Increased Frequency of ICOS⁺ CD4 T Cells as a Pharmacodynamic Biomarker for Anti-CTLA-4 Therapy

Cytotoxic lymphocyte antigen-4 (CTLA-4) is an inhibitory coreceptor expressed on T cells and acts to attenuate T-cell immune responses (1–3). Blockade of CTLA-4 with a monoclonal antibody (mAb) has been shown to enhance T-cell responses and induce tumor rejection in a number of animal models (4, 5). Ipilimumab (Bristol-Myers Squibb), a mAb to human CTLA-4, has been found to elicit objective responses in clinical trials (6–10) and was shown to improve median overall survival of patients with metastatic melanoma in two different phase III clinical trials (11, 12).

In a presurgical clinical trial in which anti-CTLA-4 (ipilimumab) was administered to patients with localized bladder cancer, we previously reported a significant increase in the frequency of T cells expressing the inducible costimulator (ICOS) molecule in both tumor tissues and blood samples obtained from patients. (13, 14) We subsequently reported that a sustained increase in the frequency of ICOS⁺ CD4 T cells was correlated with clinical benefit in a small cohort of patients with melanoma who received anti-CTLA-4 therapy (15). We also found that loss of the ICOS/ICOS-ligand pathway in a murine model led to impaired antitumor immune responses and tumor rejection (16). Another study, with a different anti-CTLA-4 antibody (tremelimumab; Pfizer), also reported an increase in ICOS⁺ CD4 T cells in blood samples obtained from patients with breast cancer (17). Collectively, these data indicated a biologically relevant role for ICOS in antitumor responses elicited by anti-CTLA-4 therapy.

ICOS is a T-cell–specific molecule that belongs to the CD28/CTLA-4 family (18, 19) and is expressed only after T-cell activation. As anti-CTLA-4 enhances T-cell activation, we hypothesized that ICOS expression on CD4 T cells could be used as a pharmacodynamic biomarker for anti-CTLA-4 therapy. To determine whether ICOS expression on CD4 T cells measured by flow cytometry could be used as a pharmacodynamic biomarker for anti-CTLA-4 therapy (ipilimumab), we analyzed blood samples from 56 individuals, consisting of 10 healthy donors, 36 patients who received ipilimumab (12 patients with bladder cancer and 24 patients with melanoma), and 10 patients with melanoma who received treatment with a different immunomodulatory agent (gp100 DNA vaccine).

Blood samples were collected after normal healthy donors provided informed consent on The University of Texas MD Anderson Cancer Center (Houston, TX) Institutional Review Board (IRB)-approved laboratory protocol 2005-0027. Blood samples from patients with cancer were obtained after they provided consent on IRB-approved protocols such that patients with bladder cancer were consented as per protocols 2006-0080 and 2005-0027, as previously published (13, 15), and patients with melanoma were either consented on the Memorial Sloan-Kettering Cancer Center (New York, NY) IRB-approved biospecimen collection protocols 95-090 or 00-144 before treatment with ipilimumab (NCT00495066), as previously published (15), or protocol 06-113 for treatment with gp100 DNA vaccine (NCT00398073), as previously published (20).

All blood samples were collected from the clinic and transported at room temperature within 3 hours after being drawn from each individual. Whole blood was collected in Vacutainer tubes containing sodium heparin (BD Vacutainer). Peripheral blood mononuclear cells (PBMC) were isolated from whole blood by density gradient centrifugation using Lymphocyte Separation Medium (Mediatech) and Leucosep tubes (Greiner Bio-one). In brief, whole blood was centrifuged at 410 × g for 10 minutes and the plasma layer was collected, centrifuged for 10 minutes at 885 × g, and the supernatant frozen at −80°C for subsequent experiments. The whole blood was diluted 1:1 with RPMI (Mediatech) and the PMBCs isolated per manufacturer's guidelines. Cells at the interface were harvested and washed once with RPMI, and red blood cells were lysed with ammonium chloride lysis buffer. Cells were washed, counted, and immediately used fresh for staining and flow cytometry analysis as previously described (13), or 10 million PBMCs were frozen per milliliter of freezing solution (10% DMSO 10% FBS-supplemented RPMI) per NUNC cryovial. PBMCs were frozen in a Nalgene Cryo 1°C freezing container per manufacturer's instructions and stored in liquid nitrogen thereafter.

PBMCs were resuspended at 1.5 × 10⁶ cells/mL in RPMI (Mediatech) supplemented with MEM nonessential amino acids, 2 mmol/L l-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, and 10% human AB serum (Life Technologies); 30 × 10⁶ PBMCs were added to appropriate T25 tissue culture flask (Corning). PMBCs were activated for 48 hours with immobilized anti-CD3 5 μg/mL clone UCHT1 and soluble anti-CD28 2 μg/mL clone CD28.2 (BD Biosciences). PBMCs were collected after detachment from the flask with 4 mmol/L NaEDTA and washed twice with PBS + 2% FBS, followed immediately by labeling for flow cytometry or freezing as above for approximately 24 hours before labeling for flow cytometry.

Antibodies used for flow cytometry analysis were CD4 PerCP Cy5.5 (BD Biosciences), ICOS PE-Cy7 (eBioscience), and FoxP3-PE. Lineage was excluded using a cocktail of FITC-conjugated antibodies, CD303 (Miltenyi Biotec), CD20, CD16, CD14, CD56, TCR γδ, and CD19 (BD Biosciences). Triplicate samples were analyzed using the FACSCanto II (Becton Dickinson). Data were analyzed using the BD FACSDiva software. Gates were set according to appropriate isotype controls. D.N. Tang and J. Sun conducted studies independently.

Summary statistics such as mean, median, SD, and 95% confidence interval (CI) were provided for the frequency of CD4⁺ICOS⁺ T cells. The histograms and bar plots were used to display ICOS expressions by different subject group. Wilcoxon signed-rank test for paired data was used to assess the difference on the frequency of ICOS⁺ CD4 T cells between prior and posttreatment. Wilcoxon rank-sum test was also used to assess the difference on the frequency of ICOS⁺ CD4 T cells at baseline between donor controls and patients with different tumor types. All P values presented are two-sided. P values less than 0.05 were considered to be statistically significant. Statistical analyses were carried out using SPLUS 7 (Insightful Corp).

In a subset of 3 to 6 patients with bladder cancer who received two doses of anti-CTLA-4 and for whom PBMCs were available, we evaluated multiple potential pharmacodynamic biomarkers, including expression of ICOS, HLA-DR, and CD25 on CD4 T-cells; frequency of total CD4 and CD8 T-cell populations; and absolute lymphocyte count. As shown in Table 1, an increased frequency of ICOS⁺ CD4 T cells was consistently found after patients received treatment with dose 1 and dose 2 of anti-CTLA-4. However, we did not observe a consistent change in the other 5 evaluated biomarkers after patients received treatment with two doses of anti-CTLA-4. These data suggested that ICOS expression on CD4 T cells was a reasonable biomarker to further pursue as a potential pharmacodynamic biomarker for anti-CTLA-4 therapy.

Because we found that anti-CTLA-4 (ipilimumab) led to an increased frequency of ICOS⁺ CD4 T cells, which was due to enhanced T-cell activation, we conducted studies to determine whether in vitro activation with anti-CD3 and anti-CD28 would also lead to an increased frequency of ICOS⁺ CD4 T cells. We conducted in vitro activation with anti-CD3 and anti-CD28 on human PBMCs obtained from normal healthy donors to determine whether our flow cytometric assay could reliably detect a change in ICOS expression on CD4 T cells. We found that in vitro activation of human PBMCs led to a detectable increase in ICOS⁺ CD4 T cells (Supplementary Fig. S1), which was similarly observed for 10 different donors (Supplementary Table S1).

To determine whether our flow cytometric assay was reproducible, we obtained blood samples from 3 healthy donors on 3 different days and evaluated frequency of ICOS⁺ CD4 T cells before and after in vitro activation for both fresh and frozen samples. Blood was obtained from donors on 3 separate days, and samples were evaluated by two independent operators before and after activation either immediately (fresh) or after being frozen for 24 hours (Fig. 1, representative donor). Blood samples from 2 additional healthy donors were also obtained on 3 separate days and underwent similar analyses (Supplementary Table S2). Each donor sample was tested in triplicate, and the coefficient of variation was found to be less than 20% for the two operators (Supplementary Table S2). In addition, we evaluated the intra-assay variability on triplicate samples studied at each timepoint for 6 patients who received treatment with anti-CTLA-4 and found that the coefficient of variation was less than 10% (Supplementary Table S3). These data indicate that ICOS expression on CD4 T cells could be reproducibly detected by flow cytometric analyses.

To determine whether there were any differences in the frequency of ICOS⁺ CD4 T cells in samples obtained from normal healthy donors as compared with baseline (pretherapy) samples obtained from patients with cancer, we compared data from a cohort of normal healthy donors (n = 10) to pretherapy data obtained from patients with localized bladder cancer (n = 12) and patients with metastatic melanoma (n = 34; 24 patients who would receive treatment with ipilimumab and 10 patients who would receive treatment with gp100 DNA vaccine). As shown in Table 2, there was not a statistically significant difference in the frequency of ICOS⁺ CD4 T cells at baseline among these cohorts. However, after treatment with anti-CTLA-4, the frequency of ICOS⁺ CD4 T cells significantly increased in patients with localized bladder cancer and metastatic melanoma (P ≤ 0.004). Patients who were treated with a different immunomodulatory agent, gp100 DNA vaccine, did not show a statistically significant change in the frequency of ICOS⁺ CD4 T cells (P = 0.92; Table 3). We used the upper bound of 95% CI obtained from the baseline samples (n = 56) to define a positive test for the frequency of ICOS⁺ CD4 T cells as more than 5.63. On the basis of this threshold, flow cytometric analysis for the frequency of ICOS⁺ CD4 T cells as a pharmacodynamic biomarker for ipilimumab therapy provided an average specificity of 96% (95% CI, 88–100) and sensitivity of 71% (95% CI, 54–85) after administration of two doses of anti-CTLA-4 (Table 4).

Pharmacodynamic biomarkers are markers of drug effect that usually encompass the pathway associated with the molecular target or represent downstream consequences of target and pathway modulation. In the case of anti-CTLA-4 therapy, the mAb blocks the inhibitory CTLA-4 molecule and prevents interactions between CTLA-4 and its ligand (B7), which subsequently enables B7 to be free for interactions with CD28, a critical costimulatory molecule for T-cell activation (21, 22). Therefore, anti-CTLA-4 therapy leads to enhanced T-cell activation. Upon T-cell activation, ICOS expression is upregulated (18, 23); therefore, ICOS can serve as a marker of T-cell activation. We were the first to report that anti-CTLA-4 therapy (ipilimumab) led to an increase in the frequency of ICOS⁺ T cells in patients with cancer (13–15). Subsequently, another group found an increase in the frequency of ICOS⁺ T cells after patients with cancer were treated with a different anti-CTLA-4 antibody, tremelimumab (17). In this study, we tested the hypothesis that an increased frequency of ICOS⁺ CD4 T cells detected by flow cytometry can be used as a pharmacodynamic biomarker for anti-CTLA-4 therapy. We showed 96% specificity and 71% sensitivity for flow cytometric measurement of an increased frequency of ICOS⁺ CD4 T cells after treatment with anti-CTLA-4.

Although we present data to show that the frequency of ICOS⁺ CD4 T cells can serve as a pharmacodynamic biomarker for anti-CTLA-4 therapy, it should be noted that ICOS⁺ T cells also play an important role in the antitumor response mediated by anti-CTLA-4 therapy. We have previously shown that mice lacking ICOS or its ligand (ICOSL) have impaired antitumor responses after treatment with anti-CTLA-4, as compared with wild-type mice (16). Therefore, based on the clear relationship between T-cell activation and upregulation of ICOS expression, as well as the biologic role of the ICOS/ICOSL pathway in mediating antitumor immune responses, we propose that ICOS expression on CD4 T cells represents a valid biomarker to be further developed in the setting of anti-CTLA-4 therapy.

Anti-CTLA-4 (ipilimumab) is currently approved by the U.S. Food and Drug Administration for the treatment of patients with metastatic melanoma. In addition, because anti-CTLA-4 targets a T-cell–specific molecule, as opposed to a tumor-specific molecule, it is being tested in multiple clinical trials in patients with different malignancies, which will require appropriate development of biomarkers for anti-CTLA-4 therapy. Also, because preclinical data suggest that combination therapies with agents that prime a T-cell response plus anti-CTLA-4 are more effective at eliciting antitumor responses as compared with monotherapy (24–26), clinical trials are under way with anti-CTLA-4 in various combination strategies, which will require appropriate biomarker development to evaluate whether the combination maintains biologic efficacy of anti-CTLA-4. Given the number of clinical trials that are ongoing or are expected to be conducted with anti-CTLA-4 antibody, appropriate biomarkers are clearly needed to enable rational decisions regarding dose, schedule, and combinations. Here, we define an increased frequency of ICOS⁺ CD4 T cells measured by flow cytometry as a pharmacodynamic biomarker that can be measured in blood samples, which can potentially be used for making decisions regarding different doses, schedules, and/or combinations tested in clinical trials with anti-CTLA-4. Our work represents a small retrospective analysis and warrants further prospective studies in a larger cohort of patients.

J.D. Wolchok has a commercial research grant and is a consultant/advisory board member for Bristol-Myers Squibb. J.P. Allison has ownership interest (including patents) in Bristol-Meyers Squibb and is a consultant/advisory board member for Kite Pharmceuticals and Jounce Therapeutics. P. Sharma has ownership interest (including patents) in Jounce and is a consultant/advisory board member for Bristol-Myers Squibb and MedImmune. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P. Sharma

Development of methodology: D.N. Tang, J. Yuan, P. Sharma

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.N. Tang, J.D. Wolchok, J. Yuan, P. Sharma

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.N. Tang, Y. Shen, J. Sun, S. Wen, J. Yuan, J.P. Allison, P. Sharma

Writing, review, and/or revision of the manuscript: D.N. Tang, Y. Shen, J. Sun, S. Wen, J.D. Wolchok, J. Yuan, J.P. Allison, P. Sharma

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.N. Tang, Y. Shen

Study supervision: J.D. Wolchok, P. Sharma

This work was supported in part by a UTMDACC Physician-Scientist Program Award, a Career Development Award from the UTMDACC Bladder Cancer SPORE, a Melanoma Research Alliance Young Investigator Award, a Doris Duke Clinical Scientist Development Award, and an American Cancer Society Mentored Research Scholar Grant (all to P. Sharma). Bristol-Myers Squibb sponsored and funded the clinical trials with ipilimumab. The work of P. Sharma, J.D. Wolchok, and J.P. Allison is supported by a Stand Up To Cancer – Cancer Research Institute Cancer Immunology Translational Cancer Research Grant (SU2C-AACR-DT1012). Stand Up To Cancer is a program of the Entertainment Industry Foundation administered by the American Association for Cancer Research.