Infectious Causes of Cancer
Human papillomavirus status in young patients with head and neck squamous cell carcinoma
The role of human papillomavirus (HPV) in head and neck squamous cell carcinoma (HNSCC) development has been recognized only in the last decade. Although younger patients develop HNSCC associated with HPV, the incidence in young patients has not been studied. Forty-five young HNSCC patients (<40 years) were tested for HPV and the expression of p16ink4a and p53 in tumor biopsies. The presence of HPV was correlated with the absence and presence of alcohol and tobacco exposure. Paraffin-embedded, archival biopsy materials from HNSCC of 45 patients younger than 40 years were analyzed. HPV subtypes were identified by PCR followed by genotyping. Expression of p16ink4a and p53 were determined by immunohistochemistry. Fourteen (31%) of the HNSCC specimens from 45 patients unequivocally exhibited HPV16 positivity. Sixty percentage of the oropharyngeal tumors and 5% of the oral cavity tumors were HPV16 positive. P16ink4a overexpression was detected in 93% of the HPV16-positive tumors. None of the HPV16 tumors showed p53 overexpression. There was no association of HPV positivity with (lack of) exposure to alcohol and smoking. HPV association was not exclusively detected in nonsmoking, nondrinking young HNSCC patients. The presence of p16ink4a accumulation and the absence of p53 overexpression are good surrogate markers for HPV-associated HNSCC.
The proportion of human papillomavirus (HPV)-induced head and neck squamous cell carcinomas (HNSCCs) seems to increase during the last decades. Although impressive clinical and molecular data have been generated concerning the role of HPV in the overall HNSCC population,1, 2 less is known about the subgroup of young onset HNSCC. This subgroup of HNSCC patients is of specific importance due to two factors. First, the incidence of HNSCC in young patients shows increasing incidences worldwide.3 Second, these patients have consumed less tobacco and alcohol, indicating a possible etiological role for HPV.4 Furthermore, studies on vaccines to prevent HPV infection5 or to induce activity against established HPV infections6 are at the moment of great interest because of their potential clinical implications for patients with HPV-related HNSCC. Therefore, it is important to investigate whether HPV is indeed more prevalent in this young age group.
HPV is a small epitheliotropic DNA virus that primarily infects transitional epithelium, present in the upper aerodigestive tract and anogenital regions. The HPV viral family consists of over 100 types, divided into low- and high-risk based on their oncogenicity. Low-risk types include HPV6 and HPV11, associated with benign anogenital condylomas and laryngeal papillomatosis. High-risk types include HPV16 and HPV18, both well-established initiators of cervical and anogenital carcinogenesis.7 Accordingly, high-risk HPV DNA is found in nearly 100% of cervical carcinomas and 84% of anal carcinomas.8, 9 HPV prevalence is lower in vaginal carcinoma (70%) and even lower in vulvar carcinomas (40%).8 The overall prevalence of HPV in penile carcinomas is 47%.10 HPV DNA is present in tumor cells either integrated into the host DNA or as episomes. Evidence of functionally active HPV can be found in the majority of these cells, including the presence of viral mRNA and proteins.11 E6 and E7 are the characteristic oncogenic proteins that confer the high-risk phenotype, through disruption of key tumor suppresor proteins including p53 (degraded by E6) and Rb (degraded by E7).11
For detection of HPV in tumor material various methods are used. Detection of HPV E6 and E7 expression is the gold standard for classifying a tumor as HPV related; however, detecting viral RNA in existing paraffin-embedded material is impractical for cancer diagnostics.12, 13 Therefore, several PCR-based, as well as in situ hybridization assays, are currently used to detect HPV DNA in tumors. Moreover, p16ink4a is suggested to be an excellent surrogate marker for HPV infection, reflecting the functional effects of E7-induced inactivation of pRb.13–16 Mutation of the p53 gene has been documented as being the most frequent genetic alteration in most carcinomas, including 70% of HNSCC.17 However, in HPV-positive HNSCC the oncoprotein E6 binds to the p53 protein, causing p53 degradation and thereby functional inactivation.18–20 Therefore, HPV-positive tumors often lack p53 mutation.
Study results on the prevalence of HPV16-related HNSCC can be confounded by the use of only p16ink4a expression as a predictive method for HPV16 positivity.14, 21 Although p16 expression is proven to be highly correlated with HPV status in HNSCC, p16 overexpression could suggest pRb pathway disturbances unrelated to HPV (e.g., mutational inactivation of retinoblastoma protein).22, 23 Using E6 and E7 mRNA levels as conclusive evidence of HPV involvement in HNSCC, p16 has shown 100% sensitivity but only 79% specificity as a surrogate marker.24
HPV16 is the most common HPV type in oropharyngeal SCCs accounting for approximately 90% of HPV-containing cases while HPV18 was found in only 2.8% of the cases.7
As HPV-positive HNSCC patients are reported to be younger and more often nonsmokers (NS) and nondrinkers (ND),3, 23, 25 the question arises whether in young patients with HNSCC, HPV is more frequent and whether in this subgroup patient characteristics and risk factor behavior are different, especially with regard to exposure to alcohol and tobacco.
In our study, we evaluated a cohort of HNSCC patients less than 40 years of age, stratified into nonsmokers/nondrinkers (NSND) and smokers/drinkers (SD). The purpose of our study was to determine whether infection with high-risk HPV was associated more often with HNSCC malignancy in young patients and if this was related with the presence or absence of tobacco and alcohol use. Second, we investigated if immunohistochemical detection of p16ink4a and p53 can be good surrogate markers for HPV in young onset HNSCC.
Patients and Methods
Tumor material and patient data
From 1977 till 2008, a total of 54 patients aged <40 years with newly diagnosed, previously untreated HNSCC were identified in the Netherlands Cancer Institute Database. Patients with cancer of the nasopharynx, external lip and salivary glands were excluded. Formalin-fixed, paraffin-embedded archival biopsy and resection materials from these patients were selected from the archives of the Department of Pathology. Sufficient tumor material of 45 patients was present for analysis. The distribution by anatomic site was as follows: two were located in the larynx, 20 in the oral cavity, three in the hypopharynx and 20 in the oropharynx. Demographic data, including age at diagnosis, gender, alcohol and tobacco exposures, were obtained from medical records. Tobacco usage was coded as dichotomous variable and was considered positive if any current or past history of active smoking was reported. Alcohol usage was also coded as a dichotomous variable and was considered positive when patients consumed more than three drinks a day. A series of 4-μm-thick sections were cut for (i) hematotoxylin–eosin staining; (ii) DNA extraction and HPV detection; (iii) immunohistochemistry (IHC) to visualize p53 and p16ink4a expression.
Human papillomavirus typing
DNA extracted from formalin-fixed, paraffin-embedded tumor tissue was used for HPV typing analysis. HPV typing was performed at the Department of Pathology, Leiden University Medical Centre, The Netherlands. To ensure lack of contamination, sections of a paraffin block without tissue, cut before each sample, served as negative control. All these controls were negative in the PCR analysis. INNO LiPA prototype research genotyping assay (Innogenetics, Gent, Belgium), a highly sensitive hybridization assay, was used for HPV typing as recommended by the manufacturer.26
Immunohistochemical staining of p53 and p16 ink4a
Immunohistochemical staining was performed on 4-μm formalin-fixed, paraffin-embedded tissue sections. In brief, sections were deparaffinized and subsequently pretreated with 0.3% H2O2 in methanol to quench endogenous peroxidase activity. Antigen retrieval was performed by microwave heating in 0.001 M citrate buffer (pH 6.0). The monoclonal antibody clone D-07 (Dako) and clone MC16-PO4 (Neomarkers) were used to detect p53 and p16ink4a, respectively. Protein expression scoring was performed by two observers (H.v.M. and L.v.V.). p53 staining was scored as positive when ≥75% of the cells exhibited a strong nuclear staining because this was shown to reflect the presence of p53 accumulation due to a mutation.26 In the literature, there is no consensus for the use of a cut-off value to score p16ink4a overexpression (Table 1). Analogous with the reasoning for p53, which proved to be valid,27 p16ink4a expression was thought to reflect activation of this pathway by HPV if most cells exhibited expression and thus a threshold of 75% used.
Table 1. Studies evaluating concordance between HPV status and p16 ink4a overexpression
Factors associated with HPV status were selected on cross-tabulations, which were analyzed by the use of the χ2 test. We assessed levels of agreement between the HPV test, p16ink4a, p53 within the sample by Kappa score. McNemar χ2 test for matched pair data was used for assessing unequal distribution of discordant results. A significant level of p ≤ 0.05 was set for all comparisons. All statistical analyses were performed using SPSS version 17.0 (SPSS, Chicago, IL).
In this population of 45 HNSCC patients under 40 years, the median age was 34.8 years with a minimum age of 21 at diagnosis. Eight patients were ≤30 years of age at diagnosis, with malignancies consisting of four oral cavity and four oropharyngeal SCCs. Males were overrepresented with 29 versus 16 female patients. Between patients with oral cavity SCCs and oropharyngeal SCCs, 50 and 65% smoked tobacco as did the two laryngeal carcinoma patients.
HPV status and the expression of p16ink4a and p53
In total, 45 primary HNSCC cases were examined for the presence of HPV by PCR followed by the INNO-LiPA genotyping to identify the HPV subtype. Fourteen out of 45 (31%) HNSCCs exhibited HPV16-specific PCR signals (Table 2). One case was positive for HPV35, located in the oropharynx and one for HPV66, located in the oral cavity. High-risk HPV types 35 and 66, showed no p16ink4a overexpression, possibly indicating that HPV was not etiologically involved in carcinogenesis in these patients. Of the 14 HPV16-positive samples, 12 (86%) were oropharyngeal SCCs, whereas two HPV-positive cases originated from nonoropharyngeal tumors (one oral cavity and one hypopharynx).
Table 2. Presence of HPV DNA in 45 HNSCC patients by location
Thirteen out of fourteen of the HPV16-positive tumors showed a p16ink4a overexpression. Five cases showed overexpression of p16 while they were HPV negative. With a threshold of 1% for p16, three more HPV negative cases were detected, warranting the threshold of 75%. Using McNemar's test, no significant difference was found for demonstrating the presence of HPV16 using either PCR or p16ink4a overexpression (p = 0.180; Table 3). None of the HPV16-positive tumors showed p53 overexpression (>75% of cells) neither did both other high-risk HPV types (35 and 66). Twelve out of 31 HPV16-negative tumors (39%) showed p53 overexpression. Using McNemar's test, no significant difference was found for demonstrating the presence of HPV16 using PCR and the absence of p53 overexpression (p = 0.052; Table 3). The presence of HPV16 and the presence of p16 overexpression had good agreement [κ = 0.58 (95% CI 0.34–0.82)].
Table 3. Concordance and association analysis between HPV16 status and tobacco, alcohol, p53 and p16ink4a
The combination of both p16 overexpression and the lack of p53 overexpression did not result in a higher sensitivity for prediction of the presence of HPV by this combined immunohistochemical staining than p16 overexpression alone; whereas, the specificity increased from 83 to 87% as of five HPV-negative tumors showing p16 overexpression, four were indeed p53 negative. The presence of HPV16 and the presence of p16 overexpression in tumors together with the lack of p53 overexpression had good agreement [κ = 0.76 (95% CI 0.56–0.95].33
Association between HPV status and risk factors
Of the 29 patients with HPV-negative tumors, 10 were NS (34%) whereas in the 16 HPV-positive patients eight were NS. In the HPV-positive patients, five of the patients were ND, whereas four were NSND patients. Statistical analysis showed no significant association between HPV16 positivity and the absence of tobacco exposure (p = 0.523) or the absence of alcohol abuse (0.367; Table 3). For patients with HPV-positive oropharyngeal SCC, 58% had a history of tobacco use and 67% used alcohol. In the HPV-negative oropharyngeal carcinomas only one case (14%) did not smoke. Of the nine NSND patients, four patients had HPV16-positive tumors (p = 0.280; Table 3).
Numerous studies have led to an appreciation of the involvement of HPV in the etiology of a subgroup of HNSCCs, especially in the oropharynx. The incidence of HPV-related HNSCC varies between different studies and populations. This variance is most likely due to the use of differing detection methods (Table 4) and differences in risk factor-associated behavior between countries. In our study, we identified HPV16 positivity in 31% of all HNSCC patients aged <40 years and in 60% of all oropharyngeal SCCs using a DNA-based technique. This supports the findings of several other studies.7, 34, 35 Hafkamp et al.36 who also studied the Dutch population found a HPV16 incidence of 21% in a patient group with a mean age of 57 (range, 27–84).
Table 4. Overview of literature on incidence and methods for detecting HPV in HNSCC
Our findings support that predominantly oropharyngeal epithelium is sensitive to HPV carcinogenesis, compared to the oral mucosa.34, 30, 39 The reason for the strong association of HPV with oropharyngeal SCC in particular remains unclear, although the easy viral accessibility of the invaginated, crypt epithelium and the presence of cytokines produced by the lymphoid tissue to stimulate viral transcription and cellular transformation have been suggested as explanations.40, 41 However, we found one HPV-positive oral cavity carcinoma and one HPV-positive hypopharyngeal carcinoma in our population. The HPV-positive oral cavity cancer was located at the inferior alveolar process indicating a possible relation with the retromolar trigone. Although previous studies have demonstrated that there has been an increase in the number of young adults with squamous cell carcinoma of the mobile tongue (even in the absence of tobacco and alcohol),42, 43 this rise could not be explained by HPV and the etiology behind this increasing incidence remains to be elucidated.
Only two tumors displayed another type of HPV than high-risk HPV16: one tumor with HPV type 35 was located in the oral cavity and one with HPV type 66 was located in the oropharynx. Both HPV 35 and HPV 66 are high risk and sexually transmittable HPV types44 and associated with cervical cancer.9 In HNSCC studies, HPV 35 and HPV 66 are rarely detected,7, 16 therefore in our study, most statistical analysis were performed considering only HPV type 16 as a causative agent. In cervical carcinomas these HPV types are highly associated with p16ink4a overexpression.45 The fact that in the HPV 35- and 66-positive tumors p16ink4a was not overexpressed in our study might indicate that these viruses did not play a role in carcinogenesis in these patients.
Previous studies presumed that mainly nonsmoking nondrinking HNSCC patients are at risk for HPV-related HNSCC3, 4; however, our study shows contradicting results (Table 3). HPV16 positivity was detected in the presence of exposure to both tobacco and alcohol use. Of the nine patients who did not have any exposure to tobacco and alcohol, four patients had HPV16-positive tumors (p = 0.280), leaving five NSND HNSCC patients without HPV positivity. Although it remains possible that the HPV detection method used in our study does not detect 100% of the cases, these cancers deserve scrutiny for genetic abnormalities. In these patients a genetic predisposition to environmental carcinogenesis reflected by chromosomal abnormalities, increased susceptibility to mutagen-induced chromosome damage or DNA repair deficiency should be considered.46
It is important to investigate accurate and manageable techniques for HPV detection, because HPV positivity has effect on prognosis and may have future implications for treatment.47 The commonly used techniques of detection include PCR for HPV DNA, RT-PCR for E6 and E7 mRNA and in situ hybridization for direct visualization of HPV in the nucleus.13 PCR-based techniques have more risk of being false positive, as HPV DNA may be present without being etiologically involved.48 In previous studies, HPV DNA detected by PCR showed incidences in oral cavity tumors of 530 to 11%,35 whereas for oropharyngeal carcinomas the incidence is reportedly between 5630 and 64%.49 In our study, 60% of the oropharyngeal carcinomas showed HPV presence detected by PCR, which is relatively high compared to these previous studies. Regarding this result, young age also appears to correlate with high incidence of HPV in oropharyngeal carcinomas. However, when the expression of E6 and E7 is regarded as the gold standard for the presence of HPV, the incidence may drop by 50%.37 We used a highly sensitive PCR hybridization assay, able to detect over 30 subtypes of HPV. This method has been proven to detect HPV DNA and HPV subtypes in both cervical smears and paraffin-embedded material.50
Because HPV influences the cell cycle and leads to p16ink4a overexpression and p53 inactivation (without the need for p53 mutations), we used p16ink4a and p53 immunohistochemistry to evaluate whether these parameters can be used as reliable surrogate markers for HPV presence. In this series, only 81% of all HPV-positive cases showed p16ink4a overexpression, however, in the HPV16-positive subgroup this was 93%. For HPV16-positive cancers, p16ink4a can be considered as a good surrogate marker in this group of young HNSCC patients. Additionally, the absence of p53 overexpression in the HPV pathway was confirmed by our study as p53 overexpression (>75% of cells) was not detected in this group. This is in accordance with findings of Fakhry and Gillison3 who found that HPV-positive tumors usually lack p53 mutation. It should be noted that the combination of IHC staining for both p53 and p16 yielded only a slightly higher specificity for HPV presence. This can be explained by the fact that tumors in which the p16 pathway is activated due to other causes than HPV, p53 mutation is not necessary for oncogenesis.
We thus confirm that p16 immunostaining followed by PCR on the p16-positive cases is the most suitable algorithm.24
A limitation of our study is the small sample size, but HNSCC in young patients remains a rare disease accounting for approximately 6% of HNSCC cases.51 It is therefore highly recommended to pool resources and data of head and neck cancer research groups around the globe for the study of young patients.
To our knowledge, this is the first report in the literature focussing on HPV and other etiologic features in a cohort of patients under 40 years of age. High-risk HPV16 infection was detected in 31% of a total of 45 HNSCC patients. HPV association was not exclusively detected in nonsmoking, nondrinking young HNSCC patients. The presence of p16ink4a accumulation and the absence of p53 overexpression are good surrogate markers for HPV-associated HNSCC in young patients. Whether vaccination with prophylactic HPV vaccines reduces the incidence of oropharyngeal SCC in young nonsmoking patients remains to be elucidated.