Plasma Proteomics for Risk Prediction and Therapeutic Target Discovery in Crohn’s Disease and Ulcerative Colitis
Inflammatory Bowel Diseases, 2025
Gan X., Zhang Y., Zhang Y., Ye Z., Yang S., Xiang H., Huang Y., Wu Y., Zhang Y., Qin X.
| Disease area | Application area | Sample type | Products |
|---|---|---|---|
Immunological & Inflammatory Diseases | Pathophysiology Patient Stratification | Plasma |  Olink Explore 3072/384 |
Abstract
Objectives
We aimed to identify plasma proteins associated with incident Crohn’s disease (CD) and ulcerative colitis (UC), develop and validate predictive models for CD and UC risk, and uncover novel protein-based drug targets.
Methods
The study included 46 523 participants from England in the UK. Biobank as the development set and 47 105 participants for internal replication. An external validation set comprised 5807 participants from Scotland and Wales. Plasma proteomic profiling was performed on 2911 proteins.
Results
In the development set, 49 and 34 proteins were significantly associated with incident CD and UC risk, respectively (Bonferroni P < .05). These findings were replicated in the internal replication set. Two-sample Mendelian randomization (MR) analysis identified three proteins (TIMP1, TNFRSF10A, and LTBR) with causal associations for CD and four proteins (CCL20, OSM, NOS2, and CD300E) for UC. Among these, TIMP1 and CD300E represent novel, undrugged targets, while the remaining five are currently druggable. The proteomic-based model, incorporating age, sex, and candidate proteins, demonstrated strong predictive performance in the external validation set, with a C-index of 0.94 (95% CI, 0.88–1.00) for CD and 0.82 (95% CI, 0.73–0.92) for UC. Integrating candidate proteins or the top 10 proteins into clinically based models significantly enhanced risk prediction for both CD and UC.
Conclusions
This study identifies novel plasma protein associations with CD and UC, supported by genetic evidence, and highlights their potential as therapeutic targets. Plasma proteomics significantly improves risk prediction for incident CD and UC compared to traditional clinical models, offering new avenues for drug discovery and personalized risk assessment.