Y-ECCO Literature Review: Colleen McGregor

Despite advances in therapeutic options for Inflammatory Bowel Disease (IBD), the clinical use of biologic therapies remains challenging, largely due to the variable and often unpredictable nature of patient responses. Vedolizumab, a monoclonal antibody targeting integrin α4β7, has been shown to induce and maintain remission in both Crohn’s Disease (CD) and Ulcerative Colitis (UC) [1, 2]. Its mechanism of action involves selectively inhibiting the trafficking of lymphocytes to the intestinal mucosa by blocking the interaction between integrin α4β7 on circulating leucocytes and mucosal vascular addressin cell adhesion molecule 1 (MadCAM-1) on the gut endothelium [3].

By sequestering gut-homing immune cells in the circulation, vedolizumab effectively reduces their presence and pathological activity within the intestinal mucosa and associated lymphoid tissue. However, a substantial proportion of patients fail to respond or achieve full remission [4, 5], and at present there are no validated biomarkers capable of reliably predicting therapeutic response. Our understanding of the immunological and molecular mechanisms underpinning treatment failure also remains limited.

In this study, Horn and colleagues undertook a comprehensive multimodal profiling study to investigate systemic immune changes induced by vedolizumab in IBD, with the aim of identifying peripheral biomarkers and immune signatures predictive of treatment response.

This prospective cohort study integrated multiple high-dimensional immunological profiling techniques, including 36-parameter mass cytometry (CyTOF), flow cytometry (FACS), single-cell RNA sequencing (scRNA-seq), single-cell T- and B-cell receptor sequencing and serum proteomics using the Olink 96-plex inflammation panel. Peripheral blood samples were collected from patients with IBD prior to and at various time points up to 50 weeks following vedolizumab initiation. The cohort included 47 patients (38 with UC and 9 with CD) in the discovery arm, and 26 patients in an independent validation cohort. Age- and sex-matched healthy controls were also included for baseline comparisons.

Treatment response was defined as a reduction of at least three points in the Harvey-Bradshaw Index for CD or two points in the partial Mayo score for UC, evaluated at 30 weeks post-treatment. Faecal calprotectin levels at 18 weeks provided an additional marker of response.

Using scRNA-seq, the authors characterised α4β7 integrin expression across immune subsets. Expression of ITGA4 and ITGB7 transcripts was observed in multiple lineages, with significant changes in ITGB7 expression in circulating myeloid cells post-treatment. Mass cytometry analysis confirmed a reduction in α4β7-positive cells pre-treatment and an upregulation of β7, but not α4, following vedolizumab exposure.

T- and B-cell receptor repertoire analyses demonstrated a significant post-treatment increase in circulating gut-homing memory CD4+ T cells, accompanied by enhanced TCR diversity in CD4+ T cells co-expressing ITGA4 and ITGB7. These changes were not observed in CD8+ T cells, and overall TCR/BCR diversity did not correlate with treatment response.

The authors then applied iterative machine learning models, specifically logistic regression with LASSO and elastic net regularisation, to integrate and analyse data across platforms. High-performing features were identified across modalities, and models built on combined data types achieved strong predictive performance, with AUC values ranging from 0.81 to 0.99 depending on the feature set and scenario tested.

A simplified FACS-based model using three features (Ki67, CXCR3 and IL4 expression on Ki67+ memory CD4+ T cells) was subsequently validated in an independent cohort (n=26). Significantly, increased frequencies of proliferating (Ki67+) memory CD4+ T cells were strongly associated with vedolizumab non-response.

To further characterise these proliferating cells, scRNA-seq was performed on FACS-sorted memory CD4+ T cells. Ki67+ subsets were found to exhibit a Th1/Th17 effector phenotype, expressing transcription factors T-bet and RORγt. Functionally, these cells showed elevated expression of integrin α4β1 rather than α4β7, suggesting a plausible mechanism of therapeutic escape—namely, continued trafficking via an alternative integrin pathway not targeted by vedolizumab.

Collectively, these findings implicate Ki67+ memory CD4+ T cells as a putative biomarker of treatment resistance and point towards a mechanistic basis for non-response involving differential integrin expression.