Clinicians managing patients with Crohn’s disease frequently face significant challenges in tracking deep tissue recovery. Therefore, selecting an accurate transmural healing assessment tool is vital for predicting long-term therapeutic success. Specifically, standard magnetic resonance enterography (MRE) metrics often yield highly variable and inconsistent results. Consequently, a novel dual-center study has validated a machine-learning-optimized model named the SYSU-score. Indeed, this tool standardizes and significantly enhances the monitoring of transmural healing.
Limitations of Conventional Crohn’s Disease Criteria
First, researchers analyzed seven conventional MRE-defined transmural healing criteria. Specifically, they evaluated metrics like the Magnetic Resonance Index of Activity (MaRIA), sMaRIA, and the C-score. Subsequently, they tested these parameters in a large cohort of 467 active disease patients. Interestingly, these traditional criteria demonstrated highly variable rates of clinical attainment. For example, the simplified MaRIA (sMaRIA) achieved an attainment rate of 41.64%, whereas the classic MaRIA achieved only 23.75%. Furthermore, patients achieving healing via these different criteria showed varying rates of disease progression. Consequently, clinicians still struggled to establish a highly reliable standard for daily practice. Therefore, this limitation highlighted the urgent need for a more optimized predictive model.
The SYSU-score: A New Standard for Transmural Healing Assessment
To solve this, the investigators developed five machine-learning models using robust clinical data. Among these, the random forest (RF) model, named the SYSU-score, demonstrated the most favorable overall performance. Additionally, researchers validated this model in semi-external cohorts of patients receiving modern therapies like Ustekinumab and Upadacitinib. Specifically, the SYSU-score achieved an outstanding area under the curve of 0.83 in the Ustekinumab cohort. Similarly, it achieved an AUC of 0.80 in the Upadacitinib cohort, outperforming traditional scores. Moreover, the SYSU-score showed a remarkably strong association with a lower risk of disease progression. For instance, in the Ustekinumab cohort, SYSU-score-defined healing predicted disease progression with a low hazard ratio of 0.07. Consequently, this model offers a powerful, objective, and validated approach to monitoring therapeutic responses.
Clinical Implications for Gastroenterologists
Clearly, adopting this machine-learning model can revolutionize daily clinical decision-making for gastroenterologists. Currently, relying solely on mucosal healing might overlook deep transmural inflammation that drives future complications. However, implementing the SYSU-score allows clinicians to identify patients at a truly low risk of disease progression. As a result, physicians can make safer decisions regarding treatment de-escalation or maintenance. Furthermore, the score helps in customizing therapy for patients on biological agents like Ustekinumab or Upadacitinib. Ultimately, this objective tool can significantly reduce the rate of bowel damage and surgical resection. Therefore, Indian gastroenterologists should familiarize themselves with these machine-learning-optimized MRE frameworks.
Frequently Asked Questions
Q1: What is the SYSU-score in Crohn’s disease?
The SYSU-score is a machine-learning-optimized model based on random forest algorithms. Specifically, it standardizes magnetic resonance enterography (MRE) evaluations to assess transmural healing accurately.
Q2: Why is transmural healing assessment crucial in Crohn’s disease?
Transmural healing assessment is crucial because deep, full-thickness bowel healing yields significantly better outcomes. Indeed, it significantly reduces the risk of disease progression, hospitalization, and future surgery.
Q3: How does the SYSU-score compare to traditional MRE criteria?
The SYSU-score significantly outperforms traditional criteria like MaRIA, sMaRIA, and C-score. For instance, it demonstrated superior predictive accuracy in validation cohorts of patients treated with Ustekinumab and Upadacitinib.
References
- Zheng Q et al. Development and validation of the SYSU-score for MRI-based transmural healing assessment in Crohn’s disease: a dual-center study. Eur Radiol. 2026 Jun 19. doi: 10.1007/s00330-026-12701-6. PMID: 42319405.
- Rimola J, Colombel JF, Bressler B, et al. Magnetic Resonance Enterography Assessment of Transmural Healing with Vedolizumab in Moderate to Severe Crohn’s Disease: Feasibility in the VERSIFY Phase 3 Clinical Trial. Clin Exp Gastroenterol. 2024;17:9-23.
- Buisson A, et al. Transmural healing as a therapeutic goal in Crohn’s disease: a systematic review. Lancet Gastroenterol Hepatol. 2021;6(8):659-667.