Introduction

Achieving Histologic Remission is a key therapeutic target in ulcerative colitis (UC) and is predictive of favourable clinical outcomes. Predicting consistent therapy responses in clinical trials and practice is an unmet need crucial for improving therapeutic decisions in UC patients.

Aims & Methods

Thus, our study aimed to develop a novel AI model for assessing histologic disease activity and predicting early therapy response in a Phase 2 Clinical trial UC cohort.​​​​We used 302 whole slide images (WSIs) from moderate-to-severe active UC patients enrolled in a phase 2 mirikizumab (anti-IL23 mAb) clinical trial to train a novel deep neural network (DNN) model for WSIs. 106 WSIs from baseline, 106 from 12 weeks of treatment and 90 from 52 weeks of treatment were included. The ability of the model to predict remission/activity and response to therapy at 12 and 52 weeks was assessed. The system output was compared with human assessment. Histologic remission was defined as Geboes <2.0 and histologic improvement as Geboes <3.0. An early response to therapy at different time points was based on histologic remission or histologic improvement.

Results

Using all available WSIs, the system distinguished histologic activity/remission with sensitivity, specificity, and accuracy of 89%, 86% and 88%, respectively. The DNN predicted response to therapy based on histologic remission with 79% sensitivity, 78% specificity, and 78% accuracy at week 12 and 80% sensitivity, 92% specificity, and 87% accuracy at week 52. Furthermore, the model predicted response to therapy based on histologic improvement with 79% sensitivity, 76% specificity, 77% accuracy at week 12 and 80% sensitivity, 92% specificity, and 87% accuracy at week 52.

Conclusion

Our study shows the remarkable ability of this novel AI histology model to distinguish histologic remission and activity in a clinical trial cohort. Notably, the model accurately predicted response to therapy at weeks 12 and 52. This AI model represents a promising tool in clinical trials and potentially in practice, offering objective guidance for personalized management in UC patients.

Introduction

Human-computer interactions (HCI) can substantially influence the performance of Artificial Intelligence (AI). Studies indicate that although endoscopists, when assisted by AI, demonstrate significant improvements in assessing Barrett’s esophagus (BE), their performance does not reach the stand-alone level achieved by AI. A critical aspect of HCI is the impact of AI on the certainty and confidence exhibited by endoscopists. Indirectly, diagnostic confidence when utilizing AI may be associated with the trust and acceptance of AI. In a study focusing on BE videos, our objective was to examine the influence of AI on endoscopists' diagnostic confidence and explore potential correlations with diagnostic performance.

Aims & Methods

Twenty-two endoscopists from twelve centers, with varying levels of BE experience, reviewed ninety-six standardized endoscopy videos. The endoscopists were categorized as experts and non-experts and were randomly assigned to evaluate the videos with and without AI. Participants were divided into two arms: Arm A assessed videos initially without AI and then with AI, while Arm B followed the opposite order. Evaluators were tasked with identifying BE-related neoplasia and providing confidence ratings both with and without AI on a scale from 0 to 9. The study was followed up by a survey.

Results

In Arm A (without AI first, with AI second), the use of AI significantly increased confidence levels for both experts and non-experts (p<0.001). Notably, only non-experts experienced a significant improvement in sensitivity and specificity (p<0.001 bzw. p<0.05) with the assistance of AI. While experts exhibited higher confidence levels without AI compared to non-experts with AI, there was no significant difference in accuracy between these two groups. In Arm B both groups demonstrated a significant reduction in confidence (p<0.001), while maintaining consistent accuracy levels. In addition, in 9% of the choice changes with AI, the wrong choice was made despite correct indications made by AI.

Conclusion

AI significantly boosted confidence levels for both expert and non-expert endoscopists - an effect that was reversible in the study model. Furthermore, experts consistently displayed higher confidence levels, with or without AI, compared to non-experts with AI, despite comparable outcomes. It was further observed that in 9% of the choice changes the investigators ignored the AI in disadvantage of the patient. These findings highlight the potential role of AI in enhancing diagnostic confidence during endoscopic assessments.

Introduction

Informed consent (IC) serves as the ethical foundation for all medical interventions, empowering patients to make informed, autonomous decisions while safeguarding clinicians. Barriers such as limited health literacy and document complexity may hinder comprehension^1,2. This study evaluated the comprehensibility, readability, and effectiveness of the documents provided for IC and instructions for endoscopic procedures.

Aims & Methods

Outpatients undergoing endoscopic procedures completed a questionnaire assessing their understanding of IC, preparation instructions, and adherence to guidelines. Patient confidence and document comprehensibility were assessed through self-reported scales. Readability was assessed using Flesch Reading Ease score adapted for portuguese (FRE-PT)³.

Results

Among 186 participants (mean age: 58.1 ± 16.1 years), 76.9% reported full-extent reading of the provided instructions and IC document. Patients with primary education (≤4th grade) were less likely to read the documents fully (59.2% vs. 83.2%, p < 0.001) and were independently associated with the need for additional assistance (OR 5.65, 95% CI: 1.92 - 16.68, p = 0.002). Full-extent reading was independently associated with greater risk awareness (OR 5.52, 95% CI: 2.37 - 11.55, p < 0.001), reduced need for additional assistance (OR 0.22, 95% CI: 0.09 - 0.51, p < 0.001), and higher compliance with medication guidelines (75.7% vs. 45.8%, OR 3.42, 95% CI: 1.28 - 9.20, p = 0.015). Patients with primary education rated document’s comprehensibility lower (median:8, IQR 7-10) compared to those with higher education (median: 10, IQR 8 - 10, p < 0.001) and reported lower levels of confidence and preparation for the procedure (median: 8, IQR 7 - 10 vs. median: 9, IQR 8 - 10; p = 0.018). Outcomes did not differ significantly between patients aged 65 years or older compared to younger patients. FRE-PT scores classified readability as ‘’very difficult’’ for both upper endoscopy (20.87) and colonoscopy (26.00).

Conclusion

Full-extent reading IC and instruction documents reduces the need for assistance, improves compliance with pre-procedural measures, risk awareness and patient confidence. However, the document’s complexity and difficult readability disproportionately challenge and discourage patients with limited education. Future research should explore strategies to foster more equitable informed decision-making.

References

1. Protheroe J, Rowlands G. Matching clinical information with levels of patient health literacy. Nurs Manag (Harrow). 2013;20(3):20-21. doi:10.7748/nm2013.06.20.3.20.e1095
2. Ley R, Forio T. The use of readability formulas in health care. Psychol Health Med. 1996;1(1):7-28.
3. Martins TBF, Ghiraldelo CM, Nunes MGV, Oliveira Junior ON. Readability formulas applied to textbooks in Brazilian Portuguese. 1996.

Introduction

Artificial intelligence (AI) algorithms aid endoscopists in detecting and characterizing colonic polyps in clinical practise and lead to an improved adenoma detection rate. AI may also support therapeutic procedures by detecting and marking anatomical structures during complex procedures like endoscopic submucosal dissection (ESD) ¹. However, a possible effect of AI on the intraprocedural blood vessel detection rate during third space endoscopy remains unclear.

Aims & Methods

This study aimed at evaluating the impact of a newly developped AI algorithm for real time intraprocedural vessel detection on the vessel detection performance of endoscopists during ESD. To this end, 5470 single images from 59 full length third space endsocopy videos were annotated for submucosal blood vessels. Together with 179681 additional unlabeled images this dataset was utilized for the training of a DeepLabV3+ neural network. The trained algorithm detected and marked submucosal blood vessels on the endoscopic image in real time. Consequently, a video test consisting of 101 video clips with 200 predefined blood vessels was used for the evaluation of the algorithm alone, as well as 19 endoscopists with and without AI support.

Results

The algorithm showed a vessel detection rate (VDR) of 93.5% and a median vessel detection time (VDT) of 0.3 sec. AI support was associated with an elevation of endoscopists‘ VDR from 56.4% to 72.4% (p<0.001). VDT decreased from 6.7 to 5.2 sec with AI support (p<0.001). False positive detections were measured on 4.5% of single frames. They had a significantly shorter duration than true positive findings (0.7 vs. 6.0 sec, p<0.001).

Conclusion

AI support led to increased VDR and reduced VDT of endoscopists. False positive findings were easily distinguishable from true positives by their short duration. Future research may elucidate a possible effect of AI on clinical parameters like intraprocedural complication rates or procedure times.

References

1. Ebigbo A, Mendel R, Scheppach MW, et al. Vessel and tissue recognition during third-space endoscopy using a deep learning algorithm. Gut 2022.

Introduction

Metabolic associated steatosis liver disease (MASLD) is the new nomenclature of non-alcoholic fatty liver diseases (NAFLD) which highlights its crucial effect on the cardiometabolic risk factors that lead to the development and progression to liver diseases. Recently SGLT2 inhibitors have been identified to reduce hepatic steatosis in patients with T2DM with MASLD

Aims & Methods

Our aim was to assess the effect of SGLT2 inhibitors on hepatic steatosis in patients with T2DM and MASLD, we enrolled 30 diabetic adults with hepatic steatosis on ultrasound confirmed and quantified by baseline magnetic resonance proton density fat fraction (MRI-PDFF) they started SGLT2 inhibitors added to the standard of care treatment for 6 months and assessment of the percentage change of hepatic steatosis with done by repeat MR-PDFF.

Results

There were statistically significant decreases in the mean body weight and BMI after treatment (86.7Kg, 33.313 Kg/m²) respectively compared to their levels before treatment (90.367kg, 35.144 Kg/m²) respectively, P <0.001. Moreover, the mean serum fasting blood sugar and alanine transaminase decreased significantly after treatment (209.267, 22.633) compared to levels before treatment (220.566, 28.03) (P= 0.017, P=0.03) respectively. Regarding hepatic steatosis there was a statistically significant decrease in the mean MRI-PDFF measurement for hepatic steatosis after treatment (13.297 ±7.15) compared to the mean levels before treatment (15.288 ± 8.72) with a mean difference (-1.9911, P = 0.006) of about 13.16% decrease of steatosis

Conclusion

Addition of SGLT2 inhibitors to the standard treatment in type 2 diabetic patients with MASLD could reduce the hepatic steatosis significantly after 6 months of treatment associated with decrease in body weight, BMI, fasting blood sugar. Thus, the therapeutic use of SGLT2 inhibitors is valuable in type 2 diabetic patients associated with MASLD

Introduction

5-HT3 receptor antagonists are a second line treatment for irritable bowel syndrome with predominant diarrhoea (IBS-D), but there is no evidence on their efficacy for functional diarrhoea (FDr) (Savarino, 2022). Randomized controlled trials show that ondansetron seems to be effective in treating IBS symptoms (Garsed 2014, Gunn 2023), but real-life data are lacking.

Aims & Methods

Our aim was to study the clinical response to ondansetron in patients with IBS-D or FDr in a real-life setting. A retrospective analysis on consecutive patients (December 2014 – February 2024) meeting criteria for IBS-D or FDr according to Rome III criteria was conducted in a single Neurogastroenterology outpatient clinic. Only patients prescribed with ondansetron were included. A standardized template was used to collect clinical data and patient-reported response to treatment. Ondansetron was prescribed at initial dosage of 4mg 1 tablet every other day with the possibility of increasing the dosage up to 6 tablets/daily. Clinical response was defined as (i) an improvement in the number of bowel movements per day vs baseline and (ii) firmer stool consistency (Bristol stool form types 3 and 4, ‘normal’).

Results

Ninety-four patients (female 68.1%, mean age 45 ±15 years old) were included. Among them, 77 (82%) were diagnosed with IBS-D and 17 (18%) with FDr. The mean follow-up duration was 19.1 months (SD±16.71). Seventy patients (74.5%) achieved a clinical response to ondansetron. Response was maintained with ondansetron at a mean dosage of 2.5 tablets/daily (SD±1.5) for a mean follow-up duration of 23.5±17 months. There was no significant difference between the number of IBS-D and FDr patients who responded to ondansetron (table 1). Patients with IBS-D reported a significantly higher improvement in abdominal pain vs baseline (p<0.01, table 1). Twenty-four patients (25.5%) did not respond to ondansetron. Adverse events were significantly higher in non-responders compared to responders (p=0.02, table 1).

Table 1. Main characteristics of responders and non-responders at baseline and post-treatment

Conclusion

This is the first real-life study assessing the response to ondansetron in IBS-D or FDr patients showing long-term reduction in the frequency of bowel movements, improvement in faecal consistency, and reduction in abdominal pain. Our data suggest that ondansetron can be used in FDr patients.

References

1. Savarino E, Zingone F, Barberio B, Marasco G, Akyuz F, Akpinar H, Barboi O, Bodini G, Bor S, Chiarioni G, Cristian G, Corsetti M, Di Sabatino A, Dimitriu AM, Drug V, Dumitrascu DL, Ford AC, Hauser G, Nakov R, Patel N, Pohl D, Sfarti C, Serra J, Simrén M, Suciu A, Tack J, Toruner M, Walters J, Cremon C, Barbara G. Functional bowel disorders with diarrhoea: Clinical guidelines of the United European Gastroenterology and European Society for Neurogastroenterology and Motility. United European Gastroenterol J. 2022 Jul;10(6):556-584.
2. Garsed K, Chernova J, Hastings M, Lam C, Marciani L, Singh G, Henry A, Hall I, Whorwell P, Spiller R. A randomised trial of ondansetron for the treatment of irritable bowel syndrome with diarrhoea. Gut. 2014 Oct;63(10):1617-25.
3. Gunn D, Topan R, Barnard L, Fried R, Holloway I, Brindle R, Corsetti M, Scott M, Farmer A, Kapur K, Sanders D, Eugenicos M, Trudgill N, Whorwell P, Mclaughlin J, Akbar A, Houghton L, Dinning PG, Aziz Q, Ford AC, Farrin AJ, Spiller R. Randomised, placebo-controlled trial and meta-analysis show benefit of ondansetron for irritable bowel syndrome with diarrhoea: The TRITON trial. Aliment Pharmacol Ther. 2023 Jun;57(11):1258-1271.

Clinical Case Summary

Introduction: Inflammatory bowel disease (IBD) is linked to cutaneous manifestations in 10% of cases. Neutrophilic dermatoses represent a diagnostic and management challenge, as it requires a dual-targeted treatment approach.
Clinical Case: We present a 67-year-old man with A3L1B1 Crohn’s disease (CD) followed in our IBD consultation. His medical history included previous tobacco abuse, hypertension, type 2 diabetes and ischemic cardiopathy with heart failure. He was medicated for his comorbidities and had been treated with budesonide.
Due to evidence of disease progression, increased fecal calprotectin (FC) at 751ug/g and a history of invasive pulmonary aspergillosis, vedolizumab was initiated.
The patient presented with recurrent erythematous, tender, nodular lesions on the abdomen, limbs and face. Dermatological consultation and cutaneous biopsies led to an initial misdiagnosis of infectious panniculitis, as methicillin sensitive Staphylococcus aureus (MSSA) was isolated and the lesions regressed with antibiotics.
Two months later, cutaneous lesions reappeared and histopathology confirmed IBD-associated erythema nodosum-like neutrophilic dermatosis with MSSA isolation. A 10-week doxycycline course induced their regression. However, recurrence transpired 5 months later, associated with an increased FC (816ug/g).
A swap for ustekinumab (UST) resulted only in transient clinical and biochemical improvement (FC 324ug/g). Despite dose escalation to every 4 weeks, nodular/papular lesions with central postulation recently reappeared, prompting initiation of daptomycin with a potential swap to another drug class.
Conclusions: IBD can be associated with neutrophilic dermatoses and its diagnosis can be challenging. This case reports an CD-associated erythema nodosum-like neutrophilic dermatosis, underscoring its complex approach and management.
Note: We are submitting this case with iconography as it underscores a rare IBD manifestation.

References

Photographs of cutaneous lesions (3-4)