Abstract
Background: Molecular and epidemiologic evidence indicates that human papillomavirus (HPV) is involved in the etiology of oral and oropharyngeal squamous cell carcinomas (SCCs). HPV+ tumors appear to be clinically distinct from HPV− tumors, conferring improved survival outcomes for patients. Determination of the HPV status of tumors may assist in patient risk-stratification and ultimately guide optimum treatment. The primary aim of this study was to examine the distribution of HPV in oral and oropharyngeal SCCs as assessed using seven different in vitro amplification assays. The secondary aim was to correlate the distribution of HPV in tumors with clinical and demographic patient data. Materials and Methods: Sixty-eight invasive oral/oropharyngeal SCCs were tested for HPV using four laboratory-developed PCR assays for HPV16 or 18 and three commercial tests, INNO-LiPA® HPV Genotyping Extra (Innogenetics), Linear Array® HPV Genotyping Test (Roche Diagnostics), and Invader® HPV16/18 ASRs (Hologic Corp.). Results: Consensus results between tests revealed that 71.9% of tumors were HPV+, primarily with HPV16 (63.2%). Other genotypes were uncommon and generally occurred coincidently with HPV16. HPV-positivity was significantly higher in oropharyngeal tumors (76.9%), particularly of the tonsils (91.7%), versus oral cavity tumors (20.0%). HPV+ tumors occurred in younger patients (average 54.4 years versus 61.1 years) and were significantly associated with lower histological differentiation (poorly, 100.0%; moderately, 65.6%; well-differentiated, 42.9%). Conclusion: A high rate of HPV-positivity, especially involving HPV16, occurred in oropharyngeal tumors, with a lower rate in oral cavity SCCs; however, solitary infections with HPV18, 33 or 45 in a minority of cases signified the potential oncogenicity of these additional genotypes and the likely need to screen for these less common genotypes in clinical specimens.
Historically, smoking and alcohol have been identified as primary risk factors for head and neck squamous cell carcinomas (HNSCCs); however, recent evidence indicates that human papillomavirus (HPV) is an important carcinogenic factor involved in the etiology of a subset of these tumors (1-7). One meta-analysis (4) estimated that 36% of oropharyngeal and 24% of oral squamous cell carcinomas (SCCs) were associated with HPV, mostly HPV16; however, considerable variation in prevalence exists between studies. Identification of HPV in HNSCC may have clinical utility, since HPV+ tumors appear to confer significantly improved clinical outcomes (2, 5, 8, 9). In a prospective study of stage III/IV oropharyngeal cancer patients receiving chemoradiation, those with HPV+ tumors had a 59% reduction in risk of death and 46% reduction in risk of progression or death (10). However, it remains to be determined whether HPV status will prompt a change in standard management of HNSCC patients.
A variety of methods have been applied for the detection of HPV in biopsies and brushings of HNSCCs. A recent study compared the analytical performance of six in vitro amplification methods for HPV determination in oral and oropharyngeal biopsies (11). The present study evaluated the HPV status of the same series of tumors utilizing previously collected data in conjunction with additional results of a reverse line blot assay (INNO-LiPA HPV Genotyping Extra). The distribution of HPV in tumors was analyzed with reference to anatomical site, differentiation and patient demographic information.
The primary aim of this study was to examine the distribution of HPV in oral and oropharyngeal SCCs, as assessed by using seven HPV detection assays. The secondary aim was to correlate the distribution of HPV with clinical and demographic patient data.
Materials and Methods
Specimens. Following Institutional Review Board approval, 68 consecutive oral cavity and oropharyngeal SCCs, dating from 2005 to 2007, were retrieved from OU Medical Center, Oklahoma City, OK and Yellowstone Pathology Institute, Billings, MT, USA. The cases were primary invasive tumors, fixed in neutral-buffered formalin and embedded in paraffin (FFPE). Cases with imprecise anatomical location (e.g., ‘tongue’) were omitted. Demographic and clinical information is presented in Table I.
DNA extraction. Hematoxylin and eosin-stained sections were used to confirm the diagnosis of record and presence of tumor in paraffin blocks. DNA was extracted from six to eight 10 μm-thick sections using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA, USA). Routine techniques to prevent sample-to-sample contamination and polymerase chain reaction (PCR) amplicon carry-over were applied throughout, as previously described (12).
HPV genotyping. Extracted DNA was used without determination of concentration in seven HPV detection procedures, including three commercial kits and four laboratory-developed tests. Commercial tests included INNO-LiPA® HPV Genotyping Extra (Innogenetics NV, Gent, Belgium), Linear Array® HPV Genotyping Test (Roche Diagnostics, Branchburg, NJ, USA), and Invader® HPV16/18 analyte specific reagents (Hologic Corp., Bedford, MA, USA). The latter two tests were modified, as previously described (13) to accommodate analysis of FFPE tissues. The Linear Array assay detects 37 HPVs (6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 42, 45, 51, 52, 53, 54, 55, 56, 58, 59, 61, 62, 64, 66, 67, 68, 69, 70, 71, 72, 73, 81, 82, 83, 84, IS39 and CP6108) using PGMY09/11 primers targeting L1 sequences. The Invader HPV16/18 assay has proprietary targets, unknown to the authors. The INNO-LiPA HPV assay uses SPF10 primers to L1 to detect 27 HPVs (6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 43, 44, 45, 51, 52, 53, 54, 56, 58, 59, 66, 68, 69, 70, 71, 73 and 82) from liquid-based cytology or FFPE specimens. DNA extraction differed from that recommended by the manufacturers of Invader HPV16/18 and INNO-LiPA assays. Also, the hybridization and detection steps for INNO-LiPA and Linear Array were performed using an Auto-LiPA staining system (Innogenetics NV). Cases were also analyzed using three PCR-gel electrophoresis methods targeting HPV16 E6 or E7, or HPV18 E6 and one real-time PCR method using hydrolysis probes for HPV16 E7, as previously described (14). Positive (HPV16 plasmid or patient sample) and negative (plasmid without insert and no-DNA) controls were run concordantly with patient samples. Samples were generally tested only once unless a very weak HPV+ signal was observed.
Statistics. Descriptive and inferential analyses, including independent t-tests, Wilcoxon rank sum, contingency table (tests of independence and McNemar's test) and tests for trend, were completed using Stata™ 11.0 statistical software (StataCorp, College Station, TX, USA). Statistical significance was established at p≤0.05.
Results
HPV results were obtained for all 68 tumors using five of seven assays (Table II). Nine cases tested by Invader and one case tested by INNO-LiPA failed due to inadequate amplification of internal controls. The highest rates of HPV positivity were observed with INNO-LiPA (85.0%), Linear Array and HPV16 E7 assays (67.6% each). HPV16 was the most prevalent genotype but prevalence varied considerably (from approximately 48.5% for TaqMan PCR to 67% for E7 PCR and INNO-LiPA) between the five assays capable of detecting this genotype. HPV18 was the second most prevalent genotype, ranging from approximately INNO-LiPA to 1.7% by Invader to 13% by E6 PCR.
The Linear Array and INNO-LiPA tests were the only methods capable of distinguishing genotypes other than HPV16 and 18; Linear Array detects 37 genotypes while INNO-LiPA distinguishes 27 genotypes, although, in the latter test, aberrant hybridization patterns may result from undetermined genotypes (designated HPV-X). HPV6, 33, 45, 52, 82, 84 and HPV-X, were detected in 13 tumors by Linear Array and/or INNO-LiPA (Table II); however, the results of only four specimens were concordant by both assays, including two cases with HPV33 alone, one case with HPV45 alone and one case with HPV82, as part of a double infection. Concurrent infection with multiple genotypes, as determined by Linear Array and/or INNO-LiPA, occurred in 12 tumors; however, the results of only five tumors were concordant by both tests: HPV18 and HPV16 occurred together in four cases and one case had an HPV16/82 double-infection.
Adjudicated HPV results were generated for each case based on the consensus (majority) results of all tests. Equivocal results (equal number of positive and negative tests) were removed from analysis. Adjudicated HPV positivity of tumors was assessed in two ways: using combined HPV16/18 results only, or using results of all genotypes. Using the former method, 66 cases were analyzed, two others being equivocal. Adjudicated HPV results based on all genotypes included only 57 cases in the analysis because of frequent non-consensus between Linear Array and INNO-LiPA for rare genotypes and because of an inability to confirm the results of some rare genotypes (e.g., HPV84 can be specifically identified by Linear Array but not by INNO-LiPA, and genotypes represented by HPV-X are not defined in INNO-LiPA). Consensus HPV results are presented in Table III. Analysis of data from the two methods of assigning HPV-positivity revealed no significant differences (McNemar's exact p=0.25). Using consensus results based on all genotypes, 41/57 (71.9%) tumors were HPV+. Single-genotype infections were limited to HPV16 (32 cases), HPV18 (2 cases), HPV33 (2 cases) and HPV45 (1 case), while double-genotype infections involved only HPV16/18 (4 cases).
Using consensus results based on all genotypes, patients with HPV+ tumors were on average younger (average 54.4 years) than HPV− patients (average 61.1 years, Wilcoxon rank-sum test p=0.02). This trend was similar but not significant when considering only oropharyngeal cases: 53.8 years for HPV+ versus 58.6 years for HPV− patients (Wilcoxon rank-sum test p=0.07). Of the patients aged less than or equal to 50 years old at biopsy, 95.0% were HPV+ compared to 59.5% for those aged more than 50 years old (Fisher's exact test p=0.005). This age difference between HPV+ and HPV− patients was even more pronounced when considering only males (53.7 versus 61.5 years of age, respectively; Wilcoxon rank-sum test p=0.008). Males predominated by a 13.1 to 1 ratio and HPV-positivity was higher in males (73.6%) than in females (50.0%), but not significantly different (Fisher's exact test p=0.312).
Consensus HPV-positivity was significantly higher in oropharyngeal (76.9%) versus oral cavity (20.0%) tumors (Fisher's exact test p=0.019 and p<0.001 respectively); 91.7% of tonsillar tumors were HPV+. HPV-positivity decreased significantly with increasing histological differentiation of tumors: poorly, 100.0%; moderately, 65.6%; well-differentiated, 42.9% (test for trend p=0.002). The same relationship was seen with oropharyngeal cases alone: poorly, 100.0%; moderately, 71.4%; well-differentiated, 50.0% (p=0.004).
Discussion
Since the first report (15) in 1983 associating HPV with HNSCC, a variety of detection techniques, including in situ hybridization, immunohistochemistry and PCR, have been employed to confirm the presence of HPV in such tumors. However, the reported rates of infection in oral and oropharyngeal SCC vary from 0 to 100% (3, 4, 8, 9, 16, 17). Much of this variation is undoubtedly due to differences in laboratory methods and specimen types. Tumor site selection and failure to classify tumor sites strictly according to the American Joint Committee on Cancer (18) also likely confound comparisons. Heterogeneity in patient populations may also account for variation between studies; e.g., HPV+ HNSCCs are particularly prevalent in Asians and in individuals without a history of tobacco use or alcohol consumption and with a history of anogenital HPV, viral co-infections, poor oral hygiene and certain sexual behaviors (1-6).
The present study analyzed 68 invasive oral and oropharyngeal SSCs in a U.S.A. population for the presence of HPV using seven detection methods. Most of the assays detect one or two genotypes (HPV16 and/or 18) but two (Linear Array and INNO-LiPA) have broad-spectrum genotype detection capabilities. As expected, detection rates varied considerably between methods (Table II). This is consistent with the variability observed between different studies. In an attempt to resolve this variability in the results between the assays examined, adjudicated HPV results were generated for each tumor using consensus HPV16/18 results only or consensus results based on all genotypes. Using adjudicated results, the prevalence of HPV in tumors was 68.2% based on HPV16/18 results and 71.9% based on all genotypes. Similar rates of HPV detection have been reported in other studies of oropharyngeal SCC (2, 8, 17, 19). In contrast, in a meta-analysis of 27 studies of oropharyngeal SCC (969 cases) and 35 studies of oral SCC (2,642 cases) using PCR-based detection methods, average HPV-positivity was substantially lower; overall HPV prevalence was 35.6% and 23.5% and HPV16 prevalence was 30.9% and 16.0%, respectively (4). In the current study, using consensus HPV16 results, HPV16 prevalence was 57.6% overall, with higher positivity in oropharyngeal (60.0%) versus oral cavity (33.3%) sites.
Most studies indicate the highest prevalence of HPV in oropharyngeal SCCs, notably those located in the tonsils and base of the tongue, with lower rates of infection in the oral cavity and laryngeal tumors (3, 4, 16, 19, 20-23). Consistent with this observation, in the current study, HPV-positivity was the highest in tonsillar tumors (91.7%). Nevertheless, this is significantly higher than that reported (approximately 50%) for most other studies. Previous reports (8, 9, 16) have also noted that HPV+ tumors are more likely to be poorly differentiated and to have basaloid histopathology, which, in part, is consistent with the present observations; HPV-positivity was 100.0% in poorly differentiated tumors versus 42.9% in well-differentiated tumors, but there was only one pharyngeal tumor with basaloid morphology which was HPV− by every assay.
The most common genotype detected in tumors was HPV16, a finding that is corroborated by all studies (2, 4, 8, 16, 17, 20, 21, 24, 25). HPV18 was the next most-prevalent type; using consensus majority results from the four tests capable of detecting HPV18, prevalence was 9.1% with all cases occurring in oropharyngeal tumors. In contrast, a meta-analysis reported HPV18 to be uncommonly associated with oropharyngeal SCC (1.0%) and more common in oral SCC (8.0%) (4). Other genotypes identified in the tumors of the present study were HPV types 33, 45, 52, 82 and 84. The former three genotypes are considered oncogenic in the context of anogenital cancers (26) and have been previously reported in oropharyngeal SCCs together with other oncogenic (HPV31, 33, 35, 45, 51, 56, 58, 59 and 68) and non-oncogenic (HPV3, 5, 6, 7, 11, 12, 13, 22, 27, 32, 44, 53, 57, 70 and 81) types (3, 4, 19, 21, 22, 27). In the current study, these other HPV genotypes occurred coincident with HPV16, except for three cases which contained only HPV33 or 45. The presence of multiple HPV genotypes in tumors, particularly combined with HPV16, has been described by other groups but such information is probably highly under-reported for most studies due to methodological limitations in detecting less common genotypes.
The possible role of these rare HPV genotypes in the biology of HNSCC is yet unknown. Despite the categorization of some as high-risk in anogenital cancers, their demonstration in oropharyngeal and oral SCCs does not necessarily confirm a causal relationship with malignant transformation. Indeed, the International Agency for Research on Cancer concluded that while there is sufficient evidence to characterize HPV16 as carcinogenic in the oral cavity and oropharynx, there is limited evidence for HPV18 and inadequate evidence for other genotypes (26). Whether tumors containing these rare genotypes are also associated with improved clinical outcomes also remains uncertain. Therefore, the importance of designing a clinical assay that accommodates their detection is currently undefined.
As is generally observed in HNSCCs, whether tobacco- or HPV-associated (23, 28), there was a predominance of males by 13.1 to 1 ratio in the examined cohort. HPV-positivity of tumors was also higher in males than in females; however, the small number of females in the study precludes conclusive correlations. Generally, patients with HPV+ oropharyngeal SCC are younger (by approximately 5 years) than smokers with oropharyngeal cancer (16, 28). The same has also been demonstrated when comparing HPV+ and HPV-tumors (16), which is intuitive considering that the majority of cases of oropharyngeal SCC in individuals with a history of smoking are HPV−. Consistent with previous observations, a similar age difference was noted in the current study: HPV+ patients were on average 6.7 years younger than HPV− ones and 7.8 years younger when only considering males.
Conclusion
This study demonstrated a high prevalence of HPV16 in oral/oropharyngeal SCC. The presence of HPV18, 33 or 45 as sole infections in a minority of tumors requires further investigation to delineate the potential oncogenic and clinical implications of rare HPV genotypes of these types of cancer.
Acknowledgements
We thank the pathologists and staff of OU Medical Center and Yellowstone Pathology Institute who helped in the diagnosis of cases and in the acquisition of blocks and reports used in the study. We also thank Innogenetics NV (Gent, Belgium) for supplying testing kits used in the study.
- Received October 19, 2010.
- Revision received November 11, 2010.
- Accepted November 12, 2010.
- Copyright© 2010 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved