B cell-guided inflammatory cascade perpetuates ICI-mediated colitis

Modeling of ICI-related intestinal inflammation in mice

Multiple studies demonstrate that patients with cancer who develop ICI-colitis have subclinical disease that rapidly becomes symptomatic upon ICI treatment.¹³ Therefore, we implemented a model of subclinical colitis that fully manifests only after ICI therapy. A total of 2.5% dextran sodium sulfate (DSS) was administered to C57BL/6 mice for 3 days to chemically induce latent inflammation that causes epithelial injury and microbiome alterations like those observed in patients with colitis-irAE.^4,14–20 On days 3, 5, and 7, mice were intraperitoneally (i.p.) injected with isotype IgG control, anti-PD-1, anti-CTLA-4, or a combination of anti-PD-1 and anti-CTLA-4 (combo ICIs). As a positive control, we employed the classical model of DSS-colitis by administering 4% DSS for 5 days without additional treatment (Figure 1A). Mice treated with isotype IgG control did not develop clinical signs of colitis such as weight loss, shortened colon lengths, diarrhea, or hematochezia. Conversely, mice treated with anti-CTLA-4 or combo ICIs experienced significant loss of body weight, decrease in colon length, severe diarrhea, and hematochezia (Figures 1B–1D; Figure S1A). Furthermore, greater histological damage was observed in mice treated with anti-CTLA-4 or combo ICIs than in other groups, with the highest severity with combo ICIs (Figures 1E and 1F). Symptoms were comparable between the combo ICIs and classic colitis groups, indicating that combo ICIs were sufficient to induce severe colitis in asymptomatic mice primed with a low dose of DSS. Notably, mice treated with anti-PD-1 monotherapy developed milder disease, as is also observed in patient studies.⁴ These findings collectively demonstrate that our model efficiently induces subclinical colonic inflammation that only becomes symptomatic after ICI administration.

ICI-induced colitis is associated with the expansion of colonic T and B cells

In patients, ICI-colitis is marked by an increase in CD4⁺ and CD8⁺ T cells in colonic tissues with type 1 and 17 effector characteristics.^6,21 We profiled the immune cell repertoire in the colon and mesenteric lymph nodes (mLNs) of healthy, isotype IgG control, ICI monotherapy, and combo ICI groups and detected a significant expansion of CD4⁺ T, CD8⁺ T, and B cells in the colons of the combo ICI-treated mice compared to healthy and isotype IgG control (Figures 1G and 1H; Figures S1B–S1D, S2A, and S2B). The elevated number of colonic T and B cells was comparable between the combo ICI and the classical colitis groups (Figures 1G and 1H). No differences were observed in the frequencies of the analyzed innate cell subsets (natural killer [NK] cells, macrophages [Mϕs], neutrophils, and dendritic cells [DCs]) or Foxp3⁺ T regulatory (Treg) cells²² with mono and combo ICI therapies (Figure 1G; Figures S1D, S1F, and S2A).

Immune profiling in the mLNs revealed a similar increase in cellularity with combo ICIs (Figure S1C), with no significant alterations in the number of Mϕs (Figure S1E). However, anti-CTLA-4 elicited an expansion of NK cells and DCs compared to IgG isotype control treatment. Furthermore, anti-CTLA-4 monotherapy and combo ICIs induced an expansion of neutrophils in the mLNs compared to healthy, isotype IgG-treated, and anti-PD-1 monotherapy groups (Figure S1E). Although alterations in the composition of the innate compartment were observed in mLNs, the most striking changes occurred in adaptive immunity (Figure 1I). In contrast to the increased colonic CD4⁺ and CD8⁺ T cells in ICI-treated mice, there was a corresponding decline in these subsets in the mLNs, suggesting preferential retention of T cells within the colon post-ICI therapy (Figure 1J). Alternatively, like in the colon, we observed a significant increase in the number of B cells and no differences in the number of Treg cells within mLNs of combo ICI-treated mice (Figures 1J and S1F). Additionally, we found that ICI therapy did not significantly alter the expression of PD-1 or CTLA-4 in the colonic or mLN CD4⁺ T, CD8⁺ T, or B cells of DSS-treated mice (Figures S1G–S1L). Since the strongest clinical and immunologic phenotypes were observed with combo ICIs, this treatment was employed for subsequent mechanistic studies.

Type 1 and 17 cytokines contribute to ICI-colitis pathogenesis

We next validated if colonic T cells have similar cytokine profiles to those seen in patients with colitis-irAE. To address this, unbiased clustering was performed on colonic CD3⁺ T cells of combo ICI- or isotype IgG control-treated mice (Figures S3A–S3C). We identified five clusters based on their expression of PD-1 and the cytokines IFN-γ, TNF-α, IL-17A, granzyme B, and granulocyte-macrophage colony-stimulating factor (GM-CSF). Notably, cluster 4, consisting of T cells enriched in PD-1, IFN-γ, TNF-α, IL-17A, and GM-CSF, was overrepresented in the colons of ICI-treated mice (Figures S3A–S3C). Quantitatively, IL-17A⁺ and IFN-γ⁺ CD8⁺ T cells were significantly enriched in the colons and mLNs of ICI-colitis mice compared to healthy or isotype IgG controls (Figures 2A–2D; Figures S3D and S3E). Also, IL-17A⁺, IFN-γ⁺, and TNF-α⁺ CD4⁺ T cells were enhanced in the colons of ICI-treated mice compared to controls, and these proinflammatory cytokine-producing CD8⁺ and CD4⁺ T cells were correlated with increased colitis severity score and decreased colon length by principal component analysis (PCA) (Figures 2A and 2B). These data demonstrate that our model not only recapitulates the clinical phenotype but also the proinflammatory T cell cytokine signatures observed in patients.

Furthermore, blockade of IL-17A, IFN-γ, or TNF-α prior to ICI administration deterred clinical development of colitis compared to untreated mice (Figures 2E–2I; Figure S3F). Additionally, we targeted other cytokines involved in promoting Th17 cell differentiation (IL-6, IL-1β, or IL-23).²³ Blockade of IL-6 and IL-1β, but not IL-23, was sufficient to improve body weight and colon length in ICI-treated mice; however, they did not significantly improve the histological score. Interestingly, IL-23 blockade was the only condition to significantly reduce histological damage within the colon (Figures S3G–S3K). Blockade of these proinflammatory cytokines underscores the clinical relevance of this model.

The role of B cells in ICI-colitis development through supporting proinflammatory T cells

While the involvement of T cells in the pathology of ICI-colitis has been extensively studied, the role of B cells is not yet fully understood. B cells express both CTLA-4 and PD-1, and ICIs lead to an increase in the number and function of circulating B cells in patients with irAE.^24,25 We observed a pronounced increase in B cells in the colons and mLNs of ICI-colitis mice and profiled their expression of immunoglobulins (Igs) and cytokines. The overall increase in B cell numbers was attributable to the increased total amount of B cells within the colons and mLNs of colitis-irAE mice rather than an expansion of a specific subset (Figures S4A and S2B). While there were no significant differences in IgG1 and IgA expression levels within colonic plasma cells between groups, both colonic and mLN B cells from the combo ICI group had significantly higher levels of IL-17A expression compared to healthy mice (Figures 3A and 3B; Figure S4B). In addition, no significant differences were noted in IL-10, IL-12p40, TNF-α, TGF-β, or IL-35 in mLN and colonic B cells, but there was a slight decrease in IL-6⁺ B cells in the mLNs of ICI-treated mice compared to healthy mice (Figures 3A and 3B; Figures S4C and S2B).

We further validated the B cell contribution to ICI-colitis and found that B cell depletion prior to ICI administration significantly decreased all clinical signs of colitis after ICI treatment (Figures 3C–3H). Lack of B cells was associated with a significant reduction of IL-17A, IFN-γ, and TNF-α in CD8⁺ T cells and a decrease of IFN-γ in colonic CD4⁺ T cells (Figures 3I and 3J). While B cell depletion did not change the expression of these cytokines in mLN T cells, it significantly enhanced the proportion of CD4⁺ and CD8⁺ T cells compared to that in colitis-irAE mice, which is likely a reflection of B cell depletion (Figures S4D–S4F). Furthermore, we found that B cell depletion does not impact colitis development in the non-ICI-mediated classic colitis model (Figures S4G–S4J). Thus, these data highlight a unique pathogenic role for B cells in ICI-colitis development that differs from idiopathic inflammatory bowel disease (IBD), which may be mediated through the regulation of the number and function of T cells.

Reciprocal regulation between B cells and proinflammatory CD8⁺ T cells

Observing dampened CD8⁺ T cell responses upon B cell depletion, we sought to determine if there is a reciprocal crosstalk between CD8⁺ T cells and B cells. We depleted CD8⁺ T cells and observed a rescue of ICI-colitis development (Figures S5A–S5F). While CD8⁺ T cell depletion dramatically decreased the number of colonic B cells in ICI-treated mice, there was only a modest decrease in IL-17A and IL-6 and an enhanced IL-12p40 expression in B cells (Figure S5G). Moreover, CD8⁺ T cell depletion led to a slight decrease in the number of CD4⁺ T cells as well as reduced IL-17A and TNF-α expression in the colon (Figure S5H). Additionally, there were no differences in the proportion of B cells or their functional status in the mLNs of ICI-colitis mice and those treated with anti-CD8 monoclonal antibodies (mAbs) (Figure S5I). Although there was profound enhancement in the proportion of CD4⁺ T cells in the mLN of CD8-depleted mice, this is likely a reflection of the loss of the CD8⁺ T cells (Figures S5J and S5K). Collectively, these data indicate that there is reciprocal regulation between CD8⁺ T cells and B cells in which B cells contribute to the proinflammatory nature of CD8⁺ T cells, while those cells influence subsequent B cell recruitment/retention. This process may form a feedback loop to not only encourage but also perpetuate inflammation in ICI-colitis.

Proinflammatory B and T cell responses are consequences of intestinal dysbiosis

Previous studies have demonstrated that ICI-colitis is marked by intestinal microbiota dysbiosis and that baseline microbial composition can be predictive of colitis development.^8,9 Thus, to determine if dysbiosis marks and/or is associated with colitis development, we performed 16S rRNA sequencing on fecal samples collected prior to any treatment, after 3 days of DSS, and following two doses of combo ICIs (Figures 4A–4C). While there were no significant differences in alpha diversity, the beta diversities of each group were distinct, suggesting that alterations in the composition of the microbiota occur both prior to and after ICI administration (Figures 4A and 4B). These compositional differences were characterized by a significant decrease in the abundance of Firmicutes in ICI-treated mice with a concomitant increased relative abundance of the Verrucomicrobiota phylum (Figures S6A and S6B). Regardless of ICI administration, colitis-susceptible mice displayed a trend toward decreased Desulfobacterota and Actinobacteriota phyla when compared to untreated mice. Furthermore, LefSe (Linear discriminant analysis Effect Size) analysis identified a cohort of differentially abundant microbes. Of interest, ICI-colitis mice were enriched in Akkermansia, the only constituent of the Verrucomicrobiota phylum in this dataset, and depleted in Negativibaciilus when compared to either control. Further, both colitis-susceptible groups prior to and after ICI displayed a depletion of Prevotellaceae and Roseburia compared to healthy controls (Figure 4C; Figures S6A and S6B). These findings indicate that mice susceptible to colitis have a perturbed microbiota composition, which is further altered by ICI use, as observed in patient studies.

Next, to assess the impact of dysbiosis at the early (pre-ICI) stage on the onset of ICI-colitis, we administered fecal microbial transplant (FMT) from healthy mice to DSS-primed mice to bolster eubiotic microbiota and subsequently treated the mice with ICIs (Figure 4D). While FMT had no impact on body weight, we observed significant improvement in disease activity scores, colon lengths, and histological scores compared to untreated mice (Figures 4E–4G; Figures S6C and S6D). Clinical improvement was accompanied by reductions in the number of B and T cells within the colon (Figures 4H–4J). Functionally, FMT reduced IL-17A- and IL-6-expressing B cells but did not change IL-12p40 expression (Figure 4H). Additionally, FMT profoundly reduced the expression of IL-17A and IFN-γ in both CD4⁺ and CD8⁺ T cells as well as reduced TNF-α expression in CD4⁺ T cells (Figures 4I and 4J). Although there were no alterations in the functional profile of mLN T and B cells in FMT mice, there was a minor reduction in IL-17A and IL-12p40 expression by B cells as well as IL-17A expression by CD8⁺ T cells (Figure S6E). Collectively, these data suggest that latent intestinal injury and microbiota perturbations may be initiating factors in ICI-colitis, which instigates expansion of proinflammatory B and T cells, as maintenance of a healthy microbiome deters this pathology.

ICI-driven colonic inflammation is preceded by alterations in B cell number and function

To gain mechanistic insight into the pathogenesis of ICI-colitis, we next assessed how the immune response develops over time in different tissues, including peripheral blood (PB), mLNs, and the colon (Figure 5A). Notably, the proportion of circulating B cells was enhanced pre-ICI (colitis-susceptible mice), prior to any clinical signs of colitis development, and significantly reduced following ICI treatment (Figure 5B). In contrast, the number of colonic and mLN B cells increased only after ICI administration (Figure 5B). In PB, the proportions of both CD8⁺ and CD4⁺ T cells significantly increased only after two ICI doses, while mLN proportions were profoundly decreased (Figures 5C and 5D). On the other hand, colonic proportions of CD8⁺ and CD4⁺ T cells began to show enhancements in the pre-ICI condition but were not yet significantly enriched following two doses of combo ICIs (Figures 5C and 5D). Importantly, colonic tissues only become enriched with IL-17A and IFN-γ producing CD4⁺ and CD8⁺ T cells after ICI administration, which coordinates with a deficit of these cells in the PB (Figure S7A).

An enhanced number of circulating B cells was one of the early events that was observed in colitis-susceptible mice (prior to ICIs), and further phenotypic characterization of B cells revealed distinct shifts in their functional status across the PB, mLNs, and the colon with colitis development (Figures 5E–5G). Prior to ICIs, colitis-susceptible mice had a greater number of circulating B cells expressing major histocompatibility complex (MHC) I/II, costimulatory molecules (CD80, CD86, CD40, and ICOSL), and the gut-homing receptor (CCR9), which all decreased with ICIs (Figures 5E–5G). This efflux of activated B cells from PB aligns with an increased number of B cells expressing gut-directing/retention markers (CCR9, LPAM-1, and CD103) and cytokines (IL-17A, IL-12p40, and IL-6) within the colon after ICIs. Additionally, after ICIs, there was an increase in the number of colonic CXCR4⁺ and CXCR5⁺ B cells as well as a decrease in the number of B cells expressing T cell immunoglobulin and mucin domain 1 (TIM-1), an immune checkpoint for B cells (Figures 5E–5G).²⁶ Altogether, these data suggest that B cells with antigen-presenting capabilities may mobilize early because of latent colon injury or inflammation and influence the number and function of T cells.

In addition to B cells, DCs and Mϕs are other antigen-presenting cells (APCs) that respond to the gut microbiota, maintain barrier integrity, and can prime adaptive immune responses.²⁷ Disruptions in the homeostatic functions of these cells have been implicated in driving ICI-colitis.¹² While we did not see a difference in the number of conventional DCs and Mϕs either pre-ICI, with the exception of DC2 in the colon, or 5 days after combo ICIs (Figures 5H–5K; Figures S1D, S1E, S2, and S7B), the proportions of colonic DC1 and DC2 were significantly enhanced 3 days post-ICIs when compared to untreated mice (Figures 5H and 5I). In addition, we observed a significant increase in the proportion of DCs expressing CCR7, a chemokine involved in the mobilization of DCs to lymphoid tissues, in both the colon and mLNs post-ICIs (Figures 5J and S7B). Conversely, we observed a decrease in the proportion of Mϕs in both the colon and mLNs (Figure 5K). Thus, the DC number and tissue localization are mainly altered post-ICIs.

Dysregulated B cell responses to intestinal injury are key to incite proinflammatory DC and T cell activity

Given the early alterations in the number and function of circulating B cells prior to ICIs, we further examined the mechanism by which B cells contribute to the tissue-specific immune environment pre- and post-ICIs (Figure 5L). With B cell depletion prior to establishing subclinical inflammation, the number of mLN DCs significantly increased both pre- and post-ICIs compared to untreated mice (Figures 5M–5O). Within the colon, there was a significant reduction in CCR7⁺ DCs and minimal changes in DC1 and DC2 post-ICIs (Figures 5M–5O). Functionally, without B cells, the expanded DCs in the mLNs displayed a more immunoregulatory phenotype after ICIs with enhanced expression of CD83 and CD200 across subsets and a decrease in IL-12p40 in DC2 (Figure 5P).^28–33 In contrast, there were no significant alterations in the functional status of colonic DCs except a decrease in IL-6 and IL-12p40 expression in CCR7⁺ DCs (Figure S7C). Collectively, these findings suggest that DC migratory/retention patterns and activation status during ICI-colitis rely on B cells.

Next, we evaluated the tissue distribution and activity of T cells in ICI-colitis mice lacking B cells over time. While there were no significant alterations in the number or functional status of colonic CD4⁺ T cells upon B cell depletion, the CD8⁺ T cells had significant reductions in the expression of IFN-γ, IL-17A, and TNF-α in the absence of B cells (Figures S7D–S7G). Additionally, mLN CD4⁺ and CD8⁺ T cell numbers were increased in both pre- and post-ICI conditions upon B cell depletion (Figures S7H and S7I). CD8⁺ T cells had increased IL-17A, and both CD4⁺ and CD8⁺ T cells had decreased TNF-α expression within mLNs (Figures S7J and S7K). On the other hand, there were no differences in Treg numbers in the colons or mLNs with B cell depletion (Figure S7L). Furthermore, after ICIs, we observed an accumulation of CD8⁺ and CD4⁺ T cells expressing α4β7 (LPAM-1) integrin and CCR9 in the mLNs of B cell-depleted mice (Figures 5Q and 5R). Both α4β7 (LPAM-1) integrin and CCR9 are essential to the homing of activated lymphocytes to the intestine through interactions with mucosal addressin cell adhesion molecule 1 (MAdCAM-1) expressed on endothelial venules and the chemokine CCL25, respectively.^34–36 This finding suggests that, during colitis development, the functional capacity and migratory patterns of DCs and T cells are reliant on B cell activation status and provides insight into the mechanism by which B cell targeting deters the development of ICI-mediated colitis (Figure S7M).

Increase in circulating B cell number is associated with subsequent ICI-colitis development in patients with cancer

Current mechanistic understanding of ICI-colitis has been derived from single-cell RNA analyses with samples collected from patients at or after the onset of toxicity.^6,21,37 These samples are often collected after immunosuppressive therapies, which may mask cellular drivers of disease. Thus, baseline samples obtained from patients with cancer prior to the administration of ICIs are imperative to identify candidate cellular and molecular biomarkers of ICI-colitis risk and to inform prophylactic and therapeutic strategies. Here, we analyzed baseline peripheral blood mononuclear cells (PBMCs) from a pilot cohort of 10 patients with melanoma. The median age was 59.5 years (range 41–78), 90% were male, and 90% had stage III/IV cutaneous melanoma (according to the American Joint Committee on Cancer Version 8 staging system). Six patients (60%) subsequently received combination ICIs and four (40%) received single-agent anti-PD-1, with varying intervals between PBMC collection and ICI initiation across patients (Figure 6A; Table S1). Following ICIs, 4 patients (40%) developed grade 3 ICI-colitis per the Common Terminology Criteria for Adverse Events (CTCAE) guidelines (all 4 occurred in the combo-ICI group), whereas six did not develop any irAE through the last available follow-up. Flow cytometric analysis revealed that patients who developed ICI-colitis had a higher proportion of circulating B cells at baseline compared to those who did not develop any irAE (Figures 6B and 6C; Figure S8). Additionally, the circulating B cells between these two groups were phenotypically distinct. In the ICI-colitis group, there was an increased number of B cells expressing the gut-homing molecules CCR9 and LPAM-1, costimulatory molecules such as CD80 and CD86, as well as the cytokines IL-17A, TNF-α, IL-6, and IL-10 when compared to the non-irAE group (Figures 6D and 6E; Figure S8). Notably, these baseline human findings were consistent with the pre-ICI stage of our mouse model, in which colitis-susceptible mice similarly showed expansion of circulating B cells expressing gut-homing features and an activated/APC-like phenotype compared to controls. Collectively, these human data, together with our mechanistic mouse studies, support the hypothesis that the abundance and functional status of circulating B cells prior to ICI exposure are associated with subsequent ICI-colitis risk. Larger prospective cohorts stratified by ICI regimen and by timing of baseline sampling are needed to validate B cell-based biomarkers and to define their predictive performance.

B cells play a role in immune cell homing to the colon during ICI-colitis

Immune cell trafficking to the gut has been shown to be a critical component in ICI-colitis pathogenesis, especially by the α4β7 integrin/MAdCAM-1 mechanism.^6,29,38 To validate that α4β7 integrin-mediated trafficking is essential to the development of ICI-colitis in our model, we administered α4β7-blocking mAbs in conjunction with ICIs to impede homing of T and B cells to the colon (Figure S9A). While α4β7 integrin blockade did not deter weight loss, there was significant improvement in all other symptomatic assessments of treated mice (Figures S9B–S9F). These clinical improvements were due to a profound reduction in the number of B and T cells as well as their proinflammatory cytokine signatures (IL-17A, IFN-γ, and TNF-α) within the colon of anti-α4β7-treated mice compared to mice receiving ICIs alone (Figures S9G–S9I). Within the mLN of these mice, there were no alterations in the proportion and function of T or B cells (Figures S9J–S9L). These data collectively implicate α4β7-mediated gut homing of T and B cells as a contributor to colonic inflammation in the ICI setting.

To address the contribution of immune cell trafficking between lymphoid tissues and the colon in ICI-colitis, we adoptively transferred mLN cells from colitis-susceptible mice to naive (untreated) mice followed by isotype IgG control or combo ICI administration (Figure 7A). Unlike isotype IgG controls, treatment of recipient mice with ICIs was sufficient to induce clinical signs of ICI-colitis (Figures 7A–7E). Interestingly, anti-α4β7 treatment prior to cell transfer improved the colon lengths, cellularity, and histological score of ICI-treated mice compared to those who did not receive integrin blockade (Figures 7A–7E). Transferred T and B cells repopulated into the mLNs, with most recovered cells being T cells (Figure S9M). Although transferred cells constituted a small proportion of the overall immune cells within the colons and mLNs of recipient mice, they displayed unique trafficking dynamics based on the therapy administered (Figures 7F and 7G). Only mice receiving combination ICIs had equivalent proportions of transferred cells homing to the colon and mLNs. In contrast, cells transferred to recipient mice, which did not develop colitis, displayed a propensity to traffic to the mLNs. Notably, integrin blockade in conjunction with ICIs enhanced transferred cell trafficking to the heart in recipient mice, which was not observed in any other group (Figures 7F and 7G). Within recipient mice, we observed an increased number of T and B cells in colons of ICI-treated mice compared to the other groups; however, CD8⁺ T cells were the only subset to show significant elevation (Figure 7H). Further, mLNs of the ICI-treated mice had a significantly larger B cell population compared to other groups, while T cells tended to decrease with significant decreases in the number of CD4⁺ T cells (Figure 7H). Functionally, compared to the other groups, colon T and B cells of ICI-treated mice were enriched with IFN-γ⁺ CD4⁺ T and CD8⁺ T cells as well as IL-17A⁺ CD8⁺ T and B cells (Figure 7I). In addition, the mLNs of these mice were also enriched in IFN-γ⁺ CD4⁺ T and CD8⁺ T cells as well as IL-17A⁺ CD8⁺ T cells compared to the other groups (Figure 7I). Furthermore, cell transfer and ICI administration, with or without integrin blockade, led to an enhancement in the number of DCs and Mϕs within the mLN, compared to the transfer-alone and healthy groups (Figure S9N).

To further understand the role of B cells in the onset of ICI-colitis, we repeated the adoptive transfer with a set of donor mice in which B cells were depleted during DSS administration (Figure 7J). Mice that received B cell-depleted mLN cells and subsequent combo ICI treatment demonstrated longer colon length, decreased colon cellularity, and decreased histological damage compared to recipient mice receiving B cell-containing mLN cells (Figures 7K–7M). Additionally, mice receiving B cell-depleted mLN cell transfer had significantly less cells migrating to the colon after ICIs than recipients of undepleted donors, indicating that B cells may alter the migratory patterns of T cells in the ICI setting (Figure 7N). Furthermore, impediment of colitis was marked by reductions in IFN-γ expression by CD4⁺ and CD8⁺ T cells as well as IL-17A expression by CD8⁺ T and B cells within the colons of ICI-treated mice with B cell-deficient mLN cell transfer (Figure 7O). Overall, the aforementioned data indicate that trafficking of T and B cells is a critical step in colitis development and that lack of B cells impairs this process and reduces the proinflammatory status of T cells in the ICI setting.

Limitations of the study

This study has potential limitations. Firstly, our findings need to be further validated with expanded patient studies. The human data in this study are exploratory and extremely limited, reflecting a small pilot cohort with clinical heterogeneity, including variability in the interval between baseline PBMC collection and ICI initiation and imbalance in treatment regimen between groups. Therefore, these data should be interpreted as hypothesis-generating rather than establishing biomarker performance. These findings will be further validated in an ongoing investigator-initiated study (NCT04990726) expanding these analyses in larger, regimen-stratified cohorts using paired standardized pre-ICI and early on-treatment post-ICI samples collected prior to colitis onset, with appropriately matched controls. Furthermore, our mouse model findings need to be corroborated using a syngeneic tumor model to understand the contributions of cancer to ICI-colitis development as our preclinical experiments were performed in non-tumor-bearing mice. Lastly, this work primarily utilizes flow cytometry-based methods that can limit the available information, which needs to be further refined using higher throughput methods (e.g., single-cell transcriptomic/proteomic and spatial profiling) to elucidate cellular states and dynamic interactions across various colonic niches.

RESOURCE AVAILABILITY

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

METHOD DETAILS

QUANTIFICATION AND STATISTICAL ANALYSIS

Data were graphed in GraphPad Prism 10 or RStudio and analyzed using t-tests, multiple t test, One-way ANOVA, or two-Way ANOVA in GraphPad Prism where applicable (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns not significant). Differences were considered significant if p value <0.05. All experiments were performed at least twice, with similar results. All results are expressed as with the mean ± SEM. All N numbers represent biological replicates. A student’s two-tailed unpaired t test was used to determine the significance between means of two groups. One way or two-way ANOVA was used to determine the significance between means of three or more groups. Bonferroni post-hoc tests were used to adjust for pairwise comparisons where applicable. Data distribution was assumed to be normal, with no normality tests performed.