溃疡性结肠炎患者对维多珠单抗响应轨迹的定义
An ulcerative colitis taxonomy defined by individual patient-reported response trajectories
研究通过多组学分析定义了溃疡性结肠炎患者对维多珠单抗的响应轨迹,发现SPP1驱动的血管炎症是治疗抵抗的关键机制。该研究为溃疡性结肠炎的精准医疗提供了新视角。
研究通过多组学分析揭示了溃疡性结肠炎患者对维多珠单抗治疗的响应轨迹及其背后的分子机制,为精准医疗提供了新视角。
Main
Uncontrolled mucosal inflammation in ulcerative colitis (UC) results in reduced quality of life and increased risk for hospitalization and colectomy1,2,3. Treatments include targeted biologic and small-molecule therapies (advanced drug therapies (ADTs)); however, most patients do not reach sustained symptomatic and endoscopic remission after 1 year4. Even fewer achieve stricter composite endpoints such as ‘comprehensive disease control’5, or ‘disease clearance’6,7, which represent a patient-level combination of symptom resolution, improvements in quality of life, normalization of biomarkers and near-complete healing of intestinal lesions on endoscopy or histology6.
ADTs for UC target distinct immune pathways and thus may fail through distinct resistance mechanisms8. Antitumor necrosis factor (anti-TNF) therapies primarily inhibit cytokine-driven inflammatory amplification and may also affect membrane-bound TNF-expressing immune cells. Therapies targeting interleukin (IL)-12/IL-23 or IL-23 pathways interfere with cytokine programs supporting T cell and innate immune activation and differentiation. Treatments targeting the Janus kinase (JAK) pathway prevent intracellular communications downstream of cytokine signaling in effector cells. By contrast, vedolizumab blocks binding between α4β7 integrin and mucosal addressin cell adhesion molecule (MAdCAM)-1, thereby modulating gut-directed immune-cell trafficking. These mechanistic differences provide a rationale for considering treatment resistance as a therapy-specific biological endotype shaped by the pathway targeted by each drug9.
Most UC studies have relied on baseline molecular/clinical features and binary responder definitions at landmark time points to classify these resistance mechanisms10,11. Such approaches may miss dynamic and treatment-specific resistance biology, particularly when persistent disease activity reflects cellular compartments not directly targeted by the therapy. For vedolizumab, nonresponse could arise from incomplete blockade of lymphocyte trafficking, but also from α4β7-independent inflammatory programs involving innate immune cells, stromal cells, endothelial activation or alternative trafficking escape mechanisms. We hypothesize that patient-level response trajectories could provide a clinical framework for resolving treatment-specific endotypes by linking response dynamics to molecular mechanisms of resistance.
Results
Using a systems immunology approach12,13,14, we aimed to use a group-based multivariate trajectory modeling (GBMT) methodology on individual patient-reported outcome (PRO) response data from phase 3 programs for vedolizumab15,16 as an ensemble learning method to determine whether early response dynamics may serve as surrogates for disease courses (Fig. 1).
A phase 3b head-to-head double-blind clinical trial for vedolizumab (n = 382; n = 231 male) versus adalimumab (n = 386; n = 216 male) in UC was used to identify treatment-specific PRO trajectory patterns. These were subsequently validated in a second phase 3 placebo-controlled clinical trial (n = 576; n = 337 male) for vedolizumab-specific associations with disease remission, clearance and long-term disease-related complications. PRO trajectories were used to identify biologically distinct response groups by linking them to bulk RNA-seq data from the phase 3b clinical trial program for vedolizumab (n = 228; n = 147 male) versus adalimumab (n = 202; n = 107 male). This identified a vedolizumab-specific nonresponder molecular signature associated with SPP1-mediated vascular inflammation, which was confirmed in the second phase 3 vedolizumab clinical trial (n = 24; n = 14 male). Peripheral blood proteomics with the Olink Explore HT (over 5,400 protein panels; n = 280 adalimumab, n = 162 male; n = 279 vedolizumab, n = 172 male) and ELISA (n = 245 adalimumab, n = 138 male; n = 248 vedolizumab, n = 153 male) confirmed associations for SPP1 as a candidate biomarker with vedolizumab nonresponse. External publicly available clinical trial datasets (n = 252) and a routine practice cohort (n = 30; n = 11 male) were subsequently used to further validate the association for this vascular inflammation signature with nonresponse to vedolizumab. scRNA-seq datasets (n = 87; n = 41 male) identified macrophages–endothelial interactions as the cellular sources for this signature. Spatial single-cell transcriptomics in three separate UC cohorts (n = 5, n = 9, n = 16, respectively) confirmed associations between SPP1+ macrophage interactions and nonresponse to vedolizumab and identified the renin–angiotensin system as a key mediator of this vascular inflammation-mediated resistance program, leading to the clinical hypothesis that concomitant use of ACEi and ARBs may be associated with improved outcomes for vedolizumab. Two independent real-world UC cohorts derived from electronic medical records (n = 4,981) and claims data (n = 3,854) demonstrated that concomitant use of ACEi and ARB was associated with reduced long-term complications for vedolizumab but not anticytokine therapies, with no effect seen for concomitant CCB or BB. Figure created in BioRender; Dulai, P. https://BioRender.com/wbvvcde (2026).
We further aimed to link these trajectory group assignments with transcriptomics and proteomics data to determine whether response-trajectory-based classification could define mechanisms of vedolizumab nonresponse and leveraged real-world data to provide proof-of-concept associations that this resistance mechanism is targetable to improve disease outcomes.
Differential response to anti-integrin therapy in UC
Rectal bleeding and stool frequency are well-established PROs and surrogates of mucosal inflammation accepted by regulatory authorities for assessment of clinical disease activity in UC (PRO-2)17. Using GBMT on individual-level PRO-2 measurements in cohort 1 (VARSITY15, phase 3b comparative trial of vedolizumab versus adalimumab), we identified four distinct vedolizumab groups (V1–V4) characterized by different response dynamics—super-responders (V1; n = 170 (45%)), responders (V2; n = 42 (11%)), partial responders (V3; n = 101 (26%)) and nonresponders (V4; n = 69 (18%); Fig. 2a,b). Clustering of patients by response dynamics demonstrated robustness, with narrow confidence intervals (CIs; Fig. 2a,b and Supplementary Fig. 1), with most patients clustering around the center of the assigned group trajectory. This observation was confirmed in cohort 3 (validation cohort; GEMINI 1 (ref. 16), phase 3 vedolizumab registration trial) showing similar response-trajectory groups and robustness with narrow CIs (V1, n = 118 (20%); V2, n = 45 (8%); V3, n = 179 (31%); V4, n = 234 (41%); Fig. 2c and Supplementary Fig. 1). No associations were detected between response-trajectory groups and baseline characteristics across both cohorts (Extended Data Tables 1 and 2).
a–c, Disease response-trajectory plots for cohort 1—vedolizumab (n = 382; a), cohort 2—adalimumab (n = 386; b) and cohort 3—vedolizumab (n = 576; c) identify four discrete groups of patients for each compound. The four trajectories that are shown in each plot were termed super responders (V1 or A1), responders (V2 or A2), partial responders (V3 or A3) and nonresponders (V4 or A4). PRO-2 is the sum of rectal bleeding and stool frequency Mayo subscores for UC. For cohort 1, PRO-2 data were collected daily and averaged across weekly measurements to be reported every 2 weeks from baseline through week 6, and then every 8 weeks up to week 52. For cohort 2, PRO-2 data were collected daily and averaged across weekly measurements to be reported every 2 weeks from baseline through week 6, followed by every 4 weeks until week 50, with an additional measurement at week 52. Shaded curves represent CIs around trajectory estimates represented by dark lines. d, Proportion of participants in each trajectory of cohort 1 who achieved disease clearance at week 52, defined as achieving a Mayo rectal bleeding subscore of 0, Mayo stool frequency subscore of 0 or 1, Mayo endoscopic subscore of 0 or 1 (without friability), a full composite Mayo score of 0–2, histologic remission (either by Geboes score <2 or by Robarts Histopathology Index <3) and CRP and/or FCP within normal range. e, Kaplan–Meier curves for hospitalization and/or colectomy for participants stratified by disease trajectories V1 + V2 and V3 + V4 in cohort 2. Shaded curves represent CIs around HR cumulative point estimates represented by dark lines. Time-to-hospitalization and/or colectomy-free survival outcomes were evaluated using survival analysis techniques. HRs along with their corresponding 95% CIs were calculated using a Cox proportional hazards model. The statistical model was constructed and executed using JMP software. Kaplan–Meier survival curves were generated and visualized using the ‘ggsurvfit’ package in R.
PRO-based response-trajectory modeling in both vedolizumab cohorts independently demonstrated concordance of response dynamics, with achievement of 1-year symptomatic, endoscopic and/or histologic disease remission (Extended Data Table 3). This included association with the most stringent outcome of disease clearance at 1 year (Fig. 2d). Patients with faster response trajectories also showed improved long-term disease modification (reduced risk for hospitalization and/or colectomy) over 7 years in the validation cohort (V1–V2 versus V3–V4 hazard ratio (HR) = 0.56; 95% CI = 0.34–0.93; Fig. 2e). In contrast, the current US Food and Drug Administration and European Medicines Agency regulatory definition of endoscopic response (that is, endoscopic improvement (Mayo endoscopic subscore = 0–1) postinduction (week 6)) were not strongly associated with a reduced risk for hospitalization and/or colectomy over 7 years in the validation dataset (HR = 0.86; 95% CI = 0.55–1.35).
Differential response to anti-TNF therapy in UC
Overall efficacy for adalimumab at week 52 was significantly lower than vedolizumab in VARSITY15. PRO-based response-trajectory modeling for adalimumab also distinguished four groups (A1, n = 27 (7%); A2, n = 80 (21%); A3, n = 257 (67%); A4, n = 22 (6%)); however, the scale of response dynamics differed compared with vedolizumab (Fig. 2a,b). This was particularly notable for the adalimumab nonresponse group (A4) that showed a wider 95% CI than the vedolizumab nonresponse group (V4), and the adalimumab super-responder group (A1) that showed a more rapid slope of reduction in symptoms relative to the vedolizumab super-responder group (V1), albeit at a lower proportion (6% versus 18%). As with vedolizumab, PRO-based response-trajectory modeling of adalimumab groups demonstrated concordance of response dynamics with achievement of 1-year outcomes (Fig. 2d and Extended Data Table 3). These findings suggest that response-trajectory analysis detects meaningful differences at the patient level, and biological differences in treatment mechanisms (broader cytokine inhibition versus selective cell trafficking modulation) may give rise to the distinct profiles of predictive response-trajectory characteristics.
Molecular basis of PRO response-defined trajectory groups identifies endotypes
Combining the PRO-based response-trajectory groups with bulk RNA sequencing (RNA-seq) identified unique molecular signatures of nonresponse for vedolizumab but not adalimumab (Supplementary Fig. 2). These signatures were most clearly enriched during the postinduction period of 14 weeks in cohort 1 (Fig. 3a). Independent assessment of bulk RNA-seq linked to vedolizumab PRO-based response-trajectory groups in cohort 2 identified similar overlap in molecular nonresponse signatures; however, comparisons across all four response groups were less robust during this earlier postinduction time point of 6 weeks (Fig. 3b). The consistency in signatures across cohorts for vedolizumab nonresponse with increasing strength and robustness in associations during the later induction time point of 14 weeks suggests that treatment resistance mechanisms for vedolizumab evolve during exposure.
a,b, Heatmap showing the comparative analysis of enriched pathways by responder category in (a) VDZ-treated and ADA-treated patients in the clinical trial (VARSITY) cohort 1 (bulk RNA-seq data available for the following: V1 n = 25, V2 n = 109, V3 n = 56, V4 n = 38; A1 n = 15, A2 n = 42, A3 n = 135, A4 n = 10) and (b) VDZ-treated patients in the clinical trial (GEMINI 1) cohort 2 (bulk RNA-seq data available for the following: V1 n = 5, V2 n = 4, V3 n = 3, V4 n = 12). Results are generated from GSEA of the MSigDB Hallmark gene sets against the ranked DEGs across different comparisons. Values displayed in the heatmap represent NES for enriched pathways across nonresponder (V4 or A4) versus partial responder (V3 or A3), responder (V2 or A2) and super-responder (V1 or A1) groups at different time points (baseline, week 6 and week 14). Pathways with positive NES indicate activation in nonresponders, whereas pathways with negative NES represent inhibition in nonresponders (upregulation in responders). Asterisks denote pathways that are statistically significant, based on an adjusted P value <0.05. Nonstatistically significant results are grayed out. c, Heatmap representation of GSEA results for vascular inflammation and selected gene sets. Analysis was performed in UC clinical trial bulk-transcriptomics data with patients treated with IFX (GSE12251, n = 23; GSE16879, n = 24; GSE23597, n = 17; GSE73661, n = 23), ADA (VARSITY cohort 1), VDZ (VARSITY cohort 1 and GEMINI cohort 2) and UST (GSE206285, n = 165) as validation datasets for drug specificity in associations for vascular inflammation with vedolizumab nonresponse. d, Differential expression of vascular inflammation signature genes in nonresponders versus responders across transcriptomic and proteomic platforms in VARSITY. Heatmap displaying log2(FC) values for the significant genes only, comparing nonresponders versus responders for the primary endpoint. Data are shown for tissue bulk RNA-seq and serum proteomics (PX, Olink) at baseline and week 14 for both VDZ and ADA treatment arms. Positive values (red) indicate higher expression in nonresponders; negative values (blue) indicate higher expression in responders. *P < 0.05; **FDR < 0.05. e,f, Baseline SPP1 levels and association with the primary endpoint in VARSITY. Serum SPP1 concentrations were measured by Olink proximity assays (e; n = 559; n = 334 male) and ELISA assay (f; n = 493; n = 291 male). It shows density distributions comparing ADA versus VDZ treatment arms at baseline (week 0), followed by box plots comparing them across Rs versus NRs. Statistical comparisons were performed using Wilcoxon rank-sum tests; FDR-corrected P values (Benjamini–Hochberg) are shown for treatment comparisons. Baseline SPP1 levels showed similar distributions between ADA and VDZ arms in both ELISA and Olink datasets (FDR = 0.4 and FDR = 0.099, respectively). Notably, VDZ NRs showed significantly higher baseline SPP1 levels compared with Rs (ELISA, P = 0.018; Olink, P = 0.00014), whereas no significant difference was observed between ADA Rs and NRs at baseline. Separate multivariable logistic regression models were constructed for each treatment cohort—ADA and VDZ and each assay (Olink SPP1 or ELISA SPP1). The primary dependent variable was the binary clinical remission status achieved at week 52. Baseline SPP1 protein expression served as the primary continuous independent predictor. To stabilize variance and account for right-skewed biomarker distributions, ELISA SPP1 expression levels were log2-transformed before model entry. Olink SPP1 was already in log2 scale. To isolate the independent predictive effect of baseline SPP1 and obtain unbiased estimates, both treatment-specific models were adjusted for prior anti-TNF exposure and concomitant steroid exposure, and smoking status. The statistical model was constructed and executed using JMP software. Box plots display the median (centerline), the 25th and 75th percentiles (box bounds, that is, IQR) and whiskers extending to the most extreme value within 1.5× IQR of the hinges. Individual points represent single patients (responders or nonresponders). P values are from two-sided Wilcoxon rank-sum tests (unpaired). VDZ, vedolizumab; ADA, adalimumab; MSigDB, Molecular Signature Database; NES, normalized enrichment scores; IFX, infliximab; UST, ustekinumab; Rs, responders; NRs, nonresponders; IQR, interquartile range.
To ensure molecular response signatures identified were not due to differential achievement in phenotypic response at varying postinduction time points across response-trajectory groups, sensitivity analyses were performed to account for endoscopic and mucosal inflammation status postinduction. These yielded identical results in cohort 1 to the main analysis (Supplementary Fig. 3). Differentially expressed genes (DEGs) demonstrated that vascular inflammation (SPP1, STC1, TIMP1, CXCL6, OLR1, THBD, PGLYRP1, PRG2, FSTL1, KCNJ8, FGFR1, PF4, CCND2 and MSX1; Extended Data Table 4) was the key postinduction resistance pathway for vedolizumab in both cohorts. Initial evaluation of baseline bulk RNA-seq data did not reveal clear pathway associations with response groups; however, re-evaluation of baseline bulk RNA-seq data using this postinduction vascular inflammation signature demonstrated an association with nonresponse to vedolizumab. Application of this vascular inflammation signature to additional anti-TNF (adalimumab15 and infliximab18,19,20,21,22) and anti-IL-12/IL-23 (ustekinumab23) trial datasets demonstrated no association with response status for anticytokine therapies (Fig. 3c).
Given the evolution of this molecular mechanism of resistance during treatment induction, we then studied dynamic variations and evolutions in the vascular inflammation gene signature during induction in a small, prospectively recruited, independent and densely sampled longitudinal cohort of ADT-naive patients with inflammatory bowel disease (IBD) treated with vedolizumab (n = 18) or an anti-TNF therapy (infliximab, n = 12; Extended Data Fig. 1; cohort extended from ref. 24). Vedolizumab induced dynamic tissue gene expression changes within hours, becoming more pronounced at weeks 6 and 14, mirroring the postinduction time points of cohorts 2 and 1, respectively. Angiogenesis and vascular inflammation pathway enrichment were observed in upregulated transcripts in vedolizumab nonresponders, but not in infliximab nonresponders, with variable dynamics in change albeit the dynamics of change were variable across patients. Targeted analysis of selected transcripts found in cohorts 1 and 2 for vascular inflammation demonstrated an association with response status for vedolizumab in UC, but not for infliximab, with dynamic patient-to-patient variation in patterns (Extended Data Fig. 2).
We then analyzed peripheral blood proteomic data with Olink (5K explore HT panel) in cohort 1 for orthogonal confirmation of tissue RNA signatures and to determine whether certain genes and/or proteins were primarily associated with vedolizumab nonresponse. Among the genes included in the vascular inflammation bulk RNA-seq signature, secreted phosphoprotein 1 (SPP1, also known as osteopontin) was the key gene that demonstrated consistency with peripheral blood proteomics at baseline and week 14 for associations with nonresponse to vedolizumab but not adalimumab (Fig. 3d). Olink-based and ELISA-based assessment of SPP1 in serum from cohort 1 confirmed that this association was not due to baseline variation in expression across the treatment arms (Fig. 3e,f). After adjusting for confounders known to influence treatment efficacy (prior anti-TNF exposure and concomitant steroid exposure) or SPP1 expression (smoking status), baseline elevations in SPP1 expression were associated with a reduced likelihood for achieving week 52 clinical remission with vedolizumab (Olink—odds ratio (OR) = 0.534, 95% CI = 0.352–0.787; ELISA—OR = 0.66, 95% CI = 0.457–0.94) as compared with adalimumab (Olink—OR = 0.771, 95% CI = 0.508–1.17; ELISA—OR = 0.862, 95% CI = 0.613–1.218)15,25,26.
Cellular resolution of the vascular inflammation signature in vedolizumab nonresponse
To better understand the cellular source of this SPP1-driven vedolizumab treatment resistance mechanism, the vascular inflammation gene signature was converted to a module score and applied to publicly available tissue single-cell RNA-seq (scRNA-seq) UC datasets27,28. The vascular inflammation module score was significantly enriched in key cells involved in the inflammatory cascade in IBD—fibroblasts, inflammatory monocytes and endothelial cells (Fig. 4a and Extended Data Fig. 3a,b).
a, Differences in vascular inflammation module scores for inflammatory monocytes (IBD inflamed—n = 38 patients, 6,224 cells; IBD noninflamed—n = 29 patients, 929 cells; healthy—n = 3 patients, 79 cells). Inflamed versus noninflamed, Padj = 4.2 × 10−16; inflamed versus healthy, Padj = 8.3 × 10−11; noninflamed versus healthy, Padj = 0.58. b, Dot plot showing vascular inflammation module scores and relative expression of selected genes in SPP1+ (2) and SPP1− (1) monocytes/macrophages across three datasets—PBMC scRNA-seq cohort (total n = 10 patients; monocyte 1 = 28,537 cells; monocyte 2 = 27 cells), rectal tissue scRNA-seq cohort (total n = 8 patients; MoMac 1 = 2,177 cells; MoMac 2 = 14 cells), VDZ CosMx cohort 1 (n = 5 patients; MoMac 1 = 782 cells; MoMac 2 = 47 cells) and VDZ CosMx cohort 2 (n = 9 patients; MoMac 1 = 972 cells; MoMac 2 = 100 cells). Dot color intensity reflects average expression level, with red indicating higher values and gray indicating lower values. Dot size indicates the percentage of cells expressing the corresponding genes. c, Differences in integrin β7 (left) and integrin α4 (right) protein expression in SPP1+ (29 cells) and SPP1− macrophages (1,358 cells) across inflamed (CD, n = 4; UC, n = 4), noninflamed (CD, n = 4; UC, n = 1) and healthy control (n = 1). Integrin β7 SPP1+ macrophage versus SPP1− macrophage—Padj = 0.0024; integrin a4 SPP1+ macrophage versus SPP1− macrophage—Padj = 0.00015. Box plots in a and c show the median (centerline), the 25th and 75th percentiles (box bounds) and whiskers extending to the most extreme value within 1.5× IQR; points beyond are plotted individually. Significance levels assessed using Wilcoxon rank-sum test, adjusted for multiple comparisons using the Benjamini–Hochberg method. **P < 0.01, ***P < 0.001, ****P < 0.0001. d,e, Scatterplot showing incoming and outgoing interaction strength of cell types in nonresponder and responder conditions based on VDZ CosMx cohort 1 (d, left; n = 5 patients) VDZ CosMx cohort 2 (d, right; n = 9 patients), and whole transcriptomic VDZ CosMx cohort 3 (e; n = 16 patients) spatial transcriptomics data. Each dot represents a specific cell type, and dot size reflects the number of significant ligand–receptor interactions. f, Line plot showing spatial co-occurrence probability of other cell types with SPP1+ macrophages in a baseline VDZ nonresponder sample (FOV 75) from VDZ CosMx cohort 2. Values above 1 indicate an increased probability of spatial proximity compared with random expectation. g, Dot plot showing pathways of interest enriched among HSVG pretreatment in VDZ nonresponders from VDZ CosMx cohorts 1 and 2. HSVG calculated using nnSVG across all FOVs, and pathway enrichment calculated using Metascape and the cumulative hypergeometric distribution (one-sided, testing over-representation) corrected for multiple testing by the Benjamini–Hochberg procedure. Dot size represents the number of genes overlapping with each pathway, the x axis shows the gene hit ratio of the given pathway and dot color indicates enrichment significance, shown as −log10(P). h, Dot plot showing Stouffer’s weighted Z scores for spatial colocalization between cell types of interest and endothelial cells in the VDZ CosMx cohort 3, stratified by baseline VDZ NRs (n = 8) and Rs (n = 8). SPP1+ macrophages NR Z = 2.68, R Z = –2.45, lymphocytes NR z = −59.36, R z = −37.91, granulocytes NR z = 20.77, R z = 16.82. i, SPP1 and ANGPT2 centered subset of the Metascape-identified PPI network. Nodes represent proteins encoded by the input genes, and edges represent known or predicted PPIs. Node size indicates the number of direct connections each node has in the full PPI network. j, Relative expression of ICAM-1 and MAdCAM-1 in cellular neighborhoods containing SPP1+ AGT+ macrophages compared with baseline expression per sample. Positive log(FC) represents enrichment of these molecules near SPP1+ AGT+ macrophages and negative log(FC) represents depletion. Statistical significance was assessed using a Wilcoxon signed-rank test to determine whether the mean per-sample log(FC) differed from zero within each group (NR, n = 8; R, n = 8). From left to right, P = 0.008, 0.195, 0.250 and 0.742. k, Bar plot comparing the per-sample percentage of SPP1+ macrophages that are AGT+ and located in CTSD+ neighborhoods in NR (n = 8) versus R (n = 8). Neighborhood CTSD expression was determined per cell as the spatial lag of CTSD in the k = 15 nearest neighbors. Bars represent group means with error bars. Statistical significance was assessed using a binomial generalized linear mixed model (**P = 0.008). NK, natural killer; NS, not significant; HSVG, highly spatially variable genes.
Using scRNA-seq and CosMx spatial scRNA data from publicly available external routine practice vedolizumab-treated UC cohorts29,30,31, we observed two unique monocyte populations in the peripheral blood and two similarly unique monocyte–macrophage populations in the tissue that were differentiated by their vascular inflammation module scores (Fig. 4b). The two populations of peripheral blood monocyte and tissue monocyte–macrophage populations also had differences in RNA expression for integrin subunit α4 (ITGA4) and integrin subunit β7 (ITGB7), the integrin targets for vedolizumab (anti-α4β7). A key marker for monocytes and monocyte–macrophages with high vascular inflammation module scores, and low RNA expression for α4 and β7, was SPP1 (refs. 32,33). Using newly generated scRNA-seq and cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) data from mucosal biopsies in patients with IBD (n = 14; Extended Data Fig. 4a,c), we confirmed that SPP1+ monocyte–macrophages had significantly lower protein expression for integrin α4 and β7 (Fig. 4c), and a significant interaction with endothelial cells (Extended Data Fig. 4d,e). Spatial multi-omic analyses confirmed spatial colocalization of SPP1+ monocyte–macrophages (CD163+ CD68+) with endothelial cells (CD31+) in vedolizumab nonresponders. These spatially colocalized endothelial cells were strongly positive for an alternative integrin, intracellular adhesion molecule (ICAM)-1 (Extended Data Fig. 4f).
Spatial analysis links the interaction of SPP1+ macrophages with the vascular niche to vedolizumab nonresponse
To understand how SPP1+ monocyte–macrophages may promote vedolizumab treatment resistance, we then investigated single-cell spatial interactions in vedolizumab-treated patients with UC. Using two publicly available external cohorts (vedolizumab CosMx cohort 1, n = 5; vedolizumab CosMx cohort 2, n = 9), we observed a stronger cellular interaction for SPP1+ monocyte–macrophages in nonresponders compared with responders pretreatment (Fig. 4d). To further validate this observation, we generated new spatial single-cell whole-transcriptome data from vedolizumab-treated patients with UC (vedolizumab CosMx cohort 3, n = 16; Extended Data Fig. 5a,c) and again observed a stronger cellular interaction for SPP1+ monocyte–macrophages in nonresponders compared with responders pretreatment (Fig. 4e). Spatial proximity confirmed spatial colocalization of SPP1+ monocyte–macrophages with endothelial cells in vedolizumab nonresponders (Fig. 4f).
Orthogonal evaluation of spatial data using spatial gene variability analyses identified angiogenesis and blood vessel morphogenesis as key pathways enriched in vedolizumab nonresponders compared with responders pretreatment in vedolizumab CosMx cohorts 1 and 2 (Fig. 4g). The increased spatial variability in these pathways demonstrates that vedolizumab nonresponders have unique pretreatment spatial niches related to vascular inflammation and remodeling driven by SPP1+ macrophage interactions. Indeed, in the vedolizumab CosMx cohort 3, we validated that SPP1+ macrophage–endothelial spatial niches were significantly depleted at pretreatment in responders and significantly enriched at pretreatment in nonresponders (Fig. 4h).
Vedolizumab nonresponders have upregulation of angiotensin pathway-mediated vascular inflammation
Predicted protein–protein interaction (PPI) analyses of spatially variable genes in vedolizumab nonresponders from CosMx cohorts 1 and 2 identified interactions with multiple alternative integrins, including α5 and β1 (Fig. 4i and Extended Data Fig. 5d). A monocyte–fibronectin interaction through α5β1 has been demonstrated to promote CXC chemokine-dependent angiogenic activity from monocytes34 and SPP1–α5β1 interactions have been demonstrated to induce PI3K signaling, which was one of the most spatially enriched pathways pretreatment in vedolizumab nonresponders35 (Fig. 4g,i). In colorectal cancer, SPP1 has been demonstrated to alter tumor-associated macrophage programming through the PI3K/AKT/CSF1 axis36. Another notable PPI for vedolizumab nonresponders was for ANGPT2 (angiopoietin 2) with CSF1R (colony-stimulating factor 1 receptor), which is known to promote pathogenic angiogenesis through macrophage interactions with endothelial cells37,38,39.
Cardiovascular research has demonstrated that ANGPT2, α5β1, PI3K and CSF1 pathway-mediated interactions have a critical role in pro-inflammatory macrophage polarization and cardiac remodeling through endothelial interactions after myocardial infarction40,41. Recruited macrophages interacting with endothelial cells at the site of injury regulate remodeling through autocrine or paracrine production of angiotensin II, which has been demonstrated to stimulate expression of alternative adhesion molecules (ICAM-1)42,43,44,45. Furthermore, the renin–angiotensin system is upregulated during differentiation of monocytes to macrophages46, and inhibition of the angiotensin-converting enzyme (ACE) with accompanying reductions in angiotensin II has been demonstrated to attenuate release of monocytes from peripheral blood reservoirs, inhibit monocyte adhesion to vascular endothelium and modulate monocyte function47,48,49. Angiotensin II has also been demonstrated to induce SPP1 and ANGPT2, and angiotensin II inhibitors reduce colitis in experimental models50,51,52,53,54,55,56,57,58. These data support the hypothesis that the renin–angiotensin system may have a central role in SPP1 macrophage-mediated resistance through upregulation of alternative trafficking pathways (ICAM-1) that bypass the target of vedolizumab (α4β7:MAdCAM-1).
We observed pretreatment SPP1+ macrophages in vedolizumab nonresponders to have significantly higher expression of angiotensinogen (AGT) compared with responders (P = 0.038; patient-level comparison), with no significant differences seen between responders and nonresponders for AGT expression in SPP1− macrophages pretreatment. These pretreatment SPP1+ macrophages in vedolizumab nonresponders, however, had significantly lower expression for renin (REN; P = 0.036; patient-level comparison), with no significant differences seen for REN in SPP1− macrophages. Because renin is the canonical rate-limiting enzyme for this reaction, the combination of high substrate (AGT) with low renin in the same cells suggests that, where the pathway is locally active, angiotensin I may be generated through a renin-independent route. Although renin has been reported to be upregulated during inflammation in UC56, our observation is specific to SPP1+ macrophages and does not exclude renin activity in other tissue compartments.
A plausible renin-independent candidate is cathepsin D (CTSD), an aspartic protease structurally related to renin that can cleave AGT and, unlike renin, has an acidic optimum pH well-suited to the inflamed, hypoxic and acidotic microenvironment of active UC. Intestinal macrophages in IBD have been demonstrated to have induction and significant upregulation of CTSD specifically during inflammation59,60,61, and inhibition of CTSD has been demonstrated to ameliorate experimental colitis59. The AGT+ SPP1+ macrophage–endothelial spatial niches enriched pretreatment in vedolizumab nonresponders and depleted in vedolizumab responders were indeed noted to have significant spatial enrichment for CTSD expression (Fig. 4j). Vedolizumab nonresponders had significantly higher relative expression for ICAM-1 versus MAdCAM-1 in endothelial cells pretreatment (P = 0.023; patient-level comparison), with no significant difference seen for relative expression of these adhesion molecules in vedolizumab responders pretreatment (P = 0.844; patient-level comparison). Furthermore, the AGT+ SPP1+ macrophage–endothelial spatial niches enriched pretreatment in vedolizumab nonresponders were noted to have significant spatial enrichment for ICAM-1 relative to MAdCAM-1 expression (Fig. 4k).
The gut-selective regulation of cellular trafficking by vedolizumab makes it an ideal model system for exploring systems immunology in UC. Our reverse human-translational study has uncovered a specific relationship among macrophages, vascular inflammation and vedolizumab nonresponders in UC, where local tissue angiotensin production may promote MAdCAM-1-independent leukocyte trafficking that could contribute to disease progression. When combined with well-established cardiology research and known biology, these data suggest that concomitant use of angiotensin II inhibitors (angiotensin-receptor blockers (ARBs) and/or ACE inhibitors (ACEi)) may improve outcomes in vedolizumab-treated patients with UC.
Targeting vascular inflammation through the renin–angiotensin system to improve disease outcomes for vedolizumab in UC
Small retrospective cohort studies in patients with IBD have been conflicting, with some studies suggesting a benefit for RAS inhibitors in IBD primarily for Crohn’s disease (CD)62,63,64, and others suggesting a lack of benefit and potentially even harm for UC63,65. This heterogeneity in prior research regarding the RAS in IBD has made it difficult to fully define potential benefits of targeting this pathway for treatment66. These prior studies are likely confounded by relatively small sample sizes, studying the concomitant use of RAS inhibitors with multiple therapies and biologics, and/or patient-specific confounders for why the medications may have been used relative to comparator cohorts. To overcome these prior limitations, we leveraged two large national cohorts of patients with UC and hypertension being treated with biologic therapy (n = 4,981 and n = 3,854) to understand the vedolizumab-specific impact of concomitant RAS inhibition on treatment outcomes in UC.
Using propensity score matching in a large multicenter, routine practice, electronic medical record-based cohort of vedolizumab-treated patients with UC with hypertension, we compared outcomes between those receiving vedolizumab (n = 1,022) with or without an ACEi or ARB (n = 325 patients per group after propensity score matching). The concomitant use of an ACEi or ARB with vedolizumab was associated with significantly higher rates of treatment persistence, lower rates of hospitalization for UC flares and lower rates of colectomy (Fig. 5). As a negative control analysis, we similarly compared outcomes for vedolizumab-treated patients with UC with hypertension with or without a β-blocker (BB; n = 880; n = 214 per group after propensity score matching) and vedolizumab-treated patients with UC with hypertension with or without a calcium channel blocker (CCB; n = 914; n = 167 per group after propensity score matching). Concomitant use of a BB (adjusted HR = 1.18; 95% CI = 0.89–1.56) or a CCB (adjusted HR = 1.07; 95% CI = 0.77–1.47) had no observed impact on risk for hospitalization and/or colectomy over 3 years. To further account for residual confounding by indication, we used a positive control analysis for alternative biologics including anti-TNF therapy (n = 1,823; n = 507 patients per group after propensity score matching) and anti-IL-12/IL-23 therapy (ustekinumab; n = 342; n = 100 patients per group after propensity score matching). This demonstrated no significant difference in these outcomes when using an ACEi or ARB alongside anticytokine therapy for patients with UC with hypertension (Fig. 5).
a–i, Comparative analysis of persistence on biologic therapy (a–c), occurrence of hospitalization for UC flare (d–f) and need for colectomy (g–i) in a multicenter routine practice UC cohort with hypertension treated with or without ACEi or ARBs. Biologic-treated UC cohorts included vedolizumab (n = 1,022), anti-TNF therapies (n = 1,823) and ustekinumab (n = 342). Patient-level data from electronic medical records were used to create propensity-matched cohorts (vedolizumab, n = 325 per group; anti-TNF, n = 507 per group; ustekinumab, n = 100 per group) with 1:1 matching to account for age, sex, race, diabetes mellitus, nicotine dependence, obesity, ischemic heart disease, heart failure, stroke, chronic kidney disease, chronic lower respiratory diseases, primary sclerosing cholangitis and other autoimmune disorders, UC-specific medications, any prior oral or intravenous steroid use before biologic initiation, hemoglobin, CRP and calprotectin as measures of mucosal inflammatory status. Treatment persistence was defined as the need for a switch to alternative ADT (biologic or small molecule); hospitalization for UC flare was defined as the need for hospitalization and initiation of intravenous steroids for a diagnosis of UC; and colectomy was defined as the need for total colectomy. Outcomes are reported as adjusted HRs (Cox proportional hazards model; solid lines) with 95% CI (shaded bands) for the estimate comparison of biologic and ACEi/ARB versus biologic without ACEi/ARB. Significance was only noted for outcomes in patients with UC treated with vedolizumab and ACEi/ARB versus vedolizumab without ACEi/ARB (a, d, g). P values were obtained from a two-sided Wald test with no adjustment for multiple comparison.
A second validation in a large, national, claims-based cohort of patients with UC with hypertension treated with vedolizumab (n = 980), anti-TNF (n = 2,460) or ustekinumab (n = 414) therapy confirmed that concomitant ACEi or ARB treatment with vedolizumab was associated with a significant reduction in colectomy at 1 year (adjusted HR = 0.31; 95% CI = 0.11–0.87, P = 0.02). Concomitant ACEi or ARB treatment with anti-TNF or ustekinumab therapy was not associated with a reduction in colectomy risk (Fig. 6).
a–c, Kaplan–Meier curves show colectomy-free survival over 1 year of follow-up for patients receiving (a) vedolizumab (n = 490 per group), (b) anti-TNF therapy (n = 1,230 per group) or (c) ustekinumab (n = 207 per group) matched with propensity score. Patients were stratified into two groups—those with concomitant ACEi/ARB use and those without. The index date was defined as the date of first biologic prescription, and outcome ascertainment started 4 weeks after the index date. One-to-one propensity score matching was used to balance baseline covariates. Covariates included age at biologic initiation, sex, race, diabetes mellitus, nicotine dependence, obesity, ischemic heart disease, heart failure, stroke, chronic kidney disease, chronic lower respiratory diseases, primary sclerosing cholangitis, other autoimmune diseases, prednisone usage, intravenous steroid usage and other biologics. Adjusted HRs, 95% CI and P values were estimated using Cox proportional hazards models. Lower colectomy-free survival indicates a higher risk of total colectomy. Solid lines represent Kaplan–Meier estimates of colectomy-free survival, and shaded bands represent the corresponding 95% CIs.
Discussion
Although vedolizumab offers a highly favorable safety profile because of its gut selectivity, personalized use to maximize treatment efficacy remains elusive. Molecular profiling studies have identified host and host–microbe mechanisms that may be associated with response to vedolizumab in IBD; however, to date, none of these observations have resulted in actionable therapeutic algorithms to improve vedolizumab efficacy19,67,68,69,70,71,72. This lack of progress in precision medicine for UC likely reflects constraints in traditional research designs that rely on binary responder classifications at landmark time points and ignore individual differences in the dynamics and depth of efficacy signals when analyzing the molecular framework of drug response in a heterogeneous population. Here we advocate for an individual classification of patients based on a response-trajectory-based framework that treats therapeutic success for each patient as a continuous dynamic phenotype24,73. This framework aligns with established prognostic principles in infectious disease, oncology and psychiatry, where early response kinetics guide therapeutic decision-making74,75,76,77,78,79,80. Despite the growing body of evidence implicating dysregulated angiogenesis and vascular inflammation in UC, the importance of these pathways and targeting them to optimize disease outcomes has not been fully explored. Vedolizumab targets the interaction of α4β7 with MAdCAM-1 on intestinal endothelial cells to inhibit immune-cell trafficking to the gut24,73. We observed that SPP1+ monocytes in peripheral blood and SPP1+ macrophages in mucosal tissue had significantly lower α4β7 integrin expression. Across multiple cohorts, we observed consistency in associations among vascular inflammation, SPP1+ macrophage–endothelial interactions and nonresponse to vedolizumab. This indicates an expanded α4β7-independent inflammatory program at baseline that could sustain cellular interactions in the gut through alternative trafficking routes, leading to dysregulated vascular inflammation as a targetable mechanism to improve treatment response. Our proof-of-concept demonstration that these pathways may be targetable with ACEi and ARBs, to optimize disease outcomes in vedolizumab-treated UC patients, provides support for further research and drug development. Dedicated clinical trials for repurposing available antihypertensive treatments for precision combination therapy in UC will clarify whether these agents boost vedolizumab activity as a combination therapy.
Although our study has many strengths, including access to individual data of well-documented large prospective clinical trial programs, high-dimensional multi-omics data and multiple validation cohorts, it is not without limitations. Our single-cell and spatial omics datasets are of limited sample size and downstream observations related to vascular inflammation and use of therapies targeting renin–angiotensin should be considered hypothesis generating. A routine practice single-center cohort study has demonstrated feasibility for application of our PRO-trajectory modeling with similar response strata81, but further validation in broader populations is needed to capture the known heterogeneity seen between clinical practice and clinical trial populations for UC82. Our main base of evidence comes from prospective, randomized clinical trials that, by concept, minimize confounding through randomization and the application of stringent inclusion/exclusion criteria. Although this approach is well established for controlling confounders and strengthening validity, it may also exclude large numbers of patients, potentially limiting the extent to which the study population reflects real-world clinical practice. On the contrary, real-world evidence derived from cohorts such as Optum or TriNetX captures routine clinical care and therefore may be more representative of broader patient populations. However, treatment selection in these datasets is based on the physician’s choice, which could be influenced by unmeasured or unrecorded confounders. Therefore, definitive confirmation that targeting vascular inflammation optimizes response to vedolizumab specifically will require further prospective randomized clinical trials. Ideally, these studies would be pragmatic in nature and combined with further validation of our PRO-trajectory modeling approaches and could establish individual adaptive strategies using early time points of measurement for PROs for individual patient-level treatment guidance. Peripheral proteomics data from our clinical trial suggest that SPP1 may be a potential biomarker to guide such treatment guidance; however, this requires further validation in a prospective manner. Further evaluation of this conceptual framework and observations will be needed in CD, particularly when subsetting the condition by ileal and colonic involvement, to understand whether certain phenotypic subtypes are primarily driven by this mechanism83,84. Finally, we could not identify clear molecular signatures of nonresponse for adalimumab and further refinement is needed to identify molecular changes that might be unique to each treatment group and allow for personalization across various mechanisms of action.
In conclusion, our findings support a taxonomic model for UC in which early clinical response dynamics, underpinned by distinct molecular programs, may inform adaptive treatment strategies including early switching or nonintuitive combination therapies in patients with unfavorable trajectories. This approach may ultimately enable discovery of new ADTs, the repurposing of drugs that enhance ADT efficacy and the use of precision-guided adaptive trials of individual combinations of ADTs in immune-mediated inflammatory disorders of the gastrointestinal tract.
Methods
Our study complies with all relevant ethical regulations. Mucosal biopsies for bulk RNA-seq and blood for proteomics from the VARSITY trial were obtained during the original clinical trial. Mucosal biopsies for CITE-seq and CosMx spatial single-cell transcriptomics were obtained through the Northwestern Digestive Health Institute Advisory Council IBD Biorepository, approved by the Northwestern University Institutional Review Board (STU00203172). Mucosal biopsies for bulk RNA-seq from the longitudinal cohort were obtained through the University of Kiel, approved by the ethics committee of the Christian-Albrechts-Universität zu Kiel (A 124/14 and AZ 156/03-2/13). Donors provided written informed consent for the use of their samples and associated data in the present study, including sharing of data, before sample collection and analysis. Appropriate protocols governing sample use were reviewed and approved by the respective institutional review boards listed above before conduct of experiments. Sex and/or gender were not considered in the design of the study.
Response trajectories and PRO-2-defined subgroups
The VARSITY trial (cohort 1) is a double-blind, double-dummy, controlled study in which 769 patients with moderately to severely active UC were randomized to receive vedolizumab or adalimumab15. Data from patients who received vedolizumab treatment in the GEMINI 1 trial (cohort 2; n = 591), where the drug was examined against placebo, were used for replication16. Longitudinal PRO-2 data (rectal bleeding and stool frequency from the adapted partial Mayo score) were extracted for trajectory modeling using the GBMT R package14, with models being fitted separately for each drug using established criteria and metrics85. The four-group structure for vedolizumab was replicated in cohort 2 (GEMINI 1). The long-term safety (LTS) data from the GEMINI 1 trial (n = 450) were used for further validation that our response-trajectory subgroups predicted long-term disease outcomes86. Baseline demographics, outcome definitions and clinical outcomes from these datasets have previously been published, with comparability in disease characteristics across trials15,16,86.
Molecular transcriptomic analyses to define vedolizumab-specific endotypes
Colonoscopy-derived or sigmoidoscopy-derived biopsies were collected for VARSITY, GEMINI 1 and GEMINI LTS with standardized collection protocols. Illumina strand-specific RNA-seq with rRNA depletion libraries were prepared and sequenced with 150 bp paired-end reads (2 × 150 bp) for VARSITY samples, achieving approximately 50 million sequencing depth per sample. VARSITY RNA-seq data underwent quality control and preprocessing, including adaptor removal, genome alignment and gene quantification, using the OmicSoft platform87. The hg38 human reference genome and gene annotations from Ensembl were used for alignment and quantification. GEMINI 1 and LTS gene expression datasets, accessible at the Gene Expression Omnibus (GEO) database under accession GSE73661, were generated through Affymetrix GeneChip Human Gene 1.0 ST arrays (Thermo Fisher Scientific)19. Sequencing data were linked to patient-level trajectory group assignments to determine trajectory-specific DEGs using DESeq2 for VARSITY samples and the limma package at the probe level for GEMINI 1 and LTS microarray data88, followed by gene set enrichment analysis (GSEA) using the Molecular Signature Database Hallmark gene sets89,90. This was done separately for both trial datasets with multiple testing correction applied using the Benjamini–Hochberg false discovery rate (FDR) method91. DEGs were reported as upregulated if they had a positive log2 fold change (FC) and FDR-adjusted P value <0.05 and vice versa for downregulated genes89,90. Genes were preranked using the sign of the log2(FC) values, thereby capturing the directionality of expression changes between groups. Gene sets were considered significantly enriched if the FDR-adjusted P value was <0.05.
Vedolizumab-specific gene modules (specifically angiogenesis, epithelial-mesenchymal transition, TNF, interferon, IL-6, JAK/STAT signaling) and their composition are provided in Extended Data Table 4. Public bulk RNA-seq datasets used to determine specificity in associations to vedolizumab included UC patients treated with infliximab and ustekinumab—(1) anti-TNF GSE73661; (2) anti-TNF GSE23597 and anti-TNF GSE12251 with patients enrolled in the ACT1 phase 3 clinical trial; (3) anti-TNF GSE16879; and (4) anti-UST GSE206285 with patients enrolled in the UNIFI phase 3 clinical trial18,20,22,92. We evaluated gene module enrichment across responders and nonresponder-treated patients. The differential gene expression results comparing nonresponders with responders (the reference group) at baseline were processed to perform a GSEA using the preranked DEGs based on the log2(FC) values.
Cohort recruitment and temporal gene-signature analysis in vedolizumab-treated and anti-TNF-treated patients with UC
In a small, densely sampled cohort of biologic-naive clinically active patients with IBD (18 receiving vedolizumab (10 with CD and 8 with UC) and 12 receiving infliximab (4 with CD and 8 with UC)) undergoing first-time ADT, we further explored the dynamics of response. In this cohort, biopsies were sampled at baseline, 2 weeks, 6 weeks and 14 weeks. In selected patients, early biopsies 24–72 h postinduction were also available. Remission was assessed clinically at 14 weeks based on Harvey–Bradshaw Index (≤4) for CD and total Mayo score (≤2, bleeding 0 and endoscopy ≤1) for UC.
Total RNA was isolated from whole biopsies according to the manufacturer’s instructions using the mirVana miRNA isolation kit (Life Technologies). Strand-specific paired-end RNA-seq libraries were prepared from total RNA using the TruSeq Stranded Total RNA kit (Illumina) and sequenced on an Illumina HiSeq 4000 platform (2 × 75 bp). Low-quality bases were filtered with the Illumina FASTQ filter, and adaptors and low-quality reads were trimmed using Trim Galore (v0.4.4), discarding sequences shorter than 35 bp. Reads were aligned to the human genome (GRCh38; v25) with STAR (v2.5.2b). Gene counts were obtained using featureCounts (v1.5.2) and normalized using the DESeq method.
Differential gene expression analysis was performed with DESeq2 (v1.40.2) using two approaches. Correction for multiple testing was performed using the Benjamini–Hochberg method, and genes with an adjusted P value <0.05 were considered significant. Functional enrichment was assessed with topGO (v2.46.0) using genes with a similar expression level as the background.
Vascular inflammation module cell-type attribution analysis
We used two publicly available scRNA-seq cohorts27,28 to investigate the clinically derived vascular inflammation module enrichment in different cell types. Here 10x-formatted count matrices and annotation information were downloaded from Zenodo (https://doi.org/10.5281/zenodo.13768607) and Single Cell Portal (SCP259). The two datasets were processed, and original annotations were added using Seurat (v5). Module scores representing vascular inflammation clinical features were calculated using the AddModuleScore function in Seurat.
Module scores were summarized at the sample level by inflammation status (inflamed, noninflamed and healthy), by disease (UC, CD and healthy controls), and/or by cell type27. Linear mixed-effects model was fitted comparing the mean module score for the target cell type versus all other cell types. The model included fixed effects for target cell-type inflammation status and condition, with a random intercept for sample to account for repeated observations from the same sample. Effect sizes were reported as estimated differences with 95% CIs. P values were adjusted across tested cell types using the Benjamini–Hochberg method. Cell types were considered enriched if they showed a positive estimated difference and Benjamini–Hochberg-adjusted P value <0.05.
CITE-seq data generation and analysis in vedolizumab nonresponders
CITE-seq was performed and processed on ileocolonic mucosal biopsies from patients with UC and CD (n = 14) at the Northwestern University NUSeq Core Facility using standard experimental processes for sample preparation and sequencing as recommended by 10x Genomics. Fresh biopsy specimens were collected in 1× Hanks′ balanced salt solution buffer (Gibco, Life Technologies). The buffer was carefully aspirated and tissue transferred to a small vial, then 500 µl of Liberase plus DNase I cocktail (MilliporeSigma) was added and the tissue was thoroughly minced using fine dissecting scissors. The tissue suspension was passed through a pipette tip ten times and incubated in a thermomixer at 37 °C at 800 rpm. The tube was centrifuged, and the resulting supernatant was passed through a 50 µl Flacon tube strainer. The tissue pellet was resuspended in another 500 µl of Liberase plus DNase I cocktail and again passed ten times through a pipette tip. The suspension was again incubated in a thermomixer at 37 °C at 800 rpm. After a second centrifugation, the supernatant was collected and passed through the 50 µl Falcon tube filter. The tissue pellet was again resuspended in Liberase plus DNase I cocktail and passed ten times through a pipette tip. A third centrifugation followed, with the supernatant again being passed through a 50 µl Falcon tube filter. The tissue pellet was resuspended in another 500 µl of Liberase plus DNase I cocktail and again passed ten times through a narrower pipette tip and then passed through the 50 µl Falcon tube filter. The cumulative post 50 µl Falcon tube filtrate containing a largely single-cell suspension was centrifuged and the supernatant was aspirated. The cell pellet was resuspended in 200 µl of cold 1× PBS as well as 2% BSA. The sample was transferred immediately to Northwestern University NUSeq Core Facility for cell counting and subsequent processing.
Cell number and viability were analyzed using Nexcelom Cellometer Auto 2000 with acridine orange and propidium iodide fluorescence staining method. A total of 16,000 cells per sample were loaded into the Chromium Controller (10x Genomics, PN-120223) on a Chromium Next GEM Chip K (10x Genomics, PN-1000127) and processed to generate single-cell gel beads in the emulsion (GEM) according to the manufacturer’s protocol. The cDNA and library were generated using the Chromium Next GEM Single Cell 5′ Reagent Kits (v2; 10x Genomics, PN-1000283) according to the manufacturer’s manual. In addition, cell surface protein and human T cell V(D)J libraries were constructed using 10 × 5′ Feature Barcode kit (10x Genomics, PN-1000256) and Chromium Single Cell Human TCR and BCR Amplification kit (10x Genomics, PN-1000252). The multiplexed libraries were pooled and sequenced on an Illumina NovaSeq X Plus sequencer with 100-cycle kits using the following read lengths: 28 bp Read 1 for cell barcode and unique molecular identifier, and 90 bp Read 2 for transcript. Raw sequencing data, in base call format (.bcl), were demultiplexed using Cell Ranger from 10x Genomics, converting the raw data into FASTQ format. Cell Ranger was also used for alignment of the FASTQ files to the human reference genome and to count the number of reads from each cell that align to each gene.
Gene expression and antibody capture data were processed using Cell Ranger (v7.2.0) with the GRCh38-2020-A reference genome and an in-house antibody reference, respectively. RNA and protein data were independently processed and clustered using the Scanpy workflow. Cell-type identities were assigned based on canonical marker gene expression and further validated by differential expression analysis of each RNA cluster. Macrophages were further subdivided based on SPP1 RNA expression—cells with detectable SPP1 expression were classified as macrophage SPP1+, whereas those lacking SPP1 expression were designated macrophage SPP1−. Comparative visualization of integrin subunit expression (β7 and α4) at the protein level was performed using the ggpubr package in R.
Peripheral blood mononuclear cell (PBMC) scRNA-seq analysis in vedolizumab-treated patients with UC
We used a publicly available cohort31 to investigate clinical features in UC PBMCs. Raw count.h5 files were downloaded from the corresponding GEO repository (GSE261334). Quality control was applied to retain high-quality cells, defined as those with ≥400 detected genes and ≤5% mitochondrial transcript content. A total of 25 PBMC samples were analyzed, including 5 from healthy controls and 5 each from post-treatment patients with UC at week 0 and week 6, stratified by responder and nonresponder status. Samples were merged, normalized and scaled; highly variable features were identified, and dimensionality reduction was performed using principal component analysis (PCA). Batch effects were corrected using Seurat’s HarmonyIntegration function with default parameters. The top principal components were used to construct a shared nearest-neighbor graph, followed by Uniform Manifold Approximation and Projection (UMAP) embedding and graph-based clustering using Seurat’s FindNeighbors and FindClusters functions (resolution = 0.2), with cluster resolution evaluated using Clustree. Cell-type identities were assigned based on canonical marker gene expression and further validated by differential expression analysis using Seurat’s FindAllMarkers function. Module scores representing clinical features were calculated using the AddModuleScore function in Seurat.
Mucosal biopsy scRNA-seq analysis in vedolizumab-treated patients with UC
We used another published cohort31 to investigate clinical features in UC rectum samples. scRNA-seq (https://figshare.com/s/3ceec45e4a640a7bceb4, https://figshare.com/s/a6142d18141697dc9650) was downloaded directly from the source provided in the original study. The scRNA-seq dataset was normalized and scaled; highly variable features were identified, and dimensionality reduction was performed using PCA. Batch effects were corrected using Seurat’s HarmonyIntegration function with default parameters. The top principal components were used to construct a shared nearest-neighbor graph, followed by UMAP embedding and graph-based clustering using Seurat’s FindNeighbors and FindClusters functions (resolution = 0.1), with resolution selection informed by Clustree analysis. The macrophage cluster was subsequently subdivided based on SPP1 expression—cells expressing SPP1 were labeled macrophage SPP1+, whereas those without detectable SPP1 expression were labeled macrophage SPP1−. Module scores representing clinical features were calculated using the AddModuleScore function in Seurat.
CosMx spatial analysis
We leveraged CosMx spatial transcriptomics data from a public UC vedolizumab dataset (CosMx cohorts 1 and 2)31 to spatially validate our findings. Seurat objects were generated from flat files downloaded from the link provided by the publication (CosMx vedolizumab cohort 1—https://doi.org/10.6084/m9.figshare.21919338; CosMx vedolizumab cohort 2—https://figshare.com/s/e8ad68ff0b4627351df0). The two CosMx datasets underwent the following workflow steps: quality control, merging, normalization and scaling; highly variable features were identified, and dimensionality reduction was performed using PCA. Batch effects were corrected using the HarmonyIntegration function with default parameters.
To achieve robust cell-type annotation in the two CosMx datasets, we used the AUCell package and a multistep approach incorporating top marker genes of general cell types by disease state from public scRNA-seq datasets using Seurat’s FindAllMarkers function, with overlapping genes from the CosMx panel and direct quantification-based gating of cell categories using the CosMx immunofluorescence protein staining.
We generated a new CosMx spatial single-cell whole-transcriptome (over 18,000 genes) dataset to further validate our spatial findings (CosMx cohort 3). Seurat objects were generated from flat files that are available for download at https://figshare.com/s/6b1fd67a7b0141dd5b32. The dataset underwent the following workflow: quality control, merging, normalization and scaling; highly variable features were identified, and dimensionality reduction was performed using PCA. Batch effects were corrected using the HarmonyIntegration function with default parameters. A shared nearest-neighbor graph was constructed using the first 30 Harmony dimensions using FindNeighbors(). Graph-based clustering was performed using FindClusters() in Seurat across a range of resolution parameters (0.1, 0.2, 0.3, and 0.5) to assess cluster granularity.
Cluster assignments at each resolution were visualized using UMAP, and a resolution of 0.2 was selected for downstream analyses based on cluster separation and biological interpretability. Cell identities were then assigned based on canonical marker gene expression. Clusters identified as low-quality or noninformative populations (clusters 11 and 18) were excluded from subsequent analyses. To further resolve macrophage heterogeneity, cells in cluster 5 were stratified according to the expression of SPP1, with positivity defined as expression >0.
Spatial cell–cell interaction analysis of the scRNA-seq, CITE-seq and CosMx vedolizumab cohorts 1, 2 and 3 was performed using CellChat (v2). For the three CosMx cohorts, individual CellChat objects were generated for each responder and nonresponder sample and subsequently merged to enable comparative analysis. Spatial coordinates of cells were scaled according to CellChat’s recommended parameters, using a conversion factor of 0.18 and a cell diameter of 1.5 µm. The standard CellChat (v2) workflow was applied to the comparison between responder and nonresponder samples.
Spatial co-occurrence probability analysis of the vedolizumab CosMx cohort 2 dataset was performed using the Squidpy package in Python (v3.13). An AnnData object was downloaded directly from the following link (https://doi.org/10.6084/m9.figshare.23896959), and cell-type annotations based on AUCell scoring with PanCK fluorescence intensity were aligned to the object. AnnData object was then subsetted to pretreatment responders (baseline) fields of view (FOVs) and the spatial co-occurrence probability of SPP1+ macrophages with other cell types was then calculated using the gr.co_occurrence and pl.co_occurrence functions from Squidpy at each FOV level.
Spatial variable gene analysis of the vedolizumab CosMx cohorts 1 and 2 was performed using the nnSVG package in R. The previously annotated Seurat objects were subsetted to include only baseline responder and nonresponder FOVs. Epithelial cells were excluded from the analysis to minimize background noise. These subsets were then converted into SpatialExperiment objects compatible with nnSVG. Spatial variability analysis was conducted using nnSVG’s default parameters without gene filtering, allowing all panel genes to be ranked for spatial variability in each FOV. To compare conditions, we calculated the mean spatial rank of each gene across FOVs. Genes with consistently higher spatial ranks in baseline vedolizumab nonresponders compared with responders (mean rank nonresponder − mean rank responder < 0) in both vedolizumab CosMx cohorts 1 and 2 were used for pathway enrichment analysis and PPI analysis using Metascape. Genes/proteins with direct PPI with SPP1 and ANGPT2 were subsetted for visualization.
Spatial colocalization between immune-cell populations and endothelial cells in CosMx cohort 3 was assessed on a per-sample basis using Squidpy. For each sample, a spatial neighborhood graph was constructed using the spatial coordinates of individual cells with 15 nearest neighbors. Cell-type neighborhood enrichment was then calculated using Squidpy’s permutation-based neighborhood enrichment test (sq.gr.nhood_enrichment) with 1,000 random permutations. This approach quantifies whether pairs of cell types occur in spatial proximity more or less frequently than expected by chance, producing a standardized enrichment z score for each cell-type pair. To summarize enrichment across samples in each clinical response group, sample-level z scores were combined using Stouffer’s weighted z method, with weights proportional to the square root of the number of cells in each sample. Positive combined z scores indicated spatial enrichment (colocalization), whereas negative z scores indicated spatial depletion.
Multi-omic CosMx data in vedolizumab-treated patients with UC
Proprietary reagents were provided by Bruker Spatial Biology, and protocols were per the Manual Slide Preparation Protocol. In summary, in situ protein expression was assessed by taking 5-µm paraffin-embedded tissue sections mounted on Apex BOND (Leica Biosystems) and quantifying expression on the CosMx spatial molecular imager platform (NanoString). Initially, formalin-fixed, paraffin-embedded tissue sections were mounted on glass slides and dried overnight at 60 °C then deparaffined with CitriSolv (Decon Labs) washed in 100% then 70% ethanol and finally 1× PBS. Sections were subject to antigen retrieval in target retrieval buffer (citate based) at 100 °C for 15 min and allowed to cool in buffer for 25 min. After PBS washes, tissue was blocked with buffer W for 1 h. Primary antibodies (64 total) and antibody markers CD298/B2M, PanCK/CD45 and CD3 were incubated on tissue for 16–18 h in buffer W. Following washes in 1× Tris-buffered saline with Tween-20 (0.00005%), fiducials were applied. After 1× PBS washes, tissue was incubated for 15 min with DSP (Pierce) solution, washed and incubated with DAPI stain, washed and treated with NHS acetate (Sulfo-NHS-Acetate, ProteoChem), before washing in 1× PBS and flow cell assembly. Flow cells were loaded onto a CosMx machine and protein run was initiated. FOV regions were selected and expression data were generated and stored in AtoMx (v2.1) for analysis.
The protein expression study was followed by RNA analysis on the same tissue slides; the flow cell cover was carefully removed with a disposable scalpel blade under 1× PBS. Tissue was then treated for 15 min with 50 mM TCEP (MilliporeSigma) and washed in 1× PBS. Tissues were incubated with Subtilisin at 40 °C for 30 min. After washing, fiducials were applied at 0.01% for 5 min, washed and incubated for 16–18 h with 1,000 plex RNA in situ hybridization probes in 0.5 nM rRNA. Tissues were washed in 50% formamide 2× saline-sodium citrate buffers, and DAPI stain was applied. After 1× PBS washes, the tissues were incubated for 1 h with antibody markers CD298/B2M, PanCK/CD45 and CD68. Tissues were washed in 1× PBS and treated with NHS acetate, washed again in 1× PBS and flow cells were assembled. Flow cells were loaded onto the same CosMx machine and placed in the same position within the machine as used for the protein run. FOV regions very closely matching the protein run were selected, and RNA expression data were generated and stored in AtoMx (v2.1) for analysis. Cells were annotated using protein expression and a cell-type marker approach, with subsequent visual confirmation of protein staining quality and examination of staining distribution patterns in AtoMx. For the RNA data paired at the single-cell level to each sample, expression of target genes was visualized using Seurat’s ImageFeaturePlot function, with the default boundary set to ‘segmentation’.
Proteomic confirmation of biomarkers and pathways in cohort 1
To assess serum protein levels of the identified tissue biomarkers, we analyzed baseline and week 14 proteomics data generated from cohort 1 patients using the Olink Explore HT NGS platform (~5,400 proteins). Tissue-derived gene signatures available in the proteomics panel were evaluated both at an individual protein level and signature-based enrichment through single-sample GSEA in nonresponder versus responder groups based on the primary clinical response. Single-sample GSEA was performed on proteins that were normalized using inverse rank transformation. Between-group differences (responders versus nonresponders) were evaluated using t tests with FDR correction at both baseline and week 14 time points. Protein/signature serum activity differences with both vedolizumab and adalimumab treatment groups are visualized in line plot format, showing mean and standard error, highlighting both baseline and week 14 values and their longitudinal changes.
Serum osteopontin (OPN; SPP1) was quantified using a microfluidic immunoassay (Ella platform, ProteinSimple) with the Human OPN 2nd Gen Simple Plex cartridge (SPCKB-PS004314; products 896055, 896098, 896959). Serum samples were stored at −80 °C, thawed at room temperature and diluted 1:100 in sample diluent before analysis. The assay uses antibody-coated glass nanoreactors to capture OPN, followed by detection with a fluorophore-labeled antibody and measurement of relative fluorescence units. Quantification was performed using lot-specific calibration curves (21–12,800 pg ml−1) fitted with a four-parameter logistic model (1/y2 weighting). Sample concentrations were calculated by interpolation from the standard curve, using the mean relative fluorescence units across triplicate nanoreactors. Quality controls included buffer and endogenous serum controls, with predefined acceptance criteria of ±20–25% accuracy and ≤20% coefficient of variation. Data were analyzed using Assist LIMS.
Multicenter cohort validation of impact for angiotensin II inhibitors
A retrospective cohort study was conducted using TriNetX, a multi-institutional de-identified federated research network and database93. The database does not include claims data or data collected from randomized clinical trials. The database contains inpatient and outpatient claims, along with prescription drug claims.
We conducted a real-time search and analysis of the US Collaborative Network in the TriNetX platform in adult patients with UC and hypertension who were initiated on vedolizumab, anti-TNF or ustekinumab therapy before 31 December 2022. Patients with UC and hypertension were identified using ‘International Classification of Diseases, Tenth Revision, Clinical Modification’ (ICD-10-CM) codes in their electronic health records for UC (K51*) and essential hypertension (I10*), respectively. Patients receiving biologics were identified using a RxNorm code (1538097). Patients were divided into two cohorts—one comprising those who started either an ACEi or ARB, and one comprising those who did not (control cohort). As a negative control, we created a second cohort of patients who started either a BB or a CCB and those who did not. Patients in the concomitant medication arms were required to have a prescription before and after initiation of the biologic to ensure ongoing use. Patients who had an ICD-10-CM code for CD were excluded to avoid misclassification bias. Patients with a history of partial or total colectomy or change in therapy within 4 weeks of initiation were excluded to account for potential insurance denials and/or inadequate trial of the therapy of interest.
The index event was the prescription of biologic, and outcome ascertainment was after 4 weeks from the index event. The outcomes included the risk of hospitalization for a UC flare, colectomy or change in therapy (treatment persistence). Patients with hospitalization for a flare were identified using RxNorm codes for methylprednisolone (6902) and hydrocortisone (5492). Patients who required total colectomy were identified using Current Procedural Terminology or ICD-10 Procedure Coding System codes. Change in therapy included a switch to an alternative advanced biologic or small-molecule therapy.
All statistical analyses were conducted using the TriNetX software using the browser-based real-time analytics feature TriNetX Live (TriNetX). One-to-one (1:1) propensity score matching was performed to balance the following covariates between groups: age, sex, race, diabetes mellitus, nicotine dependence, obesity, ischemic heart disease, heart failure, stroke, chronic kidney disease, chronic lower respiratory diseases, primary sclerosing cholangitis, other autoimmune diseases, UC medication (oral and intravenous) and steroid use before initiation of the biologic. Propensity score matching was also performed for mean hemoglobin, albumin, CRP and calprotectin. Patients with missing laboratory data were still included; however, propensity score matching adjusted laboratory covariates from patients with available data. The TriNetX platform uses input matrices of the user-identified covariates to conduct a logistic regression analysis to obtain propensity scores for all individual participants. The propensity scores generated are used to match patients using greedy nearest-neighbor algorithms with a caliper width of 0.1 pooled s.d. TriNetX randomizes the order of rows to eliminate bias resulting from nearest-neighbor algorithms. Standardized mean difference after propensity score matching indicates the success of matching a covariate between the two cohorts. A standardized mean difference of <0.1 indicates that the difference between the cohorts for the covariate is small (after matching, age, sex, race/ethnicity, comorbidities, hemoglobin, C-reactive protein, albumin and calprotectin met this threshold). After propensity score matching, the risk of each outcome was calculated and expressed as adjusted HRs with 95% CIs. Kaplan–Meier curves were generated for the ACEi/ARB cohort and control cohort after 1:1 propensity score matching.
Claims-based validation of impact for angiotensin II inhibitors
Optum’s de-identified Clinformatics Data Mart (CDM) database is built from administrative health claims collected from members enrolled in large commercial insurance plans and Medicare Advantage plans. The dataset is de-identified using the Expert Determination approach in compliance with Health Insurance Portability and Accountability Act standards and is maintained under Optum’s data use agreements. Before inclusion in Optum CDM, administrative claims submitted by physicians, hospitals and pharmacies are validated, adjudicated and stripped of identifying information. The dataset contains patient-level enrollment details and captures all medical and outpatient pharmacy claims, providing information on health care usage and costs. The covered population is geographically broad, representing individuals from all 50 states.
Patients with UC were identified using either ICD-10-CM (K51*) or ICD-9-CM (556*). Hypertension was identified using either ICD-10-CM (I10*) or ICD-9-CM (401*). Patients needed to be at least 18 years of age at the time of their first vedolizumab, anti-TNF or ustekinumab exposure. The cohort was divided into two groups—one with patients who used ACEi/ARB and one without. Patients in the ACEi/ARB group were required to have at least one refill within 3 months of biologic initiation to ensure consistent usage. Patients who had partial or total colectomy or who changed therapy within 4 weeks before initiation of biologics were excluded. Patients diagnosed with CD were also excluded. We extracted data from Optum CDM from 2000 to 2025.
The index date was defined as the prescription date of the biologic. The outcome was total colectomy identified using Current Procedural Terminology codes. The follow-up period was 1 year. Patients were censored if they died or exited insurance enrollment. The outcome was evaluated starting 4 weeks after the index date. Data were queried using SAS Studio 3.81 (Enterprise Edition). All statistical analyses were conducted using R (v4.3.1). Baseline covariates included demographics, comorbidities and IBD medication usage in the past 3 months. One-to-one propensity score matching was conducted to balance covariates between the two groups using the R package ‘MatchIt’. Covariates included age at biologic initiation, sex, race, diabetes mellitus, nicotine dependence, obesity, ischemic heart disease, heart failure, stroke, chronic kidney disease, chronic lower respiratory diseases, primary sclerosing cholangitis, other autoimmune diseases, prednisone usage, intravenous steroid usage and other biologics. We used greedy nearest-neighbor matching with a caliper of 0.1 s.d. The order of rows was randomized to avoid bias from the greedy nearest-neighbor algorithm. The risk of the outcome was represented by adjusted HRs with 95% CIs. Kaplan–Meier curves were used to visualize the two groups.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Participant-level data from the phase 3 clinical trial programs are available through established data sharing agreements via application at https://search.vivli.org. Mucosal biopsy bulk RNA-seq and peripheral blood Olink proteomic data for the VARSITY clinical trial will be made available upon request after de-identification, in compliance with applicable privacy laws, data protection and requirements for consent and anonymization. If necessary, approval by the relevant ethics committee(s) may have to be provided. Newly generated CITE-seq, scRNA-seq and spatial single-cell whole-transcriptome datasets are available through GEO. All other public datasets used are available through accession numbers outlined in the methods.
Publicly available single-cell datasets analyzed in this study can be accessed from their respective repositories—PBMC scRNA-seq data (GSE261334); rectal biopsy scRNA-seq data (https://figshare.com/s/3ceec45e4a640a7bceb4); vedolizumab CosMx cohort 2 data (https://doi.org/10.6084/m9.figshare.21919338) and its regenerated expression matrix and metadata for Seurat objects (https://doi.org/10.6084/m9.figshare.30767693); and vedolizumab CosMx cohort 2 data (https://doi.org/10.6084/m9.figshare.23896959) and its regenerated expression matrix and metadata for Seurat object (https://doi.org/10.6084/m9.figshare.32810471).
The raw data of CITE-seq and scRNA-seq used in this study are available from GEO under accession GSE331585. The processed CITE-seq expression matrices and metadata can be accessed at https://doi.org/10.6084/m9.figshare.30729362 (ref. 94). The raw data of CosMx whole-transcriptomic data (vedolizumab CosMx cohort 3) used in this study are available from GEO under accession GSE336853. The processed Seurat object can be accessed at https://figshare.com/s/6b1fd67a7b0141dd5b32 (ref. 95). CosMx multi-omics data generated in this study are available from GEO under accession GSE311636.
Survival-curve data used for comparative analysis of persistence on biologic therapy are available on figshare (https://doi.org/10.6084/m9.figshare.30766853 (ref. 96)).
Patient data and bulk RNA-seq data for the densely sampled longitudinal IBD cohort are available through the European Genome-phenome Archive (EGA) under accession number EGAS00001008553. Intermediate analysis files for this cohort can be accessed through figshare at https://figshare.com/s/23a299e0e812653b1917 (ref. 97).
Data for the VARSITY bulk RNA-seq comparative and trajectory analysis can be accessed through figshare at https://figshare.com/s/36a0c2dbd7ea269a9b3c (ref. 98).
Data for the propensity score matching survival analysis for vedolizumab/anti-TNF/ustekinumab can be accessed through figshare at https://figshare.com/s/2844f9dcc8d85b641d26 (ref. 99).
Data for spatial variable gene lists from nnSVG analysis for Vedo CosMx cohorts 1 and 2 and the corresponding Metascape cytoscape network of PPI can be accessed through figshare at https://doi.org/10.6084/m9.figshare.32810492 (ref. 100).
References
Pandey, A., Achrafie, L., Kodjamanova, P., Tencer, T. & Kumar, J. Endoscopic mucosal healing and histologic remission in ulcerative colitis: a systematic literature review of clinical, quality-of-life and economic outcomes. Curr. Med. Res. Opin. 38, 1531–1541 (2022).
Yoon, H. et al. Incremental benefit of achieving endoscopic and histologic remission in patients with ulcerative colitis: a systematic review and meta-analysis. Gastroenterology 159, 1262–1275 (2020).
Dulai, P. S. et al. Prevalence of endoscopic improvement and remission according to patient-reported outcomes in ulcerative colitis. Aliment. Pharmacol. Ther. 51, 435–445 (2020).
Shehab, M. et al. Comparative efficacy of biologics and small molecule in ulcerative colitis: a systematic review and network meta-analysis. Clin. Gastroenterol. Hepatol. 23, 250–262 (2025).
Schreiber, S. et al. Defining comprehensive disease control for use as a treatment target for ulcerative colitis in clinical practice: international Delphi consensus recommendations. J. Crohns Colitis 18, 91–105 (2024).
D’Amico, F. et al. Disease clearance as a new outcome in ulcerative colitis: a systematic review and expert consensus. Inflamm. Bowel Dis. 30, 1009–1017 (2024).
Parkes, G. et al. Correlation of mucosal healing endpoints with long-term clinical and patient-reported outcomes in ulcerative colitis. J. Gastroenterol. 58, 990–1002 (2023).
Neurath, M. F. Current and emerging therapeutic targets for IBD. Nat. Rev. Gastroenterol. Hepatol. 14, 269–278 (2017).
Schreiber, S., Aden, K., Tran, F. & Rosenstiel, P. Rise of precision medicine: can it deliver on its promise in IBD?. Gut 75, 176–188 (2025).
Mao, R. & Chen, M. Precision medicine in IBD: genes, drugs, bugs and omics. Nat. Rev. Gastroenterol. Hepatol. 19, 81–82 (2022).
Syed, S. et al. Challenges in IBD Research 2024: precision medicine. Inflamm. Bowel Dis. 30, S39–S54 (2024).
Davis, M. M., Tato, C. M. & Furman, D. Systems immunology: just getting started. Nat. Immunol. 18, 725–732 (2017).
Friedman, J. Greedy function approximation: a gradient boosting machine. Ann. Statist. 29, 1189–1232 (2000).
Nagin, D. S., Jones, B. L. & Elmer, J. Recent advances in group-based trajectory modeling for clinical research. Annu. Rev. Clin. Psychol. 20, 285–305 (2024).
Sands, B. E. et al. Vedolizumab versus adalimumab for moderate-to-severe ulcerative colitis. N. Engl. J. Med. 381, 1215–1226 (2019).
Feagan, B. G. et al. Vedolizumab as induction and maintenance therapy for ulcerative colitis. N. Engl. J. Med. 369, 699–710 (2013).
US Food and Drug Administration. Guidance for industry: patient-reported outcome measures: use in medical product development to support labeling claims https://www.fda.gov/regulatory-information/search-fda-guidance-documents/patient-reported-outcome-measures-use-medical-product-development-support-labeling-claims (2009).
Arijs, I. et al. Mucosal gene expression of antimicrobial peptides in inflammatory bowel disease before and after first infliximab treatment. PLoS ONE 4, e7984 (2009).
Arijs, I. et al. Effect of vedolizumab (anti-α4β7-integrin) therapy on histological healing and mucosal gene expression in patients with UC. Gut 67, 43–52 (2018).
Arijs, I. et al. Mucosal gene signatures to predict response to infliximab in patients with ulcerative colitis. Gut 58, 1612–1619 (2009).
Pavlidis, S. et al. I_MDS: an inflammatory bowel disease molecular activity score to classify patients with differing disease-driving pathways and therapeutic response to anti-TNF treatment. PLoS Comput. Biol. 15, e1006951 (2019).
Toedter, G. et al. Gene expression profiling and response signatures associated with differential responses to infliximab treatment in ulcerative colitis. Am. J. Gastroenterol. 106, 1272–1280 (2011).
Pavlidis, P. et al. Interleukin-22 regulates neutrophil recruitment in ulcerative colitis and is associated with resistance to ustekinumab therapy. Nat. Commun. 13, 5820 (2022).
Zeissig, S. et al. Vedolizumab is associated with changes in innate rather than adaptive immunity in patients with inflammatory bowel disease. Gut 68, 25–39 (2019).
Narula, N. et al. Adaptive steroid tapering impedes corticosteroid-free remissions compared with forced tapering in clinical trials of ulcerative colitis. J. Crohns Colitis 18, 1863–1869 (2024).
Shan, M. et al. Cigarette smoke induction of osteopontin (SPP1) mediates TH17 inflammation in human and experimental emphysema. Sci. Transl. Med. 4, 117ra119 (2012).
Smillie, C. S. et al. Intra- and inter-cellular rewiring of the human colon during ulcerative colitis. Cell 178, 714–730 (2019).
Thomas, T. et al. A longitudinal single-cell atlas of anti-tumour necrosis factor treatment in inflammatory bowel disease. Nat. Immunol. 25, 2152–2165 (2024).
Hsu, P. et al. Responsiveness to vedolizumab therapy in ulcerative colitis is associated with alterations in immune cell–cell communications. Inflamm. Bowel Dis. 29, 1602–1612 (2023).
Mennillo, E. et al. Single-cell and spatial multi-omics highlight effects of anti-integrin therapy across cellular compartments in ulcerative colitis. Nat. Commun. 15, 1493 (2024).
Horn, V. et al. Multimodal profiling of peripheral blood identifies proliferating circulating effector CD4+ T cells as predictors for response to integrin α4β7-blocking therapy in inflammatory bowel disease. Gastroenterology 168, 327–343 (2025).
Green, P. M., Ludbrook, S. B., Miller, D. D., Horgan, C. M. & Barry, S. T. Structural elements of the osteopontin SVVYGLR motif important for the interaction with α4 integrins. FEBS Lett. 503, 75–79 (2001).
Masuda, H., Takahashi, Y., Asai, S. & Takayama, T. Distinct gene expression of osteopontin in patients with ulcerative colitis. J. Surg. Res. 111, 85–90 (2003).
White, E. S., Livant, D. L., Markwart, S. & Arenberg, D. A. Monocyte-fibronectin interactions, via α5β1 integrin, induce expression of CXC chemokine-dependent angiogenic activity. J. Immunol. 167, 5362–5366 (2001).
Kim, J. et al. Secreted phosphoprotein 1 binds integrins to initiate multiple cell signaling pathways, including FRAP1/mTOR, to support attachment and force-generated migration of trophectoderm cells. Matrix Biol. 29, 369–382 (2010).
Liang, X. et al. Colorectal cancer-derived osteopontin rewires macrophages into a pro-metastatic M2 state via the PI3K/AKT/CSF1-CSF1R axis. Cell Death Discov. 12, 92 (2026).
Zhou, Y. et al. CSF1/CSF1R-mediated crosstalk between choroidal vascular endothelial cells and macrophages promotes choroidal neovascularization. Invest. Ophthalmol. Vis. Sci. 62, 37 (2021).
Lewis, C. E., Harney, A. S. & Pollard, J. W. The multifaceted role of perivascular macrophages in tumors. Cancer Cell 30, 18–25 (2016).
Kosti, C. N. et al. CSF1/CSF1R signaling mediates malignant pleural effusion formation. JCI Insight 7, e155300 (2022).
Lee, S. J. et al. Angiopoietin-2 exacerbates cardiac hypoxia and inflammation after myocardial infarction. J. Clin. Invest. 128, 5018–5033 (2018).
Kuang, Y. et al. Vascular endothelial S1pr1 ameliorates adverse cardiac remodelling via stimulating reparative macrophage proliferation after myocardial infarction. Cardiovasc. Res. 117, 585–599 (2021).
Jian, Y. et al. Crosstalk between macrophages and cardiac cells after myocardial infarction. Cell Commun. Signal. 21, 109 (2023).
Alvarez, A. et al. Direct evidence of leukocyte adhesion in arterioles by angiotensin II. Blood 104, 402–408 (2004).
Pueyo, M. E. et al. Angiotensin II stimulates endothelial vascular cell adhesion molecule-1 via nuclear factor-κB activation induced by intracellular oxidative stress. Arterioscler. Thromb. Vasc. Biol. 20, 645–651 (2000).
Pastore, L. et al. Angiotensin II stimulates intercellular adhesion molecule-1 (ICAM-1) expression by human vascular endothelial cells and increases soluble ICAM-1 release in vivo. Circulation 100, 1646–1652 (1999).
Okamura, A. et al. Upregulation of renin-angiotensin system during differentiation of monocytes to macrophages. J. Hypertens. 17, 537–545 (1999).
Kojima, C., Kawakami, A., Takei, T., Nitta, K. & Yoshida, M. Angiotensin-converting enzyme inhibitor attenuates monocyte adhesion to vascular endothelium through modulation of intracellular zinc. J. Pharmacol. Exp. Ther. 323, 855–860 (2007).
Leuschner, F. et al. Angiotensin-converting enzyme inhibition prevents the release of monocytes from their splenic reservoir in mice with myocardial infarction. Circ. Res. 107, 1364–1373 (2010).
Napoleone, E., Di Santo, A., Camera, M., Tremoli, E. & Lorenzet, R. Angiotensin-converting enzyme inhibitors downregulate tissue factor synthesis in monocytes. Circ. Res. 86, 139–143 (2000).
Spencer, A. U. et al. Reduced severity of a mouse colitis model with angiotensin converting enzyme inhibition. Dig. Dis. Sci. 52, 1060–1070 (2007).
Sueyoshi, R., Ignatoski, K. M., Daignault, S., Okawada, M. & Teitelbaum, D. H. Angiotensin converting enzyme-inhibitor reduces colitis severity in an IL-10 knockout model. Dig. Dis. Sci. 58, 3165–3177 (2013).
Arumugam, S. et al. Telmisartan treatment targets inflammatory cytokines to suppress the pathogenesis of acute colitis induced by dextran sulphate sodium. Cytokine 74, 305–312 (2015).
Katada, K. et al. Dextran sulfate sodium-induced acute colonic inflammation in angiotensin II type 1a receptor deficient mice. Inflamm. Res. 57, 84–91 (2008).
Shi, Y. et al. Activation of the renin-angiotensin system promotes colitis development. Sci. Rep. 6, 27552 (2016).
Luo, A. J., Chang, F. C. & Lin, S. L. Exploring angiopoietin-2: clinical insights and experimental perspectives in kidney diseases. Kidney Int. Rep. 9, 3375–3385 (2024).
He, L. et al. Renin-angiotensin system promotes colonic inflammation by inducing TH17 activation via JAK2/STAT pathway. Am. J. Physiol. Gastrointest. Liver Physiol. 316, G774–G784 (2019).
Collins, A. R. et al. Osteopontin modulates angiotensin II-induced fibrosis in the intact murine heart. J. Am. Coll. Cardiol. 43, 1698–1705 (2004).
Wolak, T. et al. Osteopontin modulates angiotensin II-induced inflammation, oxidative stress, and fibrosis of the kidney. Kidney Int. 76, 32–43 (2009).
Menzel, K. et al. Cathepsins B, L and D in inflammatory bowel disease macrophages and potential therapeutic effects of cathepsin inhibition in vivo. Clin. Exp. Immunol. 146, 169–180 (2006).
Kanugula, A. K., Kaur, J., Batra, J., Ankireddypalli, A. R. & Velagapudi, R. Renin-angiotensin system: updated understanding and role in physiological and pathophysiological states. Cureus 15, e40725 (2023).
Hausmann, M. et al. Cathepsin D is up-regulated in inflammatory bowel disease macrophages. Clin. Exp. Immunol. 136, 157–167 (2004).
Mantaka, A. et al. Is there any role of renin-angiotensin system inhibitors in modulating inflammatory bowel disease outcome?. Eur. J. Gastroenterol. Hepatol. 33, 364–371 (2021).
Jacobs, J. D. et al. Impact of angiotensin II signaling blockade on clinical outcomes in patients with inflammatory bowel disease. Dig. Dis. Sci. 64, 1938–1944 (2019).
Garg, M. et al. Imbalance of the renin-angiotensin system may contribute to inflammation and fibrosis in IBD: a novel therapeutic target?. Gut 69, 841–851 (2020).
Fairbrass, K. M., Hoshen, D., Gracie, D. J. & Ford, A. C. Effect of ACE inhibitors and angiotensin II receptor blockers on disease outcomes in inflammatory bowel disease. Gut 70, 218–219 (2021).
Ferreira-Duarte, M., Tonin, F. S., Duarte-Araujo, M., Fernandez-Llimos, F. & Morato, M. Heterogeneity in protocols and outcomes to study the effect of renin-angiotensin system blockers in inflammatory bowel disease: a systematic review. Fundam. Clin. Pharmacol. 37, 1139–1152 (2023).
Roosenboom, B. et al. Mucosal α4β7+ lymphocytes and MAdCAM+ venules predict response to vedolizumab in ulcerative colitis. Inflamm. Bowel Dis. 30, 930–938 (2024).
Rioux, J. D. et al. Serum proteomic and metabolomic analyses from patients with IBD identify biological pathways associated with treatment success with anti-integrin therapy. Immunol. Cell Biol. 103, 648–663 (2025).
Lee, J. W. J. et al. Multi-omics reveal microbial determinants impacting responses to biologic therapies in inflammatory bowel disease. Cell Host Microbe 29, 1294–1304 (2021).
Joustra, V. W. et al. Development and validation of peripheral blood DNA methylation signatures to predict response to biological therapy in adults with Crohn’s disease (EPIC-CD): an epigenome-wide association study. Lancet Gastroenterol. Hepatol. 10, 818–830 (2025).
Gonzalez, C. G. et al. Crohn’s patient serum proteomics reveals response signature for infliximab but not vedolizumab. Inflamm. Bowel Dis. 30, 1536–1545 (2024).
Canales-Herrerias, P. et al. Gut-associated lymphoid tissue attrition associates with response to anti-α4β7 therapy in ulcerative colitis. Sci. Immunol. 9, eadg7549 (2024).
Luzentales-Simpson, M., Pang, Y. C. F., Zhang, A., Sousa, J. A. & Sly, L. M. Vedolizumab: potential mechanisms of action for reducing pathological inflammation in inflammatory bowel diseases. Front. Cell Dev. Biol. 9, 612830 (2021).
Murray, S. A. et al. Using illness trajectories to inform person centred, advance care planning. BMJ 384, e067896 (2024).
Arnaud, A. et al. The impact of early remission on disease trajectory and patient outcomes in major depression disorder (MDD): a targeted literature review and microsimulation modeling approach based on the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study. J. Affect. Disord. 325, 264–272 (2023).
Kerr, D. J. & Yang, L. Personalising cancer medicine with prognostic markers. eBioMedicine 72, 103577 (2021).
Fang, J. & Liao, L. Association analysis of white blood cell trajectories and 30-day mortality in ventilator-associated pneumonia patients: a cohort study based on the MIMIC-IV database. APMIS 133, e70042 (2025).
Shi, Q. et al. Using group-based trajectory modeling to examine heterogeneity of symptom burden in patients with head and neck cancer undergoing aggressive non-surgical therapy. Qual. Life Res. 22, 2331–2339 (2013).
Cleeland, C. S. et al. Levels of symptom burden during chemotherapy for advanced lung cancer: differences between public hospitals and a tertiary cancer center. J. Clin. Oncol. 29, 2859–2865 (2011).
Côté, S. M. et al. Depression and anxiety symptoms: onset, developmental course and risk factors during early childhood. J. Child Psychol. Psychiatry 50, 1201–1208 (2009).
Tran, F. et al. P722 symptom-based disease trajectories identify a population of early super-responders in ulcerative colitis achieving complete disease control under targeted therapy. J. Crohns Colitis 17, i852–i854 (2023).
Ha, C., Ullman, T. A., Siegel, C. A. & Kornbluth, A. Patients enrolled in randomized controlled trials do not represent the inflammatory bowel disease patient population. Clin. Gastroenterol. Hepatol. 10, 1002–1007 (2012).
Gonzalez, C. G. et al. Location-specific signatures of Crohn’s disease at a multi-omics scale. Microbiome 10, 133 (2022).
Dulai, P. S. et al. Should we divide Crohn’s disease into ileum-dominant and isolated colonic diseases?. Clin. Gastroenterol. Hepatol. 17, 2634–2643 (2019).
Nagin, D. S., Jones, B. L., Passos, V. L. & Tremblay, R. E. Group-based multi-trajectory modeling. Stat. Methods Med. Res. 27, 2015–2023 (2018).
Loftus, E. V. Jr. et al. Long-term safety of vedolizumab for inflammatory bowel disease. Aliment. Pharmacol. Ther. 52, 1353–1365 (2020).
Hu, J., Ge, H., Newman, M. & Liu, K. OSA: a fast and accurate alignment tool for RNA-seq. Bioinformatics 28, 1933–1934 (2012).
Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 43, e47 (2015).
Mootha, V. K. et al. PGC-1α-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat. Genet. 34, 267–273 (2003).
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B Stat. Methodol. 57, 289–300 (1995).
Pavlidis, P. et al. Cytokine responsive networks in human colonic epithelial organoids unveil a molecular classification of inflammatory bowel disease. Cell Rep. 40, 111439 (2022).
Kochhar, G. S. et al. Efficacy of biologic and small molecule agents as second-line therapy after exposure to TNF inhibitors in patients with ulcerative colitis: a propensity-matched cohort study. Aliment. Pharmacol. Ther. 58, 297–308 (2023).
Wei, C. CITE_seq. Figshare https://doi.org/10.6084/m9.figshare.30729362.v1 (2026).
Simpson, K. Processed Seurat object for whole transcriptome CosMx dataset, Figshare, https://doi.org/10.6084/m9.figshare.32663904 (2026).
Wei, C. KM curve. Figshare https://doi.org/10.6084/m9.figshare.30766853.v1 (2026).
Kimmig, F. Densely sampled longitudinal cohort bulk RNA-Seq analysis results, Figshare, https://doi.org/10.6084/m9.figshare.30823292 (2026).
Farahmand, S. VARSITY response trajectory -omics analysis, Figshare, https://doi.org/10.6084/m9.figshare.30811562 (2026).
Lin, Z. Data for the propensity score matching survival analysis for vedolizumab/anti-TNF/ustekinumab, Figshare, https://doi.org/10.6084/m9.figshare.30826262 (2026).
Wei, C. SVG consistently highly ranked in both Vedo CosMx cohort 1 and 2. Figshare https://doi.org/10.6084/m9.figshare.32810492.v1 (2026).
Acknowledgements
The authors would like to thank H. Wennbo (Takeda, Cambridge, MA, USA) for supporting this work at Takeda. The authors would like to thank S. Vermeire (KU Leuven, Leuven, Belgium) for assistance with analysis of the GEMINI 1 dataset. This study includes clinical and molecular data supported by Takeda. Medical writing support was provided by P. Hassan of Envision Medical Communications, a part of Envision Pharma Group, and supported by Takeda. The study was conducted in accordance with the Good Publication Practice (GPP 2022) guidelines.
Funding
P.S.D. received funding in part from the NIDDK for work outlined in the study (R01DK135620 and U01DK134321). Takeda supported the study by providing access to data from the VARSITY and GEMINI 1. S.S. and P.R. received funding from the Innovative Medicines Initiative 2 Joint Undertaking (JU) under grant agreement 831434 (3TR) and ImmUniverse under grant agreement 853995. The JU receives support from the European Union’s Horizon 2020 research and innovation program and the European Federation of Pharmaceutical Industries and Associations. The content of this publication reflects only the authors’ views and the JU is not responsible for any use that may be made of the information it contains. Further funding was received from the DFG Excellence Cluster ‘Precision Medicine in Inflammation’ (EXC 2167) and BMBF iTREAT (01ZX1902A). Funding information for this article has been deposited with the Crossref Funder Registry. Application of our concepts to other chronic diseases and drugs will offer opportunities to develop personalized medicine approaches that reshape therapeutic efficacy in gastrointestinal and other diseases being evaluated, for which we have founded a cross-industry collaboration (3TR) to evaluate patient-level data across different mechanisms of action. Open access funding provided by Christian-Albrechts-Universität zu Kiel.
Author information
Author notes
These authors contributed equally: Vijay Yajnik, Paresh Thakker, Parambir S. Dulai.
Authors and Affiliations
Department Medicine I and Institute Clinical Molecular Biology, Kiel University, University Hospital Schleswig-Holstein, Kiel, Germany
Stefan Schreiber
Global Medical Affairs, Takeda, Cambridge, MA, USA
Vijay Yajnik, Christian Agboton & Joshua Brown
Division of Gastroenterology and Hepatology, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
Cenfu Wei, Katherine Simpson, Zitong Lin, Isabella Peshek, Frida Chavez, Libeth Rosas, Alan Robinson & Parambir S. Dulai
Computational Biology & Human Genetics, Computational Science & Data Strategy, Takeda Development Center Americas, Cambridge, MA, USA
Maria D. Paraskevopoulou & Saman Farahmand
Statistical and Quantitative Sciences, Takeda, Cambridge, MA, USA
Jie Cheng
Translational Medicine & Early Clinical Development, Gastrointestinal and Inflammation Therapeutic Area Unit, Takeda Development Center Americas, Cambridge, MA, USA
Pratik Panchal
Translational Medicine and Biomarker Research (Early Clinical Development), Gastrointestinal and Inflammation Therapeutic Area Unit, Takeda Development Center Americas, Cambridge, MA, USA
Banishree Saha, Ninfa Candela & Paresh Thakker
Northwell Health Center for Advanced IBD Care, Manhasset, NY, USA
Gursimran Singh Kochhar
Division of Gastroenterology, Hepatology & Nutrition, Allegheny Health Network, Pittsburgh, PA, USA
Aakash Desai
Institute of Clinical Molecular Biology, University of Kiel, Kiel, Germany
Franziska Kimmig & Philip Rosenstiel
Center for Human Immunobiology, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
Parambir S. Dulai
Authors
- Stefan Schreiber
- Vijay Yajnik
- Cenfu Wei
- Maria D. Paraskevopoulou
- Saman Farahmand
- Jie Cheng
- Katherine Simpson
- Zitong Lin
- Christian Agboton
- Pratik Panchal
- Banishree Saha
- Ninfa Candela
- Joshua Brown
- Isabella Peshek
- Frida Chavez
- Libeth Rosas
- Alan Robinson
- Gursimran Singh Kochhar
- Aakash Desai
- Franziska Kimmig
- Philip Rosenstiel
- Paresh Thakker
- Parambir S. Dulai
Contributions
C.W. and M.D.P. contributed equally as co-third authors in this manuscript. The concept was developed by S.S. and primary analyses were developed and carried out by the authors. P.S.D., P.T., V.Y., M.D.P., J.C., C.W., K.S. and S.S. conceived the study. P.S.D., M.D.P., C.W., J.C., S.F., Z.L., K.S., B.S., A.D., F.K., P.R. and S.S. performed the methodology. C.W., K.S., Z.L., M.D.P., J.B., J.C., S.F., A.D. and F.K. performed the software development. P.S.D., P.T., V.Y., C.W., J.C., S.F., Z.L., K.S., B.S., J.B., I.P., F.C., L.R., F.K., A.R., P.R., G.S.K., A.D. and M.D.P. were responsible for validation. P.S.D., P.T., C.W., M.D.P., J.C., S.F., Z.L., K.S., A.D., F.K. and P.R. conducted formal analysis. P.S.D., P.T., V.Y., J.C., S.F., C.A., P.P., B.S., N.C., G.K., A.D., F.K., P.R., C.W., Z.L., K.S., M.D.P. and S.S. conducted investigation. P.S.D., V.Y., P.T., M.D.P., C.W. and S.S. performed resource management. P.S.D., M.D.P., C.W., Z.L., K.S., J.C., S.F., A.D. and F.K. conducted data curation. P.S.D., C.W., Z.L., K.S., M.D.P., J.C., S.F., A.D., F.K. and P.R. visualized the study. P.S.D., P.T., V.Y. and S.S. supervised the study. P.S.D., P.T., M.D.P., J.C., V.Y. and S.S. were responsible for project administration. P.T., P.S.D. and S.S. acquired funding. P.S.D., P.T., V.Y. and S.S. contributed to writing the original draft of the manuscript. All authors contributed to the preparation, writing, review and editing of the final draft of the manuscript and approved it for publication.
Corresponding author
Correspondence to Stefan Schreiber.
Ethics declarations
Competing interests
S.S. has received consulting fees from AbbVie, Alfasigma, Amgen, Arena, Boehringer Ingelheim, Bristol Myers Squibb, Celltrion, Dr. Falk Pharma, Eli Lilly, Ferring Pharmaceuticals, Fresenius, Genentech, Gilead Sciences, GSK, I-MAB Biopharma, Janssen, Merck, Novartis/Sandoz, Pfizer, Protagonist, Takeda and Ventyx. V.Y., M.D.P., S.F., J.C., C.A., P.P., B.S., N.C., J.B. and P.T. are employees of the study sponsor, Takeda, and hold Takeda stock or stock options. P.S.D. has received consulting fees and/or research support from AbbVie, Abivax, Adiso Therapeutics, Bristol Myers Squibb, Celltrion, Cristcot, Eli Lilly, Genentech, Geneoscopy, Janssen, Merck, Sanofi and Takeda. G.S.K. has been a speaker for Eli Lilly; an advisor for Eli Lilly, Exact Sciences and GIE Medical; a consultant for Takeda, Olympus endoscopy, Boston Scientific endoscopy and Pentax endoscopy; and holds stock options in Digbi Health. The remaining authors declare no competing interests.
Peer review
Peer review information
Nature Medicine thanks Zlatko Trajanoski, who co-reviewed with Eduardo Villablanca; Jose Ordovas-Montanes; and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Anna Ranzoni, in collaboration with the Nature Medicine team.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Differential longitudinal gene expression analysis of vedolizumab (VDZ) versus infliximab (IFX) in a densely sampled inflammatory bowel disease cohort.
a, Schematic of a prospectively recruited independent cohort of biologic-naive patients with inflammatory bowel disease (n = 18 VDZ (10 Crohn’s disease (CD), 8 ulcerative colitis (UC)); n = 12 IFX (4 CD, 8 UC)) sampled at baseline (BL), 2, 6, and 14 weeks after treatment initiation; in selected patients, additional early biopsies (24–72 h) were obtained. All patients had clinically active disease at BL and received induction therapy according to standard medical criteria. b, Dot plot showing GO terms enriched in genes differentially up- or downregulated in responders (R) versus nonresponders (NR) at BL. Angiogenesis pathway enrichment (ANGPT2, TIE1, ITGA5) was observed in VDZ NR but not IFX NR. Color corresponds to the p-value of enrichment (topGO elim algorithm with one-sided Fisher’s exact test). c,d, UpSet plots of differentially expressed genes in (c) VDZ-treated patients and (d) IFX-treated patients assessed by longitudinal pairwise analysis against BL (R and NR analyzed separately). Dot plots depict GO terms enriched in corresponding gene sets for remitters. Color corresponds to the p-value of enrichment (topGO elim algorithm with one-sided Fisher’s exact test). DEG, differentially expressed gene; IgG, immunoglobulin G.
Extended Data Fig. 2 Targeted analysis of vascular inflammation-related transcripts in the longitudinal cohort.
Targeted analyses of vascular inflammation-related gene sets identified in cohorts 1 and 2 were evaluated in the longitudinal cohort. a, Combined patients with UC and CD treated with vedolizumab. b, Patients with UC treated with vedolizumab. c, Patients with UC treated with infliximab. Variance-stabilized normalized expression values are shown based on pairwise comparisons per time point. Thick blue lines represent median expression in responders (R), thick red lines in nonresponders (NR); thin lines denote individual patients. Adjusted p-values: <0.05 (*), <0.01 (**) based on DESeq. ns, nonsignificant.
Extended Data Fig. 3 Cell type enrichment of vascular inflammation module score.
a, Forest plots showing cell type enrichment of the vascular inflammation module using sample-level module scores in two scRNA-seq cohorts (ref. 27 (right) and ref. 28 (left)). For each annotated cell type, a linear mixed-effects model compared the module score in the target cell type with all other cell types, adjusting for disease and inflammation status, with each sample included as a random effector. X-axis shows adjusted mean difference. Positive estimates indicate higher module scores in the target cell type compared with other cell types, whereas negative estimates indicate lower module scores (dashed line indicates no (0) estimated difference). Dot size represents −log10(Benjamini–Hochberg (BH)-adjusted p value), and horizontal bars represent 95% confidence intervals. Colors indicate enrichment status based on the direction of the estimate and BH-adjusted significance: green, enriched; red, depleted; blue, not significant. P values were adjusted across tested cell types using the BH method. b, Violin plots showing elevated expression of pro-inflammatory mediators, including IL1RN, CD300E, TREM1, IL1B, and S100A12, in inflammatory monocytes across two public scRNA-seq cohorts: ref. 27 (right) and ref. 28 (left).
Extended Data Fig. 4 CITE-seq and spatial multi-omics.
a, UMAP visualization of the cell type annotation based on the RNA assay from the CITE-seq dataset. b, Heatmap showing the number of predicted ligand–receptor interactions among annotated cell types. X-axis represents sender/source cell types, and y-axis represents receiver/target cell types. Color intensity indicates the number of predicted interactions for each pair. The bars along the top and right summarize the total number of interactions for each receiver and sender cell type, respectively. SPP1+ macrophages showed a higher number of interactions with endothelial cells compared with SPP1− macrophages. c, Circle plot showing the number of predicted ligand–receptor interactions among annotated cell types. Node size represents the total number of interactions associated with each cell type, and edge thickness represents the number of interactions between cell type pairs. Arrows indicate the direction of signaling from sender to receiver. d, Dot plot showing the top 5 RNA markers of each RNA-based CITE-seq cell type. Each dot represents the average target expression for each cell type. Dot color intensity reflects average expression level (red = high, gray = low), and dot size indicates the percentage of cells expressing the corresponding genes. e, Dot plot showing the top 5 protein antibody markers of each RNA-based CITE-seq cell type. Each dot represents the average target expression for each cell type. Dot color intensity reflects average expression level (red = high, gray = low), and dot size indicates the percentage of cells expressing the corresponding genes. f, Multi-omics spatial analysis highlighting an SPP1+ macrophage (arrow) in close proximity to fibroblasts (fibronectin, red) and inflammatory endothelial cells (CD31, blue; ICAM1, purple). Left: CosMx proteomic image displaying DNA (gray), CD163+ macrophages (yellow), and the spatial arrangement of surrounding cell types. Right: corresponding CosMx transcriptomic map showing SPP1 (red) and CHI3L1 (blue) RNA expression.
Extended Data Fig. 5 Spatial single-cell whole transcriptome and predicted protein interaction.
a, UMAP visualization of de novo clustering based on the Harmony-integrated Vedo CosMx cohort 3. Clusters were collapsed into broad cell-type annotations as follows: cluster 0 = lymphocytes; cluster 0 = lymphocytes; clusters 1, 4, 14, 15, and 16 = epithelial cells; clusters 2, 12, and 13 = fibroblasts; clusters 3, 6, 8, 9, and 17 = plasma cells; cluster 5 = macrophages; cluster 7 = endothelial cells; cluster 10 = granulocytes; and clusters 11 and 18 = undifferentiated cells. b, Dot plot showing expression of selected inflammatory monocyte–macrophage marker genes in SPP1+ and SPP1− macrophages. Dot color represents average expression, and dot size represents the percentage of cells expressing each gene. Statistical significance between SPP1+ and SPP1− macrophages was assessed using the Wilcoxon rank-sum test, with P values adjusted for multiple comparisons using the Benjamini–Hochberg method. Significance levels are denoted as follows: ns, not significant; p < 0.05; *p < 0.01; **p < 0.001; ***p<0.0001. c, Dot plot showing the top five marker genes for each de novo cluster. Dot size represents the percentage of cells expressing each marker, and color intensity represents average expression.d, PPI network identified by Metascape using highly spatially variable genes in baseline Vedo nonresponders from both Vedo CosMx cohorts 1 and 2. Nodes represent proteins encoded by the input genes, and edges represent known or predicted protein–protein interactions. Node size indicates the number of direct connections each node.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Schreiber, S., Yajnik, V., Wei, C. et al. An ulcerative colitis taxonomy defined by individual patient-reported response trajectories. Nat Med (2026). https://doi.org/10.1038/s41591-026-04741-8
Received: 31 October 2025
Accepted: 06 October 2026
Published: 09 November 2026
Version of record: 09 November 2026
DOI: https://doi.org/10.1038/s41591-026-04741-8
来源:Nature Medicine · nature.com