Neoadjuvant anti-OX40 (MEDI6469) therapy in patients with head and neck squamous cell carcinoma activates and expands antigen-specific tumor-infiltrating T cells

Abstract

Despite the success of checkpoint blockade in some cancer patients, there is an unmet need to improve outcomes. Targeting alternative pathways, such as costimulatory molecules (e.g. OX40, GITR, and 4-1BB), can enhance T cell immunity in tumor-bearing hosts. Here we describe the results from a phase Ib clinical trial (NCT02274155) in which 17 patients with locally advanced head and neck squamous cell carcinoma (HNSCC) received a murine anti-human OX40 agonist antibody (MEDI6469) prior to definitive surgical resection. The primary endpoint was to determine safety and feasibility of the anti-OX40 neoadjuvant treatment. The secondary objective was to assess the effect of anti-OX40 on lymphocyte subsets in the tumor and blood. Neoadjuvant anti-OX40 was well tolerated and did not delay surgery, thus meeting the primary endpoint. Peripheral blood phenotyping data show increases in CD4+ and CD8+ T cell proliferation two weeks after anti-OX40 administration. Comparison of tumor biopsies before and after treatment reveals an increase of activated, conventional CD4+ tumor-infiltrating lymphocytes (TIL) in most patients and higher clonality by TCRβ sequencing. Analyses of CD8+ TIL show increases in tumor-antigen reactive, proliferating CD103+ CD39+ cells in 25% of patients with evaluable tumor tissue (N = 4/16), all of whom remain disease-free. These data provide evidence that anti-OX40 prior to surgery is safe and can increase activation and proliferation of CD4+ and CD8+ T cells in blood and tumor. Our work suggests that increases in the tumor-reactive CD103+ CD39+ CD8+ TIL could serve as a potential biomarker of anti-OX40 clinical activity.

Conflict of interest statement

A.D.W. is the founder of AgonOx, which has licensed the use of OX40 agonist patents for therapeutic use in cancer patients. The remaining authors declare no competing interests.

Figures

Fig. 1. Immune activation after OX40 administration in head and neck squamous cell carcinoma (HNSCC).

Patients with HNSCC cancer were given three doses of neoadjuvant anti-OX40 (MEDI6469) at 0.4 mg/kg, followed by resection at days 8, 12, or 19. a Trial schema of administration of anti-OX40 in the neoadjuvant setting in patients with HNSCC. b Blood lymphocyte and neutrophil counts were assessed at baseline, day of surgery (D8, D12, or D19), D34, and D55. The mean + SEM of the absolute cell counts per liter are shown for each subset. c–e Change in percentages of viable, CD3+ T cells. The gating strategy is outlined in Supplementary Fig. 2a, b. c Average percentages of CD4+ Tconv, CD8+, and CD4+ Treg cells during and after anti-OX40 administration measured by flow cytometry. d Summary of the average percentage of Ki-67 expression during OX40 treatment in CD4+ Tconv memory cells, CD4+ Treg cells, CD8+ TEM RA−, and TEM RA+ cells. e Average percentages of CD8+ TEMRA− and CD8+ TEMRA+ cells during and after anti-OX40 administration. f Summary of the percentages of total CD19+ cells, Ki-67+ CD19+ cells and plasmablasts, identified by the absence of IgD and expression of CD38. g Expression of ICOS on CD4+ Tconv and Treg cells during and after OX40 treatment. All cells were gated excluding doublets and dead cells. Error bars indicate mean + SEM (c–g); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns not significant. P values were determined by paired two-tailed Student’s t test between D1 and D12 or D19. N = 3 patients in the D8 cohort, N = 9 patients in the D12 cohort, and N = 4 patients in the D19 cohort (b–g). Black triangles represent the D8, orange circles the D12, and blue squares the D19 cohort. The same color code is applied to the values denoting significance. Source data are provided as Source Data file.

Fig. 2. Changes in TIL composition after OX40 administration.

TIL from a pretreatment biopsy and a surgical specimen after OX40 therapy were assessed for lymphocyte composition and activation markers. The gating strategy is outlined in Supplementary Fig. 3. a Percentages of CD4+ Tconv cells, CD4+ Treg cells, and CD8+ T cells in each patient before and after OX40 administration, N = 17 patients. b tSNE analysis of the pre and post specimens from patient HNOX07, gated on CD3+ cells. Blue represents the baseline sample, orange the day of surgery sample and gray is the concatenated file. The red circle indicates the population of cells expressing both CD103 and CD39. tSNE analysis was performed on N = 4 patients, one representative patient is shown here, two more patients are shown in Supplementary Fig. 4. c Flow cytometric analysis of the expression of CD103 and CD39 in CD4+ Tconv cells, CD8+ cells, and CD4+ Treg cells in one immune-responding head and neck squamous cell carcinoma (HNSCC) patient pre- and post OX40 therapy. d Summary of the flow cytometric analysis in (c), left panel depicts CD8+ CD103+ CD39+ T cells and the right panel depicts CD4+ CD39+ T cells; patients with an increase are shown on the left, patients with a decrease are on the right. e Expression of Ki-67 was assessed among memory CD4+ TIL and CD8+ TIL subsets (DN, SP, and DP) in biopsy (pre) and DOS (post) tissue (N = 17 patients). Blue histograms indicate pre, red indicate post tissues. The left graph shows a summary of the percentage of proliferating CD8+ T-cell subsets pre- and post anti-OX40 in each patient. The graph on the right indicates the expression of Ki-67 in CD4+ TIL pre- and post anti-OX40. Red symbols highlight patients that exhibit an increase. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns not significant. P values were determined by paired two-tailed Student’s t test between pre- and post samples (a, e). Source data are provided as Source Data file.

Fig. 3. Multiplex IHC analysis reveals changes in lymphocyte infiltrates in tumor and stroma after anti-OX40.

Multiplex IHC was performed on FFPE specimens from N = 15 patients to determine the composition and changes of the immune infiltrate in tumor and stroma. For each patient, on each slide, six regions of interest (ROI) were analyzed for tumor and stroma. a Fold change of CD8+ CD103+ CD39+ Ki-67+ T cells for all patients as determined by flow cytometric analysis. Filled orange circles indicate patients with an activation index above 1. Filled black circles indicate patients with an activation index below 1. IR immune responder, INR immune non-responders. b Total number of CD3+ and Foxp3+ T cells among tumor and stroma in N = 15 patients pre- and post anti-OX40. Error bars indicate mean ± SEM, *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns not significant. P values were determined by paired two-tailed Student’s t test between pre and post samples and between tumor and stroma. c, d Representative multiplex IHC of two HPV+ (c) and two HPV− (d) head and neck squamous cell carcinoma (HNSCC) patients pre and post anti-OX40, identified by the response (IR and INR) as determined in (a). e Magnified view (bottom) of cells expressing CD8, CD103, and Ki-67 in the tumor area (top). f Summary of the numbers of Ki-67+ CD103+ CD8+ cells in the tumor and stroma pre- and post anti-OX40, in patients stratified by the immunologic response. N = 4 IR, N = 10 INR. Error bars indicate mean ± SEM. The values presented in (b) and (f) represent the mean of the ROI analysis for each patient. g Comparison of the numbers of Ki-67+ CD103+ CD8+ cells enumerated by multiplex IHC in the tumor and stroma, with the percentages of the same subset determined by flow cytometry. Black dots represent the immune non-responders. Red dots represent the immune responders. Source data are provided as Source Data file.

Fig. 4. Survival analysis of patients after neoadjuvant anti-OX40 administration.

Patients were followed for disease-free (DFS) and overall survival (OS). a Kaplan–Meier estimate of DFS (orange) and OS (blue). The numbers below the graph represent the number of patients at risk. N = 17 head and neck squamous cell carcinoma (HNSCC) patients. Supplementary Table 4b contains the survival estimate for the overall and disease-free survival dataset as outlined in Gebski et al.. b DFS in immune-responding (N = 4, blue) vs non-responding (N = 13, orange) HNSCC patients. The log-rank test (Mantel–Cox) was used to compare both curves and P values < 0.05 were considered statistically significant. Source data are provided as Source Data file.

Fig. 5. TCRβ sequencing analysis reveals clonal differences following administration of anti-OX40.

TCRβ sequencing was performed on blood and TIL samples obtained before and after anti-OX40 administration. The gating strategy for the isolation of the blood and TIL subsets is outlined in Supplementary Fig. 6a, b. a Schematic of all T-cell subsets that were isolated from blood and tumor by cell sorting, followed by DNA isolation and TCRβ sequencing. b Summary of the clonality of CD4+ and CD8+ T cells in blood before anti-OX40 as well as blood and drLN at DOS in four patients. c Summary of the clonality of CD4+ TIL and CD8+ TIL subsets (DN, SP, and DP) before (pre) and after (post) anti-OX40. N = 2 pre, N = 4 post. Error bars indicate mean ± SEM in the post patients. d TCRβ repertoire overlap was calculated using the Morisita–Horn index. Overlap analysis is shown for DN CD8+ TIL compared with SP, DP, and drLN cells. The same analysis was performed for SP CD8+ TIL (with DN, DP, and drLN) and DP CD8+ TIL (with DN, SP, and drLN). Colors and symbols in (b) and (d) depict the four patients. e The top 30 clones in the DOS specimen (DN, SP, and DP) were separately compared to the same subsets pretreatment and to memory CD8+ T cells in blood at D1 and D12. Black open circles represent the blood before and after anti-OX40, orange filled circles represent the biopsy (pre) specimen, and red filled circles the DOS (post) specimen. Connecting lines indicate the presence of the same TCRβ sequence in each subset. f Red filled circles represent the number of clones among the top 30 clones in DP CD8+ TIL post treatment, that were present in the sample pretreatment. In b–f, N = 4 patients (HNOX04, HNOX05, HNOX11 and HNOX18) were analyzed. M memory, LN lymph node, DOS day of surgery.

Fig. 6. Identification of neoantigen and HPV-reactive cells in patients after anti-OX40.

Expanded DP CD8+ T cells from HNOX18 were screened for reactivity against HPV16, and neoantigens were predicted for HNOX04 and HNOX07. Peptide Pools were screened for reactivity by IFN-γ -ELISpot analysis. a Expanded DP CD8+ T cells from HNOX18 were screened with autologous PBMC transfected by electroporation with RNA encoding HPV16 E6 and E7 proteins. Anti-CD3 is a positive control, water the negative (MOCK) control. Prior to IFN-γ ELISpot development (right), cells were harvested and expression of 4-1BB and CD25 was assessed by flow cytometry (left). b Summary of HPV-specific spot-forming cells (SFC) in CD8+ TIL subsets. c HPV16 E6 and HPV16 E7 reactive CD8+ T cells were sorted based on the expression of 4-1BB and CD25. 4-1BB-CD25− and 4-1BB+ CD25+ cells were analyzed by TCRβ-sequencing and the frequency of the top 6 (E6) and top 7 (E7) clones are depicted. d Expanded DP CD8+ T cells from patient HNOX04 were screened with the addition of predicted neoantigens. Shown is the response to peptide 5, anti-CD3 as positive, DMSO as the negative control. e Summary of spot-forming cells (SFC) in CD8+ TIL subsets from HNOX04. f Expanded DP T cells from HNOX07 were screened with the addition of predicted neoantigens. Shown is the response to peptide 21, anti-CD3 as positive, DMSO as the negative control. g Summary of SFC in CD8+ TIL subsets from HNOX07. Red depicts the peptide response, gray is the anti-CD3 control, black the MOCK/DMSO control. TMG tandem minigene. Source data are provided as Source Data file.