|  Autism
is one of the most heritable complex genetic disorders in psychiatry.
Despite this high heritability, autism has a heterogeneous etiology, with
multiple genes and chromosomal regions likely to be involved. Scientists
are using both indirect and direct approaches to identify autism susceptibility
genes. Indirect approaches include the characterization of less complex
genetic disorders that share some of the symptoms of autism, including
Rett syndrome or fragile X syndrome, in the hope that these analyses will
provide clues to the more complex disorder of autism (see recent reviews
in this space, February and May 2000). Direct approaches include three
overlapping methodologies to identify genes or regions of interest in
autism: chromosomal methods, such as karyotyping and fluorescence in situ
hybridization (FISH); linkage studies, such as genome screens in affected
sibling pairs; and gene association studies, including candidate gene
studies. These approaches are yielding preliminary findings that are reviewed
here. While no specific gene variant has been identified and confirmed
that contributes to the expression of autism, it is very likely that several
will be confirmed over the next decade.
Twin
and sibling studies demonstrate heritability in autism. The monozygotic
twin of a patient with autism, who shares nearly 100% of nuclear DNA,
has an approximately 60% chance of having autism, while the concordance
for an autism spectrum disorder is greater than 90%. The dizygotic twin
of a patient, who shares 50% of the genes, has approximately the same
risk as a sibling, about 4.5%. A member of the general population has
approximately 0.2% chance of having autism. The ratios between these different
risks provide strong evidence for heritability. The dramatically diminished
risk in relatives who share 50% versus 100% of their DNA is most consistent
with an oligogenic inheritance pattern, where more than 2 and as many
as 100 genetic variants may contribute to susceptibility to developing
autism.
Genetic
heterogeneity in autism is no surprise to clinicians familiar with the
varied presentation of the disorder. Recent advances in diagnosis, such
as the Autism Diagnostic Inventory and the Autism Diagnostic Observation
Schedule, may reduce uncertainty in diagnosis, but clinical heterogeneity
remains. Symptoms and signs, rather than etiologies, currently characterize
psychiatric syndromes such as autism, schizophrenia, or attention-deficit/hyperactivity
disorder, as well as other medical syndromes, including diabetes and asthma.
Each gene may make a different contribution to the disorder, with gene
A more important for the development of social cognition and gene B more
important for language acquisition.
On
the other hand, different variants in the same gene may also produce different
clinical pictures. When clustering of risk alleles reaches a certain threshold,
an individual is at increased risk of developing the disorder. A subthreshold
number of risk alleles may result in the broader autism phenotype identified
in family members of patients with autism. It is also likely that several
variants that contribute to susceptibility to autism (and other childhood-onset
psychiatric disorders) will be relatively common in the general population
and may even be advantageous (e.g., a hypothetical variant of a gene may
heighten focused attention and add to risk for autism but in another context
be helpful).
Language
disturbance is an important characteristic of autism. A mutation in the
transcription factor FOXP2 on chromosome 7q was found in affected members
of a family with an autosomal dominant speech and language disturbance.
However, this gene was screened in a large sample of children with autism
and no evidence was found for its relationship to that disorder.
A
large number of chromosomal abnormalities have been reported in autism.
The study of these abnormalities serves dual purposes. First, these abnormalities
may be characteristic of specific subsyndromes. Second, they point to
the chromosomal location of a gene that may have other types of mutations
(i.e., point mutations or deletions too small to be visible by karyotyping).
In this way, researchers will get hints that a gene or gene system may
be involved in patients without visible chromosomal abnormalities.
Chromosomal
abnormalities have been reported on chromosomes 2q37, 7q, and 22q13, among
others. However, the most common specific cause of autism appears to be
maternally inherited duplications of chromosome 15q11-13, accounting for
1% to 3% of cases. Maternal duplication of this region is also found in
another childhood developmental disorder, Angelman syndrome. There are
several types of mutations in this region that result in Angelman syndrome.
A minority of cases have been found to be the result of a point mutation
in a gene termed UBE3A. The majority of cases are caused by a
large, visible deletion that removes several genes surrounding UBE3A,
or after inheritance of two copies of the paternal copy of chromosome
15 (uniparental disomy) (Angelman syndrome was reviewed in this column
in July 2000). In some brain areas or developmental periods, a few genes
in the region, including UBE3A and ATP10C, are active
only when they are inherited from the mother.
Linkage
studies use highly variable polymorphisms spaced evenly throughout the
genome to identify chromosomal regions shared among family members who
are affected with a particular disease. The likelihood of linkage at a
given point in the genome is based on the percentage of sharing between
affected family members at the nearest polymorphisms. Linkage results
are presented as a LOD score, which is the log (base 10) of the ratio
of the likelihood of linkage relative to no linkage. Several genomewide
linkage studies have been reported in autism over the past few years,
primarily using affected sibling pairs. Almost all studies failed to find
significant evidence for linkage. A peak on chromosome 2q achieved strict
genomewide significance in a single study (LOD 4.8, p = 1.2 3
10–6). Meta-analysis of the first five published studies
implicates two additional regions of interest on chromosome 7q and 13q.
Smaller linkage peaks have been reported in numerous chromosomal regions.
(Figure 1)
Once
significant linkage findings have been identified, the next step is to
use more closely packed polymorphisms to pinpoint the area of linkage.
Initial studies suggest that narrowing the phenotype of interest may also
help in identifying a gene. While a large number of variables could be
used to identify a subphenotype with greater evidence for linkage in a
region, investigators have begun by narrowing the diagnostic criteria.
For example, restricting the analysis to children with autism accompanied
by severe language delays increased linkage findings on both 7q and 2q.
Other variables that might be appropriate for consideration include head
circumference, platelet serotonin, seizure disorder, and restrictive or
repetitive behaviors.
Genetic
association studies are often used both in the initial approach and in
the final step in the dissection of complex genetic disease. Association
studies look for excess sharing of alleles not just in family members
with a syndrome, but among individuals with the disorder across families.
Initial genetic association studies compared allele frequencies in patients
in comparison with controls from the general population, but these studies
were liable to population stratification bias due to undetected differences
in the genetic backgrounds of the two groups of subjects. The transmission-disequilibrium
test applied to parent–child trios uses family-based controls to
avoid this type of bias.
The
initial approach to most psychiatric disorders has considered candidate
genes, thought to be involved on the basis of pathophysiology. Some studies
have merged this approach with genomic approaches, studying only those
candidate genes that lie in chromosomal regions implicated by other genetic
evidence, such as karyotyping analyses. Finally, following significant
linkage studies, the last step in identifying complex disease genes is
association mapping in regions of interest without regard to hypothetical
candidate genes. When genotyping thousands of polymorphisms in each subject
becomes economically feasible, genomewide searches will harness the increased
power to detect gene association rather than linkage.
A
number of family-based gene associations have been reported in autism.
Two of these association findings have been replicated at least once,
although negative association studies have also been reported. The serotonin
transporter gene (SLC6A4) was selected as a candidate gene because
of hyperserotonemia observed in approximately 25% of patients with autism
and the efficacy of serotonin transporter inhibitors in treating rituals
and preoccupations associated with anxiety or aggression. Most studies
of SLC6A4 have found association with a promoter variant (5-HTTLPR),
but other studies have found association with opposite alleles. Following
up on these findings, a linkage study found a single-point LOD score of
3.6 (p = 2.2 3 10–5) at a variant in the first
intron of SLC6A4. A recent study found more significant association
with other SLC6A4 polymorphisms. Association has also been assessed
at a number of genes in the chromosome 15q11-13 duplication region. A
polymorphism in the GABA type A receptor b3
subunit (GABRB3) has been associated with autism in two studies,
but association has also been reported with a separate polymorphism near
GABRB3 and a polymorphism in the nearby Angelman disorder gene,
UBE3A.
A
number of promising association studies have not yet been replicated.
Evidence for maternal transmission of alleles at three polymorphisms was
identified within the glutamate receptor 6 gene (GRIK2) on chromosome
6q21, a region with suggestive evidence for linkage in one sample. Association
has also been reported at two genes on chromosome 7q near the linkage
finding: RELN, a gene that is mutated in recessive lissencephaly;
and WNT2, a gene implicated in the development of the central
nervous system.
Initial
genetic studies in autism have been promising, but the detection of genetic
variants responsible for disease has thus far been elusive. This is not
unexpected in a complex genetic disease. Subgroups of patients may be
identified with a simpler genetic etiology. However, most patients are
likely to have at least 2 genes, and perhaps as many as 100, acting in
concert to cause susceptibility to the disorder. Several initial linkage
studies have yielded suggestive results in various chromosomal regions,
but only a few of these regions have been implicated in more than one
study, particularly at 2q and 7q. Significant association findings with
a number of candidate genes are encouraging, but efforts to replicate
these findings have been mixed.
Although
a disadvantage in studying a complex genetic disorder is the difficulty
in identifying the variants, a potential advantage is that more options
for intervention may be implicated by multiple genes, and these genes
may be related to a developmental neurobiological system that is not yet
fully understood. Chromosomal findings such as FRAXA and maternally inherited
duplications 15q11-q13 do provide additional information about recurrence
risk for many families. However, it is important to recognize that for
almost all cases, genetic testing is unlikely to establish a diagnosis
of autism in the absence of careful clinical evaluation, since FRAXA,
maternally inherited 15q11-q13 duplications, and other syndromes greatly
increase the risk for autism, but do not lead to autism spectrum disorders
in all cases.
top of page
|
