Presented at the 8th International Conference on Head and Neck Cancer; July 21-25, 2012; Toronto, Ontario, Canada.
Prognostic value of p16 expression and alcohol consumption in Japanese patients with oropharyngeal squamous cell carcinoma
Article first published online: 15 MAR 2013
Copyright © 2013 American Cancer Society
Volume 119, Issue 11, pages 2005–2011, 1 June 2013
How to Cite
Saito, Y., Yoshida, M., Ushiku, T., Omura, G., Ebihara, Y., Shimono, T., Fukayama, M., Yamasoba, T. and Asakage, T. (2013), Prognostic value of p16 expression and alcohol consumption in Japanese patients with oropharyngeal squamous cell carcinoma. Cancer, 119: 2005–2011. doi: 10.1002/cncr.28015
We thank Ms. A. Tsuyuzaki and Ms. K. Miwa (Department of Otolaryngology and Head and Neck Surgery, Faculty of Medicine, University of Tokyo, Tokyo, Japan) and Kei Sakuma (Department of Pathology, Faculty of Medicine, University of Tokyo, Tokyo, Japan) for the excellent technical assistance. We also thank Nobuhiko Kurita, Toru Sasaki, Kenji Kondo, Takashi Sakamoto, Kenta Watanabe, Asayo Furukawa, Hironobu Nishijima, Shinichi Ishimoto, Satoshi Takeuchi, Makoto Kinoshita, Chie Motoyama, and Ryosuke Murakami for their help in preparing sections.
- Issue published online: 20 MAY 2013
- Article first published online: 15 MAR 2013
- Manuscript Accepted: 28 JAN 2013
- Manuscript Revised: 14 JAN 2013
- Manuscript Received: 9 OCT 2012
- oropharyngeal carcinoma;
- head and neck squamous cell carcinoma;
- human papillomavirus;
- alcohol consumption;
The prevalence and prognostic value of human papillomavirus-related oropharyngeal squamous cell carcinoma (OPSCC) in Japan has not been evaluated.
Over a 12-year period, the authors used immunohistochemistry to evaluate the expression of p16 (a cyclin-dependent kinase inhibitor and tumor suppressor) in samples from 173 patients with OPSCC at a single institution and to determine its prevalence and influence on disease prognosis.
The prevalence of p16-positive OPSCC was 33.7% in tonsillar carcinoma, 28.6% in tongue base carcinoma, 0% in posterior wall carcinoma, and 18.8% in soft palate carcinoma. The prevalence of p16-positive OPSCC tumors increased from 15.2% during 2000 to 2003 up to 33.3% during 2008 to 2011; during the same periods, among nonsmokers, the prevalence of p16-positive OPSCC tumors increased from 21.2% to 27.8%; and, among nondrinkers, prevalence increased from 6.1% to 25%. Multivariate analysis identified p16 expression and alcohol consumption as significant, independent prognostic markers of OPSCC.
The current results suggest that the incidence of human papillomavirus-related OPSCC in Japan is increasing and indicate that p16 expression and alcohol consumption may be significant prognostic markers of survival for patients with OPSCC in Japan. Cancer 2013;119:2005–2011. © 2013 American Cancer Society.
Tobacco smoking and excessive alcohol consumption traditionally have been considered the main risk factors for head and neck squamous cell carcinoma (HNSCC). Oral and oropharyngeal squamous cell carcinoma (OPSCC) are major causes of morbidity, with an estimated annual worldwide incidence of 300,000. Evidence suggests that human papillomavirus (HPV) infection is also an independent risk factor, especially for OPSCC. Most of the available data on HPV-associated SCC were generated in the United States and Europe. The incidence of HPV-associated malignancies, such as carcinoma of the uterine cervix, varies substantially in different regions of the world; for example, the incidence of HPV-related cervical/uterine cancer is higher in African and Asian countries than in European countries. Although eastern Asia, including Japan, is known for high epidemics of several oncogenic viral infections, such as Epstein-Barr virus and hepatitis B virus, the status of HPV-related SCC in Asia has not been elucidated.
The diagnosis of HPV-related OPSCC can be made using polymerase chain reaction (PCR) analysis to detect HPV DNA2; however, although this is a logical approach, it is quite challenging in clinical practice. Other diagnostic methods are HPV in situ hybridization or quantification of expression of the HPV oncoproteins E6 and E7, which has the major advantage of detecting transcriptionally active HPV. These assays, however, are cumbersome, costly, and require highly trained personnel. A more feasible diagnostic approach is to measure the expression of p16, which is a cell cycle regulator that, in combination with retinoblastoma (Rb), suppresses cyclin-dependent kinases and is also a surrogate marker of HPV transcriptional activity in OPSCC. Overexpression of p16 in OPSCC correlates with a better therapeutic response and a favorable clinical outcome. In contrast to the diagnostic methods discussed above, p16 immunostaining is widely available and easy to perform and interpret. Usually, p16 immunostaining is classified as either strongly and diffusely positive (with cytoplasmic and nuclear staining) or completely negative. In addition, p16 is a tumor suppressor and thus, is either absent or is expressed only weakly in most non-HPV–related HNSCC because of gene mutation, deletion, or methylation silencing. The current study was designed to evaluate the prevalence of OPSCC in Japan; to assess the prognostic value of p16 expression, tobacco smoking, and alcohol consumption in OPSCC; and to compare their prognostic value with the commonly used markers of clinical staging.
MATERIALS AND METHODS
From January 2000 to December 2011, 193 patients with OPSCC received treatment in the Department of Otolaryngology and Head and Neck Surgery, Tokyo University, Tokyo, Japan. Of these, 20 patients were excluded because of the poor quality of their histologic samples. The clinical charts of the remaining 173 patients (156 men and 17 women; ages 21-86 years; median age, 64 years) were reviewed. The anatomic locations of the tissue samples were as follows: tonsil, n = 101;; base of the tongue, n = 49;; soft palate, n = 16;; and posterior pharyngeal wall, n = 7. Of the 173 patients, 155 were treated with curative intent, whereas the remaining 18 patients received only palliative treatment on the basis of their poor performance status. For the survival analysis, we excluded the 31 surviving patients because of short observation periods (<2 years). Therefore, 124 patients were included in the survival analysis, and the mean follow-up was 3.3 years (range, 0.3-12.3 years). For the patients who remained alive (n = 62), the mean follow-up was 4.6 years (range, 2.0-12.0 years); and, for the patients who died (n = 62), the mean follow-up was 2.0 years (range, 0.3-6.4 years). The TNM staging system was used to classify tumors in accordance with the American Joint Committee on Cancer classification. Patient distribution according to TNM classification is provided in Table 1 (note that there were no patients with distant metastasis). Paraffin-embedded specimens and tissue sections were retrieved from the Department of Pathology files. Institutional Review Board approval was obtained for this study, and written informed consent was obtained from all patients before their inclusion in the study.
|No. of Patients (%)|
|Total||42 (33.9)||15 (12.1)||14 (11.3)||28 (22.6)||15 (12.1)||10 (8.1)||124 (100)|
Initial treatments according to tumor (T) classification are listed in Table 2, and the treatment modalities over the 12-year study period are illustrated in Figure 1. None of our treatments were based on p16 or HPV status, because these markers are not routinely examined before treatment in our institution.
|Tumor Classification: No. of Patients (%)|
|Radiotherapy||9 (43)||33 (62)||23 (72)||9 (53)||0 (0)||74 (59.7)|
|Surgery||12 (57)||20 (38)||9 (28)||8 (47)||1 (100)||50 (40.3)|
Immunostaining for p16 was performed on representative 4-μm sections cut from formalin-fixed, paraffin-embedded tissue blocks that were obtained before patients received chemotherapy or radiotherapy. Deparaffinized sections were autoclaved for 20 minutes in Target Retrieval Solution (S1700; Dako Japan Inc., Kyoto, Japan) for antigen retrieval. Mouse anti-p16 monoclonal antibody (1:100 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, Calif) was used as the primary antibody, and Histofine Simple Stain MAX-PO (M) was used as the secondary antibody (Nichirei Corp., Tokyo, Japan), according to the instructions provided by the manufacturer. The tissue sections were counterstained with hematoxylin. A known p16-expressing HNSCC case was used as a positive control, and sections of normal tonsils were used as negative controls in each experiment. Immunostained samples were examined by a pathologist (T.U.), and p16 immunohistochemistry was scored as positive if there was strong, diffuse staining of the nucleus and cytoplasm in >70% of malignant cells.
Tobacco smokers were individuals who had smoked for ≥20 pack-years (ie, >1 pack per day for 20 years), and alcohol consumers were individuals who had a history of heavy alcohol consumption (≥15 drinks per week for ≥15 years). Potential correlations between p16 expression and various clinical features were tested using the chi-square test; however, for analyses in which there were <4 patients, the Fisher exact probability test was used. The major endpoint of the study was overall survival. Univariate overall survival was evaluated using the Kaplan-Meier method and the log-rank test. Variables that were analyzed by multivariable survival analysis using multiple Cox regression models were selected on the basis of the number of deaths; to avoid over-fitting, only 1 variable was selected per 10 deaths. The hazard ratio (HR) and 95% confidence interval (CI) were calculated to determine the effect of each variable on the outcome with an HR <1.0. P values < .05 were considered statistically significant. Recursive partitioning analysis (for censored survival data) was performed with the use of “party” software package in R (R Foundation for Statistical Computing, Vienna Austria; available at: http://www.R-project.org [Accessed January 12, 2013]) to identify the factors that were most influential for overall survival and to permit the classification of our patients. The prognostic factors in the recursive analysis were p16, smoking, and alcohol consumption. All analyses were carried out using Microsoft Excel version 2010 (Microsoft Corp., Redmond, Wash), Ekuseru-Toukei version 2010 (SSRI, Tokyo, Japan), and R (version 2.15.2).
Positive p16 Status, Demographics, and Clinical Features
Table 3 provides data on p16 expression in pretreatment specimens and includes clinical and follow-up data from 173 patients with OPSCC. Immunostaining for p16 was more common in women than in men (47.1% vs 27.6%; P = .09), more common in nonsmokers than in smokers (31.4% vs 28.7%; P = .72), and significantly more common in nondrinkers than in drinkers (58.2% vs 22.3%; P = .0001). Positive p16 status was related to a lower rate of T4 classification (5.9% vs 19.7%; P = .03). Of the 173 analyzed OPSCC specimens, the most common tumor sites were palatal tonsils (34 of 101 specimens; 34%) and base of the tongue (14 of 49 specimens; 29%). There was no significant correlation between positive p16 status and oropharyngeal tumor localization (palatal tonsil, base of the tongue, soft palate, and posterior wall of the oropharynx).
|No. of Patients (%)|
|Men||43 (28)||113 (72)|
|Women||8 (47)||9 (53)|
|Age: Median [range], y||66 [36-81]||62 [21-86]||.22|
|Nonsmokers||16 (31)||35 (69)|
|Smokers||35 (29)||87 (71)|
|Nondrinkers||20 (59)||16 (41)|
|Drinkers||31 (22)||106 (78)|
|Palatine tonsil||34 (34)||67 (64)||—|
|Soft palate||3 (23)||13 (77)||.23|
|Base of tongue||14 (29)||35 (71)||.53|
|Posterior wall||0 (0)||7 (100)||.06|
|Clinical tumor classification||.67|
|cT1||6 (24)||19 (76)|
|cT2||24 (33)||48 (67)|
|cT3||18 (37)||31 (63)|
|cT4||3 (11)||24 (89)|
|Clinical lymph node classification||.39|
|cN0||14 (27)||38 (73)|
|cN1||5 (24)||16 (76)|
|cN2||29 (33)||59 (67)|
|cN3||3 (29)||9 (71)|
|I||1 (9)||10 (91)|
|II||7 (30)||16 (70)|
|III||11 (35)||20 (65)|
|IV||33 (31)||75 (69)|
Incidence of p16-Positivite Oropharyngeal Squamous Cell Carcinoma
The prevalence of OPSCC in nonsmokers and nondrinkers and positive p16 status increased over the course of the study from 2000 to 2011 (Fig. 2). Positive p16 status increased 2-fold, from 15.2% to 33.3% (P = .05); OPSCC in nonsmokers increased 1.3-fold, from 21.2% to 27.8% (P = .47); and OPSCC in nondrinkers increased 4.2-fold, from 6.1% to 25% (P = .014).
p16 Expression and Patient Survival
Overall survival was analyzed for the 124 patients who received therapy with curative intent. The 3-year disease-specific survival rate was 47.2% (95% CI, 36%-58%) for patients with p16-negative tumors and 82.9% (95% CI, 70%-95%) for patients with p16-positive tumors (P = .0002) (Fig. 3).
We also evaluated the effect of p16 expression on prognosis after treatment in 3 groups according to treatment received: surgery with or without postoperative radiotherapy, radiotherapy, and radiotherapy with chemotherapy. Tumors that were p16-positive were correlated with a good prognosis in all 3 groups (Table 4). In the group of 50 patients who underwent surgery with or without postoperative radiotherapy, the 3-year overall survival rate was 72% (95% CI, 44%-99%) for those with p16-positive tumors and only 48.6% (95% CI, 32%-65%) for those with p16-negative tumors (P = .12). The 3-year overall survival rate for the 39 patients who received radiotherapy with chemotherapy was 83.3% (95% CI, 62%-100%) for those with p16-positive tumors and only 47.3% (95% CI, 28%-66%) for those with p16-negative tumors (P = .12). In the 35 patients who received only radiotherapy, the 3-year disease-specific survival rate was 92.9% (95% CI, 79%-100%) for those with p16-positive tumors and only 44.9% (95% CI, 22%-68%) for those with p16-negative tumors (P = .0018).
|Treatment Modality (No.)||3-Year OS, %||95% CI, %||P|
|Surgery ± PORT|
Prognostic Value of p16 in Oropharyngeal Squamous Cell Carcinoma
Multivariate survival analysis was performed to determine whether the effect of p16 protein expression on prognosis depended on other well established prognostic variables. Considering the OPSCC sample size and the total number of disease-specific deaths in this subgroup of patients with OPSCC, the effect of p16 was determined after adjustment for clinical stage, smoking, alcohol consumption, treatment modality, tumor classification, and lymph node classification. Positive p16 status and lack of alcohol consumption had significant beneficial effects on prognosis (P = .003; HR, 0.34; 95% CI, 0.16-0.70; P = .019; HR, 2.88; 95% CI, 1.19-6.95), even after considering treatment modality, smoking, clinical stage, T classification, and lymph node (N) classification (Table 5). Furthermore, a multivariate survival analysis of patients with p16-positive and p16-negative tumors was performed (Table 6) and indicated that alcohol consumption had significant impairment effects on prognosis in those who had p16-negative tumors (Table 6) (P = .033; HR, 3.15; 95% CI, 1.10-9.04).
|Variable||HR (95% CI)||P|
|p16-Positive vs p16-negative||0.34 (0.16-0.70)||.003|
|Clinical stage III/IV vs I/II||1.16 (0.42-3.17)||.778|
|cT4 vs cT1-cT3||1.05 (0.54-2.04)||.894|
|cN1-cN3 vs cN0||1.44 (0.61-3.40)||.400|
|Surgery ± PORT vs RT or CRT||0.92 (0.55-1.54)||.752|
|Smokers vs nonsmokers||1.15 (0.58-2.28)||.700|
|Alcohol drinkers vs nondrinkers||2.88 (1.19-6.95)||.019|
|Variable||HR (95% CI)||P|
|Clinical stage III/IV vs I/II||1.32 (0.44-3.97)||.627|
|cT4 vs cT1-cT3||0.89 (0.43-1.83)||.753|
|cN1-cN3 vs cN0||1.50 (0.59-3.86)||.400|
|Surgery ± PORT vs RT or CRT||0.81 (0.47-1.42)||.467|
|Smokers vs nonsmokers||0.85 (0.41-1.75)||.657|
|Alcohol drinkers vs nondrinkers||3.15 (1.10-9.04)||.033|
|Smokers vs nonsmokers||4.74 (0.57-39.0)||.148|
|Alcohol drinkers vs nondrinkers||2.35 (0.48-11.6)||.294|
Recursive partitioning analysis identified positive p16 status as the major determinant of overall survival, followed by alcohol consumption for patients who had p16-negative tumors. We classified the risk of death among Japanese patients with OPSCC into 2 categories (Fig. 4b): low risk, with an 80.8% 3-year overall survival rate (95% CI, 69%-92%); and high risk, with a 3-year overall survival rate of 43% (95% CI, 31%-55%). The patients who had p16-positive tumors and nondrinkers who had p16-negative tumors (40%) were at low risk, whereas drinkers who had p16-negative tumors (60%) were at high risk (Fig. 4a).
Viral oncogene (E6 and E7) expression is the gold standard for establishing the HPV etiology of a tumor. Commercially available in situ hybridization assays for HPV DNA are currently close to this gold standard. One potential limitation of HPV16 in situ hybridization that approximately 10% of oropharyngeal carcinomas are associated with non-HPV16, high-risk subtypes of HPV. Shi et al. observed that the concordance between p16 immunoreactivity with HPV16 in situ hybridization or E6 mRNA was 92% and 86%, respectively; and the discordant cases represented non-HPV16 disease.
Immunostaining for p16 is an alternative to HPV16 in situ hybridization. Testing and interpretation of standardized p16 analysis has been discussed extensively in the literature. El-Naggar and Westra analyzed basaloid, nonkeratinizing, or partially keratinizing oropharyngeal carcinomas and proposed that: 1) HPV testing was not required for tumors with strong, uniform p16 staining (both cytoplasmic and nuclear) in all or most tumor cells; 2) HPV testing was required for tumors with absent or weak p16 staining; and 3) HPV testing was required p16-positive, conventional keratinizing SCCs of the oropharynx. Although we have not yet performed additional HPV testing in our samples, we believe that the majority of HPV-positive tumors were included in our study.
The reported clinical features of Japanese patients with p16-positive OPSCC generally are similar to those of their US and European counterparts.[1, 2, 4] In US and European studies, OPSCC tumors were HPV-positive and p16-positive, were genetically less complex, responded well to all standard types of treatment (including radiotherapy, surgery, and chemotherapy), and were associated with favorable clinical outcomes. Similarly, the patients with p16-positive OPSCC in the current study had a significantly better prognosis after radiotherapy and tended to have a better prognosis after surgery and chemoradiotherapy. However, there was no association with the tonsils or the base of the tongue, which typically are more likely to be HPV-positive than the soft palate or the posterior pharyngeal wall. This may have been because of our relatively small sample or the relatively low numbers of patients with p16-positive OPSCC in Japan.
We observed an increase in the prevalence of p16-positive OPSCC, nonsmoker OPSCC, and nondrinker OPSCC in Japan between 2000 and 2011. What is the reason for these increases? Although we could not identify the exact reason in the current study, it is possible that these increases are related to the increased incidence of oropharyngeal HPV infection in Japan. The relation between HPV and oropharyngeal tumors is well established in other countries, where the incidence of OPSCC is on the rise, eg, in the United States and Western European countries.[4, 19] Although the rate of positive p16 status reported in our study (33.3%) was far less than that in the United States, there has been a steady increase in Japan over the last decade, and the incidence may match that in Western countries in the future. The prevalence rates per 100,000 for oral and pharyngeal cancer and for uterine cervical cancer according to national statistics are illustrated in Figure 5. The reasons for these increases in the number of oral and pharyngeal cancers and uterine cervical cancers after 2000 are unclear but may be related to the westernization of sexual habits. In other Asian countries, the p16-positive rate is reportedly 49.5% in Korea and 29% in Hong Kong.
Multivariate analysis revealed a significant correlation between p16 expression and alcohol consumption and the clinical outcome of patients with OPSCC. The prognostic value of these parameters surpassed that of clinical staging. In this regard, recent studies have demonstrated a better prognosis for patients who have HPV-related HNSCC compared with those who have non-HPV HNSCC.[10, 12, 18] Currently, the reason for the better survival of patients with HPV-positive tumors is not clear, but it may be attributable to either the high radiosensitivity of HPV-positive tumors or the active antiviral cellular immune responses observed in patients with these tumors.
Gillison et al. reported that the risk of cancer progression or death increased quantitatively with tobacco exposure at diagnosis and that the effective strength was independent of treatment by radiotherapy or chemoradiotherapy. Our report indicates that alcohol consumption, rather than tobacco smoking, was an independent factor for survival, especially among patients with p16-negative OPSCC. This mat be another racial difference between the Japanese and US and Western European populations. In Japan, Korea, and China, genomic aldehyde dehydrogenase-2 (ALDH2) polymorphism, a key enzyme for the elimination of acetaldehyde, is associated with esophageal and head and neck cancers.[25, 26] This uniquely Asian genetic alteration may render p16-negative OPSCC less responsive to therapy; however, future studies will be needed for further elucidation. Furthermore, because the number of p16-positive patients was relatively small (37 of 124 patients), the risk of smoking nullified the effects of DNA damage-induced cell death because of smoking.
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 25Allele frequencies of 25 polymorphisms pertaining to cancer risk for Japanese, Koreans and Chinese. Asian Pac J Cancer Prev.2002;3:197-206., , .
- 27A review of genetic epidemiology of head and neck cancer related to polymorphisms in metabolic genes, cell cycle control and alcohol metabolism. Acta Otorhinolaryngol Ital.2011;32:1-11., , , et al.