Autism Spectrum Disorder (ASD) cases can be clustered into two types based upon the risk factor genes involved. This division coincides with a distinctions between severe symptoms and less severe symptoms. The two clusters appear to have distinct causes. Basically, it appears that there are two different conditions that happen to have symptoms that resemble each other.
Knowing what causes a particular individual's ASD could be critical in figuring out what kind of therapies or symptom management strategies are likely to work best for a particular individual.
Autism Spectrum Disorder (ASD) is a complex neurodevelopmental disease primarily characterized by deficits in verbal communication, impaired social interaction and repetitive behaviors. It exemplifies profound clinical heterogeneity, which poses challenges in diagnosis and treatment. Genetic studies have pointed to hundreds of presumptive causative or susceptibility genes in ASD, making it difficult to find common underlying pathogenic mechanisms and suggesting that multiple different genetic etiologies for ASDs influence a continuum of traits.
Deep phenotyping analysis allowed for re-categorization of genetic variants. Our previous analysis suggested the existence of two significant subgroups within the existing ASD classification. To investigate this hypothesis in greater detail we have performed in-depth analysis using phenotypic and genetic data from Autism Genetic Resource Exchange (AGRE) and Autism Genome Project (AGP). Our initial findings on both phenotypic and genetic data (1,262 cases and 2,521 controls using familial transmission disequilibrium test) suggest existence of two groups that range in severity. Findings were replicated in a validation dataset. Genetic risk scores (GRS) were used to sum up the total effect of several single-nucleotide polymorphisms characteristic of the two clusters. The high discriminatory ability of the genetic risk score to define cluster 1 from cluster 2 case group at different combinations of sensitivity and specificity was assessed and clearly demonstrates strong signal with AUC being 0.74. There is a significant signal differentiating the 2 clusters relying on non-genetic risk factors and even greater signal when using both non-genetic risk factors and GRS. The detection and validation of the two groups allowed us focus on convergence of findings at the pathway level. ASD heterogeneity was leveraged via large scale pathway analysis within those two categories, which led to identification of a driver gene set across significant pathways. The significant pathways in cluster 1 (severe, affected = 300) include autoimmune disease, vitamin B6 metabolism, whereas in cluster 2 (non-severe, affected = 921) included oxytocin signaling pathway, WNT signaling pathway and glutamatergic synapses (all at P < 0.001). We envision that systematic study of all genomic pathways obtained given a set of redefined categories will yield profound findings for ASD even in the absence of strong individual variant information.S. Smieszek and J.L. Haines., "Autism redefined: Genomic pathway approach to autism spectrum disorder." ASHG Conference Presentation 33 (October 2016).