Introduction

Immune checkpoint inhibitors (CPI) have revolutionised cancer outcomes, but frequently trigger immune related adverse effects, including life-threatening colitis. Current immunosuppressive treatments for CPI-colitis (CPI-C) incur significant side effects and may compromise CPI efficacy, highlighting the constant need for new therapies.

Aims & Methods

RNA sequencing was performed on colonic biopsies from patients with CPI-C (n=12), CPI (no colitis, n=6), healthy controls (n=13), and UC (n=12). Gene expression, pathway level (GSEA) and cellular composition were analysed. The impact of JAK1 blockade was evaluated in vivo using our pre-clinical model of CPI-C. Clinical readouts and colonic leukocytes were analysed by flow cytometry. Colonic organoids from CPI-C and controls were cultured and treated with IFNγ ± JAK1 inhibition with Upadacitinib. Clinical and endoscopic outcomes were defined in patients with treatment-refractory CPI-C following treatment with Upadacitinib 45mg for 8 weeks. Finally, we used a novel humanised mouse model of CPI-C, whereby mice express the human CTLA4 and PD1 gene, to validate these patient findings. Mice were given Nivolumab (6mg/kg) and Ipilimumab (10mg/kg) every 3 days with and without treatment of Upadacitinib (10mg/kg) and clinical features and flow cytometry were used to assess disease severity in these mice.

Results

Gene expression profiling of colonic biopsies highlighted overactivity of the JAK1-dependent cytokine IFNγ in CPI-C. IFNγ-responsive transcripts, including CXCL9, CXCL10, IDO1, GZMB and STAT1, were among the most upregulated transcripts, with IFNg signalling identified as the top enriched pathway. In vivo administration of Upadacitinib significantly reduced pathological clinical features, with reduced colon mass and infiltrating neutrophils, although no changes in T cell cytokine responses were seen in mice treated with CPI-C. Colonic organoids from CPI-C patients expressed high baseline expression of IFNγ-responsive transcripts and significantly overexpressed these transcripts following ex vivo stimulation with IFNγ, which was corroborated at protein level. Upadacitinib treatment of stimulated organoids were able to dampen these responses. To examine the functional impact of the IFNg/JAK1 pathway in CPI-colitis we blocked IFNg (neutralizing mAbs) or JAK1 (Upadacitinib treatment) in a humanised CTLA4 and PD1 murine model of CPI-colitis, which significantly attenuated disease with reduction in colon mass and fewer infiltrating neutrophils and fewer inflammatory monocytes. Humanised mice treated with Upadacitinib also had fewer CD8⁺ T cells in the colonic lamina propria, which also produced less cytotoxic cytokines. Encouraged by these findings, we treated 8 patients with severe, treatment refractory CPI-C with Upadacitinib. All achieved rapidly resolution of disease and complete mucosal healing.

Conclusion

IFNγ/JAK1 signalling was upregulated in CPI-C patients, which was recapitulated in a novel CPI-C derived colonic organoid model and two in vivo models of disease. Upadacitinib improved organoid morphology, reduced IFNγ targets, and ameliorated CPI-C in mice. Moreover, upadacitinib rapidly rescued treatment refractory patients highlighting its potential as a promising therapeutic approach in patients with severe disease.

References

1. Lo, J.W., Cozzetto, D., Alexander, J.L. et al. Immune checkpoint inhibitor-induced colitis is mediated by polyfunctional lymphocytes and is dependent on an IL23/IFNγ axis. Nat Commun 14, 6719 (2023). https://doi.org/10.1038/s41467-023-41798-2
2. Luoma A. M. et al. Molecular pathways of colon inflammation induced by cancer immunotherapy. Cell 182, 655-671.e22 (2020).
3. Sasson, S. C. et al. Interferon-gamma-producing CD8(+) tissue resident memory T cells are a targetable hallmark of immune checkpoint inhibitor-colitis. Gastroenterology 161, 1229–1244.e1229 (2021).
4. Gupta T, Antanaviciute A, Hyun-Jung Lee C, et al. Tracking in situ checkpoint inhibitor-bound target T cells in patients with checkpoint-induced colitis. Cancer Cell. 2024 May 13;42(5):797-814.e15. doi: 10.1016/j.ccell.2024.04.010. PMID: 38744246.

Introduction

The prevention of postoperative recurrence (POR) after ileocolonic resection is one of the most relevant issues in the management of Crohn's disease (CD). The role of ustekinumab (UST) and vedolizumab (VEDO) especially in the context of real-life studies is still not entirely clear, often due to small sample sizes, thus requiring further investigation and confirmation.

Aims & Methods

Our real life single-centre retrospective study aims to evaluate the effect of UST and VEDO therapy on prophylaxis of endoscopic recurrence after ileo-colonic resection in CD. In the whole cohort UST or VEDO were used as first-line therapy for prophylaxis of recurrence. All patients (pts) underwent the first endoscopic evaluation within 6-12 months to assess postoperative recurrence (POR), using the Rutgeerts score and were subsequently evaluated at 24 (+-3) months with a second endoscopy. Persistence on therapy two years after the surgery was also evaluated.

Results

64 pts with a mean age of 38 years were included, 25 pts received VEDO (Cohort 1) and 39 UST (Cohort 2).
In the two cohorts, 89% and 91% of pts had been previously resected; 58% and 55% were active smokers. Notably, 100% had already been exposed to one or more biological drugs and the most frequent reason for surgery was the presence of stenosing disease in both cohorts.
In cohort 1, the first post-operative colonoscopy at 6-12 months showed severe POR (RS score >i2) in 28% of pts, all of whom underwent a treatment change at this time point (71% and 29% drug change and VEDO optimisation, respectively). Among the pts who continued the drug (22 pts), the second endoscopy performed at 24 (+-3) months showed severe ROP in 10% of pts, who underwent treatment modification. At ~2 years persistence on therapy in the cohort was 72%.
In cohort 2, the first post-operative colonoscopy at 6-12 months showed severe POR in 25.6% of patients, who changed therapy at this time. Among patients who continued the drug (29 pts), the second endoscopy performed at 24 (+-3) months showed severe POR in 10.3% of pts, who underwent treatment change. At ~2 years persistence on therapy in the cohort was 66.6%. No patients discontinued the drug due to adverse effects in both cohorts.
Comparing the two cohorts regarding severe endoscopic recurrence at 6-12 months and perisistence on therapy at ~2 years, no significant differences emerged between the two groups (p = 1.000, OR: 0.89,IC 95%: 0.29–2.75 and p = 0.86, OR:1.29, IC 95%: 0.43–3.85, respectively).

Conclusion

In our real life cohort of patients with complex CD, ustekinumab and vedolizumab showed high levels of persistence on therapy at ~2 years after surgery. Limited to the small sample population, no significant differences were observed between the two drugs.

Introduction

Ustekinumab has pioneered IBD treatment by targeting the p40 subunit shared by IL- 12 and IL-23, crucial cytokines in gut inflammation. Despite its lower binding affinity compared to more recent anti-IL-23 antibodies targeting the p19 subunit, ustekinumab achieves comparable remission rates in IBD. This unexpected efficacy likely stems from its dual action; IL-12 neutralization suppresses the initiation of new inflammatory gut lesions, while IL-23 inhibition maintains disease control (Eftychi et al., 2019). However, ustekinumab's sub-optimal binding affinity and immunogenicity (3-12% neutralizing ADA per Stelara Prescribing Information), presents clear opportunities to develop MVP-523 as a next-generation, bio-better therapeutic. Discovered 25 years ago, ustekinumab predates our current understanding of IBD pathology and antibody optimization techniques. Its sub-optimal performance characteristics present opportunities for improvement. Moreover, ustekinumab's extensive patient years of data and favorable safety profile make it an ideal template for next-generation drug development.

Aims & Methods

Computational protein design enabled by CCG Molecular Operating Environment and Epimatrix software.
Variants were expressed, purified and tested experimentally biochemically, in vitro cell models and in vivo in mice.

Results

Our optimization efforts have yielded MVP-523, featuring increased binding affinity/potency and reduced immunogenicity, while having the potential to extend halflife and dosing intervals. CDR amino acid substitutions in both heavy and light chains reduced KD from ~300 pM to ~50 pM, significantly improving IC50 in cell-based assays and enhancing in vivo IL-12 neutralization. Concurrently, computational protein design combined with EpiVax's ISPRI (Interactive Screening and Protein Reengineering Interface) was employed for in-silico deimmunization. This process successfully reduced the anti-drug antibody (ADA) score from 10.4% (which correlates with 3-12% observed clinically) to a predicted score of 1.5%. Considering that the majority of ustekinumab-induced ADAs are neutralizing, this substantial improvement could potentially lead to significantly enhanced outcomes in clinical trials and increased treatment durability.

Conclusion

MVP-523 represents a novel, bio-better therapeutic, built on ustekinumab's reliability and offers a de-risked clinical development candidate for IBD and other diseases. Furthermore, it provides an intriguing platform for engineering trispecifics (IL-12, IL-23, other target) while maintaining the preferred bivalent structure of classical IgG.

Disclosure

JHS BLB AFH are shareholders in Multiverse Pharma

Introduction

Janus kinase inhibitors (JAKi), including tofacitinib (TOFA, a pan-JAK inhibitor) and filgotinib (FILGO, a selective JAK1 inhibitor), represent an emerging class of orally administered agents for moderate-to-severe ulcerative colitis (UC). Comparative real-world data on their effectiveness and safety are limited.

Aims & Methods

We conducted a multicenter, real-world, propensity score-weighted cohort study using data from the Sicilian Network for Inflammatory Bowel Diseases (SN-IBD). All adult UC patients treated with TOFA or FILGO with ≥16 weeks of follow-up were included. The primary outcomes were steroid-free clinical remission and clinical response at weeks 8 and 24. Treatment persistence and adverse events were also assessed. Inverse Probability of Treatment Weighting (IPTW) was applied to adjust for baseline differences.

Results

A total of 263 patients were included (TOFA: n=171; FILGO: n=92). At week 8, steroid-free clinical remission was achieved in 53.2% of patients treated with TOFA and 43.4% of those treated with FILGO (IPTW-adjusted OR: 1.51; 95% CI: 0.88–2.61; p=0.14). Clinical response at 8 weeks was reported in 75.4% of TOFA-treated patients versus 69.6% in the FILGO group (IPTW-adjusted OR: 1.34; 95% CI: 0.73–2.46; p=0.34). At week 24, steroid-free clinical remission was observed in 58.8% of the TOFA group and 44.9% of the FILGO group (IPTW-adjusted OR: 1.37; 95% CI: 0.76–2.49; p=0.30). Clinical response rates at 24 weeks were 63.6% for TOFA and 59.0% for FILGO (IPTW-adjusted OR: 1.13; 95% CI: 0.62–2.08; p=0.69). IPTW-adjusted Kaplan–Meier and Cox proportional hazards analyses showed no statistically significant difference in the likelihood of treatment discontinuation (log-rank p=0.294; IPTW-adjusted HR for TOFA: 0.78; 95% CI: 0.49–1.25; p=0.30). Adverse event rates were 12.7 and 5.4 per 100 person-years for TOFA and FILGO, respectively, with no significant difference (p=0.152).

Conclusion

In this real-world, IPTW-based study, TOFA and FILGO were well tolerated and demonstrated comparable and high effectiveness and treatment persistence in patients with moderate-to-severe UC, supporting their use as viable oral options in routine clinical practice.

Disclosure

FSM served as an advisory board member and/or received lecture grants from AbbVie, Ferring, Galapagos, Giuliani, Janssen, Lilly, Liohealth s.r.l., Pfizer, Sandoz, Takeda Pharmaceuticals. WF served as an advisory board member and/or received lecture grants from Abbvie, MSD, Takeda, Pfizer, Biogen, Sandoz, Zambon, Ferring, Sofar. AV received lecture grants from Pfizer. MC served as an advisory board member for AbbVie, MSD, Takeda Pharmaceuticals, and received lecture grants from AbbVie, MSD, Chiesi, and Takeda Pharmaceuticals. MC served as an advisory board member for AbbVie, MSD, Takeda Pharmaceuticals, and received lecture grants from AbbVie, MSD, Chiesi, and Takeda Pharmaceuticals. FM served as an advisory board member for Janssen and Galapagos, and received lecture grants from Takeda Pharmaceuticals. SR served as an advisory board member for AbbVie, Janssen, and MSD Pharmaceuticals, and received lecture grants from AbbVie, Galapagos, Janssen, MSD, Pfizer, and Takeda Pharmaceuticals. AO served as an advisory board member for AbbVie, Ferring, Galapagos, Lilly, MSD, Janssen, Pfizer, Samsung Bioepis, and Takeda Pharmaceuticals, and received lecture grants from AbbVie, Fresenius Kabi, Galapagos, Lionheath s.r.l., MSD, Sofar, Chiesi, Janssen, Pfizer, and Takeda Pharmaceuticals. All other authors: nothing to disclose.

Introduction

Mirikizumab is a humanized IgG4 monoclonal, anti-interleukin-23p19 antibody. VIVID-2 (NCT04232553), an ongoing open-label study, has demonstrated sustained clinical and endoscopic efficacy at 2 years of treatment of mirikizumab for Crohn’s disease (CD).¹ Here we present the long-term quality of life (QoL) up to 2 years from VIVID-2.

Aims & Methods

Patients enrolled in the phase 3 VIVID-1 study who achieved an endoscopic response to mirikizumab continued maintenance treatment in VIVID-2, while endoscopic non-responders received mirikizumab induction therapy followed by maintenance treatment. Quality of life was assessed using the Inflammatory Bowel Disease Questionnaire (IBDQ). The proportion of patients who achieved IBDQ response/remission, as well as mean changes from VIVID-1 baseline in IBDQ total scores at W64 and104 were reported. EQ-5D-5L Visual Analog Scale (VAS) scores (range 0-100, with 100 being best imaginable health state) were assessed at W64. Work productivity was assessed at W64 using Work Productivity and Activity Impairment Questionnaire in Crohn's Disease (WPAI:CD) (4 scores including absenteeism, presenteeism, work productivity loss and activity impairment are derived as percentages with higher values indicating greater impairment). Modified non-responder imputation (mNRI) was used for binary missing data imputation, and modified baseline observation carried forward (mBOCF) was used for continuous missing data imputation. Estimates of mean changes from baseline were calculated with analysis of covariance (ANCOVA).

Results

Analysis of endoscopic responders showed IBDQ response and IBDQ remission were maintained through W104 (85.0% and 71.3%, respectively) while endoscopic non-responders also maintained IBDQ response and remission through the same time (72.6% and 60.8%, respectively). (Table 1) Mean change from baseline in IBDQ total score was maintained through W104 for endoscopic responders and non-responders (least squares mean (standard error); 57.2 (2.1) and 46.8 (2.4), respectively). Similarly, EQ-5D-5L VAS improvement was observed for both groups at W64 (27.0 (1.0) and 22.3 (1.2), respectively). WPAI:CD improvements in both groups ranged from −13.9 (1.8) to −34.3 (1.5) for the different scores collected at W64.

Conclusion

Patients continuing mirikizumab maintenance and those receiving reinduction achieved meaningful long-term QoL improvement that was maintained at 2 years of treatment.

References

1. Barnes E, Sands B, D'Haens G, et al. Long-term efficacy and safety of mirikizumab following 104 weeks of continuous treatment for Crohn's disease: results from the VIVID-2 open-label extension study. Gastroenterology. 2025;168(3 Suppl):S14-S15. doi:10.1053/j.gastro.2025.01.073.

Introduction

Crohn's disease (CD) is a chronic inflammatory bowel disease of unknown etiology with no cure. Reliable tools for the early identification of patients at risk, especially for long-term prediction, remain to be defined.

Aims & Methods

This study aimed to employ machine learning to develop proteomics-based models to predict future CD onset. We utilized data from 39,634 participants without baseline CD from the UK Biobank (UKB) as the training cohort. Feature selection involved Cox regression to identify proteins associated with incident CD, followed by automated machine learning-based importance ranking using five-fold cross-validation (10 repeats). Four machine learning algorithms were applied to construct predictive models. Independent validation was conducted in a geographically distinct UKB cohort (n = 13,262) and an external Southern China cohort (n =74).

Results

Among 2,736 proteins, Cox regression identified 44 proteins significantly associated with incident CD. CD274, CHI3L1, REG1B, ITGAV, PRSS8, ITGA11, GDF15, DEFA1_DEFA1B, and IL6 were ranked the top 9 in protein importance ordering. Machine learning models using four algorithms were constructed based on these proteins, with XGBoost demonstrating the best performance. In the geographically distinct UKB cohort, the protein model (area under the curve [AUC] 0.76) outperformed clinical risk models (highest AUC 0.67, P < 0.05) and was externally validated in the Southern China cohort (AUC 0.79). Combining these proteins with clinical data enabled desirable predictions up to 16 years pre-diagnosis (AUC 0.78). Our model performed better than existing PREDICTS model (AUC 0.68, P < 0.05) when applied to the UK Biobank population. Individuals in the high-risk group stratified by the protein model were 4.23 times more likely to develop CD.

Conclusion

Our findings suggest that proteomics-based models could be used to predict individuals who will develop CD up to 16 years in advance. The novel approach has important implications in identifying individuals at high-risk for evaluation of early intervention.

Introduction

Despite robust quality indicators such as adenoma detection rate (ADR) in patients without IBD, such indicators for quality in colonoscopy are so far lacking in IBD. The European Society of Gastrointestinal Endoscopy currently recommends a dysplasia detection rate (DDR) of >8%, but evidence base for this is limited.

Aims & Methods

To examine whether existing quality indicators such ADR correlate with endoscopists’ IBD specific quality indicators such as dysplasia detection rate (DDR) and dysplastic lesions per colonoscopy (DPC). Data from consecutive colonoscopies performed by gastroenterologists across 11 tertiary IBD centres were collected, selected if at least 150 non-IBD colonoscopies and 40 IBD-surveillance colonoscopies were performed over a 5-year period. Non-IBD colonoscopies were for patients aged ≥50y, and IBD-colonoscopies for patients meeting national and international criteria for eligibility for dysplasia surveillance. Colonoscopies were excluded if the patient had known dysplasia, a known or suspected familial CRC syndrome, previous colonic resection or severe inflammation in Crohn’s or UC. Correlation between endoscopists’ ADR, adenomas per colonoscopy (APC) in non-IBD colonoscopies with their DDR and DPC in IBD colonoscopies. Secondarily, detection of sessile serrated lesions (SSLs) in patients with and without IBD were analysed.

Results

21 endoscopists (Table 1) met inclusion criteria, who performed 3207 non-IBD colonoscopies and 902 IBD (515 ulcerative colitis [UC] and 387 Crohn’s disease [CD]) colonoscopies. Of the non-IBD colonoscopies, patients’ median age was 65 (IQR 58-73) years, with 1556 (49%) female. Of the IBD-colonoscopies, patients’ median age was 48 (IQR 36-59) years, with 412 (45.7%) female, median disease duration of 14 (IQR 10-21) years. Chromoendoscopy was used in 234 (26%) of IBD colonoscopies. Nontargeted segmental biopsies were performed in 854 IBD colonoscopies, of which 12 (1.42%) detected dysplasia. No cancers were detected in all 902 IBD surveillance colonoscopies.
Key quality indicators for endoscopists are outlined in table 1. ADR correlated with both DDR (Pearson r = 0.44, p = 0.047) and DPC (r = 0.51, p = 0.018). Of the 17 endoscopists with ADR ≥35%, 13 (76%) endoscopists had DDR of ≥ 8%, 6 (35%) had DDR ≥ 12%, and 9 (53%) had a DPC rate of ≥ 0.15. Sessile serrated lesion detection rate (SSLDR) and SSL per colonoscopy (SSLPC) were strongly correlated in both IBD and non IBD endoscopies.

Table 1. Characteristics of 21 endoscopists included for study

Conclusion

Individual endoscopists’ ADRs in patients without IBD correlate with their DPC and DDR in patients with IBD. Further studies are required to assess whether these indicators are correlated with post-colonoscopy colorectal cancer in patients with IBD.

Disclosure

No conflicts of interest to disclose