A framework for an evidence-based gene list relevant to autism spectrum disorder

Abstract

Autism spectrum disorder (ASD) is often grouped with other brain-related phenotypes into a broader category of neurodevelopmental disorders (NDDs). In clinical practice, providers need to decide which genes to test in individuals with ASD phenotypes, which requires an understanding of the level of evidence for individual NDD genes that supports an association with ASD. Consensus is currently lacking about which NDD genes have sufficient evidence to support a relationship to ASD. Estimates of the number of genes relevant to ASD differ greatly among research groups and clinical sequencing panels, varying from a few to several hundred. This Roadmap discusses important considerations necessary to provide an evidence-based framework for the curation of NDD genes based on the level of information supporting a clinically relevant relationship between a given gene and ASD.

Fig. 1.. Overlap between three sets of genes considered to be associated with ASD susceptibility.

At the time of the submission of the manuscript, SFARI GENE (https://gene.sfari.org) has scored 913 genes; of these, 144 genes received the highest score (Category 1; High Confidence). Iakoucheva et al. reported 106 ASD susceptibility genes, of which 82 have been evaluated by SFARI Gene to date. Yuen et al. reported 61 ASD susceptibility genes, all which have been evaluated by SFARI Gene to date. In both gene sets^, a third of genes did not yield the highest SFARI score for ASD association, and only a total of 30 genes were shared between these three gene sets.

Fig. 2.. Proposed pipeline.

Step 1, Evidence curation to evaluate strength of a gene-disease relationship based on publicly available genetic and experimental evidence. Select gene with reported association with ASD, regardless of additional phenotypic associations. We will start with genes with highest level of available evidence. Step 2, PHENOTYPE and GENOTYPE data: consists of two components, i) systematic evaluation and consensus rating of the quality of report of the ASD phenotypes in the evidence collected by the ClinGen procedure (Table 1), and ii) application of the ClinGen evaluation process for genetic evidence as outlined by Strande et al. Table 2). Step 3, LIST compilation: any NDD gene that scores as strong or definitive category for association with ASD will be included on the list.

Fig 3.. Systematic evaluation of quality of the ASD phenotype report.

Six phenotype experts independently scored eight published reports^- using this algorithm; results indicated 98% consistency regarding the rating of low, medium and high confidence in the reported ASD phenotype, and 90% consistency regarding the cognitive ability information. Proposed scoring adjustments: If default score is 2 (ClinGen default for de novo variant): low confidence (−1) and medium confidence (−0.5); if default score is 1.5 (ClinGen default for an inherited variant that is predicted/proven null): low confidence (−0.5) and medium confidence (−0.25); if default score is 0.5 (ClinGen default for an inherited variant not predicted/proven null, with some evidence of gene impact, e.g. missense variant with functional evidence supporting pathogenicity): low confidence (−0.25) and medium confidence (−0.1). For all default scores: Cognitive ability comments applied to all default scores: profound ID (c): case not counted towards curation; insufficient information regarding intellectual ability (d): scoring adjustments optional and decided by expert review. Note that these rules are tentative and may be adjusted based on increasing curation experience, but have been maintained consistently across all curations thus far. ADI-R, Autism Diagnostic Interview- Revised; ADOS, Autism Diagnostic Observation Schedule; DSM, Diagnostic and Statistical Manual of Mental Disorders; ICD, International Statistical Classification of Diseases; PDD, pervasive developmental disorder; PDD-NOS, pervasive developmental disorder-not otherwise specified.