How Far Autism Genetics Is From Anything a School Can Use

What the review found

The unit of description in this review is the synapse, the junction where one neuron passes a signal to the next. On the receiving side of that junction lies an assembly called the postsynaptic density, built from more than a thousand distinct proteins that hold receptors in position and convert an arriving signal into a change inside the cell. Nothing in the paper is measured at a larger scale.

Yuan Zhang, Rui Tang and colleagues, pathologists at Chengdu Women’s and Children’s Central Hospital and at Southwest Medical University in Sichuan, assembled the survey for the Journal of Integrative Neuroscience in September 2024. It is a narrative review rather than a systematic one, with no search protocol, no stated inclusion criteria and no pooled figures. It describes the state of a field.

More than a thousand genes or genomic loci have been associated with autism. Chromosomal abnormalities appear in roughly 2 to 5 percent of autistic individuals, and copy number variants, stretches of DNA duplicated or deleted outright, are more common; the study that first reported them found such a variant, arisen newly in the child rather than inherited, in over 10 percent of children with autism against fewer than 1 percent of controls. Duplications and deletions at 16p11.2 are the most frequent.

Several genes carry syndromes with them. Mutations in FMR1 cause fragile X syndrome and are identified in 2 to 5 percent of autistic individuals through testing, and disruption of SHANK3 defines Phelan-McDermid syndrome, more than half of whom show autistic behaviours. Mutations in CHD8 appear in almost 4 percent of autism diagnoses, with an unusually large head in 85 percent of those cases.

These genes, along with NLGN, NRXN and MECP2, do the same class of work: they build synapses, tune the strength of transmission across them, or permit the adjustments that constitute learning. The authors propose that a disturbed balance between excitatory and inhibitory signalling is the baseline mechanism these separate genetic routes arrive at.

Almost every mechanistic result behind that proposal was obtained in genetically modified mice. The human evidence is of two kinds: sequence data from living people, and post-mortem brain tissue, in which increased dendritic spine density on cortical projection neurons and an excess of excitatory over inhibitory synapses in the prefrontal cortex have been reported. No synapse in a living child has been measured.

The imbalance also runs in both directions. Mice lacking exons 6 and 7 of Shank2 show reduced NMDA receptor function; mice lacking only exon 7 of the same gene show enhanced function. Mice deficient in Cntnap2 or Scn2a transmit less at excitatory synapses, while mice carrying a 16p11.2 duplication transmit more. Opposite molecular states produce the same behavioural phenotype in the animal.

What follows from it, and what does not

The corrections reported in the review are corrections of mice. Restoring a regulator called Npas4 in mice with a 16p11.2 duplication reversed their social and cognitive deficits; a compound acting on the enzyme that controls Npas4 improved social preference in the same strain; another compound reduced repetitive behaviour in a mouse model of fragile X. No human trial of any of them is reported here.

Set against that, the review’s account of current treatment occupies three sentences. Risperidone and aripiprazole are approved in the United States, they target associated symptoms, and they are not effective for the core characteristics of autism. Early behavioural intervention remains, in the authors’ assessment, the most effective approach available.

A shared pathway gives pharmacology somewhere to aim without yet supplying a compound, and the difficulty of aiming is visible in the review’s own contents, since a drug that raised excitatory transmission would meet children whose synapses are already transmitting too much. That reading of the two-directional finding is an inference from the review rather than a claim it makes.

What genetics delivers to a family now is a name and its associations. A confirmed SHANK3 disruption or SYNGAP1 variant tells a clinician which conditions to watch for, and the review lists the usual companions: epilepsy, intellectual disability, sleep disturbance, anxiety, gastrointestinal problems. It does not alter how a child is taught to read.

Nor do the commoner variants function as a test. One much-studied single-letter change in SHANK3 was associated with autism in Iranian and Wenzhou samples and showed no association in Northern Han Chinese, Bangladeshi, Japanese or Caucasian ones. The genes with clear consequences are individually rare, and the variants that are common are weak, inconsistent, and diagnostic of nothing.

The authors close by saying that this work should eventually deepen understanding and facilitate tailored interventions. Eventually is the honest word, and they name no candidate close to a clinic. A field thirty years into the genetics of a condition, having found where more than a thousand genes converge and not yet what to do about it, is a field at an ordinary stage of its work. None of it changes what will help your child this year, and it can be left to arrive in its own time.