Introduction

Biologic therapies are pivotal for treating moderate-to-severely active UC. Drug sustainability (DS) is a surrogate marker for drug efficacy, but predicting DS, particularly on a patient-by-patient basis, has proven challenging and tools are scarce. However, with the advent of machine learning (ML) and artificial intelligence (AI), it may be possible to overcome this obstacle and predict DS with greater accuracy. In this study, we aimed to externally validate an explainable AI-derived web-based tool we have developed for predicting patient level DS of infliximab and vedolizumab.

Aims & Methods

In this study, we aimed to externally validate an explainable AI-derived web-based tool we have developed for predicting patient level DS of infliximab and vedolizumab. This retrospective, multi-center, cohort study is based on data from three large tertiary medical centers in Israel; Rabin Medical Center (RMC), Rambam Healthcare Campus (RHC) and Sheba Medical Center (SMC). The RMC cohort was used for model development and internal validation and the RHC and SMC were used for separately for double external validation. We followed patients with moderate-to-severe UC treated with infliximab or vedolizumab for at least 54 weeks. DS was defined as corticosteroid-free, hospitalization-free, and surgical-free treatment with the drug. Using readily available clinical parameters at treatment baseline machine learning was used to generate two outcomes: a patient-specific prediction of DS over a period of one year, and patient-specific explainable DS survival curves using the top five predictive features identified by the ML algorithm.

Results

The study cohort included 313 patients. The RMC data set used included a total of 246 patients and was used to build the model while the SMC and RHC test sets which included 35 and 32 patients, respectively were used separately to externally validate the model developed using the RMC data set. There was no significant difference in overall drug survival between the 3 data sets. In all data sets most patients were biologic naïve and most patients were treated with vedolizumab (although absolute numbers approximated 50%). The overall performance of the model, as depicted by the area under the ROC curve, was 0.85. The model demonstrated good overall accuracy (62.8%-80%) and good model performance, with sensitivity, specificity, PPV, and NPV ranging from 25%-90.4%, 36.3%-50%, 70.1%-86.3%, and 33.3%-41.3%, respectively, on the two external test sets.

Conclusion

We present an externally validated novel explainable AI-based tool for prediction of DS of infliximab and vedolizumab in patients with moderate-to-severe UC, using patient level parameters readily available at treatment start. The tool could serve as a clinical decision-making aid in the context of precision medicine, helping clinicians to select the most appropriate treatment for their patients and optimize disease management. The explainability of the model, further increases the applicability of the tool in different healthcare settings.

Disclosure

No relevant conflict of interests.

Introduction

A decade ago, endoscopic surveillance (ES) for Barrett’s oesophagus (BO) was recommended to improve the outcome of oesophageal adenocarcinoma through early recognition and prevention¹. The yield of ES to detect dysplasia may depend on guideline adherence and quality of endoscopy^2,3,4.

Aims & Methods

The aim is to assess whether dedicated (expert) ES (DS) is more effective than conventional (non-dedicated) ES (NS) by means of Key Performance Indicators (KPI), Dysplasia Detection Rate (DDR), use of advanced imaging and complication rate.
This multicentre (5 UK hospitals) retrospective study includes real-world ES data from 1303 procedures conducted within the last 8 years. The endoscopy and histology reports and clinical notes of consecutive procedures were accessed. The data were collected in spreadsheets and then analysed with SPSS-23. Two services were reported demographically and compared with appropriate statistical tests and logistic regression to test factors associated with DDR.

Results

Data from 680 DS and 623 NS procedures were included, comprising 906 (69.5%) males and 397 (30.5%) females (M:F=2.3:1.0). The age ranged from 14-90 yrs with a mean of 63.7 (SD±12.5) yrs. The maximum length of the BO was 18 cm with a mean of 4.1 cm (SD±3.0). There were significant differences in all the KPIs (Prague classification, Barrett’s island description, hiatus hernia delineation, visible lesion documentation, Paris classification, Seattle protocol adherence and inspection time) favoring the DS (P<0.001). Other comparisons: narrow band imaging (NBI, P<0.001), acetic acid chromoendoscopy (AAC, P<0.001), sedation (P=0.02) and acquisition of target biopsies (, P<0.001) were also significantly higher in DS while the complication rate was significantly lower (P=0.05). Importantly, the DDR was significantly higher in DS (58/680 (8.5%)) than in NS (16/623 (2.6%)) with a P<0.001. Regression analysis showed: increasing age, male sex, lesion description and Paris classification documentation were associated with higher DDR (Table 1), while other indicators showed no significant association.

Table 1. Factors associated with dysplasia detection

Conclusion

Surveillance for BO through a dedicated service with high quality endoscopy can produce a higher diagnostic yield with improved DDR and a lower complication rate. Given these benefits, large-scale prospective studies are warranted to assess the true impact on mortality, morbidity, cost benefit and patient-reported outcomes, ensuring that surveillance service provides a meaningful clinical benefit, particularly with risk stratification concerning the factors associated with disease progression.

References

1. Fitzgerald et al. (2013). BSG guidelines on Barrett’s oesophagus. Gut, 63(1), 7–42. doi:10.1136/gutjnl-2013-305372.
2. Britton et al. (2019). Improving accuracy in Barrett’s surveillance. Frontline Gastroenterol, 10(2), 128–134. doi:10.1136/flgastro-2018-101019.
3. Ratcliffe et al. (2023). Surveillance service improves dysplasia detection. Frontline Gastroenterol, 15(1), 21–27. doi:10.1136/flgastro-2023-102425.
4. Ratcliffe et al. (2023). Dedicated Barrett’s surveillance vs. standard care. iGIE, 3(1), 58–65. doi:10.1016/j.igie.2023.11.003.

Introduction

Immune checkpoint inhibitors (ICI) are widely used in the management of numerous solid tumors and act by enhancing T-cell–mediated anti-tumour responses. While this activation boosts the immune response against cancer cells, it may also result in immune-related side effects (irAEs). Gastrointestinal symptoms, mainly colitis and diarrhea, are among the most common irAEs.¹ We have recently (February 2025) opened a dedicated clinic, within our Gastroenterology outpatient service, to manage patients who experience gastrointestinal irAEs during ICI therapy for epithelial cancers.

Aims & Methods

Our aim was to evaluate characteristics and outcomes of patients referred to our dedicated clinic for the management of gastrointestinal irAEs associated with ICI therapy since its start (February 2025). All patients were clinically evaluated and were treated with standard-of-care therapy in accordance to European Guidelines². Patients were followed up for at least 8 weeks following initiation of therapy. We collected data on patient demographics (gender and age), patient’s overall health status as assessed by the Charlson Comorbidity Index (CCI)³, cancer type, oncological therapy, diarrhea severity (graded via CTCAE)⁴, time to onset of diarrhea following the initiation of oncological therapy, interruption of ICI, outcomes of colonscopy (if performed), treatment modality for ICI-related gastrointestinal toxicity, time to resolution of diarrhea (in weeks), and whether ICI therapy was subsequently resumed.

Results

We included 14 patients affected by ICI-related diarrhea or colitis (grade 1: n=2 , grade 2: n=9 , grade 3: n=2; grade 4: n=1). Four patients were affected by endometrial cancer, 3 patients melanoma, 3 lung cancer, 2 breast cancer, 1 ovarian cancer, 1 cervical cancer. Thirteen patients were treated with PD1/PDL1- inhibitors (Pembrolizumab: n=9 , Dostarlimab: n=2, Nivolumab: n=2) and one patient was treated with CTLA-4 (Ipilimumab: n=1). Ten patients interrupted immunotherapy due to ICI-related diarrhea after 17 weeks of therapy on average. As indicated by the European Guidelines², all patients with diarrhea grade G2 or higher (6 patients) underwent colonoscopy, which revealed signs consistent with ICI-induced colitis. After standard-of-care treatment,12 patients experienced diarrhea grade 1 or lower. Mean time to resolution of diarrhea was 4.2 weeks. Among the 10 patients who had to interrupt ICI due to gastrointestinal adverse events, 5 patients (50%) resumed ICI after a mean of 4 weeks. No patients experienced any adverse events related to the standard-of-care treatment as defined by CTCAE.⁴

Conclusion

Despite being preliminary, our experience shows that patients managed in a dedicated clinic for gastrointestinal irAEs associated with ICI therapy experience acceptable outcomes. Future studies, aimed at comparing whether this approach is more effective than a non-dedicated management, are warranted.

References

1. Chan KK, Bass AR. Autoimmune complications of immunotherapy: pathophysiology and management. BMJ. Published online April 6, 2020:m736. doi:10.1136/bmj.m736
2. Haanen J, Obeid M, Spain L, et al. Management of toxicities from immunotherapy: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol. 2022;33(12):1217-1238. doi:10.1016/j.annonc.2022.10.001
3. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987;40(5):373-383. doi:10.1016/0021-9681(87)90171-8
4. Freites-Martinez A, Santana N, Arias-Santiago S, Viera A. Using the Common Terminology Criteria for Adverse Events (CTCAE - Version 5.0) to Evaluate the Severity of Adverse Events of Anticancer Therapies. Actas Dermosifiliogr. 2021;112(1):90-92. doi:10.1016/j.ad.2019.05.009

Disclosure

G.I. has received personal fees for acting as speaker for Biocodex, Danone, Sofar, Malesci, Metagenics and Tillotts Pharma, and for acting as consultant and/or advisor for Ferring Therapeutics, Giuliani, Malesci and Tillotts Pharma. A.G. reports personal fees for consultancy from Eisai Srl, 3PSolutions, Real Time Meeting, Fondazione Istituto Danone, SinergieSrl, Board MRGE and Sanofi SpA personal fees for acting as a speaker for Takeda SpA, AbbVie and Sandoz SpA and personal fees for acting on advisory boards for VSL3 and Eisai. G.C. has received personal fees for acting as advisor for Ferring Therapeutics. All other authors have no conflicts of interest to disclose.

Introduction

The management of inflammatory bowel disease (IBD) often relies on colonoscopy to evaluate disease activity and detect complications. However, intestinal ultrasound (IUS) has emerged as a non-invasive, accurate alternative, endorsed by ECCO-ESGAR guidelines for use in both Crohn’s disease and ulcerative colitis. This study aimed to evaluate the impact of IUS implementation on the frequency of colonoscopies performed in IBD patients at a tertiary care hospital.

Aims & Methods

We conducted a retrospective analysis comparing two time periods: before IUS implementation (2021–2022) and after its introduction (2023–2024). The number of colonoscopies performed for IBD management in each period was recorded. Screening colonoscopies in IBD patients were not included in the analysis; therefore, the data reflect colonoscopies performed specifically for the clinical management of the disease. The percentage change in colonoscopy utilization was calculated to assess the effect of IUS integration into routine clinical practice.

Results

In the pre-IUS period (2021–2022), a total of 340 colonoscopies were performed for the management of patients with IBD. After the introduction of IUS in 2023, the number of colonoscopies over the subsequent two years decreased to 280, reflecting a 17.6% reduction. This decline suggests a meaningful shift towards non-invasive monitoring facilitated by IUS.

Conclusion

The integration of IUS into the endoscopic workflow was associated with a notable decrease in colonoscopy use among IBD patients. These findings reinforce the role of IUS as a valuable, non-invasive modality for disease monitoring, with the potential to optimize resource utilization and support faster clinical decision-making. Prospective studies are needed to further assess the long-term benefits and cost-effectiveness of widespread IUS adoption in IBD care.

References

1. Panés J, Bouzas R, Chaparro M, et al., Imaging techniques for assessment of inflammatory bowel disease: Joint ECCO and ESGAR evidence-based consensus guidelines. Journal of Crohn's and Colitis. 2013;7(7):556–585. doi: 10.1016/j.crohns.2013.01.206
2. Smith RL, Taylor SA, Horsthuis K, et al., Reduction in the use of endoscopy after implementing an intestinal ultrasound service: A retrospective single-centre study. Journal of Crohn’s and Colitis. 2024;18(Suppl 1):i1344. doi: 10.1093/ecco-jcc/jjad212.1124
3. Kucharzik T, Wittig BM, Helwig U, et al., Intestinal ultrasound for monitoring of Crohn's disease activity: A prospective multicenter study. Clinical Gastroenterology and Hepatology. 2023;21(6):1397–1406. doi: 10.1016/j.cgh.2022.12.005
4. Bryant RV, Jairath V, Kucharzik T, et al., AGA Clinical Practice Update on the Role of Intestinal Ultrasound in the Management of Inflammatory Bowel Disease: Expert Review. Clinical Gastroenterology and Hepatology. 2024. [In press] doi: 10.1016/j.cgh.2024.03.014
5. Zorzi F, Bezzio C, Mazza S, et al., The role of intestinal ultrasound in ulcerative colitis: A systematic review. World Journal of Gastrointestinal Endoscopy. 2023;12(3):97210. doi: 10.4253/wjge.v12.i3.97210

Introduction

Until now, there have only been conflicting reports regarding whether adenomas located proximal to the splenic flexure have increased association with metachronous colorectal cancer (CRC) compared to distal adenomas, and whether adenoma location should affect surveillance practice.

Aims & Methods

To address this knowledge gap, we aimed to compare metachronous colorectal cancer (CRC) and advanced adenoma (AA) risk in patients with baseline proximal vs. distal adenomas, including consideration of metachronous CRC location, baseline proximal adenomas only vs. both proximal and distal adenomas, and baseline proximal non-advanced adenoma vs. AA.
We systematically searched Pubmed, EMBASE, and Cochrane for cohort studies and clinical trials describing CRC or AA incidence at surveillance stratified according to baseline adenoma’s location. Proximal adenomas were defined as those located from the cecum to the splenic flexure. We performed a meta-analysis by calculating pooled incidence rate ratios (IRR) and corresponding 95% confidence intervals (CI) for CRC and AA using a random-effects model. Heterogeneity was assessed with the I2 statistic.

Results

The literature search yielded 5,907 studies of which 13 were included, with a total of 205,207 patients (mean age 62.3 years, 63.8% males) and a mean duration of follow-up of 8.5 years. Compared with patients with only distal adenomas, patients with any proximal adenoma and only proximal adenomas had a significantly higher risk of metachronous CRC [1.70 (95%CI: 1.17-2.46; I2=73%), and 1.97 (95%CI: 1.41-2.75; I2=0%)], respectively. Patients with any proximal AA had also a higher risk of CRC compared to those with only distal AA [2.77 (95%CI 1.77-4.34; I²=38%)]. Also, patients with any proximal adenoma had an even higher risk of proximal CRC compared to patients with baseline only distal lesions [3.65 (95%CI 1.57-8.47; I2=86%)]. Finally, patients with any baseline proximal adenoma had also a higher risk of metachronous AA [1.82 (95%CI: 1.41-2.34; I2=59%)].

Conclusion

Our findings suggest that baseline adenoma proximal location is associated with future detection of CRC and AA. Better understanding of the reasons behind this increased CRC risk, and particularly whether these can be overcome with improved colonoscopic technique or technical innovations, will inform whether baseline adenoma location should be considered in future guidelines as an independent factor to tailor surveillance recommendations.

Introduction

The fecal immunochemical test (FIT) is the preferred non-endoscopic test in programmatic screening worldwide. Compared to other methods, FIT may be better accepted with higher participation rates, which may counteract its lower one-time sensitivities. We performed a systematic review and meta-analysis to ascertain participation patterns, participation rates, and FIT performance in organized multi-round FIT-based screening programs worldwide.

Aims & Methods

We performed a systematic search using Pubmed, Embase, and Cochrane databases. Participation patterns were categorized as “consistent screeners” (participation in 100% of rounds), “intermittent screeners” (including drop-outs, delayed entrants, and intermittent re-screeners), and “consistent non-screeners” (never screened). Data regarding positivity rates and detection rates for adenomas, advanced adenomas (AA), and colorectal cancer (CRC) across subsequent rounds were extracted and summary estimates were obtained. Yields were calculated for 2 populations: 1) Overall Population; 2) Consistent Screeners. A sensitivity analysis was performed including only programs with biennial screening with FIT cut-off of 20μg Hb/g feces.

Results

The literature search yielded 4,414 studies of which 25 were included. Across all studies, the fraction of the population behaving as “consistent screeners”, “consistent non-screeners”, and “intermittent screeners” was 47.1%, 30.3%, and 22.7%, respectively. For the overall population, FIT-positivity from the 1^st through 4^th rounds was 6.7%, 5.5%, 5.4%, and 5.3%; and detection rates were 5.8%, 3.3%, 2.9%, and 2.5% for CRC; 32.8%, 28.3%, 22.7% and 23% for AA; 48.6%, 46.2%, 48.7% and 44.6% for adenoma. For Consistent Screeners, FIT-positivity across studies from the 1^st through 4^th rounds was 6.4%, 4.8%, 4.1%, and 3.6%, respectively; and detection rates were 6.9%, 4.6%, 3.6%, and 3.1% for CRC; 38.7%, 30.6%, 28.2% and 24.9% for AA; 53.4%, 46.7%, 42.6% and 42.4% for adenoma. Similar trends were observed for biennial screening with FIT cut-off of 20μg Hb/g feces.

Conclusion

Organized multi-round FIT-based CRC screening programs achieve consistent participation in a minority of participants, and a sizeable proportion of the population do not initiate screening at all after multiple offers. Among those participating, FIT-positivity, and yields for CRC, AA, and adenoma are highest in the 1^st round, but may stabilize at a clinically relevant steady state at rounds 3+.

References

1. Doria-Rose VP, Lansdorp-Vogelaar I, McCarthy S, Puricelli-Perin DM, Butera V, Segnan N, et al. Measures of longitudinal adherence to fecal-based colorectal cancer screening: Literature review and recommended approaches. Int J Cancer. 2021 Jul 15;149(2):316–26.
2. Duncan A, Turnbull D, Wilson C, Osborne JM, Cole SR, Flight I, et al. Behavioural and demographic predictors of adherence to three consecutive faecal occult blood test screening opportunities: A population study. BMC Public Health. 2014 Mar 7;14(1).

Clinical Case Summary

Coeliac disease is an autoimmune disorder triggered by gluten exposure. It often overlaps with microscopic colitis, particularly in refractory cases, posing treatment challenges. Upadacitinib, a Janus Kinase (JAK) 1 inhibitor, targets key inflammatory pathways by blocking JAK enzyme activity. We describe the efficacy of Upadacitinib in treating refractory coeliac disease with concomitant microscopic colitis, resulting in clinical and histological remission.
A 46-year-old woman diagnosed with coeliac disease in 2014, initially presenting with chronic diarrhea. Coeliac serology revealed markedly elevated anti-Gliadin IgG levels and Tissue Transglutaminase IgA. Duodenal biopsies demonstrated mild villous blunting, crypt hyperplasia, and increased number of intraepithelial lymphocytes (IELs). Following diagnosis, her symptoms resolved with strict adherence to gluten-free diet.
Despite continued dietary adherence, she experienced recurrent symptoms over subsequent years. In March 2021, repeat endoscopy appeared normal macroscopically, though histology revealed increased IELs and mild villous blunting. Concurrent colonoscopy identified features consistent with microscopic colitis.Two Budesonide courses yielded minimal improvement; coeliac serology remained elevated.
During this period, she was also initiated on Upadacitinib for psoriatic arthritis. Her symptoms subsequently resolved over several months of treatment. At approximately 12 months, serology showed normalization of coeliac antibodies, remaining negative since October 2022. Recent endoscopy in March 2025 revealed no evidence of active coeliac disease, and duodenal biopsies showed restoration of normal villous architecture and no significant increase in IELs.
This case supports the potential of JAK inhibitors in treating refractory coeliac disease with microscopic colitis, especially when standard therapies fail. Recognizing the frequent overlap of these conditions is crucial for effective management.

References

Figure 1 - Index endoscopy in March 2021. Endoscopic findings of normal duodenum examined under WLE, NBI or TXI
Figure 2 - Endoscopy findings part way through Upadacitinib therapy, demonstrating decreased folds found in duodenum with patchy, subtle changes best visualised under TXI
Figure 3 - Latest endoscopy findings in March 2025 whilst on Upadacitinib for 3 years, demonstrating normal duodenum without endoscopic evidence of coeliac disease