No evidence for human papillomavirus having a causal role in salivary gland tumors

Salivary gland malignancies are a very heterogeneous group of cancers, where histologically more than 20 different subtypes have been identified and prognosis varies greatly [1]. These tumors are rare, yet affect patients in all age groups and are mainly found in the parotid gland, but can also be found in submandibular, sublingual and accessory parotid glands [2, 3]. In Sweden about 100 new cases of salivary gland cancers are diagnosed each year, accounting for about 8% of head and neck cancers [3], which is similar to 6% of head and neck cancers in the United States [2]. Today the determination of tumor subtype is mainly based on the pathologist’s histological evaluation, and due to the large histological variability in this group of tumors, diagnoses are often determined with uncertainty. As a supportive measure genetic profiling is used in some cases, however, the sensitivity of the tests varies greatly. Moreover, it may often be with great uncertainty the prognostic tumor grade is determined, which is directly correlated to the treatment strategy. The most common treatment strategy is surgery, sometimes followed by postoperative radiation therapy [2, 3].

It has been well established by us and others that human papillomavirus (HPV) plays a causative role in a large proportion of oropharyngeal squamous cell carcinomas (OPSCC) [4‐6] and that the mutational burden is lower in HPV positive OPSCC as compared to HPV negative OPSCC [7]. It is also well established that HPV can be valuable as a prognostic tool, since patients with HPV positive OPSCC have a much better clinical outcome compared to patients with HPV negative OPSCC [8]. However, whether HPV has a causal effect in salivary gland carcinomas is not as certain. A recent study indicated that there are major differences in mutation frequencies between different salivary gland tumor subtypes, where none or few mutations were found in mucoepidermoid, adenoid cystic and acinic cell subtypes [9], suggesting that other factors such as e.g. a viral infection may be the causative factor. There are a handful small studies that have shown an association between HPV infection and certain salivary gland malignancies, however, reported HPV prevalence varied to a great extent, as summarized in Additional file 1: Table S1 [10‐17]. Moreover, these studies included only a small sample size, and most studies only included one or few specific subtypes of salivary gland tumors with very few samples in each subgroup, thus missing proper control groups. In this study we investigated the occurrence of HPV (defined as being HPV DNA positive and overexpressing p16^INK4a) in, to our knowledge, the largest cohort of parotid salivary gland tumors, including 12 different subtypes, in patients diagnosed between 2000-2009 in the counties of Stockholm and Gotland.

In total 117 tumor samples, including 10 corresponding metastases, from 12 different salivary gland tumor types, were analyzed for presence of HPV DNA (Table 1). All samples were positive for the β-globin gene used as a positive control, indicating successful DNA purification. One out of the 107 (0.93%) tested malignant salivary gland primary tumors was positive for HPV16, which was 1/7 salivary gland squamous cell carcinomas included in this study. This tumor also showed p16 overexpression. No significant overrepresentation of HPV DNA was observed in squamous cell carcinomas as compared to the other subgroups (p = 0.06, Table 1). Moreover, the HPV16 and p16 positive tumor had a corresponding metastasis located on the neck, which also was positive for HPV16 and p16.

The patient with an HPV positive tumor was a 78-year old male patient with severe co-morbidities and no history of other tumors. Tumor origin was evaluated clinically and radiologically with no evidences of other origins than the parotid gland; the oropharynx was however not assessed histologically. Since the patient died some months after diagnosis, the follow-up was limited.

In this study, 107 malignant salivary gland tumors, including 12 different subtypes, with the addition of 10 corresponding metastases were tested for the presence of 27 different HPV types (including all known high risk HPV types) using a multiplex bead based assay. One of the 107 malignant salivary gland tumors and its corresponding metastasis on the neck were positive for HPV16 DNA, and both also overexpressed p16 and therefore could be defined as being HPV positive. All other subtypes tested were HPV DNA negative. In total, only one of seven squamous cell carcinoma salivary gland tumors was HPV positive and due to the small sample size of squamous cell salivary gland tumors it was not possible to draw any reliable conclusion to whether HPV was involved in the development of this tumor type. Furthermore, all other subgroups lacked high risk HPV DNA and did not express presence of any of the other tested HPV types, suggesting that HPV does not play a major role as a causative agent in these subgroups. The latter was shown in a more limited study by Skalova et al. [13].

Nonetheless, HPV positive head and neck cancers often present with an HPV positive neck metastasis, and it has been shown that HPV positive cancer of unknown primary of the head and neck region often originate from an HPV positive OPSCC [22, 23]. Hence there is also a possibility that the HPV16 positive squamous cell salivary gland carcinoma found in this study was instead a primary tumor, or a cancer of unknown primary originating from an undetected OPSCC. This would further support that HPV does not play a major role for the development of salivary gland tumors. However, the clinical evaluation in this case did not support such an assumption.

There are some limitations to this study. Firstly, salivary gland tumors are a rare and heterogeneous group of tumors including more than 20 different subtypes, making it difficult to gather representable amounts of material for this type of analysis. Furthermore, although, to our knowledge, the presence of HPV was analyzed in the largest cohort published so far, the number of samples per subgroup was still limited, which makes it difficult to draw any clear conclusions. Secondly, the presence of p16 overexpression in samples was only tested on the one sample, and its corresponding metastasis, that both had tested positive for HPV DNA. Nevertheless, previous studies have shown p16 overexpression in malignant salivary gland tumor cohorts, still lacking or having a low correlation to presence of HPV [11, 13]. These findings may also be indicative of HPV not playing the cancer-driving role and that p16 in these tumors is a bad surrogate marker for an active HPV infection. Lastly, the HPV16 positive tumor found in this study was one out of seven squamous cell carcinomas, which is a very rare type of salivary gland tumor. Here, no significant overrepresentation of HPV was observed in squamous cell carcinomas as compared to the other sybtypes. However, the numbers of squamous cell carcinomas was restricted and the results should be interpreted with caution. With a larger cohort of squamous cell salivary gland carcinomas, perhaps more conclusive results could be drawn to whether HPV is a causative agent in the development of these tumors or not.

In conclusion, HPV DNA analysis in a large number of malignant salivary gland tumors, including 12 different subtypes, did not show any strong indications that tested HPV types have a causal role in the studied salivary gland tumor types.