Distinct molecular and immune hallmarks of inflammatory arthritis induced by immune checkpoint inhibitors for cancer therapy

Immune checkpoint inhibitors are associated with immune-related adverse events (irAEs), including arthritis (arthritis-irAE). Management of arthritis-irAE is challenging because immunomodulatory therapy for arthritis should not impede antitumor immunity. Understanding of the mechanisms of arthritis-irAE is critical to overcome this challenge, but the pathophysiology remains unknown. Here, we comprehensively analyze peripheral blood and/or synovial fluid samples from 20 patients with arthritis-irAE, and unmask a prominent Th1-CD8+ T cell axis in both blood and inflamed joints. CX3CR1hi CD8+ T cells in blood and CXCR3hi CD8+ T cells in synovial fluid, the most clonally expanded T cells, significantly share TCR repertoires. The migration of blood CX3CR1hi CD8+ T cells into joints is possibly mediated by CXCL9/10/11/16 expressed by myeloid cells. Furthermore, arthritis after combined CTLA-4 and PD-1 inhibitor therapy preferentially has enhanced Th17 and transient Th1/Th17 cell signatures. Our data provide insights into the mechanisms, predictive biomarkers, and therapeutic targets for arthritis-irAE.

0 D A description of all covariates tested D 0 A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g.means) or other basic estimates (e.g.regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g.confidence intervals) □ l'xl For null hypothesis testing, the test statistic (e.g.F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted L.'.:.J Give P values as exact values whenever suitable.0 D For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings 0 D For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes 0 D Estimates of effect sizes (e.g.Cohen's d, Pearson's r), indicating how they were calculated Our web collection on statistics for biologists contains articles on many of the points above.-Raw single-cell RNA-seq data were processed using Cell Ranger (v3.1.0;http://www.support.10xgenomics.com/sing1e-ce11-gene-expression/software/downloads/latest), including demultiplexing the FASTQ reads, aligning them to the human reference genome (GRCh38, v3.0.0, from lOX Genomics), and counting the unique molecular identifier (UMI).
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers.We strongly encourage code deposition in a community repository (e.g.GitHub).See the Nature Portfolio guidelines for submitting code & software for further information.

Field-specific reporting
Please select the one below that is the best fit for your research.If you are not sure, read the appropriate sections before making your selection.

Life sciences
Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study design
All studies must disclose on these points even when the disclosure is negative.
Sample sizes were determined with the reference, which performed scRNAsequencing of blood and matching synovial fluid and shown to be of sufficient size of sample to discreminate major immune cell subsets (Penkava et al.PMID: 32958743).
We carried out a number of steps to filter out poor quality data.First, we removed cells with high mitochondrial gene expression because damaged and dead cells often exhibit extensive mitochondrial contamination.Specifically, we calculate the proportion of UMIs from mitochondrial genes (all genes with names start with "MT-" or "mt-") for each cell, then remove cells which contain more than 15%mitochondrial UMIs.Second, we removed cells for which less than 200 genes were detected.Third, doublets were identified using a multi-step approach: 1) library complexity: cells with high complexity libraries (in which detected transcripts are aligned to more than 6500 genes) were removed; 2) Cluster distribution: doublets or multiplets likely form distinct clusters with hybrid expression features and exhibit an aberrantly high gene count; 3) cluster marker gene expression: cells of a cluster express markers from distinct lineages (e.g., cells in the T-cell cluster showed expression of myeloid cell markers and vice versa); 4) doublet detection algorithm: scrublet, an algorithm to predict doublets in scRNA-seq data, was applied to further identify and clean doublets that could have been missed by steps 1-3.
Single cell RNA sequencing from synovial fluid and matching peripheral blood was performed three times for 10X sequencing.
Cell numbers as well as data at each time were comparable.To validate the data of single cell RNA sequencing, we performed flow cytometry.Flow cytometry (n=10), multiplex (n=3), HLA B27 typing (n=2), Treg suppression assay (n=4), and migration assay (n=2) were performed multiple times independantly as indicated by n.All replication attempts were successful.
Because this was a prospective observational study, randomization was not necessary.
Because this was a prospective observational study, blinding was not necessary.
Anti-Human CD16 BUV 395 BD Biosciences Cat#: 563785 RRID: AB_2744293, Dilution 1:20 Anti-Human CD27 BUV 395 BD Biosciences Cat#: 563815 RRID: AB_2744349, Dilution 1:50 Anti-Human CD4 BUV 395 BD Biosciences Cat#: 563550 RRID: AB_2738273, Dilution 1:50 We collected residual synovial fluid (SF) and/or peripheral blood (PB) from 20 patients who newly developed arthritis after immune checkpoint inhibitor (ICI) therapy.As serum-negative controls, we collected PB samples from patients who had not developed irAEs at least 12 weeks after initiating ICI therapy.For SF supernatantnegative controls, we collected SF samples from patients with osteoarthritis.The patients met the American College of Rheumatology diagnostic criteria for osteoarthritis.Participants included both male and female patients ranging in age from 34 to 77 years.Detailed information about individual participants can be found in Table 1 and Supplementary Data file 1.
The patients were recruited prospectively from Rheumatology services (either outpatient or inpatient settings) at the University of Texas MD Anderson Cancer Center.The diagnosis of inflammatory arthritis was determined by a history and physical exam performed by a treating rheumatologist at MD Anderson (S.T.K., M.S.-A, J.H.T., and H.L.).Prior to the procedures (diagnostic arthrocentesis and/or venipuncture), participants provided written informed consent, allowing collection of residual SF and/or PB samples as well as prospective follow-up for 12 months after the sample donation.Because mild arthritis-irAE is likely managed by either a patient or by an oncologist, recruitment in this study may be biased to more severe arthritis-irAE and the results might have revealed an altered immunity of severe inflammation.
The study was approved by the institutional review board at The University of Texas MD Anderson Cancer Center (IRB No: PA16-0935).
This it not a clinical trial.
After residual synovial fluid and/or peripheral blood samples were donated, the participants were prospectively followed up for 12 months.
Information on age, sex, body mass index, tumors, pattern of arthritis, history of irAEs prior to the arthritis, onset of the arthritis, CDAI, erythrocyte sedimentation rate, C-reactive protein, anti-nuclear antibody, rheumatoid factor, and anti-cyclic citrullinated peptide antibody were obtained from the medical record.We followed patients with arthritis-irAE for 12 months after the sample collection whether the patients failed steroid monotherapy and required steroid-sparing disease modifying anti-rheumatic drugs (DMARDs).