Immunotherapy Targeting HPV16/18 Generates Potent Immune Responses in HPV-Associated Head and Neck Cancer

This article is featured in Highlights of This Issue, p. 1

Squamous cell carcinoma of the head and neck (HNSCCa) is diagnosed in over 500,000 patients worldwide each year, accounting for 5% of all malignancies (1). The immune system plays an important role in head and neck carcinogenesis. Even though HNSCCa, in general, are among the most highly immune-infiltrated cancer types, certain subsets of HNSCCa are characterized by an immunosuppressive environment, marked by T-cell dysfunction, low levels of CD4⁺ and CD8⁺ T cells, increased T-regulatory cells (Treg), cytokine alterations, and antigen presentation defects (2–4). Oncogenic human papilloma virus (HPV) infection accounts for a significant number of HNSCCa, most of which are related to the HPV16 subtype (5, 6). HPV-associated HNSCCa may be dependent particularly on aberrant immune checkpoints that create an immune-privileged site for HPV infection and function as an adaptive resistance mechanism of tumor against host (7, 8).

The availability of checkpoint inhibitors for metastatic HNSCCa has changed outcomes for this disease (9–12) but durable benefit and survival gains occur only in a subset of patients (∼15%–20%). Tumor HPV status does not seem to be a principal driver of outcome with this new therapy (13, 14).

Combination strategies may increase the percentage of patients who respond to PD-1 immunotherapy (15, 16). One such potential approach is the addition of HPV-specific immunotherapy, targeting HPV-specific “non-self” antigens to activate the immune system to recognize the cancer. The HPV E6 and E7 oncoproteins represent ideal targets for an immunotherapeutic agent because of their constitutive expression in HPV-associated tumors and their crucial role in the induction and maintenance of HPV-associated disease (17). HPV-specific immunotherapy may eliminate preexisting lesions and infections by generating cellular immunity against HPV-infected cells (18–21). Generation of E6- and E7-specific cellular responses has been demonstrated with the administration of VGX-3100 (Inovio Pharmaceuticals), a novel plasmid-based immunotherapy that targets E6/E7 using the SynCon optimization process to generate synthetic HPV16/18 E6 and E7 DNA sequences (22, 23). In a pilot study of 18 patients, with previously treated high-grade CIN, 3 doses of VGX-3100 delivered intramuscularly (i.m.) followed by electroporation with the CELLECTRA (Inovio Pharmaceuticals) constant current device induced HPV-specific CD8⁺ T cells that exhibited full cytolytic functionality (22). In a follow-up placebo-controlled, blinded randomized clinical trial of VGX-3100 conducted in 167 patients with HPV16/18-related CIN 2/3, histopathologic regression and clearance of HPV16/18 were seen in ∼50% of patients who received VGX-3100 alone (n = 125; refs. 22, 23). CELLECTRA delivers three 52-ms controlled electrical pulses in 3 different orientations directly at the plasmid injection site resulting in significantly enhanced transfection, leading to an increase in overall expression of antigen resulting in more robust immunogenicity than synthetic DNA alone (24–26).

We hypothesized that a cellular immune response similar to that demonstrated in women with HPV-associated CIN would be beneficial by increasing effector T cells and potentially overcoming immune evasion commonly seen in patients with HPV-associated HNSCCa. We designed a proof-of-concept study to evaluate the safety and immune effects of MEDI0457 (formerly INO-3112), a combination of synthetic plasmids targeting HPV-16 and HPV-18 E6/E7 antigens (VGX-3100) and a recombinant IL12 encoding molecular adjuvant (INO-9012), delivered by electroporation with the CELLECTRA device. One of the major drawbacks of DNA vaccines has been a reduced level of immunogenicity in humans, and addition of a plasmid cytokine adjuvant has been shown to enhance induction of cellular immune responses (27–29). IL12 is a cytokine that promotes the maturation and function of T cells (30), and in preclinical models, the inclusion of the IL12 plasmid as an adjuvant improved the magnitude, quality, and breadth of the antigen-specific cellular immune responses (31). Based on these observations, INO-9012, a dual promoter expression plasmid expressing the genes encoding human IL12 proteins p35 and p40 was added. The trial was conducted in 2 distinct patient populations with locally advanced HPV-associated HNSCCa: patients in cohort I received immunotherapy before and after definitive surgery, whereas patients in cohort II received immunotherapy after completion of concurrent chemoradiation. Cohort I was designed as an immunologic proof-of-concept study with unique access to paired tumor samples for analyses of tissue immune responses. Cohort II was designed to determine whether patients could mount an immune response to the antigens encoded by MEDI0457 after cisplatin-based concurrent chemoradiation therapy.

This was a prospective, single-center, open-label, phase I/IIa study (NCT02163057; Fig. 1). This study was conducted in accordance with the ethical guidelines outlined in the Declaration of Helsinki and was approved by the University of Pennsylvania's Institutional Review Board. Adult patients with histologically confirmed locally advanced p16⁺ HNSCCa, with adequate end organ function, ECOG PS of 0–1 were enrolled. Patients must have had a diagnostic surgical core biopsy prior to enrollment. Patients on immunosuppressive medications, including >10 mg of prednisone/day, active hepatitis B, C, or HIV, and history of active cardiac preexcitation syndromes were excluded. All patients provided written informed consent. Primary endpoints were safety and tolerability; secondary endpoints were cellular and humoral immune responses. MEDI0457 was administered by i.m. injection followed immediately by electroporation with the CELLECTRA device. In cohort I, patients were seen every 3 weeks before surgery for evaluation and treatment. Patients could receive 1 or 2 immunotherapy doses prior to surgery determined by timing of surgery. Surgery was not delayed to allow for immunotherapy dosing. About 4 weeks after surgery, if adequate healing was documented by the surgeon, the patient resumed immunotherapy dosing every 3 weeks, for a total of 4 doses. Patients then went on to receive standard adjuvant therapy (if indicated) based on pathologic features. In cohort II, patients started immunotherapy 2 months after completion of chemoradiation in order to allow recovery from the acute side effects of combined modality therapy. After completion of immunotherapy, patients in both cohorts were seen every 3 months for total of 6 months for safety follow-up and measurement of immune responses. Adverse events were recorded and graded according to NCI Common Terminology Criteria for Adverse Events (CTCAE) v 4.0.

Peripheral blood mononuclear cells (PBMCs) were collected in ACD-A tubes and cells were isolated within 24 hours of draw. Serum was collected in red top tubes. Samples were collected at baseline, at the time of immunotherapy dosing, and at each follow-up visit (Fig. 1) and cryopreserved for immune analyses in batches. T-cell and antibody responses to HPV16 and HPV18 E6 and E7 were determined by interferon-γ ELISpot, and ELISA as described previously (22). For flow cytometry, PBMCs were recovered after cryopreservation overnight in cell culture medium and spun, washed, and resuspended the following day. After counting, 1 × 10⁶ PBMCs were plated into a 96-well plate in R10 medium from patients with sufficient sample. For antigen-specific responses, cells were stimulated 5 days with a combination of peptides corresponding to HPV16 E6 and E7 or HPV18 E6 and E7 that had been pooled at a concentration of 2 μg/mL, while an irrelevant peptide was used as a negative control (OVA) and concanavalin A was used as a positive control (Sigma-Aldrich). No costimulatory antibodies or cytokines were added to cell cultures at any point. At the end of the 5-day incubation period, cells were stained for CD3-APCH7, CD4-PerCPCy5.5, CD14-Pacific Blue, CD-16 Pacific Blue, CD137-APC, PD1-PECy7, Granulysin-FITC (BD Biosciences), CD38-QDot705 and CD69-QDot800 (Thermo Fisher), CD8-BV605 and Granzyme A-AF700 (BioLegend), CD-19 Pacific Blue, granzyme B-PETR (Invitrogen), and perforin-PE (Abcam). Staining for extracellular markers (CD4, CD8, CD137, CD69, CD38, and PD-1) occurred first, followed by permeabilization to stain for the remaining markers. CD3 was stained intracellularly to account for downregulation of the marker following cellular activation. Prepared cells were acquired using an LSR II flow cytometer equipped with BD FACSDiva software (BD Biosciences). Acquired data were analyzed using the FlowJo software version X.0.7 or later (Tree Star).

ELISpot assays were performed using separate E6 and E7 peptide pools using a 12-hour stimulation period. Antigenic pools were divided up as follows: HPV16 E6 peptides only, HPV16 E7 peptides only, and HPV18 E6 plus E7 peptides pooled into a single stimulation. This latter stimulation was done as a method of preserving available cells for use in other assays. Results are graphed as individual responses to the combined response of HPV16 E6+HPV16 E7 or the response to the HPV18 E6+E7 peptides. Each patient's sample received the HPV16 E6 peptide pool as its own triplicate, HPV16 E7 as its own triplicate and HPV18 E6+E7 as a single triplicate. Results are represented as mean SFU/10⁶ PBMCs generated from the triplicate media control wells subtracted from mean SFU/10⁶ PBMCs generated from HPV antigen triplicate wells. Because there were 2 cohorts in the study (Fig. 1), that could vary in their dosing schedule (cohort I dosing was contingent on the timing of surgery), patients' results were graphed based on dose and not by study week. CEF (CMV, Epstein–Barr, Flu) peptides were included as a control for immune competence during the course of the study; responses to these antigens occur through natural infection/vaccination and are not linked to responses to HPV antigens.

As the surgical center was not adjacent to the tissue analytical labs, we were not able to isolate fresh tissue from resection specimens that would survive transport to another location for viable tumor-infiltrating lymphocyte (TIL) extraction, isolation, and analysis. In place of fresh TIL isolation, tissue was formalin-fixed and embedded in paraffin for analysis by IHC. All IHC assays used a polymer/multimer-based secondary detection system. For each IHC staining run, regardless of which platform was used, positive control tissues were included that were treated with the primary antibody and a Buffer Negative control in which the primary antibody was omitted. Samples were initially allocated first for staining CD8 (Dako clone C8/144B, cat. #M7103) and Foxp3 (Ebio clone 236/E7, cat. #14-4777). Remaining tissue was then allocated for staining perforin (Abcam clone dG9, cat. #ab194807) and PD-L1 (VENTATA PD-L1 SP263 assay). Whole slide image capture was performed by Histologix (Biocity) at ×20 magnification with a Hamamatsu Nanozoomer 1.0-HT digital slide scanner. Normal neoplastic regions of interest (ROI) were digitally annotated, where present, onto each section image by the study pathologist. Quantitative image analysis of IHC staining within the annotated ROI was performed by OracleBio (Biocity) using Definiens Tissue Studio software. An analysis algorithm was developed to detect positive cellular staining across each tissue image. Within the algorithm, image colors were initially separated into respective stain components (e.g., brown and blue). Cells were defined and generated based on the presence of a blue (hematoxylin) stained nucleus. A threshold level based on identified positive and negative staining in control tissues was then applied to the positive brown color intensity parameter within each cell, above which a cell was defined as positive. The number or area of positive and negative stained cells was then quantified within specific ROIs across each tissue.

Positive PD-L1 status was assigned to cases with total percentage of tumor cells with membrane staining of any intensity of greater than or equal to 25%. Tissue obtained from the diagnostic biopsy was used to determine HPV16 or HPV18 positivity via in situ hybridization (ISH) using proprietary probe sets specific for HPV16 or HPV18 manufactured by DAKO (DAKO Agilent Pathology Solutions), with a reflex to Roche Linear Array if hybridization was not positive for either HPV oncotype. Tissue samples from cohort I were additionally confirmed for HPV oncotype using RNAscope for HPV16 and HPV18 probes (Advanced Cell Diagnostics).

The primary analysis was used to estimate safety. Toxicities were recorded, graded, and summarized. Subjects who received the assigned number of doses were included in the secondary immunogenicity and survival analyses. Changes in immune parameters following immunotherapy, including immunoglobulinG (IgG) responses measured by ELISA, number of antigen-specific IFNγ-secreting cells in response to stimulation with E6 and E7 pools measured by IFNγ ELISpot assay and induction of antigen-specific CD8⁺ T-cell populations expressing granzyme A, granzyme B, and perforin, were analyzed with Wilcoxon signed-rank tests. In exploratory analyses, disease-free survival (DFS) and overall survival (OS) were measured from the time of administration of the initial immunotherapy dose.

Between May 2014 and August 2016, 27 patients were screened, and 22 were enrolled and treated: 6 in cohort I and 16 in cohort II. Each patient received 4 total doses of MEDI0457. Demographics are summarized in Table 1. In keeping with the global epidemiology of HPV-associated HNSCCa, patients on this study were predominantly male, with median age 57.5 years, and about half were never smokers. HPV genotyping was available for all patients. The majority of the patients had HPV16 (n = 19, 86%); 3 patients had other genotypes (HPV26, 33, and 35). MEDI0457 was well tolerated. All adverse events (AE) regardless of attribution are detailed in Supplementary Table S1. There were no grade 3 or higher related AEs. The most common related AEs were injection site pain (all grade 1, 21 events), which were evenly distributed among cohorts I and II. All patients are alive with median follow-up of 15.9 months; 3 patients developed progressive disease, and 12-month DFS rate was 89.4%.

Generation of antibodies to both the E6 and E7 antigens of HPV16 and HPV18 was observed (Fig. 2A) following MEDI0457 treatment. Peak seroreactivity was highest against the HPV18 E7 antigen (88.2% of subjects) followed by the HPV16 E7 antigen (64.7%). Across both cohorts, 100% of patients showed seroreactivity to at least 1 antigen. HPV16 E6 and HPV18 E6 antigens had lower seroreactivity compared with E7 antigens (Supplementary Table S2), which is consistent with previous studies of VGX-3100. Antibody titers could be detected against at least 1 of the 4 HPV antigens 3 months after the last dose of immunotherapy, indicating that administration of MEDI0457 could induce antibodies that persisted for at least 6 months after the start of immunotherapy. The longest persistence of seroreactivity against all 4 antigens in a single patient was noted 9 months following the final dose. Persistent reactivity against any antigen was noted in a single patient 23 months following the final dose of MEDI0457, when the patient was discharged from the study.

Assessment of cellular immune responses induced by MEDI0457 was done by performing an overnight IFNγ ELISpot without the addition of supportive cytokines on directly isolated PBMCs obtained prior to and following MEDI0457 dosing. Cellular reactivity against HPV16 and HPV18 antigens across the study population prior to dosing was generally low (<25SFU/10⁶ PBMCs against all 4 HPV antigens) with the exception of 2 patients in cohort I (patient 9 and patient 24) who had minimal baseline reactivity against HPV16 (45.8 and 43.3 SFU/10⁶ PBMCs; Fig. 2B). After immunotherapy, both cohorts displayed peak ELISpot responses that exceeded predose levels against both HPV16 and HPV18 antigens (Fig. 2B). Specifically, patients in cohort I developed median peak increases above baseline for HPV16 and HPV18 antigens of 63 and 75 SFU/10⁶ PBMCs, while those in cohort II exhibited median increases of 75 and 55 SFU/10⁶ PBMCs, respectively (Fig. 2C). As HPV16-associated HNSCCa constituted the majority of the study population and no patients with HPV18 genotype associated malignancy were enrolled, we focused on ELISpot analysis of responses against individual HPV16 antigens. The HPV16 E6 and E7 antigens were similarly immunogenic, as shown by a comparison of peak median ELISpot responses. Cohort I showed median increases above baseline for HPV16 E6 antigen of 30 SFU/10⁶ PBMCs and 10 SFU/10⁶ PBMCs for HPV16 E7 (Fig. 2D). Similarly, cohort II showed median increases above baseline of 32 SFU/10⁶ PBMCs for HPV16 E6 and 20 SFU/10⁶ PBMCs for HPV16 E7 (Fig. 2D). Importantly, 8 of 21 patients showed a peak response of ≥100 SFU/10⁶ PBMCs against HPV16 E6 or E7, suggesting robust HPV-specific cellular immune response. Responses appeared to be diminished when assayed 3 months after the completion of dosing in both cohorts, but remained above baseline for most patients (Fig. 2E) and persistent long-term reactivity against any antigen was noted in 1 patient 23 months following the final treatment, when the patient completed the study.

We and others have shown previously that the production of IFNγ is suggestive of a Th1 immune response but does not correlate 1:1 with lytic activity (23, 24, 32–34). It is generally accepted that a cytolytic response by CD8⁺ T cells is likely to be of importance for control and elimination of neoplastic cells. We therefore performed flow cytometry on PBMCs from patients with sufficient sample isolated prior to the first dose and following the last dose of MEDI0457 to assay the ability of HPV16- and HPV18-specific CD8⁺ T cells to load granzymes and perforin in response to treatment (total n = 8, n = 2 from cohort I; n = 6 from cohort 2). To that end, we analyzed the CD8⁺ T-cell compartment for immune activation via antigen-specific expression of cell-surface markers such as CD38, CD69, and CD137 as well as for lytic potential as determined by the presence of granzyme A (GrzA), granzyme B (GrzB), and perforin (Prf; Fig. 2A; Supplementary Fig. S1) after in vitro stimulation with cognate antigens. When assessing for activation based on expression of CD137, 3 of the 8 patients showed immune activity to HPV16 antigens prior to MEDI0457 and 3 patients showed baseline reactivity to HPV18 (Supplementary Tables S3 and S4). Following MEDI0457 administration, 3 patients showed clear elevations in activated HPV16-specific CD8⁺CD137⁺ T-cell populations expressing GrzA, GrzB, and Prf, in a range of 0.24% to 0.52% of total CD8⁺ T cells over baseline (Fig. 3B, top left and Supplementary Tables S3 and S4) after stimulation in this assay, one of whom had limited activity at baseline. Five patients showed an elevation in this population of CD8⁺ T cells specific for HPV18 antigens in a range of 0.04% to 0.35% (Fig. 3B, top right), including 3 patients who also showed HPV16 reactivity after MEDI0457. We performed an additional assessment of CD8⁺ T-cell activation based on the expression of CD69 and found that, at baseline, 2 of 8 patients showed immune activity to HPV16 antigens and 2 nonoverlapping patients showed reactivity to HPV18. Following treatment, 4 patients showed clear elevations in activated HPV16-specific CD8⁺CD69⁺ T-cell populations expressing GrzA, GrzB, and Prf, in a range of 0.28% to 1.16% of total CD8⁺ T cells over baseline after stimulation in this assay (Fig. 3B, middle left). Three of those 4 patients also showed an elevation in this population of CD8⁺ T cells specific for HPV18 antigens in a range of 0.40% to 1.21% (Fig. 3B, middle right). Finally, we assessed CD38 regulation based on antigen-specific cellular activation and found that at baseline, 3 of 8 patients showed immune activity to HPV16 antigens, with 2 of those 3 patients also showing reactivity to HPV18 at baseline. Following MEDI0457, 4 patients showed clear elevations in activated HPV16-specific CD8⁺CD38⁺ T-cell populations expressing GrzA, GrzB, and Prf, in a range of 0.50% to 1.38% of total CD8⁺ T cells over baseline after stimulation in this assay (Fig. 3B, bottom left). All 4 of these patients and an additional fifth patient showed an elevation in this population of CD8⁺ T cells specific for HPV18 antigens in a range of 0.06% to 1.33% (Fig. 3B, bottom right). In total, 5 of the 8 patients tested showed both HPV16- and HPV18-specific elevations in at least 1 of the 3 parameters tested by flow cytometry after treatment with MEDI0457 (Supplementary Tables S3 and S4). These results are not statistically significant, likely due to the small sample size. Nevertheless, the results suggest that MEDI0457 drove the induction of CD8⁺ T cells capable of activation in the context of antigenic exposure and that these cells were capable of granzyme and perforin synthesis, thus exhibiting a clear CTL phenotype.

Paired tumor samples obtained from baseline (prior to treatment with MEDI0457) and at definitive surgery (after a single dose of MEDI0457) were available on 5 cohort I patients. These paired samples were analyzed for the presence of TILs and stained for CD8 (Fig. 4A), and FoxP3 (Fig. 4B) by IHC. After MEDI0457, increases in CD8⁺ infiltrates per mm² of neoplastic tissue were noted in 2 patients (Fig. 4C), and decreases in FoxP3⁺ infiltrates were noted in 3 patients (Fig. 4C). We calculated the CD8/FoxP3 ratio within this tissue to assess the possibility of a shift toward a proinflammatory response and showed a positive change in 4 of 5 patients, including 1 patient with a greater than 3-fold rise in this measure (Fig. 4C). The fifth patient showed a CD8/FoxP3 ratio shift in the negative direction (decrease in both CD8 and FoxP3 infiltration, with the former being greater), suggesting the possibility of a dampened inflammatory response for this patient. We further assessed these surgical specimens for the presence of immune infiltrates that were perforin positive in order to determine cytolytic capacity of the infiltrates. Interestingly, all 5 patients showed increases in the number of perforin-positive immune infiltrates in neoplastic tissue (Fig. 4C and D). The presence of perforin-positive infiltrates in tissue samples from patient 21 (with the negative shift in the CD8/FoxP3 ratio) suggests the presence of a mixed immune state (increases in both pro- and anti-inflammatory markers). PD-L1 assessment by IHC staining was also performed on these paired samples, with a “positive” score being assigned to total percentage of tumor cells with membrane staining of any intensity of greater than or equal to 25% (Fig. 4E). Interestingly, all 5 patients assessed showed negative PD-L1 expression at study entry, with 1 patient showing upregulation of PD-L1 expression after treatment with MEDI0457 (Fig. 4C). Because there was no intervening therapy between the paired biopsies other than MEDI0457, the results of CD8, FoxP3, and perforin staining provide evidence that MEDI0457 directly altered the composition of TILs in tumor tissue.

Three patients exhibited disease progression while on study. One patient in cohort I, patient 21, had disease recurrence 7 months after completion of adjuvant chemoradiation therapy, including hemorrhagic dermal and lymph node metastases. Palliative radiotherapy was instituted for control of bleeding metastases followed by therapy with nivolumab. The patient had a complete radiographic response to treatment following just 4 cycles of nivolumab (Fig. 5A and B). The patient's tonsil HNSCCa was confirmed to be HPV16 associated by genotyping. In light of an unusually rapid and complete response (CR) to anti–PD-1 antibody therapy, we analyzed the immune responses in detail in this patient. Assessment of peripheral immune responses induced by MEDI0457 in this patient revealed the induction of humoral responses (endpoint titers of 1:450 and 50 for HPV16 E6 and E7, respectively) but only limited cellular responses as gauged by IFNγ ELISpot (peak ELISpot responses above baseline of 7 SFU/10⁶ PBMCs for HPV16 E6 and 7 SFU/10⁶ PBMCs for HPV16 E7 as compared with median responses of 30 and 10 SFU/10⁶ PBMCs for all of cohort I patients; Fig. 5C). Patient 21 was also the only patient in cohort I who did not show a positive shift in the CD8/FoxP3 ratio in resected tumor (Fig. 4C). This suggests the possible absence of an antitumor inflammatory state and may partly explain the disease progression that occurred. Although this noninflamed tumor profile appears to fit with the patient's early disease progression, the unusually rapid achievement of CR after PD-1 blockade was surprising. In order to investigate this paradox further, we performed additional assessments of peripheral blood samples taken from this patient prior to MEDI0457 dosing as well as 8 months following the completion of dosing (15 days prior to the first infusion of nivolumab). We specifically assessed MEDI0457-driven CD8⁺ T-cell activity and the expression of PD-1 after in vitro stimulation with HPV16 antigens. Interestingly, HPV16 antigen–specific PD-1⁺ CD8⁺ T cells were not found prior to MEDI0457 but were found at robust levels after MEDI0457 dosing (0% vs. 1.8%, Fig. 5D left). In order to determine if such cells might harbor the potential for a desirable effector response, we assessed the expression of granzyme A, granzyme B, and perforin. Results of this analysis indicate that nearly half of the HPV16-specific PD-1⁺ CD8 T cells observed in the assay also expressed these lytic proteins (0.70%, Fig. 5D right), suggesting the potential for HPV16-specific lytic activity in this T-cell subset. We hypothesize that the expression of PD-1 on peripheral CD8⁺ T cells may have allowed them to be inhibited when bound to tumor cells expressing PD-L1. Nivolumab may have relieved the inhibition of these MEDI0457-induced HPV16-specific CTLs, allowing their outgrowth and leading to the durable CR. At the time of this publication, this patient remains in CR, 24 months after initiation of anti–PD-1 therapy.

Here, we report results of a phase I/IIa immunologic proof-of-principle clinical trial of an HPV16 and HPV18 E6/E7 DNA immunotherapy in patients with HPV-associated locally advanced HNSCCa that included a novel IL12 DNA adjuvant (MEDI0457), both delivered by electroporation with the CELLECTRA device. We demonstrated that administration of the immunotherapy was safe and well tolerated both in the perioperative setting (cohort I) and after multimodality therapy with chemotherapy and radiation (cohort II). There were no treatment-related serious AEs. The immunotherapy could be safely delivered as part of clinical standard of care, and in the cohort I patients did not result in delays in time to surgery. We observed induction of strong humoral and cellular responses even after a single dose of MEDI0457. All patients showed induction of humoral responses against at least 1 HPV-specific antigen, with persistence of humoral responses for up to 23 months. That all patients responded to the immunotherapy is especially striking in cohort II. Despite receiving cisplatin and radiotherapy, which cause lymphopenia that might blunt an immune response, patients in cohort II were nonetheless able to mount a potent immune response to the immunotherapy. Although the humoral responses induced against the E6 or E7 oncoproteins may not play a major role against progression of HPV-associated HNSCCa, E6- or E7-specific antibodies do reflect immunotherapy potency and, as shown in our study, were accompanied by the parallel induction of antigen-specific cellular immune responses. Why there are differences in the induction of immune reactivity to HPV16/18 E6- and E7- antigens is currently unclear, but this may be due to a number of factors, including intrinsic immunogenic potential of these antigens as well as their relative ordering on the plasmid constructs, which, in turn, may affect the expression, processing, or presentation of the antigens.

We have previously reported the results of a phase IIb randomized, blinded, placebo-controlled study, testing VGX-3100 (HPV16 and HPV18 E6 and E7 antigens without IL12) in women with high-grade cervical dysplasia (CIN2/3) who were positive for HPV16 and/or HPV18 (23). Immune responses observed in the phase IIb study in patients with CIN2/3 were similar or greater when compared with the magnitude of the responses observed in the current study of patients with HNSCCa. Relative immune competency of the 2 patient populations may be a factor accounting for this difference as patients with an untreated locally advanced cancer or who have just completed chemoradiation may be immune suppressed by the tumor itself and/or treatment with chemotherapy or radiation (35). That said, we were encouraged by the robustness of the immune response seen in patients on the current clinical trial.

In the current study, we demonstrated that immunotherapy with MEDI0457 induced antigen-specific cytotoxic T cells that were functionally capable of granzyme and perforin synthesis. Cells with this phenotype were also noted in the randomized controlled trial in patients with CIN and, on that study, were shown to be significantly associated with clinical response to treatment (23, 32). Pre- and postimmunotherapy tissue analyses from the current study revealed that MEDI0457 altered the composition of TILs in tumor tissue tilting the cellular immune profile in most of the patients toward a proinflammatory state. These findings confirm our hypothesis that HPV-specific immunotherapy would induce potentially beneficial antitumor immune responses and are consistent with our previous finding that HPV-specific immunotherapy is able to generate tissue infiltrating immune responses that correlate with complete resolution of high-grade CIN (23, 32).

The observation of a sustained complete clinical response after just 4 doses of nivolumab (ongoing for >18 months) in a patient with progressive metastatic disease is remarkable. Complete responses after treatment with PD-1 inhibitors have been observed, but are uncommon, and usually occur later in the treatment course (9, 10, 13, 14). Our data showing expansion of antigen-specific PD-1⁺ CD8⁺ cells with cytolytic potential in this patient are evidence of the immunogenicity of MEDI0457. One hypothesis is that subsequent therapy with nivolumab facilitated the outgrowth of functional HPV16-specific CTLs that had been induced by the MEDI0457 immunotherapy, but were impeded from antitumor effector activity by coexpression of PD-1. A combination approach, using immunotherapy to “prime” the immune response by activation of antigen-specific T cells followed by therapeutic PD-1 blockade, may be one way to increase response rates to immunotherapy in HPV-associated HNSCCa.

In murine tumor models, several experimental HPV immunotherapy strategies, including vector-, peptide-, protein-, and nucleic acid–based, as well as dendritic immunotherapies and RNA replicon approaches have been shown to enhance HPV-specific immune cell activity and antitumor responses (36–41). Many of these strategies have been tested in humans in the setting of CIN- or HPV-associated cervical cancer, but evaluation as treatment for advanced HNSCCa remains understudied, despite a large unmet medical need in the face of an “epidemic” of HPV-associated HNSCCa. Axalimogene filolisbac (AXAL or ADXS11-001) is a novel immunotherapeutic based on live irreversibly attenuated Listeria monocytogenes (Lm) fused to the nonhemolytic fragment of listeriolysin O (Lm-LLO) and secretes the Lm-LLO-HPV E7 fusion protein targeting HPV-positive tumors (42). ADXS11-001 is currently being evaluated in phase II clinical trials for cervical cancer, and HPV-positive head and neck cancer (43). Preliminary data suggest robust induction of immune responses by ADXS11-001; side effects include pyrexia, anemia, vomiting, chills, and muscle pain (36, 44), and clinical responses have not yet been reported. We did not observe systemic effects such as fever, myalgia, arthralgia, or malaise with MEDI0457. Another phase I dose escalation trial is evaluating the safety and immunologic responses to peptide immunomodulatory immunotherapy GL-0810 (HPV16) and GL-0817 (MAGE-A3) in HPV16- and MAGE-A3–positive metastatic or recurrent HNSCCa patients, respectively. The immunotherapy was safe, and no significant AEs were noted. In patients who received all 4 doses, 80% of HPV16–positive and 67% of MAGE-A3–positive patients developed both a T-cell and antibody response. In both groups, there was a significant correlation between T-cell response and subsequent antibody response, but these immune responses did not translate into any clinical responses as measured by RECIST (45).

Finally, ISA 101, a synthetic long peptide immunotherapy directed against HPV16, is being studied in combination with nivolumab in patients with recurrent/metastatic HPV16-associated oropharyngeal cancer. In a recent report, an overall response rate of 36% (8/22 patients) was reported with the use of this combination (46). Immunologic biocorrelative data have not yet been reported, and it is unclear how much of the clinical benefit is attributable to the immunotherapy versus PD-1 blockade alone in this PD-1 treatment-naïve population. Nevertheless, this report provides early signals of efficacy with a therapeutic vaccine combined with PD-1 blockade in patients with metastatic HPV-associated HNSCCa.

Our study has important limitations that must be acknowledged. First, in the initial iteration of the clinical protocol, patients were not selected for HPV16 and 18 genotypes based on HPV genotyping, but were enrolled solely on the basis of p16 positivity. HPV genotyping was performed retrospectively, and consequently, 3 patients with non-HPV16 and 18 tumors were enrolled. The protocol was then amended to require demonstration of HPV16 and/or HPV18 positivity. Second, none of the patients on this study were positive for the HPV18 genotype. Third, this was a relatively small, single-institution study and our results need to be confirmed in a larger population of patients. Finally, our study was not designed to answer questions related to optimal dosing schedule, timing and duration of immunotherapy, or the contribution of the IL12 DNA adjuvant. Instead, these aspects of dosing were based on the prior experience in patients with CIN and warrant further investigation.

Our observations provide novel insights into advancing therapy for HNSCCa. Indeed, this is the first report of detailed immune responses following therapeutic HPV-specific DNA-based immunotherapy in patients with locally advanced HNSCCa. There have been very few, if any, novel approaches in the management of HPV-associated HNSCCa because the HPV epidemic was first described (47). There are currently no FDA-approved therapies specific to this population. Instead, treatments developed in patients with HPV-negative HNSCCa—a biologically different disease—are typically used. Approximately a third of patients with HPV-associated HNSCCa will go on to develop fatal metastatic disease. Based on our data, one can envision using therapeutic HPV-specific immunotherapy across the various settings that are relevant in this disease, including the preinvasive stage where immunotherapy could be used to slow progression to carcinoma or in conjunction with existing therapies (surgery, chemotherapy, and radiation) for locally advanced high-risk HPV-associated HNSCC to reduce local recurrence and metastases. Additionally, as noted in our discussion of an exceptional responder, HPV immunotherapies may be useful to increase the activity of checkpoint inhibitors by inducing the requisite proinflammatory state. A phase Ib/IIa, open-label study of MEDI0457 in combination with durvalumab (anti–PD-L1 antibody) is under way to determine whether this combined approach will improve the response rate in patients with recurrent/metastatic HPV-associated HNSCCa (NCT03162224). The consistent induction of antigen-specific humoral and functional cellular immune responses in the current study should encourage the use of this “targeted immunotherapeutic” approach in the management of HPV-associated HNSCCa.

C. Aggarwal is a consultant/advisory board member for Bristol-Myers Squibb, Celgene, and Medimmune. R.B. Cohen reports receiving commercial research grants from and is a consultant/advisory board member for Innate. M.P. Morrow, K.A. Kraynak, A.J. Sylvester, D.M. Knoblock, M. Dallas, I. Csiki, and M. Bagarazzi hold ownership interest (including patents) in Inovio Pharmaceuticals. J. Bauml reports receiving commercial research grants from Astra Zeneca, Bayer, Incyte, Janssen, Merck, Novartis, and Takeda, and is a consultant/advisory board member for Astra Zeneca, Bristol-Myers Squibb, Celgene, Genentech, Guardant Health, Merck, and Takeda. A. Lin is a consultant/advisory board member for Galera Pharmaceuticals. M.T. Esser and R. Kumar hold ownership interest (including patents) in AstraZeneca. D.B. Weiner is an employee of GeneOne, Inovio, and Medimmune, reports receiving commercial research grants from GeneOne, Inovio, and Janssen, holds ownership interest (including patents) in Inovio, and is a consultant/advisory board member for GeneOne and Inovio. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. Aggarwal, R.B. Cohen, M. Dallas, D. Weiner, M.L. Bagarazzi

Development of methodology: C. Aggarwal, M.P. Morrow, D. Weiner, M.L. Bagarazzi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Aggarwal, R.B. Cohen, M.P. Morrow, J.M. Bauml, G.S. Weinstein, A. Lin, J. Boyer, S. Tan, A. Anton, K. Dickerson, R. Vang, S. Oyola

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Aggarwal, R.B. Cohen, M.P. Morrow, K.A. Kraynyak, A.J. Sylvester, D.M. Knoblock, J.M. Bauml, L. Sakata, M. Dallas, S. Oyola, S. Duff, R. Kumar, D. Weiner, M.L. Bagarazzi

Writing, review, and/or revision of the manuscript: C. Aggarwal, R.B. Cohen, M.P. Morrow, K.A. Kraynyak, A.J. Sylvester, J.M. Bauml, A. Lin, J. Boyer, D. Mangrolia, M. Dallas, M. Esser, R. Kumar, D. Weiner, I. Csiki, M.L. Bagarazzi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Aggarwal, M.P. Morrow, A.J. Sylvester, S. Oyola, S. Duff, R. Kumar

Study supervision: C. Aggarwal, I. Csiki, M.L. Bagarazzi

We would like to thank the patients who participated and their families as well as Hoyin Mok, Jiping Zha, and Lily Cheng for their work on the ACD and IHC data. We would also like to thank Histologix and OracleBio for assistance with additional IHC staining and digital image analysis. We would like to thank Alison Berry and Kristine Mykulowycz at the University of Pennsylvania for clinical research support. This study was supported by NCI P30 Cancer Center Support grant #5-P30-CA-016520-38.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.