Autism spectrum disorder

A child carries a diagnosis of autism spectrum disorder and the family wants to know two things: is there an identifiable cause, and what is the chance it happens again. The clinical work splits along a single axis. Syndromic autism comes with dysmorphic features, intellectual disability, congenital anomalies, regression, seizures, or an abnormal growth parameter, and the diagnostic yield is high. Idiopathic or nonsyndromic autism is autism with a normal examination and often normal cognition, and it is the far larger group. Both deserve testing; the difference is what you expect to find and what you say about recurrence.

Autism is highly heritable and highly heterogeneous at the same time, which is why single answers are rare.

• Prevalence is about 1 in 31 among 8-year-olds in the most recent CDC surveillance (2022 data), up substantially over two decades, driven largely by broadened criteria and better ascertainment
• Sex ratio is roughly 3 to 4 affected males per affected female. Part of this is ascertainment, since autistic girls are diagnosed later and are more likely to camouflage, and part appears to be a genuine female protective effect: a girl needs a larger genetic burden to cross the threshold
• Twin heritability estimates run 60 to 90%, among the highest of any neurodevelopmental condition
• De novo variants account for a substantial share of simplex cases (one affected child, no family history): CHD8, SCN2A, ARID1B, SYNGAP1, ADNP, DYRK1A, POGZ, SHANK3, GRIN2B and many others. Well over a hundred genes now reach genome-wide significance
• Recurrent copy number variants contribute another slice: 16p11.2 deletion and duplication, maternal 15q11-13 duplication, 22q11.2, 1q21.1, and 7q11.23 duplication (the reciprocal of the Williams syndrome deletion)
• Common polygenic variation contributes real liability, with SNP heritability around 10 to 20%, but polygenic scores are not clinically actionable and should not be ordered
• Advanced paternal age raises risk through the de novo mutation rate, which rises with paternal age at conception

This is where the counseling most often goes wrong, because the generic multifactorial figure of 3 to 5% quoted for many traits is far too low here.

• Sibling recurrence in idiopathic autism is roughly 10 to 20%. Prospective infant-sibling studies put it near 19%
• If the proband is female, recurrence is higher, roughly double, consistent with the female protective effect: an affected girl implies more familial liability
• If two children are already affected, recurrence rises to about one in three
• A broader autism phenotype (language delay, social difficulty short of diagnosis) is seen in a further share of siblings
• When a specific cause is identified, the mechanism replaces these figures. A de novo variant carries a low recurrence, on the order of 1 to 2% from gonadal mosaicism rather than zero. An inherited variant, an unbalanced product of a parental rearrangement, or an X-linked cause each carries its own number

• Macrocephaly: the single highest-yield association, pointing to PTEN hamartoma tumor syndrome
• Microcephaly, congenital anomalies, or distinctive facial features
• Regression, particularly loss of purposeful hand use in a girl aged 6 to 18 months, pointing to Rett syndrome (MECP2)
• Seizures, especially infantile spasms, pointing to tuberous sclerosis complex
• Intellectual disability, which roughly doubles to triples the diagnostic yield of any test ordered. See intellectual disability and global developmental delay
• Male with large ears, long face, and macroorchidism after puberty, pointing to fragile X syndrome
• A family history of intellectual disability in males on the maternal side, pointing to X-linked intellectual disability

1. Exome or genome sequencing. Current professional guidance places sequencing as a first-tier test for autism with developmental delay or intellectual disability. Yield is roughly 8 to 20% and rises with comorbid intellectual disability and dysmorphism
2. Chromosomal microarray. Yield around 5 to 10%. Still valuable alongside sequencing, since it catches the recurrent CNVs and regions of homozygosity that exome coverage handles poorly. See chromosomal microarray
3. FMR1 CGG repeat sizing. Yield 1 to 2%, higher in males, and it is not reliably detected by sequencing or microarray, so it has to be ordered separately
4. MECP2 in any girl with regression, if not already covered by the sequencing panel
5. Targeted testing when a syndrome is recognizable, rather than waiting for broad testing to return
6. Karyotype only when a specific chromosomal rearrangement is suspected, for example a family history of recurrent pregnancy loss
7. Adjuncts as indicated by phenotype: EEG for suspected seizures, brain MRI for focal findings or regression, audiology in every child with language delay, and lead level where exposure is plausible

Combined first-tier yield in autism without intellectual disability or dysmorphism is around 10%. With intellectual disability, dysmorphism, or congenital anomalies it reaches 30 to 40%.