Skip to main content
Premium Trial:

Request an Annual Quote

De Novo, Inherited Autism Risk Factors Uncovered With SPARK Cohort Analysis


NEW YORK – Although the majority of de novo contributors to autism spectrum disorder (ASD) are being uncovered, there appear to be many inherited risk variants that remain undetected, according investigators who searched for both types of variants in a large set of ASD-affected families enrolled through the Simons Foundation's "Simons Powering Autism Research " (SPARK) project.

At the American Society of Human Genetics (ASHG) virtual meeting on Wednesday, Columbia University Medical Center pediatrics and systems biology researcher Xueya Zhou provided an early look at unpublished findings from genetic analyses that tapped into available SNP array and exome sequence data for 28,649 individuals from more than 9,100 simplex or multiplex ASD families enrolled through the national SPARK research cohort, including 10,242 individuals with ASD.

"The field of autism genetics has put a very heavy focus on [finding] genomic variants, and over 100 autism genes have been identified to date," Zhou said during his plenary presentation, noting that "most of these genes are also mutated in other neurodevelopmental diseases."

Nevertheless, the SPARK team — which aims to collect data from some 50,000 families in the next few years — reasoned that there may be benefits to continuing to search for yet undiscovered genetic contributors to ASD in a large cohort that includes families with one or more affected child.

Many of the ASD-related genes identified so far have been linked to the autism, intellectual disability, and other neurodevelopmental conditions through de novo variant analyses, Zhou explained. But he and his colleagues also did not discount the possibility that there might be rare, inherited and more common polygenic ASD-related variants that remain undiscovered, since the heritability of autism has been estimated at up to around 80 percent.

"There's a strong motivation to continue to search for genes that are enriched by de novo [variants], because the estimated mutational targets are on the order of 1,000 genes," Zhou said. "On the other hand, autism also has a very high estimated heritability and studies so far have only scratched the surface of inherited variants."

When they analyzed de novo variants using available data on more than 17,300 ASD-affected trios assessed through SPARK or other studies, the researchers found that LoF or deleterious missense variants that occurred de novo in individuals with ASD turned up at higher-than-usual rates in genes that are considered intolerant to LoF changes, Zhou reported, noting that he and his colleagues flagged almost four-dozen genes that were enriched for deleterious de novo changes in individuals with ASD at genome-wide significance.

The investigators also found that LoF changes were particularly common across the top 100 high-confidence autism genes delineated by the de novo variant analyses, which included previously unappreciated ASD-related genes from pathways involved in transcriptional regulation, post-transcriptional regulation, protein modification, signal transduction, and other processes.

Together with dominant genes linked to other neurodevelopmental conditions in the past, the collection of high-confidence autism genes seemed to explain roughly two-thirds of the de novo variant burden behind ASD, Zhou noted, suggesting that researchers are inching ever closer to understanding the broader suite of de novo changes that can predispose individuals to ASD.

"We would expect that the residual de novo burden will be explained in the near future, when new autism genes are identified," he said.

When it came to inherited variants with potential ties to ASD, meanwhile, the researchers began by searching for rare, LoF variants in parents, before looking at whether these alterations were turning up in ASD-affected children at different rates than their unaffected siblings in more than 7,000 parent-child trios and 184 parent-child duos from SPARK, and in thousands more families previously analyzed from the Simons Simplex Collection.

After taking steps to filter out false-positive LoF variants not involved in this transmission disequilibrium, the team did not see inherited LoF variants in the set of genes that came to the forefront in the de novo variant analysis, although a small subset of genes such as KDM5B, GIGYF, and PTEN appeared prone to both inherited or de novo variation in individuals with ASD.

"We reason that, theoretically, a gene can contribute to autism risk through both de novo and inherited variants, but with different proportions. And the proportion of de novo LoF is determined by selection coefficients," Zhou explained, noting that relaxed selection constraints appeared to be a shared feature in de novo-mutated ASD genes showing relatively high rates of rare, inherited mutations in ASD.

The investigators also saw signs that some of the same genes impacted by rare, inherited LoF variants in ASD are affected by de novo variation in not only ASD, but in other neurodevelopmental conditions such as schizophrenia.

LoF variants that were overtransmitted to ASD cases and undertransmitted from parents to their unaffected children again seemed to center on genes that are typically considered LoF intolerant, Zhou noted.

In contrast to the de novo variant analysis, though, findings from the inherited variant side hinted that there may still be many ASD-related genes and overtransmitted parent-to-child risk variants that have not yet been found.

The team's subsequent exome sequence meta-analysis led to a ubiquitin ligase enzyme-coding gene called UBR1 that appears to contribute to inherited ASD risk, but not to the risk of ASD via de novo variation, while a machine-learning-based gene set enrichment analysis pointed to the possibility of uncovering still other ASD-related genes based on transmission disequilibrium clues coupled with single-cell RNA sequence data from human brain cells.

Finally, Zhou noted that transmission patterns for both rare, inherited variants and common variants from polygenic risk scores for ASD and related traits appear to show promise for teasing out the genetic differences between ASD cases with specific symptoms such as cognitive impairment.

"This suggests that [polygenic risk scores] derived from different base phenotypes indeed capture the effects of common variants in different gene sets across the genome, and they have different effects on the phenotype spectrum," Zhou reported, adding that SPARK results support the notion that the "phenotype manifestations of autism are, indeed, a combined effect of all types of variants across different disease-associated genes."