Introduction

Artificial Intelligence (AI) -based optical polyp diagnosis (CADx) allows for pathology prediction during colonoscopies to support resecting and discarding polyps. However, it is unclear if clinical use of such CADx modules should be as an autonomous decision-making device or rather to assist endoscopists in their decision-making. We therefore evaluated outcomes for CADx optical diagnosis when used autonomously versus as an assistive device for polyps up to 10mm.

Aims & Methods

We conducted a secondary analysis of a randomized non-inferiority trial in which 1-10mm polyps found in patients undergoing diagnostic, screening, and surveillance colonoscopies were randomized to undergo optical diagnosis using: 1) an Autonomous AI-based CADx diagnosis or 2) an AI-assisted endoscopist diagnosis (AI-H). Primary outcome was the accuracy of optical diagnosis in both arms with reference standard being the histological evaluation of the polyps. Secondary outcomes included the surveillance interval agreement based on USMSTF guidelines when using CADx (with pathology as a reference standard) as well as the proportion of misclassified polyps with advanced histology.

Results

A total of 223 and 227 polyps 1-10mm in size were found amongst 268 patients randomized to the autonomous AI group versus 229 patients to the AI-H group, respectively. Accuracy for optical diagnosis for cases with high-confidence diagnoses was 77.6% (95% confidence interval [CI], 71.5-82.9) in the autonomous AI group versus 74.0% (95% CI, 67.8-79.6) in the AI-H group. When comparing surveillance interval agreement, autonomous AI-A did not exhibit superior agreement with USMSTF pathology-based recommendations compared to AI-H (88.5% [95% CI, 82.0-93.3] vs 83.4% [95% CI, 77.0-88.6]; P = 0.213), but did nearly reach the 90% PIVI recommendation benchmark to implement optical diagnosis. Prevalence of polyps with advanced histology (adenomas with villous component, high-grade dysplasia) and sessile serrated lesions was 11.2% (25/223 polyps) and 7.9% (18/227 polyps) in the autonomous AI and AI-H groups. The inability of CADx and endoscopists to specifically recognize these higher-risk lesions led to mismanagement of 3 out of 16 patients (18.8%) and 2 out of 16 patients (12.5%) with such lesions in the autonomous AI and AI-H groups, respectively.

Conclusion

In this study, autonomous use of CADx for optical diagnosis of polyps up to 10mm exhibited similar accuracy compared to its use as an assistive device. Autonomous AI did not achieve statistically significant different agreement with USMSTF guidelines-based surveillance interval assignment. Development of CADx advanced histology recognition and further evaluation of its use as an autonomous versus assistive device within multi-center studies is needed.

Disclosure

Daniel von Renteln has received research funding from ERBE Elektromedizin GmbH, Ventage, Pendopharm, Fujifilm, and Pentax and has received consultant or speaker fees from Boston Scientific Inc, ERBE Elektromedizin GmbH, and Pendopharm. The remaining authors disclose no conflicts.

Introduction

^​Intestinal ultrasound (IUS) is gaining increasing acceptance as a reliable tool for tight monitoring of ulcerative colitis (UC). Although several IUS scores have been proposed, none has been formally implemented into official recommendations of international scientific societies.

Aims & Methods

We aimed to compare an IUS score validated for Crohn’s disease (CD), the International Bowel Ultrasound Segmental Activity Score (IBUS-SAS)¹, with other IUS scores specifically developed for UC, in terms of correlation with endoscopic activity.
Patients with UC ≥ 18 years old undergoing partial or full colonoscopy between April 2023 and January 2024 were prospectively included. Endoscopic evaluation of at least the sigmoid colon was mandatory for inclusion, and patients with proctitis (Montreal E1) were excluded. Endoscopic activity was scored with both the Mayo Endoscopic Subscore (MES) and the Ulcerative Colitis Endoscopic Index of Severity (UCEIS), and defined by MES ≥ 1 and UCEIS ≥ 2. Moderate endoscopic activity was determined by a MES = 2 or an UCEIS = 5-6, while severe activity was defined by a MES = 3 or an UCEIS ≥ 7. IUS was performed within 4 weeks of endoscopy. The IBUS-SAS, the Milan Ultrasound Criteria (MUC)², the UC-IUS score³, the US Score from Parente et al.⁴, and the Hata index⁵ were included in the analysis. Applicability of the Civitelli index⁶, an IUS score developed for pediatric patients, was also evaluated. The Spearman’s rank coefficient [rho = ρ] was used to perform correlations, while receiver operating characteristic (ROC) curves were compared with the Hanley & McNeil method.

Results

Collectively, 104 patients were included in the study. The majority of patients (61.5%) had extensive colitis (Montreal E3), with half of them having an active disease at endoscopy (54.8% using MES and 48.1% using UCEIS). IUS examination was performed at a median time of 8 days (IQR 2-15) after endoscopy. All the investigated scores were significantly able to distinguish between the 3 grades of severity of colonic inflammation, using both MES and UCEIS as a reference, with the exception of the Parente’s US Score for the severe endoscopic activity assessed with UCEIS [0 (IQR 0-1) for UCEIS < 7 vs 1 (IQR 0-1) for UCEIS ≥ 7, p=0.09]. The IBUS-SAS showed a strong correlation with endoscopic activity, both with MES (ρ = 0.71, p < 0.01) and UCEIS (ρ = 0.73, p < 0.01) assessments. The Area Under the Curve (AUC) of IBUS-SAS to detect an endoscopic activity of at least moderate severity (MES ≥ 2 and UCEIS ≥ 5) was 0.85 and 0.90, respectively. The optimal cutoff to detect a MES ≥ 2 and UCEIS ≥ 5 was > 11 (sensitivity 75% and specificity 82%) and > 22 (sensitivity 92% and specificity 80%), respectively. Both cut-offs showed progressive increase from mild to severe activity, with IBUS-SAS maintaining a consistently high AUC at different thresholds. Consistent with IBUS-SAS results, all the investigated UC-developed IUS scores correlated with both the MES and the UCEIS (p < 0.01). AUC values and relative comparisons between the AUC are detailed in Table 1.

Table 1. Relative comparisons between the AUC of each score to detect a MES ≥ 2 (A, top-right) and an UCEIS ≥ 5 (B, bottom-left) are reported.
Grey rows and columns: AUC values (with 95%CI).
White cells: p-values obtained via the Hanley and McNeil method (significant p-values are shown in bold).

Conclusion

In this study we provide a successful validation of 6 IUS scores for UC. The IBUS-SAS has an optimal performance in predicting endoscopic activity also in patients with UC, despite having been developed initially for CD.

References

1. Novak KL, Nylund K, Maaser C, et al. Expert Consensus on Optimal Acquisition and Development of the International Bowel Ultrasound Segmental Activity Score [IBUS-SAS]: A Reliability and Inter-rater Variability Study on Intestinal Ultrasonography in Crohn's Disease. J Crohns Colitis 2021;15:609-616
2. Allocca M, Fiorino G, Bonovas S, et al. Accuracy of Humanitas Ultrasound Criteria in Assessing Disease Activity and Severity in Ulcerative Colitis: A Prospective Study. J Crohns Colitis 2018;12:1385-1391
3. Bots S, Nylund K, Löwenberg M, et al. Intestinal Ultrasound to Assess Disease Activity in Ulcerative Colitis: Development of a novel UC-Ultrasound Index. J Crohns Colitis 2021;15:1264-1271
4. Parente F, Molteni M, Marino B, et al. Are colonoscopy and bowel ultrasound useful for assessing response to short-term therapy and predicting disease outcome of moderate-to-severe forms of ulcerative colitis?: a prospective study. Am J Gastroenterol 2010;105:1150-7
5. Kinoshita K, Katsurada T, Nishida M, et al. Usefulness of transabdominal ultrasonography for assessing ulcerative colitis: a prospective, multicenter study. J Gastroenterol 2019;54:521-529
6. Civitelli F, Di Nardo G, Oliva S, et al. Ultrasonography of the colon in pediatric ulcerative colitis: a prospective, blind, comparative study with colonoscopy. J Pediatr 2014;165:78-84.e2

Disclosure

GD reports speaker fees from Alfasigma, Johnson & Johnson, Novartis, Pfizer, Takeda, and has served in advisory board for Celltrion Healthcare and Pfizer. The other authors have no conflict of interests to declare.

Introduction

Tumour necrosis factor-like ligand 1A (TL1A) is a novel therapeutic target for inflammatory bowel disease. Patients (pts) with moderately to severely active ulcerative colitis (UC) treated with the anti-TL1A antibody, RO7790121 (formerly PF-06480605 and RVT-3101), showed significant endoscopic improvement and adequate safety in the phase 2a TUSCANY trial.¹ Here, we report data from the global, randomised, double-blind, placebo-controlled, dose-ranging phase 2b TUSCANY-2 (NCT04090411) trial evaluating the efficacy and safety of subcutaneous (SQ) doses of RO7790121 in adult pts with moderately to severely active UC (total Mayo Score [tMS] ≥6, endoscopic subscore ≥2) who had previously failed conventional/advanced therapy.

Aims & Methods

Pts were randomised to receive RO7790121 SQ 50mg, 150mg, 450mg or placebo (PBO) monthly during the 12-week induction period and RO7790121 SQ 50mg, 150mg or 450mg monthly during the treat-through 40-week maintenance period. Efficacy endpoints were assessed at weeks 14 and 56 in the induction and maintenance intent-to-treat analysis sets, respectively. The primary efficacy endpoint was clinical remission at week 14 by tMS. Safety was assessed throughout the induction and maintenance periods. The secondary endpoint of clinical remission using modified Mayo Score (mMS) is aligned to updated FDA guidance. Other secondary endpoints included clinical remission (by tMS and mMS) at week 56 and endoscopic improvement at weeks 14 and 56. Efficacy analysis by pre-defined biomarker subgroups was also performed.

Results

Overall, 245 pts were randomised and treated; 228 completed the induction and 224 entered the maintenance period of the study. At week 14, 25.5%, 23.3% and 23.9% of pts in the RO7790121 50mg, 150mg, 450mg arms, respectively vs 11.6% in PBO experienced clinical remission by tMS (p>0.05 [Table]). The proportion of pts achieving clinical remission using mMS at week 14 was 29.8%, 35.0% and 31.8% in the RO7790121 50mg, 150mg, 450mg arms, respectively vs 11.6% in PBO (nominal p<0.05). Improvements in clinical remission were sustained throughout the maintenance period (Table). A greater proportion of pts demonstrated endoscopic improvement across all doses of RO7790121 vs PBO at week 14 and week 56. Numerically greater treatment effects were observed in biomarker-positive pts relative to all-comer pts. During the induction period, 47.8% (117/245) of pts reported ≥1 treatment-emergent adverse event (TEAE). The most common TEAEs (occurring in ≥5% pts overall) were anaemia (5.3%) and headache (5.3%). In the induction period, 10 pts experienced serious TEAEs (PBO=4; 50mg=3; 450mg=3); 2 treatment-related (PBO=1; 450mg=1). No pts discontinued the study due to TEAEs. A generally similar safety profile was observed during the maintenance period.

Conclusion

Data from the phase 2b TUSCANY-2 dose-ranging study suggests that RO7790121 has a favourable benefit/risk profile with clinically meaningful improvements in pts with moderately to severely active UC. A phase 3 study will be conducted to confirm these results.

Table. Summary of efficacy outcomes at week 14 and week 56

References

1. Danese S, et al. Clin Gastroenterol Hepatol 2021;19:2324–32.e6.

Disclosure

This study was sponsored by Pfizer Inc and Roivant Sciences Ltd. Medical writing support, under the direction of the authors, was provided by Margarita Gkrava, MSc, of Ashfield MedComms, an Inizio company, and was funded by F. Hoffmann-La Roche Ltd.
SD, consultancy/advisory role/board: AbbVie, Alimentiv, Allergan, Amgen, AMT, AstraZeneca, Athos Therapeutics, Biogen, Boehringer, BMS, Celgene, Celltrion, Dr Falk Pharma, Enthera, Ferring, Gilead, Hospira, Inotrem, Janssen, J&J, Lilly, Morphic, MSD, Mundipharma, Mylan, Pfizer, Roche, Sandoz, Sublimity Therapeutics, Takeda, Teladoc, TiGenix, UCB, Vial, Vifor. Lecture fees: Abbvie, Amgen, Ferring, Gilead, Janssen, Mylan, Pfizer, Takeda.
JRA, consulting/advisory role/board: Abbvie, Adiso, BMS, Ferring, Finch Therapeutics, GSK, Janssen, Merck, Pfizer, Roivant, Seres Therapeutics. Speakers’ bureau: Abbvie, BMS, Janssen. Research funding: Janssen, Merck, Pfizer.
SS, consulting/advisory role/board: AbbVie, Amgen, Arena Pharmaceuticals, Biogen, BMS, Celgene, Celltrion, Dr Falk Pharma, Ferring, Fresenius Kabi, Galapagos/Gilead Sciences, Hikma Pharmaceuticals, I-Mab, Janssen, Lilly, Morphic Therapeutic, MSD, Mylan, Pfizer, Protagonist Therapeutics, Provention Bio, Sandoz/Hexal, Takeda, Theravance Biopharma, and Ventyx Biosciences.
LP-B, consulting/advisory role/board : Abbvie, Abivax, Alimentiv, Alma Bio Therapeutics, Amgen, AMT, Arena, Biogen, BMS, Celltrion, Connect Biopharm, Cytoki Pharma, Enthera, Ferring, Fresenius Kabi, Galapagos, Genentech, Gilead, Gossamer, GSK, Hac Pharma, IAC Image Analysis, Index Pharmaceuticals, Inotrem, Janssen, Lilly, Medac, Mopac, Morphic, MSD, Norgine, Novartis, OM Pharma, ONO Pharma, OSE Immunotherapeuthics, Pandion Therapeuthics, Par' Immune, Pfizer, Prometheus, Protagonist, Roche, Roivant, Samsung, Sandoz, Sanofi, Takeda, Theravance, Thermo Fischer, Tigenix, Tillots, Viatris, Vifor, Vectivbio, Ventyx, Ysopia. Speakers’ Bureau: Abbvie, Amgen, Arena, Biogen, Celltrion, Ferring, Galapagos, Genentech, Gilead, Janssen, Lilly, Medac, MSD, Pfizer, Sandoz, Takeda, Tillots, Viatris, Vifor. Research funding: Celltrion, Fresenius Kabi, Takeda. Support travel: Abbvie, Amgen, Celltrion, Connect Biopharm, Ferring, Galapagos, Genentech, Gilead, Gossamer, Janssen, Lilly, Medac, Morphic, MSD, Pfizer, Sandoz, Takeda, Thermo Fischer, Tillots.
VJ, employment: Western University, London Health Sciences Centre, Alimentiv. Consulting or advisory role/board : AbbVie, Alimentiv, Arena pharmaceuticals, Asahi Kasei, Asieris, AstraZeneca, Avoro Capital, BMS, Celltrion, Endpoint Health, Enthera, Ferring, Flagship Pioneering, Fresenius Kabi, Galapagos, Gilde Healthcare, GSK, Genentech, Gilead, Innomar, JAMP, Janssen, Lilly, Merck, Metacrine, Mylan, Pandion, Pendopharm, Pfizer, Protagonist, Prometheus, Reistone Biopharma, Roche, Roivant, Sandoz, Second Genome, Sorriso, Synedgen, Takeda, TD Securities, Teva, Topivert, Ventyx, Vividion. Speakers’ bureau: AbbVie, Ferring, BMS, Galapagos, Janssen, Pfizer, Shire, Takeda, Fresenius Kabi.
GRD, consulting/advisory role/board: Alimentiv, Abbvie, BMS, Boehringer, Celltrion, GSK, Immunic, Index, J&J, Lilly, Polpharm, Takeda, Tillotts, Ventyx, Pfizer. Speakers’ bureau: Lilly, Pfizer, Takeda. Research funding: Cellitron, Lilly, Pfizer, Takeda.
JK, no COI.
RWL, consulting/advisory role/board: AbbVie, Aspen, BMS, Celgene, Celltrion, Chiesi, Ferring, Glutagen, Hospira, Janssen, Lilly, MSD, Novartis, Pfizer, Prometheus, Takeda. Research funding: Celltrion, Shire, Janssen, Takeda, Gastroenterological Society of Australia, NHMRC, Gutsy Group, Pfizer, Joanna Tiddy grant, McKusker Charitable Foundation.
AJY, consulting/advisory role/board: Abbvie, Takeda, Celltrion, BMS, J&J, Abivax.
JMM, employment/stock/other ownership interests: Roche/Genentech.
DB, employment: F. Hoffmann-La Roche.
KL, employment: Roche/Genentech.
CS, employment/stock/other ownership interests: Roche/Genentech.
KEH, employment/stock/other ownership: Pfizer.
BGF, stock/other ownership interests: Connect Biopharm, EnGene. Honoraria/consulting/advisory role/board: AbbVie, Abivax, Adiso, Agomab, Allianthera, Amgen, AnaptysBio, Arena Pharma, Avoro Capital, Atomwise, BioJamp, Biora Therapeutics, Blackbird Laboratories, Boehringer, Boxer Capital, Celsius Therapeutics, Celgene/BMS, Connect BioPharma, Cytoki, Disc Medicine, Duality, EcoR1, Equillium, Ermium, First Wave, Forbion, Galapagos, Galen Atlantica, Roche/Genentech, Gilead, Gossamer, GSK, Hinge Bio, Index Pharma, Imhotex, Immunic Therapeutics, Intercept, JAKAcademy, Janssen, Japan Tobacco, Kaleido, Klick Health, Landos Biopharma, Lenczner Slaght, LifeSci Capital, Lilly, Lument, Mage Biologics, Mestag, Millennium, MiroBio, Monte Rosa, Morgan Lewis, Morphic Therapeutics, Mylan, Nexys Therapeutics, Nimbus Therapeutics, OM Pharma, OrbiMed, Orphagen, Pandion Therapeutics, Pendopharm, Pfizer, Prometheus, Progenity, Protagonist, PTM Therapeutics, Q32 Bio, Rebiotix, REDX, Roche, Roivant/Televant, Sandoz, Sanofi, Seres Therapeutics, Silverback Therapeutics, Sobi, Spyre Therapeutics, Surrozen, Synedgen, Takeda, Teva, Thelium, Tigenix, Tillotts, Triastek, Ventyx Biosciences, Zagbio, Zealand Pharma. Speakers’ Bureau: Takeda, Janssen, Abbvie.