Abstract
Background/Aim: This study was conducted to comprehensively evaluate programmed cell death ligand 1 (PD-L1) expression, and analyze the clinical and prognostic implications of PD-L1 expression in oropharyngeal squamous cell carcinoma (OPSCC). Patients and Methods: We evaluated the expression of PD-L1 using the antibodies SP263 and SP142 in 106 patients with OPSCC, using immunohistochemistry. PD-L1 expression was subdivided into tumor cell score (TC), immune cell score (IC), and combined score (CS). Correlations between each PD-L1 expression and HPV status, clinicopathological features, and survival were analyzed. Results: The expression levels of PD-L1 SP263 and SP142 were significantly correlated. High PD-L1 SP263 TC and CS and SP142 IC and CS were associated with HPV positivity. PD-L1 expression showed no effect on survival in all patients' group. However, in the subgroup analysis, high TC and CS of both PD-L1 SP263 and SP142 were correlated with shorter time to recurrence in the HPV positive group. Conclusion: High expression of PD-L1 was associated with HPV positivity in OPSCC. In addition, high expression of PD-L1 might suggest a poorer outcome, especially in the HPV positive subgroup. PD-L1 could be a useful predictive and prognostic biomarker in OPSCC.
Oropharyngeal squamous cell carcinoma (OPSCC) is a biologically heterogeneous disease despite the presence of common histologic features. Although there has been an improvement in the therapeutic performance due to the development of treatment methods including surgical techniques, radiotherapy, and chemotherapy, the survival rate is not yet satisfactory (1, 2). Even after curative treatment, many patients with OPSCC have lifelong disabilities and deterioration of quality of life (3). Lack of an effective targeted therapeutic agent is also problematic. Therefore, it is important to define a subgroup of patients who are at increased risk of poor outcome and find new therapeutic targets and biomarkers for predicting prognosis and selecting appropriate treatment strategies.
Human papillomavirus (HPV) is a representative and major factor leading to heterogeneity in OPSCC. HPV-positive OPSCC is clinically, biologically, and histologically distinct from HPV-negative OPSCC which is commonly related to smoking. HPV-positive OPSCC is usually non-keratinizing type histologically and often shows basaloid appearance (4). Patients with HPV-positive OPSCC tend to be younger, and have a favorable response to treatment and better outcome compared to HPV-negative OPSCC (5).
Evasion of immune surveillance of cancer cells is known as a hallmark of cancer and a major mechanism of resistance to treatment (6, 7). Activation of T cells is regulated by a number of activating and inhibitory checkpoint signals (8). After T cell activation, negative inhibitory pathways are initiated to suppress hyperactivation of T cells and thereby prevent autoimmunity. Recently, programmed cell death ligand 1 (PD-L1), a ligand of programmed cell death receptor 1 (PD-1), is noted as a major immune checkpoint, which is related to immune escape of cancer cells and is spotlighted as a new therapeutic target (9). PD-L1 is a transmembrane protein that can be expressed on various cell types, including antigen presenting cells, lymphoid cells and cancer cells (10). Interaction of PD-1 and PD-L1 induces inhibition of activated T cells, apoptosis of T cells, and ultimately, promotion of immune evasion and growth of cancer cells. Clinical trials of PD-1/PD-L1 blockade have shown promising efficacy in patients with various malignancies including melanoma, non-small cell lung carcinoma, head and neck cancer, and renal cell carcinoma (11-14). The mechanisms regarding immune evasion in HPV-positive OPSCC and HPV-negative OPSCC might be different. Recently, a few studies have reported that the PD-1/PD-L1 axis is related to HPV-positive OPSCC head and neck cancer rather than HPV-negative OPSCC (15-18). However, the clinical and prognostic role of PD-L1 expression in OPSCC remains unclear.
The aim of this study was to evaluate PD-L1 expression using antibodies that were commonly used in clinical trials and analyze the clinical or prognostic implication of PD-L1 expression considering HPV status in OPSCC. In particular, we comprehensively investigated PD-L1 expression by scoring for tumor cell staining and tumor infiltrating immune cell staining, and by calculating the combined score.
Patients and Methods
Patients. In this study, a retrospective review of the medical records and pathologic reports was conducted on 106 patients who had been diagnosed with OPSCC at Kyungpook National University Hospital and Kyungpook National University Chilgok Hospital in Daegu, Korea between 2006 and 2013. The participants were selected based on the following criteria: 1) pathological diagnosis of OPSCC arising from tonsil, soft palate, back wall of the throat, and base of tongue; 2) treated with surgery, concurrent chemoradiotherapy (CCRT), chemotherapy (CTx) or radiotherapy (RTx); and 3) availability of medical records and adequate tissue sections from formalin-fixed paraffin-embedded (FFPE) tissue. The baseline clinicopathological data including age, gender, smoking history, location of tumor, TNM stage, tumor grade, modality of treatment, and follow up data were obtained. The tumor stage was re-evaluated according to the 8th edition of the Union for International Cancer Control (UICC)/American Joint Committee on Cancer (AJCC) Staging Manual. The study protocol was approved by the institutional review board (approval no. KNUCH 2018-08-029) in consideration of the retrospective nature of the evaluation of anonymized medical data.
PD-L1 expression by immunohistochemistry. PD-L1 expression was evaluated using standard autostaining protocols and the antibodies used in our study were SP263 (Roche Ventana, Tucson, AZ, USA) and SP142 (Roche Ventana). Four-μm representative tumor sections were cut from FFPE, and PD-L1 SP263 and SP142 assays were performed using BenchMark ULTRA platform (Roche Ventana) according to the manufacturer's protocol.
PD-L1 was considered positive only in cases of at least 100 viable tumor cells, if membranous staining alone or membranous and cytoplasmic staining together was present. For a comprehensive analysis of PD-L1 expression, PD-L1 expression was subdivided into tumor cell score (TC), immune cell score (IC), and combined score (CS). We defined TC as the percentage of tumor cells, which shows discernable PD-L1 membrane staining at any intensity (Figure 1A); IC as the percentage of tumor area occupied by stained immune cells, which shows discernable PD-L1 staining of tumor infiltrating immune cells at any intensity (Figure 1B); and CS as the sum of TC and IC. Although the result of CS can exceed 100, the maximum score was defined as 100. In the PD-L1 SP263 assay, TC≥50%, IC≥10%, and CS≥50% were regarded as high expression considering their mean value. In the PD-L1 SP142 assay, TC≥10%, IC≥1%, and CS 10% were considered as high expression.
HPV status by real-time PCR. The evaluation of HPV status was performed using real-time PCR. 10-μm representative tumor sections were cut from FFPE, and DNA was extracted using Maxwell® 16 FFPE Purification kit for DNA (Promega, Madison, WI, USA) and Maxwell® 16 Mdx instrument (Promega). The polymerase chain reaction was performed using CFX96 real-time PCR equipment (Bio-rad, Hercules, CA, USA) and Anyplex TM II HPV28 Detection kit (Seegene, Seoul, Republic of Korea). Anyplex TM II HPV28 Detection kit (Seegene) can detect a group of 19 types of high risk HPV including type 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 69, 73 and 82 and 9 types of low risk group HPV including type 6, 11, 40, 42, 43, 44, 54, 61 and 70.
Epstein-Barr virus (EBV) status by in situ hybridization (ISH). 4-μm representative tumor sections were cut from FFPE, and EBV ISH was performed using Ventana INFORM® EBER probe (Ventana) and Ventana BenchMark® XT platform (Ventana). Dark blue or violet color in the background of red counterstain was regarded as positive (Figure 1C).
Statistical analysis. We used chi-square or Fisher's exact test for comparisons of clinicopathological characteristics and PD-L1 expression and HPV status. We defined duration of recurrence-free survival (RFS) as the interval from the time of the first surgery to the time of any relapse of disease, and duration of overall survival (OS) as the interval from the time of the first surgery to the time of death or the time of the last follow-up. The survival rates for RFS and OS were estimated using the Kaplan–Meier method, and differences between the groups were tested for significance using the log-rank test. A multivariate survival analysis was performed using the Cox proportional hazards regression model. The hazard ratio (HR) and 95% confidence interval (CI) were estimated. A p-value of less than 0.05 was considered to be significant. All statistical analyses were performed using SPSS version 25.0 (IBM Corporation, Armonk, NY, USA).
Results
Patients characteristics. The clinical characteristics of the 106 eligible patients are summarized in Table I. The median age was 62 years (range=37-89 years), and the proportion of men (97, 91.5%) was notably higher than that of women (9, 8.5%). 79 patients (74.5%) were current smokers or ex-smokers, and 27 patients (25.5%) were never smokers. The most common anatomical subsite of the tumor was tonsil in 62 patients (58.5%), and other subsites included soft palate, back wall of the throat, and base of tongue in 44 patients (41.5%). According to the TNM stage, 15 patients (14.2%) were stage I-II, and 91 patients (85.8%) were stage III-IV at the time of diagnosis. Tumor grade was classified as well differentiated in 5 patients (4.7%), moderately differentiated in 90 patients (84.9%), and poorly differentiated in 11 patients (10.4%). A total of 63 patients (59.4%) received CCRT, 13 patients (12.3%) received surgery + RTx ± CTx, and 30 patients (28.3%) underwent only surgery.
PD-L1 expression and HPV status. Overall, 17 (16.0%), 24 (22.6%), and 19 patients (17.9%) showed high expression in TC, IC, and CS of both PD-L1 SP263 and PD-L1 SP142, respectively (Table I and II). All patients with high expression in PD-L1 SP263 also showed high expression in SP142 (p<0.001). HPV was positive in 45 patients (42.5%), and 61 patients (57.5%) were negative for HPV (Table III). Among the 45 HPV positive patients, most were type 16 HPV (41 patients, 91.1%) and type 35, type 39, type 56, and type 59 were detected in 1 patient each (2.2%). All detected types of HPV types belonged in the high-risk group.
Association between PD-L1 expression, HPV status and clinicopathological characteristics. The relationship between variable features and PD-L1 SP263 expression is shown in Table I. Both high PD-L1 SP263 TC and CS were significantly associated with HPV positivity (64.7% vs. 38.2%, p=0.043; 63.2% vs. 37.9%, p=0.044) and EBV positivity (11.8% vs. 1.1%, p=0.015; 10.5% vs. 1.1%, p=0.026). High PD-L1 SP263 IC was significantly related with tumor grade (p=0.002), but no association was found between PD-L1 SP263 IC and HPV status (p=0.395) unlike TC or CS. Regarding PD-L1 SP142, no statistically significant association was observed between TC and clinicopathological features except PD-L1 263 TC (Table II). Patients with high PD-L1 SP142 TC tended to have a higher HPV positive rate (75% vs. 39.8%), but this was not statistically significant (p=0.053). Patients with high PD-L1 SP142 IC revealed poorer tumor grade (poorly 30.8% vs. 7.5%, p=0.029) and higher HPV positivity (84.6% vs. 36.6%, p=0.001). High PD-L1 SP142 CS was also significantly related to HPV positivity (77.8% vs. 39.2%, p=0.025). HPV positivity was significantly associated with younger age (<60 years 53.3% vs. 24.6%, p=0.002), never smoker (35.6% vs. 18.0%, p=0.041), and tonsil as subsite (82.2% vs. 41.0%, p<0.001).
Survival analysis. The median follow-up duration was 26 months (range=1-169 months). In this period, 10 patients (9.4%) experienced recurrence and 13 patients (12.3%) died. The results of the Kaplan–Meier analysis showed no significant difference in OS between TC, IC, CS of PD-L1 SP263 and SP142 (SP263 TC, p=0.764; SP263 IC, p=0.212; SP263 CS, p=0.848; SP142 TC, p=0.712; SP142 IC, p=0.375; SP142 CS, p=0.919, respectively). RFS showed no association with TC, IC, CS of PD-L1 SP263 and SP142 either (SP263 TC, p=0.216; SP263 IC, p=0.106; SP263 CS, p=0.255; SP142 TC, p=0.064; SP142 IC, p=0.064; SP142 CS, p=0.137, respectively). We further evaluated RFS in subgroups according to HPV status. In the HPV positive group, high TC of PD-L1 SP263 had significantly worse RFS than low TC (p=0.038, Figure 2A). High CS of PD-L1 SP263 seemed to be related to shorter RFS, but no statistical significance was identified (p=0.054, Figure 2E). High TC (p=0.008, Figure 2B) and CS (p=0.044, Figure 2F) of PD-L1 SP142 were also significantly associated with shorter RFS than low TC and CS. IC of both PD-L1 SP263 (p=0.071) (Figure 2C) and SP142 (p=0.650) (Figure 2D) showed no significant relation with RFS. However, we found no significant difference in RFS according to PD-L1 expression in the HPV negative subgroup (SP263 TC, p=0.444; SP263 IC, p=0.260; SP263 CS, p=0.444; SP142 TC, p=0.650; SP142 IC, p=0.613; SP142 CS, p=0.650; respectively).
To determine the prognostic value of PD-L1 high expression for RFS, univariate and multivariate analyses were performed (Table IV). In the multivariate analysis on all patients, only higher tumor grade showed significant association with poorer RFS (moderately vs. well, p=0.006; poorly vs. well, p=0.041). In the HPV positive subgroup, PD-L1 SP142 TC high expression was related to shorter RFS in the univariate analysis (p=0.024), but no definite association was found in the multivariate analysis (p=0.920).
Discussion
This study presented a comprehensive evaluation of PD-L1 expression in consideration of HPV status in OPSCC. We aimed to investigate the association between PD-L1 expression and HPV status, and their clinical and prognostic implications. Furthermore, we investigated the possibility of a subset of patients with OPSCC being eligible for immunotherapy. In this study, we demonstrated that high expression of PD-L1 was significantly associated with HPV positivity in OPSCC. TC and CS of SP263 antibody staining and IC and CS of SP142 antibody staining showed a significant association with HPV status. TC of SP142 staining also showed a trend to be associated with HPV positivity, but was not statistically significant (p=0.053). Recently, a few studies reported that PD-1 or PD-L1 is related to HPV positivity in head and neck cancer (15-18). We obtained similar results, suggesting that HPV may play a key role in the immunologic microenvironment including PD-L1 in OPSCC.
There are some considerations for PD-L1 immunostaining and scoring in real clinical setting. First, proper antibody selection for accurate and reliable results should be considered. Commercially available PD-L1 antibodies vary significantly, and the degree of expression also varies according to the type of antibody. Some PD-L1 antibodies have been developed and validated simultaneously as companion diagnosis along with immunotherapy. SP263 (Roche Ventana), SP142 (Roche Ventana), 28-8 (Dako, Agilent Technologies, Santa Clara, CA, USA) and 22C3 (Dako, Agilent Technologies) are representative validated antibodies (13, 19). Suitable devices are also recommended by the manufacturer; BenchMark ULTRA platform (Ventana) for SP142, BenchMark platform (Ventana) for SP263, and Link 48 autostainer platform (Dako, Agilent Technologies) for 28-8 and 22C3 (19). Although the study focused on lung cancer, and not head neck cancer, a harmonization study called Blueprint PD-L1 IHC assay comparison study was conducted to overcome the differences in the expression rate of PD-L1 according to the antibody (19, 20). A few studies have evaluated PD-L1 expression in OPSCC using the proven antibodies, including SP142 (18, 21, 22), 28-8 (23), 22C3 (22), but all of them used platforms other than those approved for companion diagnosis. Therefore, we selected SP263 and SP142, which are the most commonly used antibodies for companion diagnosis, and each PD-L1 assay was performed according to the specified platform and method suggested by the manufacturer. The mean expression levels of SP142 were 3.6, 0.4, and 3.9 in TC, IC, and CS, and the mean expression levels of SP263 were 17.6, 5.1, and 22.1 in TC, IC, and CS. The expression level of PD-L1 SP263 and SP142 were significantly correlated in all of TC, IC, and CS in this study (p<0.001), although the expression level of SP142 was much lower than that of SP263. These differences in the expression levels of SP142 and SP263 were also consistent with the results of the harmonization study (19, 20).
Second, a predicting, reproducible, and easy-to-read cutoff value should be used in a clinical setting. Studies used 1% (17, 23, 24), 5% (18, 21, 22, 25), 20% (26) and so on as cutoff values in PD-L1 expression in OPSCC. The target cells of evaluation of PD-L1 expression also varied; tumor cells (17, 18, 22, 24, 26), immune cells (21, 23), and both tumor cells and immune cells (25). In this study, we set 50%, 10% and 50% as the cutoff in TC, IC and CS of SP263, and 10%, 1% and 10% as the cutoff in TC, IC, and CS of SP142 considering mean values of each expression levels, reproducibility and clear discrimination. The reading of PD-L1 expression was made in all of TC, IC, and CS comprehensively and was evaluated by dividing into high expression and low expression based on each cutoff value. High expression of PD-L1 was revealed in 16% in TC, 22.6% in IC and 17.9% in CS in this study. Our results showed lower expression rates in both TC and IC compared with other studies ranging 23-67.8% in TC (17, 18, 22, 25, 26) and 38.7 to 68% in IC (21, 23, 25). That is, because we divided the expression levels into high expression and low expression using a higher cutoff, rather than positive and negative using a lower cutoff level.
In addition to HPV, EBV, which is known to be related to a subset of head neck cancer, might also affect the immunologic environment. Therefore, we evaluated EBV status by EBV ISH simultaneously. EBV was positive in only three patients (2.8%), and high expression of TC and CS in SP263 was associated with EBV positivity in OPSCC (p=0.015 and p=0.026, respectively). However, we could not find any association between HPV and EBV status, and no prognostic implication of EBV status was found either.
In the survival analysis, TC, IC, and CS of PD-L1 SP263 and SP142 had no effect on both OS and RFS in the all patients' group. We further evaluated outcomes in subgroups by dividing into an HPV positive group and an HPV negative group, because we found that PD-L1 expression varied according to HPV status. Notably, high SP263 TC (p=0.038) and high TC (p=0.008) and CS (p=0.044) of SP142 were associated with shorter time to recurrence, in only the HPV positive subgroup. High SP263 CS also revealed a tendency of association with shorter time to recurrence, but there was no statistically significant difference (p=0.054). Since TC and CS of SP263 and SP142 were thought to have similar prognostic implications, we further performed univariate and multivariate analyses using TC. However, no definite association was found in the multivariate analysis. The prognostic implications of PD-L1 expression on OPSCC have been controversial so far. Hong et al. (17), Kwon et al. (18), Young et al. (21), Meulenaere et al. (22), Birtalan et al. (23) and Sato et al. (25) have reported that PD-L1 expression was associated with better outcome of patients with OPSCC. Kim et al. (26) have found that PD-L1 expression did not affect patients' survival. Li et al. (27) have conducted a systematic review and meta-analysis regarding the prognostic value of PD-L1 expression in patients with head and neck cancer. They reviewed 17 studies with 2,869 patients, and no significant correlation between PD-L1 expression and OS or RFS was identified. However, their subgroup analysis suggested that PD-L1 expression was related to poor OS and RFS in patients with head neck cancer in Asian countries. Although we could not find a definite association in the multivariate analysis, our results suggest that high expression of PD-L1 can be a predictive marker for poor outcome, especially for short RFS, as similarly suggested by Li et al. in their meta-analysis. Our results also suggest that a subset of OPSCC with high expression of PD-L1 might benefit from immunotherapy.
Our study has some limitations. First, because this is a single-center retrospective study, the number of patients in this study may not be sufficient for evaluation. Second, we included patients who were diagnosed with OPSCC from 2006 to 2013, which indicates that the specimens were old and this could have affected the results of immunohistochemical staining even though we considered the quality of the specimens. Very recent samples were not included in this study for the survival analysis. Third, the results of PD-L1 expression might have intratumoral heterogeneity. Therefore, further large-scale investigations and prospective studies are required to validate the clinical and prognostic significance of PD-L1 expression in OPSCC. Fourth, because this is not a clinical trial-based study, it is not known whether the patients with high expression of PD-L1 will benefit from immunotherapy or how the prognostic outcome will change if the patients have immunotherapy. Immunotherapy has not been commonly used in OPSCC so far except in clinical trials, which adds some limitations in evaluating the role of PD-L1 as a predictive or prognostic biomarker in a retrospective study. Further studies are needed in the era of increased use of immunotherapy for OPSCC.
In conclusion, high PD-L1 SP263 TC and CS and PD-L1 SP142 IC and CS were significantly associated with HPV positivity in OPSCC. In addition, we demonstrated that high TC and CS in both PD-L1 SP263 and SP142 might suggest poorer RFS. Our results suggest that PD-L1 could be a useful predictive and prognostic biomarker considering HPV status in OPSCC patients, although we could not find a definite association in multivariate analysis. Future prospective studies and further investigations are warranted to define the definitive role of PD-L1 in these patients in the era of immunotherapy.
Acknowledgements
This work was supported by Biomedical Research Institute grant, Kyungpook National University Hospital (2017).
Footnotes
Authors' Contributions
Conceived and designed the study: J.Y.J. Acquired clinical data: D.A. Analyzed and interpreted the data: J.Y.J. and T.I.P. Wrote the manuscript: J.Y.J. Revised the manuscript: T.I.P. and D.A. Supervised the study: J.Y.J. All Authors made substantial contributions to the design of the study, acquisition of data, analysis and interpretation of the data, writing and revising the manuscript. All authors approved the final version of the manuscript.
Conflicts of Interest
All Authors declare no conflicts of interest regarding this study.
- Received June 1, 2020.
- Revision received June 15, 2020.
- Accepted June 16, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved