Role of artificial intelligence in imaging and endoscopy for the diagnosis, monitoring and prognostication of inflammatory bowel disease: a scoping review protocol.

INTRODUCTION: Inflammatory bowel diseases (IBD) are immune-mediated conditions that are increasing in incidence and prevalence worldwide. Their assessment and monitoring are becoming increasingly important, though complex. The best disease control is achieved through tight monitoring of objective inflammatory parameters (such as serum and stool inflammatory markers), cross-sectional imaging and endoscopic assessment. Considering the complexity of the information obtained throughout a patient's journey, artificial intelligence (AI) provides an ideal adjunct to existing tools to help diagnose, monitor and predict the course of disease of patients with IBD. Therefore, we propose a scoping review assessing AI's role in diagnosis, monitoring and prognostication tools in patients with IBD. We aim to detect gaps in the literature and address them in future research endeavours. METHODS AND ANALYSIS: We will search electronic databases, including Medline, Embase, Cochrane CENTRAL, CINAHL Complete, Web of Science and IEEE Xplore. Two reviewers will independently screen the abstracts and titles first and then perform the full-text review. A third reviewer will resolve any conflict. We will include both observational studies and clinical trials. Study characteristics will be extracted using a data extraction form. The extracted data will be summarised in a tabular format, following the imaging modality theme and the study outcome assessed. The results will have an accompanying narrative review. ETHICS AND DISSEMINATION: Considering the nature of the project, ethical review by an institutional review board is not required. The data will be presented at academic conferences, and the final product will be published in a peer-reviewed journal.

Milan Ultrasound Criteria Predict Relapse of Ulcerative Colitis in Remission.

Introduction: Bowel ultrasound is a noninvasive alternative to endoscopy for assessing the disease activity of ulcerative colitis; however, it is unclear whether bowel ultrasound can predict subsequent relapse from remission. Materials and Methods: A retrospective cohort study enrolled patients with ulcerative colitis who underwent bowel ultrasound between July 2018 and July 2021 during clinical remission (patient-reported outcome-2 ≤1 and no rectal bleeding) for at least 3 months and were followed up for 1 year. Ultrasonographic findings (bowel wall thickness, bowel wall flow, bowel wall stratification, and enlarged lymph nodes), Milan ultrasound criteria, Mayo endoscopic subscore, C-reactive protein, and fecal calprotectin levels and their association with subsequent clinical relapse were assessed. Relapse was defined as rectal bleeding score ≥1, stool frequency score ≥2, or treatment intensification for symptoms. Results: 31% of the patients (18/58) relapsed within 1 year. No single ultrasonographic finding predicted relapse, whereas Milan ultrasound criteria >6.2 (p = 0.019), Mayo endoscopic subscore ≥1 (p = 0.013), and fecal calprotectin ≥250 µg/g (p = 0.040) were associated with a shorter time to relapse in the log-rank test. Milan ultrasound criteria >6.2 (hazard ratio 3.22; 95% confidence interval 1.14-9.08, p = 0.027) and Mayo endoscopic subscore ≥1 (hazard ratio 8.70; 95% confidence interval 1.11-68.1, p = 0.039) showed a higher risk of relapse according to a Cox proportional hazards model. Conclusion: Bowel ultrasound can predict subsequent clinical relapse from remission in patients with ulcerative colitis using the Milan ultrasound criteria.

Optimal Use of Serum Leucine-Rich Alpha-2 Glycoprotein as a Biomarker for Small Bowel Lesions of Crohn's Disease.

Introduction: A large proportion of small bowel lesions in Crohn's disease (CD) may exist beyond the reach of ileocolonoscopy and there is no gold standard imaging modality to screen them, suggesting the need for optimal biomarkers. We aimed to compare the usefulness of C-reactive protein (CRP), faecal calprotectin (FC), and leucine-rich alpha-2 glycoprotein (LRG) in determining small bowel lesions of CD. Methods: This was a cross-sectional observational study. CRP, FC, and LRG were prospectively measured in patients with quiescent CD who underwent imaging examinations (capsule or balloon-assisted endoscopy, magnetic resonance enterography, or intestinal ultrasound) selected by the physician in clinical practice. Mucosal healing (MH) of the small bowel was defined as a lack of ulcers. Patients with a CD activity index >150 and active colonic lesions were excluded. Results: A total of 65 patients (27, MH; 38, small bowel inflammation) were analysed. The area under the curve (AUC) of CRP, FC, and LRG was 0.74 (95% confidence interval: 0.61-0.87), 0.69 (0.52-0.81), and 0.77 (0.59-0.85), respectively. The AUC of FC and LRG in a subgroup of 61 patients with CRP <3 mg/L (26, MH; 32, small bowel inflammation) was 0.68 (0.50-0.81) and 0.74 (0.54-0.84), respectively. The cut-off of 16 µg/mL of LRG showed the highest positive predictive value of 1.00 with specificity of 1.00, while negative predictive value was highest (0.71) with sensitivity of 0.89 at the cut-off of 9 µg/mL. Conclusion: LRG can accurately detect and/or exclude the small bowel lesions with two cut-off values.

An open competition involving thousands of competitors failed to construct useful abstract classifiers for new diagnostic test accuracy systematic reviews.

There are currently no abstract classifiers, which can be used for new diagnostic test accuracy (DTA) systematic reviews to select primary DTA study abstracts from database searches. Our goal was to develop machine-learning-based abstract classifiers for new DTA systematic reviews through an open competition. We prepared a dataset of abstracts obtained through database searches from 11 reviews in different clinical areas. As the reference standard, we used the abstract lists that required manual full-text review. We randomly splitted the datasets into a train set, a public test set, and a private test set. Competition participants used the training set to develop classifiers and validated their classifiers using the public test set. The classifiers were refined based on the performance of the public test set. They could submit as many times as they wanted during the competition. Finally, we used the private test set to rank the submitted classifiers. To reduce false exclusions, we used the Fbeta measure with a beta set to seven for evaluating classifiers. After the competition, we conducted the external validation using a dataset from a cardiology DTA review. We received 13,774 submissions from 1429 teams or persons over 4 months. The top-honored classifier achieved a Fbeta score of 0.4036 and a recall of 0.2352 in the external validation. In conclusion, we were unable to develop an abstract classifier with sufficient recall for immediate application to new DTA systematic reviews. Further studies are needed to update and validate classifiers with datasets from other clinical areas.

Usefulness of colestimide for diarrhea in postoperative Crohn's disease.

Background and Aim: Crohn's disease (CD) often causes intractable diarrhea after intestinal resection. Anion exchange resins have been reported to be effective in patients with bile acid diarrhea after distal ileectomy; furthermore, bile acid metabolism has been implicated in the pathogenesis of CD. Therefore, we aimed to examine the effectiveness of colestimide in the management of postoperative CD, and to compare its impact between patients with and those without ileocecal resection. Methods: Postoperative CD patients prescribed colestimide for diarrhea between April 2017 and December 2020 were retrospectively evaluated for changes in the total Crohn's disease activity index (CDAI), each CDAI component including diarrhea frequency/week, albumin, and C-reactive protein (CRP) was evaluated before and after the administration of colestimide. Furthermore, comprehensive patient and physician assessments were reviewed from medical records during the first outpatient visit as a global clinical judgment after the initiation of colestimide therapy. Results: A total of 24 patients were included, of whom 17 had a previous history of ileocecal resection. Significant improvement was noted in CDAI and diarrhea frequency only in the ileocecal resection group (CDAI: 114.5 ± 52.7 and 95.4 ± 34.8, P < 0.05; diarrhea frequency/week 23.8 ± 14.1 and 15.4 ± 11.2, P < 0.05, respectively). There was no significant improvement in other CDAI components, albumin level, or CRP level in either group. In the global clinical judgment, 13 and 4 patients in the ileocecal and non-ileocecal resection groups, respectively, were judged as "effective," with an overall efficacy rate of 70.8%. Conclusion: Colestimide is effective for diarrhea in patients with postoperative CD, especially after ileocecal resection.

Lower effectiveness of intravenous steroid treatment for moderate-to-severe ulcerative colitis in hospitalised patients with older onset: a multicentre cohort study.

BACKGROUND: The increasing incidence of older-onset ulcerative colitis (UC), which has a higher risk of surgery, is a global health issue. However, data regarding intravenous steroid treatment, one of the important treatment options to avoid surgery, for older-onset UC is lacking. AIMS: To evaluate the association between onset age and effectiveness of intravenous steroids in UC. METHODS: This retrospective multicentre (27 facilities) cohort study included moderate-to-severe hospitalised UC patients who underwent their first intravenous steroids between April 2014 and July 2019. The primary outcome was clinical remission at day 30, using two-item patient-reported outcome scoring. The key secondary outcomes were risks of surgery and adverse events (death, infection and venous thrombosis) within 90 days. A modified Poisson regression model was used for analysis. RESULTS: Overall, 467 UC patients (384 younger-onset and 83 older-onset) were enrolled. Clinical remission at day 30 was observed in 252 (65.6%) among younger-onset patients and 43 (51.8%) among older-onset patients (adjusted risk difference, -21.7% [95% CI, -36.1% to -7.2%]; adjusted risk ratio [ARR], 0.74 [95% CI, 0.59 to 0.93]). The risks of surgery and adverse events were higher in older-onset UC (20.5% vs. 3.1%; ARR, 8.92 [95% CI, 4.13 to 19.27], 25.3% vs. 9.1%; ARR, 2.19 [95% CI, 1.22 to 3.92], respectively). Four deaths occurred, all involving older-onset UC. The risks of infection and venous thrombosis were also higher in older-onset UC (18.1% vs. 8.6%, 7.2% vs. 0.5%, respectively). CONCLUSIONS: Older-onset was associated with a lower effectiveness of intravenous steroids with higher risks of surgery and adverse events in UC.

Early improvement in bowel wall thickness on transperineal ultrasonography predicts treatment success in active ulcerative colitis.

BACKGROUND: Bowel ultrasonography is a non-invasive imaging tool that can repeatedly monitor ulcerative colitis (UC) activity. AIM: This study aimed to determine whether early transabdominal or transperineal ultrasonography changes can predict subsequent clinical response to induction therapy in patients with UC. METHODS: This single-centre prospective study explored ultrasonographic predictors for clinical remission (patient-reported outcome-2 ≤ 1 with no rectal bleeding subscore) at week 8 in patients with active UC who underwent induction therapy, in comparison with faecal calprotectin and C-reactive protein (measured at baseline, week 1 and week 8). Predictive factors were assessed using multivariable regression models and receiver-operating-characteristic curve analysis. RESULTS: A total of 100 patients were analysed, of which 54 achieved remission at week 8. Baseline biomarker and ultrasonographic-parameter values were not predictive of remission. Contrastingly, change from baseline to week 1 in rectal bowel wall thickness measured using transperineal ultrasonography was an independent predictor of remission by week 8 (adjusted odds ratio is associated with a 1-mm decrease: 1.90 [95% confidence interval, 1.22-2.95]). In a subgroup analysis of the patients who did not achieve remission in 1 week, the predictive value of change in rectal bowel wall thickness remained high (AUC = 0.77 [95% confidence interval, 0.61-0.88]). CONCLUSION: Improvement in rectal bowel wall thickness measured using transperineal ultrasonography at week 1 predicts treatment success and potentially facilitates decision making during the early course of induction therapy in UC.

Residual Short-Segment Distal Inflammation Has No Significant Impact on the Major Relapse of Extensive Ulcerative Colitis.

BACKGROUND: Mucosal healing is a treatment target for patients with ulcerative colitis. However, the relevance of proactive treatment for residual inflammation limited to the distal colon is unclear. MATERIALS AND METHODS: Patients with ulcerative colitis who had extensive colitis in clinical remission and underwent colonoscopy were retrospectively enrolled and followed up for 2 years. Patients with complete endoscopic remission (CER; Mayo endoscopic subscore [MES] of 0) and those with short-segment distal inflammation (SS; active inflammation only in the sigmoid colon and/or rectum with a proximal MES of 0) were compared for the incidence of minor (only symptomatic) and major (need for induction treatments or hospitalization) relapses. RESULTS: A total of 91 patients with CER and 54 patients with SS were identified and 63 relapses (47 minor and 16 major) were analyzed. Univariate analysis showed that minor relapses were significantly more frequent in the SS group than in the CER group (hazard ratio [HR], 2.22; 95% confidence interval [CI], 1.25-3.97), but major relapses were not more frequent in the SS group than in the CER group (HR, 1.78; 95% CI, 0.65-4.83). Multivariable analysis showed that SS was the only risk factor significantly associated with minor relapse (HR, 2.38; 95% CI, 1.31-4.36). When the SS group was stratified by MES of 1 vs 2/3, minor relapses were significantly more frequent in the subgroup with MES of 2/3 than in the CER group, whereas the incidence of major relapse remained similar. CONCLUSIONS: Residual short-segment distal inflammation is not a risk factor for major relapses as long as endoscopic remission is achieved in the proximal colon. Therefore, reactive but not proactive treatment may be appropriate for such lesions.

Increased DNA-incorporated thiopurine metabolite as a possible mechanism for leukocytopenia through cell apoptosis in inflammatory bowel disease patients with NUDT15 mutation.

BACKGROUND AND AIMS: Polymorphisms in the nucleotide diphosphate-linked moiety X-type motif 15 (NUDT15) gene are associated with thiopurine-induced leukopenia in patients with inflammatory bowel disease (IBD). NUDT15-associated subcellular thiopurine metabolism has not been investigated in primary lymphocytes. We hypothesized that NUDT15 mutation increases DNA-incorporated deoxythioguanosine (dTG) and induces apoptosis in lymphocytes. METHODS: DNA-incorporated dTG in peripheral blood mononuclear cells (PBMCs) and 6-thioguanine nucleotides (6-TGN) in red blood cells were measured in patients with IBD undergoing thiopurine treatment. The association of a single nucleotide polymorphism for NUDT15 (rs116855232) with dTGPBMC was examined. The pro-apoptotic effect of DNA-incorporated dTG was examined ex vivo in association with NUDT15 genotypes by co-culturing patient-derived peripheral CD4+ T lymphocytes with 6-thioguanine (6-TG). RESULTS: dTGPBMC was significantly higher in NUDT15 variants than in non-variants. dTGPBMC, but not 6-TGNRBC, negatively correlated with peripheral lymphocyte counts (r = - 0.31 and - 0.12, p = 0.012 and 0.173, respectively). DNA-incorporated dTG significantly accumulated to a greater extent in lymphocytes from NUDT15 variants when co-cultured with 6-TG ex vivo than in those from non-variants and was associated with decreased proliferation and increased apoptosis. CONCLUSION: Increased DNA-incorporated dTG may be responsible for thiopurine-induced leukocytopenia through cell apoptosis in IBD patients with NUDT15 mutation.

Post-marketing analysis for biosimilar CT-P13 in inflammatory bowel disease compared with external data of originator infliximab in Japan.

BACKGROUND AND AIM: CT-P13, an infliximab (IFX) biosimilar, was approved for treatment of inflammatory bowel disease. However, no comparison with the originator IFX in this indication has been conducted in Japan where endemic levels of tuberculosis and hepatitis virus infection are not low. We evaluated the safety and efficacy in real-world data of CT-P13 and compared with originator IFX data in Japan. METHODS: In a prospective post-marketing surveillance (PMS) study, patients who received CT-P13 in a 28-month period from January 2015 were followed up for 2 years. By conducting Japanese administrative database search (DBS) for the same period of PMS, data of the originator IFX including treatment persistence, tuberculosis incidence, and liver injury were analyzed retrospectively and compared with the corresponding PMS data of CT-P13. RESULTS: CT-P13 persistence in PMS (n = 640) and IFX persistence in DBS (n = 4113) were almost similar between patients who switched from the originator and patients who continued on the originator, and also between the biologics-naïve patient groups. There were no differences in the incidences of tuberculosis and hepatic injury (Tuberculosis: 2 patients [0.31%] with CT-P13, 10 patients [0.24%] with the originator, P = 0.75; Hepatic injury: 18.5% with CT-P13, 15.4% with the originator, P = 0.22). Most of the patients with hepatic injury continued treatment in PMS and DBS at similar rates (80.8% vs 83.6%, P = 0.65). CONCLUSION: The results of long-term PMS of CT-P13 compared with external reference data from an administrative database suggested that the biosimilar and its originator were comparably useful in real-world clinical practice.

Differential effects of mesalazine formulations on thiopurine metabolism through thiopurine S-methyltransferase inhibition.

BACKGROUND AND AIM: Thiopurines are often used in combination with mesalazine for the treatment of ulcerative colitis (UC). Mesalazine formulations are delivered to the digestive tract by various delivery systems and absorbed as 5-aminosalicylic acid (5-ASA). 5-ASA is known to inhibit thiopurine S-methyltransferase (TPMT) activity and to affect thiopurine metabolism. There have been no studies comparing TPMT inhibition by multimatrix mesalazine (MMX) with other formulations. We investigated the difference in TPMT inhibition by different mesalazine formulations and prospectively confirmed the clinical relevance. METHODS: Plasma concentrations of 5-ASA, N-acetyl-5-aminosalicylic acid (N-Ac-5-ASA), and TPMT activities were measured in UC patients receiving various mesalazine formulations (time-dependent or pH-dependent mesalazine or MMX) as monotherapy. Patients already on both time-dependent or pH-dependent mesalazine and thiopurines switched their mesalazine to MMX, examining 6-thioguanine nucleotide (6-TGN) and 6-methylmercaptopurine (6-MMP) 0 and 8 weeks after switching. Clinical relapse after switching was also monitored for 24 weeks. RESULTS: Plasma 5-ASA and N-Ac-5-ASA levels were significantly higher in patients receiving time-dependent mesalazine (n = 12) compared with pH-dependent mesalazine (n = 12) and MMX (n = 15), accompanied by greater TPMT inhibition. Prospective switching from time-dependent mesalazine to MMX decreased 6-TGN levels, increased those of 6-MMP, and increased 6-MMP/6-TGN ratios. Furthermore, this resulted in significantly more relapses than switching from pH-dependent mesalazine to MMX. CONCLUSIONS: Time-dependent mesalazine has higher plasma 5-ASA and N-Ac-5-ASA levels and greater TPMT inhibition than MMX. Therefore, switching from time-dependent mesalazine to MMX may lead to an increase of 6-MMP/6-TGN, which may reduce the clinical effectiveness of thiopurines, warranting close monitoring after switch.

Accuracy of Ultrasound for Evaluation of Colorectal Segments in Patients With Inflammatory Bowel Diseases: A Systematic Review and Meta-analysis.

BACKGROUND & AIMS: The accuracy of ultrasound for evaluation of individual colorectal segments in patients with inflammatory bowel diseases (IBD) has not been evaluated in a systematic review. We evaluated the diagnostic accuracy of ultrasound in different colorectal segments of patients with IBD. METHODS: We searched publication databases from inception through March 2019 for studies that assessed the accuracy of ultrasound in detection of inflammation in right, transverse, and left colon and in rectum in patients with IBD, using findings from colonoscopy as the reference standard. Subgroup analyses were performed including IBD type, patient age, body mass index, and study design. The risk of bias was assessed with the Quality Assessment of Diagnostic Accuracy Studies-2 tool. RESULTS: Nineteen studies (1101 patients) were included in the qualitative synthesis. After we assessed the risk of bias, 7 studies (comprising 84 patients with Crohn's disease and 420 patients with ulcerative colitis) were included in the meta-analysis. Bowel wall thickness ≥ 3 mm identified colorectal segments with inflammation with 86.4% pooled sensitivity (95% CI, 76.1%-92.7%) and 88.3% pooled specificity (95% CI, 58.1%-97.6%). In rectum only, bowel wall thickness ≥ 3 mm identified inflammation with 74.5% sensitivity (95% CI, 53.0%-88.3%) and 69.5% specificity (95% CI, 33.6%-91.1%). Diagnostic accuracy was comparable among subgroups. Increased bowel wall flow and loss of stratification had higher true-positive odds ratios. CONCLUSIONS: Based on meta-analysis of patient-level data, ultrasound has higher diagnostic accuracy for detecting inflammation in colon than rectum in patients with IBD. Studies are needed to increase the accuracy of ultrasound detection of inflammation in rectum.

Determining the usefulness of Capsule Scoring of Ulcerative Colitis in predicting relapse of inactive ulcerative colitis.

BACKGROUND AND AIM: The usefulness of second-generation colon capsule endoscopy (CCE2) in ulcerative colitis (UC), especially in clinically inactive patients, has been reported. Capsule Scoring of Ulcerative Colitis (CSUC) was developed as a severity index for UC. We aimed to determine whether CSUC is useful for predicting relapse during clinical remission. METHODS: Forty-one UC patients in clinical remission who underwent CCE2 were prospectively registered from April 2016 to August 2019. Patients' CSUC score was obtained; those with subsequent relapse were followed up retrospectively. The correlation of CSUC with white blood cell count, platelet count, albumin, C-reactive protein, fecal calprotectin and fecal lactoferrin levels, and fecal immunochemical test results was evaluated; their predictive values for future relapse were compared. RESULTS: The correlations of CSUC with white blood cell, platelet, albumin, C-reactive protein, fecal calprotectin, fecal immunochemical test, and fecal lactoferrin values were rs  = 0.13, 0.27, -0.25, 0.15, 0.50, 0.43, and 0.50, respectively. CSUC was higher in 12 patients who relapsed within 1 year than in 29 patients who remained in clinical remission (2.83 ± 1.95 vs 0.72 ± 1.00, P < 0.01). Receiver operator characteristic curve analysis showed that CSUC ≥ 1 was a predictor of relapse (area under the curve of 0.82, sensitivity of 83.3%, specificity of 58.6%) and maybe superior to fecal biomarkers. In the univariate analysis, patients with CSUC of 0 had a lower relapse rate than those with CSUC of ≧ 1 (P = 0.03, log-rank test). After analyzing patients who underwent CCE2 within 6 months after the successful induction treatment, results showed that those with CSUC of ≤ 1 remained in clinical remission for a year. CONCLUSION: CSUC predicts relapse within 1 year in UC patients in clinical remission, especially when used 6 months after induction treatment.

Transperineal ultrasound predicts endoscopic and histological healing in ulcerative colitis.

BACKGROUND: Transabdominal ultrasound is useful to assess inflammation in patients with ulcerative colitis (UC); however, the assessment of the rectum is challenging and a barrier for its widespread use. AIM: To evaluate if transperineal ultrasound is useful for predicting endoscopic and histological findings of the rectum in UC. METHODS: Fifty-three consecutive adults with UC who required colonoscopy were included and transperineal ultrasound was performed in combination with transabdominal ultrasound within a week before or after colonoscopy with rectal biopsy. Mayo endoscopic subscore (MES) ≤1 was defined as endoscopic healing and Geboes score <2.1, Robarts histopathology index ≤6, and Nancy index ≤1 were defined as histological healing. Limberg score and bowel wall thickness were recorded with transperineal ultrasound. Faecal calprotectin was also measured. RESULTS: Excellent correlation was confirmed between colonoscopy and transabdominal ultrasound in all segments except for the rectum. Rectal bowel wall thickness and Limberg score in transperineal ultrasound well correlated with rectal MES and histological indices. Bowel wall thickness ≤4 mm predicted endoscopic (Area under the curve [AUC] = 0.90) and histological (AUC = 0.87-0.89) healing. In multivariable logistic regression analysis, only bowel wall thickness in transperineal ultrasound was a significant independent predictor for rectal endoscopic and histologic healing (P < 0.05) and the predictability was better than faecal calprotectin. CONCLUSIONS: Transperineal ultrasound predicts endoscopic and histological healing of the rectum. The combination of transperineal ultrasound with transabdominal ultrasound visualises the entire colorectum and is an ideal modality for the treat-to-target strategy. Clinical Trials Registry number UMIN000033611 (https://upload.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000038323).

Correction to: Management of elderly ulcerative colitis in Japan.

Management of elderly ulcerative colitis in Japan.