You are here

  • Home
  • Archive
  • Volume 12, Issue 1
  • Lower frequencies of circulating suppressive regulatory T cells and higher frequencies of CD4+ naïve T cells at baseline are associated with severe immune-related adverse events in immune checkpoint inhibitor-treated melanoma

Article Text

Article menu
Immunotherapy biomarkers
Original research
Lower frequencies of circulating suppressive regulatory T cells and higher frequencies of CD4+ naïve T cells at baseline are associated with severe immune-related adverse events in immune checkpoint inhibitor-treated melanoma
  1. Magdalena Kovacsovics-Bankowski1,
  2. Johanna M Sweere2,
  3. Connor P Healy2,
  4. Natalia Sigal2,
  5. Li-Chun Cheng2,
  6. William D Chronister2,
  7. Shane A Evans2,
  8. John Marsiglio3,
  9. Berit Gibson1,
  10. Umang Swami1,
  11. Alyssa Erickson-Wayman1,
  12. Jordan P McPherson4,
  13. Yoko S Derose1,
  14. Annaleah Larson Eliason3,
  15. Carlos O Medina2,
  16. Ramji Srinivasan2,
  17. Matthew H Spitzer2,5,
  18. Ngan Nguyen2,
  19. John Hyngstrom1 and
  20. Siwen Hu-Lieskovan1
  1. 1Huntsman Cancer Institute, University of Utah School of Medicine, Salt Lake City, Utah, USA
  2. 2Teiko.bio Inc, Salt Lake City, Utah, USA
  3. 3The University of Utah School of Medicine, Salt Lake City, Utah, USA
  4. 4Department of Pharmacy, Huntsman Cancer Institute Cancer Hospital, Salt Lake City, Utah, USA
  5. 5Department of Otolaryngology-Head and Neck Cancer, University of California San Francisco, San Francisco, California, USA
  1. Correspondence to Dr Siwen Hu-Lieskovan; Siwen.Hu-Lieskovan{at}hci.utah.edu

  • MK-B and JMS are joint first authors.

Abstract

Background Immune-related adverse events (irAEs) are major barriers of clinical management and further development of immune checkpoint inhibitors (ICIs) for cancer therapy. Therefore, biomarkers associated with the onset of severe irAEs are needed. In this study, we aimed to identify immune features detectable in peripheral blood and associated with the development of severe irAEs that required clinical intervention.

Methods We used a 43-marker mass cytometry panel to characterize peripheral blood mononuclear cells from 28 unique patients with melanoma across 29 lines of ICI therapy before treatment (baseline), before the onset of irAEs (pre-irAE) and at the peak of irAEs (irAE-max). In the 29 lines of ICI therapy, 18 resulted in severe irAEs and 11 did not.

Results Unsupervised and gated population analysis showed that patients with severe irAEs had a higher frequency of CD4+ naïve T cells and lower frequency of CD16+ natural killer (NK) cells at all time points. Gated population analysis additionally showed that patients with severe irAEs had fewer T cell immunoreceptor with Ig and ITIM domain (TIGIT+) regulatory T cells at baseline and more activated CD38+ CD4+ central memory T cells (TCM) and CD39+ and Human Leukocyte Antigen-DR Isotype (HLA-DR)+ CD8+ TCM at peak of irAEs. The differentiating immune features at baseline were predominantly seen in patients with gastrointestinal and cutaneous irAEs and type 1 diabetes. Higher frequencies of CD4+ naïve T cells and lower frequencies of CD16+ NK cells were also associated with clinical benefit to ICI therapy.

Conclusions This study demonstrates that high-dimensional immune profiling can reveal novel blood-based immune signatures associated with risk and mechanism of severe irAEs. Development of severe irAEs in melanoma could be the result of reduced immune inhibitory capacity pre-ICI treatment, resulting in more activated TCM cells after treatment.

  • Immune Checkpoint Inhibitors
  • Melanoma
  • Immunotherapy
  • T-Lymphocytes, Regulatory
  • Immunologic Techniques

Data availability statement

Data are available upon reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/jitc-2023-008056

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Immune checkpoint inhibitors (ICI) can lead to severe immune-related adverse events (irAE) that hinder their clinical application, interfere with development of more effective combination immunotherapy, and compromise patients’ lifestyle, however, there has not been reliable biomarkers for irAE prediction to guide clinical practice.

    WHAT THIS STUDY ADDS

    • Our study shows that CD4+naïve T cells, CD16+natural killer (NK) cells and T cell immunoreceptor with Ig and ITIM domain (TIGIT)+regulatory T cells (Treg) cells are present at significantly different frequencies in the peripheral blood before ICI-treatment in patients with severe irAE compared with patients without severe irAE, and that peak of severe irAE is associated with activated CD4+ and CD8+ memory T cells.

    • These immune features are not associated with BRAF mutation status and are generally observed across different irAE categories.

    • Patients with severe irAE were more likely to observe clinical benefit from ICI, and patients with clinical benefit also had significantly different frequencies of CD4+naive T cells, CD16+NK cells and suppressive Treg at baseline compared with patients without clinical benefit.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • With additional validation in larger patient cohorts, these baseline peripheral immune features could be developed into potential biomarkers associated with severe irAE and/or clinical benefit and help to elucidate the mechanisms of ICI-induced irAEs.

    Introduction

    Immune checkpoint inhibitors (ICIs) have dramatically changed the treatment schema of various cancers over the last decade.1 ICIs block pathways that regulate T-cell activation, resulting in more efficient antitumor responses. The clinical use of ICIs had a major impact on the overall survival of patients with advanced melanoma, especially when used in combination.2 However, they also cause immune-related adverse events (irAEs). IrAEs can involve all tissues and severity varies from grade 0 (no irAE) to grade 5 (resulting in death of the patient). While irAEs are caused by activation of autoreactive T cells,3 the precise mechanisms of irAEs are not fully understood and are complex.

    A meta-analysis performed on more than 15,000 patients from 36 randomized clinical trials, using anti-programmed cell death protein 1 (PD-1), anti-programmed cell death ligand 1 (PD-L1), anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) alone or in combination, reported that the probability to develop any irAEs on either was 54–76% and the probability to present an irAE grade 3–4 was 28–74%.4 The management of patients developing irAEs depends on severity. Supportive care is usually provided for grade 1–2 irAEs. For grade 3–4 irAEs, holding the ICIs and administering immunosuppressive drugs is required, starting with high dose glucocorticoids. Permanently discontinuing ICIs is recommended for grade 4 irAEs.5–7 Interestingly, the occurrence of irAEs has been correlated with a favorable clinical outcome in several studies.8 On the other hand, a recent retrospective study suggests early administration of glucocorticoids for irAEs associated with anti-PD-1 treatment may result in a negative clinical outcome.9

    As some irAEs are irreversible or require permanent discontinuation of ICIs, there is major interest in identifying biomarkers that can predict their occurrence. This has become increasingly relevant as ICIs are being used in earlier stages of melanoma and other cancers, including neoadjuvant and adjuvant treatments. Several studies have demonstrated that neutrophil to lymphocyte ratio, platelet to neutrophil ratio, or absolute lymphocyte count in combination with an absolute eosinophil count at the initiation of ICI therapy can be associated with an increased probability of developing irAEs.10–14 Cytokine levels, the presence of autoantibodies and certain human leukocyte antigen types are also associated with higher risk of developing irAEs.15–17

    T cells are the main target of ICIs, and changes in the phenotype or frequency of different T-cell populations could be associated with the occurrence of irAEs. Clonal expansion of peripheral CD4+ and CD8+ T cells is associated with irAEs in patients with advanced prostate cancer treated with ipilimumab.18 19 In a cohort of 17 patients with melanoma, who received either anti-PD-1 alone or with anti-CTLA-4, the frequency of peripheral activated CD4+ T cells increased after the first cycle of immunotherapy in patients who developed irAEs.20 These studies suggest that features of peripheral blood mononuclear cells (PBMCs) can be predictive biomarkers for irAEs. However, more extensive studies that identify specific T-cell subpopulations and their functional state are needed. In addition, few studies have assessed innate immune populations, such as natural killer (NK) cells and myeloid cells, and their role in irAE presentation. Mass cytometry allows for the characterization of larger T-cell subpopulations and innate immune populations by assessing up to 45 immune markers in one sample and has been used to identify immune features associated with clinical response or overall survival.21 22

    In this study, we used a 43-marker mass cytometry panel to characterize the frequency and the functional state of PBMC subsets at different time points in 28 patients with melanoma undergoing ICI therapy. PBMCs from patients who developed severe irAE were compared with time matched PBMC samples from patients who did not develop a severe irAE. The goal of the study was to identify immune populations or functional states associated with severe irAEs.

    Methods

    Patients and samples

    Patients’ clinical information was collected in an Research Electronic Data Capture (REDCap) database, including timing of ICI therapy, occurrences of irAEs and the grading and treatment that was provided for each individual irAE. Whole blood was collected into an EDTA vacutainer before the initiation of ICIs (baseline), before the onset of irAEs (pre-irAE) and at the peak of irAEs (irAE-max). PBMCs were isolated using a standard Ficoll Hypaque density gradient protocol by staff at the Biorepository and Molecular Pathology laboratory of HCI. After isolation, PBMCs were transferred into 15 mL flip top sterile RNAse/DNAse-free centrifuge tubes (CELLTREAT) and washed twice with phosphate-buffered saline (PBS) pH 7.4 (Life Technologies) before cryopreservation in freezing media and storage in liquid nitrogen.

    Panel design and heavy-metal conjugation of antibodies

    Some antibodies were obtained preconjugated to heavy-metal isotopes from Standard BioTools (formerly Fluidigm). In-house conjugations were performed as needed using the MaxPar X8 or MaxPar MCP9 Antibody-Labeling Kits (Standard BioTools) following an optimized and updated protocol23 and according to manufacturer instructions. The allocation of targets to appropriate heavy-metal isotopes was based on the sensitivity of the mass cytometer (eg, lower abundance targets were placed on higher sensitivity channels) and to avoid potential signal spillover as previously described.24

    Sample preparation

    Cryopreserved PBMC samples were thawed, washed, and resuspended in PBS+5 mM EDTA before cisplatin staining (Sigma) for 1 min at room temperature (RT). Cells were washed and fixed with 1.6% paraformaldehyde (PFA) in PBS EDTA BSA (PEB) (PBS+5 mM EDTA+0.5% BSA) for 10 min at RT. Cells were washed twice with PEB and stored at −80°C until staining.

    Sample barcoding

    To minimize batch-to-batch variation and improve data quality, samples were barcoded prior to antibody staining using a 20-Plex Pd Barcoding Kit (Standard BioTools) following manufacturer instructions. Briefly, after counting, 1 million cells were washed with MaxPar Barcode Perm Buffer (Standard BioTools) twice and barcoded with unique combinations of Pd isotopes for 30 min at room temperature on a shaker in Maxpar Barcode Perm Buffer. Cells were washed twice with eBioscience perm buffer and once with Maxpar Cell Staining Buffer (CSB, Standard BioTools) and pooled into a single tube.

    Antibody staining

    Antibody cocktail was prepared prior to staining according to previously determined antibody titers at 25 µL of staining volume per 1 million cells. The antibody cocktail was aliquoted and stored at −80°C until staining. The pooled barcoded sample was stained for 10 min at room temperature with Fc receptor blocking solution (BioLegend) followed by 1 hour incubation with the antibody staining cocktail at RT. PBMCs were stained using a panel of 43 antibodies (online supplemental table S1). Cells were washed twice with CSB and resuspended in intercalation solution (4% PFA in PBS and 0.5 mM iridium intercalator (Standard BioTools)) for 20 min at RT or pelleted and frozen at −80°C until acquisition.

    Supplemental material

    Data acquisition

    Before acquisition, samples were washed once in CSB and twice in CAS (Standard BioTools) and filtered through a cell strainer (Falcon). Cells were then resuspended at 0.7 million cells/mL in CAS supplemented with EQ4 element calibration beads (Standard BioTools) and acquired on a Helios mass cytometer (Standard BioTools). All samples met quality control standards. Manual gating of FCS files was performed using CellEngine (CellCarta, Montreal, Canada).

    Data analysis

    PBMCs were separated into different cell populations as shown (online supplemental figure S1). To account for cell loss during sample processing, we normalized the relative abundances of immune cell subsets to the total live lymphocytes per patient. Frequencies of immune cell subsets between patients with severe irAEs and patients without severe irAEs, or patients with clinical benefit and without clinical benefit were compared using significance analysis of microarrays (SAM)25 with a false discovery rate of 0.01 as the threshold for statistical significance. Changes in population frequencies between irAE-max and baseline were assessed with SAM with a false discovery rate of 0.01. Abundances of immune cell subsets between BRAF mutation (BRAF-Mut) patients with or without severe irAEs and BRAF wildtype (BRAF-WT) patients with or without severe irAEs were compared using Kruskal-Wallis test with post hoc Wilcoxon rank-sum pairwise analysis with a statistical significant cut-off at p value<0.05. Unsupervised clustering was performed using the FlowSOM program with k=80 clusters.26 Plots were produced with the R package ggplot2.

    Supplemental material

    Results

    Patients

    In this retrospective study, we performed analysis on banked PBMCs from 28 patients with melanoma undergoing ICI therapy (figure 1, online supplemental table S2). One patient was diagnosed with stage II, 17 with stage III, and 10 with stage IV melanoma. We did not separate patients into ICI-naïve versus ICI-experienced for this study. As one patient received 2 lines of treatment, we considered the samples to represent 29 lines of therapy. PBMCs were collected at three time points: prior to the start of ICI (baseline), before the occurrence of irAEs (pre-irAE) and at the peak of irAEs (irAE-max). For 76% of the patients, samples were collected during first line ICI therapy. Treatments include neoadjuvant (n=1), adjuvant (n=18) and palliative (n=10). 22 lines of therapy were single-agent anti-PD-1, 6 were combinations of anti-PD-1 and anti-CTLA-4, and 1 was a bispecific anti-PD-1/anti-CTLA-4 antibody. At the pre-irAE time point the median number of ICI cycles per patient was 3.5 (range 1–5) for patients with severe irAEs and 2 (range 1–3) for patients without irAEs; at irAE-max, the median number of ICI cycles per patients was 4 (range 1–13) for patients with severe irAEs and 6 (range 3–8) for patients without severe irAEs.

    Figure 1
    Figure 1

    Sankey diagram of the patient cohort for this research. Of the n=28 patients total, 1 patient had two lines of therapy for total n=29 in treatment, ICI line, severe irAE and clinical benefit data. One patient on neoadjuvant training was excluded from clinical benefit analysis. CTLA-4, cytotoxic T-lymphocyte-associated protein 4; ICI, immune checkpoint inhibitor; irAE, immune-related adverse event PD-1, programmed cell death protein 1; WT, wildtype.

    Toxicities

    In this study, irAEs were classified as ‘severe irAEs’ if they were grade ≥2 and required immunosuppression and/or permanent medical intervention. Grade 2 irAEs not requiring immune suppression or permanent treatment and grade 0–1 irAEs were classified as non-severe irAEs. This classification was designed to identify immune features associated with the occurrence of irAEs that are clinically significant and/or can interfere with ICI due to impacting the patient’s lifestyle or performance status. All toxicities are summarized in (online supplemental table S3). Among the 29 lines of therapy, 18 (62.1%) were associated with 1 or more severe irAEs (figure 1, online supplemental table S2-3). Among patients who presented with severe irAEs, 7 had more than 1 severe irAE, while 11 had only 1.

    Unsupervised clusters of CD4+ naïve T cells and CD16+ NK cells are associated with severe irAEs at baseline and irAE-max

    Next, we assessed phenotypes and functional states of the banked PBMCs using a 43-marker mass cytometry panel designed to identify immune subsets, as well as their states of activation, suppression, exhaustion, senescence, migration, or proliferation. To identify immune features associated with severe irAEs in an unbiased manner, we first applied unsupervised clustering using the FlowSOM algorithm. We identified 75 unique clusters, which accounted for 99.84% of total cells (figure 2A). We compared cluster frequencies in patients with severe irAEs and patients without severe irAEs at baseline, pre-irAE and irAE-max (matched time controls for non-severe irAEs) to identify significant differences at each time point using SAM (figure 2B). We found that CD4+ naïve T-cell cluster C6, and CD16+ NK cell clusters C50, C71 and C72, were significantly different between irAE groups across all time points. Compared with patients with no severe irAE, patients with severe irAEs had a twofold higher frequency of CD4+ naïve T-cell cluster C6 at baseline and at pre-irAE, and up to fivefold more at irAE-max (figure 2C). Cluster C6 is marked by CCR7hiCD27hiCD45RAhiCD28medCD127medCD56medCD38med, suggesting a low activation status (online supplemental figure S2B-C). Conversely, patients that developed severe irAEs had threefold lower frequencies of CD16+ NK cell clusters C50, C71 and C72 at baseline and about twofold to fivefold lower frequencies at irAE-max compared with patients without severe irAEs (figure 2C). C50 represents cytotoxic NK cells expressing CD16hiCD56loCD161higranzyme B (GRB)medCD38med TbetloCD57neg. C71 and C72 are two more mature, terminally differentiated NK cell clusters that are marked by CD16hiCD56loCD57hiCD161medGRBmedCD38med; C71 is CD11cloTbetlo and C72 is CD11chiTbetmed (online supplemental figure S3B-C).

    Figure 2
    Figure 2

    PMBC clusters found with unsupervised analysis to be associated with severe irAE. (A) Uniform Manifold Approximation and Projection (UMAP) visualizing the FlowSOM-guided clustering of CD45+ non-granulocyte cells from PBMC samples. Clusters were created based on all PBMC samples across time points and lines of therapy. (B) Scaffold map of clusters at baseline, pre-irAE and irAE-max. Coloring is based on statistical differences in frequencies between patients with severe irAE and patients without severe irAE. Node size reflects the number of cells in a cluster; distance between clusters reflects relative similarity between clusters. Note that landmark nodes are not clusters, but indicate the relative phenotypic similarity of connected clusters. Highlighted clusters include: CD4+ naïve T-cell cluster 6; CD4+ effector memory T cell (TEM) cluster 30; CD4+ central memory T cell (TCM) cluster 31; memory B-cell cluster 38; CD8+ TCM clusters 43 and 44; CD8+ TEM cluster 56; CD16+ NK cell clusters 50, 71 and 72; CD56high NK cell cluster 59. (C) Frequencies of CD4+ naïve T-cell cluster 6 and CD16+ NK cell clusters 50, 71 and 72 as percentage of non-granulocytes in patients with or without severe irAE at different time points. **=false discovery rate (fdr)<0.01 as determined by Significance Analysis of Microarray (SAM). n.s.=not significant. cDC, classical dendritic cells; cMono, classical monocytes; DNT, double-negative (CD4− CD8−) T cells; DPT, double-positive (CD4+ CD8+) T cells; inMono, intermediate monocytes; irAE, immune-related adverse event; ncMono, non-classical monocytes; NK cells, natural killer cells; NKT, natural killer T cells; PBMC, peripheral blood mononuclear cell; pDC, plasmacytoid dendritic cells; TEMRA, CD45RA+effector memory T cells; transDC, transitional dendritic cells; Tregs, regulatory T cells. N of lines of therapy per comparison: n=10 no severe irAE, n=16 severe irAE at baseline; n=9 no severe irAE, n=6 severe irAE at pre-irAE; n=9 no severe irAE, n=20 severe irAE at irAE-max.

    Additionally, there were statistically significant differences in cluster frequency at pre-irAE and irAE-max in patients with severe irAEs, including threefold higher frequency of B memory cluster C38 (CCR7hiCD74medHLA-DRloCD38lo) at pre-irAE (online supplemental figure S4S-D), and 1.5-fold lower frequencies of CD4+ T effector memory (TEM) cluster C30 and CD4+ T central memory (TCM) cluster C31 at irAE-max (online supplemental figure S2D-E). C30 is marked as CD127hiCD28medCD16medCD69lo and C31 is marked as CD161hiCD127hiCD28medCCR7medCD16medCD69loCD38neg (online supplemental figure S2B-C), suggesting non-activated phenotypes. Patients with severe irAEs at irAE-max also had twofold lower frequencies of CD8+ TCM clusters C43 (activated phenotype with CD127hiCD161hiKLRG-1hiCD69hiCD28medTbetloCD56lo) and C44 (non-activated phenotype with CD127hiCD28medTbetloKLRG-1loCD74lo), threefold fewer CD8+ TEM cluster C56 (activated phenotype CD25hiCD127hiCD28med) (online supplemental figure S5), and a twofold lower frequency of activated but non-differentiated CD56hi NK cluster C59 (CD127hiCD25hiCD11chiCD38hiTbetmedCD27med CD16neg) (online supplemental figure S3). Taken together, unsupervised analysis showed that patients with severe irAEs had higher frequencies of CD4+ naïve T cells and lower frequencies of cytotoxic NK cells at baseline, higher frequencies of memory B cells at pre-irAE, and lower frequencies of non-activated CD4+ and CD8+ TCM, non-activated CD4+ TEM, and activated CD8+ TCM and TEM clusters at peak of irAEs.

    Patients with severe irAEs have lower frequencies of immunosuppressive cell populations at baseline

    To confirm the findings from unsupervised clustering and potentially find other significant cell populations not observed with clustering, we manually gated on major adaptive and innate populations and assessed expression of functional state markers in each population. In general, we found that the gated analysis corroborated the findings observed in the unsupervised analysis. Patients with severe irAEs had a up to fourfold higher frequency of CD4+ naïve T cells at all three time points compared with patients without severe irAEs (figure 3A and online supplemental figure S6A). Similarly, CD16+ NK cells frequency was significantly lower at baseline and at the irAE-max at about a twofold difference, and slightly but not significantly lower at the pre-irAE time point in patients with severe irAEs (figure 3A and online supplemental figure S6A). Finally, patients with severe irAEs had more than twofold higher frequencies of memory B cells at pre-irAE than patients without severe irAEs (figure 3B).

    Figure 3
    Figure 3

    Manually gated PBMC populations associated with severe irAE. (A) Frequencies of CD4+ naïve T cells and CD16+ NK cells as percentage of non-granulocytes in patients with or without severe irAE at different time points. (B) Frequencies of memory B cells as percentage of non-granulocytes in patients with or without severe irAE at pre-irAE. (C) Frequencies of TIGIT+ Tregs as percentage of Tregs in patients with or without severe irAE at different time points. (D) Frequencies of CD38+ CD4+ TCM as percentage of CD4+ TCM and frequencies of CD39+ CD8+ TCM and HLA-DR+ CD8+ TCM as percentage of CD8+ TCM in patients with or without severe irAE at irAE-max. For A–D **=fdr<0.01 as determined by SAM. n.s.=not significant. (E–F) Visualization of paired comparisons in cell population frequencies between irAE-max and baseline in (E) patients with severe irAE and (F) patients without severe irAE. Each dot represents a population significantly different (fdr<0.01) at irAE-max relative to baseline. Statistical assessment was done through SAM. The size of each dot represents median population size at baseline as percentage of parent population. The color of each dot represents change in frequency from baseline (CFBL). N of lines of therapy per comparison: n=10 no severe irAE, n=16 severe irAE at baseline; n=9 no severe irAE, n=6 severe irAE at pre-irAE; n=9 no severe irAE, n=20 severe irAE at irAE-max. cMono, classical monocytes; DNT, double-negative T cells; DPT, double-positive T cells; fdr, false discovery rate; inMono, intermediate monocytes; irAE, immune-related adverse event; ncMono, non-classical monocytes; NK, natural killer; NKT, natural killer T cells; SAM, significance analysis of microarrays; TCM, T central memory; TEM, T effector memory; TEMRA, effector memory T cells; Treg, regulatory T cells.

    Gated analysis also found differential immune features not identified by unsupervised clustering. At baseline, patients with severe irAEs had about twofold lower frequencies of TIGIT+ regulatory T cells (Treg) (figure 3C) and CD161+ CD4+ T cells (online supplemental figure S6B), both of which have been described as having suppressive phenotypes.27 28 Interestingly, similar trends were observed at pre-irAE and irAE-max time points (figure 3C and online supplemental figure S6B). Of note, while total frequencies of Treg were not significantly different at any time point, patients with severe irAEs had slightly higher CD4+ TCM at baseline than patients without severe irAEs (online supplemental figure S6A). In addition, at irAE-max, patients with severe irAEs had twofold higher frequency of CD38+ CD4+ TCM cells, fourfold more CD39+ CD8+ TCM and fivefold more HLA-DR+ CD8+ TCM (figure 3D). Total frequencies of CD8+ TCM were not significantly different at any time point (online supplemental figure S6A).

    To assess which immune features were directly associated with severe irAE presentation, we performed a paired analysis of baseline and irAE-max samples between patients with and without severe irAEs (online supplemental figure S6C). Overall, we observed that patients with severe irAEs had statistically significant increases in expression of several activation markers across T-cell populations (figure 3E). For example, almost all T-cell subsets had about ~5% more ICOS+ populations, and 5–10% more CD39+, HLA-DR+ and CD38+ populations. In contrast, patients without severe irAEs did not have major changes in immune features over time, except for several T-cell populations having approximately 5% fewer PD-1+ cells (figure 3F).

    Taken together, these data suggest that patients with severe irAEs have fewer phenotypically regulatory immune populations pretreatment and higher frequencies of activated T-cell populations during peak of irAEs.

    Distinct irAEs might drive the immune features associated with severe irAEs

    To investigate whether the observed immune changes were driven by specific severe irAE categories, we analyzed the contribution of each irAE category to the differential immune features observed in the gated analysis. It is important to note that, due to the small number of patients in individual irAE categories, we were not able to analyze the pre-irAE time point. We only performed statistical analysis for irAE categories with three or more patients.

    At baseline, the higher frequency of CD4+ naïve T cells (figure 4A) and CD161+ CD4+ T cells (online supplemental figure S7) were mostly observed in patients who developed severe gastrointestinal, cutaneous or type 1 diabetes irAEs. Interestingly, at irAE-max, the observed frequency difference of these two cell populations is seen across all irAE categories (figure 4A and online supplemental figure S7). The lower frequency of TIGIT+ Treg and CD16+ NK at baseline or irAE-max was seen across all individual irAE categories, except patients with pituitary irAEs (n=2) did not show differential frequencies of either cell population at baseline (figure 4B–C). At irAE-max, the higher frequencies of CD38+ CD4+ TCM, CD39+ CD8+ TCM and HLA-DR+ CD8+ TCM associated with severe irAEs were driven by gastrointestinal, hepatobiliary, musculoskeletal, pituitary, and cutaneous irAEs, but not type 1 diabetes or pneumonitis (figure 4D) which could be due to small number of patients in these two categories. Taken together, most immune features associated with severe irAEs at baseline seem to be observed across different categories of irAEs (from a few to all). At irAE-max, the differential immune features are associated with all subtypes of irAEs except type 1 diabetes or pneumonitis.

    Figure 4
    Figure 4

    Contribution of individual irAE categories to the immune features associated with severe irAE observed in analysis of gated populations. (A) Frequencies of CD4+ naïve T cells as percentage of non-granulocytes in patients with or without severe irAE at baseline and irAE-max. (B) Frequencies of CD16+ NK cells as percentage of non-granulocytes in patients with or without severe irAE at baseline and irAE-max. (C) Frequencies of TIGIT+ Tregs as percentage of Tregs in patients with or without severe irAE at baseline. (D) Frequencies of CD38+ CD4+ TCM as percentage of CD4+ TCM and frequencies of CD39+ CD8+ TCM and HLA-DR+ CD8+ TCM as percentage of CD8+ TCM in patients with or without severe irAE at irAE-max. For A–D statistical comparisons were done between individual irAE categories with n≥3 patients and patients without severe irAE. **=fdr<0.01 as determined by SAM. n.s.=not significant. N of lines of therapy per comparison: n=10 no severe irAE, n=2 gastrointestinal irAE, n=8 hepatobiliary irAE, n=4 musculoskeletal irAE, n=2 pituitary irAE, n=1 respiratory irAE, n=3 cutaneous irAE, n=1 T1DM at baseline; n=9 no severe irAE, n=3 gastrointestinal irAE, n=11 hepatobiliary irAE, n=5 musculoskeletal irAE, n=3 pituitary irAE, n=1 respiratory irAE, n=6 cutaneous irAE, n=2 T1DM at irAE-max. fdr, false discovery rate; irAE, immune-related adverse event; NK, natural killer; TCM, T central memory; Treg, regulatory T cells; T1DM, type 1 diabetes mellitus.

    Immune features associated with severe irAEs and BRAF mutation status

    Next, we wanted to see if patient molecular factors like BRAF mutation (BRAF-Mut) status impact the immune features associated with severe irAEs. We divided patients into four different subgroups: patients with a BRAF-Mut with severe irAEs (n=8), BRAF-Mut without severe irAEs (n=5), BRAF-WT with severe irAEs (n=7) and BRAF-WT without severe irAEs (n=5). The incidence of severe irAEs was similar among patients with BRAF-Mut and BRAF-WT (61% vs 58%, respectively; χ2 p value=0.870). Patients without known BRAF-Mut status (n=4) were excluded from this analysis. The majority of BRAF-Mut were V600E or V600K (n=10), but one patient had L597S and one had R506_K507insVLR mutations.

    The differences in the frequency of CD4+ naïve T cells, CD16+ NK and TIGIT+ Treg cells associated with severe irAEs at baseline were statistically significant in BRAF-WT patients (figure 5A–C). While not statistically different, they followed a similar general trend in BRAF-Mut patients (figure 5A–C). Similarly, differences in CD38+ CD4+ TCM, CD39+ CD8+ TCM and HLA-DR+ CD8+ TCM associated with severe irAEs at irAE-max were significant in BRAF-WT patients, and not significant but with similar trend in BRAF-Mut patients (figure 5D). Conversely, the difference in CD161+ CD4+ T cells associated with severe irAEs at irAE-max was significant in BRAF-Mut patients, and not significant but with similar trend in BRAF-WT patients (online supplemental figure S8). These results suggest that BRAF mutation status has little impact on immune features associated with severe irAEs.

    Figure 5
    Figure 5

    Contribution of BRAF mutation status to the immune features associated with severe irAE observed in analysis of gated populations. (A) Frequencies of CD4+ naïve T cells as percentage of non-granulocytes in patients with or without severe irAE and with or without a BRAF mutation (BRAF-Mut or BRAF-WT) at baseline and irAE-max. (B) Frequencies of CD16+ NK cells as percentage of non-granulocytes in patients with or without severe irAE and with or without BRAF-M at baseline and irAE-max. (C) Frequencies of TIGIT+ Tregs as percentage of Tregs in patients with or without severe irAE and with or without BRAF-M at baseline. (D) Frequencies of CD38+ CD4+ TCM as percentage of CD4+ TCM and frequencies of CD39+ CD8+ TCM and HLA-DR+ CD8+ TCM as percentage of CD8+ TCM in patients with or without severe irAE and with or without BRAF mutation at irAE-max. For A–D *=p value<0.05 and ** =p value<0.01 as determined by Kruskal-Wallis test with post hoc Wilcoxon rank-sum pairwise analysis. n.s.=not significant. N of lines of therapy per comparison: n=5 no severe irAE BRAF-WT, n=6 severe irAE BRAF-WT, n=4 no severe irAE BRAF-Mut, n=7 severe irAE BRAF-Mut at baseline; n=5 no severe irAE BRAF-WT, n=9 severe irAE BRAF-WT, n=3 no severe irAE BRAF-Mut, n=8 severe irAE BRAF-Mut at irAE-max. BRAF-WT, BRAF wildtype; irAE, immune-related adverse event; NK, natural killer; TCM, T central memory; Treg, regulatory T cells.

    Some immune features associated with severe irAEs are also associated with clinical benefit

    As patients who develop severe irAEs tend to have better clinical outcomes to ICI treatment, we wanted to assess whether any immune features associated with severe irAEs were also associated with clinical benefit. Given this clinical cohort contained patients with different stages of melanoma (figure 1, online supplemental table S2), we grouped them into two categories for exploratory analysis: patients who had clinical benefit to ICIs, including a partial response or a complete response to palliative ICIs or no relapsed disease with adjuvant ICIs (n=17) (median follow-up 1107 days, 380–1,285 days), and patients who had no clinical benefit, including disease progression in both palliative and adjuvant ICI settings (n=8) (median follow-up 502 days, 164–1,256 days). Three patients with stable disease and one patient who received neoadjuvant therapy were excluded from clinical benefit analysis. In this cohort, 13/15 (87%) of patients with severe irAEs showed clinical benefit, compared with 4/10 (40%) of those without severe irAEs (figure 6A, χ2 p value=0.014). These results corroborated what others have observed.8 29

    Figure 6
    Figure 6

    Manually gated PBMC populations associated with clinical benefit. (A) Percentage of patients with or without severe irAE showing clinical benefit to immunotherapy. (B) Frequencies of CD4+ naïve T cells as percentage of non-granulocytes in patients with or without clinical benefit at various time points. (C) Frequencies of CD16+ NK cells as percentage of non-granulocytes in patients with or without clinical benefit at various time points. (D) Frequencies of HLA-DR+ Tregs as percentage of Tregs in patients with or without clinical benefit at baseline. (E) Frequencies of Killer-Cell Lectin like receptor G1 (KLRG-1)+ double-negative (CD4–CD8–) T cells (DNT) as percentage of DNT in patients with or without clinical benefit at various time points. (F) Frequencies of CD28+ CD8+ naïve T cells as percentage of CD8+ naïve T cells in patients with or without clinical benefit at pre-irAE. (G) Fold change from baseline (FC from BL) in classical dendritic cells (cDC), plasmacytoid dendritic cells (pDC) and classical monocytes (cMono) frequency in patients with or without clinical benefit at pre-irAE. **=false discovery rate<0.01 as determined by SAM. n.s.=not significant. N of lines of therapy per comparison: n=8 no clinical benefit, n=15 clinical benefit at baseline; n=5 no clinical benefit, n=9 clinical benefit at pre-irAE; n=6 no clinical benefit, n=18 clinical benefit at irAE-max. irAE, immune-related adverse event; NK, natural killer; Treg, regulatory T cells.

    Through unsupervised and gated population analysis, we found a few immune features associated with clinical benefit that align with those associated with severe irAEs. At baseline and irAE-max, patients with clinical benefit had twofold more CD4+ naïve T cells than patients without clinical benefit (figure 6B). Similarly, patients with clinical benefit had twofold fewer CD16+NK cells (figure 6C) and up to fivefold lower frequencies of the unsupervised CD16+ NK clusters C50, C71 and C72 at irAE-max (online supplemental figure S9A) than patients without clinical benefit.

    However, clinical benefit was also associated with unique immune features that did not associate with severe irAEs. Patients with clinical benefit had 1.5-fold fewer HLA-DR+ Treg cells at baseline (figure 6D) and a twofold higher frequency of KLRG-1+ CD4– CD8– (double negative) T cells, which was maintained at all time points (figure 6E). At pre-irAE, patients with clinical benefit demonstrated twofold higher frequency of CD28+ naïve T cells (figure 6F) and a larger increase in frequencies of plasmacytoid dendritic cells (pDC), classical dendritic cells (cDC) and classical monocytes compared with patients without clinical benefit (figure 6G). At irAE-max, patients with clinical benefit had slightly more CD74+ pDC than patients without clinical benefit (online supplemental figure S9B).

    All results from this study are summarized in table 1.

    View this table:
    Table 1

    Result Summary

    Discussion

    In this study, we used mass cytometry to identify immune features associated with the development of severe irAEs in patients with melanoma treated with ICIs, using both unsupervised and gated population analyses. One of our main observations was that at all time points, patients with severe irAEs had higher frequencies of CD4+ naïve T cells and lower frequencies of CD16+ NK cells compared with patients without severe irAEs. Not much is known about the role of these cell populations in the development of irAEs. By their nature, naïve T cells do not exert many biological functions on their own until differentiation into helper T-cell effector phenotypes, like Th1, Th2 or Th17.30 Interestingly, several studies have linked CD4+ helper T cells to the development of irAEs. In a cohort of predominantly patients with lung adenocarcinoma treated with ICIs, Bukhari et al studied the peripheral blood with single cell RNA sequencing (RNA-seq) and found that patients with immune-related pneumonitis had more CD4+ Th2 cells at baseline, whereas patients with immune-related thyroiditis had more Th17 cells at baseline.31 Similarly, Kim et al used multicolor flow cytometry to study a cohort of patients with non-small cell lung cancer and thymic epithelial tumors and found that higher baseline frequencies of Th17 and Th1 cells were associated with development of grade 3+irAEs.32 Cytokines secreted by helper T-cell subsets play a critical role in regulating inflammation and can contribute to the development of autoimmunity. As our mass cytometry panel lacked markers necessary to distinguish individual helper T-cell subsets, such as RORγt for Th17 and GATA3 for Th2, we were not able to determine whether these CD4+ naïve T cells present in higher baseline frequencies in patients with severe irAEs went on to differentiate into specific helper subsets. It is also possible that the administration of ICIs disrupts the quiescent state of CD4+ T cells, which subsequently become autoreactive and contribute to irAE development through production of inflammatory cytokines.33

    We also found CD56loCD16+ NK cells were present in lower frequencies at all time points in patients with severe irAEs compared with patients without severe irAEs. These cells typically mediate natural and antibody-dependent cellular toxicity, exhibit high levels of perforin production and enhanced killing, and migrate to inflamed sites through expression of various chemokine receptors like CXCR3 and CXCR4. They also express high levels of killer cell immunoglobulin-like receptors, which mediate NK cell development, tolerance, and activation.34 Engagement of CD16, the low-affinity receptor for IgG1 and IgG3, triggers cytotoxic activity and production of pro-inflammatory cytokines and chemokines.35 NK cells also differentially express PD-1 and CTLA-4. Although the activation of immunoglobulin receptors can have direct downstream effects on NK cell activation, the mechanisms of how these cells are associated with the development of severe irAEs are unknown. Interestingly, when we just look at the patients treated with anti-PD-1 monotherapy in our cohort, patients with severe irAE have lower frequencies of CD16+NK cells than patients without severe irAE, both at baseline as well as irAE-max (p<0.01 and p<0.05, respectively, Wilcoxon rank-sum test, data not shown).

    We found that patients with severe irAEs had fewer suppressive TIGIT+ Treg cells and CD161+ CD4+ T cells at baseline.27 28 A recent study by Gonzalo-Nunez et al profiled PBMCs and serum proteins in 101 patients with melanoma or non-small cell lung cancer who were treated with ICIs. They reported that higher plasma levels of various cytokines and increased proliferation of CD4+ Treg and CD8+ T cells after initiation of treatment were associated with an increased risk to develop irAEs of any grade.36 While we did not see increased Treg proliferation after treatment initiation in our cohort, we did see that the differential frequencies of TIGIT+ Treg cells associated with severe irAEs was no longer present at pre-irAE and irAE-max, possibly reflecting an attempt of the immune system to increase this Treg population to counter the increasing levels of inflammation associated with development of irAEs. These findings are in line with other studies; Chaput et al show that ipilimumab-induced colitis is associated with lower baseline frequencies of Treg cells, and Grigoriou et al report that patients with melanoma with irAEs undergo extensive transcriptomic reprogramming of Treg cells towards an inflammatory signature during ICI treatment.37 38 At peak of irAEs, we did see more CD38+ CD4+ TCM cells and CD39+ CD8+ TCM and HLA-DR+ CD8+ TCM cells, which represent activated CD4+ and CD8+ T cells.39 40 Increased frequency of these cell populations has been observed by others in association with irAEs.20 Our findings are in line with a study by Lozano et al, who profiled three cohorts of 27, 26 and 18 patients with ICI-treated melanoma using mass cytometry, single cell RNA-seq and bulk T-cell receptor sequencing. They found that increased CD4+ memory T-cell abundance and T-cell receptor diversity at baseline were associated with irAEs occurrence.41 These findings are in line with the slightly higher frequencies of CD4+ TCM we observed in patients with severe irAEs at baseline. Taken together, the lower frequency of suppressive TIGIT+ Treg and CD161+ CD4+ T cells observed in our study before the initiation of ICI therapy may represent a reduced immunosuppressive capacity in these patients that allows for the expansion of reactive effector CD8+ and CD4+ T cells, which in turn may contribute to the occurrence of severe irAEs.

    Different organs can be preferentially affected by different ICIs. For example, irAEs observed in patients treated with ipilimumab, which targets CTLA-4, more frequently occur in the skin and gastrointestinal tract. On the other hand, lung, joint and certain endocrine toxicities are mostly associated with anti-PD-1 monoclonal antibodies.16 42 In addition, while most irAE eventually subside, others result in irreversible damage, like type 1 diabetes or myocarditis. Therefore, we investigated whether any observed immune features associated with severe irAE were driven by specific categories of irAE. In our study, we observed that the higher frequency of CD4+ naïve T cells at baseline was present across different types of irAEs. At the peak of irAEs, the higher frequency of activated CD4+ and CD8+ TCM cells are also associated with most of the individual irAEs. Our cohort was too small to attribute specific immune features to individual treatments within irAE categories. Our findings suggest that the dysregulated T-cell activation by ICIs may not be specific to any irAE category but may lead to overall off-target tissue damage.

    The development of irAEs in patients treated with ICIs has been correlated with improved clinical outcome.43 44 In this study, we observed a similar correlation: 87% of patients who developed severe irAEs derived clinical benefit from ICIs, whereas only 40% of patients without severe irAEs experienced clinical benefit. Patients with clinical benefit, similar to patients with severe irAEs, had an increased frequency of CD4+ naïve T cells at all time points and a lower frequency of CD16+ NK cells at irAE-max. In contrast, patients with clinical benefit had lower frequencies of HLA-DR+Treg, but not TIGIT+Treg, at baseline. However, like TIGIT+ Treg, HLA-DR+ Treg cells have been described as highly suppressive.45 It is possible that the lower frequency of highly suppressive Treg at baseline contributes to the development of a more effective antitumor response, but that the resulting activated circulating T cells cause off-target tissue damage as a result. Whereas many studies show an association with higher baseline or on treatment Treg frequencies and clinical benefit, the exact Treg phenotype that contributes to response is not clearly understood. Although clinical benefit to ICIs and presentation of severe irAEs both require activation of the immune system, additional different mechanisms likely apply. We saw changes in myeloid cell frequencies associated with clinical benefit after initiation of treatment (pre-irAE), which was not observed in association with severe irAEs. Immunotherapy has been reported to activate cDCs present in the lymph nodes, which in turn stimulate T-cell proliferation and migration of effector subsets to the tumor.46 It should also be noted that this cohort was comprised of patients who received ICIs as treatment for stage II, III or IV melanoma, which are distinct biological representations of melanoma and for which clinical response to ICI treatment is determined through different clinical measures. As such, the immune features presented as associated with clinical benefit should be interpreted as exploratory analysis.

    Overall, our study identified unique immune features differentially associated with ICI-induced severe adverse events in patients with melanoma, which provides insights to elucidate the mechanisms of irAEs and potential biomarker candidates to help clinical management of these patients. Additional studies with larger cohorts will need to be done to further expand on and independently validate these findings to assess their performance and use as clinical biomarkers. We chose mass cytometry as our primary research method due to its ability to determine the frequency of 30+cellular phenotypes and expression of 20+functional state markers. However, integrating mass cytometry data with other methods in a larger data set would increase the impact and breadth of our findings. We did not assess cytokine production, antigen specificity and clonal expansion or other cellular functions in this study. A combination of multiomic approaches, including analysis of serum proteins, T-cell clonal sequencing, and other methodologies could provide more insight on the underlying mechanisms involved in presentation of severe irAEs and clinical benefit. A larger cohort would also allow for more accurate prediction of multiparametric biomarkers, for example, through machine-learning algorithmic analysis. Finally, we assessed a limited number of clinical and molecular factors in this study, and the size of the cohort prohibited us from performing multivariate testing to assess the contribution of additional factors like individual treatments, disease stage, tumor mutational burden and tumor PD-L1 expression. Large observational studies are currently ongoing, like the iCHECKIT trial (NCT04871542), that are specifically designed to determine how certain clinical risk factors such as age, gender, other medical conditions, and the type of ICIs used affect whether a patient with a malignant solid tumor will develop mild or serious side effects to ICIs.

    This study does demonstrate the ability of mass cytometry to identify cellular immune features associated with clinical outcomes to ICI treatment in the blood of patients with melanoma. We plan to validate and expand on these findings in a larger patient cohort from the same institution.

    Supplemental material

    Data availability statement

    Data are available upon reasonable request.

    Ethics approval

    This study involves human participants and was approved by Huntsman Cancer Institute (HCI)’s Ethics board (IRB_00010924). Participants gave informed consent to participate in the study before taking part.

    References

    1. ↵
      1. Shiravand Y,
      2. Khodadadi F,
      3. Kashani SMA, et al
      . Immune Checkpoint inhibitors in cancer therapy. Current Oncology 2022;29:3044–60. doi:10.3390/curroncol29050247
      OpenUrl
    2. ↵
      1. Larkin J,
      2. Chiarion-Sileni V,
      3. Gonzalez R, et al
      . Combined Nivolumab and Ipilimumab or monotherapy in untreated Melanoma. N Engl J Med 2015;373:23–34. doi:10.1056/NEJMoa1504030
      OpenUrlCrossRefPubMed
    3. ↵
      1. Lyubchenko T,
      2. Leung DYM,
      3. Goleva E
      . Why are immune adverse events so common with Checkpoint inhibitor therapy. Ann Allergy Asthma Immunol 2021;126:608–10. doi:10.1016/j.anai.2021.03.016
      OpenUrl
    4. ↵
      1. Xu C,
      2. Chen Y-P,
      3. Du X-J, et al
      . Comparative safety of immune Checkpoint inhibitors in cancer: systematic review and network meta-analysis. BMJ 2018;363:k4226. doi:10.1136/bmj.k4226
    5. ↵
      1. Santomasso BD,
      2. Nastoupil LJ,
      3. Adkins S, et al
      . Management of immune-related adverse events in patients treated with Chimeric antigen receptor T-cell therapy: ASCO guideline. JCO 2021;39:3978–92. doi:10.1200/JCO.21.01992
      OpenUrl
    6. ↵
      1. Thompson JA,
      2. Schneider BJ,
      3. Brahmer J, et al
      . Management of Immunotherapy-related toxicities, version 1.2022, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw 2022;20:387–405.:jnccn2004gls. doi:10.6004/jnccn.2022.0020
      OpenUrlCrossRef
    7. ↵
      1. Schneider BJ,
      2. Naidoo J,
      3. Santomasso BD, et al
      . Management of immune-related adverse events in patients treated with immune Checkpoint inhibitor therapy: ASCO guideline update. J Clin Oncol 2021;39:4073–126. doi:10.1200/JCO.21.01440
      OpenUrlCrossRefPubMed
    8. ↵
      1. Freeman-Keller M,
      2. Kim Y,
      3. Cronin H, et al
      . Nivolumab in Resected and Unresectable metastatic Melanoma: characteristics of immune-related adverse events and association with outcomes. Clin Cancer Res 2016;22:886–94. doi:10.1158/1078-0432.CCR-15-1136
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Bai A,
      2. Moss A,
      3. Rothweiler S, et al
      . NADH oxidase-dependent Cd39 expression by Cd8(+) T cells modulates interferon gamma responses via generation of adenosine. Nat Commun 2015;6:8819. doi:10.1038/ncomms9819
    10. ↵
      1. Peng L,
      2. Wang Y,
      3. Liu F, et al
      . Peripheral blood markers predictive of outcome and immune-related adverse events in advanced non-small cell lung cancer treated with PD-1 inhibitors. Cancer Immunol Immunother 2020;69:1813–22. doi:10.1007/s00262-020-02585-w
      OpenUrlCrossRefPubMed
    11. ↵
      1. Pavan A,
      2. Calvetti L,
      3. Dal Maso A, et al
      . Peripheral blood markers identify risk of immune-related toxicity in advanced non-small cell lung cancer treated with immune-Checkpoint inhibitors. Oncologist 2019;24:1128–36. doi:10.1634/theoncologist.2018-0563
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Eun Y,
      2. Kim IY,
      3. Sun J-M, et al
      . Risk factors for immune-related adverse events associated with anti-PD-1 Pembrolizumab. Sci Rep 2019;9. doi:10.1038/s41598-019-50574-6
    13. ↵
      1. Diehl A,
      2. Yarchoan M,
      3. Hopkins A, et al
      . Relationships between lymphocyte counts and treatment-related toxicities and clinical responses in patients with solid tumors treated with PD-1 Checkpoint inhibitors. Oncotarget 2017;8:114268–80. doi:10.18632/oncotarget.23217
      OpenUrlPubMed
    14. ↵
      1. Nakamura Y,
      2. Tanaka R,
      3. Maruyama H, et al
      . Correlation between blood cell count and outcome of Melanoma patients treated with anti-PD-1 antibodies. Jpn J Clin Oncol 2019;49:431–7. doi:10.1093/jjco/hyy201
      OpenUrlCrossRef
    15. ↵
      1. Hommes JW,
      2. Verheijden RJ,
      3. Suijkerbuijk KPM, et al
      . Biomarkers of Checkpoint inhibitor induced immune-related adverse events-a comprehensive review. Front Oncol 2020;10:585311. doi:10.3389/fonc.2020.585311
    16. ↵
      1. Xu Y,
      2. Fu Y,
      3. Zhu B, et al
      . Predictive biomarkers of immune Checkpoint inhibitors-related toxicities. Front Immunol 2020;11:2023. doi:10.3389/fimmu.2020.02023
      OpenUrlCrossRef
    17. ↵
      1. Nakamura Y
      . Biomarkers for immune checkpoint inhibitor-mediated tumor response and adverse events. Front Med 2019;6:119. doi:10.3389/fmed.2019.00119
    18. ↵
      1. Subudhi SK,
      2. Aparicio A,
      3. Gao J, et al
      . Clonal expansion of Cd8 T cells in the systemic circulation precedes development of Ipilimumab-induced toxicities. Proc Natl Acad Sci U S A 2016;113:11919–24. doi:10.1073/pnas.1611421113
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Oh DY,
      2. Cham J,
      3. Zhang L, et al
      . Immune toxicities Elicted by CTLA-4 blockade in cancer patients are associated with early diversification of the T-cell repertoire. Cancer Res 2017;77:1322–30. doi:10.1158/0008-5472.CAN-16-2324
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Reschke R,
      2. Gussek P,
      3. Boldt A, et al
      . Distinct immune signatures indicative of treatment response and immune-related adverse events in melanoma patients under immune checkpoint inhibitor therapy. Int J Mol Sci 2021;22:15.:8017. doi:10.3390/ijms22158017
      OpenUrlCrossRef
    21. ↵
      1. Subrahmanyam PB,
      2. Dong Z,
      3. Gusenleitner D, et al
      . Distinct predictive biomarker candidates for response to anti-CTLA-4 and anti-PD-1 Immunotherapy in Melanoma patients. J Immunother Cancer 2018;6:18. doi:10.1186/s40425-018-0328-8
    22. ↵
      1. Hogan SA,
      2. Levesque MP,
      3. Cheng PF
      . Melanoma Immunotherapy: next-generation biomarkers. Front Oncol 2018;8:178. doi:10.3389/fonc.2018.00178
    23. ↵
      1. Hartmann FJ,
      2. Babdor J,
      3. Gherardini PF, et al
      . Comprehensive immune monitoring of clinical trials to advance human Immunotherapy. Cell Rep 2019;28:819–31. doi:10.1016/j.celrep.2019.06.049
      OpenUrlPubMed
    24. ↵
      1. Takahashi C,
      2. Au-Yeung A,
      3. Fuh F, et al
      . Mass Cytometry panel optimization through the designed distribution of signal interference. Cytometry A 2017;91:39–47. doi:10.1002/cyto.a.22977
      OpenUrl
    25. ↵
      1. Tusher VG,
      2. Tibshirani R,
      3. Chu G
      . Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 2001;98:5116–21. doi:10.1073/pnas.091062498
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Quintelier K,
      2. Couckuyt A,
      3. Emmaneel A, et al
      . Analyzing high-dimensional Cytometry data using Flowsom. Nat Protoc 2021;16:3775–801. doi:10.1038/s41596-021-00550-0
      OpenUrl
    27. ↵
      1. Duurland CL,
      2. Santegoets SJ,
      3. Abdulrahman Z, et al
      . Cd161 expression and regulation defines rapidly responding Effector Cd4+ T cells associated with improved survival in Hpv16-associated tumors. J Immunother Cancer 2022;10:e003995. doi:10.1136/jitc-2021-003995
    28. ↵
      1. Joller N,
      2. Lozano E,
      3. Burkett PR, et al
      . Treg cells expressing the Coinhibitory molecule TIGIT selectively inhibit proinflammatory Th1 and Th17 cell responses. Immunity 2014;40:569–81. doi:10.1016/j.immuni.2014.02.012
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Maher VE,
      2. Fernandes LL,
      3. Weinstock C, et al
      . Analysis of the association between adverse events and outcome in patients receiving a programmed death protein 1 or programmed death ligand 1 antibody. J Clin Oncol 2019;37:2730–7. doi:10.1200/JCO.19.00318
      OpenUrlCrossRefPubMed
    30. ↵
      1. Luckheeram RV,
      2. Zhou R,
      3. Verma AD, et al
      . Cd4(+)T cells: differentiation and functions. Clin Dev Immunol 2012;2012:925135. doi:10.1155/2012/925135
    31. ↵
      1. Bukhari S,
      2. Henick BS,
      3. Winchester RJ, et al
      . Single-cell RNA sequencing reveals distinct T cell populations in immune-related adverse events of Checkpoint inhibitors. Cell Rep Med 2023;4:100868. doi:10.1016/j.xcrm.2022.100868
    32. ↵
      1. Kim KH,
      2. Hur JY,
      3. Cho J, et al
      . Immune-related adverse events are clustered into distinct subtypes by T-cell profiling before and early after anti-PD-1 treatment. Oncoimmunology 2020;9:1722023. doi:10.1080/2162402X.2020.1722023
    33. ↵
      1. Grywalska E,
      2. Pasiarski M,
      3. Góźdź S, et al
      . Immune-Checkpoint inhibitors for combating T-cell dysfunction in cancer. Onco Targets Ther 2018;11:6505–24. doi:10.2147/OTT.S150817
      OpenUrlCrossRefPubMed
    34. ↵
      1. Stabile H,
      2. Fionda C,
      3. Gismondi A, et al
      . Role of distinct natural killer cell subsets in anticancer response. Front Immunol 2017;8:293. doi:10.3389/fimmu.2017.00293
    35. ↵
      1. Wang W,
      2. Erbe AK,
      3. Hank JA, et al
      . NK cell-mediated antibody-dependent cellular cytotoxicity in cancer Immunotherapy. Front Immunol 2015;6:368. doi:10.3389/fimmu.2015.00368
    36. ↵
      1. Nuñez NG,
      2. Berner F,
      3. Friebel E, et al
      . Immune signatures predict development of autoimmune toxicity in patients with cancer treated with immune Checkpoint inhibitors. Med 2023;4:113–29. doi:10.1016/j.medj.2022.12.007
      OpenUrl
    37. ↵
      1. Chaput N,
      2. Lepage P,
      3. Coutzac C, et al
      . Baseline gut Microbiota predicts clinical response and colitis in metastatic Melanoma patients treated with Ipilimumab. Ann Oncol 2017;28:1368–79. doi:10.1093/annonc/mdx108
      OpenUrlCrossRefPubMed
    38. ↵
      1. Grigoriou M,
      2. Banos A,
      3. Hatzioannou A, et al
      . Regulatory T-cell Transcriptomic Reprogramming characterizes adverse events by Checkpoint inhibitors in solid tumors. Cancer Immunol Res 2021;9:726–34. doi:10.1158/2326-6066.CIR-20-0969
      OpenUrlAbstract/FREE Full Text
    39. ↵
      1. Piedra-Quintero ZL,
      2. Wilson Z,
      3. Nava P, et al
      . Cd38: an immunomodulatory molecule in inflammation and Autoimmunity. Front Immunol 2020;11:597959. doi:10.3389/fimmu.2020.597959
    40. ↵
      1. Saraiva DP,
      2. Azeredo-Lopes S,
      3. Antunes A, et al
      . Expression of HLA-DR in cytotoxic T lymphocytes: a validated predictive biomarker and a potential therapeutic strategy in breast cancer. Cancers 2021;13:3841. doi:10.3390/cancers13153841
    41. ↵
      1. Lozano AX,
      2. Chaudhuri AA,
      3. Nene A, et al
      . T cell characteristics associated with toxicity to immune Checkpoint blockade in patients with Melanoma. Nat Med 2022;28:353–62. doi:10.1038/s41591-021-01623-z
      OpenUrlCrossRefPubMed
    42. ↵
      1. Young A,
      2. Quandt Z,
      3. Bluestone JA
      . The balancing act between cancer immunity and Autoimmunity in response to Immunotherapy. Cancer Immunol Res 2018;6:1445–52. doi:10.1158/2326-6066.CIR-18-0487
      OpenUrlAbstract/FREE Full Text
    43. ↵
      1. Zhong L,
      2. Wu Q,
      3. Chen F, et al
      . Immune-related adverse events: promising predictors for efficacy of immune Checkpoint inhibitors. Cancer Immunol Immunother 2021;70:2559–76. doi:10.1007/s00262-020-02803-5
      OpenUrlCrossRefPubMed
    44. ↵
      1. Das S,
      2. Johnson DB
      . Immune-related adverse events and anti-tumor efficacy of immune Checkpoint inhibitors. J Immunother Cancer 2019;7:306. doi:10.1186/s40425-019-0805-8
    45. ↵
      1. Baecher-Allan C,
      2. Wolf E,
      3. Hafler DA
      . MHC class II expression identifies functionally distinct human regulatory T cells. J Immunol 2006;176:4622–31. doi:10.4049/jimmunol.176.8.4622
      OpenUrlAbstract/FREE Full Text
    46. ↵
      1. Rahim MK,
      2. Okholm TLH,
      3. Jones KB, et al
      . Dynamic Cd8(+) T cell responses to cancer Immunotherapy in human regional lymph nodes are disrupted in metastatic lymph nodes. Cell 2023;186:1127–43. doi:10.1016/j.cell.2023.02.021
      OpenUrlCrossRef

    Supplementary materials

    Footnotes

    • MK-B and JMS contributed equally.

    • Contributors Study design: MK-B, JMS, RS, SH-L. Clinical team, recruitment, and consent of patients: US, AE-W, JPM, JH, SH-L, BG, ALE. Clinical database: MK-B, JM, ALE, SH-L. Generation of research data and sample processing: NS, L-CC, YSD. Data post-processing: NS, L-CC, SAE, NN. Data analysis and interpretation: MK-B, JMS, CPH, COM, MHS, NN, SH-L. Manuscript writing: MK-B, JMS, CPH, WDC, SH-L. Guarantor: SH-L.

    • Funding This project has been funded in whole or in part with Federal funds from the National Institute of Mental Health, National Institutes of Health, Department of Health and Human Services, under Grant/Contract No. 1R01A1154722-01. This publication is based on research supported by the Melanoma Research Alliance. Research reported in this publication used the Research Informatics Shared Resource and the Biorepository and Molecular Pathology (BMP) Shared Resource at Huntsman Cancer Institute at the University of Utah and was supported by the National Cancer Institute of the National Institutes of Health under Award Number P30CA042014. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

    • Competing interests SH-L has been scientific advisor/consultant for: Amgen, Ascendis, Astellas, BMS, Genmab, Merck, Nektar, Neon Therapeutics, Novartis, Regeneron, Vaccinex, Xencor; and has done contracted research through her affiliated institutions with Astellas, BioAtla, BMS, Boehringer Ingelheim, Checkmate, Dragonfly, F Star, Genentech, Kite Pharma, Merck, Neon Therapeutics, OncoC4, Pfizer, Plexxikon, Vaccinex, Vedanta, Xencor. US reports consultancy to Astellas, Exelixis, Seattle Genetics, Imvax, Sanofi, AstraZeneca and Gilead and research funding to institute from Janssen, Exelixis and Astellas/Seattle Genetics. JH conducts clinical trials from the following entities: BMS, Merck, Amgen, Philogen, Lyell, Lovance, NCI, Takara, Natera, Skyline. JM, AE-W, MK-B, ALE, JPM : None. JMS, NS, L-CC, WDC, COM and NN are currently employed by Teiko Bio and are shareholders. SAE and CPH was formerly employed by Teiko Bio and is a current shareholder. MHS and RS are founders, shareholders and board members of Teiko.bio. MHS has received a speaking honorarium from Standard BioTools. and Kumquat Bio, has been a paid consultant for Five Prime, Ono, January, Earli, Astellas, and Indaptus, and has received research funding from Roche/Genentech, Pfizer, Valitor, and Bristol Myers Squibb. WDC was employed by Fate Therapeutics within the last 24 months. NN was employed by Atreca within the last 24 months.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.