Current clinical knowledge surrounding one of the most promising immune checkpoint pathways, namely programmed cell death-1 (PD-1) and its ligands PD-L1 and PD-L2, is reviewed in the context of head and neck squamous cell carcinoma. The results of two phase III clinical trials (KEYNOTE 040 and CheckMate 141) are critically examined. The utility of predictive biomarkers of response to immune checkpoint blockade, such as PD-L1/PD-L2 protein expression, interferon-gamma gene expression signatures, and mutational and neoantigen load, is discussed. Finally, we project future directions in the immuno-oncology field by discussing other promising predictive biomarkers as well as areas where the next advances are likely to take place, such as in the implementation of immune checkpoint inhibitors earlier in the course of cancer treatment and/or in combination therapies.

Introduction

Tumor-infiltrating lymphocytes are thought to represent a host immune response directed against antigens expressed on tumor cells. However, despite the generation of effector immune cells targeting heterogeneous cancer cell populations, some cancers survive and grow through adaptation and natural selection of clonal cancer cell populations that are capable of evading immune recognition and clearance. Both positive and negative co-signaling pathways, termed immunologic ‘checkpoints’, regulate the effector arm of cancer immunotherapy. In healthy individuals, these effector cells are part of a well-orchestrated system that endows the host with the specificity and plasticity to mount a vigorous response to antigen challenge, while simultaneously limiting the extent and duration of the response, the latter as a means to prevent autoimmunity. Unfortunately, activating immune checkpoint clusters, in the attendance of cancer cells, can create an environment that is permissive to further cancer growth and thereby contribute to peripheral immune tolerance and tumor immune evasion [1,2]. By blocking the activation of these immune checkpoint pathways, the pendulum swings back, converting an exhausted immune response into an effective anticancer host immune response capable of eradicating the cancer cells.

Recently, programmed cell death 1 (PD-1), a critical immune checkpoint that down-regulates host immune responses, has taken center stage in the immuno-oncology field. PD-1 belongs to the CD28 family and is characterized as an inhibitory receptor expressed on T cells, dendritic cells (DCs), natural killer (NK) cells, macrophages, and B cells.

Relevance of the PD-1 : PD-L1/PD-L2 axis in cancer immunity

The importance of targeting the PD-1 receptor in cancer immunity was highlighted in early reports which demonstrated that blockade of the PD-1 receptor or its ligand, PD-L1, by specific monoclonal antibodies (mAbs) could reverse the anergic (exhausted) state of tumor-specific T cells, thereby enhancing antitumor immunity [1,3]. Subsequent clinical trials across multiple solid and hematologic tumor types solidified the importance and efficacy of targeting the PD-1 : PD-L1/PD-L2 axis in cancer therapeutics, ushering in a new era of immunotherapy. Based on the observed durable clinical response rates (RRs) achieved with these novel class of drugs, two therapeutic anti-PD1 mAbs, nivolumab and pembrolizumab, were approved for the treatment of recurrent and/or metastatic (R/M) head and neck squamous cell carcinoma (HNSCC) in 2016.

In a randomized, phase III clinical trial (CheckMate 141), 361 patients with R/M cancer of the oral cavity, pharynx, or larynx were randomized in a 2 : 1 ratio to receive nivolumab (N = 240) or the investigator's choice of one of the standard chemotherapies (cetuximab, methotrexate, or docetaxel) (N = 121). The study revealed that all patients progressed on or within 6 months of the last dose of platinum-based therapy. Most patients received ≥2 systemic therapies and over 90% had received prior radiation therapy. Nivolumab, which was administered intravenously at 3 mg/kg every 2 weeks, was found to double the 12-month overall survival (OS) rate compared with conventional chemotherapy [36% (95% CI: 28.5–43.4) versus 16.6% (95% CI: 8.6–26.8)] and reduce the risk of death by 30% in R/M HNSCC patients. Moreover, the median OS was 7.5 months (95% CI: 0.51–0.96) with nivolumab compared with 5.1 months (95% CI: 4.0–6.0) with conventional chemotherapy [hazard ratio (HR) = 0.70 (95% CI: 0.51–0.96) P = 0.0101]. The RR among nivolumab-treated patients was 13.3% (N = 32 of 240 patients, 95% CI: 9.3–18.3), including six complete responses (CRs) and 26 partial responses (PRs). In the standard-therapy group, the RR was 5.8% (N = 7 of 121 patients, 95% CI: 2.4–11.6), including one CR and six PRs [4].

In another randomized, phase III clinical trial (KEYNOTE 040), 495 patients with R/M cancer of the oral cavity, oropharynx, hypopharynx, or larynx were randomized in a 1 : 1 ratio to receive either pembrolizumab (N = 247) or the investigator's choice of conventional therapy (cetuximab, docetaxel, or methotrexate) (N = 248). All patients progressed within 3–6 months of the last dose of platinum-based therapy. Pembrolizumab was administered as a fixed dose of 200 mg every 3 weeks. The 12-month OS rate with pembrolizumab was 37.3% (N = 179) compared with 27.2% (N = 201) with the standard-of-care options (P = 0.0204). The RR with pembrolizumab was 14.6% (N = 36 of 247 patients), including four CRs and 32 PRs. With the standard-of-care treatment, the RR was 10.1% (N = 25 of 248 patients), including one CR and 24 PRs [5].

Most of the clinical responses achieved with immunotherapy fell into the categories of a PR (13%, N = 32 of 247, KEYNOTE 040) or stable disease (22.7%, N = 56 of 247, KEYNOTE 040). CR occurred at a very low frequency in both trials: 2.5% (N = 6 of 240) in the CheckMate 141 trial and 1.6% (N = 4 of 247) in the KEYNOTE 040 trial. One of the more important questions to ask is whether CR is the gold standard against which all cancer therapies should be measured? If so, is there a way to increase the frequency of CRs with immunotherapy? Alternatively, should we enrich for CRs through patient selection, by identifying the patients who are most likely to respond to immune checkpoint blockade?

Patient selection with predictive biomarkers

PD-L1 and PD-L2

Understanding tumor and/or patient characteristics that might predict who will or will not respond to a particular therapy facilitates personalized patient care and allows for the most efficient use of healthcare resources. In the first reported phase I study of anti-PD-1 therapy (nivolumab), 39 patients with solid tumors refractory to conventional therapy were treated with a single dose of anti-PD-1 antibody. There was one durable CR and two PRs. Among nine patient tumor biopsies studied, three of the four patients who exhibited membranous PD-L1 staining experienced tumor regression. The group also reported a good correlation between membranous PD-L1 expression on tumor cells and likelihood of response to treatment [6].

Subsequently, intratumoral PD-L1 expression was evaluated in a larger cohort of patients treated with nivolumab. Pre-treatment biopsies obtained from 42 patients were available to assess the role of intratumoral PD-L1 expression in the modulation of the PD-1 : PD-L1 axis. None of the 17 patients in the present study with PD-L1 (−) tumors had an objective response, whereas 9 of the 25 patients (36%) with PD-L1 expressing tumors did have an objective response (P = 0.006) to anti-PD1 monotherapy [7]. A correlation between response to PD-1 : PD-L1 targets and PD-L1 expression has also been observed in R/M HNSCC [4,8].

Both phase III clinical trials of anti-PD1 mAb therapy evaluated OS and RR using various thresholds for PD-L1 expression. KEYNOTE 040 (pembrolizumab) found that patients with a combined positive score (CPS) of ≥1% PD-L1 expression, which included both tumor and immune cells, had a median OS of 8.7 months (95% CI: 6.9–11.4), with a HR of 0.75 (95% CI: 0.59–0.95, P = 0.0078). Similarly, in the CheckMate 141 trial, patients with PD-L1 expression ≥1% on tumor cells alone had a median OS of 8.7 months (95% CI: 5.7–9.1), with a HR of 0.55 (95% CI: 0.36–0.83). KEYNOTE 040 also evaluated higher thresholds of PD-L1 expression, reporting that when the PD-L1 tumor proportion score (TPS) was ≥50%, median OS increased to 11.6 months (95% CI: 8.3–19.5), with a HR of 0.54 (95% CI: 0.35–0.82, P = 0.0017) [5]. Interestingly, the objective RRs (ORRs) according to RECIST v1.1 were also found to increase with increasing thresholds of PD-L1 expression. Specifically, the pembrolizumab-treated arm had an overall ORR of 14.6% (N = 247); patients with a PD-L1 CPS of ≥1% had an ORR of 17.3% (N = 196); and patients with a PD-L1 TPS of ≥50% had an ORR of 26.6% (N = 64) [5].

Similarly, in CheckMate 141, the ORR increased as the TPS scores increased [TPS ≥ 1% = ORR of 17% (N = 88); TPS ≥ 5% = ORR of 22%(N = 54); TPS ≥ 10% = ORR of 27% (N = 43); and TPS ≥ 50% = ORR of 36% (N = 19)] (Table 1). The ORR was lower when PD-L1 was not used as a selection biomarker (14.6% for pembrolizumab and 13.3% for nivolumab; Table 2). If PD-L1 expression is a predictive biomarker of response to anti-PD1 mAb, then one would expect the CRs and/or PRs to occur in patients with higher PD-L1 expression, and this was indeed the case. In KEYNOTE 040, overall, PR was achieved in 13% (N = 32 of 247) of patients and CR was achieved in 1.6% (N = 4 of 247) of patients. However, when utilizing the PD-L1 expression thresholds of CPS ≥ 1% and TPS ≥ 50%, the PR increased to 15.3% (N = 30) and 21.9% (N = 14), respectively, with nivolumab and the CR increased to 2.0% (N = 4) and 4.7% (N = 3), respectively, with pembrolizumab. However, since clinical responses were also observed in a subset of PD-L1-negative patients undergoing treatment with PD-1 blockade, efforts are ongoing to identify additional biomarkers of response, and current indications for treatment are not restricted by PD-L1 expression.

Table 1
Clinical response rates targeting the PD-1 : PD-L1 axis based on various PD-L1 expression thresholds in head and neck cancer patients
TargetPhasePD-L1 AbPD-L1 criteriaORROS at 12 monthsMedian OS (months)PFS
Pembrolizumab (KEYNOTE 040) PD-1 III 22C3 CPS ≥ 1% 17.3% (n = 196) 40.1% (n = 137) 8.7 29% (n = 170) 
TPS ≥ 50% 26.6% (n = 64) 46.6% (n = 41) 11.6 40% (n = 52) 
Nivolumab (CheckMate 141) PD-1 III 28-8 TPS ≥ 1% 17% (n = 88) NA 8.7 NA 
TPS ≥ 5% 22% (n = 54) NA 8.8 NA 
TPS ≥ 10% 27% (n = 43) NA 8.7 NA 
TPS ≥ 50% 36% (n = 19) NA NA NA 
Durvalumab (HAWK) PD-L1 II SP263 TC ≥ 25% 16.2% (n = 111) 33.6% (n = 112) 7.1 NA 
Atezolizumab PD-L1 Ia SP142 PD-L1 ≥ 5% on IC only 24% (n = 25) 36% (n = 32) 6.0 16% (n = 32) 
TargetPhasePD-L1 AbPD-L1 criteriaORROS at 12 monthsMedian OS (months)PFS
Pembrolizumab (KEYNOTE 040) PD-1 III 22C3 CPS ≥ 1% 17.3% (n = 196) 40.1% (n = 137) 8.7 29% (n = 170) 
TPS ≥ 50% 26.6% (n = 64) 46.6% (n = 41) 11.6 40% (n = 52) 
Nivolumab (CheckMate 141) PD-1 III 28-8 TPS ≥ 1% 17% (n = 88) NA 8.7 NA 
TPS ≥ 5% 22% (n = 54) NA 8.8 NA 
TPS ≥ 10% 27% (n = 43) NA 8.7 NA 
TPS ≥ 50% 36% (n = 19) NA NA NA 
Durvalumab (HAWK) PD-L1 II SP263 TC ≥ 25% 16.2% (n = 111) 33.6% (n = 112) 7.1 NA 
Atezolizumab PD-L1 Ia SP142 PD-L1 ≥ 5% on IC only 24% (n = 25) 36% (n = 32) 6.0 16% (n = 32) 

Abbreviations: CPS, combined positive score; TPS, tumor proportion score; NA, data not available.

Table 2
Clinical response rates independent of PD-L1 expression in head and neck cancer patients
PhaseDosingPD-L1
criteria
NORROS at 12 monthsMedian OS (months)Median time to response (months)
Pembrolizumab (KEYNOTE 040) III 200 mg q3w No 495 14.6% (n = 247) 37.3% (n = 247) 8.4 4.5 
Nivolumab (CheckMate 141) III 3 mg/kg q2w No 361 13.3% (n = 240) 34% (n = 240) 7.7 2.1 
PhaseDosingPD-L1
criteria
NORROS at 12 monthsMedian OS (months)Median time to response (months)
Pembrolizumab (KEYNOTE 040) III 200 mg q3w No 495 14.6% (n = 247) 37.3% (n = 247) 8.4 4.5 
Nivolumab (CheckMate 141) III 3 mg/kg q2w No 361 13.3% (n = 240) 34% (n = 240) 7.7 2.1 

Since PD-1 binds to both PD-L1 and PD-L2, Yearley et al. [9] investigated the expression of PD-L2 in different tumor types to evaluate the potential role it may play in predicting clinical responsiveness to anti-PD-1 therapies. This group reported that the prevalence and distribution of PD-L2 correlated significantly with PD-L1 (P = 0.0012), and PD-L2 could be detected in the absence of PD-L1 in some tumor types. In 172 R/M HNSCC patients treated with pembrolizumab (KEYNOTE 012), PD-L2 positivity was significantly associated with PD-L1 positivity (P < 0.001). The RRs in the PD-L1-positive patients (23%, 95% CI, 16.0–31.4) and PD-L2-positive patients (26.6%, 95% CI, 18.0–36.7) were higher than those in the PD-L1-negative and PD-L2-negative patients (5.9%, 95% CI, 0.1–28.7). Logistic regression analysis suggested that PD-L2 expression provided an additional predictive value for response (P = 0.038) when scoring was performed on both the tumor and immune cells when compared with the tumor cells alone. Patients who expressed both PD-L1 and PD-L2 demonstrated a two-fold higher ORR than patients whose tumor microenvironment was PD-L1-positive alone (27.5 versus 11.4%, respectively). Furthermore, PD-L2 positivity was a significant predictor of progression-free survival (PFS) (P = 0.005) and was associated with longer median PFS and OS. Thus, these findings suggest that clinical response to pembrolizumab in patients with HNSCC may be related in part to blockade of PD-1/PD-L2 interactions.

Although PD-L2 can be expressed on tumor cells, when PD-L2 was scored on tumor cells alone, the sensitivity to responders was poorer than when PD-L2 was scored on both tumor and immune cells. Interestingly, PD-L1 is expressed on tumor cells, DCs, and tissue macrophages [1,10]. PD-L2 is also expressed on tumor cells and antigen-presenting cells, such as DCs and macrophages [11]. These data highlights the importance of evaluating the myeloid population when determining responsiveness to anti-PD1 therapy. Given the predictive values of PD-L1 and PD-L2 expression within the tumor microenvironment, further studies are needed to more fully elucidate how the myeloid population may be modulated with anti-PD1 therapies.

Gene expression profiles

The up-regulation of PD-L1 and PD-L2 on tumor cells and/or immune cells within the tumor microenvironment is mediated primarily by interferon-gamma (IFN-γ) expression. Therefore, additional predictive biomarkers of response to anti-PD1 therapy would include IFN-γ expression or other inflammatory signatures. Ayers et al. [12] developed a clinical grade assay consisting of a T-cell-inflamed gene expression profile (GEP) or signature that contains IFN-γ related to antigen presentation, chemokine expression, and adaptive immune resistance, which identifies patients most likely to respond to PD-1 checkpoint blockade. The 18-gene set encompasses CD3D, CD3E, IL2RG, CXCR6, CCL5, STAT1, NKG7, HLA-E, CIITA, HLA-DRA, IDO1, LAG3, and CXCL13, GZMB, GZMK, CXCL10, CD2, TAGAP. This 18-gene T-cell-inflamed GEP signature was derived using a complex cross-validation penalized regression modeling strategy in which 220 pembrolizumab-treated patients were studied across nine different tumor types. The GEP was independently confirmed and compared with PD-L1 immunohistochemistry (IHC) in 96 HNSCC patients. With the multivariate approach, they found that utilizing the GEP increased the sensitivity to detect a response to anti-PD-1 therapy when compared with PD-L1 IHC alone. Interestingly, the gene signature was necessary, but not always sufficient to predict a clinical response to anti-PD1 mAb, suggesting that the presence of a T-cell infiltrate is not always predictive of clinical responsiveness to immune checkpoint blockade.

Important elements within the 18-gene set include markers of DCs and macrophages (STAT1, CCL5, CIITA, HLA-DRA). Thus, in the subset of inflamed tumors that do not clinically respond to anti-PD1 therapy, the status of the macrophages and/or DCs may be important in determining the responsiveness of T cells to anti-PD1 therapy.

An alternative gene expression signature that may be predictive of clinical response to immune checkpoint blockade is the immunological constant of rejection (ICR) signature. The ICR is derived from gene signatures that are activated across various immune-mediated tissue rejection models, specifically allograft rejection, graft-versus-host disease, and autoimmunity [13]. This gene signature incorporates the detection of immune effector mechanisms (GNLY, PRF1, GZMB, GZMA, and GZMH), Th1 signaling (IFNG, TBX21, CD8B, IRF1, IL12A, IL12B, and STAT1), and CXCR3/CCR5 chemokine ligands (CXCL9, CXCL10, and CCL5) [1317]. The advantage of the ICR is that it incorporates immune effector mechanisms not accounted for by the T-cell-inflamed GEP signature. In the setting of immune checkpoint blockade, the incorporation of adaptive immune suppressive mechanisms (PDL1, PD1, CTLA4, IDO1, and FoxP3) may provide a more comprehensive signature predictive of clinical responsiveness to immunotherapy. The ICR signature is currently under study as a prognostic biomarker in HNSCC samples treated with immune checkpoint blockade.

Mutational load

Another promising area of investigation for prognostic biomarkers of response to immunotherapy is assessing the tumor mutational landscape and neoantigen load. Retrospective analysis of advanced melanoma patients, treated with anti-CTLA-4 mAb, and non-small cell lung cancer (NSCLC) patients, treated with anti-PD-1 mAb, revealed that higher somatic mutational burden was significantly associated with an improved clinical response [18,19]. The predictive nature of tumor mutational load was validated in a prospective phase II study with pembrolizumab, wherein patients with mismatch repair-deficient colorectal cancer and other solid tumors had ORRs of 40 and 71%, respectively, compared with 0% RRs in mismatch repair-proficient colorectal cancer [20]. However, high mutational burden may be necessary, but not sufficient to predict response to immune checkpoint blockade. In melanoma, some patients with high mutational burden did not experience a clinical benefit with anti-CTLA-4 blockade; rather, the presence of an immunogenic neoantigen signature was more powerful [18].

Mutational load in the setting of clinical response to pembrolizumab has been studied in HNSCCs. In the HNSCC patients treated with pembrolizumab in KEYNOTE 012, it was revealed that a higher mutational load was significantly associated with response to pembrolizumab in non-viral associated HNSCCs. However, mutational load in viral associated HNSCCs (such as Epstein barr virus (EBV) and human papillomavirus (HPV)) was not associated with a response. Rather, IFN-γ GEP was predictive regardless of viral status [21].

While a higher burden of somatic mutations may lead to increased levels of neoantigens, not all neoantigens are immunogenic [22]. Retrospective analysis suggests that it is not just the high burden of neoantigens expressed by tumors alone; rather, it is the presence of ‘clonal neoantigens’, i.e. neoantigens that are expressed by all tumor cells that induce an immune response and predict greater clinical benefit from immune checkpoint blockade [23]. Interestingly, in patients with NSCLC, most of the neoantigens observed were attributed to smoking-induced mutations [22].

HPV-associated HNSCCs should theoretically harbor clonal neoantigens since the oncogenic HPV viral proteins, E6 and E7, are expressed in all cancer cells and drive the cellular malignant transformation. Based on the clonal expression of the foreign viral antigens within the tumors, these tumors are predicted to derive a greater clinical benefit from immune checkpoint blockade. Indeed, this is what has been observed to date. Although the benefits of nivolumab and pembrolizumab are observed in both HPV-associated and non-HPV associated HNSCC patients, HPV-associated HNSCC patients appear to have improved clinical RR with a median OS of 9.1 months with nivolumab versus 4.4 months with conventional treatments (HR, 0.56; 95% CI, 0.32–0.99), and in non-HPV associated HNSCC patients, the median OS was 7.5 months versus 5.8 months (HR, 0.73; 95% CI, 0.42–1.25) [4]. Similarly, in KEYNOTE 012, a 25% RR (N = 16, 95% CI, 7–52) was observed in HPV-associated HNSCCs when compared with a 14% RR (N = 29, 95% CI, 4–32) in non-HPV associated HNSCCs [8].

The story becomes even more complex when T-cell responses are considered. Stevanović et al. evaluated the antitumor T-cell responses responsible for complete regression of HPV-associated metastatic cervical cancer after tumor-infiltrating adoptive T-cell therapy. They theorized that the T cells directed against the foreign HPV tumor antigens were the ones driving effective immune responses and leading to CR. However, in the analysis of viral and nonviral antigens targeted by T cells in the patients, the immunodominant T-cell reactivities were directed against mutated neoantigens, or cancer germline antigens, rather than canonical viral antigens [24]. This report highlights the importance of not overlooking immune responses directed against nonviral antigens in virally associated cancers. Although HPV-positive HNSCCs may harbor more clonal tumor populations, based on the expression of common oncogenic drivers, such as HPV E6 and E7, the more effective cytolytic T-cell responses appear to derive from the expression of nonviral neoantigens.

Host immune responses result in an overall improved clinical prognosis

The biologic relevance of host anticancer immune responses in HNSCC is evidenced by the correlation between the degree of infiltration of CD4+ and CD8+ T cells and clinical prognosis [2528]. Since the presence of a host immune response is associated with an improved clinical prognosis, perhaps through the control of micrometastatic disease, it is important to understand whether the immune biomarkers utilized to predict clinical responsiveness to immune checkpoint blockade therapy also confer an overall improved clinical prognosis. A retrospective, multicenter, international study was performed to determine the prognostic value of tumoral PD-L1 expression in OS and PFS in patients with R/M HNSCC [29]. In this study, 412 archival tumor samples were stained for PD-L1 using the anti-PD-L1 antibody, SP263. Approximately 32% (N = 132 of 412) of HNSCC tumors had expression of PD-L1 ≥ 25% on the tumor cells, and PD-L1 expression was not prognostic of OS. However, when studying a different sample set of 214 HNSCCs, using different cut-offs for PD-L1 expression scored on both tumor and immune cells, PD-L1 expression on tumor cells of ≥1% was associated with improved OS [30]. Thus, our understanding of PD-L1 expression as a prognostic biomarker in HNSCC is still emerging and similar analyses have yet to be performed for PD-L2 or IFN-γ GEP. When applying various biomarkers of response to immunotherapy, the prognostic value of the same biomarkers also needs to be understood and validated in the context of non-immunotherapeutics.

Future directions

The bulk of PD-1 expression is derived from CD4+ T helper (Th) cells rather than cytotoxic CD8+ T cells in HNSCC, and the CD4+Th PD-1+ T cells co-localize with PD-L1+ tumor-associated macrophages [31]. Thus, anti-PD1 therapy in HNSCC has a greater impact on modulating the CD4+ Th population which cross-talks with the macrophages and/or DCs and modulates the activity of the cytotoxic CD8+ T cells. To date, the majority of research has focused on understanding the cytotoxic CD8+PD-1+ T cells, while little effort has been devoted to understanding how CD4+ PD-1+ Th cells are affected by anti-PD1 therapy. Given the higher frequency of CD4+PD-1+ Th cells in the tumor microenvironment, it may be that CD4+ T cells are the important regulators of overall responsiveness to immune checkpoint blockade. More research efforts need to be directed towards understanding how this T-cell population is modulated by anti-PD-1 therapy.

One strategy for increasing the frequency of CRs to immune checkpoint blockade is to administer these agents earlier in the course of cancer treatment when the majority of the tumors are of the immune-active phenotype. In both phase III clinical trials, anti-PD1 mAb was administered in the setting of R/M disease. However, there is evidence to suggest that administering immunotherapy in the newly diagnosed cancer setting may yield higher RRs. In a clinical trial opened at the Washington University School of Medicine, St Louis and the Dana-Farber Cancer Institute, pembrolizumab is being administered in the neoadjuvant setting for locally advanced, surgically resectable head and neck cancer patients (www.clinicaltrials.gov; NCT02296684). In addition, based on the presence of high-risk features in the pathology specimen (such as positive microscopic margins or extracapsular extension), patients receive pembrolizumab in conjunction with standard-of-care postoperative adjuvant chemoradiotherapy. The goal of the trial is to evaluate whether immunotherapy can improve locoregional recurrence and/or distant metastatic rates in high-risk patients with locally advanced HNSCCs treated with standard-of-care surgical approaches. A pathologic treatment response, as defined by tumor cell necrosis and/or giant cell/histiocytic reaction to keratinous debris, was observed in 42% (10 of 24) of patients, and a tumor response, as assessed by clinical examination, imaging, and/or pathology, was observed in 50% (12/24) of patients with a single dose of neoadjuvant pembrolizumab [32]. It was also reported that none of the 14 patients followed for 1 year had locoregional recurrence or distant metastatic disease, which historically has occurred in 35% of HNSCC patients. Based on these promising results, a phase III randomized clinical trial, evaluating pembrolizumab as neoadjuvant therapy in the resectable locoregionally advanced HNSCC patient population, is being launched. With an increasing number of patients receiving immunotherapy in the neoadjuvant setting and with longer follow-up, the ability of immunotherapy to achieve higher RRs will be better understood.

Combinatorial strategies with immunotherapy are also being actively explored to improve clinical responses. Since the presence of a T-cell-inflamed phenotype is necessary, but not sufficient for clinical responses to PD-1 checkpoint blockade, this finding suggests that distinct resistance mechanisms may exist for tumors that lack T-cell inflammation or an ‘immune-deserted’ tumor when compared with an activated T-cell infiltrate or ‘immune-active’ tumor that is not responsive to anti-PD-1 therapy. Since these two different tumor immune landscapes would respond differently to various combinatorial strategies, there is a need to tailor combinatorial strategies based on the pre-existing tumor landscape harbored by the patient.

Immunotherapy agents are also now being evaluated in the pre-cancerous setting. Specifically, several clinical trials are evaluating the efficacy of pembrolizumab for the management of premalignant oral lesions (NCT02882282) as well as in HPV-associated benign respiratory papillomas, which have a risk of malignant transformation (NCT02632344). As clinicians continue to gain experience with this novel class of drugs, the adverse profile with immunotherapy has become generally manageable. The immuno-oncology field will continue to evaluate and re-assess the application of immunotherapy in the context of cancer as the results of clinical trials continue to mature. It is possible the full potential of immunotherapy will be realized in the early stage setting and/or in its prevention.

Summary
  • Two phase III clinical trials targeting the PD-1 : PD-L1/PD-L2 axis have demonstrated in improved overall survival in recurrent/metastatic head and neck squamous cell carcinoma patients.

  • Complete responses are infrequent with current immunotherapies.

  • Biomarkers may help to identify patients most likely to derive clinical benefit from immune checkpoint blockade.

  • Current transcriptional based biomarkers are necessary, but not always sufficient to predict a clinical response, suggesting that other mechanisms of resistance may be activated within the T-cell-inflamed tumor microenvironment.

  • Immunotherapeutic agents administered earlier in the course of cancer therapy and/or combinatorial strategies may improve clinical response rates and result in durable treatment effects through the induction of memory T cells.

Abbreviations

     
  • CPS

    combined positive score

  •  
  • CRs

    complete responses

  •  
  • DC

    dendritic cells

  •  
  • EBV

    Epstein barr virus

  •  
  • GEP

    gene expression profile

  •  
  • HNSCC

    head and neck squamous cell carcinoma

  •  
  • HPV

    human papillomavirus

  •  
  • HR

    hazard ratio

  •  
  • ICR

    immunological constant of rejection

  •  
  • IFN-γ

    interferon-gamma

  •  
  • IHC

    immunohistochemistry

  •  
  • mAbs

    monoclonal antibodies

  •  
  • NK

    natural killer

  •  
  • NSCLC

    non-small cell lung cancer

  •  
  • ORRs

    objective RRs

  •  
  • OS

    overall survival

  •  
  • PD-1

    programmed cell death-1

  •  
  • PD-L1 and PD-L2

    PD-1 and its ligands

  •  
  • PFS

    progression-free survival

  •  
  • PRs

    partial responses

  •  
  • R/M

    recurrent and/or metastatic

  •  
  • RR

    response rate

  •  
  • SD

    stable disease

  •  
  • Th

    T helper

  •  
  • TPS

    tumor proportion score

Funding

This work was supported by National Institutes of Health/National Institute of Dental and Craniofacial Research [R01DE025340] (S.I.P.).

Acknowledgments

We thank all of those head and neck cancer patients who have participated in these and other HNSCC clinical trials to improve patient care for the next generation.

Competing Interests

S.I.P. receives research support from Abbvie, Astrazeneca/MedImmune, Cue, and Tesaro and participates in investigator-initiated immunotherapy clinical trials that are supported by Merck, Astrazeneca/MedImmune, and Aperisys. S.I.P. serves on advisory boards for Abbvie, Merck, and Astrazeneca/MedImmune. L.J.W. serves on advisory boards for Merck and Novartis.

References

References
1
Dong
,
H.
,
Strome
,
S.E.
,
Salomao
,
D.R.
,
Tamura
,
H.
,
Hirano
,
F.
,
Flies
,
D.B.
et al
(
2002
)
Tumor-associated B7-H1 promotes T-cell apoptosis: a potential mechanism of immune evasion
.
Nat. Med.
8
,
1039
2
Pardoll
,
D.M.
(
2012
)
The blockade of immune checkpoints in cancer immunotherapy
.
Nat. Rev. Cancer
12
,
252
264
3
Dong
,
H.
,
Strome
,
S.E.
,
Matteson
,
E.L.
,
Moder
,
K.G.
,
Flies
,
D.B.
,
Zhu
,
G.
et al
(
2003
)
Costimulating aberrant T cell responses by B7-H1 autoantibodies in rheumatoid arthritis
.
J. Clin. Invest.
111
,
363
370
4
Ferris
,
R.L.
,
Blumenschein
, Jr,
G.
,
Fayette
,
J.
,
Guigay
,
J.
,
Colevas
,
A.D.
,
Licitra
,
L.
et al
(
2016
)
Nivolumab for recurrent squamous-cell carcinoma of the head and neck
.
N. Engl. J. Med.
375
,
1856
1867
5
Cohen
,
E.E.
,
Harrington
,
K.J.
,
Tourneau
,
C.L.
,
Dinis
,
J.
,
Licitra
,
L.
,
Ahn
,
M.J.
et al
(
2017
)
Pembrolizumab vs standard of care for recurrent or metastatic head and neck squamous cell carcinoma: Phase 3 Keynote 040 Trial
.
ESMO Meeting
,
Madrid, Spain
,
September 8–12
, LBA45
6
Brahmer
,
J.R.
,
Drake
,
C.G.
,
Wollner
,
I.
,
Powderly
,
J.D.
,
Picus
,
J.
,
Sharfman
,
W.H.
et al
(
2010
)
Phase I study of single-agent anti-programmed death-1 (MDX-1106) in refractory solid tumors: safety, clinical activity, pharmacodynamics, and immunologic correlates
.
J. Clin. Oncol.
28
,
3167
3175
7
Segal
,
N.H.
,
Antonia
,
S.J.
,
Brahmer
,
J.R.
,
Maio
,
M.
,
Blake-Haskins
,
A.
,
Li
,
X.
et al
(
2015
)
Preliminary data from a multi-arm expansion study of MEDI4736, an anti-PD-L1 antibody
.
ASCO Meeting
,
Chicago, Illinois, May 29–June 2
,
33
,
3011
8
Seiwert
,
T.Y.
,
Burtness
,
B.
,
Mehra
,
R.
,
Weiss
,
J.
,
Berger
,
R.
,
Eder
,
J.P.
et al
(
2016
)
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.
17
,
956
965
9
Yearley
,
J.H.
,
Gibson
,
C.
,
Yu
,
N.
,
Moon
,
C.
,
Murphy
,
E.
,
Juco
,
J.
et al
(
2017
)
PD-L2 expression in human tumors: relevance to anti-PD-1 therapy in cancer
.
Clin. Cancer Res.
23
,
3158
3167
10
Keir
,
M.E.
,
Butte
,
M.J.
,
Freeman
,
G.J.
and
Sharpe
,
A.H.
(
2008
)
PD-1 and its ligands in tolerance and immunity
.
Annu. Rev. Immunol.
26
,
677
704
11
Rozali
,
E.N.
,
Hato
,
S.V.
,
Robinson
,
B.W.
,
Lake
,
R.A.
and
Lesterhuis
,
W.J.
(
2012
)
Programmed death ligand 2 in cancer-induced immune suppression
.
Clin. Dev. Immunol.
2012
,
1
8
12
Ayers
,
M.
,
Lunceford
,
J.
,
Nebozhyn
,
M.
,
Murphy
,
E.
,
Loboda
,
A.
,
Kaufman
,
D.R.
et al
(
2017
)
IFN-γ–related mRNA profile predicts clinical response to PD-1 blockade
.
J. Clin. Invest.
127
,
2930
2940
13
Wang
,
E.
,
Worschech
,
A.
and
Marincola
,
F.M.
(
2008
)
The immunologic constant of rejection
.
Trends Immunol.
29
,
256
262
14
Wang
,
E.
,
Tomei
,
S.
and
Marincola
,
F.M.
(
2012
)
Reflections upon human cancer immune responsiveness to T cell-based therapy
.
Cancer Immunol. Immunother.
61
,
761
770
15
Galon
,
J.
,
Angell
,
H.K.
,
Bedognetti
,
D.
and
Marincola
,
F.M.
(
2013
)
The continuum of cancer immunosurveillance: prognostic, predictive, and mechanistic signatures
.
Immunity
39
,
11
26
16
Bedognetti
,
D.
,
Hendrickx
,
W.
,
Ceccarelli
,
M.
,
Miller
,
L.D.
and
Seliger
,
B.
(
2016
)
Disentangling the relationship between tumor genetic programs and immune responsiveness
.
Curr. Opin. Immunol.
39
,
150
158
17
Hendrickx
,
W.
,
Simeone
,
I.
,
Anjum
,
S.
,
Mokrab
,
Y.
,
Bertucci
,
F.
,
Finetti
,
P.
et al
(
2017
)
Identification of genetic determinants of breast cancer immune phenotypes by integrative genome-scale analysis
.
Oncoimmunology
6
,
e1253654
18
Snyder
,
A.
,
Makarov
,
V.
,
Merghoub
,
T.
,
Yuan
,
J.
,
Zaretsky
,
J.M.
,
Desrichard
,
A.
et al
(
2014
)
Genetic basis for clinical response to CTLA-4 blockade in melanoma
.
N. Engl. J. Med.
371
,
2189
2199
19
Rizvi
,
N.A.
,
Hellmann
,
M.D.
,
Snyder
,
A.
,
Kvistborg
,
P.
,
Makarov
,
V.
,
Havel
,
J.J.
et al
(
2015
)
Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer
.
Science
348
,
124
128
20
Le
,
D.T.
,
Uram
,
J.N.
,
Wang
,
H.
,
Bartlett
,
B.R.
,
Kemberling
,
H.
,
Eyring
,
A.D.
et al
(
2015
)
PD-1 blockade in tumors with mismatch-repair deficiency
.
N. Engl. J. Med.
372
,
2509
2520
21
Haddad
,
R.
,
Seiwert
,
T.Y.
,
Chow
,
L.Q.
,
Gupta
,
S.
,
Weiss
,
J.
,
Gluck
,
I.
et al
(
2017
)
Genomic determinants of response to pembrolizumab in head and neck squamous cell carcinoma
.
ASCO Meeting
,
Chicago, Illinois
,
June 2–6
.
22
Segal
,
N.H.
,
Parsons
,
D.W.
,
Peggs
,
K.S.
,
Velculescu
,
V.
,
Kinzler
,
K.W.
,
Vogelstein
,
B.
et al
(
2008
)
Epitope landscape in breast and colorectal cancer
.
Cancer Res.
68
,
889
892
23
McGranahan
,
N.
,
Furness
,
A.J.S.
,
Rosenthal
,
R.
,
Ramskov
,
S.
,
Lyngaa
,
R.
,
Saini
,
S.K.
et al
(
2016
)
Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade
.
Science
351
,
1463
1469
24
Stevanović
,
S.
,
Pasetto
,
A.
,
Helman
,
S.R.
,
Gartner
,
J.J.
,
Prickett
,
T.D.
,
Howie
,
B.
et al
(
2017
)
Landscape of immunogenic tumor antigens in successful immunotherapy of virally induced epithelial cancer
.
Science
356
,
200
205
25
Gooden
,
M.J.M.
,
de Bock
,
G.H.
,
Leffers
,
N.
,
Daemen
,
T.
and
Nijman
,
H.W.
(
2011
)
The prognostic influence of tumour-infiltrating lymphocytes in cancer: a systematic review with meta-analysis
.
Br. J. Cancer
105
,
93
103
26
Badoual
,
C.
,
Hans
,
S.
,
Merillon
,
N.
,
Van Ryswick
,
C.
,
Ravel
,
P.
,
Benhamouda
,
N.
et al
(
2013
)
PD-1-expressing tumor-infiltrating T cells are a favorable prognostic biomarker in HPV-associated head and neck cancer
.
Cancer Res.
73
,
128
138
27
Nordfors
,
C.
,
Grun
,
N.
,
Tertipis
,
N.
,
Ährlund-Richter
,
A.
,
Haeggblom
,
L.
,
Sivars
,
L.
et al
(
2013
)
CD8+ and CD4+ tumour infiltrating lymphocytes in relation to human papillomavirus status and clinical outcome in tonsillar and base of tongue squamous cell carcinoma
.
Eur. J. Cancer
49
,
2522
2530
28
Ward
,
M.J.
,
Thirdborough
,
S.M.
,
Mellows
,
T.
,
Riley
,
C.
,
Harris
,
S.
,
Suchak
,
K.
et al
(
2014
)
Tumour-infiltrating lymphocytes predict for outcome in HPV-positive oropharyngeal cancer
.
Br. J. Cancer
110
,
489
500
29
Pai
,
S.I.
,
Cohen
,
E.E.
,
Lin
,
D.
,
Fountzilas
,
G.
,
Kim
,
E.
,
Mehlhorn
,
H.
et al
(
2017
)
Retrospective cohort study of PD-L1 expression in recurrent and/or metastatic squamous cell carcinoma of the head and neck (SUPREME-HN)
.
ESMO Meeting
,
Madrid, Spain
,
September 8–12
, P1048
30
Stokes
,
M.
,
Wang
,
R.
,
Wildsmith
,
S.
,
Secrier
,
M.
,
Angell
,
H.K.
,
Barker
,
C.
et al
(
2017
)
Relationship between PD-L1 expression and survival in head and neck squamous cell carcinoma (HNSCC) patients
.
ESMO Meeting
,
Madrid, Spain
,
September 8–12
, P1049
31
Mattox
,
A.
,
Lee
,
J.
,
Westra
,
W.H.
,
Pierce
,
R.H.
,
Ghoseein
,
R.
,
Faquin
,
W.C.
et al
(
2017
)
PD-1 expression in head and neck squamous cell carcinomas derives primarily from functionally anergic CD4+ TILs in the presence of PD-L1+ TAMs
.
Cancer Res.
77
,
6365
6374
32
Uppaluri
,
R.
and
Adkins,
D.
(
2017
)
Neoadjuvant pembrolizumab in surgically resectable, HPV negative locally advanced head and neck squamous cell carcinoma (HNSCC)
.
ASCO Meeting
,
Chicago, Illinois
,
June 2–6
.