Further Discussion
A recent review lists NTD genes that have been associated with NTDs in humans (M. Wang, Marco, Capra, & Kibar, 2019). Except for DVL2 , our study may appear not to corroborate findings in their review. We offer three explanations for this apparent discrepancy. First, the current study uniquely employs a gene-based mutational burden analysis. Adopting a similar methodology to those studies summarized in the review may yield similar results. Second, our subjects differ in ethnicity from many of the populations discussed in other studies. As the current study suggests, ancestry may influence which genes associate with myelomeningocele. Third, any variants that were not equally covered in both case and control populations were filtered out before analysis. If, for example, an important variant was discovered in the myelomeningocele subjects while the variant’s location was not covered in the corresponding gnomAD control population, that variant would not contribute to the mutational burden analysis.
The current study introduced 17 genes associated with risk for myelomeningocele with nominal significance (Fisher exact P values ≤ 0.05). Bonferroni correction for multiple comparisons was applied to each analysis using the number of compared genes as a denominator (173 in Hispanic comparisons and 189 in European ancestry comparisons) and 0.05 as the numerator alpha to calculate the correction threshold. While some genes came close, none of the P values fell below the correction threshold. A large enough group of myelomeningocele subjects would lend enough statistical power to achieve the strict statistical significance of a Bonferroni correction, but exome data for 511 myelomeningocele subjects is considerable, given the resource-intensive nature of gathering samples. The genes discussed above are suggested candidate genes.
Two of our quality control filters assume that each variant exists in Hardy-Weinberg equilibrium within the myelomeningocele population (inbreeding coefficient < -0.3, alternate allele depth ≥ 25%). However, if a selective pressure acts on a variant locus, that Hardy-Weinberg assumption is not met. Therefore, our analysis may exclude important variant loci that are under intense selective pressure.
A new approach was taken by limiting our analysis to a subset of human genes rather than evaluating the entire exome at once. The high concentration of current NTD candidate genes within the WNT signaling pathway prompted this focused approach. A more inclusive approach would also be valuable.
In summary, we report seventeen genes within the known WNT signaling pathways which may play a role in the development of myelomeningocele. As discussed above, the genes PORCN , DVL2 , CDH2 ,FUZ are already suspected to play a role in NTD development from studies in animal models, however the remaining thirteen genes reported here are new in their possible association with myelomeningocele.