Human papillomavirus type 16 is episomal and a high viral load may be correlated to better prognosis in tonsillar cancer

Authors


Abstract

The aim of our study was to investigate the physical state and the viral load of HPV-16 in tonsillar cancer and to correlate these findings with clinical outcome. To distinguish between integrated and episomal forms of HPV, 22 fresh-frozen tonsillar cancer samples were analysed by a method based on restriction enzyme cleavage, ligation and PCR (rliPCR). HPV-16 was detected in 11/22 and HPV-33 in 1/22 of the cancers, hence 12/22 (55%) of the tumours were HPV positive. Only extrachromosomal forms of HPV-16 were observed. Full-length episomal HPV was detected exclusively in 7/11 of the cancers, whereas both full-length and deleted forms of episomal HPV-16 were found in parallel in 2 other tumours. In 1 tumour only a deleted episomal form of HPV-16 was present. In the remaining HPV-16 positive tumour both full-length episomal as well as an 11 kbp PCR product were detected and if the 11 kbp product contained integrated HPV, or was off-size linearised episomal could not be determined. In 2 cervical cancer controls, HPV-16 was integrated and could be chromosome located. HPV-16 was quantified by real-time PCR and most tonsillar cancers contained between 10 to a few hundred copies of HPV per β-actin. The 6 patients with tumour sections with ≥190 HPV-16 copies/β-actin remained tumour free (p = 0.026) and had a better survival rate (p = 0.039) when compared to the 5 patients with tumours sections with ≤60 HPV-16 copies/β-actin. In conclusion, HPV-16 is mainly episomal in tonsillar cancer. The viral load showed a wide distribution and the clinical outcome in our study was better when the HPV load was higher. © 2002 Wiley-Liss, Inc.

Accumulating molecular and epidemiological data indicate that the high-risk types of human papillomavirus (HPV) are not only associated with cervical cancer, but may also be associated with certain subtypes of cancer in the head and neck.1, 2, 3, 4, 5 The strongest association has been found for oropharyngeal squamous cell carcinoma, especially tonsillar cancer, where HPV DNA is present in 45–70% of the cases.1, 2, 3, 5, 6, 7, 8, 9 In a nested case-control study it was found that patients who were sero-positive for HPV-16 had a 14.4 excess risk of developing oropharyngeal cancer later in life, and in another study patients with a history of HPV-related anogenital cancer had a 4.3 higher risk of developing tonsillar cancer.4, 10 Furthermore, patients with HPV positive tonsillar cancer seem less likely to be heavy smokers and drinkers, and to have a better prognosis than patients with HPV negative tonsillar cancer.3, 7, 8

Similar to cervical cancer, the oncogenes of HPV-16, E6 and E7, are generally expressed in HPV positive tonsillar cancer.1, 9, 11, 12 The viral protein E6 promotes degradation of p53, whereas E7 inactivates pRb.13, 14 P53 mutations have been reported to be less frequent in HPV positive tonsillar cancer as compared to HPV negative tonsillar cancer, although when assayed by immunohistochemistry the proportion of elevated p53 levels are similar in both HPV positive and negative tumours.3, 12, 15 In addition, HPV positive tonsillar cancers seem to have a decrease of pRb, possibly due to E7 activity.2, 9

In cervical cancer the HPV genome is mainly integrated in the host genome.16, 17 Integration leads to disruption and deletion of the viral genes E1 or E2 open reading frame (ORF), which are of importance for viral replication and viral transcription.18 Disruption of E2 may thus allow the dysregulation of E6/E7 expression, which is required in HPV associated cervical cancers and essential for maintenance of the malignant phenotype.18 Integration may result in a growth advantage for the malignant cell and could be an important step in the carcinogenesis of cervical cancer.19 In 15–30% of the HPV-16 positive cases of cervical cancer, however, only the episomal form of HPV-16 is found.16, 20, 21 The E6 and E7 transcription level is regulated by a promoter in the long control region (LCR) and is influenced by viral as well as cellular transcription factors.18 In cervical cancer biopsies with episomal HPV-16, genetic changes in the LCR, for instance in the binding sites for the transcription factor Yin Yang 1 (YY1), and subsequent elevated activity of the E6/E7 promoter have been reported.22, 23, 24

Data concerning the physical state and viral load of HPV in tonsillar cancer is so far limited. One study that included 2 HPV-16 positive biopsies of tonsillar cancer reported that both biopsies harboured episomal forms of HPV and another study has measured the viral load in 6 tonsillar cancers.8, 11 The aim of our study was therefore to investigate the physical state and the viral load of HPV in tonsillar cancer further and if possible correlate these findings to clinical outcome. Twenty-two fresh frozen tonsillar cancer samples were analysed for this purpose. A method described previously by Kalantari et al.21 based on restriction enzyme cleavage, ligation and polymerase chain reaction (PCR) was used to distinguish between integrated and episomal forms of HPV. Furthermore, 2 independent PCRs were used to investigate the integrity of E1 and E2 ORF.20, 25 The viral load was quantified by an HPV-16 specific real-time PCR.

MATERIAL AND METHODS

Patient samples and clinical background

Fresh-frozen biopsies were obtained from primary tonsillar squamous cell carcinoma (SCC) from 22 patients, diagnosed between 1990–1999 at the Department of Otorhinolaryngology, Head and Neck Surgery at Karolinska Hospital. The biopsies were taken at the time of diagnosis, before treatment. A pathologist different from the one who set the original diagnosis on the haematoxylin-eosin sections of the paraffin embedded biopsies, confirmed the histo-pathology on haematoxylin-eosin sections from the fresh-frozen biopsies and checked that all biopsies consisted of at least 80% of cancer cells. Six of the 22 patients were female (27%) and the mean age of the patients was 65 years (41–87 years). TNM stage classification was done according to the International Union Against Cancer (UICC). For a summary of the patients and their tumours see Table I. Three fresh-frozen cervical cancer biopsies, characterized previously, (F725, F826, F3155) were used as controls.21

Table I. Clinical Data, HPV Type, Viral Load and Physical State of HPV-16 in Tonsillar Cancer
PatientGenderTNM1Nodal status2Histology3HPV statusViral load4Physical state of HPV
  • 1

    TNM, tumour stage according to International Union Against Cancer.

  • 2

    Nodal status, local lymph node metastasis.

  • 3

    Cell tumour differentiation grade according to histopathology.

  • 4

    Viral load, HPV-16 copies/β-actin copy.

  • 5

    Female/male ratio.

F1FIVNegPoorly diffNeg  
F2MIIIPosUndiffHPV-16240Full-length episomal
F4MIIINegPoorly diffHPV-1615,400Full-length episomal
F5MIVPosPoorly diffNeg0 
F7FIVPosUndiffHPV-1630Deleted episomal
F8MIVPosPoorly diffNeg  
F9MIINegPoorly diffNeg  
F10MIVPosPoorly diffNeg0 
F11MIVPosUndiffHPV-1610Full-length episomal
F12MIVPosPoorly diffHPV-1630Full-length episomal + 11kb
F13MIVPosPoorly diffHPV-16720Full-length episomal
F14FIIIPosUndiffHPV-1610Full-length + deleted episomal
F15MIVPosPoorly diffNeg  
F16MIVPosUndiffHPV-33  
F17MIVPosUndiffNeg  
F18MIVNegPoorly diffHPV-1660Full-length episomal
F19MIVPosUndiffNeg  
F20FIVPosUndiffHPV-16350Full-length episomal
F21MIVPosUndiffNeg  
F23FIIINegUndiffNeg  
F24FIIIPosUndiffHPV-16230Full-length episomal
F25MIVPosUndiffHPV-16190Full-length + deleted episomal
Total6/225 (27%)   12/22 (54.5%)  

DNA extraction and HPV screening

Three 20 μm sections were cut from each biopsy, with control sections between each sample to check for contamination from the cryotome. After DNA extraction by standard phenol-chloroform and proteinase-K digestion, the samples were screened for HPV by using consensus primers GP5+/6+ (located in L1) and the samples negative for the GP5+/6+ primer pair were tested with consensus primers CpI/IIG (located in E1) to detect HPV that may have integrated and lost the L1 region.26, 27 Conditions for the GP5+/6+ were the following; the 50 μl PCR mixture contained 5 μl 10 × PCR buffer II (Applied Biosystems, Foster City, CA), 200 μM of each dNTP, 3.5 mM MgCl2, 25 pmol of each primer and 1 U of Taq DNA polymerase (AmpliTaq Gold DNA polymerase, Applied Biosystems) and 5 μl of extracted DNA. Amplification was run in an automated thermocycler (GeneAmp PCR system 9700, Applied Biosystems). The cycles consisted of an initial denaturation of 5 min at 94°C followed by 40 cycles of 95°C for 30 sec, 44°C for 60 sec, 72°C for 90 sec and finally 72°C for 10 min. The CpI/IIG PCR were run under the same conditions with the exception of that 3 mM MgCl2, 0.05% BSA, 17 pmol of CpI and 26 pmol of CpIIG, and 2.5 U of Taq DNA polymerase were used, and that the PCR program consisted of 5 min at 94°C, 35 cycles of 95°C for 60 sec, 55°C for 60 sec, 72°C for 120 sec and then 72°C for 10 min. HPV typing was done by direct cycle sequencing of the purified PCR products from the consensus primers using the Big Dye Terminator Cycle Sequencing Kit, carried out in ABI PRISM 377 DNA Sequencer (Applied Biosystems). Both DNA strands were sequenced and aligned to those available at NCBI BLAST GenBank (http://www.ncbi.nlm.nih.gov/BLAST/). The samples were also tested with HPV specific primers for HPV-16, -18 and -33 as described previously.28 To rule out false negative results, the HPV negative samples were assayed with primers GH26/27 of the HLA DQ locus.29

E2 and E1 gene detection

To investigate the integrity of E1 and E2 found in episomal HPV, both entire genes were amplified in 2 separate PCR assays as described previously.20, 25 For E2 amplification the following primers were used; 5′-AGGACGAGGACAAGGAAAA-3′ and 5′-TGTTTAGAACTATGACGTAGG-3′.20 For E1 amplification the following primers were used; 5′-TGTGCCCCATCTGTTCTCA-3′ and 5′ GGCGCATGTGTTTCCAATAG-3′.25

rliPCR

To investigate if tonsillar cancers harbour integrated or episomal HPV-16, a method described earlier by Kalantari et al.21 based on restriction enzyme digestion, ligation and inverse PCR (rliPCR) was used (Fig. 1). Running long template PCR using inverse primers, the Long A-S/AS primer pair that are specific for HPV-16, the entire episomal HPV-16 genome (7,904 bp) is amplifiable in a single PCR, because circular DNA is amplified by inverse PCR (Fig. 1a). In contrast, integrated, linear, HPV-16 will not be amplified by inverse primers (Fig. 1b). If the samples are digested with a restriction enzyme that cuts human DNA fairly regularly, however, and then circularised by self-ligation, both episomal and integrated HPV-16 can be detected at the same time (Fig. 1c). Integrated HPV detected this way will also yield PCR fragments containing virus-human junction sequences. Inverse PCR was run on undigested DNA (shows only episomal), as well as cleaved DNA and on cleaved and ligated samples (rliPCR). As controls SiHa (1-2 copies of integrated HPV-16 per cell), 2 cervical cancer samples (F826, F3155) with known integration sites, and HPV negative cell line (MCF-7) were used.21

Figure 1.

Methodical overview of restriction enzyme digestion, ligation and inverse PCR (rliPCR). (a) Circular DNA as episomes yield PCR products directly with inverse PCR, and also after cutting and recirculating as in rliPCR. (b) Integrated HPV gives no PCR product on inverse PCR. (c) RliPCR on integrated HPV gives PCR products with human-viral junctions.

Extracted DNA (0.3–1 μg) was digested with either Hind III or Taq I, in 37°C or 65°C respectively over night. (Hind III does not cleave HPV-16, and Taq I cleaves at 1 site in E6.) Ten μl of the cleaved DNA was saved for PCR and 40 μl was ligated by T4 ligase (Rapid DNA ligation Kit, Roche). The inverse PCR with Long A-S/AS primers was run using the Expand™ long Template PCR System (Roche). The forward primer Long A-S with the sequence 5′-CAGTAGTGGAAGTGGGGGAGAGGGTGTTAGTG-3′ and the backward primer Long A-AS 5′-CCTGTATCACTGTCATTTTCGTTCTCGTCATC-3′are both located at the very 5′ end of E1. Conditions for the PCR were the following; the 50μl PCR mixture contained 5 μl 10 × PCR buffer 1 (supplied in the kit), 350 μM of each dNTP, 50 pmol of each primer and 2.265 U of Taq DNA polymerase mix (included in the kit) and 5 μl of template. The PCR program was set as follows: an initial hold at 94°C for 2 min, 16 cycles of denaturation at 94°C for 15 sec and annealing/extension at 68°C for 15 min, followed by 14 additional cycles with the same conditions except for 15 sec increments per cycle of each annealing/extension step, and finally a last extension step at 72°C for 10 min. The PCR products were visualized on an 0.8% agarose gel containing ethidium bromide.

DNA sequencing by primer walking

PCR products from inverse PCR on undigested and digested and ligated DNA different from the episomal length of 7,904 bp were gel-extracted (Qiagen Kit) and cycle sequenced as described above. The sequencing was initiated using the Long A-S primer, and was continued by constructing a new primer from the sequence obtained from the previous run, i.e., “primer walking.” Sequencing was carried out for episomal HPV from the Long A-S primer until the sequence of the Long A-AS primer was reached or for integrated HPV-16 where human DNA also was sequenced, until the other end of HPV-16 was reached (reference sequence of HPV-16 complete genome in BLAST GenBank was considered NC_001526).

HPV quantification by real-time PCR

To estimate the amount of HPV-16 copies per human genome equivalent a quantitative real-time PCR method (TaqMan), based on the 5′-3′ exonuclease activity of Taq DNA polymerase was used. To create a standard with a known amount of virus copies, a PCR was run on cloned HPV-16 (pBR322) using HPV-16 specific primers located in E6 as described above.28 The HPV-16 specific product of 119 bp was then inserted into a pGEM-T Easy Vector. A dilution series of the pGEM-T plasmid with HPV-16 E6 insert was made and used in each real time assay as a standard to calculate the number of viral copies. The real-time PCR was carried out with the same HPV-16 specific primers with the addition of a fluorogenic probe (16E6TQP) located in HPV-16 E6 designed by Dr. Kalantari with the following sequence 6-FAM-CCGGTCCACCGACCCCTTATATTATGGAATCTT-TAMRA-3′-phosphate. The PCR volume of 25 μl consisted of 2.5 μl TaqMan PCR Buffer (Applied Biosystems), 200 μM of each dNTP, 1.5 mM of MgCl2, 10 pmol of each primer, 5 pmol of the probe, 0.5 U Taq Gold DNA polymerase (Applied Biosystems), and 10 μl of template. The PCR was carried out in a PE Applied Biosystems 7700 Sequence Detector with an initial step of 50°C for 2 min, 95°C for 10 min, followed by 40 cycles of 15 sec at 95°C and 1 min at 60°C. The tonsillar tumours that were found HPV-16 positive, 2 HPV-16 negative tonsillar tumours, 2 cervical cancers biopsies (F3155, F725), CaSki and negative water blanks were done in duplicates and run in parallel with the standard dilutions of HPV-16 plasmid in each analysis. As an internal positive control and as an estimation of the human genome content in the tonsillar samples, a quantitative real-time PCR was run on β-actin as instructed by the kit suppliers (Applied Biosystems).

Statistical analysis

Pearson's χ2 test was used to analyse the results when correlated to number of disease free patients. Survival analysis was done by using the Kaplan-Meier method. The significance between the difference in survival rate was analysed by the log-rank test.

RESULTS

Frequency, type and rliPCR of HPV in tonsillar cancer

HPV was detected in 12/22 (55%) of the tonsillar cancers and of the HPV positive tumours 11 were HPV-16 positive and 1 was HPV-33 positive (Table I). The 12 HPV positive tumours were all detected with the GP5+/6+ primers and no additional HPV positive tumours were found when using the CpI/CpII primers.

All HPV-16 genomes could be amplified by rliPCR. Only full-length episomal bands of 7,904 bp (full-length HPV-16 genome) were detected in 7/11 tumours (Table I and Fig. 2, demonstrated for F2, F11 and F13 in lanes 2–7). One tumour (F7) exhibited 2 bands 2,071 bp and 4,140 bp, both shorter than the full-length episomal size, and which both could be obtained using inverse PCR on undigested DNA and by rliPCR (Table I and Fig. 2, lane 12). These PCR products (both undigested as well as digested and ligated) were gel extracted and sequenced, using primer walking (sequences not illustrated). The shorter band of F7 was shown to harbour HPV-16 that was disrupted in the E1 ORF (at nucleotide [nt] 2,388), and with a deletion of 5,455 bp, the HPV genome then continued with the LCR (at nt 7,843). The longer band of F7 showed exactly the same sequence, with the disruption in E1 at nt 2,388, but was present as a dimeric deleted episomal form. A third pattern was observed in the remaining 2 tumours (F12, F14 and F25). In these tumours, both a full-length episomal band and a band different from the 7,904 bp were detected (Table I and Fig. 2, illustrated for F12 in lane 11, for F14 in lanes 9 and 10, and for F25 in lane 13). The PCR amplicons (both undigested as well as digested and ligated) differing from 7,904 bp were gel extracted and sequenced, using primer walking (sequences not illustrated). In F14 the 2,712 bp band showed a disruption of the E1 ORF (at nt 2,786), and after a deletion of 4,814 bp the sequence continued directly in the LCR (at nt 7,600). Because F14 yielded the same 2 PCR products when running an inverse PCR on undigested DNA as well as after Taq I and ligation (rliPCR), we conclude that the biopsy contained a full-length episomal and a deleted episomal form of HPV-16 (Table I and Fig. 2, lanes 9 and 10). F25 was similar to F14 and also contained full-length episomal and a deleted episomal form of HPV-16 (Fig. 2, lane 13). The deleted episomal band (1,969 bp) of F25 was disrupted in the E1 ORF at nt 2,640, and after a deletion of 5,557 bp it continued in the E6 ORF (at nt 294). F12 showed a full-length episomal band with rliPCR (Hind III cut and ligated) and in addition a 11 kbp band (Fig. 2, lane 11), which was not detected when DNA was not digested (data not shown). Unfortunately the 11 kbp band could not be gel extracted and sequenced so it could not be determined if F12 contained episomal HPV-16 as well as integrated HPV-16. In summary, exclusively full-length episomal HPV was detected in 7/11 of the cases, full-length and deleted episomal HPV was observed in 2/11 of the tumours and only deleted episomal HPV DNA was detected in 1 of the tonsillar cancers. The remaining sample (F12) contained full-length episomal HPV, but if it also harboured integrated HPV DNA could not be excluded.

Figure 2.

Agarose gel electrophoresis from inverse PCR. Lanes 1 and 8, λ-Hind III marker. Lanes 2, 3 and 6, full-length (7,904 bp) episomal HPV in uncleaved samples of tumours F2, F11 and F13 respectively. Lanes 4, 5 and 7, full-length episomal HPV after restriction cleavage and ligation, rliPCR, of respective tumours F2, F11 (Hind III cleaved and ligated) and tumour F13 (Taq I cleaved and ligated). Lane 9, uncleaved tumour F14 displaying full-length and deleted episomal HPV and lane 10, the same band products in F14 after rliPCR (Hind III cleaved and ligated). Lane 11, tumour F12 after rliPCR (Hind III cleaved and ligated). Lane 12, tumour F7 after rliPCR (Hind III cleaved and ligated). Lane 13, tumour F25 after rliPCR (Taq I and ligated).

E2 and E1 amplification in tonsillar cancer

In all HPV-16 positive samples, except for F7, both the E2 gene and the E1 gene could be amplified with the expected size (data not shown). This confirmed the results from rliPCR in the way that these genes were intact in as in episomal forms of HPV-16, with the exception of the deleted episome in F7. In tumour F7, E1 but not E2 could be detected, in line with the rliPCR results (data not shown).

rliPCR results on cervical cancers and additional controls

Two cervical cancer control samples (F826, F3155) showed no PCR product when running inverse PCR on undigested DNA, but as expected when cutting with Taq I and thereafter ligating, HPV PCR products were detected in both cervical cancers (data not shown) The PCR products were then sequenced. F826 was found to be disrupted in E1 (at nt 1,680) and integrated in chromosome 1q24 (according to GenBank chromosome location data not shown). F3155 was found to be disrupted in E2 (pos 3598) and chromosome located at chromosome 3q28 (data not shown).

SiHa, known to harbour only 1–2 copies of integrated HPV-16, showed a PCR product (5.5 kbp) only after Taq I digestion and ligation as expected (data not shown) and the MCF-7 (HPV negative cell line) lacked an HPV PCR product (data not shown).

To exclude that a low-level of integrated HPV could be obscured by a dominating amount of episomal HPV, we mixed an episomal HPV DNA (1,700 bp) from 1 tonsillar cancer with integrated HPV DNA (3,200 bp) from 1 of the cervical cancers. Even when an excess of 30–600 times more episomal HPV-16 DNA was mixed with integrated HPV DNA, both the longer integrated HPV band from the cervical cancer as well as the shorter episomal HPV band from the tonsillar cancer could be obtained at the same time (data not shown).

HPV quantification by real-time PCR

The amount of HPV-16 copies in each tumour was estimated and related to the number of human equivalent β-actin genes. All samples, standards and controls were run in duplicate and the mean value was calculated. The viral load for the tonsillar cancers varied between 10–15,500 HPV-16 copies/β-actin (Table I). As additional controls 2 cervical cancer biopsies (F3155, F725) and 1 cervical cancer cell line, CaSki, were analysed as well. Approximately 1 HPV-16 copy/β-actin was observed in F3155, 10 HPV-16 copies/β-actin were observed in F725 and CaSki had about 450 HPV-16 copies/β-actin. Two HPV-16 negative samples (F5 and F10) were also included in the run and as expected no HPV-16 could be detected in these tumours (Table I).

The median value of viral load for the HPV positive tonsillar cancer was 190 viral copies per β actin copy and the HPV positive tumours could be divided into 2 groups, 1 with values below and 1 with values above the median value. One group consequently contained tumours with 10–60 HPV-16 copies/β-actin, i.e., F7, F11, F12, F14 and F18, and the other group contained tumours with ≥190 HPV copies/β-actin, of which F2, F20, F24 and F25 had between 190-350 HPV copies/β-actin and samples F4 and F13 had >700 HPV copies/β-actin (Table I).

HPV in tonsillar cancer and correlation to relapse and survival

Twenty of the 22 patients with tonsillar cancer were included in disease free and survival analysis, 2 patients were omitted because they refused treatment. Three years after diagnosis, patients with HPV positive tonsillar cancer were more frequently disease free as compared to patients with HPV negative tonsillar cancers (data not shown). Furthermore, patients with HPV positive tumours had a better disease specific survival (only cause of death due to cancer progress was included), when compared to patients with HPV negative tonsillar cancer (data not shown). Neither the differences in disease free patients, however, nor the better survival was statistically significant (p = 0.09, χ2 test; p = 0.08, log rank test respectively). Nevertheless, the 6 patients with tumours with equal or above the median value of 190 HPV copies/β-actin were all tumour free 3 years after diagnosis as compared to 2/5 tumour free patients among the group that had tumours with ≤60 HPV copies/β-actin (p = 0.026, χ2 test) (Fig. 3a). Accordingly, patients with tumours with ≥190 HPV copies/β-actin had also significantly better survival as compared to patients with tumours with ≤60 HPV copies/β-actin (p = 0.039, log rank test) (Fig. 3b). With so few patients only descriptive statistics is possible to evaluate if age, local lymph node metastasis, tumour stage, differentiation grade and gender influence these results. Nonetheless, except for tumour stage, the data were either in disadvantage for the group that had better prognosis or were equally distributed (Table I). As for tumour stage all patients in both groups had advanced tumours (Stage III and IV). The higher viral load group had a few more patients with Stage III tumours compared to Stage IV tumours (3/6 and 3/6) than the group with lower viral load (1/5 and 4/5) (Table I). Nevertheless, the group with the higher viral load had a few more patients (5/6) with nodal metastasis than the group with the lower viral load had (4/5) (Table I). The physical state of HPV could not be correlated to clinical outcome because all HPV-16 tonsillar cancers contained extrachromosomal HPV.

Figure 3.

(a) Significantly higher number of disease-free patients 3 years after diagnosis in patients with tumours with ≥190 HPV-16 copies/β-actin copy when compare to patients with tumours with ≤60 HPV-16 copies/β-actin in tonsillar cancer (p = 0.026, χ2 test) (n = 11) (b) Kaplan-Meier graph showing significantly better disease specific survival in patients with tumours with ≥190 HPV-16 copies/β-actin when compared to patients with tumours with ≤60 HPV-16 copies/β-actin in tonsillar cancer (p = 0.039, log rank test, n = 11).

DISCUSSION

Twenty-two fresh-frozen tonsillar cancer samples were analysed for the presence of HPV by PCR. The HPV-16 positive samples were tested for the physical state of the virus by the rliPCR technique and for the viral load by a quantitative real-time PCR. The data were then correlated to clinical outcome.

Presence of HPV-16 by PCR was observed in 11/22, whereas HPV-33 was found in 1/22 of the tonsillar cancers with the use of GP5/GP6 primers, which is in line with previous publications.2, 3, 4, 6, 7, 8, 9 The use of CpI/IIG with primers located in the E1 region to detect possibly additional HPV should the L1 region have been deleted in these copies did not result in the finding of any additional presence of HPV. Only extrachromosomal forms of HPV-16 were detected in tonsillar cancer (Table I, Fig. 2). In 7/11 cancers full-length episomal HPV was detected exclusively, and in 2 other tumours (F14, F25) both full-length and deleted forms of episomal HPV-16 were found in parallel, whereas in 1 tumour (F7) only a deleted form of episomal HPV-16 was present. In the remaining HPV-16 positive tumour (F12) both a full-length episomal as well as an 11 kbp was PCR product were detected and it was not possible to determine if the 11 kbp product contained integrated HPV with human–viral junctions, or was off-size linearized episomal HPV. The fact that integrated HPV-16 was not detected in tonsillar cancer, was not due to methodological insensitivity. Traditionally the demonstration of the physical state of HPV is shown by Southern blot and 2D gel electrophoreses, which require a fairly large amount of DNA. In situ hybridisation is also quite commonly used, and can be a supportive method but it is relatively insensitive. In our study, we chose to use the sensitivity of a PCR based method, described previously by Kalantari et al.21 With this method we could detect HPV-16 in SiHa, known to harbour only 1–2 copies of integrated HPV per cell, as also described previously.21 Furthermore, 2 cervical cancer control biopsies (F3155 and F826) were both demonstrated to contain integrated HPV-16 and in addition, the human sequences flanking HPV-16 in these tumours were isolated. F3155 was chromosome-located at 3q28 as published previously by Kalantari et al..21 F826 was chromosome-located at 1q24 in our study and not located at 1q25 as in the earlier study by Kalantari et al..21 We believe that this minor discrepancy is because we used the GenBank for human genome BLAST as a chromosome location method and not FISH (fluorescent in situ hybridisation) as was done by Kalantari et al.21 Finally, we could also show that integrated HPV could still be detected when mixed together with 600× more of episomal HPV (data not shown).

Our results are thus similar to the findings by Snijders et al.11 where the physical state of HPV-16 in 2 tonsillar cancer biopsies was investigated by Southern blot and 2-dimensional gel electrophoresis analysis, and where both tumours contained only episomal forms of HPV. In the same study, however, the 2 HPV-33 positive tumours that were analysed contained integrated and mixed integrated and episomal HPV. In that study all biopsies displayed E6/E7 transcription.

In cervical cancer approximately 15–30% of all HPV-16 positive tumours have been found to carry HPV in its episomal form, which suggests that HPV-16 can transform even without integration into the host genome.16, 20, 21, 30 Why the presence of episomal HPV-16 is more frequent in tonsillar cancer is presently unknown. Nonetheless, one possible mechanism to how extrachromosomal HPV can dysregulate its oncogenes is by genetic modifications in the LCR, which may in turn influence the promoter activity for E6/E7 transcription. Dong et al.22 investigated the LCR of episomal HPV-16 DNA in primary cervical cancer biopsies and in lymph-node metastases from 6 patients. The analysis revealed mutations of the cellular transcription factor YY1 binding sites in 4 cases, and in 3 of the cases an increased activity of the viral oncogene promoter could be shown.22 Another study of episomal HPV-16 in cervical cancer biopsies reported sequence variations in the YY1 binding sites as well as in other sites in the LCR and some of these modifications led to increased E6/E7 promoter activity.24 These findings imply that deletions or mutations of YY1-binding or at other LCR sites may play a significant role in the over-expression of oncogenes in episomal HPV-16. Notably, the deleted forms of episomal HPV in our study were all disrupted at the end of E1 ORF, and the E2 ORF was missing. The deleted forms of F7 and F14 continued in the LCR (at 7,843 and 7,600 respectively) and in F25 in the E6 ORF. The biological implications of these E2 and LCR deletions are uncertain and purely speculative. The deletion of F7 in the LCR, however, included a deletion of the NF1 binding sites as well as a deletion of 2 and part of the fourth YY1 binding sites. Future mutation analysis of the LCR of HPV-16 in full-length episomal HPV in tonsillar cancer may reveal viral gene modifications of interest.

The copy number of HPV (HPV-16 per β-actin) in tonsillar cancer displayed a wide distribution. Most of the tonsillar cancers contained between 10 to a few hundred HPV copies per β-actin, which is quantitatively similar to the study of Klussman et al.8 In In the study of Klussman et al 6 tonsillar cancers and their metastasis were analysed, and the viral copy number per β-goblin varied between 5.8–152.6. The HPV positive tumours in our study were divided into 2 groups depending on if their viral load was below or above the median value of 190 copies per β-actin i.e., between 10–60 copies per β-actin and ≥190 copies per β-actin. One of our tumours displayed an extremely high value of 15,000 HPV copies per human equivalent, and it is possible that there is virus replication in this tumour. The cervical cancer biopsy F3155, known to harbour integrated HPV, displayed 1 HPV copy/β-actin, and the cervical cancer sample F725 known to harbour episomal HPV-16 contained 10 HPV copies/β-actin.21 The HPV-16 assay, however, and the β-actin real time assay were not run simultaneously in the same tube and there could be differences in the sensitivity of the 2 assays. Furthermore, we did not microdissect our material although all biopsies were checked for a content of >80% of cancer cells. Consequently the number of virus copies per human gene copy should be interpreted with some caution.

Similar to our previous study, a higher number of disease free patients 3 years after diagnosis as well as a better survival were observed among patients with HPV positive tonsillar cancer as compared to patients with HPV negative tonsillar cancers.7 In our study, however, these differences were not statistically significant, most likely due to the limited number of patients. We could not correlate the better prognosis in the surviving patients to physical status, because HPV was episomal in all HPV-16 positive tumours. Notably, the 6 patients with tumours with ≥190 HPV copies/β-actin showed a significantly better clinical outcome, both regarding remaining tumour free 3 years after diagnosis (p = 0.026) as well as better survival rate (p = 0.039) as compared to the 5 patients with tumours with ≤60 HPV copies/β-actin (Fig. 3a,b). Age, local lymph node metastasis, differentiation grade and gender do not influence these results, because the data were either in disadvantage for the group that had better prognosis or were equally distributed (Table I). As for tumour stage all patients in both groups had advanced tumours. The group with the higher viral load had a few more patients with Stage III compared to Stage IV tumours than the group with the lower viral load, but on the other hand the former group had more patients with nodal metastasis than the latter group (Table I). It is possible that the better prognosis for patients with a high viral load in tonsillar cancer is because a better immune response against the cancer can be induced. This result, however, should be confirmed in extended studies because few patients were involved (n = 11).

In conclusion, HPV-16 is episomal and the viral load shows a wide range. Furthermore, our results suggest that a high viral load could be a favourable prognostic factor. Mutation analysis of the LCR with subsequent estimation of E6/E7 expression, and investigations of the transforming ability of episomal HPV-16 isolated from tonsillar cancer, may help clarify some of the tumour molecular biology of HPV in tonsillar cancer.

Acknowledgements

The authors thank Dr. E. Ogris (Vienna, Austria) for helpful discussions and methodological and material contribution. We also thank Prof. E. Wintersberger (Vienna, Austria) for letting H. Mellin work in his department. We are grateful to Prof. H. zur Hausen, Dr. E-M de Villier (Heidelberg, Germany) and Dr. G. Orth (Paris, France) for the cloned HPV-16 and HPV-33.

Ancillary