phospholipids [[42], [43], [44], [45], [46], [47], [48]]. However, whether these brain abnormalities correlate with somatic metabolic problems in ASD is still unclear and poorly studied to date. Future studies focusing on this topic may help characterize the relationships between systemic and brain metabolic abnormalities in ASD pathophysiology as well as the contributions this level of functional features may make to ASD’s behavioral phenotypic features.

As early as in the late 1920s, Otto Warburg discovered that cancer cells had high rate of anaerobic glycolysis (now known as Warburg effect) [49[49]. It is now well known that tumor cells have altered glucose metabolism [50[50]. The alteration of cancer cell metabolism contributes to supporting their rapid proliferation and expansion. The metabolic changes are considered to be a result of mutated oncogenes and loss function of tumor suppressor genes [51[51]. In parallel, it has been observed that obesity is associated with increased risk of cancers [52[52]. Hence it could also be that alterations in metabolism trigger the gene expression dysfunction and lead to cancer. We cannot state from existing evidence whether metabolic alteration is the cause or the result of cancer; it is possible that the influence goes both ways, in a self-amplifying positive feedback loop. Similar considerations are pertinent with ASDs. The links between ASD and metabolism are also multifaceted (genes, pathways, and metabolic changes) and we do not presently have enough systems-level information to infer predominant directions of causality.

Mitochondria are central to glucose metabolism. Mitochondrial dysfunction has been associated with ASDs [[53], [54], [55]], as well as with cancer [[56], [57]]. A recent study utilizing whole-exome sequencing data found genetic evidence for mitochondrial DNA mutations in ASDs [58[58]. MAPK signaling, discussed above as being linked to both ASD and cancer, has also been found to be associated with and even to lead to mitochondrial dysfunction [59[59]. Mitochondrial dysfunction has also been associated with defects in PI3 K-Akt signaling [60[60], mTOR signaling [61[61] and Ras signaling pathways [62[62], also discussed above as associated with both autism and cancer. This association with mitochondrial dysfunction of signaling pathways generally considered in relation to their functions in other contexts (e.g. cancer, brain or metabolism) further supports the importance of looking for pleiotropic effects in our studies of genes, gene pathways and pathway networks.

Taken together, the signal transduction pathways MAPK, PI3 K-Akt, mTOR, and Ras signaling pathways are all involved in mitochondrial function and metabolic modulation, and are all linked to both ASD and cancer. We hypothesis that: 1) MAPK, PI3 K-Akt, mTOR, and Ras signaling pathways are important underlying contributors to both ASDs and cancer pathophysiology; 2) their function in mitochondria related metabolic processes may contribute importantly to the underlying mechanism(s) shared by ASD and cancer; 3) systems network analysis among these pathways together with overlapping genes of ASD and cancer with regard to mitochondrial function and neurodevelopment may shed light that helps us uncover the common mechanism(s) shared by ASDs and cancer (Fig. 2Fig. 2).

Fig. 2

Proposed systems network analysis for common mechanism of ASDs and cancer.

View Large Image | View Hi-Res Image | Download PowerPoint Slide

All these pieces – common genes, shared pathways, abnormal metabolism, impaired mitochondrial function – together seem to paint a picture that autism and cancer may be superficially different deviations in development and function that at a deeper level derive from overlapping sources – an overlapping group of genes and pathways. This suggests a group of genes and pathways that have critical roles and functions, which vary depending upon location in the organism and its systems as well as upon the part of the life course; depending on myriad possible combinations of these elements, a variety of different outcomes may emerge. This makes the genes and pathways like a group of actors playing different roles on different stages – we see different shows, but in fact some of the major players are the same. If this hypothesis is true, it would follow that overcoming the aberrations in signaling (considering the heavy involvement of these signaling pathways) could have major impacts on autism and cancer, reducing severity or perhaps even leading to remission – and that common approaches might have impacts on diverse conditions.

In summary, we have discussed the overlaps between ASD and cancer at various biological levels including genes, proteins, pathways, and networks, as well as by drawing upon studies from biological, clinical, and epidemiological literatures. Through our discussion of the existing evidence, we noted that signaling transduction pathways may play important roles in both ASD and cancer, and that of those already identified, involvement in mitochondrial function of metabolic processes is common. Signal transduction pathways are involved with communication processes both those internal to the organism and those involved in transmitting signals into the organism from the environment. Physiological and environmental factors could impact this information processing, with types and degrees of vulnerability that vary with age, organ system and other contextual features, and that could thus lead to a range of outcomes. A systems understanding of the similarities and differences between ASDs and cancer regarding “when” and “where” the genes and pathways are influenced by such factors, and “how” they respond, could be helpful in our attempts to understand how ASDs and cancer share so many features in common and yet on the other hand are so different in phenotype. Systems biological studies that can connect knowledge from many sources and disciplines should help push the research further, and thereby help us find ways to facilitate mobilization of repair and regeneration processes, thus increasing the likelihood of generating improved outcomes.