Human papillomavirus is a favourable prognostic factor in tonsillar cancer and its oncogenic role is supported by the expression of E6 and E7

Abstract

From 1970 to 2002 in the Stockholm area, we revealed a parallel three‐fold increase in the incidence of tonsillar cancer and the proportion of human papillomavirus (HPV) positive tonsillar cancer cases, indicating a possible role of HPV infection in this disease. We have now examined whether HPV and viral load in pre‐treatment tonsillar cancer biopsies correlates to disease prognosis, and whether the presence of HPV‐16 E6 and E7 mRNA could be ascertained. The presence of HPV‐16, but not viral load, in tonsillar cancer was shown to be a favourable prognostic factor for clinical outcome. Moreover, E6 and/or E7 were expressed in almost all assessable HPV‐16 positive cases, supporting an oncogenic role of HPV‐16 in tonsillar cancer.

1. Introduction

The incidence of tonsillar cancer has increased in many countries (Frisch and Goodman, 2000; Hammarstedt et al., 2007; Syrjanen, 2004). Moreover, in the Stockholm area during 1970–2002, we found a parallel three‐fold increase in the incidence of tonsillar cancer and the number of HPV‐positive tumours, indicating a possible role of HPV infection in this disease (Hammarstedt et al., 2006). The latter is in line with data showing that HPV high‐risk types, similar to those in cervical cancer, are associated with a subset of head and neck squamous cell carcinoma (HNSCC) (Alani and Munger, 1998; Gillison et al., 2000; Gillison and Shah, 2001; Hammarstedt et al., 2006; Mellin et al., 2000; Mork et al., 2001). HPV is mostly found in oropharyngeal cancer, particularly in tonsillar and base of tongue cancer, where HPV, and mainly HPV‐16, is present in 40–72% of cases (Dahlgren et al., 2004; D'Souza et al., 2007; Gillison et al., 2000; Mellin et al., 2000). Patients with HPV‐positive HNSCC smoke to a lesser extent (Gillison et al., 2000; Klussmann et al., 2001) and have a better prognosis than those without HPV in their tumours (Dahlgren et al., 2004; Gillison, 2004; Mellin et al., 2000). Further, in two reports all with very few tonsillar cancer cases, a high, rather than a low HPV‐16 load was correlated to better survival (Kim et al., 2007; Mellin et al., 2002). While HPV‐16 E6 and E7 mRNA expression is essential for transformation in cervical cancer (de Boer et al., 2007), those studies which have addressed this issue have only had a few oropharyngeal or tonsillar cancer cases (Snijders et al., 1992; van Houten et al., 2001; Venuti et al., 2004; Wiest et al., 2002). In the current study, 203 tonsillar cancer patients from the Stockholm 1970–2002 study (Hammarstedt et al., 2006) were re‐examined to see whether the presence of HPV or HPV‐16 viral load influences clinical outcome. HPV‐16 E6 and E7 mRNA expression was also determined in order to further elucidate the possible oncogenic and aetiological role of HPV‐16 in tonsillar cancer.

2. Results

2.1. Patient and tumour features

Information regarding the 203 patients, their treatment, response to treatment and their tumour characteristics, such as presence of HPV and its distribution over time, are shown in Table 1. Of the HPV‐positive cases 86/99 (87%) were HPV‐16 (Hammarstedt et al., 2006). Complete case reports regarding treatment were retrieved for 192 patients (Table 1). HPV distribution was not dependent on tumour stage (TNM stage I and II, compared to TNM stage III and IV) (P=0.63, Pearson's χ ² test). HPV‐positive tumours were more often poorly differentiated in comparison to HPV‐negative tumours and patients had metastasis in regional lymph nodes at diagnosis (P=0.02, Pearson's χ ² test, and P=0.03, Pearson's χ ² test respectively) (Table 1).

Table 1.

Patients and tumour characteristics

Characteristics	Number of patients
	(N=203)	(N=99)	(N=104)
	All	HPV+	HPV−
Number of journals retrieved	203	99 (49%)	104 (51%)
Number of complete case reports	192	99 (51%)	93 (49%)
Age at diagnosis (median, years)	61.5	55.0	65.0
Age at diagnosis (age span, years)	30–88	30–84	47–88
Sex
Male	145	70 (48%)	75 (52%)
Female	58	29 (50%)	29 (50%)
TNM stagea
I	2	0 (0%)	2 (100%)
II	15	10 (67%)	5 (33%)
III	51	34 (67%)	17 (33%)
IV	95	43 (45%)	52 (55%)
Unknown	40	12 (30%)	28 (70%)
Histopathology gradeb
Highly differentiated	21	9 (43%)	12 (57%)
Moderately differentiated	86	34 (40%)	52 (60%)
Low differentiated	87	54 (62%)	33 (38%)
Undifferentiated	3	1 (33%)	2 (67%)
Unknown	6	1 (17%)	5 (83%)
Regional lymph nodes at diagnosis
Yes	120	71 (59%)	49 (41%)
No	44	17 (37%)	27 (63%)
Presence of HPV over time
1970–1979	30	7 (23%)	23 (77%)
1980–1989	42	12 (28%)	30 (72%)
1990–1999	84	48 (52%)	36 (48%)
2000–2002	47	32 (68%)	15 (32%)
Primary treatmentc
Pre‐op radiotherapy	87	62 (71%)	25 (29%)
Post‐op radiotherapy	17	11 (65%)	6 (35%)
Radiotherapy only	46	18 (39%)	28 (61%)
Surgery only	2	1 (50%)	1 (50%)
No treatment or palliative treatment only	30	2 (7%)	28 (93%)
Other treatment/unknown	21	5 (24%)	16 (76%)
Response to radiotherapy
Complete response	96	62 (65%)	34 (35%)
Non‐complete response	30	17 (57%)	13 (43%)
No treatment, palliative treatment or response to radiotherapy unknown	77	20 (26%)	57 (74%)
Disease free 3years after diagnosis	88	66 (75%)	22 (25%)
Recurrence
Recurrence	82	24 (29%)	58 (71%)
Median time to recurrence (months)	6	10	3
Outcome
Dead free of disease	49	22 (45%)	27 (55%)
Dead of disease	72	17 (24%)	55 (76%)
Alive and no evidence of disease	71	60 (85%)	11 (15%)
Outcome unknown	11	0 (0%)	11 (100%)
Follow‐up time
Mean follow‐up time (months)	59.4	77.5	40.8

^aTNM stage, tumor stage according to International Union Against Cancer 1997.

^bTumor differentiation grade according to WHO: International Histological Classification of Tumors.

^cTreatment was less uniform in the 1970s and 1980s as compared the 1990s, and during the latter period 91% of all HPV positive cases were included.

2.2. Presence of HPV in tonsillar cancer is a positive prognostic factor for clinical outcome

In the 5‐year survival analysis, 150 patients were included, since 53 patients either lacked reliable survival data or received palliative treatment. A better disease specific survival was observed in patients with HPV positive tonsillar cancer (81%) compared to patients with HPV negative tumours (36%) (Figure 1, P<0.001, log‐rank test). These results were independent of age, gender and tumour stage according to Cox multivariate regression analysis (HR 0.17; 95% CI 0.09–032) (Table 2). The improved survival for patients with HPV positive tumours was consistent over time and statistically significant (Pearson's χ ² test) for the 1980s (P<0.0001), the 1990s (P=0.036) and for 2000–2002 (P<0.0001), but not for the 1970s (P=0.098), due to fewer cases (data not shown). Relapses were less common in the HPV‐positive group (26%) than in the HPV‐negative group (69%) (P<0.01, Pearson's χ ² test) (Table 1). For patients who did relapse, median to recurrence was 10 months in the HPV‐positive group and 3months in the HPV‐negative group (Table 1).

Figure 1.

Patients with HPV‐positive tumours had a better disease‐specific survival than patients with HPV‐negative tumours, independent of age, gender and tumour stage.

Table 2.

Patient and tumour characteristics and HPV status correlated to survivala

Characteristics	HRb Cox univariate (95% CIc)	HRb Cox multivariate (95% CIc)
HPV	0.17 (0.10–0.30)	0.17 (0.09–0.32)
TNM staged	3.82 (0.93–15.69)	4.06 (0.99–16.71)
Age	1.03 (1.01–1.06)	1.01 (0.98–1.03)
Gender	1.05 (0.63–1.76)	1.15 (0.64–2.06)

^aSurvival hazard ratio according to Cox proportional hazards uni‐ and multivariate regression.

^bHazard ratio.

^c95% confidence interval.

^dTNM stage, tumor stage according to International Union Against Cancer 1997 and TNM stage I–II compared to III–IV.

Data for smoking habits in 119 cases were available, classified as smokers (including ex‐smokers) or non‐smokers; all but one patient with an HPV‐negative tumour received planned curative treatment (data not shown). There were 12 non‐smokers and 42 smokers in the HPV‐positive group, and 3 non‐smokers and 62 smokers in the HPV‐negative group. All non‐smokers in the HPV‐positive group were alive and disease‐free with a follow‐up time of at least two years, compared to 71% smokers in the same group, and 34% smokers in the HPV‐negative group (P<0.05 and P<0.05 respectively, log‐rank test). The 42 smokers in the HPV‐positive group had a better survival than the 61 smokers in the HPV‐negative group (P<0.001, log‐rank test).

2.3. Viral load and survival in HPV‐16 positive tonsillar cancer patients

Between 0.08 and 130 copies/genome‐equivalent were detected in the 86 HPV‐16 positive tumour samples. Depending on viral load, using quartiles as cut‐offs, the patients were divided into four groups: 0.08–2.5 copies/genome‐equivalent, 2.5–11 copies/genome‐equivalent, 11–22 copies/genome‐equivalent, and 22–130 copies/genome‐equivalent in their tumours. There was no difference between the four groups as regards survival. Patients with HPV‐positive tumours had better survival when compared to patients with no HPV in their tumours (P<0.001, log‐rank test).

2.4. Expression of E6‐ and E7 mRNA in HPV‐16 positive tonsillar cancer samples

Extraction of RNA and expression of mRNA, monitored by presence of the housekeeping gene RNase P, was successfully determined in 53/54 of the 86 HPV‐16 positive tonsillar cancer samples. In 20 of the remaining 32 HPV‐16 positive tumours, paraffin blocks were no longer available and in the other 12 there was insufficient material. Directly after RNA extraction and before producing cDNA, a regular HPV‐16 type specific PCR was run with the same primers as used for detection of E6 in the quantitative PCR; no contamination of viral DNA in the samples was seen (data not shown). Expression of HPV‐16 E6 mRNA (estimated as 0.06–9.94 copies/1000 copies RNAsP) and/or E7 mRNA (0.6–68.4 copies/1000 copies RNAsP) was detected in 50/53 (94%) of the samples, i.e. 42 samples expressed both E6 and E7, and 8 samples expressed only E7 (Table 3). In three samples, neither E6 nor E7 could be detected, despite high levels of the housekeeping gene. The levels of HPV‐16 E6 or E7 expression, with a predominance of E7, varied considerably and were not correlated to viral load in the tumours. Further, these values were regarded as only being positive or negative, because only one housekeeping gene was used. It was thus not possible to correlate to patient survival.

Table 3.

Number of HPV‐16 positive tonsillar cancer biopsies also positive for HPV‐16 E6 mRNA and E7 mRNA

	E6 positive	E7 positive	E6 and/or E7 positive	E6 and E7 negative	Total
No. of cases (%)	42 (79%)	50 (94%)	50 (94%)	3 (6%)	53 (100%)

3. Discussion

In this study, from tonsillar biopsies from 203 patients diagnosed with tonsillar cancer between 1970 and 2002, the presence of HPV and HPV‐16 viral load was examined and correlated to patient survival. In addition, HPV‐16 E6 and E7 mRNA expression were analysed. HPV, with HPV‐16 in most samples, was present in approximately half the tumour biopsies. The presence of HPV in tonsillar cancer, independent of viral load, was confirmed to be a strong positive prognostic factor for clinical outcome. This was statistically significant separately for each of the three most recent decades. HPV‐16 E6 and/or E7 mRNA expression could be analysed in 53 tumours, and was found in all but three (94%) of the examined HPV‐16 positive samples, emphasising the role of HPV‐16 in the oncogenesis of tonsillar cancer.

The better response to primary treatment in patients with HPV‐positive tumours could not be explained by tumour stage, which was evenly distributed in the two groups, or by the tumour's ability to metastasise, which was more common in patients with HPV‐positive tumours. Further, patients with HPV‐positive tumours relapsed less often, similar to previous reports (Li et al., 2003). Further investigation is necessary to ascertain whether HPV tumours are more radiosensitive, more immunogenic, and/or have a different natural development. There are now substantial data to support the hypothesis that HPV‐positive tonsillar cancer is a separate different disease entity with better survival (Andl et al., 1998; D'Souza et al., 2007; Gillison et al., 2000; Li et al., 2003; Mellin et al., 2000).

Previous studies have confirmed that a history of smoking is more common in HPV‐negative HNSCC patients (D'Souza et al., 2007; Gillison et al., 2000; Klussmann et al., 2001; Ritchie et al., 2003). In this study, non‐smokers with HPV‐positive tonsillar cancer had a better survival than those who were smokers with either HPV‐positive or HPV‐negative tumours. However, these data must be interpreted with great caution, since the number of patients was limited and the data was collected retrospectively. A larger prospective study addressing these issues would be of clinical value.

The estimated HPV‐16 viral load in this tonsillar cancer study, with 86 patients with HPV‐16 positive tumours, was similar to that reported previously for tonsillar and oesophageal cancer (Klussmann et al., 2001; Mellin et al., 2002; Weinberger et al., 2006; Si et al., 2003). However, we could not confirm that a high viral load correlated with better survival, as was shown previously in studies containing limited numbers of patients (Kim et al., 2007; Mellin et al., 2002).

Expression of HPV‐16 E6 and E7, with a possible slight dominance of E7, both known to be essential for transformation in cervical cancer (de Boer et al., 2007), was found in 50/53 (94%) of assessable HPV‐16 positive tonsillar cancer samples. The samples used for mRNA expression (and for estimation of viral load) had been carefully selected to contain more than 70% tumour material; paraffin‐embedded blocks were microdissected in order to ensure a minimum contamination of normal cells. Contamination of viral DNA in the RNA preparations was also excluded by a control HPV‐16 specific PCR before preparing cDNA. We assayed only for the presence or absence of HPV‐16 E6 and E7 expression, rather than absolute quantities, since limited tumour material only allowed us to test for one of the three human housekeeping genes (Bustin, 2000).

In line with what is known for cervical cancer, our findings strongly support the hypothesis that HPV‐16 is an aetiological factor in the pathogenesis of a proportion of tonsillar cancers. The situation in tonsillar cancer appears to be more complex than in cervical cancer; in that HPV seems to be one aetiological factor, with smoking and alcohol as other factors. Tonsillar cancer is also globally much less common than cervical cancer. However, should the incidence of tonsillar cancer and the proportion of HPV positive tonsillar cancer continue to increase (Hammarstedt et al., 2006), a future scenario may be a considerable incidence, and that the majority of tonsillar cancers are HPV positive. This may warrant different treatment strategies for these patients, as well as different strategies for preventive HPV vaccination, possibly including vaccination of both sexes.

In conclusion, in this, to our knowledge the largest group of tonsillar cancer specimens studied so far, we show that HPV, with a dominance of HPV‐16, is commonly found and is independent of HPV‐16 viral load, a strong positive prognostic factor for patient survival. Moreover, HPV‐16 E6 and/or E7 mRNA was expressed in most of the HPV‐16 positive cancers evaluated, indicating an important role of HPV‐16 in the carcinogenesis of tonsillar cancer.

4. Experimental procedures

4.1. Patients, samples, and response to treatment

This study, which included the paraffin‐embedded pre‐treatment biopsies of 203 patients, (Hammarstedt et al., 2006) was approved by the Regional Ethics Committee at Karolinska Institutet. HTX‐eosin stained sections, cut from formalin‐fixed paraffin‐embedded tumour tissues, were evaluated by a pathologist (J.L.). The degree of histopathological differentiation was estimated and specimens consisting of less than 70% cancer cells were microdissected to minimise contamination of normal cells in the sample. In our previous study, 99 samples were defined as HPV positive by GP5+/6+ PCR, of which 86 were defined as HPV‐16 by HPV‐16 specific PCR and sequencing, three were HPV‐33 positive, one HPV‐35 positive and one HPV‐45 positive, while eight could not be typed (Hammarstedt et al., 2006). The 86 HPV‐16 positive cancer biopsies were further analysed regarding viral load and E6 and E7 mRNA expression. Data regarding age at diagnosis, tumour stage and grade, response to therapy, disease‐free time and survival were collected and correlated to the presence of HPV, HPV‐16 DNA viral load, and presence of HPV‐16 E6 and E7 mRNA expression. For the survival analysis, 30 patients listed as receiving no treatment or palliative treatment only, 21 patients listed as receiving other or unknown treatment, and 2 patients listed being surgically treated only, were excluded (Table 1).

4.2. Quantification of HPV‐16 DNA in tonsillar cancer

DNA extraction was performed, as described previously, on tumour samples described above using tumour biopsies containing >70% tumour material or having being microdissected to avoid the presence of normal tissue (Hammarstedt et al., 2006). The number of HPV‐16 copies/genome‐equivalent was analysed by a quantitative real‐time PCR method (TaqMan) based on the 5′–3′ exonuclease activity of Taq DNA polymerase (Mellin et al., 2002; Dahlgren et al., 2006). Samples were run in triplicate in parallel with standard dilutions of HPV‐16 plasmid (Dahlgren et al., 2006; Mellin et al., 2002). A commercial kit was used according to the manufacturer's instructions to estimate the expression of the human RNase P gene (TaqMan^® RNase P Detection Reagents Kit, Applied Biosystems, Stockholm, Sweden). Calculation of viral load/genome‐equivalent was performed as described by Si et al. (2003).

4.3. Detection of HPV‐16 E6 and E7 mRNA in tonsillar cancer

RNA was extracted using Roche High Pure RNA kit (Roche Diagnostics, Stockholm, Sweden) according to the manufacturer's instructions. Before cDNA synthesis, a regular HPV‐16 type specific PCR (Hammarstedt et al., 2006) was run on the samples to eliminate the possibility of contaminating viral DNA. cDNA was then synthesised from extracted RNA using SuperScript^® III First‐Strand Synthesis SuperMix for qRT‐PCR kit (Invitrogen, Copenhagen, Denmark). A quantitative real‐time PCR with a SybrGreen protocol in an iCycler iQ (iCycler iQ Real‐time PCR Detection System; BioRad, Sundbyberg, Sweden) was used to estimate cDNA viral load as a measurement of mRNA quantity in the samples. New primers were designed in Primer Designer 4 (version 4.2) from Scientific and Educational Software, Durham, NC. E6 primers were 5′‐GAGCGACCCAGAAAGTTACCA‐3′ and 5′‐AAATCCGCAAAAGCAAAGTCA‐3′ and for E7 5′‐ACCGGACAGAGCCCATTACAA‐3′ and 5′‐GTGCCCATTAACAGGTCTTCC‐3′. The 25μl PCR mix consisted of 12.5μl iQ SYBR Green Supermix (Bio‐Rad, Sundbyberg, Sweden) and 1.25μl (10pmolμl⁻¹) each of the HPV‐16 primers. The programme consisted of 50°C for 2min and 95°C for 10min, followed by 40 cycles of denaturation at 95°C for 15s, and annealing and elongation at 60°C for 1min. Finally, a melting curve, starting at 40°C and increasing by 0.5°C every 10s until 120°C was reached, was run to verify the specificity of the obtained amplicons. Samples were run in triplicate with a standard dilution series for DNA quantification. The RNase P kit above was used to estimate expression of the human gene RNase P internal control.

4.4. Statistical analysis

The association between the HPV status and the other variables was examined using Pearson's χ ² test. All significance testing was performed at the 0.05 level and two‐sided P‐values were reported. Survival analysis was performed using the Kaplan–Meier method and the significance between the differences in survival rates was analysed by the log‐rank test. Cox regression multivariate analysis was used to evaluate factors influencing the mortality risk. All analyses were performed using the Statistica version 7.0 statistical computer program.

Lindquist David, Romanitan Mircea, Hammarstedt Lalle, Näsman Anders, Dahlstrand Hanna, Lindholm Johan, Onelöv Liselotte, Ramqvist Torbjörn, Ye Weimin, Munck-Wikland Eva, Dalianis Tina, (2007), Human papillomavirus is a favourable prognostic factor in tonsillar cancer and its oncogenic role is supported by the expression of E6 and E7, Molecular Oncology, 1, doi: 10.1016/j.molonc.2007.08.005.