Innovative computational approaches shed light on genetic mechanisms underlying cognitive impairment among children born extremely preterm

Abstract

Background: Although survival rates for infants born extremely preterm (gestation < 28 weeks) have improved significantly in recent decades, neurodevelopmental impairment remains a major concern. Children born extremely preterm remain at high risk for cognitive impairment from early childhood to adulthood. However, there is limited evidence on genetic factors associated with cognitive impairment in this population.

Methods: First, we used a latent profile analysis (LPA) approach to characterize neurocognitive function at age 10 for children born extremely preterm. Children were classified into two groups: (1) no or low cognitive impairment, and (2) moderate-to-severe cognitive impairment. Second, we performed TOPMed-based genotype imputation on samples with genotype array data (n = 528). Third, we then conducted a genome-wide association study (GWAS) for LPA-inferred cognitive impairment. Finally, computational analysis was conducted to explore potential mechanisms underlying the variant x LPA association.

Results: We identified two loci reaching genome-wide significance (p value < 5e-8): TEA domain transcription factor 4 (TEAD4 at rs11829294, p value = 2.40e-8) and syntaxin 18 (STX18 at rs79453226, p value = 1.91e-8). Integrative analysis with brain expression quantitative trait loci (eQTL), chromatin conformation, and epigenomic annotations suggests tetraspanin 9 (TSPAN9) and protein arginine methyltransferase 8 (PRMT8) as potential functional genes underlying the GWAS signal at the TEAD4 locus.

Conclusions: We conducted a novel computational analysis by utilizing an LPA-inferred phenotype with genetics data for the first time. This study suggests that rs11829294 and its LD buddies have potential regulatory roles on genes that could impact neurocognitive impairment for extreme preterm born children.

Keywords: Cognitive impairment; Genetic mechanisms; Genome-wide association study (GWAS); Latent profile analysis (LPA); Neurodevelopment; Preterm children.

Fig. 1

Manhattan plot. The Manhattan plot visualizes the association of SNPs along the genome with the LPAx trait. X-axis represents genomic location and y-axis represents -log10(p value). Each dot represents a SNP tested. SNPs above the red horizontal line, which marks the 5 × 10⁻⁸ are considered genome-wide significant. This plot was generated using the R package karyoploteR [57]. NCBI build 38

Fig. 2

QQ plot. A quantile-quantile (Q-Q) plot is used to characterize the extent to which the observed distribution of the test statistics follows the expected null distribution. This plot was generated using the R package qqman [58]

Fig. 3

Locus zoom plots for the two genome-wide significant loci. Colors represent linkage disequilibrium r² values calculated from TOPMed individuals with the lead SNP in each plot. a Locus zoom plots with linkage disequilibrium r² values calculated from TOPMed European ancestry individuals. b Locus zoom plots with linkage disequilibrium r² values calculated from TOPMed African ancestry individuals. NCBI build 38

Fig. 4

Virtual 4C plots. Centered at a rs79453226 b rs12322215 in adult cortex and fetal cortex. The bin containing the anchor position is indicated as a thick grey vertical bar. Different genes or regions can be highlighted in yellow. On the top is gene expression data with gene locations. Each gene is indicated by an arrow pointing the direction of transcription. The start site is indicated by the tail of the arrow. Each gene is labeled by its common name and highlighted in red indicating the expression level: the deeper the red color the higher the expression. On the bottom is the chromatin interaction Hi-C data that is plotted as a virtual 4C plot with the given anchor position. The black line shows the observed counts, the red line shows the expected counts, and the blue line shows the -log10(p value). The range of the -log10(p value) is plotted on the y-axis on the right while the range of the count data is shown on the left. The x-axis is the genomic location in Mb. NCBI build 37