Tumor Histological Grade and Immunotherapy Response in Patients With Recurrent or Metastatic Head and Neck Squamous Cell Carcinoma

Key Points

Question  Is tumor histological grade associated with immunotherapy response in patients with recurrent or metastatic mucosal head and neck squamous cell carcinoma?

Findings  In this cohort study of 60 patients with recurrent or metastatic head and neck squamous cell carcinoma treated with immune checkpoint inhibitors, patients with high-grade tumors were more likely to have a clinically beneficial response to immunotherapy than patients with low-grade tumors.

Meaning  These findings suggest that high-grade tumors in recurrent or metastatic head and neck squamous cell carcinoma may have improved immunotherapy response compared with low-grade tumors.

Importance  Tumor histological factors that predict immunotherapy response in patients with recurrent or metastatic head and neck squamous cell carcinoma (HNSCC) are not well defined.

Objective  To investigate the association between tumor grade and immunotherapy response in patients with recurrent or metastatic mucosal HNSCC.

Design, Setting, and Participants  In this retrospective cohort study, the medical records of 60 patients with recurrent or metastatic mucosal HNSCC treated with immune checkpoint inhibitors at Johns Hopkins Hospital between July 1, 2015, and January 22, 2020, were reviewed.

Exposures  High-grade tumors (HGTs) vs low-grade tumors (LGTs) in recurrent or metastatic HNSCC.

Main Outcomes and Measures  Patients were divided into 2 groups: those with LGTs (well differentiated and moderately differentiated) and those with HGTs (poorly differentiated). The main outcome was a clinically beneficial immunotherapy response, defined as complete response or partial response. Univariable and multivariable logistic regressions were conducted to calculate odds ratios for each variable’s association with immunotherapy response. Survival differences were evaluated using Kaplan-Meier survival curves with multivariable Cox proportional hazards regression models.

Results  The 60 patients (35 with HGTs and 25 with LGTs) had a mean (SD) age of 64.6 (8.88) years; 51 were male (85%); and 38 were current or former smokers (63%). The oropharynx was the most common primary tumor site both in patients with HGTs (22 of 35; 63%) and those with LGTs (12 of 25; 48%). Bivariate analysis showed the proportion of patients having a beneficial response to immunotherapy was greater for patients with HGTs (12 of 35; 34.3%) than those with LGTs (2 of 25, 8.0%) (difference, 26.3%; 95% CI, 7.3%-45.3%). Upon multivariable analysis, patients with HGTs had 5.35-fold increased odds (95% CI, 1.04-27.37) of having a clinically beneficial response to immunotherapy. Among patients with available tumor genomic profiling data, the mean tumor mutational burden was greater for patients with HGTs (mean [SD], 8.6 [5.4] mut/Mb; n = 8) than patients with LGTs (mean [SD], 3.6 [1.1] mut/Mb; n = 4) (difference = 5.0 mut/Mb; 95% CI −1.4 to 11.4 mut/Mb; Cohen d = 1.2).

Conclusions and Relevance  In this cohort study, tumor grade was independently associated with immunotherapy response in patients with recurrent or metastatic mucosal HNSCC. These findings highlight the potential role of tumor grade in predicting immunotherapy response in mucosal HNSCC.

The use of immune checkpoint inhibitors (ICIs) has shown promise as an emerging treatment option for patients with advanced head and neck cancer. These ICIs include antibodies that antagonize the PD-1/PD-L1 axis or CTLA-4.¹ Previous studies of head and neck squamous cell carcinoma (HNSCC) tumors have demonstrated a clinically beneficial response to ICIs in patients with disease that is refractory to traditional therapies, including surgery, chemotherapy, and radiation therapy.^2-15

One biomarker that has been associated with a beneficial response to ICI across solid tumors is tumor mutational burden (TMB).¹⁶ This association between increased TMB and a beneficial response to ICI has been demonstrated in numerous cancer types and tumor histologies.^17,18 One hypothesis explaining such an association is the higher probability of generating mutation-associated tumor neoantigens with higher TMB. This probability increases the likelihood of immune detection of these foreign tumor neoantigens, thus facilitating a more robust immune response against the tumor cells.^19-22

While tumor grade is considered a factor of prognostic and therapeutic significance in numerous cancer types,^23-27 its role in mucosal HNSCC is more controversial, and it is therefore not used as a staging criterion.^28-32 High-grade tumors (HGTs) proliferate rapidly and display greater degrees of genomic instability, facilitating the accumulation of more mutations and subsequently a higher TMB. Therefore, we hypothesized that tumor histological grade may be associated with responses to an immune checkpoint blockade. Tumor mutational burden has been positively correlated with a beneficial response to ICIs, which implies that HGT may be also associated with a clinically beneficial response to ICIs, thus underscoring its clinical utility as a predictive biomarker. Though similar associations have been demonstrated in other histologies, this association has yet to be characterized in head and neck cancer.^33-36 To our knowledge, the association between tumor histological grade and ICI response has not been examined in patients with mucosal HNSCC. The objective of this study is to examine the association between tumor grade and immunotherapy response in patients treated with ICI for recurrent or metastatic mucosal HNSCC.

This single-institution retrospective cohort study was approved by the Johns Hopkins Hospital institutional review board. Patients treated with ICIs for recurrent or metastatic HNSCC at the Johns Hopkins Hospital between July 1, 2015, and January 22, 2020, were included in the analysis. The ICIs included in this study were pembrolizumab, nivolumab, ipilimumab, and durvalumab. Patients were excluded if they met any of the following exclusion criteria: the ICI was not for a primary mucosal HNSCC tumor; there was an absence of tumor grade data in the patient medical records; the patient died prior to imaging; the patient was under 18 years of age; the patient was lost to follow-up; or the patient had an unknown primary tumor. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

The electronic medical records of eligible patients were reviewed. Baseline patient characteristics at immunotherapy start date included age, sex, race and ethnicity, body mass index (BMI; calculated as weight in kilograms divided by height in meters squared), smoking history, and human papillomavirus (HPV) tumor status. Patients’ HPV status was determined via p16 immunohistochemical analysis or in situ hybridization. Cancer characteristics included the primary tumor site, tumor grade, T-stage, and the sites and total number of distant organ metastases. Well-differentiated and moderately differentiated tumors were classified as low-grade tumors (LGTs), and poorly differentiated tumors were classified as HGTs; these served as the 2 comparison groups in this study. There were no patients in this cohort with undifferentiated tumors. Clinician progress notes and radiology reports were reviewed to collect data on disease progression or remission after treatment. Patients whose cancers had a complete response or partial response according to the RECIST guidelines (version 1.1) were defined as having a clinical benefit from immunotherapy.³⁷ Patients with progressive disease or stable disease after treatment with an ICI were considered immunotherapy nonresponders.

All statistical analyses were completed using Stata, version 15.1 (StataCorp LLC). Univariate and multivariable standard logistic regressions were conducted to calculate odds ratios (ORs) for the association of each variable with a clinically beneficial immunotherapy response. Tumor grade and HPV status were adjusted for in the multivariable model. Owing to small sample size leading to a separation of data for patients with hypopharyngeal tumors (n = 4) in the standard logistic regression model, these patients were excluded from the standard logistic regression, and a Firth’s logistic regression including these patients was subsequently conducted to serve as a sensitivity analysis. Kaplan-Meier analyses were performed to assess progression-free survival (PFS) and overall survival (OS), and median survival times with 95% CIs were calculated for the LGT and HGT groups. Multivariable Cox proportional hazards regression models for PFS and OS was used to generate adjusted hazard ratios (HRs); consistent with the standard logistic regression, the 4 patients with hypopharyngeal tumors were excluded from the multivariable Cox proportional hazards regression models. Results were reported with appropriate effect size, and 95% CIs were used to define the precision of the estimates.

The baseline characteristics of the study cohort are summarized in Table 1. Sixty patients were analyzed, including 25 with LGTs and 35 with HGTs. The mean (SD) age was 64.6 (8.88) years. Fifty-one patients were male (85%), and 9 were female (15%); 1 was Asian (2%), 12 were Black (20%), 1 was Hispanic (2%), and 46 were White (77%). Thirty-eight patients (63%) had a current or former smoking status. The most common primary tumor site was the oropharynx in patients with LGTs (12 of 25; 48%) and in those with HGTs (22 of 35; 63%). The HPV positivity rate was 32% (8 of 25) in patients with LGTs and 54% (19 of 35) in patients with HGT (difference of 22%). Bivariate analyses showed that the HGT and LGT groups were similar in mean age, sex, race, BMI, smoking status, primary tumor site, HPV status, presenting T-stage, presenting N-stage, and metastases to the lung, bone, and liver. Prior to treatment with ICIs, patients most commonly received surgery, chemotherapy, and radiation therapy (26 of 60; 43%) or chemotherapy and radiation therapy (28 of 60; 47%) (eTable 1 in the Supplement).

The clinically beneficial immunotherapy response rate, defined as complete response or partial response, was 23.3% (14 of 60) (Table 2). Bivariate analysis showed that patients with HGTs were more likely to have a clinically beneficial immunotherapy response (12 of 35; 34.3%) than those with LGTs (2 of 25; 8.0%) (difference, 26.3%; 95% CI, 7.3%-45.3%). When stratifying by tumor grade, the proportion of patients having a clinically beneficial immunotherapy response increased with increasing grade: 0% of well-differentiated tumors, 10.5% of moderately differentiated tumors, and 34.3% of poorly differentiated tumors had a beneficial response (eTable 2 in the Supplement). Progressive disease occurred at a higher rate in patients with LGTs (17 of 25; 68%) than HGTs (10 of 35; 28.6%) (eTable 3 in the Supplement).

Univariate analysis found that patients with HGT had 6.0 times increased odds of having a clinically beneficial immunotherapy response (95% CI, 1.19-30.15) compared with patients with LGTs (Table 3). Multivariable analysis was performed, adjusting for tumor grade and HPV status. Upon multivariable analysis, HGTs (adjusted OR, 5.35; 95% CI, 1.04-27.37) remained associated with increased odds of a clinically beneficial immunotherapy response. Age, sex, primary tumor site, HPV status, smoking history, T-stage, total number of metastases, and presence of metastases to the lung, bone, or liver were not associated with increased or decreased odds of a clinically beneficial immunotherapy response in univariable or multivariable analyses.

Within the study cohort, there were 4 patients with LGTs and 8 patients with HGTs with available TMB data. Figure 1 depicts the distribution of TMB for patients with LGTs and HGTs. Among patients with available TMB data, the mean TMB was greater for patients with HGTs (mean [SD], 8.6 [5.4] mut/Mb; n = 8) than LGTs (mean [SD], 3.6 [1.1] mut/Mb; n = 4) (difference, 5.0 mut/Mb; 95% CI, −1.4 to 11.4; Cohen d = 1.2). The HGT group also had a higher median TMB than the LGT group (median = 10.4 mut/Mb [IQR, 3.0-12.0 mut/Mb] and median = 3.5 mut/Mb [IQR, 2.5-5.0 mut/Mb], respectively).

Across the full study cohort, the median PFS was 5.9 months (95% CI, 3.8-11.1 months) for patients with HGTs (n = 33) vs 3.3 months (95% CI, 1.7-4.7 months) for patients with LGTs (n = 23). Multivariable analysis showed a decreased risk of disease progression for patients with HGTs compared with LGTs (adjusted HR, 0.69; 95% CI, 0.39-1.22) (Figure 2A). The median OS for patients with HGTs vs LGTs was 16.6 months (95% CI, 11.1-25.0 months) vs 15.0 months (95% CI, 8.2-22.8 months). The risk of death was slightly lower in the HGT group compared with the LGT group (adjusted HR, 0.94; 95% CI, 0.47-1.89) (Figure 2B).

Tumor grade is associated with poorer prognosis in numerous cancer types; however, its prognostic value is regarded as less pertinent in HNSCC. This cohort study demonstrates an increased response to ICIs in patients with HGTs compared with LGTs in mucosal HNSCC. To our knowledge, these findings are the first to demonstrate this association in HNSCC. Unlike genomic biomarkers for response to immunotherapy, such as TMB, which can be costly and not easily available, tumor grade is routinely evaluated for head and neck tumors and is readily available to physicians. This availability underscores the clinical utility of tumor grade as potential predictive biomarker in the clinical setting to help guide treatment decisions and patient counseling.

On multivariable analysis, patients with HGTs had increased odds of having a clinically beneficial immunotherapy response. The magnitude of this OR was greater than any of the other covariates assessed (primary tumor site, T-stage, site and number of distant organ metastases, sex, age, BMI, HPV-status, and smoking history). A sensitivity analysis adjusting for the presence of advanced nodal disease was performed and did not alter the results or findings presented in Table 3. Additional sensitivity analyses adjusting for tumor grade, primary tumor site, HPV status, smoking status, and BMI were also performed and did not meaningfully alter the results or findings presented in Table 3. Likely limited by small sample bias, none of the 4 patients with primary tumors of the hypopharynx had a clinically beneficial response to immunotherapy. Thus, these patients were excluded from the standard logistic regression, as it is not possible to produce an OR estimate when there is complete or quasicomplete separation of the data.³⁸ A Firth’s logistic regression that included these 4 patients was conducted to serve as a sensitivity analysis for the standard logistic regression (eTable 4 in the Supplement). Firth’s logistic regression supported the same findings as those shown in Table 3, and the magnitudes of ORs for all variables were comparable. It is notable that numerous ORs in Table 3 have wide confidence intervals owing to small sample size. Though Table 3 suggests that HGTs are associated with increased odds of a clinically beneficial immunotherapy response, a definitive conclusion cannot be drawn from these data.

With increased TMB having been shown to be associated with improved ICI response,^16-18 it is notable that patients with HGTs had a higher mean TMB compared with patients with LGTs. Owing to the limited subset of patients for whom TMB data were available, analyses of tumor grade with immunotherapy response while controlling for TMB in both groups could not be reliably conducted. Furthermore, as increasing tumor grade contributes to genomic instability and resultant increased tumor mutational burden, tumor grade and TMB are not entirely independent variables.

Tumor mutational burden is commonly used as a biomarker to estimate response to treatment with ICIs. If a potential association between TMB and tumor grade could be demonstrated in a larger cohort of patients, it is possible that tumor grade may be used as an inexpensive and readily available, albeit imperfect, surrogate method for estimating low vs high TMB. Although no studies have examined the association between TMB and tumor grade in patients with mucosal HNSCC primary tumors, Johnson et al³³ demonstrated in a comparison of 157 patients with high-grade gliomas and 125 patients with low-grade gliomas that high-grade gliomas had a greater median TMB (1.8 mut/Mb vs 0.9 mut/Mb for high-grade vs low-grade gliomas, respectively) and percentage of patients with intermediate TMB (6-19 mut/Mb; 12.1% vs 3.2% for high-grade vs low-grade gliomas, respectively) and high TMB (≥20 mut/Mb; 5.7% vs 0.8% for high-grade vs low-grade gliomas, respectively). Similarly, this association between high-grade tumors and increased tumor mutations has been shown in numerous other tumor types, such as ovarian tumors,³⁴ lung adenocarcinoma,³⁵ and meningiomas.³⁶

Because of the small sample size of immunotherapy responders included in the Kaplan-Meier survival analyses, we are limited in the conclusions and comparisons we can make between groups regarding PFS and OS. Although patients with HGTs had a markedly higher response rate to ICIs compared with those with LGTs, Kaplan-Meier survival analyses suggested that patients with HGTs had slightly increased median PFS and OS compared with those with HGTs. Cox regression analysis suggests that HGTs may confer a protective effect against disease progression in patients with immunotherapy; however, the confidence interval spanning across 1.0 and the small sample size limit any definitive conclusion being made from these data. Although our study is sufficiently powered to detect a robust difference in clinical response rate to immunotherapy between HGTs and LGTs, it may not be sufficiently powered to detect potential survival differences. As data mature and additional cohorts of patients with greater sample sizes are analyzed, meaningful differences in survival may be observable.

Although the findings and implications of this study are promising, it is important to acknowledge and discuss the study’s limitations as well. First, this study is limited by its small sample size and small number of patients who experienced a clinically beneficial immunotherapy response, which limits the number of independent variables that can be incorporated in the multivariable models as well as our ability to draw any meaningful conclusions for PFS and OS differences between the HGT and LGT groups. Additionally, this study is an observational study and thus can establish correlation but not direct causality. The lack of control group further limits the interpretation of the results. Furthermore, the retrospective design of the study depends on the accuracy of documentation in the patients’ electronic medical records. In addition, this is a single-institution study conducted at a tertiary care center, which may limit the generalizability of the findings. The presence of TMB data in only a subgroup of patients limits our ability to draw any conclusions regarding any potential association between tumor grade and TMB. Finally, these patients had differing sites and patterns of metastasis, further contributing to intragroup and intergroup heterogeneity.

In this cohort study, we found that tumor grade was independently associated with increased odds of having an immunotherapy response in patients with recurrent or metastatic primary mucosal HNSCC. As immunotherapy continues to be used in the treatment of patients with recurrent or metastatic HNSCC, these findings suggest the potential role of tumor grade in estimating immunotherapy response. This study also highlights the need for additional confirmatory studies to further define the association between tumor grade, tumor mutational burden, and response to immunotherapy in head and neck cancer.

Accepted for Publication: March 12, 2022.

Published Online: April 28, 2022. doi:10.1001/jamaoto.2022.0640

Corresponding Author: Rajarsi Mandal, MD, Department of Otolaryngology–Head and Neck Surgery, Johns Hopkins School of Medicine, 601 N Caroline St, Johns Hopkins Outpatient Center, Sixth Floor, Baltimore, MD 21287 (rmandal6@jhmi.edu).

Author Contributions: Mr Alkhatib and Dr Mandal had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Alkhatib, Zhu, Guller, Pardoll, Mandal.

Acquisition, analysis, or interpretation of data: Alkhatib, Maroun, Amin, Zhu, Guller, Herberg, Wu, Seiwert, Rooper, Eisele, Fakhry, Mandal.

Drafting of the manuscript: Alkhatib, Amin, Guller, Mandal.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Alkhatib, Zhu, Guller.

Obtained funding: Mandal.

Administrative, technical, or material support: Maroun, Zhu, Guller, Herberg, Wu, Seiwert, Fakhry, Mandal.

Supervision: Maroun, Eisele, Pardoll, Mandal.

Other: Rooper.

Conflict of Interest Disclosures: Dr Wu reported receiving personal fees from AstraZeneca outside the submitted work. Dr Seiwert reported receiving grants and institutional trial funding from Merck, Bristol Myers Squibb, CUE Biopharma, AstraZeneca, Roche, Nanobiotix and advisory board honoraria from Merck, Vir Biotechnology, Innate, Nanobiotix, and Kura outside the submitted work. No other disclosures were reported.