Partial duplication and deletion of neurodevelopmental genes and their clinical interpretation

Chromosomal microarrays are routinely used to detect copy number variants (CNV) genome wide in patients with neurodevelopmental disorders including intellectual disability, learning disabilities, and autism spectrum disorder. For these conditions, the diagnostic yield is between 15-20% (Miller et al. 2010). Gene dosage, a gene fully deleted or duplicated, has clear molecular consequences with a significant decrease or increase in transcription respectively. CNVs encompassing complete neurodevelopmental (ND) genes known to be dosage sensitive are mostly observed de novo, albeit less frequently in duplications. However, CNVs involving only a portion of a known ND gene are frequently observed in the diagnostic setting and whether these variants are gene disrupting is difficult to establish, especially for duplications. Our aim is to quantify the deleteriousness of partial deletions and duplications involving known ND genes. We used a list of 614 ND genes based on literature (McRae JF et al. 2015, Coe et al. 2019, Satterstrom et al. 2020), panelapp.genomicsengland.co.uk and clinicalgenome.org. We studied parental transmission of CNVs in 8113 individuals with microarrays performed at the CHU Sainte-Justine diagnostic laboratory. We compared the proportion of inherited and de novo CNVs by Fisher's test. The analysis was replicated in the DECIPHER database. Partial duplications of ND genes (n=38) almost never occurred de novo (3%) and this was not significantly different from the frequency of de novo partial duplications observed in non-ND genes (1%), which is also the probability that a neutral CNV is observed de novo. On the other hand, the proportion of duplications fully including a ND gene observed de novo was significantly higher (43%, p= 2.68 × 10^-5); also higher (p=1.07 × 10^-10) than the complete duplications of non-ND genes (14%). On the other hand, the proportion of de novo complete (80%) and partial (71%) deletions involving ND genes were both very high and not significantly different from one another but their comparison with the complete and partial deletions of non-ND genes was statistically significant (respectively p= 1.72 × 10^-26 and p=1.91 × 10^-12). Partial deletion of ND genes are also more de novo than partial duplication of ND genes (p= 1.08 × 10^-9). These results were replicated in the DECIPHER database. Complete CNVs of ND genes are more often de novo which suggest that haploinsufficient genes are also triplosensitive. Complete and partial deletion of ND genes have a similar pattern of inheritance, we conclude that the mechanism of action is probably the same and is by loss of function. The partial duplications of ND genes are rarely de novo with a rate similar to the probability of a neutral CNV occuring de novo and their transmission is similar to the non-ND genes. As opposed to partial deletions, partial duplications appear to be benign in the majority of the cases. This could be explained by the fact that the duplicated segment is in tandem and do not disturb the function of the gene or the segment is located elsewhere in the genome and doesn't disrupt the function of a gene. In conclusion, the partial duplications of ND genes can be interpreted as benign in most cases and clinicians should pursue additional genetic testing in order to explain the phenotype of the patient.