Introduction

Inflammatory bowel disease (IBD) is associated with elevated myeloid lineage leucocytes, which show activation behaviour¹ including the CD14+CD16+DR++ phenotype known as proinflammatory monocytes and a major source of tumor necrosis factor-α.² Accordingly selective depletion of myeloid leucocytes by granulomonocytapheresis (GMA) with an Adacolumn is expected to promote remission and enhance drug efficacy. However, hitherto studies in IBD have reported contrasting efficacy outcomes, ranging from an 85%³ to statistically insignificant.⁴ Patients' demographic features should guide to select responder patients.

Aims & Methods

In ulcerative colitis patients, we were interested to identify relevant demographic features, which potentially could indicate a favourable response to GMA. Therefore, in a retrospective setting, we looked at the baseline clinical and endoscopic features of responders and non-responders to adsorptive GMA in 149 consecutive ulcerative colitis patients who had undergone GMA as remission induction therapy, 76 steroid naive, 71 steroid dependent, and 2 steroid refractory. Patients had received up to an 11 GMA sessions over 10 weeks. At entry and week 12, patients were clinically and endoscopically evaluated, allowing each patient to serve as her or his own control. Clinical activity index (CAI) ≤4 was defined as remission. Biopsies from endoscopically detectable inflamed mucosa were processed to see the impact of GMA on leucocytes within the mucosa.

Results

At entry, the average CAI was 12.8, range 10-17. Ninety-five patients (63.8%) had responded to GMA, 53 of 76 steroid naïve (69.7%), 41 of 71 steroid dependent (57.7%), and 1 of the 2 steroid refractory cases. On average remission was sustained for 8.6 months in steroid naïve patients and for 10.4 months in steroid dependent cohort. Mucosal biopsies revealed that infiltrating leucocytes were mostly neutrophils and monocytes. There was a marked reduction of infiltrating leucocytes in the biopsies from the responder patients. Patients with extensive deep UC lesions together with loss of the mucosal tissue at the lesions were non-responders. Patients with the first UC episode were identified as the best responders, followed by steroid naive patients. Further, a short duration of active UC prior to GMA marked a patient as a likely responder.

Conclusion

Patients who responded well to GMA attained a favourable long-term clinical course. GMA was more effective if applied immediately after a relapse than after a lag time. In general, GMA is favoured by patients for its safety profile and for being a non-drug remission induction option. Patients with extensive deep ulcers, with long duration of UC refractory to multiple pharmacologics are unlikely to benefit from GMA. In therapeutic settings, knowing baseline features, which may identify responder patients should help to avoid futile use of medical resources.

References

1. McCarthy DA, et al. Peripheral blood neutrophils in inflammatory bowel disease: morphological evidence of in vivo activation in active disease. Clin Exp Immunol 1991; 86: 489-93.
2. Belge KU, et al. The proinflammatory CD14+CD16+DR++ monocytes are a major source of TNF. J Immunol 2002;168:3536-42
3. Cohen RD. Treating ulcerative colitis without medications - look mom, no drugs. Gastroenterology 2005;128:235-36.
4. Sands BE, et al. A randomized, double-blind, sham-controlled study of granulocyte/monocyte apheresis for active ulcerative colitis. Gastroenterology 2008;135:400-09.

Disclosure

Nothing to declare

Introduction

Ulcerative colitis (UC) is a chronic disease characterized by inflammation. Despite the advent of anti-inflammatory biologics, most patients do not achieve a deep response to therapy. Furthermore, predicting response to therapy is challenging. Quantitative histologic features of the inflammatory microenvironment in UC have potential to inform treatment selection and maintenance efficacy once a regimen has started. To address this hypothesis, we compared such features derived from machine-learning (ML) models with endoscopic response status (Mayo endoscopic score; MES<2) in patients with UC treated with adalimumab (ADA).

Aims & Methods

A retrospective analysis of a phase III clinical trial of ADA (NCT02065622) was performed. A suite of ML models, collectively known as IBD Explore^TM (PathAI, Boston¹)* previously trained to quantify tissue regions (e.g., granulation tissue) and cells (e.g., goblet cells) were deployed on 345 hematoxylin and eosin (H&E)-stained whole slide images (WSI) of biopsies from the rectum and sigmoid colon at ADA end-of-induction (week 8). Human interpretable features (HIFs) were extracted from each WSI. Features were then correlated with end-of-maintenance (week 52) region-level endoscopic response status (MES<2) using univariate logistic regression. Multivariable logistic regression predictive models were also developed on a subset of the samples (277 WSIs). Predictive accuracy was assessed on a held-out test dataset (66 WSIs).

Results

Of the 657 model-derived HIFs assessed at end-of-induction, 251 were associated (p<0.05) with end-of-maintenance region level MES response status after applying Benjamini-Hochberg (BH) false discovery rate correction. Some example HIFs and their associated odds ratio with end-of-maintenance MES response are noted in Table 1.

Table 1: Example HIFs taken at end-of-induction and associated odds ratio to end-of-maintenance MES response.

Feature	Odds ratio	BH p-value
Median density of goblet cells in crypts	2.85	< 1e-10
Median area ratio of goblet cell mucin to epithelium + crypt abscess in crypts	2.50	< 1e-9
Proportion of neutrophils within 20 μm of goblet cells	2.40	< 1e-8
Proportion of inflamed crypts	0.46	< 1e-7
Proportion of plasma cells + intraepithelial lymphocytes in mucosa	0.66	< 1e-3
Area proportion of granulation tissue in mucosa	0.72	0.032
Heterogeneity of crypt area	0.71	0.01

To develop predictive models of region-specific end-of-maintenance endoscopic response, 5-fold cross validation was used. Biopsies from each subject were placed in a single split to minimize data leakage. The selected model (L1 penalized, class balance weighted) achieved a sensitivity and specificity of 0.64 and 0.67, respectively, on WSI not seen during predictive model development.

Conclusion

A quantitative understanding of the histologic changes in the colonic microenvironment shortly after treatment initiation is critical to advance our collective knowledge of UC and may aid in the prediction of a patient’s response. Here, quantitative features of goblet cells and goblet cell mucin were especially predictive of endoscopic response from end-of-induction. ML-derived quantitative histology features have the potential to guide therapeutic decisions in UC that leverage mechanistic rationale derived from histology, contributing to precision medicine in immunology.

References

*Gaitan et al., 2023. Machine learning-based prediction of Geboes score and histologic improvement and remission thresholds in ulcerative colitis. Digestive Disease Week 2023 Poster Presentation

Disclosure

¹PathAI Employee and owner of stock options
²AbbVie Employee and owner of stock options

Introduction

While histo-endoscopic improvement is achieved with Upadacitinib (Upa) treatment of ulcerative colitis (UC), its effect on the inflammatory microenvironment (IME) has yet to be fully resolved. Unlocking quantitative insights of histologic features of the IME has the potential to reveal biomarkers for treatment selection. Here, we assessed machine learning (ML)-derived histologic IME features associated with endoscopic response (ER) using UC biopsies at baseline and induction in two Upa clinical trials.

Aims & Methods

IBD Explore¹ (PathAI, Boston, version 1.3), a suite of ML models previously trained to quantify tissue regions (e.g., epithelium) and cells (e.g., goblet cells), was deployed on hematoxylin and eosin (H&E)-stained whole-slide images (WSI) of baseline and induction biopsies from patients with UC (N=732) enrolled in two clinical trials (NCT02819635, NCT03653026). Human interpretable features (HIFs) were extracted from each WSI and associated with Upa treatment and ER measured at induction (week 8; Mayo Endoscopic score≤1). Analysis of Covariance (ANCOVA) was used to associate change from baseline (CBL) of HIFs with Upa treatment, and logistic regression was used to associate baseline HIFs with ER, using baseline clinical data and total Geboes scores as covariates. Multivariable classification models, using baseline HIFs, were also trained to predict ER in Upa treated patients. Models were trained using 5-fold cross-validation and evaluated on the held-out test set using average predictions across these folds.

Results

ML-derived HIFs (N=402) displayed a significant CBL after Upa treatment, and 193 HIFs were prognostic of ER. These HIFs included fraction of goblet cell mucin area, fraction of normal epithelium, density of goblet cells in epithelium, fraction of infiltrated epithelium relative to all epithelium, fraction of crypts with abscess (CA) + complete granulation tissue (CGT) area relative to mucosa, and density of neutrophils (Table 1). HIFs quantifying the fraction of eosinophils and neutrophils within 20 μm of goblet cells were both independently prognostic and Upa- dependent predictive of ER. Baseline HIFs were also prognostic, as a group, of ER in the treatment group (mean validation AUROC = 0.78, test-set AUROC = 0.84).

Table 1. IME features associated with response to Upa

Conclusion

Here, using an ML-based approach to assess the UC IME, we identified a significant association between features relating to goblet cells, goblet cell mucin, and immune cells near goblet cells and Upa ER. These findings underscore the importance of granular IME characterization to better understand treatment mechanisms. These results also highlight the potential of histological biomarkers for predicting therapeutic outcomes. Further validation of these potential biomarkers in larger cohorts is necessary.

References

¹IBD Explore is for research use only. Not for use in diagnostic procedures.

Disclosure

PathAI authors are employees at PathAI and holders of company stock. ​
AbbVie authors are employees at AbbVie and holders of company stock

Introduction

Not all patients with ulcerative colitis (UC) respond initially to treatment with biologic agents, and predicting their efficacy prior to treatment is difficult. Vedolizumab, a humanized monoclonal antibody against alpha 4 beta 7 (α4β7) integrin, suppresses immune cell migration by blocking the interaction between α4β7 integrin and mucosal addressin cell adhesion molecule 1, and is used to treat patients with UC with moderate-to-severe activity. Reports about histological features that predict vedolizumab efficacy are scarce.

Aims & Methods

We evaluated the infiltration of inflammatory cells in the colonic mucosa prior to the induction of vedolizumab and examined the association between histological features and vedolizumab efficacy.
This was a multicenter, retrospective study. Patients with UC who were treated with vedolizumab between November 2018 and April 2021 were enrolled in this study. All eligible patients had moderate-to-severe disease activity at the time of the initial treatment with vedolizumab. Biopsy specimens taken from the colonic mucosa prior to vedolizumab induction were used, and the areas positively stained for CD4, CD68, and CD45 per unit area of the lamina propria were calculated. Clinical and histological features were compared between those with and without remission at week 22, and the factors associated with clinical outcomes were identified.

Results

A total of 42 patients were included in this study. 17 (40.5%) patients achieved clinical remission at week 22. Significantly more CD4+ cells were found in the colonic mucosal tissues of patients with remission than in those without remission (odds ratio [OR]=1.44, P=0.014). In contrast, the CD68-positive area and the CD45-positive were not significantly different between the two groups. The concomitant use of corticosteroids and high Mayo scores had a negative association with the vedolizumab response (OR=0.11, P=0.008 and OR=0.50, P=0.009, respectively).

Conclusion

A high CD4+ infiltration in the colonic mucosa was associated with remission after vedolizumab treatment. Histological evaluation for CD4+ cell infiltration may be helpful in selecting patients who can benefit from vedolizumab.

Introduction

Vedolizumab is one of the current treatments for patients with inflammatory bowel disease (IBD). The efficacy and safety, along with its gut specificity, make this drug an appealing therapeutic option for IBD patients with moderate to severe disease. However, as observed for other biologic treatments, a significant proportion of patients do not have an initial response to vedolizumab treatment. Currently, there is a lack of reliable predictive tools for vedolizumab treatment response, although this would help to alleviate the socioeconomic costs derived from this disease.

Aims & Methods

The primary aim of this study is to show putative transcriptional signatures associated with vedolizumab treatment response.
Methods: For the realization of this study, we used RNA-seq datasets from the Gene Expression Omnibus (GEO) database: GSE191328. This dataset includes samples from peripheral blood cells from responder and non-responder ulcerative colitis (UC) patients treated with vedolizumab and infliximab at baseline. Gene Set Enrichment Analysis (GSEA) was performed to compare the differential expression and the enrichment analysis was calculated using ssGSEA of the different groups under study.

Results

The analysis of the transcriptome of peripheral blood cells from responder (R) and non-responder (NR) vedolizumab-treated patients at baseline revealed interesting enrichment results. We found 20 gene sets positively enriched in NR versus R and only 9 gene sets negatively enriched in the same phenotypes. In order to find vedolizumab specific transcriptional signatures, we performed a similar analysis in samples from infliximab-treated patients. After that second round of analysis, we observed that around 50% of the enriched gene sets were similar for both biological treatments. Nevertheless, several interesting biological functions such as peroxisome function, reactive oxygen function or beta-catenin signalling are specifically enriched in NR patients to vedolizumab treatment.

Conclusion

Our data suggest that vedolizumab NR patients specifically show a transcriptional enrichment profile, which differs from vedolizumab R patients. Also, the comparative analysis with infliximab-treated patients reveals that those specific transcripts are modulated in response to this α4β7 integrin antibodies in peripheral blood cells. Therefore, we found a solid system to search for vedolizumab therapy predictive response markers using low invasive techniques.

Introduction

Ulcerative colitis (UC) is increasing in prevalence and incidence in Korea, and each patient shows a different course and treatment response. Since the 2000s, biologics have been developed and treatment options have diversified. Biologics and small-molecule therapeutics for immunological targets are currently under research and development. In the past, the goal was clinical improvement, including symptoms, but now endoscopic mucosal healing and, furthermore, the continuation of histological remission are gradually suggested. Accordingly, the purpose of this study was to identify clinical factors that could predict histological remission in patients receiving a first-line biologic agent, as well as confirm the patient's response to treatment and prognosis.

Aims & Methods

We retrospectively analyzed the medical records of 440 patients who had been diagnosed with UC and exposed to biologics for the first time in our hospital without using biologics at other hospitals between January 01, 2010 and December 31, 2022. Patients who did not undergo pre- and post-exposure endoscopy and biopsy or whose follow-up was discontinued were excluded. Clinical characteristics, blood tests, endoscopic findings, and pathological examination results of a total of 72 patients were collected. Endoscopic response was defined as a decrease in the Mayo Endoscopic Subscore (MES) compared to the score before the first exposure to the biological agent. Histological remission was defined as the result of no cryptitis, no crypt abscess, or no inflammatory cells observed on histological examination, and histological response was defined as remission in patients previously observed with active inflammation.

Results

The ratio of males to females was 53:19, and the median age at the time of diagnosis was 41 years. The average Mayo score at the time of diagnosis was 7.2, and the group with a histological response was 6.0, showing a statistically significant difference. The average albumin was 3.88 and was 4.06 in the histological response group, with a statistically significant difference. The median time to exposure to the first biologic was 1520 days. A colectomy was performed in three cases. Biologics were started with infliximab in 46.7% of patients, followed by adalimumab and vedolizumab in 22.7% and 21.3%, respectively. There was no statistical correlation between histological response and each biologic agent.
MES before exposure to biological agents was 50% in 2 and 44.4% in 3, but there was no statistical difference between each group. Multivariate logistic regression analysis was performed to identify predictors of endoscopic and histological responses after the first exposure to biological agents. As a result, it was confirmed that the Mayo score, ESR, and albumin had a causal relationship with the endoscopic response, respectively, and the odds ratio was 0.781, 1.051, and 3.319, respectively (p-value 0.044, 0.007, 0.049). In addition, it was confirmed that the Mayo score had a causal relationship with the histological response with an odds ratio of 0.796 (p-value 0.046).

Conclusion

This study confirmed that disease activity before treatment was a predictor of histological remission in UC patients using first-line biologics. It was confirmed that the histological response as well as the endoscopic response were well induced when the disease activity was low. Therefore, early escalation to biologics before the patient's condition worsens may be the basis for improving the overall prognosis.

References

Park, Sang Hyoung et al. “Clinical Characteristics and Long-term Prognosis of Elderly-Onset Ulcerative Colitis in a Population-Based Cohort in the Songpa-Kangdong District of Seoul, Korea.” Gut and liver vol. 15,5 (2021): 742-751. doi:10.5009/gnl20289
Park, Sunhee et al. “Histological Disease Activity as a Predictor of Clinical Relapse Among Patients With Ulcerative Colitis: Systematic Review and Meta-Analysis.” The American journal of gastroenterology vol. 111,12 (2016): 1692-1701. doi:10.1038/ajg.2016.418

Introduction

Mirikizumab (miri), a p19-directed IL-23 antibody, showed efficacy in ulcerative colitis (UC)¹. Miri modulates transcripts associated with disease activity and anti-TNF resistance in UC and increases expression of transcripts associated with healthy mucosa². Here, we evaluate the transcription profile of patients with UC who respond to placebo (PBO, 20.6%) or miri 200mg treatment (59.7%) versus non-responders and build a model that can predict response in patients with UC.

Aims & Methods

In the phase 2 AMAC trial (NCT02589665), patients were randomized 1:1:1:1 to receive intravenous PBO (N=63), 50mg miri (N=63), or 200mg miri (n=62) with possible exposure-based dose increases or fixed 600mg miri (n=61) at Week (W)0, 4 and 8. Colonic biopsies were collected at baseline, W12, and W52. An Affymetrix HTA2.0 microarray measured gene expression. Limma comparing values at baseline and W12 determined differentially expressed genes (DEGs). DEGs maintaining their W12 expression level through W52 among responders in either the PBO and miri 200mg arms were designated as similarly expressed genes (DEGSEGs).
XGBoost models³ were trained to predict clinical response (decrease in 9-point Mayo subscore [rectal bleeding, stool frequency, endoscopy] of ≥2 points and ≥35% from baseline [BL] with either a decrease of RB subscore of ≥1 or an RB subscore of 0 or 1) using baseline levels of DEGSEGs and TNF biomarkers. Leave-one-out cross validation of the models using AMAC data and validation experiments by training on the PROTECT data⁴ then predicting in AMAC were used to evaluate our models.

Results

Miri 200mg responders were associated with 84 DEGSEGs, PBO responders were associated with 26 DEGSEGs, and 21 DEGSEGs were common to both. When using the PROTECT data as the training data and AMAC data as the validation cohort, we were able to achieve high ROC-AUC and PR-AUCs using the top 36 (FC > 1.5) miri DEGSEGs (PR-AUC=0.78), or combined (PR-AUC=0.93) gene sets as the features and found that the response prediction was significantly higher in AMAC responders vs. non-responders.

Conclusion

Baseline transcript features were predictive of clinical response in patients with UC treated with miri 200mg and our model distinguished responders from non-responders. Future studies are underway to validate this model.

References

1. Sandborn WJ, et al. Gastroenterology. 2020;158(3):537-549.e10.
2. Steere B, et al. J Crohns Colitis. 2020;S103-S104.
3. Chen T, et al. Proceedings of the 22^nd ACM SIGKDD International Conference. 2016;7285-794.
4. Hyams JS, et al. Lancet Gastroenterol Hepatol. 2017;2(12):855-868.

Disclosure

TJ and KH report no conflicts of interest. WR reports grant support and personal fees from Abbott Laboratories, AbbVie, Aesca, Centocor, Falk Pharma GmbH, Immundiagnostik, and Merck Sharp & Dohme, and personal fees from Aptalis, Celltrion, Danone Austria, Elan, Ferring, Mitsubishi Tanabe Pharma Corporation, Otsuka, PDL BioPharma, Pharmacosmos, Schering-Plough, Shire, Takeda, Therakos, Vifor, Yakult, Amgen, AM Pharma, Astellas, AstraZeneca, Avaxia, Bioclinica, Biogen Idec, Boehringer Ingelheim, Bristol-Myers Squibb, Cellerix, Chemocentryx, Celgene, Covance, Galapagos, Genentech, Gilead Sciences, Grünenthal, ICON, Index Pharma, Inova, Janssen, Johnson & Johnson, Kyowa Hakko Kirin Pharma, Lipid Therapeutics, MedImmune, Millennium, Nestle, Novartis, Ocera, Pfizer, Procter & Gamble, Prometheus, Robarts Clinical Trial, Second Genome, SetPoint Medical, TiGenix, UCB, Zyngenia, and 4SC. BS, CH, NM, and VK are employees and shareholders of Eli Lilly and Company. CM is a former employee of Eli Lilly and Company. GD has served as an advisor for: AbbVie, Ablynx, Active Biotech, AgomAb Therapeutics, Allergan, AlphaBiomics, Amakem, Amgen, AM-Pharma, Applied Molecular Therapeutics, Arena Pharmaceuticals, AstraZeneca, Avaxia Biologics, Biogen, Boehringer Ingelheim, Bristol Myers Squibb/Celgene, Celltrion, Cosmo Pharmaceuticals, Dr. Falk Pharma, DSM Pharmaceuticals, Echo Pharmaceuticals, Eli Lilly and Company, enGene, Exeliom Biosciences, Ferring Pharmaceuticals, Galapagos NV, Genentech/Roche, Gilead Sciences, GlaxoSmithKline, Gossamer Bio, Immunic Therapeutics, Johnson & Johnson, Kintai Therapeutics, Lycera, Medimetrics, Medtronic, Merck Sharp & Dohme, Mitsubishi Pharma, Mundipharma, Nextbiotics, Novo Nordisk, Otsuka, Pfizer, PhotoPills, ProciseDx, ProDigest, Progenity, Prometheus Laboratories/Nestlé, Protagonist Therapeutics, RedHill Biopharma, Robarts Clinical Trials, Salix Pharmaceuticals, Samsung Bioepis, Sandoz, Seres Therapeutics/Nestec/Nestlé, Setpoint, Shire, Takeda, Teva, TiGenix, Tillotts Pharma AG, Topivert, Versant, and Vifor Pharma; and received speaker fees from: AbbVie, Biogen, Ferring Pharmaceuticals, Galapagos NV/Gilead Sciences, Johnson & Johnson, Merck Sharp & Dohme, Mundipharma, Norgine, Pfizer, Samsung Bioepis, Shire, Millenium/Takeda, Tillotts Pharma AG, and Vifor Pharma.