ISA101 and nivolumab for HPV-16+ cancer: updated clinical efficacy and immune correlates of response.

BACKGROUND: The combination of ISA101, a human papilloma virus (HPV) 16 peptide vaccine, and nivolumab showed a promising response rate of 33% in patients with incurable HPV-16+ cancer. Here we report long-term clinical outcomes and immune correlates of response.
METHODS: Patients with advanced HPV-16+ cancer and less than two prior regimens for recurrence were enrolled to receive ISA101 (100 µg/peptide) on days 1, 22, and 50 and nivolumab 3 mg/kg every 2 weeks beginning day 8 for up to 1 year. Baseline tumor samples were stained with multiplex immunofluorescence for programmed death-ligand 1 (PD-L1), programmed cell death protein-1 (PD-1), CD3, CD8, CD68, and pan-cytokeratin in a single panel and scanned with the Vectra 3.0 multispectral microscope. Whole transcriptome analysis of baseline tumors was performed with Affymetrix Clariom D arrays. Differential gene expression analysis was performed on responders versus non-responders.
RESULTS: Twenty-four patients were followed for a median of 46.5 months (95% CI, 46.0 months to not reached (NR)). The median duration of response was 11.2 months (95% CI, 8.51 months to NR); three out of eight (38%) patients with objective response were without progression at 3 years. The median and 3-year overall survival were 15.3 months (95% CI, 10.6 months to 27.2 months) and 12.5% (95% CI, 4.3% to 36%), respectively. The scores for activated T cells ((CD3+PD-1+)+(CD3+CD8+PD-1+)), activated cytotoxic T cells (CD3+CD8+PD-1+), and total macrophage ((CD68+PD-L1-)+(CD68+PD-L1+)) in tumor were directly correlated with clinical response (p<0.05) and depth of response with the two complete response patients having the highest degree of CD8+ T cells. Gene expression analysis revealed differential regulation of 357 genes (≥1.25 fold) in non-responders versus responders (p<0.05). Higher expression of immune response, inflammatory response and interferon-signaling pathway genes were correlated with clinical response (p<0.05).
CONCLUSIONS: Efficacy of ISA101 and nivolumab remains promising in long-term follow-up. Increased infiltration by PD-1+ T cells and macrophages was predictive of response. Enrichment in gene sets associated with interferon-γ response and immune infiltration strongly predicted response to therapy. A randomized trial is ongoing to test this strategy and to further explore correlates of immune response with combined nivolumab and ISA101, versus nivolumab alone. TRIAL REGISTRATION NUMBER: NCT02426892. © Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

Background

Human papilloma virus driven (HPV+) cancers are highly prevalent worldwide with the most common tumors being oropharyngeal and anogenital cancers. The E6 and E7 viral proteins play a critical role in HPV-induced carcinogenesis and have constitutive expression in HPV-associated tumors representing an ideal target for therapeutic vaccines.

The ISA101 vaccine consists of HPV-16 E6 and E7 synthetic long peptides which are efficiently processed by dendritic cells to activate CD4+ and CD8+ T-cells and drive HPV-16-specific antitumor immunity (online supplemental figure 1). ISA101 can eradicate HPV-16+ pre-malignant vulvar lesions and lesion clearance is correlated with the strength of the T cell response.1 However, despite increasing the HPV-16 immune response, ISA101 monotherapy has limited efficacy in the advanced cancer setting, likely due to dominant immunosuppressive signals within the tumor microenvironment (TME).2

Immune checkpoint inhibitors (ICI) have shown survival benefit in the treatment of incurable oropharyngeal squamous cell carcinoma (OPSCC), irrespective of HPV status, however, only 15%–20% of patients experience objective responses.3–5 HPV+ OPSCC has a significantly better prognosis in comparison with HPV− OPSCC and presents with higher T-cell infiltration.6 The recognition of tumor-specific antigens by T-cells is integral to the success of cancer immunotherapy. In this context, HPV-16 vaccination with ICI could enhance tumor regression through increasing the frequency of tumor-specific vaccine-activated T cells whose longevity and effector function is subsequently preserved and amplified by programmed cell death protein-1 (PD-1) blockade. This rationale supported our phase II trial with ISA101 and nivolumab for patients with advanced HPV-16+ tumors. We observed an objective response rate (ORR) of 33%, durable duration of response, promising survival and favorable safety profile.7

Discovering a deeper understanding of the TME composition and metabolic pathways has become a fundamental step in discovering genetic and epigenetic adaptations in cancer, as well as in revealing distinct immunologic signatures and immune-resistance mechanisms, such as loss of interferon-γ (IFN-γ) signaling and T-cell exhaustion. Here we report long-term efficacy data and immune correlatives of response from this phase II study of ISA101 and nivolumab in patients with HPV-16+ solid tumors.7

Methods

Study design and patients

The design of this phase II trial was reported previously.7 Eligible patients had incurable HPV-16+ solid tumor with ≤1 regimen for recurrent disease, an Eastern Cooperative Oncology Group performance status of 0–1, and measurable disease per Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1.

Patients received ISA101 subcutaneously, 100 µg/peptide, for a total of three doses on days 1, 22, and 50 and nivolumab intravenously, 3 mg/kg starting on day 8 and administered every 2 weeks for 12 months or until progression of disease, toxic effects, or withdrawal of consent (online supplemental figure 2).7 Tumors were assessed radiologically at baseline, prior to cycle six of nivolumab and then every 6 weeks according to RECIST. Tumor biopsies were mandatory pretreatment and planned at first restaging. The primary endpoint was ORR.

Multiplex immunofluorescence

Multiplex immunofluorescence (mIF) analysis was performed using previously validated methods.8 Formalin-fixed paraffin-embedded 4 μm thick were stained with an automated system (BOND-MAX; Leica Microsystems) using antibodies against programmed death-ligand 1 (PD-L1), PD-1, CD3, CD8, CD68, and pan-cytokeratin AE1/AE3 in a single panel. Slides were imaged using the Vectra 3.0 spectral imaging system (PerkinElmer), and analyzed by the imaging inForm 2.3.1 software (PerkinElmer).

Transcriptomic analysis

The presence of tumor in baseline biopsies was confirmed by a pathologist, and total RNA was extracted using the RNeasy Mini Kit (Qiagen). RNA quality was assessed via bioanalyzer (Agilent) and whole transcriptome analysis was performed on Affymetrix Clariom D Pico microarray by Affymetrix using their standard protocol. The raw CEL files were imported into R statistical software for pre-processing and normalization with Robust Multichip Average algorithm under oligo package. Batches were further normalized and integrated using the ComBat algorithm implemented in inSilicoMerging package.9 Differential gene expression analysis was performed on responders versus non-responders via Student’s t-test in collaboration with the Baylor College of Medicine Multi-Omics Data Analysis Core. Genes were considered as differentially expressed if the p value was <0.05 and the absolute fold change was ≥1.25 (≤1/1.25). Differentially expressed genes were further evaluated for association with survival by comparing the bottom 50% and top 50% with log-rank test. Clustering and heatmaps were generated using Matplotlib, NumPy and SciPy libraries under python. Gene Set Enrichment Analysis (GSEA) using Molecular Signature Database (MSigDB)10 was performed to find enriched pathways (q<0.25).

Responders and non-responders

We segregated patients based on clinical benefit for the mIF and gene expression analyses. Responders: patients with (complete response (CR), N=2), partial response (PR, N=3), or stable disease (SD, N=1) with progression free survival (PFS) ≥6 months, according to RECIST V.1.1. Non responders: patients who had progressive disease (PD, N=9) or SD with PFS <6 months (N=2). The patients with SD in the responders group had regression of targeted lesions that did not meet RECIST defined PR and PFS of 9.4 months. The two patients with SD in the non-responders group did not have tumor regression and had PFS of 3.8 and 5.3 months.

Statistical analysis

Continuous variables were summarized by median and range, while categorical variables were summarized by frequency and percentage; t-test or Mann-Whitney U test was performed to evaluate the difference in continuous variable between patient groups according to variable distribution. Fisher’s exact test was used to evaluate the association between categorical variables. The Kaplan-Meier method was used to calculate PFS, overall survival (OS), and duration of response. The log-rank test was performed to assess the difference of time-to-event outcome among different groups. P values <0.05 are considered statistically significant. Statistical software SAS V.9.4 (SAS), S-Plus 8.2 (TIBCO Software), and R V.3.4.4 were used for all the analyses.

Results

Long-term efficacy outcomes

Baseline characteristics of the 24 patients were described previously.7 The median follow-up time was 46.5 months (95% CI, 46.0 to not reached (NR)) as of data lock July 31, 2020. The median duration of response was 11.2 months (95% CI, 8.51 to NR); At the 3-year follow-up, the three progression-free patients at 18 months remained without progression; two patients, both with CR remain progression-free under surveillance at 46 and 47 months. One patient with PR expired from an intracranial hemorrhage, unrelated to cancer, 40 months from treatment start. Three days prior to his death CT imaging remained stable with one 2.1 cm pulmonary nodule (figure 1A).

Figure 1

Treatment response and survival. (A) Duration of response for each patient (blue bars). The asterisk indicates a patient with response ongoing who died from a non-cancer related cause (intracranial hemorrhage). (B) and (C) are Kaplan-Meier curves for progression free-survival (PFS) and overall survival (OS), respectively. CR, complete response; OPSCC, oropharyngeal squamous cell carcinoma.

The median PFS was 2.66 months (95% CI, 2.5 to 9.4 months, figure 1B) and PFS rates at 2 and 3 years were identical at 12.5% (95% CI, 4.3% to 36%). The median OS was 15.3 months (95% CI, 10.6 to 27.2 months) (figure 1C). The rates of OS at 2 and 3 years were 33% (95% CI, 18.9% to 58.7%) and 12.5% (95% CI, 4.3% to 36%), respectively.

The TME landscape

We evaluated the TME of HPV+ tumors from 17 evaluable baseline biopsies, 6 (35%) of which were from responders and 11 (65%) from non-responders. Combining tumor and stromal compartments, scores for activated T cells ((CD3+PD-1+)+(CD3+CD8+PD-1+)), activated cytotoxic T cells (CD3+CD8+PD-1+), and total macrophages ((CD68+PD-L1−)+(CD68+PD-1+)) were significantly higher among responders (p<0.05)(figure 2A–C). The highest scores of both total and cytotoxic activated T cells were present in the tumors from the two patients with CR, indicating an association with depth of response. There was no correlation of PD-L1 expression in tumor, stroma or combined compartments with response, differing from the correlation we observed with immunohistochemistry.7

Figure 2

Tumor immune microenvironment. (A), (B) and (C): Comparison of immune cell-types quantities at baseline according to treatment response (responders (CR, PR and SD ≥6 months) versus non-responders (PD and SD <6 months)). Each dot represents a patient. CR, complete response; OPSCC, oropharyngeal squamous cell carcinoma; OS, overall survival; PD, progressive disease; PFS, progression free survival; PR, partial response; SD, stable disease.

Differentially expressed genes and enriched biological pathways

Pre-therapy biopsies from 16/24 patients were evaluable for gene expression. Among the 16 specimens, 6 were from responders (37.5%) and 10 were from non-responders (62.5%). Gene expression analysis (figure 3A) revealed differential regulation of 357 genes between responders and non-responders (p<0.05). Of these, 237 genes were upregulated in responders, and 120 genes were downregulated (figure 3A).

Figure 3

Gene expression analysis. (A) Gene Set Enrichment Analysis heatmap. Expression data set sort by correlation with treatment response. Criteria is p=0.05, log2 FC ≥1.25 and max intensity 1. Rows represent genes and columns represent samples. Red, upregulated genes; blue, downregulated genes; white, unchanged gene expression. (B) Five most enriched gene sets of non-responders and responders. CR, complete response; GO, Gene Ontology - biological process gene sets from Molecular Signatures Database (MSigDB); HM, hallmark gene sets from Molecular Signatures Database (MSigDB); NES, normalized enrichment score; PD, progressive disease; PR, partial response; SD, stable disease <6 months, SD*, SD ≥6 months.

GSEA (figure 3B) revealed that the five most highly enriched gene sets among responders were connected to cellular immune response, inflammation response and interferon signaling pathway, consistent with previous literature showing that immunogenic gene expression is correlated with benefit from ICI.11 12 The most highly enriched gene sets in tumors from non-responders were involved in olfactory signaling pathway, which has been associated with tumor proliferation.13 Notably, hypoxia-related genes were upregulated in tumors from responders (figure 3B) in contrast to other studies showing the relationship of hypoxia to immune resistance.14 However, the top seven upregulated genes of the hypoxia gene set set belong to the proteasome group. Proteasomes play a role in degrading proteins that are consequently exposed on the cellular surface and recognized by T cells and have been associated with IFN-γ pathway activation and response to ICI in patients with melanoma.15 Notably, HIF1α, HIF2, and VHL, commonly upregulated in hypoxia, were not enriched in the responders.

The expression of CD68, a pan-macrophage marker, was associated with improved survival (p=0.0058, online supplemental figure 3). This is consistent with the observation that CD68 levels were significantly higher in responders than non-responders (figure 2C). A positive correlation with survival was similarly found for other genes involved in macrophage recruitment or activation or adaptive immune response (CHI3L1, ANGPTL2, TAP1, IL15RA) (online supplemental figure 3). In contrast, eight genes encoding zinc finger proteins (ZFPs) were associated with reduced survival (online supplemental table 1). ZFPs form one of the largest families of transcription factors and function in many biological processes. Recent studies have demonstrated that ZFPs are closely associated with different stages of cancer development, including signal transduction cascades, carcinogenesis, and cell proliferation.16

The TME composition was inferred by xCell, a computational tool to deconvolute cell type proportion using gene expression data. A total of 64 immune and stromal cells were deconvoluted. A significant correlation was found between M0-polarized and M1-polarized macrophage levels and response to treatment (p<0.05), whereas no significant difference was found in M2 macrophage levels between groups (online supplemental figure 4). These observations support the correlation between CD68 density and therapeutic benefit observed in our mIF panel. There were no significant differences of T cell, monocyte, or neutrophil levels between groups (online supplemental figure 4). Lastly, the cytolytic score, calculated from gene expression levels of granzyme A and perforin, positively correlated with response (p=0.0282, online supplemental figure 5).

Discussion

In this updated analysis, the benefit of adding ISA101 to nivolumab remains promising among patients with recurrent or metastatic (R/M) HPV-16+ cancer, with a median OS of 15.3 months and 2-year OS rate of 33%. These results are encouraging when compared with those from contemporaneously performed studies using ICI for HPV+ cancers, especially HPV+ OPSCC, although admittedly our patient population is more heterogeneous and totaled only 24 patients. Checkmate-141, a phase III trial using nivolumab monotherapy showed a median OS (mOS) of 9.1 months and 2-year OS rate of 16.9% among patients with HPV+ OPSCC.17 Although our study showed longer survival, only 79% of our patients were platinum-refractory versus 100% in Checkmate-141. Similarly, KEYNOTE-055, showed an mOS of 8 months in platin and cetuximab-refractory HPV+ OPSCC.18 Recently, preliminary results from a phase Ib/IIa trial with similar design to ours combining MEDI0457, an HPV vaccine plus IL-12, with durvalumab, an anti-PD-L1 antibody, in patients with HPV-16+ or 18+ R/M OPSCC were reported in abstract form.19 Patients were mixed as regards platin-sensitivity and were required to have ≥1 line of prior therapy. ORR was 22% (6/27), and disease control rate at 24 weeks was 26%.19 Finally, triple combination of PDS0101, a liposomal therapeutic vaccine targeting HPV-16 E6/E7, M9241, an immune-cytokine targeting DNA release of from necrotic tumor cells, and bintrafusp alfa, showed promising activity in patients with refractory HPV +solid tumors with an ORR of 71% (9/14), but at the cost of a higher rate of adverse events (grade 3–4=36%).20

CD8+ cytotoxic T cells have a central role in targeting cancer and, thus, are the focus of cancer immune suppression. Elevated T cell density in the TME typical of ‘hot’ cancers has been linked to improved prognosis in many solid tumors, including HPV+ OPSCC21 22 and to higher frequencies of ICI responses.22 We found that higher rates of PD-1+ T cells and PD-1+ cytotoxic T cells significantly correlated with treatment response.23 This correlation between activated T cell infiltration and response to ISA101 plus nivolumab supports the hypothesis that this combination may enhance adaptive antitumor immunity via expansion of HPV-specific T cells in response the ISA101 vaccine coupled with anti-PD-1-mediated protection of these T cells from attenuation in the TME.24

Macrophages and their precursors constitute a heterogeneous and dynamic population of the stroma that can operate across a functional spectrum from pro-inflammatory antigen presentation to tumor-supportive immune suppression. They are classically divided in two main phenotypic poles: the inflammatory/classical M1, associated with Th1 response and anticancer effects, and the tumor-associated macrophage M2, associated with tumor progression through Th2-responses and inhibition of pro-inflammatory T cells.25 Kim et al showed increased infiltration by M1 macrophages in HPV+ OPSCC tumors that were immune rich based on gene expression clustering. Furthermore, these immune rich tumors were correlated with better survival and response to immunotherapy.22 In our study, baseline expression of the pan-macrophage marker CD68 was directly correlated with clinical response and xCell analysis confirmed that the resting M0-macrophage and polarized-M1 macrophage fractions were higher among responders. We speculate that in an environment enriched with resting M0-macrophages, an HPV-16 vaccine may promote macrophage polarity towards the pro-inflammatory M1 phenotype, perhaps driven by pro-inflammatory cytokine production from active, infiltrating E6/E7 specific T cells, favoring tumor destruction.

Our transcriptional profiling of baseline tumor tissue revealed that 357 genes were differently expressed according to treatment response. Among responders, enrichment in gene sets associated with immune response, inflammation and interferon signaling pathway predicted response to treatment, as has been reported previously with ICI.22 23 These data suggest that the putative impact from HPV vaccination may be to qualitatively and quantitatively improve T cell infiltration for the subset of tumors whose TMEs are susceptible to such, as evidenced by baseline inflammation. In contrast, genes associated with olfactory signaling were significantly upregulated among non-responders. Aberrant expression of olfactory-related genes has been reported in some types of solid cancer, such as breast, prostate, and colorectal, where it is associated with tumor development and progression.13 26 For instance, OR7C1 has been shown to play an essential role in maintenance and proliferation of colorectal cells to the extent that it has been proposed as a potential target for colorectal cancer immunotherapy.26

Surprisingly, hypoxia gene sets were upregulated in tumors from responders, specifically the proteasome genes within the set. Constitutive proteasomes are upregulated in several types of malignancy, reflecting their high metabolism and demand for protein degradation. By degrading proteins, proteasomes produce immunogenic peptides that may be presented to APCs through MHC class I. Indeed, overexpression of proteasomes has been correlated with the presence of IFN-γ-secreting tumor-infiltrating lymphocytes and with ICI response.15 Moreover, important genes related to hypoxia, such as HIF1α, HIF2, and VHL, were not enriched in responders. Given the high complexity of gene expression with multiple functions being performed by a single gene, we believe our hypoxia result should be interpreted cautiously since it may be driven by an enrichment of genes playing a different role than hypoxia, such as immune-response and other metabolic pathways.

We are aware that our study has important limitations, mainly relying on the small sample size, the absence of post-treatment specimens and the single-arm design. A randomized phase II trial of ISA101 and PD-1 blocking antibody cemiplimab is ongoing to confirm these findings and further explore correlates of immune response (NCT03669718). If results are promising, ISA will file for Food and Drug Administration registration as accelerated approval.

In conclusion, with a median follow-up of nearly 4 years, ISA101 and nivolumab therapy resulted in durable responses and promising survival in patients with R/M HPV-16+ cancer. The addition of ISA101 may potentially enhance antitumor immunity through generation of E6 and E7 targeted lymphocytes that home to tumor sites and act as a priming agent. Lastly, our data that immune and inflammatory responses were associated with response add to the growing literature regarding HPV+ OPSCC immunoregulation.