Journal of Autism and Developmental Disorders (2021) 51:4321–4332

https://doi.org/10.1007/s10803-020-04685-z

S:I AUTISM IN REVIEW: 1980-2020: 40 YEARS AFTER DSM-III

Genetic Advances in Autism

Published online: 17 September 2020

Abstract
In the last 40 years, there has been a huge increase in autism genetics research and a rapidly growing number of discover-
ies. We now know autism is one of the most highly heritable disorders with negligible shared environmental contributions.
Recent discoveries also show that rare variants of large effect size as well as small effect common gene variants all contribute
to autism risk. These discoveries challenge traditional diagnostic boundaries and highlight huge heterogeneity in autism. In
this review, we consider some of the key findings that are shaping current understanding of autism and what these discover-
ies mean for clinicians.

Over the last 40 years, our understanding of autism has 1%. This is thought to reflect changes in ascertainment and
evolved enormously. We have moved from a time when the the broadening of diagnostic criteria (Rutter 2007; Rutter
role of genetics was unknown to an era when the first twin 2011, 2013a); these issues are important to consider when
and family studies showed autism to be one of the most we come to interpreting genetic study findings. Both DSM-5
highly heritable disorders (Rutter 2011). These family-based (APA 2013) and ICD-11 (WHO 2019) now use the umbrella
studies motivated molecular genetic investigations, that most term “autism spectrum disorder”. Another consideration is
recently have led to an increasing number of reported autism how we deal with monogenic disorders. Previously Rett syn-
gene discoveries and that are accompanied by a growing lit- drome (RTT) was considered as a form of autism that affects
erature on potential biological insights. For those interested females. However, there are some key clinical differences
in details of autism risk loci, implicated genes and hypoth- from typical autism, in that it is a progressive neurological
esised biological mechanisms, the reader is directed to exist- disorder with very characteristic features including loss of
ing, comprehensive reviews on these topics (Vorstman et al. purposeful hand use and repetitive movements. Rett syn-
2017; Sestan and State 2018a; Woodbury-Smith and Scherer drome is now known to be caused by variants in the methyl-
2018; Quesnel-Vallières et al. 2019; Vicari et al. 2019). Our CpG binding protein 2 (MECP2) gene. Given its distinctive
aim in this review is to consider how recent findings are clinical presentation and single known cause, RTT appro-
shaping our understanding of autism and how discoveries priately is no longer grouped with autism in DSM-5 and
might inform clinicians. ICD-11. There are an additional group of monogenic disor-
The concept of autism has gradually broadened since ders, such as Tuberous Sclerosis and Fragile X syndrome,
the time of Leo Kanner’s first clinical descriptions in his that have very distinctive physical features (e.g. tubers) and
1943 seminal paper (Harris 2018). The prevalence of autism which can be accompanied by autism. Readers interested
remained low for very many years but has risen over the last in the clinical features of these disorders and research on
few decades from around 2–4 in 10,000 to an estimate of monogenic disorders that has moved from gene identifica-
tion to reversal of deficits in animal models are directed else-
where (Sztainberg and Zoghbi 2016).
* Anita Thapar Some consider these disorders as syndromal autism or
[email protected] high penetrance forms of autism. As we will discuss later
1
however, new genetic and biological findings have high-
Division of Psychological Medicine and Clinical
Neurosciences and MRC Centre for Neuropsychiatric
lighted that there is no clear-cut distinction between rare
Genetics and Genomics, Cardiff University, Hadyn Ellis monogenic and common multifactorial autism.
Building, Cardiff, Maindy Road, Wales CF24 4HQ, UK
2
Social, Genetic and Developmental Psychiatry Centre, Kings
College London, London, UK

13
Vol.:(0123456789)
4322 Journal of Autism and Developmental Disorders (2021) 51:4321–4332

The Heritability of Autism: From Early authors acknowledge, that the design could only examine
to Modern Twin and Family Studies genetic factors shared by sisters and other types of design
are required to robustly assess the contribution of early life
Although Kanner is reported to have viewed autism as exposures. Taken together, all these twin studies provide
an innate disorder (Rutter 2013; Harris 2018), a strong strong evidence of a mainly genetic contribution to autism
psycho-analytic tradition led to the growing belief that and negligible shared environmental effects.
“refrigerator” mothers might be to blame. The first twin
study of autism conducted by Folstein and Rutter (Folstein
and Rutter 1977) was ground-breaking because it clearly What Family and Twin Studies Have Told us
showed a predominantly genetic contribution to autism. About the Autism Phenotype
The most recent meta-analysis of all published twin stud-
ies of autism/autism spectrum disorder conducted by Tick Twin and family studies of autism were important in show-
and colleagues (Tick et al. 2016) also yielded a large herit- ing early on that the biological relatives of probands with
ability estimate of 64–91% and no significant shared envi- autism were not just at heightened risk for autism itself but
ronmental contribution. These authors demonstrated that also showed elevated rates of milder autistic-like features.
if the estimated prevalence rate of autism is incorrectly This led to the appreciation of there being a broader autism
specified for the study population (1% instead of 5% which phenotype characterised by features like those of autism but
is the appropriate figure for a broader autism phenotype), less severe than in affected individuals (Le Couteur et al.
this essentially results in an increased non identical (dizy- 1996).
gotic DZ) twin correlation but does not affect identical Family studies have suggested that familial liability to
(monozygotic MZ) twin correlations, thereby resulting in autism explains the higher rate of pragmatic language dif-
a reduced heritability estimate and a stronger shared envi- ficulties (Miller et al. 2015), social abnormalities and unu-
ronmental contribution. Thus the shared environmental sual personality features such as shyness and aloofness in
contribution observed in two outlying studies (Hallmayer relatives of probands with autism (Le Couteur et al. 1996).
et al. 2011; Frazier et al. 2014) appeared to be explained However, the broader phenotype is different to autism in
by the assumption of prevalence and an overinclusion of several key aspects. First, it is not associated with epilepsy,
concordant DZ twins. The study by Tick and colleagues second there is no association with lower IQ or specific
is also important in showing that if the autism broad phe- learning problems. Although now there are ways of assess-
notype is clinically recognised, then that ought to be taken ing the broader autism phenotype (de Jonge et al. 2015), a
into account by assessing different thresholds when fitting key challenge lies in knowing where its boundaries lie given
statistical models. that autism genetic liability appears to operate across a con-
A more recent study combined extensive family and tinuum and confers risk for a range of other neurodevelop-
twin population-based data across five different countries: mental and psychiatric disorders which we will discuss later.
Denmark, Finland, Sweden, Israel and Western Australia Another striking finding from twin studies was the obser-
(Bai et al. 2019). The authors again observed a high vation of very high variability in the clinical features of
median heritability of 80.8% for autism with only modest autism (e.g. IQ, clinical symptoms) among MZ twins who
country-specific variation in estimates varying from 50.9% share all their inherited DNA (Le Couteur et al. 1996). This
in Finland to 86.8% in Israel. Shared environment contri- suggests that the clinical manifestation of autism even given
butions were negligible. The authors conducted sensitivity the same level of genetic liability maybe subject to stochastic
analyses on Finland and Western Australia because these factors or environmental factors that are not shared by MZ
yielded lower heritability estimates when compared to the twins. It has been argued that we should not be surprised
other countries. They further showed that a random under- by chance or stochastic events as contributors to health and
ascertainment of autism may result in an underestimate of disease, given they are likely to have important evolutionary
true heritability and increase the observed shared environ- advantage (Davey Smith 2011).
ment contribution. This study by Bai et al. also examined The third clinically relevant finding that emerged from
maternal contributions to autism that was enabled by the autism family studies deals with clinical indices of genetic
inclusion of offspring from sisters. Surprisingly perhaps, heterogeneity. In general, more severe autism (indexed by
given the hypothesised role of prenatal exposures and autistic symptom severity or lower verbal IQ- a measure of
risk for autism, there was negligible maternal contribu- overall language/communication skills) has been observed
tion to autism risk. This observation replicated a previous to be associated with greater familial loading (Rutter
Swedish study that had also observed limited maternal 2000). However, there was interest in whether famil-
contributions to autism (Yip et al. 2018). However, the ial loading was different for probands who also showed
profound intellectual disability (global intellectual and

13
Journal of Autism and Developmental Disorders (2021) 51:4321–4332 4323

adaptive functioning difficulties); that is, whether profound Autism Overlap with Other Childhood
intellectual disability indexed a discontinuity in terms of Neurodevelopmental Disorders
genetic liability for autism. That is important in refining
estimated recurrence risks of autism in affected families. It is now well-recognised that autism shows a high level
Family study findings here have been mixed with the larg- of comorbidity and population -based twin studies have
est study suggesting that the discontinuity in terms of consistently observed that autism traits show strong
familial loading appears to apply to autism accompanied genetic correlation with other neurodevelopmental traits
by very severe language deficits (Pickles et al. 2000). and diagnoses (Thapar and Rutter 2015a). A twin study
A final issue that has been studied is whether autism in Sweden for example, showed that autism was not only
should be viewed as a discrete diagnostic entity. Although highly heritable but that three quarters of its genetic vari-
for clinical purposes autism is defined categorically, it can ance was shared with ADHD and that genetic factors also
also be viewed as a continuously distributed dimension. contributed to the overlap between autism and learning,
Twin studies have been used to investigate the validity of motor co-ordination problems and tic disorders (Lichten-
a dimensional approach by examining whether “disorder” stein et al. 2010).
lies at the extreme of a dimension. Several twin studies A subsequent analysis of Swedish Registry family
have utilised population data on autism to assess this. Most data further highlighted important links between autism
have suggested that heritability estimates are consistent and ADHD (Ghirardi et al. 2017). This study included
across the typical population range and extreme autism 899 654 individuals in Sweden with diagnoses recorded
scores or show a strong genetic correlation between autism nationally by clinical services. The authors observed that
trait and diagnosis (Lundström et al. 2012; Colvert et al. those with autism were at higher risk of having ADHD com-
2015). However, one study (Frazier et al. 2014) yielded pared with individuals who did not have autism (odds ratio
different findings where autism showed higher heritability (OR) = 22.33, 95% confidence interval (CI) 21.77–22.92).
at the high end of the continuum when compared to low Almost half the individuals with autism also received a diag-
scorers. However, as already discussed, the twin sample nosis of ADHD. They further established that the monozy-
here was highly selected rather than population-based. The gotic co-twins of those with autism showed an increased
most recent and largest population-based study examined risk of ADHD (OR = 17.77 95% CI 9.8–32.22) compared
scores on the Childhood Autism Spectrum Test (CAST) to dizygotic co-twins (OR = 4.33 95% CI 3.21–5.86). These
at age 8 years in 2,256 MZ twin pairs and 4157 DZ pairs associations were most prominent for those with higher
(Tick et al. 2016). Here, heritability estimates for high functioning autism rather than low functioning autism (with
autism scores were not substantially different to those with intellectual disability). The findings highlight that while
low scores. Another recent large twin study in Sweden relatives of those with autism have long been known to be
also showed a modest genetic correlation (0.48 (95% CI at elevated risk for autism and the broader autism pheno-
0.44–0.53),) between autism and a trait measure of autism type, they also are at high risk for ADHD and other neu-
(Taylor et al. 2019a, b). Thus, so far most of the twin rodevelopmental disorders. That is, autism genetic liability
research suggests that autism can be viewed as lying on can manifest not just as autism but also as ADHD and other
a continuously distributed dimension in the population as neurodevelopmental disorders. These overlaps will be fur-
well as a category for clinical purposes and that these are ther considered in the light of molecular genetic studies.
similarly heritable and share similar although not identical Observations from family and twin studies however do lend
genetic contributions. weight to the stance taken by both DSM-5 and ICD-11 in
These findings together with family study observations grouping child neurodevelopmental disorders and in now
of the broader autism phenotype highlight that there is no enabling ADHD to be co-diagnosed with autism.
clear-cut boundary that demarcates a diagnosis of autism or
autism (Rutter and Pickles 2016). Although rigorous diag-
nostic instruments such as the ADI and ADOS are invaluable
for research, extremely lengthy protracted assessments in Gene‑Environment Interplay
practice that search for an “accurate” diagnosis thus are not
justifiable when intervention is a priority. Although autism is highly heritable, it is not entirely
That is not to say, we do not value careful assessment, explained by genetics, environmental factors also con-
but rather that there needs to be an appreciation that dimen- tribute. The role of environment in autism risk has been
sional and categorical approaches both are valid and that the reviewed extensively elsewhere (Mandy and Lai 2016).
diagnosis of autism cannot be defined accurately as a disease Here we will consider how environmental risks might
with discrete boundaries no matter how many assessments work together with genetic liability.
are conducted (Rutter 2011).

13
4324 Journal of Autism and Developmental Disorders (2021) 51:4321–4332

It now is known that many environmental risks are 2017) -they do have important risk effects on depression and
correlated with genetic liabilities and thus, some of the could provide one explanation for phenotype and genetic links
prenatal and early life factors that have been observed to observed between autism and depression (Thapar and Rutter
be associated with autism potentially could arise through 2019). However, that requires explicit investigation.
gene-environment correlation (Rutter 2015). For example, Gene-environment interaction is a different concept that
passive gene-environment correlation would arise when refers to the phenomenon where the effect of environmen-
a mother’s genetic background influences environmental tal exposures on phenotype is modified by the background
exposures associated with autism risk, such as medical genotype or genetic liability. Although shown to contribute in
conditions or behaviours in pregnancy (e.g. dietary intake animal studies (Thapar and Rutter 2015b, 2019), to date con-
of folic acid). Maternal genetic liability for ADHD has vincing findings that gene-environment interaction contributes
been shown to be associated with many prenatal expo- to autism risk have been lacking.
sures; for example, smoking in pregnancy (Thapar et al.
2009; Thapar and Rice 2020). However, thus far, simi- Molecular Genetic Approaches
lar findings have not been observed for maternal autism to Understanding Autism
genetic liability (Leppert et al. 2019). One issue that
requires discussion relates to the observation that older The last decade has witnessed an enormous surge in published
maternal and paternal age or delayed paternity are associ- molecular genetic findings on autism. Genome-wide stud-
ated with risk for autism and older paternal age also has ies across medicine, psychiatry and the social sciences have
been linked to a higher risk of spontaneous or de novo rare involved the interrogation of genomic variation to search for
variants. Such variants have been observed to contribute to links between specific variants and disorder or traits. Genomic
autism risk (see later). However, recent genetic epidemio- variation can be characterised by its population frequency as
logical findings suggest that age-related de novo variants well as by whether the variation involves DNA structure or
do not appear to be a primary explanatory mechanism for sequence (State and Thapar 2015). Genome-wide associa-
the paternal age findings (Gratten et al. 2016); and in one tion studies (GWAS) involve comparing the frequencies of
study it was estimated that shared genetic liability between hundreds of thousands of common gene variants, known as
father and offspring could contribute to the association single nucleotide polymorphisms (SNPS; frequency > 5%) in
(Gratten et al. 2016). This is an important issue for older cases and controls (Sullivan et al. 2018). Given the very large
fathers who are concerned about risk of autism in off- number of statistical tests required for so many variants and
spring. Interestingly, although autism shows such strong because common variants each have small effect size (e.g.
comorbidity and shared genetic liability with ADHD, it is odds ratio 1.1–1.2), extremely large sample sizes have been
younger rather than older parental age that is associated required to detect genome-wide significant variants. Other
with ADHD. genome-wide studies have examined the contribution of rare
Active and evocative gene-environment correlation arise structural and sequence variants that have larger effect size
when the offspring’s genetic liability is associated with an envi- using family-based designs as well as case–control cohorts.
ronmental exposure; for example, where an individual seeks Rare DNA variants are sometimes referred to as mutations in
out specific environments or evokes environmental exposures the literature although there are recommendations the term
depending on their genetic propensity. Children with autism variant should be used (Richards et al. 2015).
for example, are at higher risk for maltreatment and bullying Autism genetic liability can be viewed as a risk contin-
victimisation (Hoover and Kaufman 2018; McDonnell et al. uum in the population where those with clinical disorder lie
2019). Genetic studies suggest that these exposures are cor- at one extreme of this liability curve. Common gene variants
related with background family and genetic liability (Dinkler appear to contribute to most of the population risk; environ-
et al. 2017; Ohlsson Gotby et al. 2018). These adversities could mental and stochastic influences will also contribute and as
arise from both passive gene-environment correlation (e.g. we will discuss, rare variants act against a background of
via parental neurodevelopmental impairments) or evocative these other influences to shift individual liability along the
gene-environment correlation (child genetic background). The risk continuum towards disorder.
findings highlight that genetic and environmental influences
are not independent of each other. For clinicians, the phe-
nomenon of gene-environment correlation means that where Common Gene Variant Contribution
social adversity accompanies autism, it does not necessarily to Autism
mean that the social adversity was causal or that the autism is
a different type of adversity-related autism. While early social Although autism is highly heritable, and despite common
adversities, unless unusually extreme (Rutter et al. 2007) have gene variants having been considered to contribute sub-
not been demonstrated to be causal for autism (Dinkler et al. stantially to population risk, individual variants have only

13
Journal of Autism and Developmental Disorders (2021) 51:4321–4332 4325

recently been identified. The most recent genome-wide show a negative genetic correlation with IQ and educa-
meta-analysis of 18,381 people with autism and 27, 969 tional achievement. Prior to the advent of GWAS findings,
controls led to the identification of five genome-wide sig- it was well recognised that around a third of those with
nificant loci (Grove et al. 2019). The problem with GWAS autism have been reported to manifest outstanding cogni-
is that genome-wide significant findings represent just the tive skills, so called “savants” (Howlin et al. 2009). Other
start because it does not identify causal genes or mecha- striking clinical findings include the observation that,
nisms; much further work is needed to uncover what genes a proportion of those with autism show early language
are likely causal and how gene variation leads to disorder. regression and that associated epilepsy typically onsets in
Also, SNPs only capture a very small proportion of the total adolescence (Rutter and Pickles 2016). How these clinical
genetic variance as a result of which SNP heritability for observations link to the recent genetic findings on autism
autism is low (0.118) and common genetic liability has no and IQ remains unknown.
predictive utility at present. However, GWAS findings do What has emerged consistently from GWAS of autism
highlight that common as well as rare variants contribute and psychiatric disorders is evidence that genetic influ-
to the genetic architecture of autism. Also, there is grow- ences transcend diagnostic boundaries, in keeping with
ing interest in using GWAS to generate composite measures findings from twin and family studies. The most recent
of common gene risk variants nominally associated with meta-analysis of eight psychiatric/neurodevelopmental
a given disorder, known as polygenic risk scores. In other disorders that included anorexia nervosa (AN), ADHD,
areas of medicine, polygenic risk scores when combined autism, major depression, obsessive compulsive disorder
with clinical variables are being considered as potentially (OCD), schizophrenia (SCZ) and Tourette syndrome (TS)
useful predictors of disease onset, for example in high-risk observed substantial pleiotropy with over 100 loci associ-
groups, and for estimating prognosis (Lewis and Vassos ated with more than one disorder and prominently involved
2020). Thus, it is plausible that with larger GWAS discov- in neurodevelopment and expressed in fetal life (Lee et al.
ery samples sizes and more powerful PRS, these could when 2019). Interestingly autism showed strongest genetic cor-
combined with other measures, have clinical utility in the relations with ADHD (rg = 0.44), depression (rg = 0.45) and
future. to a lesser extent with schizophrenia (rg = 0.22). Yet autism
The authors of the largest autism GWAS further inter- unlike any of these disorders is not amenable to treatment
rogated the Danish registry ICD-10 diagnostic data for the by medication and even those that do show improvements
Danish iPsych cohort that included 13,076 cases and 22,664 do not respond to the same treatments (e.g. stimulants for
controls. They observed SNP heritability was three times ADHD, SSRIs for depression and atypical antipsychotics
higher for autism without intellectual disability than for for schizophrenia). Some of the pleiotropic loci, including
those with autism who also had intellectual disability. These two shared between SCZ and autism showed evidence of
findings are difficult to equate to those from twin study find- opposite direction effects and autism was implicated in 36%
ings because twin heritability includes all inherited genetic of the pleiotropic loci.
variation although some family studies had suggested higher Interestingly the authors utilised the genetic observations
familial loading in less intellectually or language impaired to investigate the structure of different psychiatric disorders
probands. Intriguingly there was also some suggestion of using exploratory factor analysis. This identified three cor-
possible heterogeneity across the different ICD-10 diagnos- related factors: one comprised disorders characterised by
tic subgroups (e.g. Asperger’s, atypical autism) but caution compulsive/perfectionistic behaviour (AN, OCD and more
is required about these findings because they have not been weakly TS), the second factor included mood disorders and
replicated and are based on clinically ascertained subjects. psychosis (depression, bipolar, schizophrenia) and the third
One of the most striking findings about autism com- factor encompassed neurodevelopmental disorders (autism,
mon genetic liability is that it shows a strong positive ADHD and TS) but surprisingly also depression (Lee et al.
genetic correlation with IQ and educational attainment. 2019). This structure is interesting because it does argue
This is puzzling given that autism itself is associated with in support for the DSM-5 grouping of placing autism and
lower IQ. The observation is not explained by the arte- ADHD together under neurodevelopmental disorders. How-
fact of selection bias or population stratification effects ever, the prominent genetic overlaps fuel the argument that
because when parent offspring trios are examined, over- diagnostic classification ought not be reified.
transmission of alleles associated with higher educational Another finding from GWAS is that autism diagnosis
attainment is observed in affected vs. unaffected siblings genetic liability as captured by common variants (using link-
(Weiner et al. 2017). These findings are a puzzle and this age disequilibrium (LD) score regression and polygenic risk
relationship with educational achievement is very differ- scores) shows overlap with population social-communica-
ent to the pattern observed for neuropsychiatric disorders. tion traits (Robinson et al. 2016; St Pourcain et al. 2018) and
For example, ADHD and schizophrenia, as expected, autism traits (Taylor et al. 2019a, b). Thus, molecular genetic

13
4326 Journal of Autism and Developmental Disorders (2021) 51:4321–4332

studies converge with twin study findings in suggesting that within families which is in keeping with family and twin
autism lies at the extreme of a continuum. study observations. The same variant does not necessarily
manifest in two affected siblings with autism. Nevertheless
there are some recognised recurrent autism associated de
Rare Genetic Variants novo CNVs (Sanders et al. 2015). Replicated CNV regions
include 1q21.1, 3q29, 7q11.23, 16p11.2, 15q11.2–13 and
In contrast to GWAS findings, autism investigations of 22q11.2 (Sanders et al. 2015). Copy number variants typi-
rare genetic variation (< 1% frequency) have yielded many cally encompass multiple genes so while de novo CNVs are
more discoveries so far. Generally rare variants tend to show thought to have a high probability of being causal, we can-
larger effect sizes relative to common variation. Initial whole not deduce the mechanisms that lead to autism without fur-
genome rare variant searches focused on a type of varia- ther investigation. Also autism-associated CNVs are highly
tion known as copy number variation (CNVs). These copy pleiotropic, with many of the same CNVs also being associ-
number variants are regions of DNA containing thousands ated with risk for intellectual disability, schizophrenia and
to millions of base pair variants (the building blocks of ADHD (Williams et al. 2010; Marshall et al. 2017; Chawner
DNA) that are duplicated or deleted relative to a reference et al. 2019).
genome. These deletions and duplications can span many
different genes and although large they are too small to be
seen by light microscopy. More recent sequencing studies Sequencing Studies
have focused on rare variants that involve changes to a single
base pair known as single nucleotide variants (SNVs) and Recent genetic investigations of autism have focused on
insertion or deletion of base pairs (indels). Rare variants sequencing all DNA variation within the coding region of
can be transmitted from parent to offspring (inherited) but the genome (exome). Exome sequencing studies of sim-
also can be de novo in origin where the variant first arises in plex families and case–control comparisons have observed
the parent germline (oocytye or spermatozoa) or later, after de novo and inherited rare variants associated with autism
fertilization when they are known as post zygotic somatic risk. The study by Sanders et al. 2015 (Sanders et al. 2015)
variants (State and Thapar 2015; Lim et al. 2017). All these combined analysis of de novo CNVs, and variants identified
variants appear to contribute to autism risk. from exome sequencing, that included indels (small inser-
tions and deletions) and single nucleotide variants (SNVs),
Copy Number Variants and yielded 71 autism risk loci. Findings from the largest
autism exome sequencing study to date involved analysing
Genome-wide searches for rare variants associated with 11,986 autism cases that included 6, 430 proband-parent
autism risk have involved simplex families where only one trios and 5556 cases with 8809 controls. Integrating and
proband is affected, consanguinous as well as multiplex fam- analysing these data has led to the implication of 102 autism
ilies where multiple siblings are affected. It is worth recog- risk genes (Satterstrom et al. 2020). The authors observed
nising that such designs that enhance variant discovery may a significant 3.5 fold increase in de novo protein truncating
mean that cases included are not necessarily typical of every variants (PTVs) and a non-significant 1.2 fold enrichment
clinician’s clinic group. of inherited PTVs.
An initial study conducted by Sebat et al. 2007 (Sebat With the advent of whole genome sequencing (Yuen et al.
et al. 2007) involved 264 families, including 118 “simplex” 2017; Werling et al. 2018), the number of implicated genes
families containing a single child with autism, 47 “multi- is set to rise further to several hundreds at least (Sestan and
plex” families with multiple affected siblings, and 99 control State 2018). Initial findings suggest possible contributions
families with no diagnoses of autism. The authors identified from non-coding variants as well as tandem repeat sequences
an increased burden of rare de novo chromosomal struc- (Trost et al. 2020) (repeated sequences of nucleotides such
tural variants consisting of deletions and duplications (copy as seen in Fragile X syndrome). However as whole genome
number variants; CNVs) in individuals with autism when sequencing involves interrogating many more variants than
compared with healthy controls (1% rate); they observed a whole exome sequencing, even larger sample sizes will be
de novo CNV rate of 10% in simplex cases and 3% in cases required to yield high confidence genetic discoveries (Sear-
from multiplex families. Subsequent studies observed simi- les Quick et al. 2020).
lar findings with an increased rate of rare de novo CNVs in There are several observations that emerge from these
autism especially in simplex families (Marshall et al. 2008; rare variant studies of autism. First, for autism, there has
Sanders et al. 2011). been a much greater discovery rate of rare variants com-
What has emerged clearly from these studies is the pared with common variants. The under-identification
high degree of etiological heterogeneity for autism even of autism common gene variants likely is due to much

13
Journal of Autism and Developmental Disorders (2021) 51:4321–4332 4327

smaller sample size availability than for other disorders associated also with schizophrenia risk (e.g. 22q11 micro-
(e.g. schizophrenia, hypertension) because investigations deletion), intellectual disability and ADHD (e.g. 15q 11.13)
suggest that at a population level, polygenic inheritance (Chawner et al. 2019). Furthermore, initial whole genome
remains an important contributor to population risk for sequencing suggests that in multiplex families, more than
autism. Moreover polygenic variation still appears to con- half of the affected siblings carry different autism related
tribute additively to risk of autism in those who possess variants (Yuen et al. 2015). Nevertheless, as we will dis-
a strong de novo variant (Weiner et al. 2017). However, cuss, most scientists who work in this area are optimistic
in clinics where the cohort includes affected individuals that clinical translation is feasible (Sestan and State 2018;
only rather than the whole population, there is enrichment Quesnel-Vallières et al. 2019; Wiśniowiecka-Kowalnik
for rare variants. It is estimated that around 10–40% of and Nowakowska 2019). However, we view that a lack of
individuals diagnosed with autism could be explained by detailed clinical information beyond simply autism diagnos-
de novo rare variants (Sestan and State 2018). However, tic interviews may be one key barrier in translating genetic
de novo variants cannot explain the familial and genetic findings into clinical practice. Detailed clinical descriptions
aggregation of autism. as well as physical investigations and imaging data will help
The second issue is that unlike common variants, the us better understand and characterise the different variants
detected rare variants have larger effect size (e.g. odds ratio and enable clinicians to interpret their clinical and long-
of > 20 (De Rubeis et al. 2014)) although penetrance for term impacts and such studies are underway (D’Angelo et al.
many mutations appears to be highly variable. Some have 2016).
proposed that individuals with the same variant may show As rare variants, especially de novo ones are often idi-
clinical heterogeneity because of “second” or “multiple” hits osyncratic to families, and multiple different common and
where additional variants modify the clinical picture by add- rare variants contribute risk, we do not know as yet whether
ing to risk or having a protective effect. there are common final common developmental and bio-
This makes it difficult to predict risk for the purpose of logical pathways to autism that could be ultimately targeted
genetic counselling (see later). Large effect size and delete- safely by treatment at the appropriate developmental stage.
rious de novo variants (e.g. result in loss of function of the We will discuss this next.
gene product) are over-represented in those with autism yet
would be subject to natural selection where they tend to be
removed from the gene pool over generations. This would From Genes to Biology and Treatment
explain why in general a higher rate of these variants has
been observed in simplex families. A strong motivation for identifying autism risk genes is
A third observation is that autism associated de novo rare to provide insights into its at present unknown biological
variants although over-represented in those with comorbid underpinnings, pathogenesis and to pave the way for treat-
intellectual disability are present across the spectrum of ment. Rare variants are considered attractive for potentially
intellectual ability. This means that rare variants are still providing clues into potential biology because of their large
relevant for higher IQ individuals with autism but this group effect size and especially de novo variants that appear causal.
may not be a high priority for genetic testing (see genetic However, the problem is that rare variants do not act in isola-
testing later). tion in any given affected individual (e.g. polygenic back-
A fourth point relates to the male preponderance in ground), there are so many genes involved and variants are
autism. Family and twin studies originally suggested that pleiotropic. Also, so far, common variants and rare variants
the siblings of females with autism are at higher risk for have not been definitively shown to converge on the same
autism than the siblings of males (Robinson et al. 2013). biological systems and we do not know whether different
This suggested that females might in some way be protected rare variant carriers show a similar type of autism and under-
against developing autism despite inherited liability. Inves- lying biology to each other and to those who do not carry a
tigation of rare variants are also consistent with the female rare autism variant at all.
protective effect hypothesis as a mechanism for the increased There is growing interest however in examining how dif-
male prevalence of autism because affected females have ferent gene variants converge on the same gene expression
been observed to also carry an increased burden of de novo and protein networks to identify potential key biological
variants. pathways that underlie autism. These approaches have also
Finally, what is very clear from molecular genetic studies involved examining how gene variants impact on different
is that autism is not only clinically heterogenous, it is highly brain cell types, in different places across the brain and at
heterogeneous in terms of genetic etiology at a molecular different developmental periods. A growing number of bio-
level and autism-associated rare variants like common vari- informatic resources enable researchers to infer what identi-
ants are pleiotropic. Autism-associated de novo CNVs are fied gene variants do which is less costly and time intensive

13
4328 Journal of Autism and Developmental Disorders (2021) 51:4321–4332

than examining the function of one gene variant at a time Given the growing number of rare variants implicated
in model organisms and cellular models. Clearly with so in autism risk there are some potential benefits of further
many autism risk genes involved, identifying autism biologi- molecular testing. These include for example, more indi-
cal underpinnings is going to be complex especially as the vidually tailored recurrence risk estimates of autism, access
phenotype manifestations of autism are not easily recapitu- to support groups, a greater understanding of how autism has
lated by animal and cellular models. Nevertheless, experts arisen in the affected proband and enhanced early recogni-
in this area are optimistic that systems biological approaches tion and treatment of medical conditions known to be associ-
that examine the convergence of autism associated genes, ated with the variant (e.g. occult congenital heart disease)
proteins, cells, circuit and behaviour will yield important as well as increased vigilance about comorbid psychiatric
biological insights. So far the autism associated risk genes disorders (e.g. elevated risk of psychosis in those with a
implicate synaptic proteins, and those involved in chromatin 22q11 deletion). In some countries e.g. the United States,
and transcriptional (the conversion of DNA to RNA) regula- guidelines currently recommend that all those with a diag-
tion (Sestan and State 2018), are mainly expressed early in nosis of autism are screened for CNVs using chromosomal
brain development during prenatal life and encode a very microarrays (Schaefer and Mendelsohn 2013). In other coun-
wide variety of proteins (Ruzzo et al. 2019); (Sestan and tries including the UK, current guidelines (e.g. (“Overview
State 2018). Autism spectrum disorder in under 19 s: recognition, referral
and diagnosis Guidance NICE” n.d.) do not recommend rou-
tine genetic testing for autism unless there is accompanying
Genetic Testing and Counselling intellectual disability or dysmorphic features. At the same
time, there is growing interest in the clinical utility of whole
Research advances have led to a widespread appreciation genome sequencing to identify deleterious rare variants (e.g.
now that genetic contributions are important in the etiology in unwell new-born infants) and health providers for some
of autism as a result of which genetic testing and counsel- nations, including NHS England (not across all devolved
ling have become salient to clinicians and affected families UK nations), have expressed the intention of making whole
(see (Griesi-Oliveira and Sertié 2017); (Nurnberger et al. genome sequencing a routine part of medical care.
2019)). Traditionally, where families have wanted to make However, there are certainly many challenges to genetic
reproductive decisions or were concerned about risk in sib- testing for autism. First, it is difficult to clinically interpret
lings, the clinician has relied on recurrence risks reported in findings. Rare variants associated with autism risk show
family studies. One challenge here is the reported estimates variable penetrance, expressivity and are highly pleiotropic
vary widely across studies and country (Jokiranta-Olkoniemi (Rosenfeld et al. 2013; Kirov et al. 2014; Kirov 2015; Fer-
et al. 2016) and very much depend on the sample ascer- nandez and Scherer 2017; Woodbury-Smith et al. 2017).
tained (e.g. whether simplex or multiplex families). Typi- This means that carriers of a given variant and relatives of
cally the rate of autism in siblings of probands has varied those affected could remain healthy, show the same pheno-
from between 10–15% (Vorstman, et al. 2017). However the type but with a very different level of severity or display a
risk of autism in siblings is higher if the proband is female, different phenotype altogether (e.g. ADHD or schizophrenia
in keeping with the female protective effect, and is higher in rather than autism), as we have already discussed.
male than in female siblings (Werling and Geschwind 2015; There are studies which are investigating the effects of
Jokiranta-Olkoniemi et al. 2016; Palmer et al. 2017). Also, specific recurrent variants (e.g. 16p11.2) (D’Angelo et al.
the recurrence risk is much higher if two siblings already are 2016). However, even for these, polygenic background and
affected, reported to rise to around 30–50% (Ozonoff et al. stochastic factors remain relevant influences on the pheno-
2011; Werling and Geschwind 2015). type. Also, inherited CNVs and SNVs may be presumed to
The other problem with recurrence risks is that the risk have different implications for reproductive decisions than
estimate is not individually tailored. This leads us to the de novo variants. A further consideration for genetic coun-
question of molecular genetic testing. At present, there selling is whether de novo variant arises in the germline or
is no clinical rationale for testing common gene variants after fertilization (somatic). Finally, as already mentioned,
because of their limited predictive utility. However, the different autism associated de novo variants can occur in the
situation is different for rare variants. Cytogenetic testing same family (Yuen et al. 2015). Thus, providing accurate
and screening for syndromes such as Fragile X syndrome information to families is challenging.
and Tuberous Sclerosis have long been part of routine clin- Second, observed variants may be known to be associated
ical investigations when these syndromes are suspected. with autism risk but it is not always clearly known if they
Chromosomal microarray investigations are now widely are causal for a given individual. Third, it is important to
available as the first line of genetic testing across many consider potential negative aspects of genetic testing. For
countries. example, what are the impacts of a “negative” test where

13
Journal of Autism and Developmental Disorders (2021) 51:4321–4332 4329

the clinician fails to detect a known pathogenic variant? were made. The images or other third party material in this article are
Will that serve as a disappointment to expectant families included in the article’s Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in
who seek an answer for why their child has autism? Alter- the article’s Creative Commons licence and your intended use is not
natively, for those who carry a rare variant or where one is permitted by statutory regulation or exceeds the permitted use, you will
inherited, will this have negative impacts that include guilt, need to obtain permission directly from the copyright holder. To view a
shame, anxiety as well as potential detrimental effects on copy of this licence, visit http://creat​iveco​mmons​.org/licen​ses/by/4.0/.
life insurance and life prospects including future health and
reproduction? It has been highlighted that despite very rapid
advances in genetic discoveries for complex disorders, this References
has not been accompanied by high quality clinical research
APA. (2013). Diagnostic and statistical manual of mental disorders
on genetic testing in child health and psychiatry including DSM-5. American Psychiatric Association.
how clinicians should be trained about this, how findings Bai, D., Yip, B.H.K., Windham, G.C., Sourander, A., Francis, R.,
should be shared with families and what the clinical utility Yoffe, R., et al. (2019). Association of genetic and environmen-
and long term risks and benefits of testing are. Overall, our tal factors with autism in a 5-country cohort. JAMA Psychiatry,
76(10).
view is that referral to clinical genetics services for investi- Chawner, S. J. R. A., Owen, M. J., Holmans, P., Raymond, F. L.,
gation and counselling is appropriate for autism accompa- Skuse, D., Hall, J., et al. (2019). Genotype–phenotype asso-
nied by ID or complex presentations (comorbid dysmorphic ciations in children with copy number variants associated with
features, a medical condition) but we believe that judgement high neuropsychiatric risk in the UK (IMAGINE-ID): A case-
control cohort study. The Lancet Psychiatry, 6(6), 493–505.
on referral will change rapidly and may depend on the local Colvert, E., Tick, B., McEwen, F., Stewart, C., Curran, S. R., Wood-
context and clinic case-mix. Future criteria and national house, E., et al. (2015). Heritability of autism spectrum disorder
decisions about referral will depend on findings that emerge in a uk population-based twin sample. JAMA Psychiatry, 72(5),
not just from high-tech genetic discoveries but also through 415–423.
D’Angelo, D., Lebon, S., Chen, Q., Martin-Brevet, S., Snyder, L. G.,
research on the clinical utility of genetic testing as well as Hippolyte, L., et al. (2016). Defining the effect of the 16p112
on available healthcare and social resources especially as at duplication on cognition, behavior, and medical comorbidities.
present there is currently no evidence of cost effectiveness JAMA Psychiatry, 73(1), 20–30.
for genetic tests for autism (Ziegler et al. 2017). Davey Smith, G. (2011). Epidemiology, epigenetics and the ‘Gloomy
Prospect’: Embracing randomness in population health research
and practice. International Journal of Epidemiology, 40(3),
537–562.
de Jonge, M., Parr, J., Rutter, M., Wallace, S., Kemner, C., Bailey,
Conclusion A., et al. (2015). New Interview and observation measures of
the broader autism phenotype: Group differentiation. Journal of
Autism and Developmental Disorders, 45(4), 893–901.
Much progress has been made in our understanding of the De Rubeis, S., He, X., Goldberg, A. P., Poultney, C. S., Samocha, K.,
genetics of autism in the last 40 years. We know now that it Cicek, A. E., et al. (2014). Synaptic, transcriptional and chro-
is one of the most heritable of disorders and that typically it matin genes disrupted in autism. Nature, 515(7526), 209–215.
Dinkler, L., Lundström, S., Gajwani, R., Lichtenstein, P., Gillberg,
is multi-factorial in origin. Both common and rare genetic C., & Minnis, H. (2017). Maltreatment-associated neurode-
variants contribute to risk and there is strong interest in uti- velopmental disorders: A co-twin control analysis. Journal of
lising gene discoveries to gain insights into the underlying Child Psychology and Psychiatry, and Allied Disciplines, 58(6),
biology of autism. However, autism shows enormous clinical 691–701.
Fernandez, B. A., & Scherer, S. W. (2017). Syndromic autism spec-
as well as genetic heterogeneity. While the genetic discover- trum disorders: Moving from a clinically defined to a molecu-
ies represent a huge advance, there is a need to link this work larly defined approach. Dialogues in Clinical Neuroscience,
with clinical research. Given the public interest in genetics, 19(4), 353–371.
another pressing clinical issue is how genetic information is Folstein, S., & Rutter, M. (1977). Infantile autism: A genetic study
of 21 twin pairs. Journal of Child Psychology and Psychiatry,
best shared with families and used in a way that is ethical 18(4), 297–321.
and clinically useful. Frazier, T. W., Thompson, L., Youngstrom, E. A., Law, P., Hardan,
A. Y., Eng, C., et al. (2014). A twin study of heritable and
Acknowledgments Dr Thapar’s research is supported by grants shared environmental contributions to autism. Journal of Autism
from the Wellcome Trust, Medical Research Council and Waterloo and Developmental Disorders, 44(8), 2013–2025.
Foundation Ghirardi, L., Brikell, I., Kuja-Halkola, R., Freitag, C.M., Franke, B.,
Asherson, P., et al. (2017). The familial co-aggregation of ASD
Open Access This article is licensed under a Creative Commons Attri- and ADHD: a register-based cohort study. Molecular Psychiatry.
bution 4.0 International License, which permits use, sharing, adapta- Gratten, J., Wray, N. R., Peyrot, W. J., McGrath, J. J., Visscher, P.
tion, distribution and reproduction in any medium or format, as long M., & Goddard, M. E. (2016). Risk of psychiatric illness from
as you give appropriate credit to the original author(s) and the source, advanced paternal age is not predominantly from de novo muta-
provide a link to the Creative Commons licence, and indicate if changes tions. Nature Genetics, 48(7), 718–724.

13
4330 Journal of Autism and Developmental Disorders (2021) 51:4321–4332

Griesi-Oliveira, K., & Sertié, A. L. (2017). Autism spectrum disor- autism spectrum disorder. American Journal of Human Genetics,
ders: An updated guide for genetic counseling. Einstein (Sao 82(2), 477–488.
Paulo, Brazil), 15(2), 233–238. McDonnell, C. G., Boan, A. D., Bradley, C. C., Seay, K. D., Charles,
Grove, J., Ripke, S., Als, T. D., Mattheisen, M., Walters, R., Won, J. M., & Carpenter, L. A. (2019). Child maltreatment in autism
H., et al. (2019). Identification of common genetic risk variants spectrum disorder and intellectual disability: Results from a popu-
for autism spectrum disorder. Nature Genetics, 51(3), 431–444. lation-based sample. Journal of Child Psychology and Psychiatry
Hallmayer, J., Cleveland, S., Torres, A., Phillips, J., Cohen, B., and Allied Disciplines, 60(5), 576–584.
Torigoe, T., et al. (2011). Genetic heritability and shared envi- Miller, M., Young, G. S., Hutman, T., Johnson, S., Schwichtenberg,
ronmental factors among twin pairs with autism. Archives of A. J., & Ozonoff, S. (2015). Early pragmatic language difficul-
General Psychiatry, 68(11), 1095. ties in siblings of children with autism: implications for DSM-5
Harris, J. (2018). Leo Kanner and autism: A 75-year perspective. Inter- social communication disorder? Journal of Child Psychology
national Review of Psychiatry, 30(1), 3–17. and Psychiatry, 56(7), 774–781.
Hoover, D. W., & Kaufman, J. (2018). Adverse childhood experiences Nurnberger, J. I., Austin, J., Berrettini, W. H., Besterman, A. D.,
in children with autism spectrum disorder. Current Opinion in DeLisi, L. E., Grice, D. E., et al. (2019). What should a psy-
Psychiatry, 31(2), 128–132. chiatrist know about genetics? Review and recommendations
Howlin, P., Goode, S., Hutton, J., & Rutter, M. (2009). Savant skills from the residency education committee of the international
in autism: Psychometric approaches and parental reports. Philo- society of psychiatric genetics. Journal of Clinical Psychiatry.
sophical Transactions of the Royal Society B: Biological Sciences, Physicians Postgraduate Press Inc.
364(1522), 1359–1367. Ohlsson Gotby, V., Lichtenstein, P., Långström, N., & Pettersson, E.
Jokiranta-Olkoniemi, E., Cheslack-Postava, K., Sucksdorff, D., (2018). Childhood neurodevelopmental disorders and risk of
Suominen, A., Gyllenberg, D., Chudal, R., et al. (2016). Risk coercive sexual victimization in childhood and adolescence—
of psychiatric and neurodevelopmental disorders among siblings a population-based prospective twin study. Journal of Child
of probands with autism spectrum disorders. JAMA Psychiatry, Psychology and Psychiatry, 59(9), 957–965.
73(6), 622–629. Ozonoff, S., Young, G.S., Carter, A., Messinger, D., Yirmiya, N.,
Kirov, G. (2015). CNVs in neuropsychiatric disorders. Human Molecu- Zwaigenbaum, L., et al. (2011). Recurrence risk for autism
lar Genetics, 24(R1), R45–R49. spectrum disorders: A baby siblings research consortium study.
Kirov, G., Rees, E., Walters, J. T. R., Escott-Price, V., Georgieva, L., Pediatrics, 128(3).
Richards, A. L., et al. (2014). The penetrance of copy number Palmer, N., Beam, A., Agniel, D., Eran, A., Manrai, A., Spettell,
variations for schizophrenia and developmental delay. Biological C., et al. (2017). Association of sex with recurrence of autism
Psychiatry, 75(5), 378–385. spectrum disorder among siblings. JAMA Pediatrics, 171(11),
Le Couteur, A., Bailey, A., Goode, S., Pickles, A., Robertson, S., 1107–1112.
Gottesman, I., et al. (1996). A broader phenotype of autism: the Pickles, A., Starr, E., Kazak, S., Bolton, P., Papanikolaou, K., Bai-
clinical spectrum in twins. Journal of Child Psychology and Psy- ley, A., et al. (2000). Variable expression of the autism broader
chiatry and Allied Disciplines, 37(7), 785–801. phenotype: Findings from extended pedigrees. Journal of Child
Lee, P. H., Anttila, V., Won, H., Feng, Y.-C. A., Rosenthal, J., Zhu, Psychology and Psychiatry and Allied Disciplines, 41(4),
Z., et al. (2019). Genomic relationships, novel loci, and pleio- 491–502.
tropic mechanisms across eight psychiatric disorders. Cell, 179(7), Quesnel-Vallières, M., Weatheritt, R. J., Cordes, S. P., & Blencowe,
1469–1482.e11. B. J. (2019). Autism spectrum disorder: Insights into convergent
Leppert, B., Havdahl, A., Riglin, L., Jones, H. J., Zheng, J., Davey mechanisms from transcriptomics. Nature Reviews Genetics,
Smith, G., et al. (2019). Association of maternal neurodevelop- 20(1), 51–63.
mental risk alleles with early-life exposures. JAMA Psychiatry, Richards, S., Aziz, N., Bale, S., Bick, D., Das, S., Gastier-Foster, J.,
76(8), 834–842. et al. (2015). Standards and guidelines for the interpretation of
Lewis, C. M., & Vassos, E. (2020). Polygenic risk scores: from research sequence variants: A joint consensus recommendation of the
tools to clinical instruments. Genome Medicine, 12(1), 44. American College of Medical Genetics and Genomics and the
Lichtenstein, P., Carlström, E., Råstam, M., Gillberg, C., & Anck- Association for Molecular Pathology. Genetics in Medicine: Offi-
arsäter, H. (2010). The genetics of autism spectrum disorders and cial Journal of the American College of Medical Genetics, 17(5),
related neuropsychiatric disorders in childhood. The American 405–424.
Journal of Psychiatry, 167(11), 1357–1363. Robinson, E. B., Lichtenstein, P., Anckarsäter, H., Happé, F., & Ronald,
Lim, E. T., Uddin, M., De Rubeis, S., Chan, Y., Kamumbu, A. S., A. (2013). Examining and interpreting the female protective effect
Zhang, X., et al. (2017). Rates, distribution and implications of against autistic behavior. Proceedings of the National Academy
postzygotic mosaic mutations in autism spectrum disorder. Nature of Sciences of the United States of America, 110(13), 5258–5262.
Neuroscience, 20(9), 1217–1224. Robinson, E. B., St Pourcain, B., Anttila, V., Kosmicki, J. A., Bulik-
Lundström, S., Chang, Z., Råstam, M., Gillberg, C., Larsson, H., Anc- Sullivan, B., Grove, J., et al. (2016). Genetic risk for autism
karsäter, H., et al. (2012). Autism spectrum disorders and autistic spectrum disorders and neuropsychiatric variation in the general
like traits: Similar etiology in the extreme end and the normal population. Nature Genetics, 48(5), 552–555.
variation. Archives of General Psychiatry, 69(1), 46–52. Rosenfeld, J. A., Coe, B. P., Eichler, E. E., Cuckle, H., & Shaffer, L.
Mandy, W., & Lai, M.-C. (2016). Annual research review: The role of G. (2013). Estimates of penetrance for recurrent pathogenic copy-
the environment in the developmental psychopathology of autism number variations. Genetics in Medicine, 15(6), 478–481.
spectrum condition. Journal of Child Psychology and Psychiatry Rutter, M. (1970s). Genetic studies of autism: from the 1970s into the
and Allied Disciplines, 57(3), 271–292. millennium. Journal of Abnormal Child Psychology, 28(1), 3–14.
Marshall, C. R., Howrigan, D. P., Merico, D., Thiruvahindrapuram, B., Rutter, M. (2007). Incidence of autism spectrum disorders: Changes
Wu, W., Greer, D. S., et al. (2017). Contribution of copy number over time and their meaning. Acta Paediatrica, 94(1), 2–15.
variants to schizophrenia from a genome-wide study of 41,321 Rutter, M. (2011). Progress in understanding autism: 2007–2010. Jour-
subjects. Nature Genetics, 49(1), 27–35. nal of Autism and Developmental Disorders, 41(4), 395–404.
Marshall, C. R., Noor, A., Vincent, J. B., Lionel, A. C., Feuk, L., Rutter, M. (2013). Changing concepts and findings on autism. Journal
Skaug, J., et al. (2008). Structural variation of chromosomes in of Autism and Developmental Disorders, 43(8), 1749–1757.

13
Journal of Autism and Developmental Disorders (2021) 51:4321–4332 4331

Rutter, M. (2015). Some of the complexities involved in gene-envi- Thapar, A., & Rice, F. (2020). Family-based designs that disentangle
ronment interplay. International Journal of Epidemiology, 44(4), inherited factors from pre- and postnatal environmental expo-
1128–1129. sures: in vitro fertilization, discordant sibling pairs, maternal
Rutter, M., Kreppner, J., Croft, C., Murin, M., Colvert, E., Beckett, versus paternal comparisons, and adoption designs. Cold Spring
C., et al. (2007). Early adolescent outcomes of institutionally Harbor Perspectives in Medicine. https​://doi.org/10.1101/cshpe​
deprived and non-deprived adoptees. III. Quasi-autism. Jour- rspec​t.a0388​77.
nal of Child Psychology and Psychiatry, and Allied Disciplines, Thapar, A., Rice, F., Hay, D., Boivin, J., Langley, K., van den Bree,
48(12), 1200–7. M., et al. (2009). Prenatal smoking might not cause attention-
Rutter, M., & Pickles, A. (2016). Annual research review: Threats deficit/hyperactivity disorder: Evidence from a novel design.
to the validity of child psychiatry and psychology. Journal of Biological Psychiatry, 66(8), 722–727.
Child Psychology and Psychiatry and Allied Disciplines, 57(3), Thapar, A., & Rutter, M. (2015a). Neurodevelopmental disorders.
398–416. Rutter’s child and adolescent psychiatry (pp. 31–40). Chichester:
Ruzzo, E. K., Pérez-Cano, L., Jung, J. Y., Wang, L. K., Kashef- John Wiley & Sons Ltd.
Haghighi, D., Hartl, D., et al. (2019). Inherited and de novo Thapar, A., & Rutter, M. (2015b). Using natural experiments and ani-
genetic risk for autism impacts shared networks. Cell, 178(4), mal models to study causal hypotheses in relation to child mental
850–866.e26. health problems. In A. Thapar, D. S. Pine, J. F. Leckman, S. Scott,
Sanders, S. J., Ercan-Sencicek, A. G., Hus, V., Luo, R., Murtha, M. Snowling, & Tay (Eds.), Rutter’s child and adolescent psychia-
M. T., Moreno-De-Luca, D., et al. (2011). Multiple recurrent try (6th ed., pp. 145–162). Oxford: John Wiley and Sons Limited.
de novo cnvs, including duplications of the 7q1123 williams Thapar, A., & Rutter, M. (2019). Do natural experiments have an
syndrome region, are strongly associated with autism. Neuron, important future in the study of mental disorders? Psychological
70(5), 863–885. Medicine, 49(07), 1079–1088.
Sanders, S. J., He, X., Willsey, A. J., Ercan-Sencicek, A. G., Samo- Tick, B., Bolton, P., Happé, F., Rutter, M., & Rijsdijk, F. (2016). Her-
cha, K. E., Cicek, A. E., et al. (2015). Insights into autism spec- itability of autism spectrum disorders: a meta-analysis of twin
trum disorder genomic architecture and biology from 71 risk studies. Journal of Child Psychology and Psychiatry and Allied
loci. Neuron, 87(6), 1215–1233. Disciplines, 57(5), 585–595.
Satterstrom, F. K., Kosmicki, J. A., Wang, J., Breen, M. S., De Trost, B., Engchuan, W., Nguyen, C.M., Thiruvahindrapuram, B.,
Rubeis, S., An, J.-Y., et al. (2020). Large-scale exome sequenc- Dolzhenko, E., Backstrom, I., et al. (2020). Genome-wide detec-
ing study implicates both developmental and functional changes tion of tandem DNA repeats that are expanded in autism. Nature,
in the neurobiology of autism. Cell, 180(3), 568–584.e23. 1–9.
Schaefer, G. B., & Mendelsohn, N. J. (2013). Clinical genetics evalu- Vicari, S., Napoli, E., Cordeddu, V., Menghini, D., Alesi, V., Loddo, S.,
ation in identifying the etiology of autism spectrum disorders: et al. (2019). Copy number variants in autism spectrum disorders.
2013 guideline revisions. Genetics in Medicine, 15(5), 399–407. Progress in Neuro-Psychopharmacology & Biological Psychiatry,
Searles Quick, V. B., Wang, B., & State, M. W. (2020). Leverag- 92, 421–427.
ing large genomic datasets to illuminate the pathobiology of Vorstman, J. A. S., Parr, J. R., Moreno-De-Luca, D., Anney, R. J. L.,
autism spectrum disorders. Neuropsychopharmacology. https​ Nurnberger, J. I., & Hallmayer, J. F. (2017). Autism genetics:
://doi.org/10.1038/s4138​6-020-0768-y. Opportunities and challenges for clinical translation. Nature
Sebat, J., Lakshmi, B., Malhotra, D., Troge, J., Lese-Martin, C., Reviews Genetics, 18(6), 362–376.
Walsh, T., et al. (2007). Strong association of de novo copy Weiner, D. J., Wigdor, E. M., Ripke, S., Walters, R. K., Kosmicki, J.
number mutations with autism. Science, 316(5823), 445–449. A., Grove, J., et al. (2017). Polygenic transmission disequilibrium
Sestan, N., & State, M. W. (2018). Lost in translation: Traversing the confirms that common and rare variation act additively to cre-
complex path from genomics to therapeutics in autism spectrum ate risk for autism spectrum disorders. Nature Genetics, 49(7),
disorder. Neuron, 100(2), 406–423. 978–985.
St Pourcain, B., Robinson, E. B., Anttila, V., Sullivan, B. B., Maller, Werling, D. M., Brand, H., An, J.-Y., Stone, M. R., Zhu, L., Glessner,
J., Golding, J., et al. (2018). ASD and schizophrenia show dis- J. T., et al. (2018). An analytical framework for whole-genome
tinct developmental profiles in common genetic overlap with sequence association studies and its implications for autism spec-
population-based social communication difficulties. Molecular trum disorder. Nature Genetics, 50(5), 727–736.
Psychiatry, 23(2), 263–270. Werling, D. M., & Geschwind, D. H. (2015). Recurrence rates provide
State, M., & Thapar, A. (2015). Genetics. In A. Thapar, D. S. Pine, evidence for sex-differential, familial genetic liability for autism
J. F. Leckman, S. Scott, M. J. Snowling, E. Taylor (Eds.), Rut- spectrum disorders in multiplex families and twins. Molecular
ter’s Child and Adolescent Psychiatry, 6th edition. Oxford: John Autism, 6, 27.
Wiley and Sons Limited. WHO. (2019). WHO international classification of diseases, 11th revi-
Sullivan, P. F., Agrawal, A., Bulik, C. M., Andreassen, O. A., Bør- sion (ICD-11). World Health Organization: WHO.
glum, A. D., Breen, G., et al. (2018). Psychiatric genomics: An Williams, N. M., Zaharieva, I., Martin, A., Langley, K., Mantripragada,
update and an agenda. American Journal of Psychiatry, 175(1), K., Fossdal, R., et al. (2010). Rare chromosomal deletions and
15–27. duplications in attention-deficit hyperactivity disorder: A genome-
Sztainberg, Y., & Zoghbi, H. Y. (2016). Lessons learned from study- wide analysis. The Lancet, 376(9750), 1401–1408.
ing syndromic autism spectrum disorders. Nature Neuroscience, Wiśniowiecka-Kowalnik, B., & Nowakowska, B. A. (2019). Genetics
19(11), 1408–1417. and epigenetics of autism spectrum disorder-current evidence in
Taylor, J. L., Debost, J., Morton, S., Wigdor, E., Heyne, H., Lal, D., the field. Journal of Applied Genetics, 60(1), 37–47.
et al. (2019a). Paternal-age-related de novo mutations and risk Woodbury-Smith, M., Nicolson, R., Zarrei, M., Yuen, R. K. C., Walker,
for five disorders. Nature Communications, 10(1), 3043. S., Howe, J., et al. (2017). Variable phenotype expression in a
Taylor, M. J., Martin, J., Lu, Y., Brikell, I., Lundström, S., Lars- family segregating microdeletions of the NRXN1 and MBD5
son, H., et al. (2019b). Association of genetic risk factors for autism spectrum disorder susceptibility genes. NPJ Genomic
psychiatric disorders and traits of these disorders in a Swedish Medicine, 2, 17.
population twin sample. JAMA Psychiatry, 76(3), 280.

13
4332 Journal of Autism and Developmental Disorders (2021) 51:4321–4332

Woodbury-Smith, M., & Scherer, S. W. (2018). Progress in the genetics sequencing resource identifies 18 new candidate genes for autism
of autism spectrum disorder. Developmental Medicine and Child spectrum disorder. Nature Neuroscience, 20(4), 602.
Neurology, 60(5), 445–451. Ziegler, A., Rudolph-Rothfeld, W., & Vonthein, R. (2017). Genetic test-
Yip, B. H. K., Bai, D., Mahjani, B., Klei, L., Pawitan, Y., Hultman, C. ing for autism spectrum disorder is lacking evidence of cost-effec-
M., et al. (2018). Heritable variation, with little or no maternal tiveness. Methods of Information in Medicine, 56(03), 268–273.
effect, accounts for recurrence risk to autism spectrum disorder
in Sweden. Biological Psychiatry, 83(7), 589–597. Publisher’s Note Springer Nature remains neutral with regard to
Yuen, R. K. C., Thiruvahindrapuram, B., Merico, D., Walker, S., Tam- jurisdictional claims in published maps and institutional affiliations.
mimies, K., Hoang, N., et al. (2015). Whole-genome sequencing
of quartet families with autism spectrum disorder. Nature Medi-
cine, 21(2), 185–191.
Yuen, R. K., Merico, D., Bookman, M., Howe, J. L., Thiruva-
hindrapuram, B., Patel, R. V., et al. (2017). Whole genome

13