Y-ECCO Literature Review: Lotte Oldenburg
Preclinical protein signatures of Crohn’s disease and ulcerative colitis: A nested case-control study within large population-based cohorts
Grännö O, Bergemalm D, Salomon B, et al.
Gastroenterology 2025;168:741–53. doi: 10.1053/j.gastro.2024.11.006
Introduction
Inflammatory Bowel Disease (IBD), encompassing Crohn’s Disease (CD) and Ulcerative Colitis (UC), often manifests after a prolonged subclinical phase. Identification of biomarkers that predict disease onset could revolutionise early diagnosis and intervention. Proteomics provides insights into functional molecular processes by measuring proteins, the key effectors of biological processes, which are often not reflected at the transcriptomic level. Despite its potential, proteomics has been an underexplored molecular layer in IBD research compared to the more commonly used transcriptomic approaches. Grännö et al. conducted a nested case-control study within large Swedish population-based cohorts to identify preclinical protein signatures predictive of future CD and UC diagnoses.
Methods and key findings
This case-control study used four cohorts (n≥180,000) to identify and validate preclinical protein signatures in blood from healthy individuals who later in life were diagnosed with IBD (preclinical cases). These preclinical cases were compared to individuals who remained disease free (controls). The study included a discovery cohort [Malmö diet and cancer (MDC) cohort], a validation cohort [Northern Sweden Health and Disease Study (NSHDS)], an inception cohort of newly diagnosed IBD patients [Swedish Inception Cohort (SIC IBD)] and a twin cohort (TwinGene study) to assess the influence of genetic and shared environmental factors.
A total of 178 proteins were measured using the Olink platform, focusing on inflammatory and cancer-related markers. Data preprocessing involved pairwise mean normalisation to correct for batch effects.
Proteins associated with a future diagnosis of IBD
In the discovery cohort (MDC), 34 proteins were associated with a future CD diagnosis, of which nine were validated in the validation (NSHDS) cohort. Six proteins were upregulated and three were downregulated. The proof-of-concept analysis showed that six of the protein markers (CXCL9, IL6, MMP-10, CCL20, DNER, CX3CL1) were significantly altered. However, upon assessment of correlations between protein levels and time to diagnosis by linear correlation analysis, only the downregulated proteins (DNER, GPNMB, CX3CL1) showed significant negative correlations.
For UC, 45 proteins in the discovery cohort were associated with a future UC diagnosis. The associations were confirmed for 12 proteins in the validation (NSHDS) cohort. These 12 were also significantly upregulated in patients with newly diagnosed UC from the inception (SIC IBD) cohort. However, only MMP-10 demonstrated a correlation with time to diagnosis of UC.
Protein signatures predictive of a future diagnosis of IBD
Regularised logistic regression was performed in the preclinical discovery cohort (MDC) to identify a 29-protein signature, which predicted future CD diagnosis with a high accuracy: an AUC of 0.85 in the discovery cohort (MDC) and 0.87 in the validation cohort (NSHDS), with a sensitivity of 77% and a specificity of 87% at the optimal cutoff. In the preclinical cohorts, the performance of the logistic regression model increased toward diagnosis. A high predictive capacity (AUC 0.82) was still observed when the analysis was restricted to >16 years before diagnosis. Evaluation of the model in the newly diagnosed CD cohort (SIC IBD) resulted in a high AUC of 1.00.
For UC, the predictive model had an AUC of 0.77 in the preclinical discovery (MDC) cohort and 0.67 in the validation (NSHDS) cohort. The model performed better in older patients (AUC 0.79) than in younger patients (AUC 0.55). In the inception cohort, the logistic regression model performed well (AUC 0.95); however, it was difficult to differentiate between CD and UC in newly diagnosed patients.
Impact of genetic and shared environmental factors on predictive biosignatures
To assess the influence of genetic and environmental factors on the predictive protein markers, a comparative analysis was performed on twins who later developed IBD (preclinical CD) and matched healthy twins (external twin analysis). The same preclinical CD twins were then compared with their healthy twin siblings. In CD, the model yielded an AUC of 0.89 in the external twin analysis, but the AUC fell to 0.58 when comparison was with the healthy co-twin (p=0.04), probably due to genetic and shared environmental factors. It is to be noted that only ten twins with preclinical CD were included in this analysis.
For UC, the reduction in AUC, from 0.74 to 0.58 (p=0.17), was relatively minor, suggesting a limited impact of genetic and shared environmental factors on the logistic regression protein signature.
Discussion
This study demonstrates that specific protein signatures can predict the onset of UC and CD years before diagnosis. The findings are particularly robust for CD, with predictive models maintaining high accuracy even in samples collected over a decade prior to diagnosis. The authors used multiple population-based cohorts with high heterogeneity, introducing high generalisability of their protein signatures. Of note, the high median age at diagnosis (>40 years old) could influence the predictive factors and hamper comparability with other studies (PREDICTS, GEM) [1, 2].
Results from the twin cohort indicate that genetic and environmental risk factors may contribute to the predictive signature of future CD, but this did not appear to be the case for UC.
Conclusion
Grännö and colleagues have provided compelling evidence of the existence of preclinical protein signatures in IBD, particularly in CD. These findings pave the way for further research into early detection and preventive strategies.
References
- Torres J, Petralia F, Sato T, et al. Serum biomarkers identify patients who will develop inflammatory bowel diseases up to 5 years before diagnosis. Gastroenterology, 2020;159:96–104. https://doi.org/10.1053/j.gastro.2020.03.007
- Leibovitzh H, Lee S-H, Raygoza Garay JA, et al. Immune response and barrier dysfunction-related proteomic signatures in preclinical phase of Crohn's disease highlight earliest events of pathogenesis. Gut 2023;72:1462–71. https://doi.org/10.1136/gutjnl-2022-328421
Profile
Lotte Oldenburg is a medical doctor and works currently as a PhD-fellow at the Gastroenterology and Hepatology department of the Amsterdam University Medical Center in The Netherlands. Her research interests include postoperative recurrence in Crohn’s Disease and the prediction of therapy response in IBD patients.