Genetic variants for head size share genes and pathways with cancer

Abstract

The size of the human head is highly heritable, but genetic drivers of its variation within the general population remain unmapped. We perform a genome-wide association study on head size (N = 80,890) and identify 67 genetic loci, of which 50 are novel. Neuroimaging studies show that 17 variants affect specific brain areas, but most have widespread effects. Gene set enrichment is observed for various cancers and the p53, Wnt, and ErbB signaling pathways. Genes harboring lead variants are enriched for macrocephaly syndrome genes (37-fold) and high-fidelity cancer genes (9-fold), which is not seen for human height variants. Head size variants are also near genes preferentially expressed in intermediate progenitor cells, neural cells linked to evolutionary brain expansion. Our results indicate that genes regulating early brain and cranial growth incline to neoplasia later in life, irrespective of height. This warrants investigation of clinical implications of the link between head size and cancer.

Keywords: cancer; genetics; genome-wide association study; head circumference; head size; intracranial volume; meta-analysis.

Graphical abstract

Figure 1

Genome-wide association studies on human head size (A) Circos Manhattan plot of the European ancestry head size GWAS, with gray lines corresponding to genome-wide significant (p < 5 × 10⁻⁸) or sub-significant (p < 1 × 10⁻⁶) p value thresholds. Known variants are in blue, novel ones in red. For each lead variant, the nearest gene is presented, with the color corresponding to its position to the lead variant: exonic (red), 3′-UTR (green), intronic (blue), intergenic including up- and downstream, exonic and intronic non-coding RNA (gray). Nearest genes for more than one locus are denoted with an asterisk (∗). (B) Circos heatmap showing the betas of lead variants in African, Asian, and European ancestry meta-analyses, as well as the transancestral meta-analysis. Differences between the height-unadjusted (model 1) and -adjusted (model 2) meta-analysis are also shown. (C) Bar plot of the genetic correlation coefficient (ρ_genetic) of the height-unadjusted and -adjusted head size GWAS with the height GWAS, with their accompanying 95% confidence intervals.

Figure 2

Genetic loci for head size and effects on regional brain volumes (A) Heatmap showing head size loci that overlap with previously identified loci for global brain volumes (red), subcortical volumes (blue), and cortical region of interest volumes (green). (B) UpSet plot of associations between head size lead variants and brain volumes. Intersection size corresponds to the frequency of the combination depicted below the bar. Set size corresponds to the frequency of associations with one of the brain volume categories (i.e., global, subcortical, or cortical). (C) Plot showing the subcortical shape analysis of rs111939932 using log Jacobian determinants. Colors correspond to t values, with positive associations depicted in blue, and negative ones in red. Letters point to different subcortical structures: a, putamen; b, pallidum; c, caudate; d, amygdala; e, hippocampus; f, thalamus; g, accumbens.

Figure 3

Gene sets enriched in human head size loci (A) Bar plots presenting enriched KEGG gene sets. –log₁₀ of adjusted p value and proportion of nearby genes overlapping with the gene set are presented. Cancer gene sets are depicted in pink, cell growth and death gene sets in yellow-green, and signal transduction gene sets in turquoise. (B) Network graph showing enriched KEGG gene sets and their included genes near genetic lead variants. Gene sets are shown in squares with arrows to overlapping genes. Colors correspond to gene set categories: only cancer gene sets (pink), only cell growth and death gene sets (yellow-green), only signal transduction gene sets (turquoise), cancer gene sets and cell growth and death gene sets (dark blue), cell growth and death and signal transduction gene sets (green), or all three gene set categories (orange). Sphere size corresponds to the number of gene sets linked to that gene. (C) Schematic overview of enriched signaling pathways with proteins encoded by genes near (<10 kb) identified genetic loci. Proteins encoded by these genes are colored (green, ErbB pathway; red, p53 pathway; blue, Wnt pathway), other proteins are depicted in gray. Circles next to protein names provide the locus number of the encoding gene. Locations of lead variants and variants in LD (r² > 0.6) are shown in squares next to the proteins: exonic (e; red), 3′-UTR (3′; green), 5′-UTR (5; light green), intronic (i; blue), intergenic including up- and downstream, exonic and intronic non-coding RNA (g; gray). For Frizzled, not only FZD2 but also FRZB is taken into consideration.

Figure 4

Gene enrichments stratified by distance from head size lead variants (A) Enrichment of OMIM macro- and microcephaly genes and COSMIC tier 1 genes near identified genetic loci. Depicted are enrichments of genes within 1 Mb (orange), 100 kb (purple), or 10 kb (pink) of identified genetic loci, genes with intragenic genetic variants (light green) and genes with intragenic genetic lead variants (yellow) in comparison with genes in the reference genome (dark green). ∗p < 0.05; ∗∗p < 0.0125 (0.05/4); ∗∗∗p < 0.0025 (0.05/4/5). (B) Violin plots showing DOMINO autosomal dominance scores of different gene sets. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001. (C) Venn diagram showing genes within 10 kb of genetic loci that overlap with OMIM microcephaly genes (yellow) or macrocephaly genes (green) or COSMIC cancer tier 1 genes (red). Genes with intragenic lead variants are depicted in black, others in gray. (D) Bar plot showing enrichments of gene sets for genes differentially expressed in neurons and progenitors. ∗p < 0.05; ∗∗p < 0.025 (0.05/2); ∗∗∗p < 0.003 (0.05/2/8). (E) Bar plots showing enrichments of gene sets for the various cell types in the human cortical brain using single-cell RNA-sequencing data. ∗p < 0.05; ∗∗FDR < 0.05; ∗∗∗p < 0.0007 (0.05/9/8).