In Europe, 90,000 new cases of head and neck cancer are diagnosed each year.1 Multiple genetic changes are involved in the development of head and neck squamous cell carcinoma (HNSCC) and known etiologic agents are tobacco and alcohol. During the past 2 decades, however, the role of high-risk human papilloma virus (HPV) has been studied and the data supporting HPV as a causative agent in the development and progression of a sub-set of these cancers has accumulated.2, 3, 4 In HNSCC the overall incidence of HPV is around 25–30%, with a considerable variability in frequency depending on tumor location.5, 6
Human papilloma virus is observed most frequently in tonsillar cancer where 45–100% of the tumors are HPV-positive.6, 7, 8, 9, 10, 11 Furthermore, HPV has been shown to be of favorable prognostic value in patients with HPV-positive tonsillar cancer, but not for patients with other HPV-positive HNSCC, compared to patients with HPV-negative tumors and this is independent of tumor stage.3, 7, 12 In addition, the genetic imbalances in tonsillar cancer vary depending on the HPV status of the tumor,10 indicating that there could be important tumor biological differences between HPV-positive and HPV-negative tumors.
Human papilloma virus is also found commonly in oral cancer where approximately 30% of the tumors are HPV-positive, but similar to HNSCC as a whole, the prevalence of HPV in oral cancer varies considerably between different studies.13 Human papilloma virus has been reported to be present in 0–100% of carcinoma of the tongue,3, 4, 11, 14, 15, 16, 17, 18, 19 but most of these studies included samples from various sites of the head and neck, and samples originating from the tongue were few. Moreover, in some reports it was not clear if the analyzed samples originated from the oral (mobile) tongue, from the base of tongue or from both locations. The rate of HPV infection could be expected to be more frequent in tumors of the base of tongue, because the base of tongue is histologically related to the tonsil, and the two belong to the lymphoid area called Waldeyer's ring, which is a histological/functional entity.20
The first aim of our study was to determine the overall frequency of HPV in tongue carcinoma and to see if the presence of HPV infection varied in carcinoma of the tongue at different sub-locations. Secondly, the influence of HPV on clinical outcome was addressed, particularly in base of tongue cancer, which histologically resembles tonsillar cancer where HPV has been found to have a favorable influence on prognosis.7, 8, 10 In addition, the influence of HPV in mobile tongue cancer Stage I was addressed. When our study was initiated our hypothesis was that mobile tongue cancer Stage I, a disease with a relatively good prognosis, would frequently be HPV-positive. A subgroup therein, however, with a DNA aberration comparable to that of advanced carcinomas and prone to local or regional recurrence21 was assumed to be HPV-negative.
To pursue these studies, a well-defined study material was collected, consisting of tumors from the base of the tongue as well as tumors of the mobile tongue. These tumors were then tested for presence of HPV DNA with a polymerase chain reaction (PCR) using general HPV primers and, when present, HPV was typed using HPV-type specific PCR and sequencing. Thereafter the clinical outcome of the patients was reviewed for possible correlation to the presence of HPV.
MATERIAL AND METHODS
Patient samples, clinical background and treatment
Paraffin-embedded, pre-treatment biopsies of primary squamous cell carcinoma of the tongue were obtained from 110 patients treated between 1970 and 2002 at the Karolinska University Hospital. This research was conducted according to ethical permission (no. 99-237). Two criteria had to be fulfilled for patients to be included in the study; no prior tumor disease in the head and neck region and a verified diagnosis. Patients with tumors originating from the dorsal surface and lateral borders, anterior to vallate papillae (anterior two-thirds, ICD-O code C02.0,1) and the inferior (ventral) surface (ICD-O code C02.2) or the base of the tongue i.e., posterior to the vallate papillae (posterior third, ICD-O code C01), were included in the study. For simplicity sites encoded by C02.0.1 and C02.2 are hereafter called the “mobile tongue.” Eighty-five patients were diagnosed with mobile tongue cancer (34 females, mean age = 65 years, range = 32–92; 51 males, mean age = 61 years, range = 30–84) and 25 were diagnosed with base of tongue cancer (7 females, mean age = 62 years, range = 52–81; 18 males, mean age = 62 years, range = 35–76 years). A pathologist confirmed the diagnosis on hematoxylin-eosin (H&E) sections and verified that the sections contained >70% tumor cells. For details of the tumor site, differentiation grade, tumor stage, treatment and response to therapy see Table I. For 3 patients with a Stage I cancer, the available tissue block contained only cancer in situ lesions, however, the patients had invasive cancer and were treated accordingly and were hence included in the HPV analysis, and subsequent analysis. TNM stage classification was done according to the International Union against Cancer (UICC 5th ed., 1997) and the differentiation grade according to the WHO International Histological Classification of tumors. The clinical data of the patients were retrieved from the files of Radiumhemmet and of the Department of Oto-Rhino-Laryngology, Head and Neck Surgery at the Karolinska University Hospital (Table I). The majority of the patients with a Stage I tumor of the mobile tongue were treated with local resection, whereas most patients with more advanced tumors and tumors at the base of the tongue received pre-operative radiotherapy followed by surgery. Only a limited number of patients received no or palliative treatment (Table I).
Table I. Clinical Data of the Patients and Tumor Characteristics
Number of patients
Mobile tongue (n = 85)
Base of tongue (n = 25)
Values are percent of mobile tongue patients and base of tongue patients respectively.–
TNM stage, tumor stage according to International Union Against Cancer 1997.–
Tumor differentiation grade according to WHO: International Histological Classification of tumors.
DNA was extracted from 5 × 5 μm tumor sections. Before and after the sections a 4-μm section was taken for staining with H&E for tumor tissue verification. To check for HPV contamination an empty block was cut between each tumor block and the sections were collected and treated as if they had been tumor sections. DNA was extracted according to a standard phenol extraction protocol or according to the manufacturer's instruction for the High Pure RNA Paraffin Kit (Roche Molecular Biochemicals, Mannheim, Germany) with exclusion of the DNase treatment.
Verification of amplifiable DNA by S14 PCR
All tumor samples were run in a S14 PCR for verification of the presence of amplifiable DNA as previously described.10 Water was used as negative control and DNA extracted from normal tissue as positive control. The PCR products were run on a 2.5% agarose gel stained with ethidium bromide and visualized with UV-light. Samples with a band length of 127 base pairs (bp) were considered to contain amplifiable DNA and subsequently screened for presence of HPV.
HPV detection by PCR
Human papilloma virus was detected as previously described8 with the general primer pair GP5+/6+.22 As positive controls cloned plasmids of either HPV-6 or -16 were used and water was used as negative control. The products were detected on an agarose gel as described above and samples with a band at 130–150 bp were considered to be positive. Positive samples were either sequenced as previously described8 or run in HPV type-specific PCRs for types −16, −18 and −33 as previously described.8 For verification of the typing, some samples were both sequenced and run in HPV type-specific PCR.
Samples negative in the GP5+/6+ PCR were run in another general PCR with the CPI/CPIIG primer pair23 to exclude false negatives as a result of a disrupted L1 gene. The PCR was run as previously described.8 The products were detected on an agarose gel as described above and samples with a band at 187 bp were considered to be HPV-positive and were run in HPV type-specific PCR or sequenced as previously described.8 To verify the sensitivity of the PCR protocol a dilution series of paraffin embedded SiHa cells were included in the GP5+/6+ and CPI/CPIIG runs.
Exact logistic regression and Fischer's exact 2-tailed test were used to correlate the frequency of HPV in tongue cancer patients to clinical data. The association between diagnosis and different endpoints were assessed using a Cox proportional hazard regression model controlling for confounders. The assumption of time independent hazard ratio was investigated by including the covariates as a function of time. Kaplan-Meier curves were used for graphical presentation of survival and Log-rank test was used to analyze significant differences in survival rate. Cox regression model was used for 3- and 5-year survival analysis. For the multivariate Cox proportional hazard regression model a backward selection was used, and the factors analyzed were grade of differentiation, presence of HPV, gender, stage and age at diagnosis. For the exact logistic regression tests the LogExact 5 statistical software was used and for the Cox proportional hazard regression models SAS 8.2 was used.
Prevalence and type specificity of HPV in tongue cancer
One hundred ten biopsies (85 mobile tongue tumors and 25 base of tongue tumors) had amplifiable DNA and were screened for the presence of HPV DNA (Table I). Only 12/110 (10.9%) samples were found to be HPV-positive by either the L1 specific consensus primers GP5+/6+ (10 samples) or by the E1 specific consensus CPI/CPIIG primer pair (2 samples). However, HPV was significantly more frequently found in base of tongue cancer samples (10/25, 40%) compared to mobile tongue cancer samples (2/85, 2.4%), (p < 0.001, Fischer's exact 2-tailed test). This was irrespective of gender or age group and when analyzed separately (Stage I + II tumors and Stage III + IV tumors) true for Stage I + II tumors (p = 0.0068, Fischer's exact 2-tailed test). Notably, none of the mobile tongue cancer Stage I–II samples were HPV-positive. For advanced tumors a discrepancy in the frequency of HPV could not be found between mobile tongue cancer (2/13) and base of tongue cancer (7/16), however, the number of cases was limited. When analyzing all tumors it was found that HPV was more common in undifferentiated tumors than in moderately differentiated (p = 0.0077, Exact logistic regression) and well differentiated tumors (p = 0.0009, exact logistic regression) and in large tumors (Stage III + IV) compared to small tumors (Stage I + II) (p < 0.001, Fischer's exact 2-tailed test). Nevertheless, the only factor that could explain presence of HPV in a multivariate analysis was the base of tongue cancer diagnosis with an odds ratio (OR) for base of tongue patients to be HPV-positive at 26.4.
HPV type 16 was predominant in both mobile tongue cancer and base of tongue cancer. The HPV type-specific PCR and sequencing showed that 9 samples (2 from mobile tongue and 7 from base of tongue) harbored the high risk HPV type 16, and one sample each (both from the base of tongue) harbored the high risk HPV types 33 and 35. One sample was not HPV typed due to limited amount of material.
Tumor characteristics in correlation to clinical outcome
Patients with a mobile tongue cancer Stage I tumor were followed for 11–175 months (mean 67 months), whereas patients with a Stage II–IV mobile tongue tumor were followed for 0–48 months (mean = 12 months). Patients with a base of tongue tumor for were followed for 3–142 months (mean = 38 months), with a mean of 59 months for HPV-positive patients and 24 months for HPV-negative patients.
Ninety-four patients were disease free 1 month after primary treatment (76/85 mobile and 17/25 base of the tongue cases), whereas 6 patients with large tumors in the mobile tongue and 3 patients with base of tongue cancer had residual disease. The remaining 8 patients had received either no or palliative treatment, or could not be evaluated successfully after treatment.
An analysis of the recurrence rate in HPV-positive vs. HPV-negative patients could not be carried out for patients with cancer of the mobile tongue due to too few HPV-positive cases.
An analysis of the recurrence rate was, however, carried out separately for base of tongue cancer patients. In the 10 HPV-positive patients with recurrence the mean time to recurrence was 17 months, whereas for the 15 HPV-negative patients the mean time to recurrence was only 8 months.
A 3-year follow-up period was used for calculation of the disease-specific survival, because most patients that suffer from recurrence do this within the first 2 years after diagnosis. The disease-specific survival for the whole group was, as expected, influenced by tumor location, tumor differentiation and tumor stage.
A 3-year survival analysis was also carried out separately for patients with base of tongue cancer. Patients were included if they had been followed for 36 months or had died of the disease within 3 years from diagnosis. Nineteen patients (7 HPV-positive and 12 HPV-negative) fulfilled the criteria and were evaluated. HPV as well as tumor stage were of importance for the 3-year disease-specific survival of these patients. Patients with HPV-positive base of tongue cancer survived significantly longer than patients with HPV-negative base of tongue cancer (6/8 HPV-positive patients were alive after 3 years vs. 2/11 HPV-negative patients, p = 0.0159, Cox regression, multivariate). Seventeen patients were included in the survival analysis regarding tumor stage. Four patients had tumors in Stages I–II and 13 patients had tumors in Stages III–IV. Patients with tumor in Stages I–II had a better disease-specific survival than patients with tumors Stage III–IV (p = 0.0376, Cox regression, multivariate). At 5 years after diagnosis, however, HPV was still a positive factor for disease-free specific survival (p = 0.0362, Cox regression, multivariate; Fig. 1), whereas tumor stage was no longer of importance (p = 0.0863, Cox regression, multivariate).
In our study the prevalence of HPV in tongue carcinoma as well as the variation of HPV at different locations of the tongue was examined. In addition, the clinical outcome of the patients was reviewed and correlated to the presence of HPV.
The overall frequency of HPV in carcinoma of the tongue in our study was found to be 10.9%, which is within the range reported previously.3, 4, 11, 14, 15, 16, 17, 18, 19 The reported frequency varies considerably between studies, possibly due to variation in the composition, the preparation and the storage of the material as well as due to the use of different detection methods. In the study by Kantola et al.,24 none of the analyzed 105 (paraffin-embedded) mobile tongue tumors, including tumors in different stages, were HPV-positive, whereas in a study by Koskinen et al.,11 where the location of the tongue was not defined 73% (11/15) were found to be HPV-positive (fresh frozen).
In the material analyzed in our study, tumor location played a significant role, and only 2 of 85 (2.4%) of the tumors originating from the mobile tongue were HPV-positive, whereas 10 of 25 (40%) tumors from the base of tongue were HPV-positive (p < 0.001). The OR for patients with a base of tongue cancer to be HPV-positive was 26.4 with a 95% confidence interval (CI) of 4.95–271. The large span in CI is possibly indicating a low number of patients. Similarly, in a study where 17 patients with base of tongue cancer were analyzed for presence of HPV, by serology, a similar OR of 20.7 and a 95% CI of 2.7–160.1 was obtained.4 When tumors from the tongue (base of tongue excluded) were analyzed in the same study,4 a lower risk of being HPV-positive was seen among those patients (OR = 2.8) compared to base of tongue cancer patients. The difference in HPV frequency between tumors of the mobile tongue and of the base of tongue in our study was therefore not unexpected. In fact, our results are in line with studies published previously where a higher frequency of HPV-associated tumors has been reported in the oropharynx compared to that in other parts of the head and neck.3, 25 Moreover, in one serological study26 an increased HPV type 16 seropositivity was found for patients with oropharyngeal cancer, but not for patients with a cancer in the oral cavity.
The results in our study are compatible with published reports27, 28 proposing that HPV 16 is found primarily in cancers originating from locations with inflammatory activity, such as the cervix, the tonsil and the base of tongue. To be speculative one can suggest that HPV-infected cells, in the vicinity of immunologically competent cells that may respond frequently to various foreign antigens inducing growth factors, may also be stimulated to divide and that this may facilitate tumor development.
In our study the frequency of HPV infection in base of tongue cancer was found to be 40% and most samples typed HPV-16, which was similar to the results published previously in tonsillar cancer.3, 7, 8, 9, 10, 11 In addition, similar to what has been observed previously in tonsillar cancer,3, 7, 8, 10 the presence of HPV in base of tongue cancer positively influenced the 3-year disease-specific survival (p = 0.0159). Furthermore, after 5 years, HPV-positive base of tongue cancer patients still had an advantage over HPV-negative base of tongue cancer patients in disease-specific survival (p = 0.0362), whereas the tumor stage at the time of diagnosis no longer had an impact on the survival rate (p = 0.0863). To our knowledge, this is the first report where HPV has been suggested to be a positive factor for clinical outcome in base of tongue cancer patients. This is possibly due to that previous HPV studies on clinical outcome have not addressed this issue in as many base of tongue cancer cases.
None of the Stage I mobile tongue tumors were HPV-positive and hence, a correlation of HPV to prognosis could not be done.
In conclusion, HPV is significantly more common in base of tongue cancer than in cancer of the mobile tongue, and the frequency and positive impact of HPV on disease-specific survival in patients with base of tongue cancer resembles that observed for tonsillar cancer patients.