Associations of 2923 plasma proteins with incident inflammatory bowel disease in a prospective cohort study and genetic analysis

Abstract

The prospective relationship between proteomics and inflammatory bowel disease (IBD) remains largely underexplored, presenting potential of therapeutic interventions. Using data from 48,800 IBD-free participants in the UK Biobank Pharma Proteomics Project (UKB-PPP), we assessed associations between 2923 plasma proteins and incident IBD risk using Cox analysis. Mendelian randomization (MR) meta-analysis, integrating cis-protein quantitative trait loci data from the UKB-PPP with IBD genome-wide association study data from the International Inflammatory Bowel Disease Genetics Consortium and FinnGen studies, identified causal proteins. Colocalization analysis strengthened evidence of shared common causal variants. Cohort study revealed associations of 673, 295, and 125 proteins with the risk of IBD, Crohn's disease (CD), and ulcerative colitis (UC), respectively. MR and colocalization analyses prioritized IL12B, CD6, MXRA8, CXCL9, IFNG, CCN3, RSPO3, and IL18 as having causal and high colocalization evidence with IBD or its subtypes. Our findings advance understanding of IBD's molecular etiology and highlight potential therapeutic targets.

Fig. 1. Schematic overview of the study design.

cis-pQTLs cis-protein quantitative trait loci, LD linkage disequilibrium, GWAS Genome-Wide Association Studies, IIBDGC International Inflammatory Bowel Disease Genetics Consortium, IBD inflammatory bowel disease, CD Crohn’s disease, UC ulcerative colitis, SNPs single nucleotide polymorphisms, MR Mendelian randomization, FDR false discovery rate, PPH4 posterior probability of Hypothesis 4.

Fig. 2. Volcano plots of individual protein associations with inflammatory bowel disease and its subtypes in UK Biobank.

Unadjusted any variable in estimating the hazard ratios of each protein for (a) inflammatory bowel disease (IBD), (d) Crohn’s disease (CD), and (g) ulcerative colitis (UC). Adjusted for age, sex, assessment center, education, employment status, household income, and TDI in estimating the hazard ratios of each protein for (b) IBD, (e) CD, and (h) UC. Additional adjusted for smoking status, alcohol consumption, PA, healthy diet, sleep during, and BMI in estimating the hazard ratios of each protein for (c) IBD, (f) CD, and (i) UC. Red, blue, and gray dots denote positive significant, inverse significant, and nonsignificant associations, respectively. FDR false discovery rate, HR hazard ratio. Source data are provided as Supplementary Data 1 and Source Data file.

Fig. 3. Observational analyses and MR analyses between candidate proteins and inflammatory bowel disease and its subtypes.

The forest plots showed the results of observational analyses and MR analyses between candidate proteins and (a) inflammatory bowel disease, (b) Crohn’s disease, and (c) ulcerative colitis. Cox regression was used to evaluate the association between baseline plasma protein and the incidence of IBD and its subtypes during study follow-up in UKB-PPP study (baseline participants: n = 48,800; incident IBD events: n = 336; incident CD events: n = 125; incident UC events: n = 232) adjusted for age, sex, assessment center, education, employment status, household income, TDI, smoking status, alcohol consumption, PA, healthy diet, sleep during, and BMI. The measure of centre and error bars in observational analyses presented the HRs and 95% CI of IBD and its subtypes risk increase associated with per 1-SD higher protein level. MR analysis, including Wald ratio or inverse variance weighted method, was used to evaluate the association between cis-pQTLs from UKB-PPP and IBD and its subtypes from IIBDGC (with results shown in green) and FinnGen (with results shown in purple). Meta-analysis was used to combine MR analysis results of IIBDGC and FinnGen (with results shown in red). The measure of centre and error barsin in MR analyses presented the ORs and 95% CI of IBD and its subtypes per 1-SD higher protein level for these proteins from in MR analyses. All tests were two sided and adjusted for multiple comparisons, with an FDR < 0.05 considered as significant. UKB-PPP UK Biobank Pharma Proteomics Project, IIBDGC International Inflammatory Bowel Disease Genetics Consortium, HR hazard ratio, OR odds ratio, CI confidence interval. Source data are provided as Supplementary Data 1, 9, and Source Data file.

Fig. 4. Single-cell type expression in colonic mucosa tissue in inflammatory bowel disease and healthy controls for the genes encoding the 22 candidate proteins.

a A total of nine cell types were identified in colonic mucosa tissue using graph-based clustering by the tSNE. b The t-SNE plot shows the expression of each candidate protein-coding gene across different cell types in the colon mucosa of IBD patients and healthy controls. c The bubble plot shows the expression proportion and levels of each candidate protein-coding gene across different cell types in the colon mucosa of IBD patients and healthy controls. d Cell-specific enrichment of each candidate protein-coding gene. e Comparison of each candidate protein-coding gene expression levels between IBD and health controls in the corresponding specifically enriched cell types. The Wilcoxon rank-sum test was used for analysis. All tests were two sided and adjusted for multiple comparisons. Log2FC > 0.25 and an FDR < 0.05 was considered as significant. The symbols * represent FDR < 0.05, ** represent FDR < 0.01, and *** represent FDR < 0.001. Source data are provided as Supplementary Data 16 and Source Data file.