Enhanced neoepitope-specific immunity following neoadjuvant PD-L1 and TGF-β blockade in HPV-unrelated head and neck cancer

Abstract

BACKGROUNDHead and neck squamous cell carcinoma not associated with HPV (HPV-unrelated HNSCC) is associated with a high rate of recurrence and poor survival.METHODSWe conducted a clinical trial in 14 patients with newly diagnosed HPV-unrelated HNSCC to evaluate the safety and efficacy of neoadjuvant bintrafusp alfa, a bifunctional fusion protein that blocks programmed death ligand 1 (PD-L1) and neutralizes TGF-β.RESULTSBintrafusp alfa was well tolerated, and no treatment-associated surgical delays or complications occurred. Objective pathologic responses (PRs) were observed, and 12 of the 14 (86%) patients were alive and disease free at 1 year. Alterations in Treg infiltration and spatial distribution relative to proliferating CD8+ T cells indicated a reversal of Treg immunosuppression in the primary tumor. Detection of neoepitope-specific tumor T cell responses, but not virus-specific responses, correlated with the development of a PR. Detection of neoepitope-specific responses and PRs in tumors was not correlated with genomic features or tumor antigenicity but was associated with reduced pretreatment myeloid cell tumor infiltration. These results indicate that dual PD-L1 and TGF-β blockade can safely enhance tumor antigen-specific immunity and highlight the feasibility of multimechanism neoadjuvant immunotherapy for patients with HPV-unrelated HNSCC.CONCLUSIONOur studies provide insight into the ability of neoadjuvant immunotherapy to induce polyclonal neoadjuvant-specific T cell responses in tumors and suggest that features of the tumor microenvironment, such as myeloid cell infiltration, may be a major determinant of enhanced antitumor immunity following such treatment.TRIAL REGISTRATIONClinicalTrials.gov NCT04247282.FUNDINGThis work was funded by the Center for Cancer Research, the NCI, and the Intramural Research Program of the NIDCD, NIH. Bintrafusp alfa was provided by the health care business of Merck KGaA (Darmstadt, Germany), through a Cooperative Research and Development Agreement with the NCI. Additional funding was provided by ImmunityBio through a Cooperative Research and Development Agreement with the NIDCD.

Keywords: Clinical Trials; Immunology; Immunotherapy.

Figures

Figure 1. Clinical and correlative study design.

Schema illustrates the neoadjuvant immunotherapy clinical trial design for the 14 study patients as well as the correlative studies performed on pre- and post-treatment tumor biopsies.

Figure 2. Primary tumor and LN responses.

(A) Waterfall plot shows primary tumor PRs (n = 14), ranked from left to right by decreasing percentage of tumor regression within the tumor bed. A PR was defined as a pTR of 50% or greater; a non-PR was defined as a pTR of less than 50%. (B) Waterfall plot shows primary tumor radiographic responses (n = 14), calculated as the percentage of change in the product of the longest primary tumor lengths and widths, ranked from left to right in the same order as in A. The asterisks indicate that the primary tumor was not visible on the CT scan. (C) Waterfall plot shows LN PRs (n = 14), ranked from left to right by decreasing percentage of primary tumor regression within the tumor bed. A PR was defined as a pTR of 50% or greater. The pound signs indicate patients who had suspicious LNs on pretreatment workup but were found to be pathologically N0 and considered to have a possible LN CR. (D) Waterfall plot shows LN radiographic responses (n = 14), calculated as the percentage of change in the product of the largest bidirectional diameter of a target suspicious LN, ranked from left to right in the same order as in A. In patients with LNs positive for carcinoma, the volume of all positive nodes was considered. For patients with suspicious nodes negative for carcinoma pathologically, only the radiographically suspicious LN volume was considered. The asterisks indicate that no clinically suspicious or pathologically positive LNs were found. The bottom black (yes) and white (no) boxes indicate patients with a disease-free status 1 year after completing the study.

Figure 3. MIF analysis of tumor TGF-β pathway activation.

(A) Heatmap shows patterns of log2 FC in nuclear H-scores of TGF-β pathway signaling proteins after treatment compared with before treatment for each patient (n = 14). (B) Dot plot shows the log2 FC in nuclear H-scores of TGF-β pathway signaling proteins in the stroma or tumor parenchyma after treatment compared with before treatment (n = 14). Most stromal cells were immune (mean, 79%; range, 59%–93%), based on size and morphology. Significance was determined with a 1-sample t test; the P value in red is significant. (C) Representative photomicrographs of pre- and post-treatment tumor Ki67 expression for patient 13, measured by immunofluorescence. Scale bars: 50 μm. (D) Box-and-whisker plots show the log2 FC in the tumor or stromal nuclear Ki67 H-score after treatment compared with before treatment in patients who did (n = 5) or did not (n = 9) have a PR. Significance was determined with a 2-tailed Mann-Whitney U test.

Figure 4. MIF analysis of tumor T cell infiltration and spatial distribution.

(A) Dot plot shows the log2 FC in immune cell density (cells/mm2) in the stroma or tumor parenchyma after treatment compared with before treatment (n = 14). Significance was determined with a 1-sample t test; P values in red are significant. (B) Box-and-whisker plots show the log2 FC in tumor CD8+ density as well as the tumor, stroma, and whole-slide CD8/Treg ratio after treatment compared with before treatment in patients who did (n = 5) or did not (n = 9) show a PR. Significance was determined with a 2-tailed Mann-Whitney U test. (C) Representative high-magnification photomicrographs of CD8+ staining for patient 3, who had a large increase in tumor CD8+ T cell infiltration after treatment compared with before treatment. Scale bars: 50 μm. (D) Representative image of a HALO spatial plot used to perform proximity analysis (inset) of FoxP3– T cells and FoxP3+ Tregs. A representative photomicrograph of a Treg directly interacting with a Ki67–CD8+ T cell is shown below. (E) Assessment of FoxP3–CD8+ or CD4+ T cells within a 100 μM radius of all FoxP3+CD4+ Tregs, with dot plot showing the mean distance between Ki67+ or Ki67– CD8+ or CD4+ T cells and Tregs in pretreatment tumors as determined by MIF (n = 14). Lines connect pre- and post-treatment measurements from individual tumors. Significance was determined with a Wilcoxon signed-rank test. (F) Assessment of all FoxP3–CD8+ or CD4+ T cells within a 100 μM radius of all FoxP3+CD4+ Tregs. Box-and-whisker plots show the log2 FC in the mean distance between Ki67+CD8+ or CD4+ T cells and Tregs after treatment compared with before treatment in patients who did (n = 5) or did not (n = 9) have a PR. Significance determined with a 2-tailed Mann-Whitney U test. (G) Distribution plots show the probability that a Ki67+CD8+ or CD4+ T cell will be at a given distance from a Treg in pretreatment (gray line) and post-treatment (blue line) tumors. Patient 3 is a representative example of a patient who developed a PR; patient 4 did not develop a PR.

Figure 5. MIF analysis of tumor myeloid cell infiltration.

(A) Representative photomicrographs of pretreatment tumor myeloid cell infiltration measured by immunofluorescence. Patient 10 showed high myeloid infiltration, and patient 7 showed low infiltration. Scale bars: 50 μm. (B) Box-and-whisker plots show quantification of pretreatment density (cells/mm2) of PD-L1+ or PD-L1– PMNs or macrophages in patients who did (n = 5) or did not (n = 9) have a PR. Significance determined with a 2-tailed Mann-Whitney U test. (C) Box- and-whisker plots show the log2 FC in the stromal Ki67+ to PD-L1+ PMN or macrophage ratio after treatment compared with before treatment in patients who did (n = 5) or did not (n = 9) have a PR. Significance was determined with a 2-tailed Mann-Whitney U test. MΦ, macrophage.

Figure 6. Neoepitope-specific TIL responses.

(A) Dot plot on the left shows the number of distinct neoepitopes eliciting positive IFN-γ responses in an ELISpot analysis of pre-treatment and post-treatment TILs for each patient (n = 12), ranked from left to right by decreasing percentage of tumor regression within the tumor bed. Dot plot on the right shows the same in patients who did (n = 5) or did not (n = 7) have a PR. Significance was determined with a 2-tailed Mann-Whitney U test. (B) Dot plot on the left shows the number of neoepitope-specific cumulative IFN-γ spots in an ELISpot analysis of pre- and post-treatment TILs for each patient (n = 12), ranked from left to right by decreasing percentage of tumor regression within the tumor bed. Dot plot on the right shows the same in patients who did (n = 5) or did not (n = 7) have a PR. Significance was determined with a 2-tailed Mann-Whitney U test. (C) Dot plots show the in silico–predicted IC50 and TPM counts for putative neoepitopes that elicited (n = 24) or did not elicit (n = 84) IFN-γ responses in TILs. Horizontal bar indicates the median. Significance was determined with a 2-tailed Mann-Whitney U test. (D) Dot plots show the total number of mutations and total number of predicted neoepitopes (IC50 <500 nM) in tumors that did (n = 8) or did not (n = 4) display detectable neoepitope-specific T cell responses in TILs. Significance was determined with a 2-tailed Mann-Whitney U test. (E) Dot plots show the total number of mutations and total number of predicted neoepitopes (IC50 <500 nM) in patients who did (n = 5) or did not (n = 97) have a PR. Significance was determined with a 2-tailed Mann-Whitney U test. (F) Dot plot shows pretreatment tumor quantification of total stromal or tumor myeloid cells in patients whose TILs did (n = 8) or did not (n = 4) display detectable neoepitope-specific T cell responses. Significance was determined with a 2-tailed Mann-Whitney U test. (G) Dot plot shows pretreatment tumor quantification of stromal cells or tumor Tregs from patients whose TILs did (n = 8) or did not (n = 4) display detectable neoepitope-specific T cell responses. Significance was determined with a 2-tailed Mann-Whitney U test.