Purpose:

Patients with resected, local–regionally advanced, head and neck squamous cell carcinoma (HNSCC) have a one-year disease-free survival (DFS) rate of 65%–69% despite adjuvant (chemo)radiotherapy. Neoadjuvant PD-1 immune-checkpoint blockade (ICB) has demonstrated clinical activity, but biomarkers of response and effect on survival remain unclear.

Patients and Methods:

Eligible patients had resectable squamous cell carcinoma of the oral cavity, larynx, hypopharynx, or oropharynx (p16-negative) and clinical stage T3-T4 and/or two or more nodal metastases or clinical extracapsular nodal extension (ENE). Patients received neoadjuvant pembrolizumab 200 mg 1–3 weeks prior to surgery, were stratified by absence (intermediate-risk) or presence (high-risk) of positive margins and/or ENE, and received adjuvant radiotherapy (60–66 Gy) and concurrent pembrolizumab (every 3 weeks × 6 doses). Patients with high-risk HNSCC also received weekly, concurrent cisplatin (40 mg/m2). Primary outcome was one-year DFS. Secondary endpoints were one-year overall survival (OS) and pathologic response (PR). Safety was evaluated with CTCAE v5.0.

Results:

From February 2016 to October 2020, 92 patients enrolled. The median age was 59 years (range, 27–80), 30% were female, 86% had stage T3–T4, and 69% had ≥N2. At a median follow-up of 28 months, one-year DFS was 97% (95% CI, 71%–90%) in the intermediate-risk group and 66% (95% CI, 55%–84%) in the high-risk group. Patients with a PR had significantly improved one-year DFS relative to patients without response (93% vs. 72%, hazard ratio 0.29; 95% CI, 11%–77%). No new safety signals were identified.

Conclusions:

Neoadjuvant and adjuvant pembrolizumab increased one-year DFS rate in intermediate-risk, but not high-risk, HNSCC relative to historical control. PR to neoadjuvant ICB is a promising surrogate for DFS.

Translational Relevance

Patients with resected, local–regionally advanced, head and neck squamous cell carcinoma (HNSCC) have a poor disease-free survival (DFS) despite adjuvant (chemo)radiotherapy. Neoadjuvant PD-1 immune-checkpoint blockade (ICB) has demonstrated clinical activity in this patient population, but biomarkers of response and effect on survival remain unclear. Compared with historical controls, the addition of pembrolizumab improved the one-year DFS rate in intermediate-risk, but not high-risk, resected HNSCC. Importantly, pathologic response (PR) to neoadjuvant pembrolizumab was associated with PD-L1 status and significantly higher DFS. Therefore, PR to neoadjuvant PD-1 ICB is a likely surrogate for DFS in resectable HNSCC. The effect of neoadjuvant and adjuvant PD-1 ICB on survival of local–regionally advanced, resectable, HNSCC and associated biomarkers of PR warrants investigation in a randomized clinical trial.

Approximately 745,000 new cases and 364,340 deaths from head and neck squamous cell carcinoma (HNSCC) occurred in 2020 (1). A majority of patients present with local–regionally advanced HNSCC for which the current standard of care is surgical resection followed by adjuvant (chemo)radiation (RT). Two randomized control trials demonstrated increased disease-free survival (DFS) in resectable HNSCC when cisplatin was administered concurrent with adjuvant RT versus RT alone, albeit with higher toxicity (2, 3). A comparative analysis showed the survival benefit was restricted to patients with high-risk pathologic features of positive margins and/or extracapsular nodal extension (ENE; ref. 4). NCCN guidelines therefore recommend chemoRT for the high-risk group and RT alone in those with intermediate-risk features (e.g., multiple lymph node involvement, perineural invasion, lymphovascular invasion, T3 or T4 stage). Unfortunately, risk of progression remains high in both groups: in RTOG 9501, one-year DFS was 65% for the high-risk group with chemoRT and 69% in the intermediate-risk group with RT alone [provided by Jonathan Harris (ACR)]. Novel treatments are necessary to improve outcomes in these patient populations.

A majority of HNSCCs express the inhibitory immune-checkpoint programmed death ligand-1 (PD-L1), which binds to the PD-1 receptor expressed on cytotoxic T cells to suppress their function. PD-1 immune-checkpoint blockade (ICB) with pembrolizumab or nivolumab increased overall survival (OS) in patients with recurrent/metastatic or platinum-refractory HNSCC, respectively (5, 6). In murine HNSCC xenograft models, upregulation of PD-L1 was observed in response to RT, and the addition of PD-L1 ICB to RT led to superior survival (7). Therefore, we hypothesized that the addition of pembrolizumab to adjuvant RT could improve survival in HNSCC patients after primary surgical resection and that neoadjuvant pembrolizumab would allow for interrogation of potential correlative biomarkers of response.

We performed a multicenter phase II window-of-opportunity clinical trial (NCT02641093) to estimate one-year DFS in patients with local–regionally advanced, resectable, HNSCC when neoadjuvant and adjuvant pembrolizumab was added to standard-of-care (chemo)RT. Secondary outcomes included OS, toxicity, pathologic response (PR), and tumor immune microenvironmental changes after neoadjuvant pembrolizumab.

Study design and treatment

The study was designed as a multicenter, open-label, nonrandomized, two-arm phase II trial of the addition of neoadjuvant and adjuvant pembrolizumab with or without concurrent cisplatin in patients with surgically resectable previously untreated, local–regionally advanced HNSCC. Patients and investigators were not masked. The trial was registered on clinicaltrials.gov (NCT02641093). The study was approved by the institutional review boards of all participating sites and was conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki. All patients were required to sign written informed consent. The study was monitored by the University of Cincinnati Cancer Center Data Safety Monitoring Board.

Eligibility criteria included: age ≥ 18 years; newly diagnosed histologically or cytologically confirmed HNSCC; local–regionally advanced-stage III/IV AJCC 8th edition T3 or T4 or ≥N2 disease or clinical evidence of ENE on diagnostic imaging; disease was determined resectable by the treating head and neck surgeon with no involvement of skull base or T4b stage; Eastern Cooperative Oncology Group (ECOG) performance status ≤ 1; and adequate organ function. Key exclusion criteria included: human papillomavirus (HPV+) oropharyngeal cancer (HPV-positive disease outside of oropharynx was allowed, but testing was not required); nasopharyngeal cancer; metastatic disease determined by chest CT and/or PET/CT; autoimmune disease; active intercurrent illness (e.g., significant cardiovascular disease, viral infections, or major psychiatric illness); and steroid use (> prednisone 10 mg daily). Please refer to full eligibility criteria in the Supporting Information. Full clinical trial protocol is available upon request.

All patients received one dose of pembrolizumab (200 mg i.v.) 7–21 days prior to surgery (schema; Supplementary Fig. S1A). After surgery, patients were stratified into two arms based on the presence or absence of high-risk pathologic features (e.g., positive margins and/or ENE). Patients received adjuvant pembrolizumab 200 mg i.v. every 3 weeks for a total of up to six doses starting 1 week prior to intensity-modulated RT to 60–66 Gy in 30–33 fractions over 6 weeks. High-risk patients received concurrent weekly cisplatin (40 mg/m2) for up to six doses during radiation. Dose reductions, modifications, and/or interruptions were performed per standard of care for cisplatin and investigational brochure for pembrolizumab.

The first 8 patients in the intermediate- and high-risk groups were enrolled in a safety lead-in designed to investigate dose-limiting toxicity (DLT), defined as an adverse event attributable to pembrolizumab that resulted in a delay in initiation of standard-of-care treatment (e.g., >7 days for surgery, >3 days for RT, >7 days for cisplatin). The study arm was to be discontinued if more than 2 of 8 patients developed a DLT. The primary endpoint was one-year DFS and secondary endpoints included OS and change in the tumor immune microenvironment after neoadjuvant pembrolizumab.

Pathologic analyses

The resected tumor specimen was evaluated by pathologists as part of the standard of care for the presence of pathologic risk features, including ENE, margin status, lymphovascular invasion (LVI), perineural invasion (PNI), and number of involved lymph nodes.

Pathologic treatment effect (TE) and PD-L1 status were determined on central review by a board-certified pathologist at the University of Cincinnati (BH). TE was defined as tumor showing necrosis with associated histiocytic inflammation and/or giant cell reaction to keratinaceous debris. TE percentage was determined by dividing estimates of the total area showing the latter features by the total area showing residual viable tumor and TE. Based on the percentage of TE seen, PR was divided into no (NPR, <20%), partial (PPR, ≥20% and <90%), or major (MPR, ≥90%) PR. PR was also determined for those patients with positive lymph nodes and tissue available for analysis.

PD-L1 expression was evaluated using the 22C3 antibody clone (Agilent, Dako) and pharmDx IHC assay at either Caris Life Technologies or Neogenomics and confirmed by BH. PD-L1 expression on both tumor and tumor-infiltrating immune cells was evaluated. For tumor cells, membranous staining of any intensity was considered positive. Only inflammatory cells infiltrating invasive tumor and in adjacent intra- and peritumoral stroma were scored. Combined positivity score (CPS) was calculated by summing the numbers of PD-L1–positive tumor cells and immune cells and dividing by the total number of viable tumor cells (8).

mRNA expression analysis

Paired formalin-fixed and paraffin-embedded tumor specimens from before and after neoadjuvant pembrolizumab were selected from the first 23 available patients with PPR/MPR (n = 11) versus NPR (n = 12). Purified total RNA was evaluated by a hybridization-based digital counting assay (NanoString nCounter platform) using the PanCancer IO 360 panel (for research use only) that measures 770 immune-related genes and controls. Raw data counts were normalized using the geometric mean of the housekeeping genes, and each gene was adjusted based on IO360 panel standards, and normalized data were log2 transformed. Gene-expression signature scores for 48 signatures measuring immune cell abundance, immune signaling, and tumor and stromal biology (9) were calculated as weighted averages of the signature genes with a signature-specific constant added to express values in a similar range (10). Signature scores were compared between pre- and post-pembrolizumab treatment and between those with PPR/MPR and NPR.

Statistical analysis

The all-patients-as-treated (APat) population including any patient receiving at least one dose of pembrolizumab was used for safety analysis. Safety endpoints included all adverse events (AE) graded 1 to 5 per CTCAE v5.0. The proportion of AEs among patients treated with pembrolizumab and RT and pembrolizumab and chemoRT were compared with historical control patients treated with RT only and chemoRT, respectively. Descriptive analysis of grade ≥3 AEs were compared with data from RTOG 9501 (e.g., 46% for RT and 78% for chemoRT; ref. 3).

All patients who received ≥1 dose of adjuvant pembrolizumab were considered evaluable for efficacy per protocol and served as the primary analysis population in this study. Patient attrition is included in the consort diagram in Supplementary Fig. S1. The primary efficacy endpoint was one-year DFS, defined as time from treatment allocation to documented relapse or death. Key secondary endpoints include one-year and two-year OS, defined as time from treatment allocation to death due to any cause. With an expected sample size of approximately N = 40 in both risk groups, we expected to reach over 80% power to detect an increase of 19%–21% in DFS at year 1 with the addition of pembrolizumab. Kaplan–Meier method was used to estimate survival rates for one-year and two-year DFS and OS. The survival rates were compared with historical censored data (RTOG 9501) using the log-rank test. We used estimates of the survival probabilities at specific time points of the RTOG 9501 trial for DFS in order to compare our results. All data and events up to 4 years of follow-up in RTOG 9501 were used to be consistent with our study. We repeated these analyses by limiting the follow-up to one year and censoring all patients who recurred after that time. The hazards were compared using Wald test in a Cox proportional hazards model.

Statistical analysis of gene-expression signature scores was performed using the Empirical Bayes Linear Model as implemented in the limma R package (11). Signature scores at baseline and difference in score before and after neoadjuvant pembrolizumab were compared in patients with PPR/MPR versus NPR. For the baseline comparison, the signature scores were normalized by subtracting the median score for each sample, and were then compared using Empirical Bayes two-sample t test. For the comparison of differences, the patient-level differences were calculated by subtracting the before scores from after scores for each patient separately, and differences were compared using the Empirical Bayes two-sample t test. Gene signature differences with FDR (false discovery rate)-adjusted (12) P values less than 0.1 were considered statistically significant.

The PD-L1 expression was categorized as 0, 1–19, and ≥20 for CPS and 0, 1–49, and ≥50 for TPS, and associations with PR, DFS, and OS were evaluated. Fisher exact test was used to compare associations between PD-L1 expression and PR. Kaplan–Meier estimates and log-rank test were used to determine difference in survival probabilities for DFS and OS.

Raw data for mRNA signatures were generated in a core facility (NanoString). Patient-level data for biosignatures are provided in Supplementary Table S9. All other data were generated by the authors and included in the article or supplementary data.

Patient population

Between January 2016 and October 2020, 92 patients were enrolled (Table 1). Median age was 59 years (27–80). The majority were male (70%), White (95%), diagnosed with oral cavity cancer (86%), had T3–T4 stage (86%), and cervical nodal metastases (81%). No patients were enrolled based on clinical ENE alone. Four patients did not proceed with surgical resection: two due to rapid progression of disease, one due to presence of unresectable disease at the skull base that was not identified on preoperative scans, and one patient due to withdrawal of consent (Supplementary Fig. S1B). Of the 92 APaT patients evaluable for toxicity and who received at least a single dose of neoadjuvant pembrolizumab, 42 patients were in the intermediate-risk and 50 were in the high-risk group (including the 4 patients who did not proceed with surgery). Of the resected patients, 75 (31 intermediate-risk and 44 high-risk) patients received adjuvant treatment and therefore were evaluable for efficacy per protocol. The most common reasons for not proceeding with adjuvant treatment per protocol included identification of secondary malignancy at time of surgery (thyroid cancer), and withdrawal of consent (Supplementary Fig. S1B).

Table 1.

Baseline patient and disease characteristics (n = 92).

Patient characteristicsN (%)
Median age, year (range) 59 (27–80) 
Sex 
 Male 64 (70) 
 Female 28 (30) 
Race 
 White 87 (95) 
 African American 2 (2) 
 Unknown/Other 3 (3) 
Smoking history (> 10pk per year) 
 Yes 59 (64) 
Alcohol history (> 5 drinks per week) 
 Yes 41 (45) 
Primary disease site 
 Larynx 10 (11) 
 Oral cavity 79 (86) 
 Oropharynx 2 (2) 
 Hypopharynx 1 (1) 
Tumor classification 
 Tx 1 (1) 
 T1 3 (3) 
 T2 9 (10) 
 T3 18 (20) 
 T4 61 (66) 
Lymph node classification 
 Nx 1 (1) 
 N0 16 (17) 
 N1 11 (12) 
 N2 60 (65) 
 N3 4 (4) 
ECOG performance status 
 0 55 (60) 
 1 37 (40) 
Patient characteristicsN (%)
Median age, year (range) 59 (27–80) 
Sex 
 Male 64 (70) 
 Female 28 (30) 
Race 
 White 87 (95) 
 African American 2 (2) 
 Unknown/Other 3 (3) 
Smoking history (> 10pk per year) 
 Yes 59 (64) 
Alcohol history (> 5 drinks per week) 
 Yes 41 (45) 
Primary disease site 
 Larynx 10 (11) 
 Oral cavity 79 (86) 
 Oropharynx 2 (2) 
 Hypopharynx 1 (1) 
Tumor classification 
 Tx 1 (1) 
 T1 3 (3) 
 T2 9 (10) 
 T3 18 (20) 
 T4 61 (66) 
Lymph node classification 
 Nx 1 (1) 
 N0 16 (17) 
 N1 11 (12) 
 N2 60 (65) 
 N3 4 (4) 
ECOG performance status 
 0 55 (60) 
 1 37 (40) 

Survival outcomes

At a median follow-up of 28 months among all evaluable patients (N = 75), the one-year DFS rate was 80% (95% CI, 71%–90%; Fig. 1A) with a hazard ratio of 0.60 (95% CI, 0.39–0.93; P = 0.0233), which was significantly higher when compared with the entire cohort from RTOG 9501. Forty-four (58%) had high-risk pathologic features: 39 (52%) had ENE and 17 (23%) had positive margins (Supplementary Table S1). One-year DFS was 96% (95% CI, 90%–100%) for the intermediate-risk group (N = 31). This was significantly higher than the one-year DFS of 69% (95% CI, 59%–78%) observed in the intermediate-risk group treated with RT alone in RTOG 9501 (P = 0.0007; Fig. 1B) with a hazard ratio of 0.23 (95% CI, 0.09–0.58; P = 0.0018). One-year OS was also higher relative to historical control in the intermediate-risk group (Fig. 1C). Similar results were seen when comparing the intent-to-treat (ITT) population (N = 96; Supplementary Fig. S2).

Figure 1.

Survival stratified by pathologic adverse features. KM curves representing all patients DFS (A), as well as DFS (B) and OS (C) stratified by high- and intermediate-risk disease. P value by KM method provided. Hazard ratios (HR) were calculated by comparing high risk to intermediate risk.

Figure 1.

Survival stratified by pathologic adverse features. KM curves representing all patients DFS (A), as well as DFS (B) and OS (C) stratified by high- and intermediate-risk disease. P value by KM method provided. Hazard ratios (HR) were calculated by comparing high risk to intermediate risk.

Close modal

In contrast, one-year DFS rate in the high-risk group was 66% (95% CI, 55%–84%), which was similar to that observed for this population in RTOG 9501 treated with chemoRT (65%; 95% CI, 57%–64%; Fig 1B) with a hazard ratio of 0.86 (95% CI, 0.52–1.44; P = 0.5736). Among the high-risk group, DFS was not significantly different among patients stratified by the presence of ENE and/or positive margins (Supplementary Fig. S3).

Analysis of PR

Representative images of primary tumors from each PR group are shown in Fig. 2A. PR (PPR/MPR) was observed in 39% of evaluable patients, and MPR in 7% (Fig. 2B). Rates of PR were higher in the intermediate-risk group (55%) than in the high-risk group (28%; Fig. 2B and C). Of those patients with primary site and lymph node (LN) resection tissue available for evaluation, 23 of 28 (82%) patients had concordance of PR between both primary site and LNs (Supplementary Table S2). The other 5 patients had a PR in primary site but not LNs. In general, PR was lower in LNs compared with primary site.

Figure 2.

Pathologic responders have increased survival. A, Representative H&E pictures of patients with no (NPR), partial (PPR), and major (MPR) PR characterized by <20%, 20%–90%, and ≥90% TE, respectively. Images were all at 200×. TE was defined as tumor necrosis with associated histiocytic inflammation and/or giant cell reaction to keratinaceous debris. TE percentage was determined by dividing estimates of area showing these features by the total area showing residual viable (VT) and TE. B, Proportion of patients with NPR, PPR, or MPR. C, Percent treatment effect in intermediate- and high-risk patients. Survival curves comparing NPR and PPR/MPR for DFS (D) and OS (E).

Figure 2.

Pathologic responders have increased survival. A, Representative H&E pictures of patients with no (NPR), partial (PPR), and major (MPR) PR characterized by <20%, 20%–90%, and ≥90% TE, respectively. Images were all at 200×. TE was defined as tumor necrosis with associated histiocytic inflammation and/or giant cell reaction to keratinaceous debris. TE percentage was determined by dividing estimates of area showing these features by the total area showing residual viable (VT) and TE. B, Proportion of patients with NPR, PPR, or MPR. C, Percent treatment effect in intermediate- and high-risk patients. Survival curves comparing NPR and PPR/MPR for DFS (D) and OS (E).

Close modal

When compared with the clinical stage at enrollment, pathologic downstaging after neoadjuvant pembrolizumab was more frequent in patients with intermediate-risk than with high-risk features (55% vs. 7%, P < 0.0001; Supplementary Table S3). This difference was less pronounced when rates of pathologic downstaging were compared in patients with (PPR/MPR, 34%) and without a PR (NPR, 22%; Supplementary Table S4).

We investigated associations between PR to neoadjuvant pembrolizumab and DFS and OS. Importantly, those that had a PPR/MPR had significantly increased one-year DFS rate compared with those with NPR [93% (95% CI, 84%–100%) vs. 72% (95% CI, 59%–87%), P = 0.0086; Fig. 2D], respectively. OS was also significantly different in the two groups [100% (95% CI, 100%–100%) vs. 93% (95% CI, 85%–100%), P = 0.004; Fig. 2E].

Toxicity

No DLTs were observed during the eight-patient safety lead-in in either group. Treatment-related adverse events (TRAE), defined as attributed to pembrolizumab, radiation, and/or cisplatin, of any grade occurred in 78 (85%) patients (Supplementary Table S5). Grade ≥3 TRAEs occurred in 15 of 42 (36%) patients in the intermediate-risk group and in 32 of 50 (64%) patients in the high-risk group (Table 2). Comparable rates in RTOG 9501 were 46% and 78%, respectively. Surgical wound complications including dehiscence, fistulas, and/or infections were reported in 33 (36%) patients. All grade pembrolizumab-related AEs were reported in 27 patients in the intermediate-risk group and 35 patients in the high-risk group. No immune-related AE-associated deaths were noted. One patient in each group discontinued immunotherapy due to gastrointestinal toxicity attributable to pembrolizumab.

Table 2.

Adverse events by pathologic risk.

Intermediate-risk group (RT+ pembrolizumab)High-risk group (RT+ cisplatin+ pembrolizumab)
an = 42bn = 50
 Grade 3 Grade 3 Grade 4 
Any adverse event: n (%) 15 (36) 31 (62) 5 (10) 
Acute kidney injury 
Adrenal insufficiency 
Anemia 
Anorexia 
Aspiration or lung infection 
Autoimmune hepatitis 
Bone infection 
Colitis/duodenal ulcer 
Dehydration 
Dental caries 
Dermatitis radiation 
Dysphagia 
Dyspnea 
Fatigue 
Febrile neutropenia/neutrophil count decreased 14 
Hearing impaired 
Hypokalemia 
Hyponatremia 
Hypophosphatemia 
Lymphocyte count decreased 
Failure to thrive 
Mucositis oral 
Nausea/vomiting 
Pancreatitis 
Platelet count decreased 
Salivary duct inflammation 
Sinusitis 
Skin Infection 
Tracheal obstruction 
Weight loss 
Wound complication 
Intermediate-risk group (RT+ pembrolizumab)High-risk group (RT+ cisplatin+ pembrolizumab)
an = 42bn = 50
 Grade 3 Grade 3 Grade 4 
Any adverse event: n (%) 15 (36) 31 (62) 5 (10) 
Acute kidney injury 
Adrenal insufficiency 
Anemia 
Anorexia 
Aspiration or lung infection 
Autoimmune hepatitis 
Bone infection 
Colitis/duodenal ulcer 
Dehydration 
Dental caries 
Dermatitis radiation 
Dysphagia 
Dyspnea 
Fatigue 
Febrile neutropenia/neutrophil count decreased 14 
Hearing impaired 
Hypokalemia 
Hyponatremia 
Hypophosphatemia 
Lymphocyte count decreased 
Failure to thrive 
Mucositis oral 
Nausea/vomiting 
Pancreatitis 
Platelet count decreased 
Salivary duct inflammation 
Sinusitis 
Skin Infection 
Tracheal obstruction 
Weight loss 
Wound complication 

aNo grade 4 events.

bIncludes patients who did not proceed with surgery.

Association between PD-L1 expression and PR

Among 72 evaluable patients, PD-L1 CPS in the baseline tumor tissue was 0 in 20 (28%) and ≥1 in 52 (72%) patients. Higher PD-L1 (CPS ≥ 1 and ≥20) was associated with PPR/MPR (P = 0.0183; Table 3). TPS was also calculated and showed similar trends with significant differences between TPS and PR (P = 0.0074). Supplementary Table S6 includes all patients for which CPS and TPS were available for the ITT group. However, PD-L1 expression was not associated with DFS or OS in univariate analysis (Supplementary Fig. S4). No difference in PD-L1 CPS was detected between high-risk and intermediate-risk groups (Fig. 3A).

Table 3.

CPS and TPS by PR.

CPSNPRPPR/MPR
PD-L1 = 0 16/20 (80%) 4/20 (20%) 
PD-L1 = 1–19 23/35 (66%) 12/35 (34%) 
PD-L1 ≥ 20 6/17 (35%) 11/17 (65%) 
CPSNPRPPR/MPR
PD-L1 = 0 16/20 (80%) 4/20 (20%) 
PD-L1 = 1–19 23/35 (66%) 12/35 (34%) 
PD-L1 ≥ 20 6/17 (35%) 11/17 (65%) 
TPSNPRPPR/MPR
PD-L1 = 0 27/34 (79%) 7/34 (21%) 
PD-L1 = 1–49 10/23 (43%) 13/23 (57%) 
PD-L1 ≥ 50 4/10 (40%) 6/10 (60%) 
TPSNPRPPR/MPR
PD-L1 = 0 27/34 (79%) 7/34 (21%) 
PD-L1 = 1–49 10/23 (43%) 13/23 (57%) 
PD-L1 ≥ 50 4/10 (40%) 6/10 (60%) 
Figure 3.

PD-L1 expression and immune gene-expression signature compared with PR. A, PD-L1 was measured using the 22c3 antibody and CPS was determined. Comparison of PD-L1 CPS in intermediate- and high-risk groups. There was no statistical difference between groups. B, Comparison of most differentially expressed gene signatures in NPR versus PPR/MPR tissues at baseline prior to treatment. The values are row-centered by subtracting average scores for each signature. C, Comparison of the gene signatures most changed between pre- and posttreatment tissues in NPR versus PPR/MPR patients. Red designates an increase in expression while blue designates a decrease in gene expression. Changes in B and C were highly significant at FDR <0.1.

Figure 3.

PD-L1 expression and immune gene-expression signature compared with PR. A, PD-L1 was measured using the 22c3 antibody and CPS was determined. Comparison of PD-L1 CPS in intermediate- and high-risk groups. There was no statistical difference between groups. B, Comparison of most differentially expressed gene signatures in NPR versus PPR/MPR tissues at baseline prior to treatment. The values are row-centered by subtracting average scores for each signature. C, Comparison of the gene signatures most changed between pre- and posttreatment tissues in NPR versus PPR/MPR patients. Red designates an increase in expression while blue designates a decrease in gene expression. Changes in B and C were highly significant at FDR <0.1.

Close modal

Association between PR and the tumor immune microenvironment

In order to better understand and characterize patients who develop PR compared with those who do not, gene-expression levels were analyzed in a subset of patient tissues pre- and post-pembrolizumab. Measured gene-expression signatures (GES) were grouped into 48 pathways representing immune cell phenotypes, and tumor and stromal cell characteristics. The heat map in Supplementary Fig. S5 shows the GES scores for all 48 pathways in all patient samples. Upon further analysis, the most significantly (FDR < 0.1) differentially expressed gene/GES between PPR/MPR and NPR patients at baseline (pretreatment tissue) are shown in Fig. 3B. Single gene expression of IDO-1, PD-L1, and PD-L2 and interferon gamma signaling GES were all significantly higher in those with PPR/MPR versus NPR at baseline using the Empirical Bayes Linear Model. When comparing changes in expression from posttreatment (surgical tissue after pembrolizumab) to pretreatment, the signatures with the most significant difference (FDR < 0.1) in pre/post scores between NPR and PPR/MPR are shown in Fig. 3C. GES for macrophages, major histocompatibility complex-2 (MHC-2), and mast cells were significantly higher posttreatment, whereas proliferation was decreased posttreatment in patients with PPR/MPR. The values for biosignatures at baseline and the change from baseline are included in Supplementary Tables S7 and S8. Patient-level data for biosignatures are also provided in Supplementary Table S9.

When neoadjuvant and adjuvant pembrolizumab is added to standard-of-care (chemo)RT, one-year DFS is estimated at 97% and 66% in patients with intermediate-risk and high-risk, resected HNSCC, respectively. These estimates are significantly higher in comparison with historical controls from RTOG 9501 for the entire cohort, and when stratified by pathologic risk features, the intermediate-risk group only. This is true even when time of enrollment is adjusted to RT completion for more direct comparison with RTOG 9501 as well as when the ITT population is compared (Supplementary Fig. S2). Importantly, PR to neoadjuvant pembrolizumab is associated with a significant improvement in DFS relative to NPR patients. Additionally, rates of PPR/MPR increase significantly with increased PD-L1 CPS. Our data indicate that PD-L1 CPS may be a predictive biomarker of PPR/MPR and that PPR/MPR is a surrogate marker of long-term disease control. These data confirm and extend results from similar window-of-opportunity studies in HNSCC and other tumor types (13–20).

Uppaluri and colleagues (14) demonstrated previously that patients with high-risk resected HNSCC who received neoadjuvant and adjuvant pembrolizumab had a lower rate of one-year distant or local relapse rate (16.7%) when compared with historical controls (35%). In contrast, the one-year relapse rate of the high-risk group in our study (32%) was similar to historical controls. Several factors may contribute to this difference in outcomes. A higher proportion of patients in our study had oral cavity primary cancers, which have a worse outcome than other anatomic sites. In our study, pembrolizumab was administered concurrently with chemoRT, whereas in the Uppaluri trial, pembrolizumab was administered after completion of chemoRT. Immunosuppression associated with cytotoxic chemotherapy may impair immune responses associated with PD-1 ICB. In support of this hypothesis is the lack of improvement of OS when the PD-L1 inhibitor avelumab was administered concurrent with chemoRT in local–regionally advanced HNSCC (21). This is in contrast to combined chemotherapy and PD-1 blockade in recurrent and metastatic HNSCC that did result in an improvement in OS for those with PD-L1 CPS ≥1 (22), which may suggest inherent differences in treatment-naïve and -refractory patients.

Patients in the intermediate-risk group had an improvement in one-year DFS and OS with the addition of pembrolizumab when compared with historical controls. This intermediate-risk group also had relatively high rates of downstaging when pathologic stage was compared with clinical stage at enrollment (Supplementary Table S2). In addition, the rate of high-risk pathology (59%) in this current study was somewhat lower compared with historical studies RTOG 9501 and EORTC 22931 for which rates were 60%–70% (4). Therefore, it is possible that response to neoadjuvant pembrolizumab is associated with resolution of high-risk pathologic features. Given that intermediate-risk patients are known to have favorable outcomes compared with high-risk patients and that current trials are investigating additional treatments in this intermediate-risk group (cetuximab in RTOG 0920), further interrogation to understand the population of patients who convert to a more favorable risk status is imperative.

Given the evidence of downstaging, it is also important to consider that the high-risk group survival endpoints were underestimated due to previously high-risk patients being assigned to the intermediate-risk group after neoadjuvant pembrolizumab. In addition, our high-risk group had increased poor prognostic risk factors in general with a higher proportion of oral cavity (86%) compared with historical controls (30%), and oropharyngeal cancers were all HPV negative.

The impact of PR to neoadjuvant immunotherapy on long-term disease control remains an active area of investigation for several tumor types. Complete (cPR) and MPR are considered the most clinically significant outcome measures. In lung cancer, MPR rates as high as 45% have been reported after neoadjuvant PD-1 ICB (23). In contrast, MPR rates are consistently <10% in HNSCC after PD-1 ICB (ref. 19; Fig. 2B). The findings from this study suggest that a PPR, defined by a TE as low as 20%, may be a clinically meaningful measure of benefit from neoadjuvant PD-1 ICB and a surrogate for long-term disease control. DFS and OS at one-year were 93% and 100%, respectively, among the 39% of patients who experienced a PPR, regardless of pathologic risk features or PD-L1 status. These data suggest that pembrolizumab-mediated PR may be a stronger predictor of survival than pathologic risk features or tumor PD-L1 status.

Patient selection for neoadjuvant PD-1 ICB as well as those most likely to benefit from adjuvant treatment would be improved should a predictive biomarker of PR be identified at diagnosis. As mentioned, PD-L1 was strongly correlated with PR (Fig. 3A) but not directly related to survival. Therefore, PD-L1 expression alone is not sufficient to predict long-term disease control. Our gene-expression data also suggest IFNγ, IDO-1, PD-L1, and PD-L2 expression at baseline was predictive of PR. An IFNγ gene-expression profile appears necessary but insufficient to predict pembrolizumab response in metastatic/recurrent HNSCC and is currently being validated in clinical trials as a predictor of PD-1 ICB response (24). Macrophages and mast cells also appeared to be important mediators of PPR/MPR as they increased substantially upon treatment with pembrolizumab. Our data indicate that the additional activation of the myeloid compartment may enhance the response and survival achieved with PD-1 ICB alone.

Hyperprogression upon ICB treatment has been described in the literature and has been associated with EGFR amplification (25). Three patients out of 92 were unable to proceed with curative intent surgery due to either involvement of the skull base not initially identified on preoperative scans, or disease that progressed quickly during the 7–21-day window prior to surgery. Therefore, it is prudent to consider that a small portion of patients may experience more aggressive growth of disease upon ICB treatment in the neoadjuvant setting, necessitating the development of biomarkers to predict potential hyperprogression in these patients. Given the rare occurrence on this study, it is unclear if these patients truly experienced hyperprogression or had more aggressive biology at baseline.

This study has several important limitations. Use of historical controls, prior to improvement in standard and supportive therapies, is a poor substitute for prospective randomization and comparison with placebo control, and therefore our findings should be interpreted with caution. Our median follow-up time and outcomes were shorter by design as a signal-seeking phase II trial. Ongoing randomized clinical trials will clarify the effect of the addition of PD-1 ICB on longer-term survival outcomes and potential ICB-related toxicities when combined with surgery. These include a randomized, placebo-controlled trial of neoadjuvant and adjuvant PD-1 ICB in patients with high-risk HNSCC (NCT03765918) and RTOG 1216 (NCT01810913). In the latter, patients with high-risk resected HNSCC are randomized 2:1:1 to receive adjuvant atezolizumab and cisplatin versus docetaxel and cetuximab versus standard-of-care cisplatin RT. However, there is no current randomized study restricted to patients in the intermediate-risk group. Randomized studies are needed to determine the validity of using PR, PD-L1 CPS, and macrophage infiltration as reliable biomarkers for survival.

T.M. Wise-Draper reports grants and personal fees from Merck & Co, as well as grants from NIH, DoD, and ACS during the conduct of the study; T.M. Wise-Draper also reports grants from BMS, Tesaro/GSK, and AstraZeneca, as well as personal fees from Rakuten, Shattuck Labs, Exicure, and Caris Life Sciences outside the submitted work. F.P. Worden reports personal fees and other support from Merck during the conduct of the study. F.P. Worden also reports personal fees and other support from Bristol Myers Squibb, grants from Bayer and Pfizer, and nonfinancial support and other support from Eli Lilly outside the submitted work. J.M. Kaczmar reports personal fees from Bicara outside the submitted work. A.M. Gillenwater reports other support from UT MDACC during the conduct of the study; A.M. Gillenwater also reports other support from Berkshire Hathaway and Yap Therapeutics, Inc., as well as personal fees from Cortexyme, GT Medical Technologies, and Rakuten Medical Inc. outside the submitted work. J.J. Lee reports grants from NCI during the conduct of the study. V. Takiar reports grants from Merck during the conduct of the study, as well as grants from Merck outside the submitted work. M. Gillison reports other support from Kura Oncology, Agenus, Genocea Biosciences, Inc, Roche, Genentech, and Bristol Myers Squibb during the conduct of the study; other support from Kura Oncology, Shattuck Labs, Inc, Nektar Therapeutics, Ispen Biopharmaceuticals, Inc., EMD Serono, Inc., Gilead Sciences, Inc., Eisai Medical Research Inc., Istari Oncology, Inc., LLX Solutions, LLC, OncLive (owned by Intellisphere, LLC), Seagen (formerly Seattle Genetics), Debiopharm, Mirati Therapeutics, Sensi Biotherapeutics, Inc., and BioNTech AG outside the submitted work; and research grant support from Agenus, Kura Oncology, Genocea Biosciences, Inc., Roche, and Bristol Myers Squibb. No disclosures were reported by the other authors.

T.M. Wise-Draper: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, writing–original draft, writing–review and editing. S. Gulati: Writing–original draft, writing–review and editing. S. Palackdharry: Data curation, writing–review and editing. B.H. Hinrichs: Data curation, formal analysis, validation, investigation, writing–review and editing. F.P. Worden: Investigation, writing–review and editing. M.O. Old: Investigation, writing–review and editing. N.E. Dunlap: Investigation, writing–review and editing. J.M. Kaczmar: Investigation, writing–review and editing. Y. Patil: Investigation, writing–review and editing. M.K. Riaz: Investigation, writing–review and editing. A. Tang: Investigation, writing–review and editing. J. Mark: Investigation, writing–review and editing. C. Zender: Investigation, writing–review and editing. A.M. Gillenwater: Investigation. D. Bell: Investigation, writing–review and editing. N. Kurtzweil: Validation, writing–review and editing. M. Mathews: Validation, writing–review and editing. C.L. Allen: Formal analysis, validation, writing–review and editing. M.L. Mierzwa: Conceptualization, investigation, writing–review and editing. K. Casper: Conceptualization, investigation, writing–review and editing. R. Jandarov: Formal analysis, writing–review and editing. M. Medvedovic: Formal analysis, writing–review and editing. J.J. Lee: Formal analysis, writing–review and editing. N. Harun: Formal analysis, writing–review and editing. V. Takiar: Investigation, writing–review and editing. M. Gillison: Conceptualization, data curation, supervision, investigation, writing–review and editing.

Dr. Wise-Draper was funded by Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., for the conduct of the clinical trial. Dr. Wise-Draper was also supported by a National Institutes of Health/Translational Science Award KL2 Training Grant TR001426 for a portion of this work, Research Scholars Grant, RSG-19-111-01-CCE from the American Cancer Society, start-up funds provided by the University of Cincinnati, philanthropic funds from the Wiltse family and by the Office of the Assistant Secretary of Defense for Health Affairs, through the Peer Reviewed Cancer Research Program, under Award No. W81XWH-17-1-0377. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the Department of Defense. The U.S. Amy Medical Research Acquisition Activity, 820 Chandler Street, Fort Detrick, MD 21702-5014, is the awarding and administering acquisition office. REDCap was supported by the Center for Clinical and Translational Science and Training grant support (2UL1TR001425-05A1). V. Takiar is supported, in part, by a Career Development Award from the United States Department of Veterans Affairs Biomedical Laboratory Research and Development Service [IK2 BX004360]. S. Gulati was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health, under Award Number 2KL2TR001426-05A1 and in part by the ASCO Conquer Cancer Foundation Career Development Award. We would like to acknowledge the UCCC clinical trials office, especially Sheena Lanverman, Shireen Desai, Sarah Wilson, Aubrey Hamilton, Christine Vollmer, and Aly Sklenar for their important contributions to this work. We would also like to thank the histopathology core for their assistance.

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

1.
Sung
H
,
Ferlay
J
,
Siegel
RL
,
Laversanne
M
,
Soerjomataram
I
,
Jemal
A
, et al
.
Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2021
;
71
:
209
49
.
2.
Bernier
J
,
Domenge
C
,
Ozsahin
M
,
Matuszewska
K
,
Lefèbvre
J-L
,
Greiner
RH
, et al
.
Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer
.
N Engl J Med
2004
;
350
:
1945
52
.
3.
Cooper
JS
,
Pajak
TF
,
Forastiere
AA
,
Jacobs
J
,
Campbell
BH
,
Saxman
SB
, et al
.
Postoperative concurrent radiotherapy and chemotherapy for high-risk squamous-cell carcinoma of the head and neck
.
N Engl J Med
2004
;
350
:
1937
44
.
4.
Bernier
J
,
Cooper
JS
,
Pajak
TF
,
van Glabbeke
M
,
Bourhis
J
,
Forastiere
A
, et al
.
Defining risk levels in locally advanced head and neck cancers: a comparative analysis of concurrent postoperative radiation plus chemotherapy trials of the EORTC (#22931) and RTOG (#9501)
.
Head Neck
2005
;
27
:
843
50
.
5.
Ferris
RL
,
Blumenschein
G
,
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
;
375
:
1856
67
.
6.
Cohen
EEW
,
Soulières
D
,
Le Tourneau
C
,
Dinis
J
,
Licitra
L
,
Ahn
M-J
, et al
.
Pembrolizumab versus methotrexate, docetaxel, or cetuximab for recurrent or metastatic head-and-neck squamous cell carcinoma (KEYNOTE-040): a randomised, open-label, phase 3 study
.
Lancet
2019
;
393
:
156
67
.
7.
Oweida
A
,
Lennon
S
,
Calame
D
,
Korpela
S
,
Bhatia
S
,
Sharma
J
, et al
.
Ionizing radiation sensitizes tumors to PD-L1 immune checkpoint blockade in orthotopic murine head and neck squamous cell carcinoma
.
Oncoimmunology
2017
;
6
:
e1356153
.
8.
Paver
EC
,
Cooper
WA
,
Colebatch
AJ
,
Ferguson
PM
,
Hill
SK
,
Lum
T
, et al
.
Programmed death ligand-1 (PD-L1) as a predictive marker for immunotherapy in solid tumours: a guide to immunohistochemistry implementation and interpretation
.
Pathology
2021
;
53
:
141
56
.
9.
nanoString PanCancer IO 360 biological signatures supplement
.
2020
.
Available from
: https://www.nanostring.com/wp-content/uploads/2020/12/FL_MK0572_IO360_BioSigDescription_r6.pdf.
10.
Damotte
D
,
Warren
S
,
Arrondeau
J
,
Boudou-Rouquette
P
,
Mansuet-Lupo
A
,
Biton
J
, et al
.
The tumor inflammation signature (TIS) is associated with anti-PD-1 treatment benefit in the CERTIM pan-cancer cohort
.
J Transl Med
2019
;
17
:
357
.
11.
Ritchie
ME
,
Phipson
B
,
Wu
D
,
Hu
Y
,
Law
CW
,
Shi
W
, et al
.
limma powers differential expression analyses for RNA-sequencing and microarray studies
.
Nucleic Acids Res
2015
;
43
:
e47
.
12.
Benjamini
Y
,
Drai
D
,
Elmer
G
,
Kafkafi
N
,
Golani
I
.
Controlling the false discovery rate in behavior genetics research
.
Behav Brain Res
2001
;
125
:
279
84
.
13.
Horton
JKH
,
Armeson
K
,
Kaczmar
J
,
Paulos
C
,
Neskey
D
.
Neoadjuvant presurgical PD-1 inhibition in oral cavity squamous cell carcinoma
.
J Clin Oncol
2019
;
37
:
2574
.
14.
Uppaluri
R
,
Campbell
KM
,
Egloff
AM
,
Zolkind
P
,
Skidmore
ZL
,
Nussenbaum
B
, et al
.
Correction: neoadjuvant and adjuvant pembrolizumab in resectable locally advanced, human papillomavirus-unrelated head and neck cancer: a multicenter, phase II trial
.
Clin Cancer Res
2021
;
27
:
357
.
15.
Huang
AC
,
Orlowski
RJ
,
Xu
X
,
Mick
R
,
George
SM
,
Yan
PK
, et al
.
A single dose of neoadjuvant PD-1 blockade predicts clinical outcomes in resectable melanoma
.
Nat Med
2019
;
25
:
454
61
.
16.
Ferrarotto
R
,
Bell
D
,
Rubin
ML
,
Lee
JJ
,
Johnson
JM
,
Goepfert
R
, et al
.
Checkpoint inhibitors assessment in oropharynx cancer (CIAO): safety and interim results
.
J Clin Oncol
2019
;
37
:
6008
.
17.
Ferris
RL
,
Spanos
WC
,
Leidner
R
,
Gonçalves
A
,
Martens
UM
,
Kyi
C
, et al
.
Neoadjuvant nivolumab for patients with resectable HPV-positive and HPV-negative squamous cell carcinomas of the head and neck in the CheckMate 358 trial
.
J Immunother Cancer
2021
;
9
:
e002568
.
18.
Zuur
CL
,
Elbers
JBW
,
Vos
JL
,
van der Leun
A
,
Qiao
X
,
Karakullukcu
B
, et al
.
Feasibility and toxicity of neoadjuvant nivolumab with or without ipilimumab prior to extensive (salvage) surgery in patients with advanced head and neck cancer (the IMCISION trial, NCT03003637)
.
J Clin Oncol
2019
;
37
:
2575
.
19.
Uppaluri
R
,
Campbell
KM
,
Egloff
AM
,
Zolkind
P
,
Skidmore
ZL
,
Nussenbaum
B
, et al
.
Neoadjuvant and adjuvant pembrolizumab in resectable locally advanced, human papillomavirus-unrelated head and neck cancer: a multicenter, phase II trial
.
Clin Cancer Res
2020
;
26
:
5140
52
.
20.
Merlino
DJ
,
Johnson
JM
,
Tuluc
M
,
Gargano
S
,
Stapp
R
,
Harshyne
L
, et al
.
Discordant responses between primary head and neck tumors and nodal metastases treated with neoadjuvant nivolumab: correlation of radiographic and pathologic treatment effect
.
Front Oncol
2020
;
10
:
566315
.
21.
Lee
NY
,
Ferris
RL
,
Psyrri
A
,
Haddad
RI
,
Tahara
M
,
Bourhis
J
, et al
.
Avelumab plus standard-of-care chemoradiotherapy versus chemoradiotherapy alone in patients with locally advanced squamous cell carcinoma of the head and neck: a randomised, double-blind, placebo-controlled, multicentre, phase 3 trial
.
Lancet Oncol
2021
;
22
:
450
62
.
22.
Burtness
B
,
Harrington
KJ
,
Greil
R
,
Soulières
D
,
Tahara
M
,
de Castro
G
, et al
.
Pembrolizumab alone or with chemotherapy versus cetuximab with chemotherapy for recurrent or metastatic squamous cell carcinoma of the head and neck (KEYNOTE-048): a randomised, open-label, phase 3 study
.
Lancet
2019
;
394
:
1915
28
.
23.
Forde
PM
,
Chaft
JE
,
Pardoll
DM
.
Neoadjuvant PD-1 blockade in resectable lung cancer
.
N Engl J Med
2018
;
379
:
e14
.
24.
Ayers
M
,
Lunceford
J
,
Nebozhyn
M
,
Murphy
E
,
Loboda
A
,
Kaufman
DR
, et al
.
IFN-gamma-related mRNA profile predicts clinical response to PD-1 blockade
.
J Clin Invest
2017
;
127
:
2930
40
.
25.
Economopoulou
P
,
Anastasiou
M
,
Papaxoinis
G
,
Spathas
N
,
Spathis
A
,
Oikonomopoulos
N
, et al
.
Patterns of response to immune checkpoint inhibitors in association with genomic and clinical features in patients with head and neck squamous cell carcinoma (HNSCC)
.
Cancers
2021
;
13
:
286
.

Supplementary data