- Email: [email protected]
Defining an inflamed tumor immunophenotype in recurrent, metastatic squamous cell carcinoma of the head and neck
Oral Oncology 67 (2017) 61–69
Contents lists available at ScienceDirect
Oral Oncology journal homepage: www.elsevier.com/locate/oraloncology
Defining an inflamed tumor immunophenotype in recurrent, metastatic squamous cell carcinoma of the head and neck Glenn J. Hanna a, Hongye Liu a, Robert E. Jones a,b, Alyssa F. Bacay a, Patrick H. Lizotte a,b, Elena V. Ivanova b, Mark A. Bittinger a,b, Megan E. Cavanaugh a,b, Amanda J. Rode a,b, Jonathan D. Schoenfeld c, Nicole G. Chau a, Robert I. Haddad a, Jochen H. Lorch a, Kwok-Kin Wong a,b, Ravindra Uppaluri a,d,⇑, Peter S. Hammerman a a
Department of Medical Oncology, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA 02215, USA Robert and Renee Belfer Center for Applied Cancer Science, Dana-Farber Cancer Institute, 360 Longwood Avenue, Boston, MA 02215, USA Department of Radiation Oncology, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA 02215, USA d Division of Otolaryngology-Head & Neck Surgery, Brigham & Women’s Hospital, 75 Francis Street, Boston, MA 02215, USA b c
a r t i c l e
i n f o
Article history: Received 21 December 2016 Received in revised form 24 January 2017 Accepted 7 February 2017
Keywords: Head and neck cancer Immunotherapy Biomarkers PD-1 Survival
a b s t r a c t Objectives: Immune checkpoint inhibitors have demonstrated clinical benefit in recurrent, metastatic (R/ M) squamous cell carcinoma of the head and neck (SSCHN), but lacking are biomarkers that predict response. We sought to define an inflamed tumor immunophenotype in this R/M SCCHN population and correlate immune metrics with clinical parameters and survival. Methods: Tumor samples were prospectively acquired from 34 patients to perform multiparametric flow cytometry and multidimensional clustering analysis integrated with next-generation sequencing data, clinical parameters and outcomes. Results: We identified an inflamed subgroup of tumors with prominent CD8+ T cell infiltrates and high PD-1/TIM3 co-expression independent of clinical variables, with improved survival compared with a non-inflamed subgroup (median overall survival 84.0 vs. 13.0 months, p = 0.004). The non-inflamed subgroup demonstrated low CD8+ T cells, low PD-1/TIM3 co-expression, and higher Tregs. Overall nonsynonymous mutational burden did not correlate with response to PD-1 blockade in a subset of patients. Conclusion: R/M SCCHN patients with an inflamed tumor immunophenotype demonstrate improved survival. Further prospective studies are needed to validate these findings and explore the use of immunophenotype to guide patient selection for immunotherapeutic approaches. Ó 2017 Elsevier Ltd. All rights reserved.
Introduction Even with current treatments and in the era of favorable prognosis human papillomavirus (HPV)-associated disease, there remains a significant risk of locoregional recurrence or distant metastases in squamous cell carcinoma of the head and neck (SCCHN) [1,2]. Moreover, 5-year survival in patients with recurrent or metastatic (R/M) SCCHN remains dismal [3]. In recent years, it has become evident that tumor progression is promoted by immune evasion and abrogation of an effective immune response against cancer cells [4]. SCCHN appears closely linked with immunosuppression, and patients often demonstrate impaired immune cell function that correlates with poor outcomes [5,6]. Mechanisms of tumor immune evasion in SCCHN include the ⇑ Corresponding author at: Harvard Medical School, Dana-Farber Cancer Institute, 450 Brookline Ave., Boston, MA 02115, USA. E-mail address: [email protected] (R. Uppaluri). http://dx.doi.org/10.1016/j.oraloncology.2017.02.005 1368-8375/Ó 2017 Elsevier Ltd. All rights reserved.
development of T cell tolerance, modulation of inflammatory and angiogenic cytokines, downregulation of antigen-processing machinery, and the expression of immune checkpoint ligands or receptors to promote immune evasion [7–9]. These mechanisms are currently serving to define immunotherapy targets for clinical development. Immune checkpoint receptors inhibit normal T cell activation and costimulation to maintain a homeostatic immune response [10]. Cytotoxic T lymphocyte-associated protein 4 (CTLA-4) and programmed death-1 (PD-1) are expressed on the surface of immune cells and interact with their respective ligands on antigen presenting or tumor cells. High tumor expression of the ligands of PD-1 (PD-L1, L2) and/or PD-1 expression by T lymphocytes can attenuate T cell activation and drive T cell exhaustion favoring tumor immune evasion [11]. Studies have estimated PD-L1 expression in SCCHN at 30–70% [12,13] with HPV+ tumors more frequently harboring infiltrating immune cells that express PD-1, despite its favorable prognosis [14]. Two recently published trials
62
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
have established the efficacy of PD-1 blockade in advanced SCCHN. Anti-PD-1 blockade with pembrolizumab in a heavily pre-treated SCCHN population that was PD-L1+( 1% by immunohistochemistry [IHC] in tumor or immune cells) resulted in an overall response rate of 18%, and 25% among those who were HPV+ [15]. The CheckMate-141 trial in platinum-refractory SCCHN using the anti-PD-1 antibody nivolumab demonstrated an improvement in median overall survival (OS) compared with standard chemotherapy – which was greater in those with HPV+ disease and in those with 1% tumoral PD-L1 expression by IHC [16]. These data were the foundation for Food and Drug Administration approval of these agents in 2016. These studies illustrate that response rates to antiPD-1 monotherapy in SCCHN patients are low. Additionally, there is significant cost and toxicity associated with these agents, together highlighting the critical need to identify predictive biomarkers or immunophenotypes to guide patient selection. The tumor immune microenvironment is built on spatially diverse and complex immune relationships that appear dynamic. Thus alternative immune checkpoint upregulation by T cells represents just one mechanism to explain why capturing PD-1/PD-L1 interactions alone may not predict response to inhibitors [17]. Recent work in other solid tumors has suggested two main tumor immunophenotypes: an inflamed tumor category comprised of a rich T cell infiltrate, a type 1 interferon signature, and a diverse chemokine profile, and a non-inflamed category that lacks these features [18]. These so-called inflamed tumors may respond more favorably to therapies targeting immune checkpoint mechanisms. Here we prospectively characterized the tumor immune microenvironment using multiparametric flow cytometry and a multidimensional clustering algorithm to define an inflamed phenotype in R/M SCCHN and correlated these findings with mutational burden, smoking and HPV status, and survival. Our observations strengthen our understanding of the tumor immune microenvironment specific to SCCHN. Along with advances in genomic biomarkers, immunophenotyping may serve to guide the selection and sequencing of immunotherapies, and identify candidate biomarkers to predict response to checkpoint inhibitors.
The surface antibody against CD45 (2D1) was purchased from eBioscience and the surface antibody against LAG-3 (polyclonal) and its isotype control (polyclonal) were purchased from R&D Systems. The intracellular antibody FOXP3 (236A/E7) was purchased from eBioscience. Cells were analyzed within 72 h of fixation on a BD FACSCanto II HTS or BD Fortessa cell analyzer with FACSDiva software v8.0.1 (BD Biosciences) and analyzed using FlowJo software v10. Massively parallel genomic sequencing Targeted next-generation sequencing was performed on DNA extracted from formalin-fixed paraffin-embedded (FFPE) archival tumor samples. Total gDNA concentration was determined using the PicoGreen dsDNA quantification assay (ThermoFisher Scientific, Waltham, MA), and samples with sufficient starting material (> 100 ng) were taken forward to library construction [20]. Briefly, gDNA was fragmented by ultrasonication to 150 base pairs and purified. Size-selected DNA was ligated to specific adapters during library construction (Illumina, Inc., San Diego, CA). Libraries were quantified using qPCR (Kapa Biosystems, Inc., Woburn, MA) and captured using the OncoPanel_v2 bait set using the Agilent SureSelect hybrid capture kit (Agilent Technologies, Santa Clara, CA). Oncopanel_v2 consists of 504 genes and 15 intronic regions with known or potential importance in cancer. The captured libraries then underwent paired end 100 (2 100 nt) sequencing on a Hiseq 2500 (Illumina Inc.). Read pairs were aligned to the reference sequence b37 edition from the Human Genome Reference Consortium using the Burrows-Wheeler Aligner [21] and de-multiplexed using Picard tools. Mutation analysis for single nucleotide variants and nonsynonymous mutational burden was performed using MuTect v1.1.4 in paired mode for tumors with matched germline or single mode for samples without matched germline, and annotated using Oncotator [22]. The alignments were further refined using the GATK tool for localized realignment around indel sites [23]. As part of quality metrics, all samples had to meet a minimum requirement that 80% of targets have a minimum coverage of 30.
Materials and methods Statistical analysis Subjects Patients with SCCHN undergoing biopsy to confirm metastatic disease and those undergoing surgical resection for locoregional recurrence were identified prospectively. Prior to biopsy, informed consent was obtained for institutional review board-approved protocols. Peripheral blood controls were obtained at the time of biopsy in a subset of patients. Patient demographics and clinical characteristics were recorded. Multiparametric flow cytometry At the time of biopsy or surgical resection, fresh tissue samples from each patient were placed in RPMI-1640 with 10% fetal bovine serum (FBS). Tumor was confirmed by hematoxylin and eosin staining. A non-fixed cell suspension was then prepared for staining with fluorescently-conjugated antibody cocktails, as previously described [19]. Single cell suspensions were stained using mouseanti-human antibodies. Surface antibodies against CD3 (HIT3a; UCHT1), CD8 (RPA-T8), CD14 (M5E2; MphiP9), CD45 (HI30), CD56 (B159), CD279 (EH12.1) and its isotype control (MOPC-21), and HLA-DR (G46-6) were purchased from BD Biosciences. Surface antibodies against CD4 (RPA-T4), CD16 (3G8), CD19 (HIB19), CD33 (WM53), CD66b (G10F5), CD123 (6H6), and TIM-3 (F38-2E2) and its isotype control (MOPC-21) were purchased from BioLegend.
Fisher’s exact test for categorical variables (Wilcoxon rank sum test for continuous variables) and one-way ANOVA on ranks was used to assess for differences between individual immune subgroups. Spearman’s Rho was used to measure the strength of association between variables. All statistical tests used a significance of <0.05 and were two-sided. An unsupervised non-linear dimension reduction method, t-Distributed Stochastic Neighbor Embedding (t-SNE), was used to investigate in reduced dimension space how tumors locate in relation to one another based on multiparametric flow data – using a tree-based algorithm [24]. With n = 20 samples included in t-SNE, the power to differentiate at least 8 markers was 0.8. Overall survival (OS) was determined from the date of diagnosis to death from any cause, otherwise this was censored at date of last follow-up. Time to recurrence (TTR) was determined from the date of completion of initial treatment to recurrence or metastatic disease – and Kaplan-Meier statistics were applied. Data were analyzed using R software package (version 2.15.3) [25]. Results Clinical characteristics of the cohort Demographics, clinical characteristics and survival information from 34 patients are summarized in Table 1. With a median age of
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69 Table 1 Demographics, clinical characteristics and survival information in patients with R/M SCCHN. Characteristic
(%)a, n = 34
Age (median, y)
57.5 (26–84)
Gender Male Female
23 (67.6) 11 (32.4)
Smoking status Never or <10 pack-year Former (10 pack-year) Current
14 (41.1) 16 (47.1) 4 (11.8)
Primary site of disease Oral cavity Oropharynx Nasopharynx Larynx Cutaneous Unknown
12 (35.3) 8 (23.5) 4 (11.8) 5 (14.7) 3 (8.8) 2 (5.9)
Initial staging at diagnosis Stage I, II Stage III, IV
11 (32.4) 23 (67.6)
HPV statusb HPV+ HPV-
10 (29.4) 24 (70.6)
Initial treatment regimen Surgery Surgery + radiation Surgery + CRT CRT IC + CRT (sequential) Targeted sequencing performed TP53 mutation NOTCH mutation PIK3CA mutation
6 (17.6) 4 (11.8) 4 (11.8) 13 (38.2) 7 (20.6) 20 (58.8) 9 (45.0) 7 (35.0) 5 (20.0)
Site of recurrence Locoregional Distant
24 (70.6) 10 (29.4)
Subsequent treatment with immune checkpoint therapy
9 (26.5)
Overall survival (median, in months) HPV+ HPV# died Median TTR (mos)
22.0 (3.0–202.0) 28.5 (10.0–87.0) 21.0 (3.0–202.0) 17 (50.0) 9.0 (1.0–187.0)
a
Except for age, survival and time to recurrence. HPV status was tested in all oropharyngeal and unknown primaries and assumed to be negative at other sites; HPV = human papillomavirus, CRT = concurrent chemoradiation, IC = induction chemotherapy, TTR = time to recurrence. b
57.5 (26–84 years) at diagnosis, this predominantly male cohort included >50% former or current smokers and 29.4% of patients had HPV+ disease. Most patients presented with locoregional involvement and were initially treated with multimodality therapy. Locoregional recurrences accounted for 70.6% of patients (median TTR of 9.0 months). Seventeen (50.0%) patients in the cohort had died at the time of analysis (median follow-up was 9.0 months). Following biopsy, 9 (26.5%) patients went on to be treated with immune checkpoint blockade. Tumor immune composition of the cohort Multiparametric flow cytometry performed on tumor samples demonstrated variable cell viability (median: 18.7% [5.1, 61.2]) with a predominant leukocyte population composed of mostly granulocytes among individual tumors (Fig. 1A). We identified a CD3+ T cell infiltrate (median: 18.0% [3.5, 39.4] of leukocytes) in the majority of tumors, with a median of 40.3% [34.7, 54.2] of T cells showing CD8 positivity and 50.1% [35.6, 60.6] of T cells expressing CD4 (Fig. 1B) resulting in a median CD8/4+T cell ratio of 1.0 [0.6, 1.5]. A high degree of immune cell heterogeneity was
63
observed when considering key clinicopathologic features of this cohort: there was no difference in CD8+ T cell abundance based on primary disease site (p = 0.283), smoking history (p = 0.570), or HPV status (p = 0.119). Matched peripheral blood samples were available in a small subset of patients (n = 5) and were composed of few CD3+ T cells (median: 1.9%, CD8/4+ T cell ratio: 0.5). Defining tumor immunophenotypes Next we utilized the t-distributed stochastic neighbor embedding (t-SNE) algorithm [24] to analyze tumor immunophenotypic data obtained from multiparametric flow cytometry. The t-SNE algorithm is a tool to visualize high-dimensional data, whereby each datapoint is plotted on a two-dimensional (2D) map. Fig. 2A–D shows a 2D spatial map and plots individual tumor samples (n = 23) at a given location based on comprehensive immunophenotypic data obtained by flow cytometry. This map spatially depicts the similarity between the overall tumor immune composition of each sample (using a minimum of 9 immune parameters), in an unbiased manner – whereby tumors with a similar overall immune profile locate closer together on the map. The map is then interrogated for a given immune parameter(s) among individual tumor samples to understand expression levels of key immune markers. Fig. 2A–D shows the same spatial map, with each panel highlighting a single immune population of interest that defined three tumor clusters: one with a high CD4+(low CD8 +) T cell infiltrate with low co-expression of PD-1/TIM3 checkpoints (CP) on CD8+ T cells (CD8lowCPlow). Another subgroup with a high CD8+ T cell population and high co-expression of PD-1/ TIM3 (CD8highCPhigh). Finally, a third subgroup was comprised of an intermediate CD8+ T cell population with variable coexpression of PD-1/TIM3 checkpoint receptors (CD8intCPint). Fig. 2E shows flow cytometry contour plots characterizing key immune cell populations corresponding to each of the three defined tumor clusters shown on the 2D spatial map. Defining an inflamed tumor phenotype We next sought to investigate both immune cell and immune checkpoint expression patterns among the 3 tumor clusters. Fig. 3A depicts immune cell composition among each tumor cluster, showing a higher CD3+ T cell abundance among CD8highCPhigh tumors (p < 0.001), but there was no significant difference in absolute T cell number between each of the subgroups (p = 0.109) – suggesting heterogeneity among tumor leukocyte populations. The CD8highCPhigh tumor subgroup demonstrated the greatest CD8+ T cell population among the 3 immune clusters (median: 75.1%, p = 0.008), but with >50% median CD8+ T cell abundance among the CD8intCPint group as well (Fig. 3B). The CD8lowCPlow subgroup had a greater infiltrate of FOXP3+ CD25+ T regulatory cells (Treg) (median: 22.6%, p = 0.026), although CD8+/Treg ratios were similar among all 3 tumor clusters (p = 0.528). CD8highCPhigh tumors demonstrated a greater population of PD-L1+ and PD-L2+ tumor-infiltrating monocytes compared with CD8lowCPlow tumors (Fig. 3C, p = 0.01 and p = 0.04, respectively), with both CD8highCPhigh and CD8intCPint subgroups showing >50% PD-L1 expression among monocytes. However, PD-L1/2 expression on tumor cells was not significantly different between tumor clusters despite a trend towards greater tumoral PD-L1 expression among CD8highCPhigh tumors. PD-1+ CD8+ T cell expression correlated with PD-L1 expression on monocytes among all 3 tumor clusters (R = 0.46, p = 0.014), but PD-L2 status did not. The CD8highCPhigh subgroup also had higher PD-1 expression and PD-1/TIM-3 co-expression among CD8+ T cells (median: 91.5%, 43.4%; p < 0.001, p = 0.006, respectively), but again these values were higher in the CD8intCPint compared with the
64
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
Fig. 1. Clinicopathologic features coupled with multiparametric flow cytometry data from 34 patients with R/M SCCHN showing the (A) proportion of immune cells relative to the CD45+ leukocyte population in each tumor sample, and (B) the proportion of CD4+ and CD8+ T cells relative to the CD3+ T cell population in each tumor sample. Color tiles above indicate smoking status and a plus sign indicates human papillomavirus (HPV)-associated disease. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
CD8lowCPlow subgroup (Fig. 3D). Given the higher CD8+ T cell infiltrates and similar checkpoint co-expression patterns among the CD8highCPhigh and CD8intCPint groups, these two clusters were combined to represent an inflamed tumor immunophenotype. Fig. 4 integrates the CD8highCPhigh and CD8intCPint groups (so-called inflamed phenotypes) and compares CD8+ T cell subsets, markers of T cell activation, inhibition and proliferation with that of the CD8lowCPlow or non-inflamed phenotype. While CD8+ T cells were more abundant in inflamed tumors, T cell subsets were similar (Fig. 4A). The majority of CD8+ T cells among inflamed tumors showed evidence of activation (median of 81.4% were CD38+, 76.9% were CD69+, Fig. 4B). Inflamed tumors exhibited significantly greater PD-1+ CD8+ T cell presence and more often co-expressed PD-1/TIM3 on CD8+ T cells (p < 0.001, Fig. 4C). CTLA-4 expression on CD8+ T cells was low among all immunophenotypes, but PD-1/CTLA-4 co-expression on CD4+ T cells was increased in non-inflamed tumors (p = 0.019, Fig. 4D).
number of smokers and HPV+ tumors between these cohorts (p = 0.204 and p = 0.203, respectively). Similarly, CD8+ T cell abundance correlated with OS (R = 0.41, p = 0.023) for the entire cohort (Fig. 5B), but PD-L1+ status did not (Fig. 5C). Nine (26.5%) of 34 patients were subsequently treated with single agent PD-1 blockade following biopsy. Four of these patients had inflamed tumor phenotypes and 5 had non-inflamed tumors. Median OS was 9.5 months from the time of starting PD-1 blockade (with a median OS of 54.0 months from the date of diagnosis) among the inflamed phenotype patients who received PD-1 inhibition, compared with 2.0 months (median OS of 14.0 months from diagnosis) among the non-inflamed patients (p = 0.026). PD-L1+ monocytes were also higher among the inflamed phenotype patients (range 30.0– 76.2%) treated with checkpoint blockade.
Survival outcomes by tumor immunophenotype
When considering clinical features, multidimensional clustering analysis showed no clear pattern based on smoking or HPV status – although oral cavity and laryngeal tumors tended to cluster within the non-inflamed subgroup (n = 7 of 10). There was no significant difference in CD8+ T cell infiltrates by HPV status (p = 0.119) and similarly there was no difference in co-expression of PD-1/TIM3 on T cells when separated by smoking (p = 0.854) or HPV status (p = 0.312). When integrating clinical outcomes with genomic
When the CD8highCPhigh and CD8intCPint tumor samples (inflamed phenotypes) with >50% CD8+ T cell infiltrates and >50% PD-1/TIM3+ CD8+ T cell co-expression were combined, there was significantly improved median OS vs. the non-inflamed subgroup (84.0 vs. 13.0 months, HR 0.09; 95% CI 0.02–0.54, p = 0.004) (Fig. 5A). Importantly, there was no difference in the
Integrating clinicopathologic and genomic data among immunophenotypes
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
65
Fig. 2. (A-D) t-distributed stochastic neighbor embedding (t-SNE) algorithm applied to visualize how R/M SCCHN tumors relate to one another based on multiple immune parameters, with each panel interrogating a key immune population among individual tumors (heatmap indicates the % expression level) that collectively define 3 tumor clusters (colored boxes). Each datapoint represents an individual tumor sample. Multiparametric flow cytometry data (including: CD3, CD8, CD14, CD45, CD56, NKp46, CD279, HLA-DR, CD4, CD16, CD19, CD33, CD66b, CD123, PD-1, TIM-3, CTLA-4, CD45, LAG-3, and intracellular FOXp3) for each sample was analyzed to generate a location on the two-dimensional (2D) spatial map. This allows for visualization of the similarity between the overall immune composition of each sample. Tumors with a similar overall immune signature locate closer together. The same 2D spatial map and clustering pattern is shown highlighting: (A) CD3+, (B) CD8+, (C) CD4+, and (D) PD-1/TIM3+ CD8+ T cell populations. (E) Flow cytometric contour plots profiling R/M SCCHN tumors, highlighting key immune cell and immune checkpoint expression patterns among the 3 tumor clusters; each row corresponds to a tumor cluster shown in (A–D). SSC = side scatter. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
and immune profiling data, there was no correlation between overall non-synonymous mutational burden using a targeted nextgeneration sequencing platform [26] comprised of 504 genes (range: 0–8 mutations per sample) and either CD8+ T cell abundance or PD-1+ CD8+ T cell expression at this cohort size (R = 0.25, p = 0.288; R = 0.14, p = 0.551, respectively) (Fig. 6A and B). Among anti-PD-1 treated patients, mutational burden did not appear to correlate with survival while on therapy (R = 0.13, p = 0.789). Additionally, there was no difference in mutational burden between those patients demonstrating a clinical benefit from PD-1 blockade (defined here as > 6 month OS from the start of anti-PD-1 therapy) and those without clinical benefit (p = 0.600). Fig. 6C depicts the overall mutational landscape among tumor samples arranged by CD8+ T cell abundance.
Discussion We prospectively profiled the tumor immune microenvironment in R/M SCCHN, correlating immune metrics with targeted sequencing data, clinical parameters, and outcomes. We found that higher CD8+ T cell abundance in this previously treated, immunotherapy-naïve population correlated with OS (p = 0.023), a relationship that has been observed in other solid malignancies [27,28]. Additionally, we identified a subgroup of tumors with an inflamed immune composition characterized by a robust CD8+ T cell infiltrate with high checkpoint co-expression (PD-1/TIM3+) independent of HPV or smoking status. These CD8+ T cells often displayed an activated phenotype despite the presence of multiple
immune checkpoint markers. While higher natural killer (NK) cell abundance appears to correlate with survival in SCCHN [29] and NK cell infiltrates were low in this R/M cohort, we did observe significantly higher NK cell infiltrates among our inflamed tumors (median 4.7 vs. 1.1%, p = 0.024). Our focus on R/M disease may explain differences in NK cell populations observed in other SCCHN studies. In addition, infiltrating monocytes in inflamed tumors had significantly higher PD-L1/2 expression. These inflamed immunophenotype patients had significantly improved median OS compared with a non-inflamed group (84.0 vs. 13.0 months, p = 0.004), and had delayed TTR (median: 28.0 months). Although a preliminary observation, patients with an inflamed immunophenotype had longer survival from the time of starting PD-1 blockade compared with non-inflamed treated patients (median 9.5 vs. 2.0 months). Together these findings are suggestive of an immunologically defined subgroup of R/M SCCHN patients who may be more likely to achieve clinical benefit with checkpoint inhibitors which requires an existing CD8+ T cell infiltrate – similar findings have been observed in advanced melanoma [30]. The infiltrate in our cohort appears regulated by high checkpoint co-expression which may reflect a prior T cell response that was later inhibited. A similar phenotype has been observed in non-small cell lung tumors with squamous histology [31]. Combined immune checkpoint blockade may add benefit to improve survival in this subgroup. Beyond the R/M setting, SCCHN is a heavily immuneinfiltrated tumor type comparatively and higher CD8+ T cell infiltrates have been correlated with improved outcomes in SCCHN patients treated with adjuvant chemoradiotherapy (CRT), regardless of HPV status [29,32]. Keck and colleagues have recently
66
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
Fig. 3. (A) Major immune cell lineages, (B) T lymphocyte subpopulations, (C) immune checkpoint ligand, and (D) immune checkpoint receptor expression patterns among R/ M SCCHN tumors grouped by immunophenotype using multiparametric flow cytometry data. Median and interquartile ranges are displayed.
Fig. 4. (A) CD8+ T cell lineage, (B) immune markers of T cell activation, inhibition and proliferation, (C) immune checkpoint co-expression patterns in CD8+ T cells and (D) CD4+ T cells among R/M SCCHN tumors using multiparametric flow cytometry data. Tumor samples are grouped by inflamed (I) and non-inflamed (NI) tumor immunophenotypes. Median and interquartile ranges are displayed.
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
67
Fig. 5. (A) Overall survival (in months) among R/M SCCHN patients with inflamed vs. non-inflamed tumor immunophenotypes using multiparametric flow cytometry data for immune profiling. Correlation between (B) CD8+ T cell abundance (%), (C) monocyte PD-L1 expression (%) and overall survival (in months) among R/M SCCHN patients. Red dotted lines indicate confidence intervals.
Fig. 6. Correlation between non-synonymous mutational burden using a targeted next-generation sequencing platform (A) and % of CD8+ T cells or (B)% of PD-1+ CD8+ T cells among R/M SCCHN tumor (blue diamond = the corresponding patient did not receive anti-PD-1 treatment, red open circle = anti-PD-1 non-responder, red filled circle = antiPD-1 responder). (C) Mutational landscape among tumors arranged by CD8+ T cell abundance (%). Each column represents a single tumor sample, and the blue color coding reflects the known clinical significance of the genomic alteration. The % positivity of T cells is color coded according to the heatmap shown. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
described an inflamed immune signature in SCCHN using gene expression analysis [33], which may encompass our inflamed immunophenotype tumors. While results from CheckMate-141 using nivolumab in R/M SCCHN demonstrated improved survival in those patients with tumors exhibiting 1% PD-L1 expression [16], we did not find a significant correlation between quantitative PD-L1/2 expression on tumor or immune cells and survival at this cohort size (R = 0.23, p = 0.283; R = 0.03, p = 0.885). Many immune biomarker studies to date have utilized IHC for PD-L1 scoring but this
may not completely capture PD-L1 expression and the interpretation of these results can be subjective. However, we did observe higher PD-L1/2 expression among infiltrating monocytes in our inflamed patients, and PD-1+ CD8+ T cell expression correlated significantly with PD-L1 expression on monocytes for the entire cohort (R = 0.46, p = 0.014). Immune profiling using flow cytometry offers several key advantages over IHC in that the method is rapid, relies on automation and permits quantitative analysis. Recent work has begun to validate flow cytometric immune profiling data with IHC results [19].
68
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
Also of interest was the non-inflamed subgroup which demonstrated lower CD8+ T cell abundance and low co-expression of checkpoint receptors. However, PD-1/CTLA-4 co-expression among CD4+ T cells was increased in this subgroup – likely coinciding with the Treg population. Median OS was significantly lower in non-inflamed immunophenotype patients (13.0 months), with a median TTR of 4.5 months. While heterogeneous in terms of smoking and HPV status, oral cavity and laryngeal cancer patients tended to cluster in this subgroup. We also observed that five CD8lowCPlow or non-inflamed patients who went on PD-1 blockade had a median OS of 2.0 months from the time of starting therapy – but larger cohorts are needed to draw definitive conclusions. While increased PD-1+ T cell infiltrates have been observed in oral cavity tumors previously, lower levels of IFN-c were detected [34]. Although speculative, this could reflect a lower CD8+ T cell infiltrate with a less activated phenotype predicting poor responses to single agent checkpoint blockade. In addition, this subgroup had higher Treg infiltrates, known to promote immunosuppression and a primary obstacle for effective tumor immunotherapy [35]. Greater PD-1/CTLA-4 co-expression on CD4+ T cells could warrant the use of CTLA-4 inhibitors in these non-inflamed tumors. The majority of our cohort (70.6%) was treated with prior CRT. Our group has evaluated blood samples in SCCHN patients undergoing definitive radiotherapy and found that circulating CD8+ T cells and soluble PD-L1 levels increased during treatment, as did T cell exhaustion markers [36]. There was no difference in the number of patients treated with CRT between immune subgroups in our study, but median TTR was longer in the inflamed phenotype patients (28.0 months), which could partially reflect persistent immunomodulatory effects of prior CRT. We integrated genomic and immune metrics with survival and observed no correlation between overall non-synonymous mutational burden and CD8+ T cell or PD-1+ CD8+ T cell expression at this cohort size and among those treated in our cohort with PD-1 blockade, mutational burden did not appear to correlate with survival or clinical benefit from therapy. These results should be interpreted cautiously as whole-exome sequencing (WES) analysis in other tumor types has correlated mutational burden with response to PD-1 blockade [37] and WES results from a SCCHN specific population have not been published. Of interest, recent work from Spranger and colleagues has failed to show decreased mutational burden in non-inflamed tumors and therefore other mechanisms may explain the lack of T cell recruitment and response to checkpoint inhibitors in these tumors [38]. Our study has several important limitations to consider. While our sample size is notable for this specific head and neck cancer population our findings are largely descriptive, but represent an important cohort of R/M patients with clinical heterogeneity that integrates genomic and immune microenvironment data with outcomes. Our approach to tissue preparation was standardized, but at times sample yield was low. Sampling error and intratumoral heterogeneity are other potential limitations to our approach. A disadvantage of using flow cytometry for immune profiling is that it does not provide tissue architecture to understand neighboring tumor and immune cell interactions, which appears important. Not every patient agreed to separately consent for genomic profiling and adequate tissue may have been exhausted. Although our genomic assay captures several hundred genes, WES with neoantigen characterization is ongoing. We describe a cohort of R/M SCCHN patients with an inflamed immunophenotype characterized by a dominant CD8+ T cell infiltrate and immune checkpoint co-expression that portends improved survival independent of HPV or smoking status. While preliminary, those in this subgroup treated with single agent PD-1 blockade appeared to benefit. We hypothesize that combined checkpoint inhibition may further improve their outcomes. We
also observed a non-inflamed subgroup with a lower immune checkpoint co-expressing CD8+ T cell infiltrate and higher Treg abundance characterized by poor survival – often comprised of oral cavity or laryngeal cancer patients. These patients may benefit from priming prior to the initiation of immune checkpoint blockade. Further studies are needed, but these preliminary observations suggest that immune-based metrics could be useful to stratify R/M SCCHN patients to identify those likely to respond to immunotherapies. Conflict of interest The authors declare no potential conflicts of interest. Financial support Dr. Schoenfeld: institutional support from Merck, BMS; paid scientific advisory board for BMS, Debiopharm. Dr. Haddad: institutional support from Merck, BMS, Astra Zeneca, Celgene; consulting honoraria from Merck, BMS, Pfizer, Eisai, Astra Zeneca, Celgene. Dr. Lorch: institutional support from Novartis, Millennium; consulting honoraria from Eisai. Dr. Wong: Robert A. and Renée E. Belfer Foundation, Expect Miracles Foundation, Starr Consortium for Cancer Research, the Damon Runyon Cancer Research Foundation, and Stand Up 2 Cancer Foundation (SU2C). Dr. Uppaluri: research support and consultant fees from Merck. Dr. Hammerman: Damon Runyon Cancer Research Foundation, NCI K-08-163677. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. References [1] Haddad RI, Shin DM. Recent advances in head and neck cancer. N Engl J Med 2008;359:1143–54. [2] Seiwert TY, Cohen EE. State-of-the-art management of locally advanced head and neck cancer. Br J Cancer 2005;92:1341–8. [3] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin 2016;66:7–30. [4] Galon J, Costes A, Sanchez-Cabo F, Kirilovsky A, Mlecnik B, Lagorce-Pagès C, et al. Type, density, and location of immune cells within human colorectal tumors predict clinical outcome. Science 2006;313:1960–4. [5] Kuss I, Hathaway B, Ferris RL, Gooding W, Whiteside TL. Decreased absolute counts of T lymphocyte subsets and their relation to disease in squamous cell carcinoma of the head and neck. Clin Cancer Res 2004;10:3755–62. [6] Ferris RL. Immunology and immunotherapy of head and neck cancer. J Clin Oncol 2015;33:3293–304. [7] Duffy SA, Taylor JM, Terrell JE, Islam M, Li Y, Fowler KE, et al. Interleukin-6 predicts recurrence and survival among head and neck cancer patients. Cancer 2008;113:750–7. [8] Stanley M. Immunobiology of HPV and HPV vaccines. Gynecol Oncol 2008;109: S15–21. [9] O’Brien PM, Campo M Saveria. Evasion of host immunity directed by papillomavirus-encoded proteins. Virus Res 2002;88:103–17. [10] Ramsay AG. Immune checkpoint blockade immunotherapy to activate antitumour T-cell immunity. Br J Haematol 2013;162:313–25. [11] Leach DR, Krummel MF, Allison JP. Enhancement of antitumor immunity by CTLA-4 blockade. Science 1996;271:1734–6. [12] Lyford-Pike S, Peng S, Young GD, Taube JM, Westra WH, Akpeng B, et al. Evidence for a role of the PD-1:PD-L1 pathway in immune resistance of HPVassociated head and neck squamous cell carcinoma. Cancer Res 2013;73:1733–41. [13] Feldman R, Gatalica Z, Knezetic J, Reddy S, Nathan CA, Javadi N, et al. Molecular profiling of head and neck squamous cell carcinoma. Head Neck 2016;38. E1625-38. [14] Badoual C, Hans S, Merillon N, Van Ryswick C, Ravel P, Benhamouda N, et al. PD-1-expressing tumor-infiltrating T cells are a favorable prognostic biomarker in HPV-associated head and neck cancer 2013;73:128–38. [15] Seiwert TY, Burtness B, Mehra R, Weiss J, Berger R, Eder JP, et al. Safety and clinical activity of pembrolizumab for treatment of recurrent or metastatic squamous cell carcinoma of the head and neck (KEYNOTE-012): an open-label, multicentre, phase 1b trial. Lancet Oncol 2016;17:956–65. [16] Ferris RL, Blumenschein G Jr, Fayette J, Guigay J, Colevas AD, Licitra L, et al. Nivolumab for recurrent squamous-cell carcinoma of the head and neck. N Engl J Med 2016 [Epub ahead of print].
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69 [17] Koyama S, Akbay EA, Li YY, Herter-Sprie GS, Buczkowski KA, Richards WG, et al. Adaptive resistance to therapeutic PD-1 blockade is associated with upregulation of alternative immune checkpoints. Nat Commun 2016;7:10501. [18] Gajewski TF, Schreiber H, Fu YX. Innate and adaptive immune cells in the tumor microenvironment. Nat Immunol 2013;14:1014–22. [19] Awad MM, Jones RE, Liu H, Lizotte PH, Ivanova EV, Kulkarni M, et al. Cytotoxic T cells in PD-L1-positive malignant pleural mesotheliomas are counterbalanced by distinct immunosuppressive factors. Cancer Immunol Res 2016. Nov 14. [20] Abo RP, Ducar M, Garcia EP, Thorner AR, Rojas-Rudilla V, Lin L, et al. BreaKmer: detection of structural variation in targeted massively parallel sequencing data using kmers. Nucleic Acids Res 2015;43:e19. [21] Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics 2010;26:589–95. [22] Cibulskis K, Lawrence MS, Carter SL, Sivachenko A, Jaffe D, Sougnez C, et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotechnol 2013;31:213–9. [23] McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing nextgeneration DNA sequencing data. Genome Res 2010;20:1297–303. [24] Pezzotti N, Lelieveldt B, van der Maaten L, Hollt T, Eisemann E, Vilanova A. Approximated and User Steerable tSNE for Progressive Visual Analytics. IEEE Trans Vis Comput Graph 2016 [Epub ahead of print]. [25] R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing 2013 V, Austria. ISBN 3-900051-07-0, URL
69
[29] Mandal R, Senbabaoglu Y, Desrichard A, Havel JJ, Dalin MG, Riaz N, et al. The head and neck cancer immune landscape and its immunotherapeutic implications. JCI Insight 2016;1:e89829. [30] Tumeh PC, Harview CL, Yearley JH, Shintaku IP, Taylor EJ, Robert L, et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature 2014;515:568–71. [31] Lizotte PH, Ivanova EV, Awad MM, Jones RE, Keogh L, Liu H, et al. Multiparametric profiling of non-small-cell lung cancers reveals distinct immunophenotypes. JCI Insight 2016;1:e89014. [32] Balermpas P, Rodel F, Rödel C, Krause M, Linge A, Lohaus F, et al. CD8+ tumourinfiltrating lymphocytes in relation to HPV status and clinical outcome in patients with head and neck cancer after postoperative chemoradiotherapy: a multicentre study of the German cancer consortium radiation oncology group (DKTK-ROG). Int J Cancer 2016;138:171–81. [33] Keck MK, Zuo Z, Khattri A, Stricker TP, Brown CD, Imanguli M, et al. Integrative analysis of head and neck cancer identifies two biologically distinct HPV and three non-HPV subtypes. Clin Cancer Res 2015;21:870–81. [34] Malaspina TS, Gasparoto TH, Costa MR, de Melo EF, Jr Ikoma MR, Damante JH, et al. Enhanced programmed death 1 (PD-1) and PD-1 ligand (PD-L1) expression in patients with actinic cheilitis and oral squamous cell carcinoma. Cancer Immunol Immunother 2011;60:965–74. [35] Zou W. Regulatory T cells, tumour immunity and immunotherapy. Nat Rev Immunol 2006;6:295–307. [36] Sridharan V, Margalit DN, Lynch SA, Severgnini M, Zhou J, Chau NG, et al. Definitive chemoradiation alters the immunologic landscape and immune checkpoints in head and neck cancer. Br J Cancer 2016;115:252–60. [37] Rizvi NA, Hellmann MD, Snyder A, Kvistborg P, Makarov V, Havel JJ, et al. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 2015;348:124–8. [38] Spranger S, Luke JJ, Bao R, Zha Y, Hernandez KM, Li Y, et al. Density of immunogenic antigens does not explain the presence or absence of the T-cell-inflamed tumor microenvironment in melanoma. Proc Natl Acad Sci USA 2016. Nov 11.
Contents lists available at ScienceDirect
Oral Oncology journal homepage: www.elsevier.com/locate/oraloncology
Defining an inflamed tumor immunophenotype in recurrent, metastatic squamous cell carcinoma of the head and neck Glenn J. Hanna a, Hongye Liu a, Robert E. Jones a,b, Alyssa F. Bacay a, Patrick H. Lizotte a,b, Elena V. Ivanova b, Mark A. Bittinger a,b, Megan E. Cavanaugh a,b, Amanda J. Rode a,b, Jonathan D. Schoenfeld c, Nicole G. Chau a, Robert I. Haddad a, Jochen H. Lorch a, Kwok-Kin Wong a,b, Ravindra Uppaluri a,d,⇑, Peter S. Hammerman a a
Department of Medical Oncology, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA 02215, USA Robert and Renee Belfer Center for Applied Cancer Science, Dana-Farber Cancer Institute, 360 Longwood Avenue, Boston, MA 02215, USA Department of Radiation Oncology, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA 02215, USA d Division of Otolaryngology-Head & Neck Surgery, Brigham & Women’s Hospital, 75 Francis Street, Boston, MA 02215, USA b c
a r t i c l e
i n f o
Article history: Received 21 December 2016 Received in revised form 24 January 2017 Accepted 7 February 2017
Keywords: Head and neck cancer Immunotherapy Biomarkers PD-1 Survival
a b s t r a c t Objectives: Immune checkpoint inhibitors have demonstrated clinical benefit in recurrent, metastatic (R/ M) squamous cell carcinoma of the head and neck (SSCHN), but lacking are biomarkers that predict response. We sought to define an inflamed tumor immunophenotype in this R/M SCCHN population and correlate immune metrics with clinical parameters and survival. Methods: Tumor samples were prospectively acquired from 34 patients to perform multiparametric flow cytometry and multidimensional clustering analysis integrated with next-generation sequencing data, clinical parameters and outcomes. Results: We identified an inflamed subgroup of tumors with prominent CD8+ T cell infiltrates and high PD-1/TIM3 co-expression independent of clinical variables, with improved survival compared with a non-inflamed subgroup (median overall survival 84.0 vs. 13.0 months, p = 0.004). The non-inflamed subgroup demonstrated low CD8+ T cells, low PD-1/TIM3 co-expression, and higher Tregs. Overall nonsynonymous mutational burden did not correlate with response to PD-1 blockade in a subset of patients. Conclusion: R/M SCCHN patients with an inflamed tumor immunophenotype demonstrate improved survival. Further prospective studies are needed to validate these findings and explore the use of immunophenotype to guide patient selection for immunotherapeutic approaches. Ó 2017 Elsevier Ltd. All rights reserved.
Introduction Even with current treatments and in the era of favorable prognosis human papillomavirus (HPV)-associated disease, there remains a significant risk of locoregional recurrence or distant metastases in squamous cell carcinoma of the head and neck (SCCHN) [1,2]. Moreover, 5-year survival in patients with recurrent or metastatic (R/M) SCCHN remains dismal [3]. In recent years, it has become evident that tumor progression is promoted by immune evasion and abrogation of an effective immune response against cancer cells [4]. SCCHN appears closely linked with immunosuppression, and patients often demonstrate impaired immune cell function that correlates with poor outcomes [5,6]. Mechanisms of tumor immune evasion in SCCHN include the ⇑ Corresponding author at: Harvard Medical School, Dana-Farber Cancer Institute, 450 Brookline Ave., Boston, MA 02115, USA. E-mail address: [email protected] (R. Uppaluri). http://dx.doi.org/10.1016/j.oraloncology.2017.02.005 1368-8375/Ó 2017 Elsevier Ltd. All rights reserved.
development of T cell tolerance, modulation of inflammatory and angiogenic cytokines, downregulation of antigen-processing machinery, and the expression of immune checkpoint ligands or receptors to promote immune evasion [7–9]. These mechanisms are currently serving to define immunotherapy targets for clinical development. Immune checkpoint receptors inhibit normal T cell activation and costimulation to maintain a homeostatic immune response [10]. Cytotoxic T lymphocyte-associated protein 4 (CTLA-4) and programmed death-1 (PD-1) are expressed on the surface of immune cells and interact with their respective ligands on antigen presenting or tumor cells. High tumor expression of the ligands of PD-1 (PD-L1, L2) and/or PD-1 expression by T lymphocytes can attenuate T cell activation and drive T cell exhaustion favoring tumor immune evasion [11]. Studies have estimated PD-L1 expression in SCCHN at 30–70% [12,13] with HPV+ tumors more frequently harboring infiltrating immune cells that express PD-1, despite its favorable prognosis [14]. Two recently published trials
62
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
have established the efficacy of PD-1 blockade in advanced SCCHN. Anti-PD-1 blockade with pembrolizumab in a heavily pre-treated SCCHN population that was PD-L1+( 1% by immunohistochemistry [IHC] in tumor or immune cells) resulted in an overall response rate of 18%, and 25% among those who were HPV+ [15]. The CheckMate-141 trial in platinum-refractory SCCHN using the anti-PD-1 antibody nivolumab demonstrated an improvement in median overall survival (OS) compared with standard chemotherapy – which was greater in those with HPV+ disease and in those with 1% tumoral PD-L1 expression by IHC [16]. These data were the foundation for Food and Drug Administration approval of these agents in 2016. These studies illustrate that response rates to antiPD-1 monotherapy in SCCHN patients are low. Additionally, there is significant cost and toxicity associated with these agents, together highlighting the critical need to identify predictive biomarkers or immunophenotypes to guide patient selection. The tumor immune microenvironment is built on spatially diverse and complex immune relationships that appear dynamic. Thus alternative immune checkpoint upregulation by T cells represents just one mechanism to explain why capturing PD-1/PD-L1 interactions alone may not predict response to inhibitors [17]. Recent work in other solid tumors has suggested two main tumor immunophenotypes: an inflamed tumor category comprised of a rich T cell infiltrate, a type 1 interferon signature, and a diverse chemokine profile, and a non-inflamed category that lacks these features [18]. These so-called inflamed tumors may respond more favorably to therapies targeting immune checkpoint mechanisms. Here we prospectively characterized the tumor immune microenvironment using multiparametric flow cytometry and a multidimensional clustering algorithm to define an inflamed phenotype in R/M SCCHN and correlated these findings with mutational burden, smoking and HPV status, and survival. Our observations strengthen our understanding of the tumor immune microenvironment specific to SCCHN. Along with advances in genomic biomarkers, immunophenotyping may serve to guide the selection and sequencing of immunotherapies, and identify candidate biomarkers to predict response to checkpoint inhibitors.
The surface antibody against CD45 (2D1) was purchased from eBioscience and the surface antibody against LAG-3 (polyclonal) and its isotype control (polyclonal) were purchased from R&D Systems. The intracellular antibody FOXP3 (236A/E7) was purchased from eBioscience. Cells were analyzed within 72 h of fixation on a BD FACSCanto II HTS or BD Fortessa cell analyzer with FACSDiva software v8.0.1 (BD Biosciences) and analyzed using FlowJo software v10. Massively parallel genomic sequencing Targeted next-generation sequencing was performed on DNA extracted from formalin-fixed paraffin-embedded (FFPE) archival tumor samples. Total gDNA concentration was determined using the PicoGreen dsDNA quantification assay (ThermoFisher Scientific, Waltham, MA), and samples with sufficient starting material (> 100 ng) were taken forward to library construction [20]. Briefly, gDNA was fragmented by ultrasonication to 150 base pairs and purified. Size-selected DNA was ligated to specific adapters during library construction (Illumina, Inc., San Diego, CA). Libraries were quantified using qPCR (Kapa Biosystems, Inc., Woburn, MA) and captured using the OncoPanel_v2 bait set using the Agilent SureSelect hybrid capture kit (Agilent Technologies, Santa Clara, CA). Oncopanel_v2 consists of 504 genes and 15 intronic regions with known or potential importance in cancer. The captured libraries then underwent paired end 100 (2 100 nt) sequencing on a Hiseq 2500 (Illumina Inc.). Read pairs were aligned to the reference sequence b37 edition from the Human Genome Reference Consortium using the Burrows-Wheeler Aligner [21] and de-multiplexed using Picard tools. Mutation analysis for single nucleotide variants and nonsynonymous mutational burden was performed using MuTect v1.1.4 in paired mode for tumors with matched germline or single mode for samples without matched germline, and annotated using Oncotator [22]. The alignments were further refined using the GATK tool for localized realignment around indel sites [23]. As part of quality metrics, all samples had to meet a minimum requirement that 80% of targets have a minimum coverage of 30.
Materials and methods Statistical analysis Subjects Patients with SCCHN undergoing biopsy to confirm metastatic disease and those undergoing surgical resection for locoregional recurrence were identified prospectively. Prior to biopsy, informed consent was obtained for institutional review board-approved protocols. Peripheral blood controls were obtained at the time of biopsy in a subset of patients. Patient demographics and clinical characteristics were recorded. Multiparametric flow cytometry At the time of biopsy or surgical resection, fresh tissue samples from each patient were placed in RPMI-1640 with 10% fetal bovine serum (FBS). Tumor was confirmed by hematoxylin and eosin staining. A non-fixed cell suspension was then prepared for staining with fluorescently-conjugated antibody cocktails, as previously described [19]. Single cell suspensions were stained using mouseanti-human antibodies. Surface antibodies against CD3 (HIT3a; UCHT1), CD8 (RPA-T8), CD14 (M5E2; MphiP9), CD45 (HI30), CD56 (B159), CD279 (EH12.1) and its isotype control (MOPC-21), and HLA-DR (G46-6) were purchased from BD Biosciences. Surface antibodies against CD4 (RPA-T4), CD16 (3G8), CD19 (HIB19), CD33 (WM53), CD66b (G10F5), CD123 (6H6), and TIM-3 (F38-2E2) and its isotype control (MOPC-21) were purchased from BioLegend.
Fisher’s exact test for categorical variables (Wilcoxon rank sum test for continuous variables) and one-way ANOVA on ranks was used to assess for differences between individual immune subgroups. Spearman’s Rho was used to measure the strength of association between variables. All statistical tests used a significance of <0.05 and were two-sided. An unsupervised non-linear dimension reduction method, t-Distributed Stochastic Neighbor Embedding (t-SNE), was used to investigate in reduced dimension space how tumors locate in relation to one another based on multiparametric flow data – using a tree-based algorithm [24]. With n = 20 samples included in t-SNE, the power to differentiate at least 8 markers was 0.8. Overall survival (OS) was determined from the date of diagnosis to death from any cause, otherwise this was censored at date of last follow-up. Time to recurrence (TTR) was determined from the date of completion of initial treatment to recurrence or metastatic disease – and Kaplan-Meier statistics were applied. Data were analyzed using R software package (version 2.15.3) [25]. Results Clinical characteristics of the cohort Demographics, clinical characteristics and survival information from 34 patients are summarized in Table 1. With a median age of
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69 Table 1 Demographics, clinical characteristics and survival information in patients with R/M SCCHN. Characteristic
(%)a, n = 34
Age (median, y)
57.5 (26–84)
Gender Male Female
23 (67.6) 11 (32.4)
Smoking status Never or <10 pack-year Former (10 pack-year) Current
14 (41.1) 16 (47.1) 4 (11.8)
Primary site of disease Oral cavity Oropharynx Nasopharynx Larynx Cutaneous Unknown
12 (35.3) 8 (23.5) 4 (11.8) 5 (14.7) 3 (8.8) 2 (5.9)
Initial staging at diagnosis Stage I, II Stage III, IV
11 (32.4) 23 (67.6)
HPV statusb HPV+ HPV-
10 (29.4) 24 (70.6)
Initial treatment regimen Surgery Surgery + radiation Surgery + CRT CRT IC + CRT (sequential) Targeted sequencing performed TP53 mutation NOTCH mutation PIK3CA mutation
6 (17.6) 4 (11.8) 4 (11.8) 13 (38.2) 7 (20.6) 20 (58.8) 9 (45.0) 7 (35.0) 5 (20.0)
Site of recurrence Locoregional Distant
24 (70.6) 10 (29.4)
Subsequent treatment with immune checkpoint therapy
9 (26.5)
Overall survival (median, in months) HPV+ HPV# died Median TTR (mos)
22.0 (3.0–202.0) 28.5 (10.0–87.0) 21.0 (3.0–202.0) 17 (50.0) 9.0 (1.0–187.0)
a
Except for age, survival and time to recurrence. HPV status was tested in all oropharyngeal and unknown primaries and assumed to be negative at other sites; HPV = human papillomavirus, CRT = concurrent chemoradiation, IC = induction chemotherapy, TTR = time to recurrence. b
57.5 (26–84 years) at diagnosis, this predominantly male cohort included >50% former or current smokers and 29.4% of patients had HPV+ disease. Most patients presented with locoregional involvement and were initially treated with multimodality therapy. Locoregional recurrences accounted for 70.6% of patients (median TTR of 9.0 months). Seventeen (50.0%) patients in the cohort had died at the time of analysis (median follow-up was 9.0 months). Following biopsy, 9 (26.5%) patients went on to be treated with immune checkpoint blockade. Tumor immune composition of the cohort Multiparametric flow cytometry performed on tumor samples demonstrated variable cell viability (median: 18.7% [5.1, 61.2]) with a predominant leukocyte population composed of mostly granulocytes among individual tumors (Fig. 1A). We identified a CD3+ T cell infiltrate (median: 18.0% [3.5, 39.4] of leukocytes) in the majority of tumors, with a median of 40.3% [34.7, 54.2] of T cells showing CD8 positivity and 50.1% [35.6, 60.6] of T cells expressing CD4 (Fig. 1B) resulting in a median CD8/4+T cell ratio of 1.0 [0.6, 1.5]. A high degree of immune cell heterogeneity was
63
observed when considering key clinicopathologic features of this cohort: there was no difference in CD8+ T cell abundance based on primary disease site (p = 0.283), smoking history (p = 0.570), or HPV status (p = 0.119). Matched peripheral blood samples were available in a small subset of patients (n = 5) and were composed of few CD3+ T cells (median: 1.9%, CD8/4+ T cell ratio: 0.5). Defining tumor immunophenotypes Next we utilized the t-distributed stochastic neighbor embedding (t-SNE) algorithm [24] to analyze tumor immunophenotypic data obtained from multiparametric flow cytometry. The t-SNE algorithm is a tool to visualize high-dimensional data, whereby each datapoint is plotted on a two-dimensional (2D) map. Fig. 2A–D shows a 2D spatial map and plots individual tumor samples (n = 23) at a given location based on comprehensive immunophenotypic data obtained by flow cytometry. This map spatially depicts the similarity between the overall tumor immune composition of each sample (using a minimum of 9 immune parameters), in an unbiased manner – whereby tumors with a similar overall immune profile locate closer together on the map. The map is then interrogated for a given immune parameter(s) among individual tumor samples to understand expression levels of key immune markers. Fig. 2A–D shows the same spatial map, with each panel highlighting a single immune population of interest that defined three tumor clusters: one with a high CD4+(low CD8 +) T cell infiltrate with low co-expression of PD-1/TIM3 checkpoints (CP) on CD8+ T cells (CD8lowCPlow). Another subgroup with a high CD8+ T cell population and high co-expression of PD-1/ TIM3 (CD8highCPhigh). Finally, a third subgroup was comprised of an intermediate CD8+ T cell population with variable coexpression of PD-1/TIM3 checkpoint receptors (CD8intCPint). Fig. 2E shows flow cytometry contour plots characterizing key immune cell populations corresponding to each of the three defined tumor clusters shown on the 2D spatial map. Defining an inflamed tumor phenotype We next sought to investigate both immune cell and immune checkpoint expression patterns among the 3 tumor clusters. Fig. 3A depicts immune cell composition among each tumor cluster, showing a higher CD3+ T cell abundance among CD8highCPhigh tumors (p < 0.001), but there was no significant difference in absolute T cell number between each of the subgroups (p = 0.109) – suggesting heterogeneity among tumor leukocyte populations. The CD8highCPhigh tumor subgroup demonstrated the greatest CD8+ T cell population among the 3 immune clusters (median: 75.1%, p = 0.008), but with >50% median CD8+ T cell abundance among the CD8intCPint group as well (Fig. 3B). The CD8lowCPlow subgroup had a greater infiltrate of FOXP3+ CD25+ T regulatory cells (Treg) (median: 22.6%, p = 0.026), although CD8+/Treg ratios were similar among all 3 tumor clusters (p = 0.528). CD8highCPhigh tumors demonstrated a greater population of PD-L1+ and PD-L2+ tumor-infiltrating monocytes compared with CD8lowCPlow tumors (Fig. 3C, p = 0.01 and p = 0.04, respectively), with both CD8highCPhigh and CD8intCPint subgroups showing >50% PD-L1 expression among monocytes. However, PD-L1/2 expression on tumor cells was not significantly different between tumor clusters despite a trend towards greater tumoral PD-L1 expression among CD8highCPhigh tumors. PD-1+ CD8+ T cell expression correlated with PD-L1 expression on monocytes among all 3 tumor clusters (R = 0.46, p = 0.014), but PD-L2 status did not. The CD8highCPhigh subgroup also had higher PD-1 expression and PD-1/TIM-3 co-expression among CD8+ T cells (median: 91.5%, 43.4%; p < 0.001, p = 0.006, respectively), but again these values were higher in the CD8intCPint compared with the
64
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
Fig. 1. Clinicopathologic features coupled with multiparametric flow cytometry data from 34 patients with R/M SCCHN showing the (A) proportion of immune cells relative to the CD45+ leukocyte population in each tumor sample, and (B) the proportion of CD4+ and CD8+ T cells relative to the CD3+ T cell population in each tumor sample. Color tiles above indicate smoking status and a plus sign indicates human papillomavirus (HPV)-associated disease. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
CD8lowCPlow subgroup (Fig. 3D). Given the higher CD8+ T cell infiltrates and similar checkpoint co-expression patterns among the CD8highCPhigh and CD8intCPint groups, these two clusters were combined to represent an inflamed tumor immunophenotype. Fig. 4 integrates the CD8highCPhigh and CD8intCPint groups (so-called inflamed phenotypes) and compares CD8+ T cell subsets, markers of T cell activation, inhibition and proliferation with that of the CD8lowCPlow or non-inflamed phenotype. While CD8+ T cells were more abundant in inflamed tumors, T cell subsets were similar (Fig. 4A). The majority of CD8+ T cells among inflamed tumors showed evidence of activation (median of 81.4% were CD38+, 76.9% were CD69+, Fig. 4B). Inflamed tumors exhibited significantly greater PD-1+ CD8+ T cell presence and more often co-expressed PD-1/TIM3 on CD8+ T cells (p < 0.001, Fig. 4C). CTLA-4 expression on CD8+ T cells was low among all immunophenotypes, but PD-1/CTLA-4 co-expression on CD4+ T cells was increased in non-inflamed tumors (p = 0.019, Fig. 4D).
number of smokers and HPV+ tumors between these cohorts (p = 0.204 and p = 0.203, respectively). Similarly, CD8+ T cell abundance correlated with OS (R = 0.41, p = 0.023) for the entire cohort (Fig. 5B), but PD-L1+ status did not (Fig. 5C). Nine (26.5%) of 34 patients were subsequently treated with single agent PD-1 blockade following biopsy. Four of these patients had inflamed tumor phenotypes and 5 had non-inflamed tumors. Median OS was 9.5 months from the time of starting PD-1 blockade (with a median OS of 54.0 months from the date of diagnosis) among the inflamed phenotype patients who received PD-1 inhibition, compared with 2.0 months (median OS of 14.0 months from diagnosis) among the non-inflamed patients (p = 0.026). PD-L1+ monocytes were also higher among the inflamed phenotype patients (range 30.0– 76.2%) treated with checkpoint blockade.
Survival outcomes by tumor immunophenotype
When considering clinical features, multidimensional clustering analysis showed no clear pattern based on smoking or HPV status – although oral cavity and laryngeal tumors tended to cluster within the non-inflamed subgroup (n = 7 of 10). There was no significant difference in CD8+ T cell infiltrates by HPV status (p = 0.119) and similarly there was no difference in co-expression of PD-1/TIM3 on T cells when separated by smoking (p = 0.854) or HPV status (p = 0.312). When integrating clinical outcomes with genomic
When the CD8highCPhigh and CD8intCPint tumor samples (inflamed phenotypes) with >50% CD8+ T cell infiltrates and >50% PD-1/TIM3+ CD8+ T cell co-expression were combined, there was significantly improved median OS vs. the non-inflamed subgroup (84.0 vs. 13.0 months, HR 0.09; 95% CI 0.02–0.54, p = 0.004) (Fig. 5A). Importantly, there was no difference in the
Integrating clinicopathologic and genomic data among immunophenotypes
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
65
Fig. 2. (A-D) t-distributed stochastic neighbor embedding (t-SNE) algorithm applied to visualize how R/M SCCHN tumors relate to one another based on multiple immune parameters, with each panel interrogating a key immune population among individual tumors (heatmap indicates the % expression level) that collectively define 3 tumor clusters (colored boxes). Each datapoint represents an individual tumor sample. Multiparametric flow cytometry data (including: CD3, CD8, CD14, CD45, CD56, NKp46, CD279, HLA-DR, CD4, CD16, CD19, CD33, CD66b, CD123, PD-1, TIM-3, CTLA-4, CD45, LAG-3, and intracellular FOXp3) for each sample was analyzed to generate a location on the two-dimensional (2D) spatial map. This allows for visualization of the similarity between the overall immune composition of each sample. Tumors with a similar overall immune signature locate closer together. The same 2D spatial map and clustering pattern is shown highlighting: (A) CD3+, (B) CD8+, (C) CD4+, and (D) PD-1/TIM3+ CD8+ T cell populations. (E) Flow cytometric contour plots profiling R/M SCCHN tumors, highlighting key immune cell and immune checkpoint expression patterns among the 3 tumor clusters; each row corresponds to a tumor cluster shown in (A–D). SSC = side scatter. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
and immune profiling data, there was no correlation between overall non-synonymous mutational burden using a targeted nextgeneration sequencing platform [26] comprised of 504 genes (range: 0–8 mutations per sample) and either CD8+ T cell abundance or PD-1+ CD8+ T cell expression at this cohort size (R = 0.25, p = 0.288; R = 0.14, p = 0.551, respectively) (Fig. 6A and B). Among anti-PD-1 treated patients, mutational burden did not appear to correlate with survival while on therapy (R = 0.13, p = 0.789). Additionally, there was no difference in mutational burden between those patients demonstrating a clinical benefit from PD-1 blockade (defined here as > 6 month OS from the start of anti-PD-1 therapy) and those without clinical benefit (p = 0.600). Fig. 6C depicts the overall mutational landscape among tumor samples arranged by CD8+ T cell abundance.
Discussion We prospectively profiled the tumor immune microenvironment in R/M SCCHN, correlating immune metrics with targeted sequencing data, clinical parameters, and outcomes. We found that higher CD8+ T cell abundance in this previously treated, immunotherapy-naïve population correlated with OS (p = 0.023), a relationship that has been observed in other solid malignancies [27,28]. Additionally, we identified a subgroup of tumors with an inflamed immune composition characterized by a robust CD8+ T cell infiltrate with high checkpoint co-expression (PD-1/TIM3+) independent of HPV or smoking status. These CD8+ T cells often displayed an activated phenotype despite the presence of multiple
immune checkpoint markers. While higher natural killer (NK) cell abundance appears to correlate with survival in SCCHN [29] and NK cell infiltrates were low in this R/M cohort, we did observe significantly higher NK cell infiltrates among our inflamed tumors (median 4.7 vs. 1.1%, p = 0.024). Our focus on R/M disease may explain differences in NK cell populations observed in other SCCHN studies. In addition, infiltrating monocytes in inflamed tumors had significantly higher PD-L1/2 expression. These inflamed immunophenotype patients had significantly improved median OS compared with a non-inflamed group (84.0 vs. 13.0 months, p = 0.004), and had delayed TTR (median: 28.0 months). Although a preliminary observation, patients with an inflamed immunophenotype had longer survival from the time of starting PD-1 blockade compared with non-inflamed treated patients (median 9.5 vs. 2.0 months). Together these findings are suggestive of an immunologically defined subgroup of R/M SCCHN patients who may be more likely to achieve clinical benefit with checkpoint inhibitors which requires an existing CD8+ T cell infiltrate – similar findings have been observed in advanced melanoma [30]. The infiltrate in our cohort appears regulated by high checkpoint co-expression which may reflect a prior T cell response that was later inhibited. A similar phenotype has been observed in non-small cell lung tumors with squamous histology [31]. Combined immune checkpoint blockade may add benefit to improve survival in this subgroup. Beyond the R/M setting, SCCHN is a heavily immuneinfiltrated tumor type comparatively and higher CD8+ T cell infiltrates have been correlated with improved outcomes in SCCHN patients treated with adjuvant chemoradiotherapy (CRT), regardless of HPV status [29,32]. Keck and colleagues have recently
66
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
Fig. 3. (A) Major immune cell lineages, (B) T lymphocyte subpopulations, (C) immune checkpoint ligand, and (D) immune checkpoint receptor expression patterns among R/ M SCCHN tumors grouped by immunophenotype using multiparametric flow cytometry data. Median and interquartile ranges are displayed.
Fig. 4. (A) CD8+ T cell lineage, (B) immune markers of T cell activation, inhibition and proliferation, (C) immune checkpoint co-expression patterns in CD8+ T cells and (D) CD4+ T cells among R/M SCCHN tumors using multiparametric flow cytometry data. Tumor samples are grouped by inflamed (I) and non-inflamed (NI) tumor immunophenotypes. Median and interquartile ranges are displayed.
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
67
Fig. 5. (A) Overall survival (in months) among R/M SCCHN patients with inflamed vs. non-inflamed tumor immunophenotypes using multiparametric flow cytometry data for immune profiling. Correlation between (B) CD8+ T cell abundance (%), (C) monocyte PD-L1 expression (%) and overall survival (in months) among R/M SCCHN patients. Red dotted lines indicate confidence intervals.
Fig. 6. Correlation between non-synonymous mutational burden using a targeted next-generation sequencing platform (A) and % of CD8+ T cells or (B)% of PD-1+ CD8+ T cells among R/M SCCHN tumor (blue diamond = the corresponding patient did not receive anti-PD-1 treatment, red open circle = anti-PD-1 non-responder, red filled circle = antiPD-1 responder). (C) Mutational landscape among tumors arranged by CD8+ T cell abundance (%). Each column represents a single tumor sample, and the blue color coding reflects the known clinical significance of the genomic alteration. The % positivity of T cells is color coded according to the heatmap shown. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
described an inflamed immune signature in SCCHN using gene expression analysis [33], which may encompass our inflamed immunophenotype tumors. While results from CheckMate-141 using nivolumab in R/M SCCHN demonstrated improved survival in those patients with tumors exhibiting 1% PD-L1 expression [16], we did not find a significant correlation between quantitative PD-L1/2 expression on tumor or immune cells and survival at this cohort size (R = 0.23, p = 0.283; R = 0.03, p = 0.885). Many immune biomarker studies to date have utilized IHC for PD-L1 scoring but this
may not completely capture PD-L1 expression and the interpretation of these results can be subjective. However, we did observe higher PD-L1/2 expression among infiltrating monocytes in our inflamed patients, and PD-1+ CD8+ T cell expression correlated significantly with PD-L1 expression on monocytes for the entire cohort (R = 0.46, p = 0.014). Immune profiling using flow cytometry offers several key advantages over IHC in that the method is rapid, relies on automation and permits quantitative analysis. Recent work has begun to validate flow cytometric immune profiling data with IHC results [19].
68
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69
Also of interest was the non-inflamed subgroup which demonstrated lower CD8+ T cell abundance and low co-expression of checkpoint receptors. However, PD-1/CTLA-4 co-expression among CD4+ T cells was increased in this subgroup – likely coinciding with the Treg population. Median OS was significantly lower in non-inflamed immunophenotype patients (13.0 months), with a median TTR of 4.5 months. While heterogeneous in terms of smoking and HPV status, oral cavity and laryngeal cancer patients tended to cluster in this subgroup. We also observed that five CD8lowCPlow or non-inflamed patients who went on PD-1 blockade had a median OS of 2.0 months from the time of starting therapy – but larger cohorts are needed to draw definitive conclusions. While increased PD-1+ T cell infiltrates have been observed in oral cavity tumors previously, lower levels of IFN-c were detected [34]. Although speculative, this could reflect a lower CD8+ T cell infiltrate with a less activated phenotype predicting poor responses to single agent checkpoint blockade. In addition, this subgroup had higher Treg infiltrates, known to promote immunosuppression and a primary obstacle for effective tumor immunotherapy [35]. Greater PD-1/CTLA-4 co-expression on CD4+ T cells could warrant the use of CTLA-4 inhibitors in these non-inflamed tumors. The majority of our cohort (70.6%) was treated with prior CRT. Our group has evaluated blood samples in SCCHN patients undergoing definitive radiotherapy and found that circulating CD8+ T cells and soluble PD-L1 levels increased during treatment, as did T cell exhaustion markers [36]. There was no difference in the number of patients treated with CRT between immune subgroups in our study, but median TTR was longer in the inflamed phenotype patients (28.0 months), which could partially reflect persistent immunomodulatory effects of prior CRT. We integrated genomic and immune metrics with survival and observed no correlation between overall non-synonymous mutational burden and CD8+ T cell or PD-1+ CD8+ T cell expression at this cohort size and among those treated in our cohort with PD-1 blockade, mutational burden did not appear to correlate with survival or clinical benefit from therapy. These results should be interpreted cautiously as whole-exome sequencing (WES) analysis in other tumor types has correlated mutational burden with response to PD-1 blockade [37] and WES results from a SCCHN specific population have not been published. Of interest, recent work from Spranger and colleagues has failed to show decreased mutational burden in non-inflamed tumors and therefore other mechanisms may explain the lack of T cell recruitment and response to checkpoint inhibitors in these tumors [38]. Our study has several important limitations to consider. While our sample size is notable for this specific head and neck cancer population our findings are largely descriptive, but represent an important cohort of R/M patients with clinical heterogeneity that integrates genomic and immune microenvironment data with outcomes. Our approach to tissue preparation was standardized, but at times sample yield was low. Sampling error and intratumoral heterogeneity are other potential limitations to our approach. A disadvantage of using flow cytometry for immune profiling is that it does not provide tissue architecture to understand neighboring tumor and immune cell interactions, which appears important. Not every patient agreed to separately consent for genomic profiling and adequate tissue may have been exhausted. Although our genomic assay captures several hundred genes, WES with neoantigen characterization is ongoing. We describe a cohort of R/M SCCHN patients with an inflamed immunophenotype characterized by a dominant CD8+ T cell infiltrate and immune checkpoint co-expression that portends improved survival independent of HPV or smoking status. While preliminary, those in this subgroup treated with single agent PD-1 blockade appeared to benefit. We hypothesize that combined checkpoint inhibition may further improve their outcomes. We
also observed a non-inflamed subgroup with a lower immune checkpoint co-expressing CD8+ T cell infiltrate and higher Treg abundance characterized by poor survival – often comprised of oral cavity or laryngeal cancer patients. These patients may benefit from priming prior to the initiation of immune checkpoint blockade. Further studies are needed, but these preliminary observations suggest that immune-based metrics could be useful to stratify R/M SCCHN patients to identify those likely to respond to immunotherapies. Conflict of interest The authors declare no potential conflicts of interest. Financial support Dr. Schoenfeld: institutional support from Merck, BMS; paid scientific advisory board for BMS, Debiopharm. Dr. Haddad: institutional support from Merck, BMS, Astra Zeneca, Celgene; consulting honoraria from Merck, BMS, Pfizer, Eisai, Astra Zeneca, Celgene. Dr. Lorch: institutional support from Novartis, Millennium; consulting honoraria from Eisai. Dr. Wong: Robert A. and Renée E. Belfer Foundation, Expect Miracles Foundation, Starr Consortium for Cancer Research, the Damon Runyon Cancer Research Foundation, and Stand Up 2 Cancer Foundation (SU2C). Dr. Uppaluri: research support and consultant fees from Merck. Dr. Hammerman: Damon Runyon Cancer Research Foundation, NCI K-08-163677. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. References [1] Haddad RI, Shin DM. Recent advances in head and neck cancer. N Engl J Med 2008;359:1143–54. [2] Seiwert TY, Cohen EE. State-of-the-art management of locally advanced head and neck cancer. Br J Cancer 2005;92:1341–8. [3] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin 2016;66:7–30. [4] Galon J, Costes A, Sanchez-Cabo F, Kirilovsky A, Mlecnik B, Lagorce-Pagès C, et al. Type, density, and location of immune cells within human colorectal tumors predict clinical outcome. Science 2006;313:1960–4. [5] Kuss I, Hathaway B, Ferris RL, Gooding W, Whiteside TL. Decreased absolute counts of T lymphocyte subsets and their relation to disease in squamous cell carcinoma of the head and neck. Clin Cancer Res 2004;10:3755–62. [6] Ferris RL. Immunology and immunotherapy of head and neck cancer. J Clin Oncol 2015;33:3293–304. [7] Duffy SA, Taylor JM, Terrell JE, Islam M, Li Y, Fowler KE, et al. Interleukin-6 predicts recurrence and survival among head and neck cancer patients. Cancer 2008;113:750–7. [8] Stanley M. Immunobiology of HPV and HPV vaccines. Gynecol Oncol 2008;109: S15–21. [9] O’Brien PM, Campo M Saveria. Evasion of host immunity directed by papillomavirus-encoded proteins. Virus Res 2002;88:103–17. [10] Ramsay AG. Immune checkpoint blockade immunotherapy to activate antitumour T-cell immunity. Br J Haematol 2013;162:313–25. [11] Leach DR, Krummel MF, Allison JP. Enhancement of antitumor immunity by CTLA-4 blockade. Science 1996;271:1734–6. [12] Lyford-Pike S, Peng S, Young GD, Taube JM, Westra WH, Akpeng B, et al. Evidence for a role of the PD-1:PD-L1 pathway in immune resistance of HPVassociated head and neck squamous cell carcinoma. Cancer Res 2013;73:1733–41. [13] Feldman R, Gatalica Z, Knezetic J, Reddy S, Nathan CA, Javadi N, et al. Molecular profiling of head and neck squamous cell carcinoma. Head Neck 2016;38. E1625-38. [14] Badoual C, Hans S, Merillon N, Van Ryswick C, Ravel P, Benhamouda N, et al. PD-1-expressing tumor-infiltrating T cells are a favorable prognostic biomarker in HPV-associated head and neck cancer 2013;73:128–38. [15] Seiwert TY, Burtness B, Mehra R, Weiss J, Berger R, Eder JP, et al. Safety and clinical activity of pembrolizumab for treatment of recurrent or metastatic squamous cell carcinoma of the head and neck (KEYNOTE-012): an open-label, multicentre, phase 1b trial. Lancet Oncol 2016;17:956–65. [16] Ferris RL, Blumenschein G Jr, Fayette J, Guigay J, Colevas AD, Licitra L, et al. Nivolumab for recurrent squamous-cell carcinoma of the head and neck. N Engl J Med 2016 [Epub ahead of print].
G.J. Hanna et al. / Oral Oncology 67 (2017) 61–69 [17] Koyama S, Akbay EA, Li YY, Herter-Sprie GS, Buczkowski KA, Richards WG, et al. Adaptive resistance to therapeutic PD-1 blockade is associated with upregulation of alternative immune checkpoints. Nat Commun 2016;7:10501. [18] Gajewski TF, Schreiber H, Fu YX. Innate and adaptive immune cells in the tumor microenvironment. Nat Immunol 2013;14:1014–22. [19] Awad MM, Jones RE, Liu H, Lizotte PH, Ivanova EV, Kulkarni M, et al. Cytotoxic T cells in PD-L1-positive malignant pleural mesotheliomas are counterbalanced by distinct immunosuppressive factors. Cancer Immunol Res 2016. Nov 14. [20] Abo RP, Ducar M, Garcia EP, Thorner AR, Rojas-Rudilla V, Lin L, et al. BreaKmer: detection of structural variation in targeted massively parallel sequencing data using kmers. Nucleic Acids Res 2015;43:e19. [21] Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics 2010;26:589–95. [22] Cibulskis K, Lawrence MS, Carter SL, Sivachenko A, Jaffe D, Sougnez C, et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotechnol 2013;31:213–9. [23] McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing nextgeneration DNA sequencing data. Genome Res 2010;20:1297–303. [24] Pezzotti N, Lelieveldt B, van der Maaten L, Hollt T, Eisemann E, Vilanova A. Approximated and User Steerable tSNE for Progressive Visual Analytics. IEEE Trans Vis Comput Graph 2016 [Epub ahead of print]. [25] R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing 2013 V, Austria. ISBN 3-900051-07-0, URL
69
[29] Mandal R, Senbabaoglu Y, Desrichard A, Havel JJ, Dalin MG, Riaz N, et al. The head and neck cancer immune landscape and its immunotherapeutic implications. JCI Insight 2016;1:e89829. [30] Tumeh PC, Harview CL, Yearley JH, Shintaku IP, Taylor EJ, Robert L, et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature 2014;515:568–71. [31] Lizotte PH, Ivanova EV, Awad MM, Jones RE, Keogh L, Liu H, et al. Multiparametric profiling of non-small-cell lung cancers reveals distinct immunophenotypes. JCI Insight 2016;1:e89014. [32] Balermpas P, Rodel F, Rödel C, Krause M, Linge A, Lohaus F, et al. CD8+ tumourinfiltrating lymphocytes in relation to HPV status and clinical outcome in patients with head and neck cancer after postoperative chemoradiotherapy: a multicentre study of the German cancer consortium radiation oncology group (DKTK-ROG). Int J Cancer 2016;138:171–81. [33] Keck MK, Zuo Z, Khattri A, Stricker TP, Brown CD, Imanguli M, et al. Integrative analysis of head and neck cancer identifies two biologically distinct HPV and three non-HPV subtypes. Clin Cancer Res 2015;21:870–81. [34] Malaspina TS, Gasparoto TH, Costa MR, de Melo EF, Jr Ikoma MR, Damante JH, et al. Enhanced programmed death 1 (PD-1) and PD-1 ligand (PD-L1) expression in patients with actinic cheilitis and oral squamous cell carcinoma. Cancer Immunol Immunother 2011;60:965–74. [35] Zou W. Regulatory T cells, tumour immunity and immunotherapy. Nat Rev Immunol 2006;6:295–307. [36] Sridharan V, Margalit DN, Lynch SA, Severgnini M, Zhou J, Chau NG, et al. Definitive chemoradiation alters the immunologic landscape and immune checkpoints in head and neck cancer. Br J Cancer 2016;115:252–60. [37] Rizvi NA, Hellmann MD, Snyder A, Kvistborg P, Makarov V, Havel JJ, et al. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 2015;348:124–8. [38] Spranger S, Luke JJ, Bao R, Zha Y, Hernandez KM, Li Y, et al. Density of immunogenic antigens does not explain the presence or absence of the T-cell-inflamed tumor microenvironment in melanoma. Proc Natl Acad Sci USA 2016. Nov 11.