Highlight
1. The IBDome project presents a large-scale, integrated multiomic and multimodal dataset encompassing 1002 patients with inflammatory bowel diseases (IBD) and controls.
2. It reveals distinct transcriptomic inflammatory signatures specific to Crohn’s disease and ulcerative colitis across gut regions.
3. A serum proteomic inflammatory protein severity signature was developed, correlating with gut molecular inflammation.
4. Advanced deep learning models accurately predict histologic disease activity and differentiate between Crohn’s disease and ulcerative colitis using H&E-stained tissue images.
Study Background
Inflammatory bowel diseases, primarily Crohn’s disease (CD) and ulcerative colitis (UC), are chronic, relapsing inflammatory disorders of the gastrointestinal tract characterized by heterogeneous clinical manifestations and variable responses to therapy. Despite advances, precision diagnostics and targeted treatment remain challenging due to the diseases’ complexity and interplay between genetics, immune response, and environmental triggers. Multiomic technologies, integrating molecular, histopathological, and clinical parameters, hold promise to delineate disease mechanisms, refine diagnostic accuracy, and personalize therapeutic regimens, yet comprehensive datasets that unite these data layers at scale are scarce.
Study Design
This multicohort, integrative study by the TRR241 IBDome Consortium involved 1002 clinically annotated participants, including patients diagnosed with Crohn’s disease, ulcerative colitis, and non-IBD control subjects. The study incorporated several high-dimensional modalities:
– Whole-exome sequencing of normal and inflamed gut tissue to capture genomic aberrations.
– RNA sequencing of corresponding tissues to detail transcriptomic profiles and inflammation-associated gene expression.
– Serum proteomics profiling to assess circulating inflammatory proteins.
– Histopathological analysis using whole-slide imaging of hematoxylin and eosin (H&E)-stained tissues.
Endpoints included identification of molecular signatures distinguishing disease subtypes and inflammation states, development of a serum-based biomarker panel reflecting intestinal inflammation severity, and deployment of deep learning algorithms for automated histologic scoring and subtype classification.
Key Findings
Distinct Transcriptomic Signatures Across Disease Types and Sites
The study identified unique inflammatory gene expression profiles in Crohn’s disease and ulcerative colitis, showing site-specific differences between normal and inflamed gut tissues. These transcriptomic signatures underline differential pathophysiological mechanisms driving the diseases. Particularly, RNA sequencing revealed distinctive immune cell infiltration patterns and cytokine expression alterations pertinent to disease subtype and gut segment affected.
Serum Proteomic Inflammatory Severity Signature
By integrating serum proteomics data, researchers developed an inflammatory protein signature that accurately reflects molecular inflammation within intestinal tissues. This blood-based biomarker panel offers potential for less invasive disease activity monitoring, correlating strongly with tissue-level transcriptomic markers, and may guide personalized treatment strategies.
Deep Learning Enhances Histopathological Diagnosis and Disease Classification
Advanced foundation model-based deep learning algorithms were employed to analyze digital images of H&E-stained intestinal biopsies. These models demonstrated high accuracy in predicting histologic disease activity scores conventionally assessed by expert pathologists. Additionally, they reliably distinguished Crohn’s disease from ulcerative colitis based on subtle tissue morphological differences, potentially supporting clinical decision-making and reducing diagnostic variability.
Publicly Available Resource Fostering Research
The IBDome multiomic dataset has been made publicly accessible, providing an unprecedented resource to propel research exploring the molecular underpinnings, diagnostic biomarkers, and therapeutic targets in IBD.
Expert Commentary
This comprehensive multiomic atlas exemplifies the transformative potential of integrative approaches in inflammatory bowel diseases. By combining genomic, transcriptomic, proteomic, and histopathological data with advanced computational analyses, the consortium has advanced understanding of disease heterogeneity and its manifestations. The identification of site-specific transcriptomic differences challenges the traditional concept of IBD as a uniform disease entity, underscoring the need for tailored therapies.
The serum proteomic signature offers a clinically convenient biomarker for real-time monitoring, which if validated prospectively, could mitigate the reliance on invasive endoscopic assessment. The successful application of deep learning to histopathology heralds a move towards AI-augmented pathology, potentially standardizing disease activity scoring and subtype classification, which have historically suffered from interobserver variability.
However, some limitations deserve attention. While multiomics integration is powerful, translating complex datasets into routine clinical algorithms requires careful validation in diverse populations. Furthermore, the cross-sectional design limits insight into longitudinal disease evolution and treatment response. Future studies should incorporate prospective cohorts and interventional trials to establish causal links and therapeutic implications. Additionally, the deep learning models necessitate external validation across varied histopathological preparation standards.
Conclusion
The IBDome study represents a landmark effort in establishing a comprehensive, multidimensional atlas of inflammatory bowel diseases, leveraging cutting-edge molecular profiling and computational pathology. This integrative approach elucidates distinct molecular and histological signatures that differentiate Crohn’s disease from ulcerative colitis and reflect underlying inflammation severity. The emergent serum biomarker panel and AI-powered histologic assessment tools promise to enhance disease monitoring and diagnostic precision. Importantly, the publicly available dataset lays the groundwork for future innovations in personalized IBD management, advancing toward a precision medicine paradigm. Clinical adoption awaits further prospective validation and real-world implementation studies.
Funding
This work was supported by the TRR241 IBDome Consortium and associated academic and clinical institutions. Detailed funding disclosures are available in the original publication.
References
1. Plattner C, Sturm G, Kühl AA, et al. IBDome: An Integrated Molecular, Histopathological, and Clinical Atlas of Inflammatory Bowel Diseases. Gastroenterology. 2026 Jun 10. PMID: 42269946.
2. Neurath MF. Current and emerging therapeutic targets for IBD. Nat Rev Gastroenterol Hepatol. 2017;14(5):269-278.
3. Jairath V, Feagan BG. Management of moderate to severe ulcerative colitis: a review of recent developments. Gut. 2021;70(5):1063-1071.
4. Becker C, Neurath MF. Novel insights into pathogenesis and treatment of IBD. J Clin Invest. 2022;132(12):e159596.
5. Kather JN, Calderaro J, Bankhead P, et al. Deep learning can predict microsatellite instability directly from histology in gastrointestinal cancer. Nat Med. 2019;25(7):1054-1056.
6. Atreya R, Neurath MF. Signaling molecules in IBD pathogenesis. Expert Rev Clin Immunol. 2021;17(5):439-450.
