Analysis of human papillomaviruses and human polyomaviruses in lung cancer from Swedish never-smokers

Introduction

With an estimated number of 1.8 million new cases every year, lung cancer accounts for about 13% of the global cancer burden [1]. Lung cancer is the leading cause of cancer death among males, whereas among females, lung cancer is the leading cause of cancer death in more developed countries and second leading cause, after breast cancer, in less developed countries [1]. Smoking is the major pathogenic factor, with about 70–80% of lung cancers among women and 80–90% among men being attributable to cigarette smoking [2]. However, about 15% of all lung cancer cases occur in never-smokers [1]. Lung cancer arising in never-smokers has been suggested to represent a different disease entity due to differences in histology (predominantly adenocarcinoma), gender (female predominance), ethnicity (more common in Asian populations) and mutational spectra (e.g., less frequently KRAS-mutated, but harboring the majority of EGFR mutations and ALK rearrangements) [3]. Knowledge on the tumor biology behind these particular tumors has become increasingly important, with implications for preventive measures, diagnostics and therapy, especially due to the increasing incidence of smoking-independent lung cancer, both in Sweden and worldwide [4,5]. Potentially etiological factors that have been implicated in lung cancer of never-smokers include, e.g., radon, indoor or outdoor pollution, occupational exposures (including asbestos and various metal or coal fumes), environmental tobacco smoke, hormonal or reproductive factors, and heredity [6–8]. Furthermore, a possible role for oncogenic infections in lung cancer has been investigated in a large number of studies, with contradictory results [9–19].

High-risk mucosal human papillomaviruses (HPVs) are recognized as being associated both with anogenital cancer and oropharyngeal cancer, and more specifically tonsillar and base of tongue cancer [20–22]. In addition, several studies have implicated HPV as a causative factor in lung cancer, while other studies have contradicted these results and found no or at most very limited involvement of HPV in lung cancer [11–14].

The polyomavirus (PyV) family, currently with 14 species described from human material, has at least one member, Merkel cell polyomavirus (MCPyV) with the potential to induce tumors in humans [23]. Moreover, some animal PyVs can potentially be oncogenic in their respective hosts, e.g., murine polyomavirus (MPyV) and Raccoon polyomavirus (RacPyV) that can cause tumors in mice and raccoons, respectively [24,25]. MCPyV, discovered in 2008, is a causative factor of Merkel cell carcinoma (MCC), mostly in immunosuppressed or elderly patients [23]. Most HPyVs were discovered during the last 12 years, and although most of them are common in the human population, their potential disease association is still being evaluated. Notably, non-tumor diseases related to HPyVs, e.g., Trichodysplasia spinulosa caused by TSPyV or progressive multifocal leukoencephalopathy (PML) caused by JCPyV is mainly found in immunodeficient or immunosuppressed transplant patients [26,27]. It has therefore been suggested that also other HPyVs could potentially be oncogenic in immunosuppressed patients [27]. There are some studies investigating especially MCPyV, KIPyV, WUPyV and JCPyV as causative factors in lung cancer [15–19,28,29], but there are few studies where a larger number of HPyVs have been analyzed specifically for their potential association to lung cancer among never-smokers.

In order to investigate if HPV or HPyV can be a causative factor for lung cancer, we analyzed a regional population-based series of surgically resected primary lung cancers from never-smokers for the presence of these viruses.

Material and methods

Patients and tumor material

To identify all patients surgically treated for primary lung cancer (excluding carcinoids) between 2005 and 2014 at Karolinska University Hospital in Stockholm and registered as never-smokers (n = 113), we utilized the National Lung Cancer Registry in Sweden 2015 [5]. All cases with archived tumor tissue available (n = 94) were reviewed regarding histology and stage, by a thoracic pathologist (COV). Never-smoking status was verified through the review of corresponding patient charts, and defined as an estimated total lifetime dose of less than 100 cigarettes. In all, 87 cases with tumor tissue available fulfilled these criteria and were included in subsequent analyses. The clinicopathological baseline characteristics of these patients are summarized in Table 1. The study was performed according to permissions from the Regional Ethical Review Boards in Lund (546/2014), Sweden.

Table 1. Baseline characteristics of 87 early stage, surgically resected primary lung cancers from patients who never smoked.

	n (%)
Sex
Female	61 (70%)
Male	26 (30%)
Age at diagnosis
Mean	68 years
Range	37–83 years
Histology (WHO 2015)
AC	82 (94%)
Other	5 (6%)
EGFR status (TST26 Illumina©)
Mutated	36 (41%)
Wild type	49 (56%)
Not conclusive	2 (2%)
Stage (TNM 8th ed.)
IA	40 (46%)
IB	14 (16%)
IIA	8 (9%)
IIB	13 (15%)
IIIA	12 (14%)

Sample preparation and DNA purification

From the archived formalin-fixed paraffin embedded (FFPE) tissue blocks, a representative area with high proportion of malignant cells was identified, sectioned, and subjected to DNA extraction using the Qiagen AllPrep kit for FFPE and automated on the QIAcube instrument (Qiagen, Hilden, Germany).

Bead-based multiplex HPV assay

Analysis of HPV DNA was performed by PCR followed by a bead-based multiplex assay on a MagPix instrument (Luminex Inc., Austin, TX, USA), as described earlier [30]. In this assay, the presence of the L1 region, from 27 different HPV types, including all known high-risk types, and in addition the HPV16 and 33 E6 regions, were evaluated. To confirm the DNA quality of the sample, the presence of the β-globin, was also assayed for, as presented in Ramqvist et al. [31].

From each of the 87 samples, we included 10 ng of DNA in the PCR reaction, together with Multiplex PCR Master Mix (Qiagen, Hilden, Germany) and HPV and β-globin primers in a total volume of 25 µl. DNA from the HPV16 positive SiHa cell line, carrying 1–2 copies of the HPV16 genome, was included as a positive control both for HPV16 and β-globin. Five microliters of the PCR reaction was then incubated together with a mixture of 30 different MagPlex bead types (Luminex Corporation, Austin, TX, USA), each coupled to a specific probe (corresponding to L1 for each of the 27 different HPV types, HPV16 and 33 E6, or β-globin). A median fluorescent index (MFI), of more than 1.5× background (a distilled water sample)+15 was considered as a positive value.

Bead-based multiplex HPyV assay

Presence of PyV DNA was similarly assessed by PCR followed by a bead-based multiplex assay on a MagPix instrument, as described in Gustafsson et al. [32] and further modified according to Franzén et al. [33]. This assay separately targets both the ST and VP1 regions of 10 different HPyVs; BKPyV, JCPyV, KIPyV, WUPyV, MCPyV, HPyV6, HPyV7, TSPyV, HPyV9 and HPyV10 as well as the primate viruses SV40 and LPyV. The primers and probes utilized are described in Gustafsson et al. [32] and Franzén et al. [33]. Also here the β-globin gene was assayed for to confirm the DNA quality. PCR was performed as for the HPV analysis above, but with PyV specific primers. As positive controls, a cellular DNA sample, and MCPyV DNA, corresponding to 5, or 50 viral genomes were included in the assay. Five microliters of the PCR reaction was then incubated together with a mixture of 25 different MagPlex bead types (Luminex, Austin, TX, USA), each coupled to a specific ST, VP1 or β-globin. An output value, MFI, of more than 2× background (a distilled water sample)+100 was considered as positive. For MCPyV, the cutoff corresponded to approximately five genomes and values.

Results

Eighty-seven lung cancer samples from never-smokers (Table 1) were analyzed for the presence of DNA from 27 mucosal HPV types, including all high-risk types, and from 10 species of HPyV. All samples were positive for the β-globin gene, confirming the presence, amplification and detection of cellular DNA. Furthermore, all samples were negative for the HPV types included in the assay.

HPyV DNA was found in the material and was restricted to the identification of DNA from the ST and/or VP1 region of MCPyV in 15 samples and in addition a very weak signal for KIPyV VP1 in one sample. More specifically, 15 samples were positive for the MCPyV ST and/or VP1 region. However, 14 of these had a signal of around or below 10 genomes/10 ng cellular DNA (approx. 1500 cellular genomes), while one sample had a slightly stronger signal. This sample was titrated and the number of genomes were found to correspond to at most 0.03 viral genomes/cellular genome. Consequently, none of the samples had one genome of viral DNA per cell genome. MCPyV DNA is common both on human skin and all kinds of surfaces, e.g., fridge door handles, and thus samples, during preservation of the tumor in FFPE, cutting of samples and onward are easily contaminated with such DNA [34]. For this reason, these low amounts of MCPyV DNA were most likely derived not from the tumor, but from non-tumor sources.

Discussion

In the present study, 87 lung cancer samples from never-smoking patients were analyzed for the presence of DNA from 27 mucosal HPV types, including all high risk types, as well as 10 different HPyV species. The poor understanding of factors behind lung cancer in never-smokers brings a rationale to the search for potentially oncogenic virus DNA in this specific, and increasingly important, group. However, none of the examined tumors was found to be caused by any of these viruses.

HPV as a potential factor for lung cancer development has been the focus of a number of studies the past decades with varying results and conclusions, and is still a controversial question. A relationship between HPV infection and risk of lung cancer in nonsmoking women has indeed been demonstrated in several case-control-studies. Cheng et al., for example, published 2001, a study on lung cancer from nonsmoking women in Taiwan, where a majority, 55%, had tumors positive for HPV16 or 18 [35,36]. However, other studies presented much lower figures [37]. In a meta-analysis by Syrjanen, 22% of 7381 lung cancer samples were HPV positive, with variation especially depending on region and histological type of cancer [11]. Likewise, in a study by Ragin et al., on 3249 lung cancer samples, a large geographical variation in HPV prevalence was presented with the lowest (3%) in Europe, and highest (22%) in South and Central America [12]. No significant difference was found in the prevalence between never and ever smokers (9.9 vs. 7.6%). Furthermore, in a meta-analysis by Hasegawa et al., where lung cancer among nonsmokers was specifically evaluated, an HPV-prevalence between 0 and 100% was presented, with lower values for Europe [14]. Whereas some of the included studies reported a higher HPV prevalence among never smokers, e.g., see ref. [38], other showed the reverse, e.g., see ref. [39]. In concordance, with a low HPV prevalence in lung cancer in Europe, a study by Koshiol et al., on 399 lung cancer samples from patients in Italy, presented no HPV positive cases while a study on 334 tumors from Denmark by Sagerup et al. reported a 3.9% prevalence with no correlation to smoking [13,40]. In addition, both Tang et al. and Khoury et al. analyzed RNA expression data from 575 and 444 lung cancer cases, respectively, with unselected smoking status, in the TCGA dataset for expression of viral genes [9,10]. In both studies only one sample, a squamous cell carcinoma, was found to express viral genes from HPV16. The data in our study are therefore in line with the studies that present no or very low prevalence of HPV in lung cancer [10,13,14]. In addition, since no HPV-positive cancer was found our data do not support a higher HPV-prevalence in lung cancer from never smokers vs. smokers. It can also be noted that Colombara et al. analyzed potential associations between of prior HPV infection and development of lung cancer but no associations were found [41].

Some studies report a potential relationship between EGFR mutations (in turn heavily associated with never-smoking status) and HPV infections in lung cancer [42]. However, in our regional population-based set of tumors from surgically treated never-smoking lung cancer patients, regardless of EGFR status, we found no HPV positive cases (Table 1).

The association between HPyV and lung cancer has also been addressed before, but in fewer studies as compared to those investigating HPV, and mostly not specifically among never-smoking patients. A few studies have reported association between MCPyV or JCPyV and lung cancer [16,19,43]. However, other studies have found no such associations, e.g., in the study by Tang et al., noted above, no HPyV was found to be expressed in 575 lung cancer samples [10]. Furthermore, although there was an initial report on the presence of KIPyV in lung cancer, several later studies on KIPyV, WUPyV, MWPyV and STLPyV in lung cancer have generated negative results [18,29,44]. Notably, DNA from both MCPyV and KIPyV has been detected at low frequencies also in normal lung tissue [15,16,29,44]. In a study by Toptan et al., tissues from 1184 cancers, including 236 lung cancer samples, were analyzed for expression of HPyV antigens by immunohistochemistry [45]. Only 5% of these lung cancers were shown to express HPyV antigens in some individual cells, and the authors concluded that HPyV was not a major causative factor for any of these tumors. In a study by Malhotra et al., potential associations between the seroprevalence for nine HPyVs and lung cancer among never-smoking lung cancer patients compared to healthy donors were evaluated but no associations were found [46]. Similarly, in study by Colombara et al., no association between of prior infection with MCPyV, KIPyV or WUPyV and development of lung cancer was detected [41]. The result of our study is in agreement with these latter studies although specifically for lung cancer among never-smoking patients.

There are several limitations in this study. The number of analyzed samples was limited and thus, although all samples were regarded as negative, it is possible that a larger sample set would have included HPV or HPyV positive samples, in line with the study by Tang et al. where 1/575 lung cancers were actively transcribing HPV. Notably, no patients in the analyzed cohort had undergone immunosuppression and transplant patients may show a higher incidence of HPV or HPyV in lung cancer. It should also be noted that since all the patients were treated in Stockholm the result reflects the situation in this area.

In conclusion, our study shows no evidence for either HPV or HPyV, including MCPyV and another nine HPyVs, in the etiology of lung cancer in Swedish never-smokers.

Disclosure statement

No potential conflict of interest was reported by the authors.