Genome-wide association study identifies distinct genetic contributions to prognosis and susceptibility in Crohn's disease

Abstract

For most immune-mediated diseases, the main determinant of patient well-being is not the diagnosis itself but instead the course that the disease takes over time (prognosis). Prognosis may vary substantially between patients for reasons that are poorly understood. Familial studies support a genetic contribution to prognosis, but little evidence has been found for a proposed association between prognosis and the burden of susceptibility variants. To better characterize how genetic variation influences disease prognosis, we performed a within-cases genome-wide association study in two cohorts of patients with Crohn's disease. We identified four genome-wide significant loci, none of which showed any association with disease susceptibility. Conversely, the aggregated effect of all 170 disease susceptibility loci was not associated with disease prognosis. Together, these data suggest that the genetic contribution to prognosis in Crohn's disease is largely independent of the contribution to disease susceptibility and point to a biology of prognosis that could provide new therapeutic opportunities.

Figure 1. Within-cases GWAS identifies four loci that are associated with prognosis in CD.

(a) Plot of genome-wide association results. –log₁₀ (P values) from the Wald statistic (logistic regression model) are plotted against chromosomal position for the combined association analysis (n = 1,762 poor prognosis CD, 972 good prognosis CD). Each point represents a SNP. Dotted red line indicates genome-wide significance threshold. The four new loci identified in this study are indicated. (b – e) Characteristics of the associated genomic regions. Upper panel, chromosomal position; middle panel, – log₁₀ (P values) for individual SNPs at each locus (left y axis), rate of recombination indicated by red line (right y axis); lower panel, gene position within the locus (in Panel e only selected genes from class I and II regions are shown for clarity). SNPs are coloured according to LD with the most associated variant.

Figure 2. Association signal at the MHC is linked to the ancestral 8.1 haplotype

(a) The residual MHC association after conditioning on the lead SNP (rs9279411). (b) Allelic associations at class I and II HLA genes. For each allele, odds ratios (OR), 95% confidence intervals, and P values are shown. The significance threshold was corrected for the number of independent tests (P < 2.5 x 10^-4). Alleles that are components of the ancestral MHC AH8.1 haplotype are shaded blue (full AH8.1 annotation: HLA A*01:01, C*07:01, B*08:01, DRB1*03:01, DRB3*01:01, DQA1*05:01, DQB1*02:01). HLA DRB1*0103 (the strongest IBD susceptibility allele, ref 28) is included for comparison (shaded grey). (c) The residual MHC association after cross-conditioning on the lead HLA allele (HLA-B*08:01). In panels a and c the -log₁₀(P) of SNPs are plotted against chromosomal position. Only selected genes from class I and II regions are shown for clarity.

Figure 3. Pathway analysis implicates regulation of immune responses and mononuclear phagocytes in CD prognosis

(a) Horizontal bar plot of –log10 (P values) for the top 40 most enriched pathways in an analysis of 29 LD-pruned prognosis-associated SNPs (meta P < 1 x 10^-5) across 1,751 pathways annotated by Gene Ontology. (b) Bar plot of –log10 (P values) for 155 specific cell-types and conditions. Significant enrichment was observed in “monocyte-derived macrophages stimulated by M-CSF” and “monocyte-derived macrophages stimulated by M-CSF and interferon-gamma”. Other non-immune cell-types included neural, skin, lung, liver, stem cells, smooth muscle, stromal cells, bone marrow progenitors, endothelial and epithelial cells. Dotted lines represent Bonferroni-corrected significance thresholds. Analyses performed using SNPsea.

Figure 4. Distribution of CD susceptibility alleles does not differ between the prognostic subgroups

(a) “Box and Whiskers” plot of weighted Genetic Risk Scores between good prognosis and poor prognosis CD subgroups. Box represents median and interquartile range. Whiskers represent maximum and minimum values. Weights were based on beta coefficients calculated from odds ratios reported in Liu et al. (2015) for 170 CD susceptibility SNPs. n = 2,413 (b) Distribution of unweighted susceptibility allele counts between good prognosis and poor prognosis CD subgroups. Purple histogram bars represent the poor prognosis CD subgroup, yellow histogram bars represent the good prognosis CD subgroup. Statistical significance was assessed using unpaired two-tailed Student's t-test and was stratified for disease location.