Twin studies are commonly used to estimate genetic and environmental contributions to health outcomes. In a twin study, disease incidence is measured in monozygotic (MZ) and dizygotic (DZ) twin pairs. MZ twins (identical twins) have identical genes (100% genetic match); DZ twins (fraternal twins) have 25% or 50% genetic match. The precisely known percentages of shared genes enables a calculation of the genetic and environmental contributions to disease risk, if certain assumptions are made. In this article I show that a critical assumption made in all the autism twin studies is wrong, resulting in an overestimation of the role of genes in autism.
Twin studies are the foundation for claims that autism is highly heritable. For example, in an April 2017 essay for Autism Awareness Month, Dr Joshua Gordon (Director of the National Institute of Mental Health, NIMH) wrote:
“For some time, we have known that ASD is a heritable condition—that is, it runs in families. We know this from a variety of studies—including twin studies, which demonstrate that if one identical twin has ASD, the other twin almost always does also. Indeed, studies suggest that up to 90 percent of the variation in developing ASD is due to genetic factors.1 “
–Dr Joshua Gordon, Director, NIMH
The citation (1) is a meta-analysis of autism-twin studies (Tick 2016), which concludes that genetic factors explain up to 91% of autism. Tick 2016 states:
“The…heritability estimates ranged between 64% and 91% and are in line with previous reports (Bailey et al., 1995; Colvert et al., 2015; Folstein & Rutter, 1977; Le Couteur et al., 1996; Lichtenstein et al., 2010; Nor- denbæk et al., 2014; Ritvo, Freeman, Mason-Broth- ers, Mo, & Ritvo, 1985; Rosenberg et al., 2009; Steffenburg et al., 1989; Taniai, Nishiyama, Miyachi, Imaeda, & Sumi, 2008).”
Paper (Tick 2016): Heritability of autism spectrum disorders: a meta-analysis of twin studies
Thats a lot of cited studies supporting high heritability! The consistent results suggest that high heritability is well-established and unquestionable. But thats not the case. All the twin studies rely on the same wrong assumption: that gene X environment (GXE) interactions do not occur in autism. As explained below, gene X environment interactions erroneously inflate the calculated heritability. Recent research shows that gene X environment interactions do occur in autism (and other mental illnesses). This necessarily means that the heritability estimates are too high.
All of the studies included in the Tick meta-analysis use the well-known “classical twin method” (CTM) for calculating the genetic and environmental risk contributions. The CTM calculation is based on a simple model of how genes and environment contribute to risk. Specifically, the CTM assumes that risk factors combine additively, without “multiplicative” (i.e. synergistic) interactions. The “additive risk assumption” is necessary for solving the mathematical equations in the CTM. If multiplicative/synergistic interactions occur, the calculations cannot be solved (unless much more data is obtained, but the needed data is almost always unobtainable).
For example, say that a genetic factor and environmental factor are, in isolation, associated with 10% and 20% risk of autism, respectively. If the risks are additive, then when both are present the risk of autism is 10% + 20% = 30%. Simple!
If the factors have a GXE interaction, then the risk is higher than 30%. For example, the risk if both factors are present may be 60% or 70%. The risk of the combination of factors exceeds the sum of the risks individually.
Twin method researchers assume risks are additive because it’s the only way to calculate anything. Unless the additive assumption is made, calculations are not possible and the researchers might as well go home and do something else.
The additive risk assumption is not based on evidence that GXE interactions are absent in autism. In fact, accumulating evidence shows that GXE interactions do occur in autism. GXE interactions are common in many, and likely all diseases.
Twin Method Calculations
In the literature on twin studies, the genetic and environmental factors each comprise two parts:
Genetic (collectively referred to as “G”) :
A=Sum of genetic influences that can be combined additively.
D=Interactions among genetic factors.
Environmental (sometimes collectively referred to as “E”):
C= Environmental factors shared by twin pairs. Vaccines are in this category, because twins almost always have exactly the same vaccine exposure. Other examples are socioeconomic status, diet etc.
E= Environmental factors unique for each person of a twin pair. Examples include accidents, differential parental treatment, differential activities etc.
NOTE: For clarity, I will use “E” for environmental factors generally (shared and unique). Don’t get confused by the use of C and E.
A review paper on twin study methods (Rijsdijk 2002) explains how the CTM calculation separates the genetic and environmental factors:
“Twin data enable the different variance components to be estimated, because MZ and DZ twins have different degrees of correlation for the genetic components A and D but the same degrees of correlation for the environmental components C and E. MZ pairs correlate 1 for both A and D, whereas DZ pairs correlate 1/2 and 1/4 for these components, respectively. Both MZ and DZ pairs correlate 1 for C and E is uncorrelated for both types of twins. Since the phenotypic differences between MZ twins can only be due to unique environmental influences, this gives us an estimate for E. Assuming that MZ and DZ twins experience the same degree of similarity in their environments, any excess of similarity between MZ and DZ twins can be interpreted as due to the greater proportion of genes shared by MZ twins, and thus gives us an estimate for A. An estimate for C is given by the difference in MZ correlation and the estimated effect of A.”
Paper (Rijsdijk 2002): Analytic approaches to twin data using structural equation models
Got that? It is confusing. Understanding this passage is not essential for understanding this article. I quote this passage to provide a general sense of how the calculation is performed.
Again, the CTM calculation absolutely requires the additive risk assumption, i.e. that gene and environmental risks combine additively, and GXE interactions are absent.
Rijsdijk 2002 continues:
“A number of assumptions are made in the classical twin study. It is important to be aware of the implications of such assumptions and of the extent to which they are realistic in relation to the trait in question. The assumptions include the following: Gene–environment correlations and interactions are minimal for the trait.”
(Emphasis added. Complete list of assumptions not quoted. The”trait” in this context is autism)
Tick 2016 also notes that GXE interactions must be absent, and that heritability will be overestimated (In Appendix S1) if GXE interactions occur. Specifically, it is the shared environmental factors (C) that erroneously increase the heritability estimate. Keep in mind that vaccines are part of C, since twin pairs almost always receive the same vaccines. Tick 2016 states:
“The assumptions are that: (i) MZ and DZ twin pairs share their environments to the same extent; (ii) Gene-environment correlations (passive/active) and interactions are minimal for the trait in question (if not they get incorporated into the other variance components, e.g. GxE interaction effects will increase the E variance, but GxC interaction will increase the heritability estimate) (iii) Twins are no different from the general population.”
-Tick 2016 (Emphasis Added)
NOTE: “G” refers to all genetic factors (A + D in combination)
NOTE: C=environmental factors shard by twin pairs. Tick uses “E” to indicate unique (non-shared) environmental factors.
As another example, a review paper on twin studies and heritability (Kaprio 2012) says the same thing:
“In basic twin models, gene–environment interactions are assumed not to exist, and if present, they are included as part of the additive genetic variance, inflating heritability estimates.”
“In some contexts, gene–environment interactions, i.e., that environments modify the effects of genes on the trait being studied, may account for a substantial part of the apparent heritability.”
Paper (Kaprio 2012): Twins and the mystery of missing heritability: the contribution of gene-environment interactions
As another example, a large twin study of autism (Colvert 2015) reports high heritability (56-95%) and states that GXE interactions may be responsible:
“…genetic modeling assumes that no gene-environment interactions or correlations exist; if they exist, the estimates of environmental and genetic effects may be inflated.”
(Colvert 2015): Heritability of Autism Spectrum Disorder in a UK Population-Based Twin Sample
Clearly, Dr Joshua Gordon’s blog statement of “up to 90% heritability” is correct only if GXE gene-environment interactions do not exist in autism.
Vaccination is a shared environmental factor (part of “C”). This is because twin pairs almost always have identical vaccine exposure. If autism is vaccine injury caused by a genetic vulnerability to vaccination (i.e. a “gene X vaccine” interaction), then the reported heritability figures are too high. Accordingly, the high heritability estimates are consistent with vaccines causing autism via a gene X vaccine interaction.
The autism heritability estimates in the twin studies are wrong and unsalvageable if gene-environment interactions occur. Tick 2016 and other twin studies of autism heritability provide no justification for the additive risk assumption. In fact, the assumption is contradicted by evidence for GXE interactions in autism and a general understanding of how genes interact with environmental factors. It is understood today that GXE interactions are common in many diseases and disorders.
Gene-Environment (GXE) Interactions In Autism
It is difficult to detect a GXE interaction in humans. Large numbers of subjects and genetic testing are required. A recent review of gene-environment interactions in autism (Kim 2015) states:
“Studies involving GXE are power-intensive. Even testing for a single GXE specified a priori, the exponential growth in the number of comparisons requires large samples (131). For example, in an unmatched case-control study with a log- additive inheritance model, sample sizes required to examine a GXE effect size of OR 1.5 and 2.0 with 80% power and two- tail p value .05 (without multiple comparison corrections) are 31,084 and 9550, respectively…”
Note: “E” here refers to environmental factors generally, not the unique environmental factors as defined above.
Paper: (Kim 2015): Genetic Epidemiology and Insights into Interactive Genetic and Environmental Effects in Autism Spectrum Disorders
In view of the difficulty of finding GXE interactions, absence of evidence should not be construed as evidence of absence. Nevertheless, there is substantial and accumulating evidence for GXE interactions in autism.
Kim 2017 reports a synergistic GXE interaction between air pollution and copy number variation (a type of genetic defect associated with autism), in autism risk:
“We found a significant interaction in which a 1SD increase in duplication burden combined with a 1SD increase in ozone exposure was associated with an elevated autism risk (OR 3.4, P < 0.005) much greater than the increased risks associated with either genomic duplication (OR 1.85, 95% CI 1.25–2.73) or ozone (OR 1.20, 95% CI 0.93–1.54) alone.” SD=standard deviation, OR= odds ratio
Paper (Kim 2017): The Joint Effect of Air Pollution Exposure and Copy Number Variation on Risk for Autism
Notice that the risk for the combination (OR 3.4) exceeds the risks added together (OR 1.85 + OR 1.2). The risk of the combination is greater than the risks added together. This is the definition of a synergistic interaction.
Autism research is increasingly pursuing GXE interactions. Publications in recent years report evidence for interactions in human and animal studies. Kim 2015 continues:
“Both human and animal studies suggest that GXE plays a role in ASD pathogenesis.”
“…one group reported three GXE from the same study population. In the study of 429 children with ASD and 278 typical children, maternal MTHFR 677TT (rs1801133), CBS rs234715 GT1TT, and child COMT 472AA (rs4680) genotypes conferred greater ASD risk when the mother did not take vitamins periconceptionally…”
“[another study] reported interactions between high pollution and nitrogen dioxide and the MET CC genotype (rs1858830) in 251 cases and 156 control subjects.”
Another review paper on GXE interactions in autism (Chaste 2012) states:
“[Additional] evidence for the contribution of G×E to autism risk comes from animal models. In a first study, 106 mice haploinsufficient for the TSC2 gene demonstrated a lack of normal social approach behavior only when exposed to maternal immune activation.”
“In another animal model, prenatal maternal immune activation and expression of a mutant DISC1 protein interacted to produce an altered pattern of sociability. This neurobehavioral profile was absent in untreated mice expressing the mutant.”
Paper (Chaste 2012): Autism risk factors: genes, environment, and gene-environment interactions
Here are more papers supporting a role for GXE interactions in autism and other neuro/psychiatric disorders:
“GeneXenvironment interactions play critical roles in the emergence of autism and schizophrenia pathophysiology.” (Michel 2012)
Paper (Michel 2012): Immune system gene dysregulation in autism & schizophrenia
“These studies have demonstrated that genetic predispositions for a number of psychiatric disorders interact with environmental influences to manifest disorder.” (Dick 2011)
Paper (Dick 2011): Gene-Environment Interaction in Psychological traits and Disorders
“Growing evidence supports GxE interactions in these disorders.” (referring to schizophrenia and bipolar) (Geoffroy 2013)
Paper (Geoffroy 2013): Gene x environment interactions in schizophrenia and bipolar disorder: evidence from neuroimaging
“Multiple lines of evidence suggest that the roles of genetic and environmental factors depend on each other. Gene–environment interactions may underlie the paradox of strong environmental factors for highly heritable disorders, the low estimates of shared environmental influences in twin studies of serious mental illness, and the heritability gap between twin and molecular heritability estimates.” (Uher 2014)
Paper (Uher 2014): Gene–environment interactions in severe mental illness
These studies have limitations and they are not replicated. But the existence of these results, and the growing support in the scientific literature for GXE interactions in autism and other psychiatric disorders, contradict the additive risk assumption. The reports of GXE interactions in human studies and animal models renders the assumption untenable, and the heritability estimates unreliable. The heritability estimates for autism are too high.
GXE Interactions and Immune Function
Autism is associated with variants in immune system-related genes such as human leukocyte antigen (HLA) genes. The HLA genes strongly influence individual immune system characteristics, by impacting the “major histocompatibility complex” (Needleman 2012). Paper (Needleman 2012): The major histocompatibility complex and autism spectrum disorder
Also, genes involved in regulating microglial function are associated with autism (Gupta 2014). Microglia are immune system cells in the brain. Microglia play a role in autism, likely by secreting cytokines such as IL-6. Paper (Gupta 2014): Transcriptome analysis reveals dysregulation of innate immune response genes and neuronal activity-dependent genes in autism
These facts support the plausibility of a gene X vaccine interaction in autism. Vaccination is an environmental factor (part of E, or more specifically C). Vaccines obviously interact with the immune system. A genetically-vulnerable immune system may produce abnormally strong or persistent inflammation in response to vaccination or vaccine aluminum adjuvant.
Kim 2015 argues that environmental factors impact genes related to immune and inflammatory responses in autism:
“Compared with individuals with-out ASD, brains of individuals with ASD demonstrated down- regulation of 209 genes, enriched for gene categories related to synaptic function, whereas 235 genes implicated in immune and inflammatory response were upregulated. The former group was significantly enriched for association signals in GWAS, whereas the latter group was not. These studies suggest that relevant immunologic changes are likely caused by environmental factors…”
-Kim 2015 (emphasis added)
GWAS=genome-wide association studies (a research technique in which people are scanned for genetic variations, which are then tested for statistical associations with health outcomes).
In view of this evidence, immune system genes are plausible candidates for gene X vaccine interactions.
The claim that autism is highly heritable is based on twin studies using the classical twin method (CTM). The CTM calculations require the absence of GXE interactions. But GXE interactions are present in autism. The heritability estimates are therefore erroneously inflated.
Gene X vaccine interactions can explain the high heritability estimates from the twin studies. Vaccination is a shared environmental factor that may have synergistic interactions with genetic factors.
Due to the erroneous additive risk assumption, the high heritability estimates cannot rule out gene X vaccine interaction as a significant or the primary cause of autism.
Gene X vaccine interactions are biologically plausible because autism is associated with immune system gene variations.