Profiling of VEGF receptors and immune checkpoints in Recurrent Respiratory Papillomatosis

Abstract

Objectives:

Recurrent respiratory papilomatosis (RRP) is caused by HPV infection of the aerodigestive tract that significantly impacts quality-of-life including the ability to communicate and breathe. Treatment was traditionally limited to serial ablative procedures in the O.R. with possible local adjuvant therapy, but new systemic therapies, such as VEGF inhibitors, are showing significant promise. This study aims to determine whether rationale exists for combination therapeutic approaches using VEGF inhibitors and/or immune checkpoint blockade.

Methods:

Using fresh specimens from the O.R., we performed flow cytometry on papilloma, normal adjacent tissue, and blood. Papilloma and surrounding tissue were examined for expression of PD-L1, PD-L2, Galectin-9, VEGFR2, and VEGFR3. CD8+ and CD4+ T cells were assayed for expression of PD-1, TIGIT, LAG3, and TIM3.

Results:

Our data shows that papilloma tissue exhibits significantly higher levels of PD-L1 and PD-L2 compared to adjacent tissue. Elevated levels of the VEGF receptor VEGFR3 was also observed in papilloma tissue. When examining T cells within the papilloma, elevated PD-1 and TIGIT expression was observed on CD8+ T cells, while levels of PD-1, TIGIT, and TIM3 were elevated on CD4+ T cells compared to PBMCs. Heterogenous marker expression was observed between individuals.

Conclusions:

Our analysis shows that RRP tissue shows elevated levels of multiple immune check point targets and VEGFR3, with varied patterns unique to each papilloma patient. Some of these immune checkpoint markers already have novel immunotherapies available or in development, providing molecular rationale to offer these systemic treatments to selected patients affected by RRP alongside VEGF inhibitors.

Lay summary:

Recurrent respiratory papillomatosis (RRP) is a relatively rare disease caused by infection of the airway with HPV. RRP manifests in patients as benign tumors in the airway that greatly impact patients lives and is typically managed surgically. In this study, we explore potential targets for medical treatment of RRP.

INTRODUCTION

Recurrent Respiratory Papillomatosis (RRP) is a disease caused by infection of the aerodigestive tract with a low-risk human papilloma virus (HPV), specifically genotypes 6 and 11 ¹. The prevalence of RRP is estimated to be between 1 and 4 per 100,000 as recently as 2010 ². RRP is often split into juvenile onset (<12 years old) and adult-onset disease. Juvenile RRP is most often maternally transmitted through the anogenital tract during delivery, with increased risk factors attributed to the first born of young mothers, and low socioeconomic status ^3,4. The severity of disease is associated with the age of onset and HPV 11, with young patients at diagnosis being more severely affected ^1–4. Adult RRP is associated with adult acquisition of HPV infection, with risk factors including increased number of sexual partners and oral sex. Adult RRP disproportionately affects males, and lower socioeconomic groups ⁵.

Current effective treatment is limited to serial ablative procedures in the operating room to remove the papilloma. Dating back to 1995, there were an estimated 15,000 surgical procedures and estimated cost of $150 million annually for the management of RRP (which is over $250 million in 2019 USD) ⁶. Fortunately, the increasing use of Gardasil, Gardasil9 and related prophylactic vaccines targeting HPV6 and 11 is likely to dramatically reduce the incidence of RRP. Unfortunately vaccination rates have lagged and there is, and will remain, a population of afflicted individuals for the foreseeable future. While the number of individuals impacted by this disease is relatively small, the functional deficit, psychological toll, and economic burden on patients and society at large is significant ⁷. In particular, with aggressive and heavy burden of disease, surgical intervals can be short, and the risk of permanent hoarseness and scarring is high. In a subset of cases, distal spread to the trachea and lungs occurs, which then becomes impossible to manage surgically ⁸.

Recently, systemic therapies have been pursued for those individuals with surgical treatment-resistant disease or distal spread beyond the larynx. Vaccination with Gardasil(9) has not proven effective for treatment, likely because the basal cells harboring infection do not express L1. For cells infected with HPV to be eliminated, it is believed that cellular immunity directed towards the oncoproteins E6/E7 is required, and this is not achieved by prophylactic HPV vaccines such as Gardasil as these vaccines are tailored towards humoral immunity to prevent infection. In the past, systemic medical therapies have been used, including interferon alpha and cidofovir, however, neither provided consistent improvement and have been associated with severe side effects ⁹. A newer approach, systemic bevacizumab (Avastin), has been shown to lengthen symptom-free intervals and decrease the burden of disease, both radiographically and clinically ^10,11. Bevacizumab targets Vascular endothelial growth factor (VEGF), a key player in the angiogiogenesis pathway. RRP is known to have a high degree of neovascularization to support its growth, and this might explain bevacizumab’s demonstrated clinical promise ¹². Understanding the expression pattern of VEGF and its corresponding receptors in papillomatous tissue is important as it may lead to better pre-treatment assessment for viable candidates for bevacizumab therapy. Used individually, bevacizumab has shown promising results in delaying disease recurrence and increasing the interval time between operative procedures, but it is not a cure since with cessation of bevacizumab, the disease recurs ¹⁰.

Recently, novel immune checkpoint inhibitors have become a mainstay of treatment in numerous solid tumor pathologies, with their uses broadening daily. Within papilloma tissue, there exists a strong anti-inflammatory, immunosuppressive microenvironment that acts as a shield to productive immunity while permitting recurrent disease ¹³. Research specifically in RRP has shown that one particular immune checkpoint pathway, programmed cell death-1 (PD-1) and its ligand programmed cell death ligand-1 (PD-L1), have upregulated expression in RRP compared to normal controls ¹⁴. PD-1 is expressed on activated T cells and serves as a pathway to curtail T cell effector function and ablate activation when appropriate ^15,16. Recently, Avelumab, an anti-PD-L1 monoclonal antibody has shown in limited, small clinical trials to have a positive effect on treatment outcomes ¹⁷. In addition to the PD-1/PD-L1 pathway, there are many other immune checkpoint targets currently being explored in a number of different cancers from a basic science and therapeutic standpoint ¹⁸.

Given that targeting the PD-1/PD-L1 and VEGF pathways with modern therapies both show promise in RRP, in our study, we sought to better understand the complex immunologic milieu of RRP to explore the possibility of combinatorial therapy. This involved specifically investigating the profile of immune checkpoint inhibitory ligands, PD-1 pathway and other checkpoint pathways including TIM3, TIGIT, and LAG3 given their potential clinical significance ¹⁸, on papilloma tissue and immune checkpoint inhibitory receptors on papilloma-infiltrating CD4+ and CD8+ T cells from RRP patients. In addition, the expression of VEGF-receptors on papilloma was explored. We hope that by better understanding these pathways in RRP, therapeutic targets can be identified and targeted alongside systemic bevacizumab and surgical resection to improve the management of RRP.

MATERIALS AND METHODS

Patient samples

Twelve patients with biopsy-proven, HPV-6/11 RRP were selected. Each patient underwent signed informed consents, which were approved by the Johns Hopkins Institutional Review Board. Patient demographics are summarized in Table 1. Disease severity was based on the Derkay score classification, which measures both the patient’s symptoms and amount of corresponding papilloma in different laryngeal subsites. This scoring system is well validated and been used in prior clinical trials^17,19. In the operating room, fresh papilloma specimens were taken from the upper aerodigestive tract or proximal trachea. Additionally, a small amount of adjacent non-papillomatous control tissue from the same patient was collected. Simultaneously, peripheral blood was taken from the patient in the operating room. These fresh specimens were then immediately transported to the laboratory, where sample prep and staining for flow cytometry was performed.

Table 1.

Patient demographics.

Variable	Patients, n = 12
Age (years)
Mean (SD)	44 (14)
Range	13 – 64
Gender
Male (%)	6 (50%)
Female (%)	6 (50%)
Disease Onset
Juvenile (<12 yrs)	6 (50%)
Adult	6 (50%)
Disease Severity
Mild	0 (0%)
Moderate	4 (33%)
Severe	8 (66%)

Tissue processing

Upon sample retrieval from the operating room, papilloma tissue and normal adjacent control were washed extensively to remove any blood. Tissue was then cut into smaller pieces using scissors. Samples were then added to tissue dissociation buffer containing collagenase I, collagenase IV, and DNase1. Following 20 minutes of incubation, homogenization was done on a gentleMACS dissociator (human tumor protocol per manufacturer instructions) using C Tubes to obtain single cell suspensions. Samples were then treated with RBC lysis buffer, washed extensively, counted, and plated in equal numbers for flow cytometric staining.

PBMC isolation

PBMCs collected in the operating room were brought to the lab and centrifuged at 500 x g for 10 minutes. Buffy coat layer was collected and treated with RBC lysis buffer for 10 minutes. Samples were then washed extensively with FACS buffer (0.5% BSA/PBS), counted, and plated in equal numbers for flow cytometric staining.

Flow cytometry

Multi-dimensional flow cytometry analysis was performed on the indicated samples. Cells were first incubated with zombie aqua live dead and Fc blocking antibodies for 20 minutes at 4 degrees (BioLegend and performed per manufacturer instructions). Cells were then stained with the antibodies indicated in Supplementary Tab. 1 as a cocktail for 30 minutes at 4 degrees. After extensive washing, cells were acquired on a Beckman Coulter CytoFLEX S machine equipped with a plate loader.

Statistical analysis

All data are expressed as means ± 95% confidence interval (CI). Results between papilloma or normal adjacent and papilloma-infiltrating or PBMC were evaluated by unpaired t test. All P values < 0.05 were considered significant. Of note, *, **, and *** indicate P values less than 0.05, 0.01, and 0.001, respectively; N.S., not significant. Specific n is equal throughout all figures. All statistical calculations were performed in GraphPad Prism 9. All flow analysis, including tSNE was performed in FlowJo version 10.

RESULTS

Papilloma tissue expresses elevated levels of VEGFR3 and B7 family members PD-L1 and PD-L2

To better understand the immunosuppressive microenvironment in RRP, we began by analyzing the expression of inhibitory molecules on the papilloma cells. Expression was compared to that on normal adjacent tissue taken from a healthy appearing portion of the upper aerodigestive tract. To identify cells making up the papilloma or normal adjacent epithelial tissue verses immune infiltrating cells, flow cytometric gating was performed (Supplementary Fig. 1). Briefly, after double discrimination, CD45- live cells (determined by viability dye exclusion) were gated for forward and side scatter parameters, and then analyzed for expression of suppressive markers including PD-L1, PD-L2, and Galectin-9. (Fig. 1a–f). When looking at B7 family members, we observed elevated levels of PD-L1 and PD-L2 on papilloma tissue (Fig. 1a,b). This finding is consistent with published literature and supports the argument that PD-L1 and PD-L2 on papilloma tissue could in theory engage with markers of T cell exhaustion on T cells infiltrating the papilloma tissue and limit cytotoxic activity. We did not observe statistically significant levels of Galectin-9 on papilloma tissue (Fig. 1e,f). We also explored the expression of VEGF receptors VEGFR2 and VEGFR3 on these tissues. While no statistically significant expression of VEGFR2 was observed (Fig. 1g,h), we did observe statistically significant elevated levels of VEGFR3 on papilloma tissue compared to normal adjacent (Fig. 1i,j). We believe that this could suggest a role of local VEGF expression in the papilloma directly engaging with VEGFR3 on papilloma cells and modulating vascular remodeling pathways, but this would need to be empirically explored. In addition, given the multitude of markers we stained on our papilloma cells, we performed dimensionality reduction using t-distributed stochastic neighbor embedding (tSNE). To assist in visualizing the data, tSNE plots were colored according to expression of various markers including PD-L1 (Fig. 1k), PD-L2 (Fig. 1l), Galectin-9 (Supplementary Fig. 2a), VEGFR2 (Supplementary Fig. 2b), and VEGFR3 (Supplementary Fig. 2c). This analysis revealed that while overlapping, the clusters of cells expressing each of these inhibitory or angiogenesis-associated ligands was distinguishable, and a possible avenue for future studies.

Figure 1. Papilloma tissue expresses elevated levels of VEGFR3 and B7 family members PD-L1 and PD-L2.

To evaluate the levels of VEGF receptors and immune checkpoint ligands on papilloma tissue, single cells were prepared. Samples were stained with an antibody cocktail including CD45 for immune cell dumping and a viability dye and analyzed by flow cytometry. Representative histogram of expression normalized to mode in papilloma tissue (red) or normal adjacent tissue (purple) of PD-L1 (a), PD-L2 (c), Galectin-9 (e), VEGFR2 (g), and VEGFR3 (i). Bar graph showing expression across patients (individual points) as well as median expression of PD-L1 (b), PD-L2 (d), Galectin-9 (f), VEGFR2 (h), and VEGFR3 (j). tSNE plots showing the heterogeneity in expression of PD-L1 (k) and PD-L2 (l) in normal adjacent (top row) and papilloma tissue (bottom row) in participants (rows).

Profile of T cells infiltrating recurrent respiratory papilloma tissue

Having profiled the papilloma cells themselves, we then characterized the cytotoxic T cells infiltrating the papilloma. We compared the expression of various relevant immunologic markers of residence and exhaustion on papilloma infiltrating cytotoxic T cells to that on PBMCs (peripheral blood mononuclear cells). This comparison was done to validate a truly unique tumor microenvironment in comparison to the immunologic milieu of circulating blood. It was logistically not possible to compare immune cells in the papilloma to those in the lesion-adjacent tissue by flow cytometry due to the extremely low number of T cells in the small piece of healthy mucosa that we collected. To identify these immune cells, flow cytometric gating was performed including doublet discrimination, selection of live CD45+ cells, scatter gating, and identification of cytotoxic T cells by CD3 and CD8 expression (Supplementary Fig. 3). Strikingly, there was a significant upregulation in levels of PD-1 (Fig. 2a,b) and TIGIT (Fig. 2c,d) on CD8+ T cells infiltrating the papilloma tissue compared to PBMCs. We did not observe a significant upregulation in expression of LAG3 (Fig. 2e,f) or TIM3 (Fig. 2g,h) on papilloma-infiltrating cytotoxic T cells. Importantly, the CD8+ T cells we analyzed from the papilloma expressed high levels of CD69 (Fig. 2i,j), a marker of activation and tissue residence. This supports that the CD8+ T cells were resident in the papilloma tissue. In addition to characterizing CD8+ T cells, we also characterized expression of exhaustion markers on CD4+ T cells. CD4+ T cells were identified using the same strategy discussed above and shown (Supplementary Fig. 3). Similar to the CD8+ T cells, CD4+ T cells in the papilloma expressed statistically significantly higher levels of PD-1 (Fig. 3a,b) and TIGIT (Fig. 3c,b). LAG3 expression was also not elevated on CD4+ T cells (Fig. 3e,f). However, in contrast to CD8+ T cells, a portion of CD4+ T cells did express statistically higher levels of TIM3 (Fig. 3g,h). Again, as was seen for CD8+ T cells, the CD4+ T cells we examined also expressed high levels of CD69, suggesting activation and tissue residence (Fig. 3i,j). Finally, we also performed dimensionality reduction using tSNE for the T cells infiltrating the papilloma. We did not split CD4+ and CD8+ T cells for this analysis given the small number of infiltrating immune cells in some of our tissue samples. tSNE plots were colored by either PD-1 (Supplementary Fig. 4a), TIGIT (Supplementary Fig. 4b), LAG3 (Supplementary Fig. 4c), TIM3 (Supplementary Fig. 4d), or CD69 (Supplementary Fig. 4e) expression. From this analysis, it appeared that there was overlap in expression of checkpoint markers in some clusters, with PD-1 and CD69 being highly associated.

Figure 2. Expression of immune checkpoint markers on papilloma-infiltrating CD8+ T cells.

After single cell suspensions were prepared and appropriate gating was performed to identify CD8+ T cells, expression of immune exhaustion markers and CD69 was evaluated. Representative histogram of expression normalized to mode in papilloma-infiltrating CD8+ T cells (red) or CD8+ T cells from PBMCs (purple) of PD-1 (a), TIGIT (c), LAG3 (e), TIM3 (g), and CD69 (i). Bar graph showing expression (individual points) as well as median expression across patients of PD-1 (b), TIGIT (d), LAG3 (f), TIM3 (h), and CD69 (j).

Figure 3. Immune checkpoint marker expression on CD4+ T cells.

As with CD8+ T cells, CD4+ T cells were identified by appropriate gating and expression of immune exhaustion markers and CD69 was evaluated. Representative histogram of expression normalized to mode in papilloma-infiltrating CD4+ T cells (red) or CD4+ T cells from PBMCs (purple) of PD-1 (a), TIGIT (c), LAG3 (e), TIM3 (g), and CD69 (i). Bar graph showing expression (individual points) as well as median expression across patients of PD-1 (b), TIGIT (d), LAG3 (f), TIM3 (h), and CD69 (j).

Correlation of immunologic profile with disease severity

Across essentially all markers analyzed, there was significant heterogeneity and significant patient-to-patient variability. This suggested that each individual patient’s immunologic profile was unique, and we sought to determine if any patterns of immunologic activation / suppression were associated with disease severity. Across all markers analyzed, there was no significant differences in immunologic markers between patients with moderate versus severe disease.

DISCUSSION

Overall, our study demonstrated that there is statistically significant upregulation of multiple immune checkpoint pathways on T cells and corresponding ligands on papilloma tissue in the papilloma microenvironment. When exploring the papilloma tissue, elevated levels of PD-L1 and PD-L2 were observed. In addition, VEGFR3 expression was also higher on papilloma tissue than normal adjacent. An important aspect of this data to consider is that while a statistically significant upregulation was observed for a number of markers, not every individual patient exhibited high expression of these markers on their papilloma cells. In fact, between patients, a wide spread of expression was seen, suggesting heterogeneity for expression of these markers in recurrent respiratory papillomatosis. When examining the CD8+ T cells, elevated exhaustion-associated PD-1 and TIGIT was observed, while elevated PD-1, TIGIT, and TIM3 was seen on CD4+ T cells. As with the papilloma tissue, between individuals, there was a wide range of expression. Some individuals expressed high levels, while other expressed levels no different than the baseline circulating levels. Even within an individual’s papilloma infiltrating T cells, a range of expression was observed. We explored whether this wide range of expression levels for each marker between patients could be explained by disease severity, but we could not detect any statistically significant correlations. It is likely that with our relatively small sample size, we do not have enough participants to detect differences by disease severity, and this could be an area for us to explore in future studies with more participants. In addition, stratifying these inidividuals by viral genotype and anatomic distribution could also help tease apart differences in future studies.

There are some important limitations to our data - while we examined the papilloma as a whole, it is important to consider that there are many different cell populations making up the papilloma microenvironment. Each of these cell types likely have their own set of factors that contribute to the suppressive papilloma environment. In future studies, we aim to sort different cell lineages such as epithelial, endothelial, fibroblast, and stem-like cells within the papilloma, and characterize their expression profiles. This would be particularly interesting when considering the VEGFR3 expression, which cells are responsible for its overexpression, and how that could be electively modulated therapeutically. We would propose expanding analysis of different cell populations in the papilloma using single cell RNAseq, so that more transcriptomic information could be acquired and VEGFR3 as well as other angiogenic contributors could be interrogated as it is a large family of proteins with many potential players. These data could then be validated using flow cytometry.

Similarly, while we explored the profile of papilloma-infiltrating T cells, we did not perform analysis of regulatory T cells or other immune cells in the papilloma tissue. Given the crucial role of humoral immunity in prevention HPV infection, it would be interesting to assay the profile of B cells in recurrent papilloma tissue, their activation status, and how they could be contributing to the suppressive environment. Beyond this, innate immune cells including myeloid-derived suppressor cells, macrophage subsets, dendritic cells, and others could be explored given their known impact on immune responses

Unsurprisingly perhaps, we have found that individuals with RRP manifest significant disease heterogeneity and individualized immunologic profile diversity. These individual patterns could be reflective of differential host epithelium / virus interactions that can represent different disease severity phenotypes ²⁰. We don’t yet understand why any given patient’s immune cells may preferentially upregulate PD-1 and another may upregulate TIM3, or what regulates the dynamic interplay of corresponding inhibitory ligands on the papilloma tissues as well as other factors such as VEGF receptor expression. In future work, collecting sufficient quantities of normal adjacent tissue would be cricual in understanding these immunological nuances in individuals, as it is hard to compare tissue derived (papilloma or normal adjacent) vs peripherial T cells phenotypically, which is an additional limitation of our current study.

We hope that in the future, we can be to the point where effective systemic therapy could be delivered with a more personalized approach, as more and more specific inhibitors of immunologic pathways are developed and tested. Each individual patient could have an assessment of their immune phenotype, and therapy targeted towards their own specific immunosuppressive environment. This approach to systemic therapy will likely fall along the lines of a multi-dimensional immunologic profiling for each patient – analogous to the “tumor proportion score” used in head and neck cancer immunotherapy – where treatments can be tailored on an individualized basis and not a “one size fits all” treatment strategy. This targeted approach is important also as these therapies can come with significant side effects and costs. We hope that by considering each papilloma patient individually, prognosis can be improved, surgical interventions and side effects can be limited, and patient quality of life can be improved. While we are not currently to this point in RRP, we hope that with our current and future studies, as well as with the work from other groups, we can one day move the field in the direction of targeted combinaton therapy.

CONCLUSIONS

Our analysis shows that recurrent respiratory papilloma tissue shows elevated levels of immune check point targets and VEGFR3, with varied patterns unique to each papilloma patient. Some of these immune checkpoint inhibitors already have novel immunotherapies available or in development, and there is justifiable molecular rationale to examine these systemic treatments for benefit to patients affected by RRP alongside VEGF inhibitors.

Supplementary Material

Supinfo

1. Supplementary Table 1. Flow cytometry antibody list.

2. Supplementary Figure 1. Flow gating strategy for identification of papilloma cells within papilloma tissue.

3. Supplementary Figure 2. tSNE plots generated from papilloma cells.

4. Supplementary Figure 3. Flow gating strategy for identification of T cells infiltrating papilloma tissue.

5. Supplementary Figure 4. tSNE plots generated from papilloma infiltrating T cells.