Human Papillomavirus Infection and EGFR Exon 20 Insertions in Sinonasal Inverted Papilloma and Squamous Cell Carcinoma

This study aimed to clarify the roles of high-risk human papillomavirus (HR-HPV) infection and epidermal growth factor receptor (EGFR) exon 20 mutations in sinonasal inverted papilloma (IP) and sinonasal squamous cell carcinoma (SNSCC). Samples were collected from 20 cases with IP, 7 with IP and squamous cell carcinoma (IP-SCC), and 20 with SNSCC and examined for HPV infection and EGFR exon 20 mutations. Low- or high-risk HPV DNA was observed in 25% of IP, 57.1% of IP-SCC, and 35% of SNSCC cases. Transcriptionally active HR-HPV infections in IP-SCC and SNSCC, accompanied by p16 overexpression, were observed in 28.5% and 25% of cases, respectively. Heterozygous EGFR exon 20 amino acid insertions (ex20ins), located between amino acids 768–774, were observed in 45% of IP, 28.5% of IP-SCC, and 0% of SNSCC and chronic sinusitis cases. EGFR phosphorylation sites were located at tyrosine (Y) 845, Y1068, Y1086, and Y1197 and induced PI3K/AKT/mTOR activation. The phosphorylation pattern of EGFR with ex20ins resembled that of HPV-related SNSCC and oropharyngeal cancer. The transcriptionally active HR-HPV infection and ex20ins might be responsible for the pathogenesis of IP-SCC cases with different fashions. Since IP-SCC might be a multifactorial disease, further investigation is needed to understand IP-SCC etiology.


Introduction
Inverted Schneiderian papilloma (IP) is a benign neoplasm arising in the nasal cavity. Its etiology remains unclear, but several risk factors have been proposed, including human papillomavirus (HPV) infection and chronic inflammation [1,2]. A recent review revealed that 25-40% of cases with sinonasal papilloma have HPV infection [3].
In the clinical setting, IP sometimes accompanies synchronous and metachronous squamous cell carcinoma (IP-SCC). The rate of malignant transformation from IP to SCC has been estimated to be approximately 10% [3,4]. The detailed mechanism of malignant transformation is unclear, but several hypotheses have been proposed. The prevailing view is that high-risk (HR)-HPV infection might play a role; however, several contradictory results have been reported regarding HPV infection [5,6].
Udager et al. first reported the presence of activating EGFR mutations in 88% of IP and 77% of IP-SCC cases, and this phenomenon might be distinct in IP and IP-malignant transformation among sinonasal squamous tumors [10]. Recent investigations also demonstrated that EGFR mutations, especially ex20ins, are frequently observed in IP and IP-sinonasal SCC (SNSCC) but not in SNSCC without IP lesions [6,[13][14][15][16]. The significance of EGFR mutations in the etiology of IP and its malignant transformation remains unclear, especially ex20ins and the downstream phosphorylation targets of ex20ins. On the contrary, approximately 90% of EGFR mutations in NSCLC are observed within the tyrosine kinase domain, and they are the possible genetic mutation targets for treatment. EGFR exon 19 deletions and L858R point mutation in exon 21 are predominant EGFR mutations, representing the favorable response to EGFR tyrosine kinase inhibitors (TKIs). The ex20is has been reported as the subset of driver genes that account for 4% to 9% of NSCLC [17]. Although previous target therapies against the ex20ins in NSCLC were intensively investigated, the response rates and prognosis have been unsatisfactory compared to classical TKIs to other EGFR mutations [17]. Thus, it is important to clarify the status of the ex20ins in sinonasal lesions. Interestingly, several reports have suggested that EGFR mutations and HPV infection are mutually exclusive [6,9,18].
The present study investigated HR-HPV infection and EGFR exon 20 mutations to shed light on their roles in the pathogenesis of IP and IP-SCC.

Subjects
Samples were collected from the following patients at our hospital from 2008 to 2021: 20 with IP, 7 with IP-SCC, 20 with SNSCC, and 6 with chronic sinusitis as controls. These patients had no history of previous surgery, chemotherapy, or radiation therapy. SNSCC patients were independent from those with IP-SCC, and the SNSCC samples did not contain any IP part. All specimens were frozen immediately in liquid nitrogen after surgical excision or biopsy and stored at −80 • C until analysis.
This study was conducted with the approval of the Institutional Review Board of the University of Ryukyus (project identification code 156). Written informed consent was obtained from all participants prior to sample collection.

HPV Infection Analysis
HPV-related tumors for IP, IP-SCC, or SNSCC were defined as being positive for HPV DNA detected by PCR or DNA in situ hybridization (ISH) and p16 overexpression.
2.2.1. Detection of HPV DNA, Measurement of Viral DNA Load, and Physical Status of HPV Extracted DNA was subjected to PCR using the degenerate consensus primer sets MY09/MY11 and GP5+/GP6+, which were designed to amplify the L1 region, as in previous studies [19,20]. We developed a new quantitative real-time PCR assay system for the E6 and E2 genes of HPV-18, -33, and -52 in addition to HPV-16 [19]. See the Supplementary Materials for details (Table S1)

Immunohistochemistry for EGFR and p-EGFR
Four-micrometer-thick sections of FFPE samples were deparaffinized in xylene and hydrated in a graded alcohol series. Primary antibodies against EGFR (1:50 dilution, the same antibody used for western blotting) and p-EGFR (Y845, 1:50 dilution, the same antibody used for western blotting) were used. The immunohistochemical reaction was performed with a SAB-PO Kit (Nichirei Bioscience, Tokyo, Japan) [20].

Statistical Analysis
Statistical analysis was conducted with SPSS 25.0 (SPSS, Inc., Chicago, IL, USA). Pearson's χ 2 -test was used for categorical data, and the Mann-Whitney U-test was used for continuous variables. p-values less than 0.05 were considered significant. (Table 1) Sixteen (80%) of 20 IP cases had Krouse T3 or T4 stage. Although all cases underwent endoscopic sinus surgery, one (5%) recurred after surgery. In IP-SCC, the chief tumor location was the maxillary sinus (four cases, 57%), and there were four cases (57%) alive without disease. All SNSCC cases were advanced with stage III (36.4%) or stage IV (63.6%), and there were 13 cases (65%) alive without disease. There was no significant difference in the primary tumor site among the three groups.

HPV Infection in IP
Of the 20 patients with IP, HPV DNA was detected in 5 samples (25%) by PCR, 3 with HR-HPV (2 HPV-16 and 1 HPV-33), and 2 with low-risk HPV (HPV-6 and -11). However, the viral load could not be determined in the three HR-HPV cases because they were below the detection limit ( Table 2).
The p16-positive cell rates in IP are shown in Table S2. Twenty-five percent of cases had ≥75% IP cells with p16 immunoreactivity; however, these cases had only weak staining, and there was no case with p16 overexpression ( Figure 1A). There was also no significant relationship between p16 immunoreactivity and the presence of HR-HPV DNA (Table 2, Figure 1B). From these findings, no case was judged to have an HPV-related tumor.

HPV Infection in IP-SCC
Of the seven patients with IP-SCC, four (57.1%) had HPV DNA: three with HR-HPV (HPV-16, -18, and -33) and one with low-risk HPV (HPV-11). The viral load of IP-SCC-1 (HPV-16) was low ( Table 2), and p16 overexpression was not observed in this case. In the IP-SCC group, there were two cases with p16 overexpression (28.5%) who also had HR-HPV DNA by PCR and/or ISH.
The HPV-18-positive case (IP-SCC-3) had a viral load of 2.63 × 10 3 copies/ng DNA and an E2/E6 ratio of 8.0 × 10 −4 . In addition, p16 expression was moderate in the IP area ( Figure 1C), whereas it was strong in the SCC area ( Figure 1D). Thus, two IP-SCC patients (28.5%) were considered to have HPV-related tumors.
Six different heterozygous EGFR ex20ins were observed in IP (Table 3, Figure S1), which were concentrated in the region encoding amino acids 768-774. Three cases had N771_H773dup, followed by two cases each with D770_N771insGF or H773_V774dup, and one case each with H773_V774insTH or S768_D770dup.
Two of the seven IP-SCC cases (28.6%) were heterozygous for EGFR ex20ins. Interestingly, the same mutation (S768_D770dup) was detected in both IP-SCC cases. Amino acid substitutions were also observed frequently in four of the seven IP-SCC cases (57.1%).
AKT (S473) was highly phosphorylated in all IP, IP-SCC, and SNSCC cases, regardless of HR-HPV infection and ex20ins status. 4E-BP1 (T37/46) was phosphorylated in all IP, IP-SCC, and SNSCC cases. p-4E-BP1 levels were much higher in IP-SCC (lanes 7-9) cases than in SNSCC and oropharyngeal carcinoma with HR-HPV infection cases. p-4E-BP1 levels were higher in IP with ex20ins cases than in IP without ex20ins (lane 1) and chronic sinusitis cases. STAT3 was expressed in all samples examined, while it was weakly expressed in IP-5 and IP-9 with H773_V774dup and H773_V774insTH, respectively. STAT3 (Y705) was similarly phosphorylated, and p-STAT3 levels were lower in IP cases than in chronic sinusitis cases.
AKT (S473) was highly phosphorylated in all IP, IP-SCC, and SNSCC cases, regardless of HR-HPV infection and ex20ins status. 4E-BP1 (T37/46) was phosphorylated in all IP, IP-SCC, and SNSCC cases. p-4E-BP1 levels were much higher in IP-SCC (lanes 7-9) cases than in SNSCC and oropharyngeal carcinoma with HR-HPV infection cases. p-4E-BP1 levels were higher in IP with ex20ins cases than in IP without ex20ins (lane 1) and chronic sinusitis cases. STAT3 was expressed in all samples examined, while it was weakly expressed in IP-5 and IP-9 with H773_V774dup and H773_V774insTH, respectively. STAT3

EGFR and p-EGFR (Y845) Immunohistochemistry
Representative findings of EGFR and p-EGFR (Y845) immunohistochemistry in IP (IP-19, Table 3), IP-SCC (IP-SCC-2, Table 3) and SNSCC (SNSCC-17, Table 3) cases are shown in Figure 4. EGFR expression was usually observed in the basal cell layer of IP ( Figure 4A); however, p-EGFR (Y845) was not noted in IP ( Figure 4B). IP-SCC cases with S768_D770dup showed EGFR and p-EGFR (Y845) immunoreactivity in the IP and SCC parts ( Figure 4C-F). The SNSCC case with neither HR-HPV infection nor ex20ins demonstrated strong EGFR expression ( Figure 4G) but not p-EGFR (Y845) ( Figure 4H). These findings were consistent with the western blotting results. HR-HPV-positive samples (lanes 7, 9, 15, and 16 in Figure 3) showed robust p16 expression in western blotting. The EGFR phosphorylation pattern of the HR-HPV-positive samples resembled that of the samples with ex20ins (lanes 2 and 8).

Discussion
This study aimed to shed light on IP and IP-SCC in view of HR-HPV infection and EGFR exon 20 mutations.
HPV infection was observed in 25% of IP cases, 57.1% of IP-SCC cases, and 35% of SNSCC cases by PCR analysis. Because HPV was not transcriptionally active and had a

Discussion
This study aimed to shed light on IP and IP-SCC in view of HR-HPV infection and EGFR exon 20 mutations.
HPV infection was observed in 25% of IP cases, 57.1% of IP-SCC cases, and 35% of SNSCC cases by PCR analysis. Because HPV was not transcriptionally active and had a low viral load in IP, it is still controversial whether HPV DNA is involved in the etiology of IP. On the contrary, HR-HPV infection in IP-SCC and SNSCC accompanied by p16 overexpression was observed in 28.5% and 25% of cases, respectively, suggesting that HR-HPV infection is a crucial factor in sinonasal cancer, consistent with previous studies [14,23,24].
Regarding the impact of HPV type, HPV-16 was the primary type, followed by HPV-33 and HPV-18. The observed HPV types were consistent with the HR-HPV types detected in oropharyngeal carcinoma in our institution [22]. Information on viral load in head and neck cancers is limited. In our previous study of oropharyngeal carcinoma, the viral loads of HPV-16-positive samples ranged from 0.24 to 2.69 × 10 5 copies/ng DNA (mean 3.49 × 10 4 and median 3.1 × 10 2 copies/ng DNA) [25]. In the present study, the viral loads of HPV-16 and -33 in IP-SCC and SNSCC ranged from 0.7 to 1.63 × 10 6 copies/ng DNA. Although the sample number was limited, the viral loads in IP-SCC and SNSCC were not significantly different from those in oropharyngeal carcinoma. The results for p16 overexpression and HPV viral load suggest a subgroup of HPV-related lesions in IP-SCC and SNSCC.
EGFR ex20ins are heterogeneous but comprise mainly in-frame insertions and duplications that cluster between amino acids 762 and 774 [26]. The amino acids changed by ex20ins in the present study were located in the tyrosine kinase domain, which contains the C-helix and the loop following the C-helix. These ex20ins result in the internal rotation of the C-helix, thereby promoting the constitutive activation of EGFR [26]. Qin et al. reported that 2.24% of non-small cell lung cancer cases had unique ex20ins, with A767_V769dup (25.1%) and S768_D770dup (17.6%) as the most prevalent [27]. ex20ins resulting in amino acid changes were observed in up to 50% of IP cases and 57% of IP-SCC cases, and ex20ins were identified only in IP and IP-SCC cases. The most common subtype of ex20ins was S768_D770dup in the IP-SCC group. Although ex20ins are not common in sinonasal cancer (6.15%) [26], IP-SCC has a high incidence of ex20ins (30-90%) compared with other cancers [6,10,28]. In the present study, ex20ins were observed frequently in the IP group, suggesting that they are involved in the etiology of IP.
Y845 is an Src-induced transphosphorylation site on EGFR, which is located in an activation loop that influences kinase activity and is also recognized as a key residue for the oncogenic properties of EGFR [29]. When EGF binds to the extracellular domain of EGFR, the regulatory C-helix rotates from an outward to an inward orientation, leading to the constitutive activation of EGFR [30]. ex20ins has been considered to promote C-helix rotation by pushing the α-C helix, resulting in a conformational change of EGFR in the absence of EGF binding [17,30]. In the present study, ex20ins was located between amino acids 768 and 774 and may induce continuous EGFR activation without ligand binding. The active conformation of the tyrosine kinase domain leads to the autophosphorylation of tyrosine residues (including Y703, Y920, Y992, Y1045, Y1068, Y1086, Y1148, and Y1173) in the intracellular C-terminal tail of EGFR [31]. Autophosphorylation of Y1068 and Y1197 has been reported in ex20ins cases [10,32], which was also observed in our series. However, the most prominent phosphorylation site was Y845 in IP-SCC cases with S768_D770dup. Y845 in EGFR is similarly located in the activation segment of the kinase domain but has distinctive features. Y845 phosphorylation is not mediated by EGFR autophosphorylation but is co-mediated by c-Src [33]. c-Src binds to EGFR and phosphorylates its tyrosine residues to induce cell transformation and carcinogenesis [34]. The conformational change of the kinase domain induced by S768_D770dup allows c-Src access to Y845, which may lead to EGFR phosphorylation (Figures 2 and 3) [35]. The present report is the first to demonstrate Y845 phosphorylation in IP and IP-SCC cases with EGFR ex20ins. In addition, HPV-related tumors, IP-SCC-3, SNSCC-2, and oropharyngeal carcinoma showed Y845 phosphorylation despite the lack of ex20ins. Although the cause of Y845 phosphorylation in HPV-related carcinomas remains unclear, it might be due to carcinogenesis induced by HPV infection.
In the present study, active HR-HPV infection was observed in 28.5% of IP-SCC and 25% of SNSCC but not in IP and sinusitis, while sinusitis samples were different from the nasal mucosa in the healthy subjects. On the contrary, ex20ins was observed in 45% of IP, 28.5% of IP-SCC, and 0% of SNSCC and sinusitis. IP and IP-SCC are diagnosed by histological findings. Thus, these phenomena reflect that IP-SCC might be a multifactorial disease. The active HR-HPV infection and ex20ins might be responsible for the pathogenesis of half of IP-SCC cases from the present results. Although case numbers were limited, patients with active HR-HPV infection did not have ex20ins, except for one IP-SCC case (IP-SCC-1 in Table 3). The finding is in accordance with the previous reports [6,9,17]. Further investigation is needed to understand IP-SCC etiology fully, including the roles of HR-HPV and EGFR mutations.

Conclusions
HPV infection status and EGFR ex20ins were investigated in IP, IP-SCC, and SNSCC. HR-HPV infection in IP was observed with a low viral load and was not transcriptionally active. On the contrary, IP-SCC and SNSCC had transcriptionally active infections with high viral load. These results suggest that HPV integration may occur during persistent infection in IP and eventually induce malignant transformation to IP-SCC. EGFR ex20ins was frequently detected in IP and IP-SCC but not in SNSCC and chronic sinusitis. Since EGFR ex20ins in IP and IP-SCC induced EGFR phosphorylation and activated the PI3K/AKT/mTOR activation, IP and IP-SCC might have a different etiology from SNSCC. The present report is the first to demonstrate EGFR Y845 phosphorylation in IP and IP-SCC cases with EGFR ex20ins. Patients with active HR-HPV infection did not have ex20ins, except for one IP-SCC case. Since IP-SCC is histologically determined, further investigation is needed to fully understand IP-SCC etiology.
Supplementary Materials: The following supporting information can be downloaded at https:// www.mdpi.com/article/10.3390/jpm13040657/s1: Figure S1. Localization of EGFR ex20ins between amino acids 768 and 774 in IP and IP-SCC; Table S1. Primers used for HPV infection analysis and Sanger sequencing of EGFR exon 20; Table S2. Relationship between p16 immunoreactivity and HR-HPV in IP; and Table S3. EGFR amino acid insertions/duplications and substitutions.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets generated and/or analyzed during the present study have not been made publicly available. However, data can be made available from the corresponding author upon reasonable request.