Infectious Causes of Cancer
p16INK4a overexpression predicts translational active human papillomavirus infection in tonsillar cancer
Version of Record online: 20 JAN 2010
Copyright © 2010 UICC
International Journal of Cancer
Volume 127, Issue 7, pages 1595–1602, 1 October 2010
How to Cite
Hoffmann, M., Ihloff, A. S., Görögh, T., Weise, J. B., Fazel, A., Krams, M., Rittgen, W., Schwarz, E. and Kahn, T. (2010), p16INK4a overexpression predicts translational active human papillomavirus infection in tonsillar cancer. Int. J. Cancer, 127: 1595–1602. doi: 10.1002/ijc.25174
- Issue online: 4 AUG 2010
- Version of Record online: 20 JAN 2010
- Manuscript Accepted: 8 JAN 2010
- Manuscript Received: 7 OCT 2009
- human papillomavirus;
- head and neck squamous cell carcinoma;
The causal role of human papillomaviruses (HPV) in squamous cell carcinogenesis of tonsillar cancers (TSCC) depends on the activity of the viral oncoproteins E6 and E7, leading to inactivation of the cellular tumor suppressor p53 and the retinoblastoma gene product pRb. Because of the negative feedback mechanisms, the pRb inactivation causes an increase of the inhibitor of the cyclin-dependent kinases p16INK4a. In 39 TSCC specimens, genotyping based on the amplification of HPV DNA was carried out using PCR by applying HPV type-specific oligonucleotides. Subsequently, amplicons were hybridised with fluorescence-labeled complementary probes using the Southern blot technology. For HPV E6/E7 mRNA expression, Northern hybridization and RT-PCR were performed, and for p16INK4a detection, immunohistochemistry was performed. With 21/39 (53%) HPV-positives, the detection rate is within the range that can be expected in TSCC. The E6/E7 oncogene mRNA was detectable in 11 cases, 10 of which showed positive signals after p16INK4a staining. Albeit the small study group was investigated, the correlation of the HPV DNA status with the p16INK4a expression was of statistical significance (p = 0.02). Kaplan-Meier estimations revealed better survival outcome for patients with HPV-positive tumors with detectable E6/E7 mRNA and p16INK4a overexpression (p = 0.02, median observation time 29 months). As mRNA expression tests are not routinely available in many clinical diagnostic laboratories, and based on the high correlation of p16INK4a staining with HPV E6/E7 mRNA expression, in conclusion we suggest for a deeper exploration for the use of p16INK4a as a surrogate marker with the potential to impact the standard of care of HPV DNA-positive head and neck carcinomas.
Malignant tumors of the upper aerodigestive tract account for ∼7% of all cancers worldwide,1 with squamous cell carcinomas (SCCs) constituting the largest portion among the different tumor entities.2 Even though a majority of SCCs of the head and neck (HNSCC) seems associated to etiologic factors as carcinogenetic substances in tobacco smoke and alcohol,3 an increase of especially younger patients without reported misuse of these environmental factors in history has been recognized.4 Therefore, an increase in knowledge of other factors such as the infection with human papillomaviruses (HPV) promoting carcinogenesis in especially the latter patient group gains in importance.
Epidemiologic data demonstrate HPV DNA prevalences of about 20–30% concerning all anatomical tumor sites of the head and neck region,4–6 with the SCC of the Waldeyer's tonsillar (squamous cell carcinogenesis of tonsillar cancers, TSCC) ring showing an HPV DNA-positive rate of up to 60% as particularly predestined for an infection with HPV.4, 5, 7 Recent data from Sweden describe HPV DNA infections in tonsillar carcinomas in even up to 85% of the investigated patients.8
HPV DNA-positive HNSCCs, again with accentuation of those derived from the Waldeyer's tonsillar ring, are considered as being a specific tumor entity, specifically in terms of molecular biology, carcinogenesis and biological behavior.4, 9 Albeit initially presenting with an apparently more aggressive biological behavior of the tumor by means of an early metastatic spread into the locoregional lymph nodes, favorable if not, better recurrence free and overall survival times have been reported for patients with HPV DNA-positive HNSCC.10–13 The better course of disease is discussed to be assigned to a specifically good responsiveness of the SCC to the applied therapy particularly with regard to the radiotherapy in case of present HPV infection.11
Interactions of the high-risk HPV oncogene products E6 and E7 with two cardinal cellular regulators of the cell cycle, the tumor suppressor protein 53 (p53) and the retinoblastoma gene product (pRb), respectively, are of prime importance for HPV-associated HNSC carcinogenesis.14, 15 The E6 protein binds wild type (wt)-p53, triggering its ubiquitin-mediated degradation or directly inactivates wt-p53 by complex formation. The induced malfunction of wt-p53 causes the accumulation of DNA mutations, thus increasing the genetic instability of the cells, an instance likewise described for mutated p53.15, 16 The E7 protein inactivates pRb, which is in control of the crucial G1-S-phase transition by binding the transcription factor E2F when pRb is unphosphorylated. Both, pRb phosphorylated by cyclin-dependent kinases and pRb bound by E7 release the E2F transcription factor, subsequently leading to progression of the cell into the S-phase, facilitating synthesis of viral DNA.17 Additionally, E7 binds to inhibitors of cyclin-dependent kinases (p16, p21), increasing the level of phosphorylated pRb.18 In this way, HPV oncoproteins induce the failure of cell cycle regulation with lack of p53 mutations, otherwise a common feature in HNSCC and many other human malignancies. This malfunction of wt-p53 as well as reduced pRb and consecutive p16 (INK4a) overexpression have been observed in patient-derived HPV DNA-positive HNSCC.19, 20 p16INK4a, s function as an inhibitor of the cyclin-dependent kinases, decreasing pRb phosphorylation, is antiproliferative in normal cell cycle life. Because of the negative feedback mechanisms between pRb and p16INK4a, an overexpression of p16INK4a can be verified immunohistochemically in case of inactivation of the pRb by the E7 gene product.21 However, some additional mechanisms may further modulate these effects.22, 23
Based on these observations, several studies propose p16INK4a determined by immunohistochemistry in tumor tissue-derived material to be a useful surrogate marker for HPV-associated disease in HNSCC.24, 25 Exclusively, those HNSCCs being positive for both HPV DNA and p16INK4a staining are expected to show the typical clinical HPV-related behavior, as mentioned earlier.
The intention of our study is to further investigate the HPV activity in tumor tissues by determining the expression of p16INK4a in correlation with the HPV mRNA expression of the E6/E7 oncogenes. The study design, therefore, aims to help to clarify the reliability of p16INK4a staining as a surrogate marker for HPV-associated head and neck cancer.
Material and Methods
Patients, tissue specimens and DNA and RNA extraction
Tissue samples (n = 45) of histopathologically confirmed TSCC were obtained during surgery in the years from 2004 to 2007 from 39 patients (31 males; 8 females; 44 to 85 years of age, mean 62.1 ± 8.6 years) receiving treatment at the Department of Otorhinolaryngology, Head and Neck Surgery at the Christian-Albrechts-University of Kiel, Germany. All samples were retrieved following informed consent approved by the local ethics committee. Regarding the tumors, a dual work-up was followed: a part of the tumorous tissue was snap frozen in liquid nitrogen and stored at −80°C for further analysis. The remaining part of each tumor was processed for routine histopathology, and the tissue detection methods were performed on this routinely processed material. DNA was extracted from 25 mg of frozen tissue sample, as described previously.6 Total RNA was extracted using peqGOLD TriFast reagent (PeqLab, Erlangen, Germany), according to the manufacturer's recommendation.
The anatomical location of the primary tumors was as follows: palatine tonsil (n = 30) and lingual tonsil (n = 9). For further clinicopathological features see Table 1.
Detection of HPV DNA
The presence of amplifiable DNA in every individual sample was confirmed using PCR by applying β-globin primers.
For the identification of multiple high- and low-risk genital HPV genotypes in a single reaction, multiplex human papillomavirus genotyping (MPG), a quantitative and sensitive high-throughput PCR procedure, was carried out. MPG is based on the amplification of HPV DNA by a general primer PCR (GP5+/6+) and the subsequent detection of the products with type-specific oligonucleotide probes coupled to fluorescence-labeled polystyrene beads (Luminex suspension array technology). Up to 100 different HPV types can be simultaneously detected with MPG yielding quantitative results. For details of the method applied see Ref.26. Furthermore, PCR was performed using type-specific primers for HPV16, a combination of type specific (for HPV6, 11 and 18) and consensus primer (MY09 and MY11). To enhance the specificity of the detection, PCR results were confirmed by Southern blot hybridization using HPV gene-specific probes, as described previously.6 These procedures were added to minimize false negative results. Serial dilutions of DNA isolated from the HPV-positive cervical carcinoma cell line SiHa and reactions without HPV DNA were run in parallel as positive and negative controls.
Detection of HPV E6 and E7 mRNA
For the identification of HPV E6 and E7 mRNA expression, Northern hybridization was carried out using P32-labeled HPV16 Pstl-3 (URR E6E7) probes. In case of missing signals in Northern blot analysis, RT-PCR by applying the primer pairs E6for2TW + E7rev1TW PCR [product size: 793 bp (E6), 611 bp (E6*)] was achieved, subsequently performing Southern blot hybridization on the RT-PCR products. The signals were classified as being weak (+), moderate (++) and strong (+++), respectively (Table 1).
Serial sections of formalin-fixed and paraffin-embedded samples were deparaffinized in xylene and stepwise incubated in 100–70% ethanol. Epitope retrieval was performed by heating at 110°C for 10 min in 10 mM citrate buffer (pH 6.0). Endogenous peroxidase activity was blocked by incubating the sections in 3% hydrogen peroxide in PBS for 10 min. Primary monoclonal mouse anti p16INK4a antibody (MTM Laboratories AG, Heidelberg, Germany) was added to the slides for 60 min at room temperature followed by incubation with a secondary horseradish peroxidase-conjugated goat anti-mouse antibody (MTM Laboratories) at room temperature for 1 h. After counterstaining of the peroxidase complex, sections were dehydronized and embedded with Eukitt (Fluka, Deisenhofen, Germany).
Interpretation of p16INK4a staining
Immunostaining of the formalin-fixed, paraffin-embedded sections was reviewed by three-blinded independent observers, and strong nuclear as well as cytoplasmic staining was considered as a positive reaction. The p16INK4a antibody reactivity was scored on a semiquantitative scale, according to Klaes et al.21 as follows: negative (<1% of the cells were positive), sporadic (isolated cells were positive, but <5%), focal (small cell clusters, but <25% of the cells were positive) and diffuse (>25% of the cells were stained). Depending on this criteria, the results were further classified as being negative, weak (+), moderate (++) and strong (+++), respectively (Table 1).
The patients were followed for a median of 29 months (range, 0–78 months). In survival analysis using the Kaplan-Meier method, the primary statistical end points were recurrence free and overall survival. Factors tested for prognostic value included sex, tumor size [T, according to the TNM Classification of the UICC 1992], incidence and extent of lymph node metastasis (N, according to the TNM Classification of the UICC 1992), location of the primary tumor and HPV DNA status in the tissue samples. Log rank test was used to test for significant differences between the groups. Finally, the Fisher's exact test was used relating HPV positivity to p16INK4a and E6/E7 mRNA expression.
HPV DNA analysis
HPV DNA analysis was performed in altogether 45 tumor specimens derived from TSCC [patients investigated: n = 39 (see Table 1)]. Because of the large primary tumors, in each of the 6 patients, 2 separate tissue specimens could be derived from opposite anatomical sites of the same tumor.
Of the 39 investigated patients, HPV DNA could be detected in 21 patients (53.8%). HPV16 and HPV35 were identified as the only infecting HPV types in 19 cases (90.5%) and 1 case (4.8%), respectively. A double infection with HPV16 and HPV35 was demonstrated in an additional case. After PCR by applying consensus primer and additionally type-specific primer pairs, the remaining tissue specimens derived from 18 patients were tested negative for HPV DNA.
With respect to the anatomical tumor site, of 30 tissue specimens derived from palatine tonsillar carcinomas, 18 (60%) tested positive for HPV DNA. The 2 cases with detected HPV35 infection could be allocated to this specific tumor site as well. Of the 9 tissue specimens derived from lingual tonsillar carcinomas, 3 were infected with HPV16.
The clinicopathological assessment, including the quantitative results of the MPG, the results of the E6/E7 mRNA analysis and the semiquantitative classification of the p16INK4a immunohistochemistry are summarized in Table 1.
Expression of E6/E7 mRNA
Only HPV DNA positive cases (n = 21) were subjected to HPV E6/E7 mRNA analysis after they were treated with DNAse to remove contaminating DNA. Strong hybridization signals were detected in 11 cases (52%). The intensity of the positive signals was described as being weak, moderate and strong in 4, 1 and 6 cases as analyzed by three different observers. Interestingly, 10 of the 21 HPV DNA cases did not show detectable HPV mRNA in the tissue specimens investigated.
p16INK4a expression in tonsillar carcinomas
Immunohistochemistry revealed 12 of 45 tissue samples with p16INK4a overexpression. According to the criteria of Klaes et al.,21 a weak, moderate and strong p16INK4a expression was detected in 1, 2 and 9 of the positive cases, respectively (Fig. 1).
p16INK4 expression in correlation to HPV DNA and E6/E7 mRNA expression
All but 1 case (11/12) with elevated p16INK4a expression could be assigned to HPV DNA positive tumors. The correlation of p16INK4a overexpression and HPV-positivity proofed to be statistically highly significant (p = 0.0018). Ten HPV DNA positive cases with p16INK4a overexpression expressed E6/E7 mRNA in the tumor tissue as well. In a single case with p16INK4a overexpression, the analysis of the tumor showed HPV DNA presence but no E6/E7 mRNA expression. Finally, 1 case showed HPV DNA and E6/E7 mRNA expression with the p16INK4a immunohistochemistry remaining negative. In this case, 2 tissue specimens from different sites of the same carcinoma were investigated giving identical results.
In those cases (n = 6) in which 2 tissue specimens of the same tumor could be investigated, all results were identical even to the extent of corresponding (semi)quantitative signal intensities.
Nine HPV DNA positive cases did not show detectable E6/E7 mRNA or elevated p16INK4a expression.
Presence of HPV DNA, E6/E7 mRNA expression and p16INK4a overexpression and survival
First Kaplan-Meier calculations demonstrated no impact on survival data when the presence of HPV DNA was focused as a single factor (p = 0.922). However, the parameter describing an activity of the viral oncogenes in terms of E6/E7 mRNA expression or p16INK4a overexpression clearly showed statistical significance for better overall survival times of the patients (p = 0.027 and p = 0.021, respectively). Interestingly, the combination of the parameters HPV DNA status and p16INK4a overexpression showed significant better survival times for patients (p = 0.036) when positive for both factors, whereas the combination of HPV DNA status and E6/E7 mRNA scantly did not (p = 0.059). Therefore, it might be hypothesized that p16INK4a staining is a more precise instrument to detect HPV DNA activity in tumor cells as mRNA analysis procedures are more susceptible in handling.
According to the described statistical results, the Fisher' exact test demonstrated strong dependencies between HPV DNA and p16INK4a (p = 0.0018), HPV DNA and E6/E7 mRNA expression (p = 0.00022) and p16INK4a with E6/E7 mRNA expression (p = 0.0000019).
The described results once more confirm the involvement of mainly HPV16 in about half of HNSCC. At least in those cases in which in addition to the HPV DNA-positivity, the expression of the viral oncogenes E6 and E7 as well as the concomitant overexpression of p16INK4a could also be demonstrated; however, the direct involvement of HPVs in HNSC carcinogenesis seems to be solidly demonstrated.
This implication of HPV in the carcinogenic process is underlined by results presented by Braakhuis and colleagues27 who demonstrated for HPV DNA- and E6/E7 mRNA-positive HNSCC lack of p53 mutations and missing alterations of the crucial genes located on 3p, 9p (INK4a) and/or 17p (TP53), allelic losses of which are considered early events in HNSC carcinogenesis. However, the fact that 9 of 21 HPV DNA-positive cases did not show E6/E7 mRNA expression and/or p16INK4a overexpression clearly demonstrates that mere HPV DNA detection may not be sufficient to demonstrate a biologically active virus infection in HNSCC at the time of diagnosis, taking exception of possible effects of HPV DNA integration into the host genome. These observations might well account for inconsistent and conflicting HPV DNA-based data reported on HPV impact on, for instance, recurrence-free and overall survival times of the patients with the majority of reports favoring a survival advantage of HPV-positive cases.10–13, 19
Using strictly deterministic semiquantitative methods, we could neatly demonstrate that the intensities of positive signals for HPV DNA detection, E6/E7 mRNA expression and staining intensities of the p16INK4a immunohistochemistry correlate strongly, including the agreement of the results in those cases in which 2 tissue specimens derived from different sites of the same tumor were analyzed. In case of weak, moderate or strong E6/E7 mRNA expression, the intensities of p16INK4a staining were weak, moderate or strong, respectively, as well. Interestingly, all HPV DNA-positive cases lacking detectable E6/E7 mRNA expression and/or p16INK4a overexpression showed moderate (n = 1) or weak (n = 8) signals after HPV DNA analysis. However, the meaning of these observations remains unclear.
The fact that HPV DNA detection in HNSCC does not (i.e., differently from what happens in cervical carcinomas) almost automatically mean that the viral oncogenes are expressed requires diagnostic consideration. In our study, 10 of 21 HPV DNA positive cases were negative for mRNA. This observation and the very good correlation between viral mRNA expression and p16INK4a positivity are in excellent agreement with the data published by Smeets et al.24 The design of the latter study investigating paraffin-embedded SCC of the oral cavity (n = 30) and the oropharynx (n = 18) taken from 48 patients from The Netherlands was very similar to our study. With 30% (9/30) HPV DNA-positive SCC of the oral cavity, Smeets et al.24 demonstrate a HPV prevalence rate that is at the upper range that should be expected for this tumor entity. However, with reported 38% (7/18) HPV DNA-positives investigating SCC of the oropharynx, the HPV prevalence seems to be relatively low, be it that tonsillar SCC with an expected HPV-positive rate of about 55% are included into that “oropharyngeal” group. Of 16 HPV DNA-positive cases, 12 expressed E6/E7 mRNA, subsequently resulting in intense positive staining in p16INK4a immunohistochemistry. As in our study, the authors reported only a single case with strong positive staining for p16INK4a, yet without detectable HPV DNA and/or HPV mRNA. However, unlike the Dutch study, the results from our study demonstrated 1 HPV DNA-positive case with p16INK4a overexpression, but lacking mRNA expression and 1 HPV DNA-positive case with mRNA expression but missing positive p16INK4a staining. The meaning in terms of acting molecular mechanisms of these very rare but regularly observed exceptions to the high correlation between simultaneous HPV DNA-positivity, mRNA expression and p16INK4a overexpression remains unclear.
Notwithstanding slight differences in results between our and the Dutch study, the algorithm proposed by the latter recommending initial p16INK4a immunohistochemistry on tumor tissue specimens followed by HPV DNA analysis in case of positive p16INK4 staining and thus, albeit in an indirect way, detecting a clinically relevant HPV infection can fully be supported by our data. Alternatively, and to avoid underdetection of HPV, simultaneous HPV DNA and p16INK4a detection may be performed. However, to be strictly correct, at this point of discussion, the following should be noted. Although it seems plausible that in those HPV DNA-positive cases with demonstration of E6/E7 mRNA and overexpression of p16INK4a, a biological activity of the virus can be assumed and—vice versa—in those HPV DNA-positive cases lacking detectable E6/E7 mRNA and/or p16INK4a overexpression a viral activity is less likely, it cannot be fully excluded that (i) in the latter cases, the E6/E7 mRNA and p16INK4a levels may be just below the level that can be easily detected with the described detection methods applied here, and (ii) p16INK4a overexpression not necessarily needs to be linked to steady high levels of E7 mRNA. Therefore, the statements and conclusions made so far need to be seen as being hypothetical until further confirmation by multicenter prospective studies. The recommended algorithm for HPV analysis using p16INK4a staining as a surrogate marker for active HPV infections only is accurate until direct HPV detection methods are implemented into diagnostics on a routine basis in all cancer-treating institutions.
It can only be assumed that a distinct portion of the SCC investigated by Smeets et al.24 are derived from the palatine and/or lingual tonsils, which is unfortunately not mentioned throughout the article. Although sophisticated, it seems to be crucial to clearly separate between the termini “oral cavity” and “oropharynx” both describing an anatomical region in the head and neck area and the terminus “Waldeyer's tonsillar ring” or just “tonsillar”, describing a functional aspect of the specific lymphoepithelial character of the tonsils. This circumstance is complicated by the fact that 2 of the 3 tonsillar regions of the Waldeyer's ring (lingual tonsil and palatine tonsil) must be ascribed to the anatomical region of the oropharynx. Moreover, as SCC of the Waldeyer's tonsillar ring by growth might affect the anatomical region of the oral cavity, as well, it seems particularly important to neatly classify the tumor entities investigated. The frequently observed use of termini describing different aspects, i.e., anatomical regions versus functional characteristics of investigated tissues, complicates meta-analysis. Therefore, we propose to strictly use the terminus “tonsillar” for all specimens derived from tissue of the true Waldeyer's tonsillar ring (palatine, lingual and nasopharyngeal tonsil) and the terminus “oropharynx” for nontonsillar-derived tissue specimens taken from the other anatomical sites of the oropharynx (lateral and posterior wall of the pharynx). SCC derived from these anatomical sites present with HPV DNA prevalences in between 20 and 30%,6, 28 thus generating an additional subgroup of patients with cancer that might possibly benefit from future insight in HPV-associated disease in terms of, for instance, therapy management (i.e., radiotherapy and vaccination).
Applying Kaplan-Meier estimations, statistical analysis of our data revealed no influence on survival times when only the HPV DNA status is considered. However, a trend toward better overall survival times for HPV DNA-positive patients when also expressing E6/E7 mRNA and/or showing positive p16INK4 staining in the tumor tissue could be seen. Although we found no statistical significant difference, the data are along the same line of the gained knowledge with HPV infections only achieving impact on initiation and course of HPV-associated disease when a biological activity of the virus can be shown. The better course of disease of the latter patients might be explained by the hypothesis that radiation and other toxic agents may decrease the capacity of HPV E6/E7 to interfere with p53 and/or pRb and other host protein, rendering the tumors more susceptible to therapy. Although there are reports on better survival times of patients exclusively stratified for HPV DNA presence,11–13 our results lacking this correlation are in congruence to our previously published data.5 Besides methodical reasons for discrepancies of survival data published so far, there also seem to be population-based differences in the extent of HPV impact on the course of disease, which should be resolved by multicenter studies on the field.
In practice, in many clinical settings, no HPV DNA diagnosis is performed after resection of the HNSC tumors, let alone the determination of viral oncogenes expression. Therefore, we adhere to the proposal by Smeets et al.24 to determine p16INK4a overexpression and HPV DNA to be able to take advantage of prognostic value attributed to biologically active HPV infections in HNSCC. The value of these determinations and the importance of the infrequent but existing exceptions should be carefully analyzed in multicenter studies to be able to eventually impact the standard of care using p16INK4a as a surrogate marker for HPV DNA expression in future and to be able to establish the potential value and cost-effectiveness of the use of HPV vaccines in males and females from head and neck perspective.