J. Peter Klussmann's current address is: Department of Otorhinolaryngology and Head and Neck Surgery, University of Giessen, Germany
Tonsillar squamous cell carcinoma (TSCC) is frequently associated with human papillomavirus (HPV) and chromosome instability. Data from cellular model systems are, however, controversial concerning a relation between HPV and chromosome instability development. Here we studied this association in 77 primary TSCC with known clinical outcome and cell cycle protein expression profiles. Thirty-two tumors (42%) showed HPV16-integration. All 77 cases were analyzed by fluorescence in situ hybridization using chromosome 1- and 7-specific centromere DNA probes to detect chromosome instability, indicated by the presence of chromosome imbalances and/or polyploidization for these chromosomes. In addition, eight HPV-positive dysplasias, seven of which were adjacent to a carcinoma, were analyzed. Disomy for chromosome 1 and 7 was present in 29 out of 77 TSCC (38%), of which 19 were HPV16-positive (p = 0.002). Aneusomy was observed in the remaining 48 TSCC, of which 13 were HPV-positive. Aneusomies correlated significantly with tobacco- and alcohol consumption (p = 0.001 and p = 0.016, respectively) and a higher T-stage (p = 0.018). Both HPV-positivity and chromosome disomy were significantly associated with a favorable disease-free survival (p = 0.001 and p = 0.025, respectively). Particularly in the HPV16-positive group chromosome instability is a very strong indicator for an unfavorable prognosis (p = 0.032). In the dysplasias an identical HPV and chromosome copy number status was identified as in the adjacent tumors. We conclude that HPV-positive TSCC and their precursor lesions are more often genetically stable than HPV-negative lesions and that these tumors are associated with a favorable prognosis. Chromosome instability is an indicator for unfavorable prognosis, particularly in the HPV-positive patient group.
Chromosome instability or aneuploidy is a characteristic of most solid tumors, including head and neck squamous cell carcinoma (HNSCC). Chromosome instability refers to changes in chromosome number (gains or losses), amongst others caused by failures in mitotic chromosome transmission and disturbances in the polarity of cell division.1 Previous flow cytometric studies have shown that chromosome instability is often present in primary HNSCC and their metastases, with an incidence ranging from 44 to 90%.2,3 Similar frequencies were found using other technologies detecting chromosome instability, such as microsatellite analysis, comparative genomic hybridization (CGH), fluorescence in situ hybridization (FISH), or using chromosome instability-associated p53-immunostaining.4–6 We and others have shown that alterations in chromosome 1 and 7 copy numbers detected by FISH are frequently found in head and neck (pre)malignancies.7,8 On the basis of imbalances in chromosome 1 and 7 copy number ratios, we were able to reliably predict malignant progression and tumor recurrence.8–10 In these cases, the chromosome 1 copy number most often equalled the average copy number of all chromosomes (ploidy) in the lesions, whereas generally higher copy numbers for chromosome 7 were detected. The latter finding is also reflected by data of Benchekroun et al. reporting increased epidermal growth factor receptor (EGFR) gene copy numbers on chromosome 7 in premalignant oral leukoplakia that progressed to carcinoma.11
Mutations in, e.g., the p16 and p53 genes in early head and neck carcinogenesis are associated with well-known risk factors, such as smoking and alcohol consumption, whereas wildtype p16 and p53 are often found in tumors associated with oncogenic human papillomavirus (HPV). The deregulation of the p53 and pRb pathways in both HPV related and -unrelated carcinogenic routes will ultimately result in a deregulation of the cell cycle and of the apoptotic process, finally leading to chromosome instability.12,13
The occurrence of chromosome instability is considered to be a late event in HPV related carcinogenesis. In vitro studies with HPV-E6/E7 immortalized keratinocytes and raft cultures have shown that a persistent HPV infection is not sufficient for malignant transformation.14–16 This transformation can be achieved by treating them for example with tobacco mutagens.17 However, patients with HPV-positive carcinomas are significantly less likely found in smokers and drinkers.18–20 Alternatively, several in vitro transcription studies have shown chromosome and centrosome alterations to occur as a result of high levels of the HPV-oncoproteins E6 and E7,21–23 although the role of the individual proteins in this process remains controversial.24,25
We and others have shown that in uterine cervix cancer oncogenic HPV integration is associated with the transition from cervical intraepithelial neoplasia to invasive carcinoma.26,27 Viral integration is often accompanied by an upregulation of the oncoproteins E6 and E7, and subsequently chromosome instability.26,28 Nevertheless, viral integration has been detected in both diploid and aneuploid cervical cancer lesions.27 Upregulation of E6 and E7 due to viral integration is also suggested to occur during the development of tonsillar squamous cell carcinoma (TSCC), in which there is a high prevalence of oncogenic HPV (30–75%).18,20,29–32 The question therefore is if HPV integration will also lead to chromosome instability in primary HPV-positive TSCC, and whether or not chromosome instability can be detected in the same frequency as in HPV-negative tumors?
The aim of the present study was to analyze a series of 77 TSCC for chromosome instability by means of FISH using chromosome 1 and 7-specific centromere probes. Chromosome copy number-data were correlated with the HPV16 status, as well as clinical and lifestyle parameters.31,33 Furthermore, we were able to select eight additional HPV-positive tonsillar dysplasias, seven of which were found adjacent to a TSCC, for comparison of chromosomal (in)stability and the physical status of HPV.
CGH: comparative genomic hybridization; CI: confidence interval; FISH: fluorescence in situ hybridization; HNSCC: head and neck squamous cell carcinoma; HPV: human papillomavirus; HR: hazard ratio; SSC: saline sodium citrate; TSCC: tonsillar squamous cell carcinoma
Material and Methods
Tumor material and patient data
We analyzed formaldehyde fixed, paraffin embedded archival biopsy and resection material of primary TSCC from 77 patients, selected from the archives of the department of Pathology, Maastricht University Medical Center, The Netherlands and the department of Pathology, University of Cologne, Germany.
Patients were diagnosed between 1992 and 2001. Sections had been previously examined for HPV16 and -18 association by type specific FISH and PCR and cell cycle protein expression profiles.19,33 Patient material was used according to the code for proper secondary use of human tissue (Federation of Medical Scientific Societies, The Netherlands; 2003). Demographic data, tumor site, degree of differentiation, TNM classification and clinical follow-up data of all patients were available.19 Patients have been treated by surgery, radiotherapy, chemotherapy or a combination, irrespective of HPV status. Fifty-five patients were male and 22 patients were female. The mean age at diagnosis was 60 (range 39–87) years. Smoking and alcohol-intake data were available for 77 and 76 patients, respectively. Sixty-one (79%) of the 77 patients were smokers (≥1 cigarette, pipe, and/or cigar per day), 42 (55%) of 76 patients were classified as drinkers (consumption of > 2 whiskey equivalents per day (1 whiskey equivalent ∼10 g alcohol)) and 38 (49%) patients used both. Thirty-eight (49%) patients had a tumor with a diameter ≥4 cm, and 57 (74%) had lymph node metastasis at time of diagnosis. The male/female ratio and the age at first diagnosis were similar for the HPV-positive and HPV-negative group. Smoking and drinking was significantly correlated with an HPV-negative tumor (p < 0.0001 and p = 0.002, respectively).
In addition, we analyzed one dysplasia and seven tumors and adjacent tissue that had the appearance of - and were classified as dysplasia. Also the tumor areas were analyzed for comparison. This tissue material was obtained from the archives of the Department of Pathology, University of Cologne, Germany.
A series of 4 -μm-thick sections was cut from the specimens for hematoxylin-eosin staining and detailed histopathological classification, according to the criteria of the World Health Organization.34 On consecutive sections HPV-status was determined by HPV-specific PCR followed by FISH analysis. Double-target FISH analysis was applied using chromosome 1- and 7-specific centromere DNA probes.
Chromosome 1 and 7 copy number analysis
FISH was performed on 4 -μm-thick tissue sections as described previously.9,35 Briefly, sections were deparaffinized, pretreated with 85% formic acid/0.3% H2O2, 1 M NaSCN and 4 mg/ml pepsin, postfixed in 1% formaldehyde in PBS (pH 7.4), dehydrated in an ethanol series and hybridized with a mixture of digoxigenin-labeled human centromere 1-specific and biotin-labeled centromere 7-specific DNA probes (1 ng/μl 60% formamide, 2 × saline sodium citrate (SSC), 10% dextran sulfate and 50 × excess salmon sperm carrier DNA). After hybridization the preparations were washed stringently in 2 × SSC at 42°C (two times 5 min) and 0.1 × SSC at 60°C (two times 5 min).
Both probes were detected simultaneously, using (i) mouse anti-digoxin (Sigma, St. Louis, MO), (ii) rabbit anti-mouse rhodamine (DAKO A/S, Glostrup, Denmark) and (iii) swine anti-rabbit rhodamine (DAKO A/S) for the digoxigenin-labeled probe and (i) avidin fluorescein (Vector Laboratories, Burlingame, CA), (ii) biotinylated goat anti-avidin (Vector) and (iii) avidin fluorescein for the biotin-labeled probe. Preparations were mounted in Vectashield (Vector Laboratories, Burlingame, CA) containing 4,6-diamidino-2-phenyl indole (DAPI; Sigma: 0.2 g/ml).
Slides were evaluated under the microscope and images were recorded with the Metasystems Image Pro System (black and white CCD camera; Sandhausen, Germany) mounted on top of a Leica DM-RE fluorescence microscope equipped with DAPI, fluorescein and rhodamine filters.
Evaluation of the FISH results was carried out according to the criteria previously described.10,35 FISH signals were scored per color and nucleus for the presence of aberrant copy numbers. The highest copy number per nucleus was determined and set if ≥20% of the nuclei (50–200 nuclei are counted, depending on the size of the lesion) showed this number of FISH signals. Based on this evaluation, areas were categorized as either monosomic, disomic, trisomic, tetrasomic or polysomic (>4 signals per nucleus) for the respective probe targets. Next, the lesions were divided into a group of lesions without evidence of chromosome alterations (disomic), and a group of lesions showing either altered but balanced chromosome copy numbers (trisomy or tetrasomy) or characteristics of chromosome instability, i.e., containing chromosome copy number imbalances and/or chromosome polysomy (indicative for aneuploidy).9
HPV FISH was performed on 4µm-thick tissue sections as described previously.31 Briefly, sections were pretreated as described above and hybridized with a digoxigenin-labeled HPV 16-specific probe (PanPath, Budel, The Netherlands) according to the manufacturer's instructions. After stringent washing in 50% formamide, 2 × SSC, pH 7.0 at 42°C (two times 5 min), probe hybridization was visualized by subsequent incubations with mouse anti-digoxin (Sigma), peroxidase-conjugated rabbit anti-mouse IgG, peroxidase-conjugated swine anti-rabbit IgG (both DAKO A/S) and finally rhodamine-labeled tyramide. Preparations were mounted, evaluated and recorded as described above.
Evaluation of nuclear hybridization signals was performed by two investigators (JM and EJMS) according to previously described criteria31: nuclear punctate signals were considered to indicate integrated HPV DNA and diffuse signals to indicate episomal HPV DNA. Also a granular FISH pattern, defined as the presence of several nuclear signals varying significantly in size and intensity, was observed in a few cases, which may indicate a combination of viral integration and episomal DNA and/or viral RNA.31
Previously assessed clinical and molecular parameters were correlated with chromosome instability and HPV-status using cross-tabulations and the two-tailed Fisher exact test and/or Chi-square test.33 We regarded a p value ≤ 0.05 as level of significance. The Kaplan–Meier method was used to calculate disease-specific survival curves. Survival was calculated from the date of diagnosis until patient's death or until the last date the patient was known to be alive (this ranged from 16 to 141 months). Patients that died of other causes than tonsillar carcinoma were considered censored observations in the disease-specific survival analyses. Disease-free survival was calculated from the date of diagnosis until the date of recurrence (local, regional or distant, whichever occurred first). Patients without recurrence were censored at the date of the last follow-up or the date of death. The statistical significance of differences between survival times was determined by the log rank test in univariate analysis. Multivariate analyses were performed using the Cox proportional hazards model. Variables remained in the model if their p values were below 0.10. We used the SPSS Base System version 18.0 for all calculations.
HPV16-positivity in TSCC is correlated with chromosome stability
Figure 1 shows the distribution of the HPV and chromosome copy number status amongst the 77 TSCC examined in this study. We have analyzed chromosome 1 and 7 alterations and HPV16 integration status using FISH, and the expression of p16INK4A using immunohistochemistry. We found 32 HPV16-positive cases, as indicated by overexpression of p16INK4, and a punctate HPV16 signal in 31 cases and a granular pattern in 1 case (Fig. 1). Predominantly single HPV16 integration sites per tumor cell nucleus were found, except for six cases that exhibited two integration sites per nucleus by FISH. Two of these were found in an aneusomic tumor. The other 4 were observed in diploid TSCC, which also exhibited tumor areas containing cells with only 1 HPV16 integration site by FISH. These TSCC might have undergone tetraploidization with duplication of the viral integration site, despite the fact that we could not find the tetraploidization of chromosomes 1 and 7 in these cases.
Disomy for chromosome 1 and 7 was found in 29 out of 77 TSCC (38%). Nineteen of these were HPV16-positive (p = 0.002), including 11 nonsmokers (of 16 nonsmoking patients in total). Figures 2a–2c show a typical example of an HPV16-positive tumor in which a strong positivity for the HPV surrogate marker p16INK4A (Fig. 2a), a punctate FISH HPV pattern indicating integration (Fig. 2b) and disomy for chromosome 1 and 7 (Fig. 2c) can be observed.
Aneusomies for both chromosomes were observed in 48 out of 77 TSCC (62%), of which 28 tumors showed balanced and 20 showed unbalanced copy numbers (19 cases with a higher chromosome 7 copy number). Aneusomies correlated significantly with HPV-negative tumors (p = 0.002), (Figs. 1 and 2d and 2e), tobacco- and alcohol consumption (p = 0.001 and p = 0.016, respectively) and a higher T-stage (p = 0.018). Strong p53 immunostaining was also associated with HPV-negative tumors (p = 0.003) and smoking and/or alcohol abuse (p = 0.005), but not with aneusomy and T-stage.33
HPV16-positive TSCC and adjacent dysplasias show similar HPV patterns and chromosome copy numbers
We were able to compare the occurrence of chromosome instability between HPV-positive tumors and their adjacent dysplastic lesions in seven cases. In both tumors and dysplasias strong nuclear and cytoplasmatic p16INK4A immunostaining (Figs. 2f–2i) was observed and found to be associated with HPV16-positivity as determined by FISH (Table 1). A punctate HPV pattern was predominantly detected, indicating viral integration in both the dysplastic and malignant lesions. In addition, similar chromosome copy numbers were identified in the tumors and the adjacent dysplastic areas, including cases with disomy, tetrasomy and aneusomy. Interestingly, two cases showed both disomic and aneusomic cell clusters adjacent to each other already in the dysplastic lesion. Case number 6 (Table 1) showed different HPV patterns (punctate and granular) in relation to disomic and aneusomic cells, respectively, and only the disomic population in the TSCC showed a punctate HPV pattern. In contrast, case number 7 (Table 1) showed cell populations with a punctate HPV pattern and disomy or aneusomy in both the dysplasia and the carcinoma. In the dysplasia that was not associated with a TSCC (Table 1, Case 1) cells exhibited p16INK4A immunopositivity, a punctate HPV16 pattern and disomy.
Table 1. p16INK4A-, HPV16- and chromosome 1 and 7 copy number status in dysplastic lesions and their adjacent TSCC; furthermore, the smoking behaviour of the eight patients is indicated
HPV16-positivity and chromosome stability are associated with a favorable prognosis
We correlated HPV16 and chromosomal copy number status with the disease-specific survival (DFS) data of patients with TSCC to determine the role of these two parameters as indicators of prognosis. Follow-up time ranged from 0 to 141 months, with a mean of 33 months. Thirty-nine (48%) out of 75 patients died as a consequence of TSCC. The DFS after 5 years was 30% for patients with an HPV-negative carcinoma and 74% for patients with an HPV-positive carcinoma (Hazard ratio (HR) = 0.3; 95% Confidence Interval (CI) = 0.1–0.6) (Fig. 3a). Also, chromosome disomy was significantly associated with a favorable outcome in the entire group of TSCC (HR = 0.4; 95% CI = 0.2–0.9) (Fig. 3b). Interestingly, in the patient group with HPV-positive tumors chromosome disomy was an even stronger indicator of favorable prognosis (HR = 0.2; CI = 0.03–0.86) (Fig. 3c) as compared to the whole patient group. No prognostic association was observed for chromosome instability in the HPV-negative TSCC (Fig. 3d). The HPV-positive aneusomic tumors (Fig. 3c) did not show significantly more often p53 overexpression.
TSCC can be divided into at least two etiological subgroups, i.e., smoking/alcohol induced and HPV induced tumors. Despite the different initiating factors and their different biological behavior,12,36 both subgroups have been reported to develop chromosome instability as a late carcinogenic event, although some studies suggest that less genetic mutations occur in HPV-positive tumors.37,38 Data on the mechanisms of how HPV induces chromosome instability are particularly based on cell model experiments, in which the viral oncoproteins E6 and E7 are overexpressed.21–23 In the present study we analyzed the relation between the presence of HPV and the occurrence of chromosome instability in 77 primary TSCC, of which 32 (42%) were HPV16-positive. We used FISH analysis to determine chromosome 1 and 7 copy number imbalances and/or polysomy, which is often associated with a more general chromosome instability.9,10 Our data show that chromosome stability was significantly associated with HPV-positivity. In contrast, chromosome instability was shown to correlate strongly with tobacco- and alcohol consumption, a higher T-stage and an unfavorable prognosis. Thus, primary HPV-positive TSCC predominantly exhibits a stable genome, which is a favorable prognosticator.
Several in vitro cell model studies using human keratinocytes and fibroblasts transfected with HPV16 E6, E7 or both, have reported that high expression levels of these viral oncoproteins can render these cells genomically unstable, thus predisposing them to accumulate chromosomal alterations.21–23 In this respect HPV16 E6 expressing cells exhibited evidence for structural chromosomal changes only after prolonged culturing, whereas HPV16 E7 expressing cells rapidly induced centrosome-related mitotic defects and chromosome instability. In the present study, however, we predominantly found chromosome stability in primary HPV-positive TSCC, which might be the result of lower expression levels of E6 and E7 than generated in the cell model studies. Indeed, a study by Lace et al.,15 examining HPV persistence induced by HPV16, 18, 31 and 11 episomes in primary human keratinocytes from the foreskin, cervix and tonsils revealed that cell growth could be promoted by steady state E6 and E7 expression without the induction of chromosome instability. This raises the question which factor(s) besides the presence of HPV is (are) required for malignant progression in those cases where chromosome instability is not involved? Tobacco mutagens are not the only factor, because in this subgroup the highest frequency of nonsmokers is found. As stated above, chromosome instability was particularly observed in HPV-negative TSCC from heavy smokers. More likely factors for malignant progression in the HPV-positive group include the accumulation of confined chromosomal alterations, such as extra copies of chromosome 3q found in HPV-positive uterine cervical cancers and TSCC,39–41 deletion of 16q,40,41 and/or modification of the methylation status of the host cell DNA.42,43 In addition, molecular alterations underlying lymphangiogenesis and epithelial mesenchymal transition might contribute to malignant tumor progression in these cases.44,45
Because most HPV-positive TSCCs are genomically stable, it could be expected that also their associated precursor lesions show this feature. Studies on premalignant tonsillar lesions are scarce, because patients usually enter the clinic with invasive tumors. Nevertheless, we were able to examine seven cases of HPV-positive TSCC with tumor-adjacent dysplasia found in the resection margins, and detected indeed in four cases disomy for chromosomes 1 and 7 in both the tumor and dysplastic lesion. This finding is in accordance with data on uterine cervical carcinogenesis,27 showing that HPV-integration can take place in diploid precursor lesions. In addition, we identified tetrasomy in two dysplastic lesions and their adjacent malignancies, and aneusomy for chromosomes 1 and 7 in three dysplastic lesions and two TSCC along with HPV integration, suggesting that HPV can also integrate into an unstable cellular genome. This has also been observed in uterine cervical preneoplasia,26,46 and might be the result of persistent infections and high viral loads in the (para)basal cells of the squamous epithelium, leading to chromosome instability before integration occurs. Another explanation might be that preneoplasia with chromosome instability is the result of DNA damage induced by for example heavy smoking, such as observed in case 8 (Table 1), although a larger series of dysplasias needs to be studied to underscore this hypothesis.
Also in our series of 77 TSCC, 13 out of 32 HPV-positive cases showed chromosome instability and these patients exhibited a significantly worse prognosis than those with chromosome stable tumors, although they were treated with comparable therapeutic regimes. Strikingly, 10 of these 13 tumors were diagnosed in smokers. In the literature smoking has been associated with an unfavorable prognosis in HPV-positive tumors as well.31,47
Because also other prognostic parameters have been reported, including T and N status31,48 and EGFR expression,47,49 the construction of multiparameter models, based on large patient series, will be essential to optimally classify patients with a low, intermediate or high risk of cancer specific death.48,50
For HPV-positive HNSCC, Weinberger et al.13 proposed that HPV-positive and chromosome instability-containing tumors in smokers might be initially smoking-induced, after which HPV infection leads to its integration into the cellular genome (“hitchhiker model”). The typical characteristics of a smoking induced HNSCC comprise p53 mutations and downregulation of p16INK4A. However, we and others have shown that HPV-positive tumors, also in smokers, predominantly express high p16INK4A levels and exhibit wild type p53.19,51 Therefore it is more likely that HPV integration occurs as an initial step in tetraploid or aneuploidy containing epithelial cells, most probably induced by high viral loads and not by p53 mutations as a result of smoking.
On the basis of our results, we propose a hypothetical model for HPV16 integration in relation to chromosome (in)stability during TSCC tumorigenesis (Fig. 4), in which the latter is a key feature of survival in this patient group. In nonsmoking or sporadically smoking patients, TSCC predominantly show single HPV integration signals in a chromosome stable background. These TSCC most likely develop from diploid premalignant lesions. Some diploid TSCC may undergo duplication of the viral integration site or a second integration event.
In smoking patients, more often single HPV16 integration signals in a chromosome unstable background are detected. These TSCC most likely develop from tetraploid or aneuploid premalignant lesions, induced by high HPV load and/or smoking-related toxins. During tumor progression these carcinomas undergo further genomic multiplication including the viral integration site. In conclusion, our data show that chromosome stability is more often found in HPV-positive TSCC than in HPV-negative tumors, and that chromosome instability is an indicator for unfavorable prognosis, particularly in the HPV-positive patient group.
Therefore, we recommend that ploidy-status should be taken into consideration as an additional prognostic factor in future risk models for cancer specific death, particularly in HPV-positive TSCC.