Human papillomavirus infection in head and neck cancer: The role of the secretory leukocyte protease inhibitor

We previously showed that secretory leukocyte protease inhibitor (SLPI) gene and protein expression is significantly lower in metastatic versus non-metastatic head and neck squamous cell carcinoma (HNSCC). However, we did not assess the human papillomavirus (HPV) status of these cases. Since SLPI plays a role in HIV and herpes simplex virus (HSV) infections, we hypothesized that SLPI may be involved in HPV-infected HNSCC. In HNSCC tissue (n=54), HPV DNA was determined and correlated with SLPI expression. Additionally, to investigate a possible role of smoking on SLPI expression in clinically normal mucosa, 19 patients treated for non-malignant diseases (non-HNSCC) were analyzed for SLPI expression and correlated with smoking habits. In HNSCC patients, SLPI expression showed a significant inverse correlation with HPV status. In patients with moderate/strong SLPI expression (n=19), 10.5% were HPV-positive. By contrast, patients with absent/weak SLPI expression (n=35), 45.7% were HPV-positive. Low SLPI expression was correlated with metastasis (P=0.003) independent of HPV status. HPV-positivity was clearly associated with lymph node status (81.3% N1-3 cases). In smoking non-HNSCC patients (n=7), 42.9% showed absent/weak and 57.1% moderate/strong SLPI staining. In non-smoking non-HNSCC patients (n=10) 83.3% showed absent/weak and 16.7% moderate/strong SLPI expression. For the first time, a correlation between SLPI downregulation and HPV infection was demonstrated, suggesting that high levels of SLPI, possibly induced by environmental factors such as tobacco smoking, correlate with protective effects against HPV infection. SLPI may be a potential biomarker identifying head and neck cancer patients not at risk of developing metastases (SLPI-positive), and those at risk to be infected by HPV (SLPI-negative) and likely to develop metastases.


Introduction
We previously demonstrated that secretory leukocyte protease inhibitor (SLPI) is significantly associated with head and neck squamous cell carcinoma (HNSCC) (1). Although we did not assess human papillomavirus (HPV) status, the viral etiological agent in a subset of HNSCC (2)(3)(4)(5), we, and others, repeatedly showed a correlation between HPV infection and metastases of these tumors (6)(7)(8). Additionally, SLPI has been associated with other viral infections and it has been demonstrated that the prevalence of oral HIV is reduced in cases of elevated SLPI expression (reviewed in ref. 9). By contrast, herpes simplex virus (HSV) downregulates SLPI levels in a cell culture model (10). Therefore, we evaluated the association between HPV status and SLPI expression in a series of tumors using multiple markers of HPV infection across multiple anatomical sites within the head and neck and correlated this with occurrence of metastases.
The SLPI protein, also known as antileukoprotease, is a 11.7-kDa (107 amino acids) non-glycosylated kazal-type serine protease inhibitor of neutrophil elastase, cathepsin G, chymotrypsin and trypsin (11). It is produced by different cell types including breast, lung, endometrium, ovary, salivary glands and by various host inflammatory and immune cells such as macrophages, neutrophils and B lymphocytes (12)(13)(14). In particular, the role of SLPI as a potent inhibitor of neutrophil elastase appears to be an important factor resulting in the protection of the mucosa and skin against proteolysis (13,15).
In HNSCC, two possible types of carcinogenesis are discussed: i) an HPV-driven carcinogenesis, leading to biologically more aggressive tumors including early metastatic spread into the locoregional lymph nodes, but with comparatively better survival rates of the patients (8,16); and ii) a tobacco/alcoholdriven carcinogenesis, leading to delayed onset of disease with a time lapse of approximately 10 years (17). However, recent data suggest a synergism between HPV and smoking (16).
Moreover, a recent report showed a positive correlation between cigarette smoke and SLPI-expression in a rat model (18). This finding motivated us to further study SLPI expression in healthy mucosal tissue of non-HNSCC patients and to correlate SLPI expression levels with smoking habits.

Materials and methods
Patients, tissue specimens, DNA and RNA extraction. From 2004 to 2009, tissue samples of histopathologically confirmed HNSCC were obtained from 54 patients (42 male and 12 female; range, 43-76 years; median, 58 years) during surgery at the Department of Otorhinolaryngology, Head and Neck Surgery at the Christian-Albrechts-University Kiel, Germany. All samples were obtained following informed consent approved by the local Ethics Committee (D 438/10). Tissue samples were treated as follows: one section of the tumor tissue was snap-frozen in liquid nitrogen and stored at -80˚C for further analysis. The remaining section of each tumor was processed for routine histopathology. DNA was extracted from 25 mg frozen tissue samples, as previously described (20,19). Total RNA was extracted using peqGOLD TriFast reagent (PeqLab, Erlangen, Germany) according to the manufacturer's protocol.
The anatomical location of the primary tumors was: tonsils [(n=20); palatine tonsil (n=14), lingual tonsil (n=6)], larynx (n=22), tongue (n=8), and soft palate (n=4). SCC of the tongue (anatomical localization of the oral cavity) and SCC of the soft palate (per definition anatomical localization of the oropharynx) were investigated in one subgroup (n=12) since the biological behavior of soft palate SCC is more comparable to oral cavity SCC than to tonsillar SCC (per definition anatomical localization of the oropharynx). Additional clinicopathological characteristics and tobacco smoking habits are presented in Table I.
We also obtained tissue samples from clinically normal mucosa of the aerodigestive tract from 19 patients (12 male and 7 female; range, 2-63 years; median, 43 years) who were treated for non-malignant diseases at our Department. Additional clinicopathological characteristics and tobacco smoking habits are presented in Table II.
Detection of HPV DNA. Two different PCR-based detection methods for HPV DNA were applied to increase precision in validity and reliability of HPV DNA diagnostics: multiplex HPV genotyping (MPG) assay and HPV Type 3.5 LCD-Array, as previously described (20-22; Table I).
Detection of HPV E6*I mRNA in fresh frozen tissue. In cases with RNA available for analysis (n=37), total RNA was subjected to ultra-short type-specific RT-PCR assays which were developed for 12 HR-and 8 pHR-HPV to amplify ~65 bp of the E6*I mRNA (23). RNA integrity was assessed by co-amplification of 81 bp of ubiquitin C transcripts. PCR products were detected by Luminex hybridization. The cut-off value (5 net MFI) to define HPV E6*I mRNA-positivity was applied as previously described (23).
The slides were reviewed by two experienced members of the Department of Otorhinolaryngology, Head and Neck Surgery, Kiel, and by an experienced pathologist. To assess SLPI protein levels, 300 cells in at least five areas were analyzed. The mean percentage of positive tumor cells was determined and cases were assigned to one of the following categories: (-) <5%; (+) 5-30%; (++) 31-75%; and (+++) >75%. Cases with a score of (-) were considered negative, cases with scores of (+, ++, +++) positive.
Algorithm for defining HPV-positivity. In the present study, specimens were classified as HPV DNA-positive only if the HPV DNA was detected by the HPV Type 3.5 LCD-Array and/or in the presence of high viral loads by BSGP5+/6+-PCR-MPG assay. HPV infections were classified as active if HPV E6*I mRNA was detected. When RNA was not available, HPV DNA-positive cases with p16 INK4A overexpression were classified as active.
Statistical analyses. Fisher's exact test was performed to test for differences between groups, using the SAS 9.2 software. All tests were performed two-sided, and P-values <0.05 were considered to indicate statistically significant differences.

Results
Clinicopathological characteristics and smoking habits of HNSCC patients. The clinicopathological assessment, including the HPV detection, HPV E6*I mRNA analysis, SLPI immunohistochemistry and smoking habits are summarized in Table I. The inverse correlation between smoking habits and HPV status is statistically significant (P= 0.02), whereas the correlation between SLPI expression and smoking habits is not. However, it is evident that all cases with moderate/strong SLPI expression can be assigned to heavy smokers and that most of the non-smokers showed absent/weak SLPI expression.
HPV DNA analysis. HPV DNA was detected in 33.3% of the analyzed cases. HPV16 and HPV18 were the only infecting HPV types in 88.9 and 11.1% of all cases, respectively. From the two HPV18-positive cases, 1 was HPV18 DNA alone and the other was a double infection with HPV16.
Anatomical distribution of the tumor and their HPV levels was: from 20 tonsillar carcinomas, 60% tested positive for HPV DNA, including the HPV18 case and the case with double infection (both palatine tonsil). From the 22 specimens derived from laryngeal carcinomas, 18.1% were infected   (Table III). The cases with absent/weak and the cases with moderate/strong SLPI staining were combined for further correlation analysis: out of 19 cases with moderate/strong SLPI reactivity, 89.5% were HPV-negative and 10.5% were HPV-positive; 54.3% cases with absent/weak SLPI expression (n=35) were HPV-negative, whereas 45.7% were HPV-positive. Thus, these data suggest that low SLPI expression is associated with HPV infection, whereas elevated SLPI expression appears to prevent such an infection. In the case of classifying the 3 cases with HPV RNA but no HPV DNA as active HPV infections, this correlation would be even stronger, since 2 of these cases showed absent/weak SLPI expression.

Lymph node-(N) status depending on SLPI expression, HPV infection status, alone and in combination.
The distribution of the N0 vs. the N1-3 cases among the group of absent/weak and moderate/strong was 20 vs. 80%, and 63.2 vs. 36.8% (P= 0.003), respectively. Elevated SLPI was associated with lower metastasis whereas absent/weak abundance of SLPI was associated with increased tumor burden of the neck. The distribution of patients with (n=22) and without (n=16) metastasis among the HPV-negative cases was similar. However, out of the HPV-positive cases (n=16), 13 developed metastases. The correlation between the HPV status, SLPI expression and the N-status was highly significant (P=0.003), despite the relatively small number of cases in each group (Table IV). Independent of HPV status, reduced SLPI levels were associated with tumor masses in the neck. However, the association between higher SLPI and lower prevalence of metastasis was only found in the absence of an HPV infection.
SLPI expression in clinically normal mucosa. SLPI immunohistochemistry of clinically normal mucosa obtained from 7 non-HNSCC patients with history of tobacco smoking showed absent/weak and moderate/strong staining in 42.9 and 57.1% cases, respectively. Patients without a history of tobacco smoking (n=12) showed absent/weak and moderate/strong staining patterns in 83.3 and 16.7% cases each (Table II). Although the absolute differences did not reach statistical significance (P=0.085) due to the small sample size the trend indicates a strong correlation between smoking habits and SLPI expression (P<0.0001).

Discussion
To our knowledge, this is the first report showing a correlation between: i) SLPI expression and HPV infection; ii) SLPI expression, HPV infection and lymph nodal disease in head and neck cancer; and iii) SLPI expression and tobacco consumption habits in clinically normal mucosa of non-HNSCC patients.
The data presented here demonstrate that high SLPI expression is correlated with reduced prevalence of HPV infection in the investigated tumor specimens of HNSCC, while the HPV-positive cases predominate in the group of absent to low SLPI expression. Therefore, even in the presence of HPV DNA, SLPI expression may prevent an HPV infection of the upper aerodigestive tract. These findings are in agreement with the current available literature regarding the role of SLPI in oral HIV infections. For example, McNeely et al (24) suggested that SLPI inhibits HIV infections by blocking HIV binding to the host cells. In agreement with the effects of SLPI   Total   HPV-positive  9  7  2  18  HPV-negative  5  14  17  36  Total  14  21 19 54 SLPI antibody reactivity was scored on a semi-quantitative scale, according to Cordes et al (1). The correlation shown here is statistically significant (P=0.005). When analyzing the SLPI negative (-) cases together with the cases with weak (+) SLPI expression against cases with moderate to strong (++/+++) expression the P=0.01. SLPI, secretory leukocyte protease inhibitor.  The correlation is statistically significant (P=0.003). SLPI, secretory leukocyte protease inhibitor.
on HIV, Woodham et al recently showed that SLPI incubation of epithelial cells reduces HPV entry into these cells in a dose-dependent manner. Furthermore, SLPI was shown to bind to the Annexin A2 heterotetramer (A2t). In turn, A2t is associated with HPV insertion into and infection of epithelial cells (25). Therefore, our data support the theory of the existence of a cellular receptor for HPV and the role of SLPI as a competing agent for such receptor binding sites. In addition, Wahl et al (26) suggested that SLPI mediates its anti-viral activity by affecting the host cells rather than the virus itself. In line with these findings, we found only 2 cases with high SLPI expression that were also HPV-positive. However, it remains unclear why these 2 cases had a higher lymph nodal status, suggesting that in cases with HPV infection, SLPI might lose its protective role against metastasis. Such a protective role of SLPI has previously been described (27,28), and has been assigned to the theory that SLPI affects the invasive activity of cancer cells by inhibiting enzymes promoting cancer invasion and progression. It is postulated that the absence or repression of the SLPI-antileukoprotease function, as we described here in metastatic tumor specimens, promotes spreading of the tumor by enabling degradation of surrounding tissues by proteases, secreted from the tumors.
We previously reported that significant elevation of SLPI is associated with non-metastasized HNSCC (1), a phenomenon which could be confirmed on mRNA as well as protein levels. Similarly, high statistical significance was again obtained when stratifying absent/weak and moderate/strong SLPI expression against tumor burden of the neck, using the data presented here. In contrast to our own previously published data on SLPI in metastasized and non-metastasized HNSCC showing no correlation between the expression of SLPI and the degree of tumor differentiation (1), other researchers (29,30) demonstrated such a correlation in human epidermal tumors, leading to the hypothesis that SLPI protein levels represent a negative surrogate marker for tumor progression, thus corroborating our present data regarding SLPI expression and metastasis in HNSCC. However, the exact mode of action of SLPI in carcinogenesis and metastasis remains unclear.
The role of SLPI as a potent inhibitor of neutrophil elastase appears to be an important factor resulting in the protection of the mucosa and skin against proteolysis. In addition, Chan et al (18) showed that total neutrophil elastase concentration and activity as well as SLPI activity was increased following exposure to cigarette smoke. This finding led us to investigate the role of SLPI in healthy mucosa of non-HNSCC patients and to correlate SLPI expression with smoking habits. In accordance with the data obtained in a rat model (18), we indeed showed that moderate/strong SLPI expression was mostly found in patients with smoking habits. In HNSCC, it is well documented that HPV-driven carcinogenesis occurs in younger patients when compared to patients with alcohol/tobacco-driven carcinogenesis (4,17). According to our data, it can be assumed that smoking-induced SLPI expression prevents HPV infection, resulting in delayed agentdependent cancer with poorer prognosis.
In conclusion, in the present study we identified for the first time a statistically significant inverse correlation between HPV infection and SLPI expression levels, suggesting that reduced expression of SLPI facilitates HPV infection. The previously described correlation between SLPI reduction, HPV infection, and metastasis was confirmed. In addition, we provided preliminary evidence that tobacco smoking is associated with SLPI expression.