OP05 High-dimensional single-cell analysis identifies cellular signatures associated with response to vedolizumab therapy in ulcerative colitis

Pibiri, P.(1)*;Sun, S.(1);Karmi, N.(1);Oelen, R.(2);de Jong, S.(1);Bangma, A.(1);Hermoso, M.(1);Paraskevopoulou, M.(3);Farahmand, S.(3);Weersma, R.(1);Bigaeva, E.(1);Festen, E.(1);Xia, G.(3);Teunis, J.(4);Bleck, B.(3);

(1)University Medical Center Groningen, Hepatology and Gastroenterology, Groningen, The Netherlands;(2)University Medical Center Groningen, Genetics, Groningen, The Netherlands;(3)Takeda Pharmaceuticals, GI Drug Discovery Unit, Cambridge MA, United States;(4)University Medical Center Groningen, Flow Cytometry Unit, Groningen, The Netherlands;

Background

The unpredictability of response to biological therapy in ulcerative colitis (UC) patients, including vedolizumab (VDZ), an anti-α4β7 integrin antibody, poses a serious challenge for medical care. This study aims to identify the relevant cellular and molecular signatures that mark response to VDZ by combining single-cell transcriptomics (scRNAseq) and high-dimensional flow cytometry (Cytek). 

Methods

Gut biopsies and peripheral blood mononuclear cells (PBMCs) from patients with active UC were collected 2 weeks prior and 14 weeks after starting VDZ therapy. Response was evaluated at week 14 based on endoscopy and Physician Global Assessment. Transcriptome profiles of 159.188 and 95.057 high-quality single cells were generated from biopsies and PBMCs, respectively. Additional 797.670 CD45+ mucosal cells and 900.000 PBMCs were processed for validating scRNAseq results by Cytek.

Results

From a total of 25 UC patients (mean age 41±12.7 years, 52% female, 36% anti-TNFα naive), 15 patients (60%) reached response at week 14. Patients with no prior use of anti-TNFα were more likely to achieve remission with VDZ compared to anti-TNFα exposed patients (89% vs 44% response rate, respectively). Analysis of cell abundances revealed a partial recovery of the epithelial and stromal cell mucosal compartments in responders at week 14. Both scRNAseq and Cytek show that in responders VDZ prevented T cells (CD8+IL17+ and CD4+PD1+ subsets) and innate immune cells (NKs, DCs and inflammatory monocytes) from entering mucosa, resulting in their enrichment in PBMCs. In contrast, non-response to VDZ was marked by an expansion of inflammatory fibroblasts and activated endothelial cells, and depletion of most epithelial subsets. VDZ retained CD4 T cells in PBMCs in non-responders, but failed to halt active mucosal inflammation at week 14. Strikingly, involvement of the innate immune cells in UC inflammation prior to VDZ treatment was much more prominent in non-responders who showed higher baseline abundance of NKs, ILCs, γδ-T cells, DCs and inflammatory monocytes in their peripheral blood and mucosa compared to responders. While VDZ inhibited integrin β7 expression on circulating T cells in all patients, only in responders it coincided with the lower mucosal abundance of lymphocytes, suggesting that T cells in non-responders might employ alternative, β7-independent trafficking routes.

Figure 1: A) Biopsy and PBMCs from 25 UC patients were collected 2 weeks before and 14 weeks after Vedolizumab treatment. Cells for both biopsies and PBMCs were analysed using 10x genomic scRNA-seq and Cytek Aurora high-dimensional flow cytometry. B) Pie charts describe the composition of the cohort for different macro-categories.Figure 2: Uniform Manifold Approximation and Projection (UMAP) representation of single cells present in scRNA-seq biopsy and PBMC dataset, categorized by cell type labels.
Figure 3: Cell abundances illustrated per patient group across mucosal cells and PBMCs (scRNAseq datasets). Each bar represents a fraction of the total number of cells that are featured in samples from a specific patient group and time point. A) Relative abundances of epithelial cells (top) and stromal cells (bottom) in the mucosa of responders and non-responders during treatment. B) Relative abundances of innate and adaptive immune cells in the mucosa (top) and PBMCs (bottom) in responders and non-responders during treatment.
Figure 4: Cell abundances illustrated per patient group across mucosal cells and PBMCs (Cytek datasets). Each bar represents a fraction of the total number of cells that are featured in samples from a specific patient group and time point. Relative abundances of innate and adaptive immune cells in the mucosa and PBMCs in responders and non-responders during treatment.

Conclusion

This is the largest prospective scRNAseq study addressing the response to biological therapy in UC. We showed that active, innate immune cell-mediated inflammation marks the primary non-response to VDZ that persists prior to and throughout the treatment, accompanied by β7-independent lymphocyte migration.