T-cell CX3CR1 expression as a dynamic blood-based biomarker of response to immune checkpoint inhibitors

Abstract

Immune checkpoint inhibitors (ICI) have revolutionized treatment for various cancers; however, durable response is limited to only a subset of patients. Discovery of blood-based biomarkers that reflect dynamic change of the tumor microenvironment, and predict response to ICI, will markedly improve current treatment regimens. Here, we investigate CX3C chemokine receptor 1 (CX3CR1), a marker of T-cell differentiation, as a predictive correlate of response to ICI therapy. Successful treatment of tumor-bearing mice with ICI increases the frequency and T-cell receptor clonality of the peripheral CX3CR1+CD8+ T-cell subset that includes an enriched repertoire of tumor-specific and tumor-infiltrating CD8+ T cells. Furthermore, an increase in the frequency of the CX3CR1+ subset in circulating CD8+ T cells early after initiation of anti-PD-1 therapy correlates with response and survival in patients with non-small cell lung cancer. Collectively, these data support T-cell CX3CR1 expression as a blood-based dynamic early on-treatment predictor of response to ICI therapy.

Conflict of interest statement

Igor Puzanov: Amgen consultant. Marc S. Ernstoff receives clinical trial support from Merck, Bristol Myers Squibb, Alkermes, EMD Serono and serves on a BMS DMSC and consultant for ImmuNext, Alkermes, EMD Serono, Merck, and BMS. Kunle Odunsi co-founder of Tactiva Therapeutics and receives research support from Astra Zeneca and Tessaro. Saby George reports consulting fees from and advisory roles for Pfizer, Exelixis, Bristol-Myers Squibb, Sanofi/Genzyme, Genentech, Bayer, Corvus, EMD Serono, Seattle Genetics/Astellas, Eisai, Aveo, and Merck, and institutional grant support from Bristol-Myers Squibb, Novartis, Bayer, Pfizer, Merck, Seattle Genetics/Astellas, Eisai, Calithera Biosciences, Immunomedics, Corvus Pharmaceuticals, Surface Oncology, and Agensys. The remaining authors have no competing interests.

Figures

Fig. 1. Effective immune checkpoint inhibitor therapy correlates with the increased frequency of circulating CX3CR1+ CD8+ T cells.

a Experimental scheme of treatment with immune checkpoint inhibitors (ICI). b Individual tumor growth and survival curves in MC38 and CT26 tumor-bearing mice treated with isotype antibody (Ab) (NT) or anti-PD-L1 Ab and anti-CTLA-4 Ab (ICI). c, d Representative flow cytometry plots and data panel showing the frequency of CX3CR1+ cells among CD8+ T cells (c) and tetramer (Tet)+ CD8+ T cells (d) in peripheral blood (PB) of MC38 and CT26 tumor-bearing mice in different treatments as indicated. Numbers denote percent CX3CR1+ cells. The gating strategy is presented in Supplementary Fig. 1. PB was harvested 2 weeks after initiation of the treatment. n = 5 mice in all groups (b–d). e, f Frequency of the PB CX3CR1+ subset among CD8+ T cells (e), tumor growth curves (mean) and survival curves (f) in CT26-bearing mice in different treatment groups as indicated. n = 9 mice (NT), 6 mice (anti-CTLA-4 Ab), 6 mice (anti-PD-L1 Ab), and 5 mice (combo). g Frequency of the PB CX3CR1+ subset among CD8+ T cells in MC38 tumor-bearing mice treated with isotype control Ab (NT) or anti-PD-L1 Ab. n = 7 mice (NT) and 6 mice (anti-PD-L1 Ab). h Tumor growth curves (mean) and survival curves in MC38-bearing mice in different treatment groups as indicated. n = 5 mice in all groups. Data shown in b–h are representative of two independent experiments. PB was harvested 1 week after initiation of the treatment (e, g). Arrows indicate initiation of treatment (b, f, h). P-values were determined by a log-rank (Mantel–Cox) test (b, f, h), a two-tailed Mann–Whitney U-test (c, d, g) or Kruskal–Wallis with Dunn’s multiple comparisons (e). Data in f, h are presented as mean ± SEM. Box plots: dot, single PB; hinges, 25th and 75th percentiles; middle line, median; whiskers, minimum to maximum value (c–e, g). Source data are provided as a Source Data file.

Fig. 2. Phenotypic analysis of peripheral CD8+ T cells in mice treated with immune checkpoint inhibitor (ICI) therapy.

a Gating strategy for phenotypic analysis of three subsets (CD27lo CX3CR1−, CD27hi CX3CR1−, and CX3CR1+) of peripheral CD8+ T cells in mice. b, c Mice bearing 10-day established CT26 tumors were treated with anti-PD-L1 antibody (Ab) and anti-CTLA-4 Ab every 3 days and every other day, respectively. Peripheral blood (PB) (b) and spleen (c) were harvested 2 weeks after initiation of the treatment. Representative flow-cytometric plots of three subsets (CD27lo CX3CR1−, CD27hi CX3CR1−, and CX3CR1+) of PB (b) and splenic (c) CD8+ T cells are shown. Data panels show frequency among CD8+ T cells. NS, not significant, *P < 0.05, **P < 0.005, ***P < 0.0001, by one-way repeated measures ANOVA with Tukey’s multiple comparisons (b, c). n = 9 mice in all groups (b, c). Data shown b, c are representative of two independent experiments. Source data are provided as a Source Data file.

Fig. 3. Phenotypic analysis of PB CX3CR1+ CD8+ T cells before and during ICI therapy.

a–c CT26 (a–c) or MC38 (c) tumor-bearing mice were treated with anti-PD-L1 Ab and anti-CTLA-4 antibody (Ab) every 3 days and every other day, respectively. Peripheral blood (PB) was obtained before and 1, 2, and 3 weeks after the initiation of the treatment. Gating strategy: all cells > size lymphocytes > singlets > live > CD3+ CD8+ > CD27, CX3CR1. a Ki-67 expression of CD27lo CX3CR1− (green), CD27hi CX3CR1− (blue), and CX3CR1+ (red) CD8+ T cells in PB 2 weeks after ICI therapy. Numbers denote percent Ki-67+ cells. The data panel shows the median fluorescence intensity (MFI) of Ki-67+ cells in each subset. n = 4 mice in all groups. b Box and whiskers plots showing MFI of Ki-67+ (blue) and frequency (red) of the CX3CR1+ subset in CD8+ T cells (upper) and Tet+ CD8+ T cells (lower) at days 0, 7, 14, and 21 in PB. n = 3 mice (day 0) and n = 4 mice (days 7, 14, and 21). a, b Data shown are representative of two independent experiments. c Frequency of PB Tet+ CD8+ T cells in the CD27lo CX3CR1− (green), CD27hi CX3CR1− (blue), and CX3CR1+ (red) subsets before and 1, 2, and 3 weeks after the initiation of the treatment in CT26 (upper) and MC38 (lower) tumor-bearing mice. For CT26 tumor-bearing mice, n = 31 (day 0), n = 18 (day 7), n = 9 (day 14), and n = 9 (day 21) derived from three independent experiments. For MC38 tumor-bearing mice, n = 14 (day 0), n = 29 (day 7), n = 19 (day 14), and n = 8 (day 21) derived from three independent experiments. P-values were determined by one-way repeated measures ANOVA with Tukey’s multiple comparisons (a, c). Box plots: dot, single PB; hinges, 25th and 75th percentiles; middle line, median; whiskers, minimum to maximum value (a–c). Source data are provided as a Source Data file.

Fig. 4. Effective ICI therapy induces a high degree of TCR sequence similarity and clonality between tumor-infiltrating CD8+ T cells and peripheral CX3CR1+ CD8+ T cells.

a–d MC38 tumor-bearing mice were treated with anti-CTLA-4 antibody (Ab) and anti-PD-L1 Ab. Three subsets of splenic CD8+ T cells determined by CD27 and CX3CR1 expression (CD27lo CX3CR1−, CD27hi CX3CR1−, and CX3CR1+), and CD8+ tumor-infiltrating lymphocytes (TILs) were isolated 2 weeks after the initiation of the treatment for TCR repertoire and clonality analysis. a TCR repertoire overlap by Morisita’s index (left) and representative pairwise scatter plots of the frequency of TCRβ CDR3 amino acid (AA) sequences between each subset of splenic CD8+ T cells and CD8+ TILs (right). n = 3 independent experiments. b TCR clonality analysis of three subsets of splenic CD8+ T cells and CD8+ TILs by top sequence plot (left), Gini index (center), and Lorenz curve (right). The most abundant 100 AA sequences are colored while other less frequent clones are in purple in the top sequence plot. n = 3 independent experiments for Gini index. The top sequence plot and Lorenz curve are representative of three independent experiments. c Representative overlapped weighted TCR repertoire dendrograms by ImmunoMap analysis between three subsets of splenic CD8+ T cells (blue) and CD8+ TILs (red). The distance of the branch ends represents sequence distance, and the size of circles denotes the frequency of sequence. The data shown are representative of three independent experiments. d Number of dominant motifs within top 100 productive sequences shared between three subsets of splenic CD8+ T cells (blue) and CD8+ TILs (red) in (c). The data table shows the number of dominant motifs shared between three subsets of splenic CD8+ T cells and CD8+ TILs from three independent experiments. One-way repeated-measures ANOVA with Tukey’s multiple comparisons (a, b). Values are mean ± SEM. Source data are provided as a Source Data file.

Fig. 5. Expansion of the CX3CR1+ subset in PB CD8+ T cells correlates with response to anti-PD-1 therapy and better survival in patients with NSCLC.

a Gating strategy for identifying CX3CR1+ CD8+ T cells in peripheral mononuclear blood cells. Cells were first gated for lymphocytes (SSC-A vs. FSC-A) and for singlets (FSC-H vs. FSC-A). b Overall survival (OS) of patients with high (n = 18) and low (n = 18) pretreatment frequency of the CX3CR1+ subset in PB CD8+ T cells. Cut-points by median baseline frequency of the CX3CR1+ subset in PB CD8+ T cells. c The largest % change of the CX3CR1+ subset in PB CD8+ T cells from baseline by the given time point in responders (CR/PR: n = 13) and non-responders (SD/PD: n = 23) of 36 NSCLC patients treated with anti-PD-1 therapy. CR/PR: complete and partial response, SD/PD: stable and progressive disease. P-values were calculated by a two-tailed Mann–Whitney U-test. Values are median ± SEM. d Percent change of the CX3CR1+ subset in PB CD8+ T cells from baseline (CX3CR1 score) in responders and non-responders. e Objective response rate (ORR) for high and low PD-L1 tumor proportion score (TPS) and PB CX3CR1 score at 3, 6, 9, and 12 weeks. ORR was analyzed by Fisher’s exact test. f Progression-free survival (PFS) and OS for high vs. low CX3CR1 score. P-values were calculated by a log-rank (Mantel–Cox) test (b, f). NS, not significant. Source data are provided as a Source Data file.