Expression of CK7, CK19 and p16 in HPV-mediated oropharyngeal squamous cell carcinoma

Samples

A total of 109 patients including 69 cases of OPSCC and 40 cases of OSCC, were recruited from The First Affiliated Hospital of Guangxi Medical University and Guangxi Medical University College of Stomatology (Nanning, China) from July 2009 to December 2021. The criteria for patient inclusion were as follows: (1) patients were diagnosed as squamous cell carcinoma (SCC) by hematoxylin-eosin (HE) staining and postoperative pathology was consistent with the preoperative diagnosis; (2) the primary sites of OPSCC patients were tonsil, base of the tongue and soft palate and of OSCC patients were tongue, buccal and gingiva; there were no other primary and/or distant metastases cancers during the same period; (3) patients had not received preoperative radiotherapy, chemotherapy and immunosuppressant; (4) the case data were detailed and paraffin specimens were complete. Case data and pathological specimens were obtained from database and the pathology departments of both hospitals. All the pathological specimens were neutral-formalin-fixed and paraffin-embedded, and from each specimen, 10 serial paraffin sections at 5 µm were taken for HE and IHC staining.

Ethics statement

Permission was obtained from the Ethics Committee of The First Affiliated Hospital of Guangxi Medical University (No. 2022021) and Guangxi Medical University College of Stomatology (No. 2022-KY-E- (149)). The ethical review board approved the consent procedure and execution of this project. This study was granted an exemption from obtaining consent.

IHC

The primary antibodies of CK7 (Kit-0021), CK19 (Kit-0030) and p16 (MAB-0673), as well as secondary antibody (Kit-5010) were purchased from Maixin (Fujian, China). IHC staining for CK7, CK19 and p16 was performed on serial paraffin sections from the primary tumor of OPSCC/OSCC cases. After deparaffinization and rehydration, antigen retrieval was performed using Tris-EDTA (pH 8.0). Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 5–7 min. After washing in PBS, sections were incubated in primary antibodies overnight at 4 °C and in secondary antibody for 20 min at 37 °C. Color was developed with diaminobenzidine, and the sections were counterstained with hematoxylin. PBS, as a substitute for the primary antibody, was used as a negative control.

IHC staining assessment

p16 According to The 8th Edition AJCC Cancer Staging Manual (Amin, 2017) and Guideline For Pathological Diagnosis Of HPV Detection In Head And Neck Carcinomas From The College Of American pathologists (Lewis et al., 2018), p16 in non-keratinising SCC in which more than 70% of tumor cells show moderately or strongly positive staining in the nucleus and cytoplasm without HPV DNA/RNA detection is defined as HPV mediated (p16+) SCC. However, it has reported that the proportion of keratinising SCC in HPV+OPSCC patients’ specimens in Asia is higher than that in other regions (Toman et al., 2017; Meng et al., 2018). Therefore, in this study, we defined keratinising or non-keratinising SCC in which more than 70% of tumor cells show moderately or strongly positive staining in the nucleus and cytoplasm without HPV DNA/RNA detection as HPV mediated (p16+) SCC.

CK7 & CK19 CK7 and CK19 IHC staining of tumor cells was scored using the H-score system by two pathologists at a multi-headed microscope by consensus. The intensity of staining present in tumor cells was given a Score from 0 to 3 (0 = none, 1 = mild, 2 = moderate, 3 = intense). The H-score was then derived from the cross-product of the intensity and the percentage of tumor cells staining (0–100%) (Santoro et al., 2015; Woods et al., 2017, 2022).

Statistical analysis

Statistical analysis was performed using SPSS 25.0 for Windows. The consistency of CK7, CK19 and p16 was analyzed by the Kappa consistency test. The chi-squared test was used on 2 × 2 contingency tables to analyze the differences in positive rates of CK7/CK19 among HPV-mediated (p16+) OPSCC, non-HPV-mediated OPSCC and OSCC groups. Because H-scores of CK7 and CK19 were not normally distributed, the Kruskal-Wallis test was used to analyze the differences between groups. P-values < 0.05 were considered statistically significant.

Demographic characteristics

The study group comprised 109 patients, including 69 patients with OPSCC and 40 patients with OSCC. Among the specimens of 69 OPSCC patients, 14 exhibited non-keratinizing morphology with p16+ tumor cells comprising more than 70%, and five exhibited keratinizing morphology with p16+ tumor cells comprising more than 70%, whereas the remaining 50 were either p16− or keratinising/non-keratinising morphology with p16+ tumor cells comprising less than 70%. Thus, the final HPV status was determined as 19 HPV mediated (p16+) OPSCC and 50 non-HPV-mediated OPSCC. Demographic and clinical details of patients with OPSCC according to HPV-status are given in Table 1.

Expression of CK7/CK19/p16 in the cancerous nests and pericancerous normal epithelium of OPSCC/OSCC

Oropharyngeal mucosa, especially the tonsil and base of the tongue, is covered with stratified epithelium. Crypt epithelium, existing in the tonsil and the base of the tongue mucosa, consists of reticular epithelium. In contrary, oral mucosal epithelium is stratified squamous epithelium whitout crypt structure. In the oropharyngeal normal mucosal tissue (amygdalitis) control group, CK7, CK19 and p16 were positive in the normal tonsils (Figs. 1B–1D). However, CK7 and CK19 but not p16, were expressed in the oral mucosal epithelium (Figs. 2B–2D). In the pericancerous normal epithelium of OPSCC, CK7 was predominantly localized in the surface layer of the epithelium (Fig. 1F), while CK19 was primarily expressed in the subepithelial layer and basal cells (Fig. 1G). The expression of p16 was observed in the lower 1/3 to the entire layer of the epithelium (Fig. 1H), with significant overlap with the CK19 site. Conversely, the expression patterns of CK7, CK19 and p16 in the pericancerous normal epithelium of OSCC exhibit similarities but lack typical characteristics (Figs. 2F–2H). Additionally, our study has unveiled the presence of soft palatal lymphatic follicles and crypt epithelial structures that are positive for CK7, CK19 and p16 deep within the follicles, a novel finding in this context (Fig. 3).

The co-expression of CK7, CK19, and p16 in OPSCC/OSCC was presented in Table 2. The expression patterns of CK7, CK19, and p16 within the cancerous nests of HPV-mediated (p16+) OPSCC, non-HPV-mediated OPSCC, and OSCC were illustrated in Figs. 3–8. In addition to CK7+CK19+p16+ (Figs. 3–5), other combinations such as CK7−CK19+p16+ (Fig. 6) and CK7+CK19+p16− (Fig. 7) were also observed in both the cancerous nests and pericancerous normal epithelium of OPSCC/OSCC. Furthermore, individual positivity for CK7, CK19 or p16 was noted, with only one case exhibiting the expression of CK7+CK19−p16−. It is noteworthy that in the cancerous nest and pericancerous normal epithelium of OPSCC/OSCC, only the expression situation of CK7+CK19−p16+ could not be observed (Table 2 and File S1).

Comparison of CK7/CK19 expression between OPSCC and OSCC

The positive rates and H-score details of CK7 and CK19 in OPSCC/OSCC were given in Table 3.

The positive rates and H scores of HPV-mediated (p16+) OPSCC were substantially higher than those of non-HPV-mediated OPSCC and OSCC, as indicated by CK7. However, there was no significant difference in the positive rates and H scores between non-HPV-mediated OPSCC and OSCC.

The positive rates and H scores of HPV-mediated (p16+) OPSCC were significantly greater than those of non-HPV-mediated OPSCC, specifically for CK19. Moreover, the incidence of non-HPV-mediated OPSCC was greater than that of OSCC in terms of both positive rates and H scores.

Expression correlation of CK7, CK19 and p16 in OPSCC/OSCC

In OPSCC, the co-expression of CK7 with CK19 (P < 0.001), CK7 with p16 (P = 0.002), and CK19 with p16 (P = 0.019) showed similar results (Table 4) while only the expression of CK7 was shown to be consistent with CK19 in OSCC (Kappa = 0.410, P = 0.005) (Table 5).

Expression relationship of CK7, CK19 and p16 in the pericancerous normal epithelium of OPSCC

In this study, we observed the expression of CK7, CK19 and p16 in the pericancerous surface epithelium of OPSCC. CK7 was located in the surface layer of epithelium (Fig. 1F). CK19 detected in the subepithelium layer and basal cells (Fig. 1G). p16, primarily coincides with the CK19 location, mainly located in the lower 1/3 to the entire layer of the epithelium (Fig. 1H). Lee, Lee & Cho (2017) also observed that both the simultaneous expression of CK7 in the epithelial surface cells and CK19 in the basal cells of the cervical intraepithelial neoplasia (CIN). CK7 expression descends toward the basal layer (top-down expression), while CK19 ascends from the basal to superficial layer (bottom-up expression) as CIN progresses. Additionally, p16 expression coincides with CK19. They suggested that one potential explanation would be concurrent neoplastic transformation of CK7+ cells in the upper layer derived from SCJ and CK19+ basal stem cells in the lower layer infected by different or multiple high-risk HPV (HR-HPV).

However, Woods et al. (2017) found that CK7 was exclusively expressed in the surface layer of the crypt epithelium around the normal tonsil and the tonsil SCC, but CK7 did not appeared in the surface epithelium. Therefore, they believed that the carcinogenesis caused by HPV infection should originate from the crypt epithelium rather than the surface epithelium. In addition, in this study, CK7 was not only observed in the oropharyngeal crypt epithelium (Figs. 4F and 4F), but also in the surface epithelium (Fig. 1F). But further investigations are needed to investigate whether the expression of CK7 in the surface epithelium can serve as a new pathway for HPV infection.

It was reported that CK19 is expressed in the crypt epithelium of normal tonsil and the columnar cells of SCJ in the cervix (Lee, Lee & Cho, 2017; Woods et al., 2017). In vitro experiments, Chatterjee et al. (2019) discovered a significant upregulation of CK19 mRNA and protein in HPV-infected cervical and tonsillar cells, suggesting that these specific areas expressing CK19 are vulnerable to the development of cancer due to HPV infection. In this study, we also observed strong positive expression of CK19 in both the crypt epithelium (Fig. 4I) and surface epithelium (Fig. 1G) of OPSCC. We propose that CK19 could be served as an additional indicator for oropharyngeal crypt epithelial cells.

In addition, a study examining 262 cases of tonsil tissue without tumors showed that 25% of the cases had p16+ in the crypt epithelium while all the cases had p16− in the superficial epithelium (Klingenberg et al., 2010). However, we observed that p16 was positive in both the pericancerous crypt epithelium (Fig. 4L) and surface epithelium (Fig. 1D) of tonsil SCC. Furthermore, the presence of CK7 positive in surface epithelium observed above, indicates that the surface epithelium could potentially be acted as a pathway for HPV infection. In the case of carcinogenesis caused by tonsils with HPV infection, the surface epithelium is infected in a reverse manner by HPV from cancerous nests, according with the traditional view that HR-HPV first infects exposed basal cells and transformed basal cells divide and migrate upward following the squamous epithelial differentiation program (Lee, Lee & Cho, 2017). Another possible explanation is that the presence of CK19 and p16 in the cancerous nests may inversely cause the surface epithelium to express CK19 and p16.

In this study, we noticed the presence of CK7+CK19+p16+ in the pericanerous crypt epithelium of tonsil and base of the tongue SCC in serial sections. Additionally, we discovered lymphoid follicles in the soft palate, together with the presence of crypt epithelial structure of CK7+CK19+p16+ deep into the follicle (Fig. 6). It was the first time that lymphoid follicles and crypt structure were observed in human being soft palate tissue. According to reports, lymphoid follicles have been observed in the soft palate of pig and cattle, but not in humans (Casteleyn et al., 2011). Observing the lymphoid follicles and crypt structure in the soft palate is challenging in HE sections because of their atypical nature compared to the structure in tonsil and base of the tongue. The structure may potentially be served as an anatomical basis for HPV-mediated carcinogenesis in the soft palate.

Based on the above-mentioned, we propose that the susceptibility of the oropharynx to HPV infection is caused by the specific expression of CK7 in the reticular epithelium of the oropharynx. The crypt epithelium of tonsil and base of the tongue near the lymphoid follicles is similar to the SCJ of the cervix and anus. In terms of anatomical structure, the crypt structure increases the chance of HPV remaining in the epithelium. The susceptibility of the crypt structure to HPV-mediated carcinogenesis is enhanced by the unique binding mechanism of CK7 and E7 (Kanduc, 2002; Favia et al., 2004; Lucchese et al., 2010).

The relationship between OPSCC and expression of CK7, CK19 and p16

E7 is the predominant oncoprotein of HPV. HPV mRNA binds with the 6-mer peptide SEQIKA, which corresponds to amino acids 91–96 of the CK7 sequence. This interaction prevents the degradation of E7 mRNA. Only CK7 and E7 exhibit this specific response in the keratin family (Kanduc, 2002). CK7 is specifically expressed at the junction site in epithelium, such as the cervical SCJ epithelium and tonsil crypt epithelium, and it binds to E7 mRNA and locks its transcription at the junction site infected with HPV (Mirkovic et al., 2015; Yang et al., 2015; Woods et al., 2017). In contrast, CK19, as a heterodimer of CK7, unlocks and promotes the translation of E7 mRNA into the oncoprotein by binding to the CK7 SEQIKA peptide (Favia et al., 2004; Lucchese et al., 2010). The E7 oncoprotein interacts with and promotes the degradation of pRb, relieving the inhibition of p16 transcription and causing p16 overexpression (Chen et al., 2012; Serra & Chetty, 2018). The mechanisms that CK7 and CK19 regulate E7 until p16 overexpression during HPV carcinogenesis are yet unknown. In this study, except CK7+CK19+p16+, the expression patterns of CK7−CK19+p16+ and CK7+CK19+p16− could also be found in both the cancerous nests and pericancerous normal epithelium of OPSCC. Surprisingly, only the expression pattern of CK7+CK19-p16+ could not be observed in any of the cases (Table 2 and File S1). The statistical results showed that expression level (positive rate and H-score) of CK7 and CK19 in HPV-mediated (p16+) OPSCC was higher than those in non-HPV-mediated OPSCC and OSCC (Table 3), and the pairwise expressions of the three proteins were all consistent in OPSCC (Table 4). This indicates a strong correlation of CK7, CK19 and p16 with the development of HPV-mediated OPSCC. Thus, we hypothesize the existence of a link termed the “CK7→CK19→p16 expression axis” during the process of CK7 and CK19 involved in the regulation of E7. However, due to the limited of study, it has not been verified whether this expression axis follows a strict temporal order.

The expression pattern of CK7+CK19+p16− was observed in both OPSCC (Fig. 7) and OSCC (Fig. 8). Probably because CK7 and CK19 have paired properties, they can be expressed simultaneously. And p16− indicates that the case might not be attributed to carcinogenesis resulting from HPV infection. We also observed the expression pattern of CK7−CK19+p16+ in the HPV-mediated (p16+) OPSCC (Fig. 6). A study on the carcinogenesis process of cervical cancer also found that CK7 was strongly expressed in the precancerous lesion, and with the cancerous progression, nearly 1/3 of the cases were CK7− but still CK19+ and along with p16+ (Lee, Lee & Cho, 2017).

There are several possibilities for this phenomenon. Huang et al. (2016) found that CK7 is a marker of transformation to malignancy in low-grade CIN, while p16 is not involved. Moreover, Hashiguchi et al. (2019) also reported that CK7 disappearing indicates the carcinogensis progression and poor prognosis in cervical cancer. A study showed that CK7 was expressed strongly and specifically in the tonsil crypt epithelium, while CK7− and p16+ were found in tonsil carcinoma (Woods et al., 2017). This study suggested that CK7 gradually disappears with tumor progression and cell differentiation. However, Santoro et al. (2015) provided the first evidence of a strong association between CK19 up-regulation and HR-HPV+OPSCC. They suggested that the CK19 test may be valuable in detecting HR-HPV infection. Based on the explanations provided above and our research findings, CK7 may be served as an “early event” and a necessary condition for HPV infection. As carcinogenesis progresses, the expression of CK7 gradually decreases. CK19 may be served as an intermediary between the preceding and the following event. And p16, closely related to CK19, is a “late event” in the carcinogenic process of HPV infection.

According to Woods et al.’s (2017) study, there is a strong correlation between CK7 and p16 in OPSCC. This finding aligns with our own data (Table 4). In addition, our investigation revealed a strong positive expression of CK7, CK19 and p16 in OPSCC (Figs. 3–5). This strong positive expression might represent the highest level of expression within the tumor, and then CK7 gradually vanished. Thus, we deducet that CK7 needs the participation of CK19 to cause HPV-mediated carcinogenesis. Without the facilitation of CK19 on E7mRNA translation, CK7 would be unable to induce p16 expression. Therefore, CK7 and CK19 have a significant impact in carcinogenesis caused by HPV.

In this study, we found that the CK7 expression level of non-HPV association OPSCC and OSCC was similar (Table 3). Woods et al. (2022) also found a similar situation in the comparison of HPV negative OPSCC and OSCC, suggesting that HPV negative OPSCC and OSCC may originate from surface epithelial cells with no or low expression of CK7. There is a potential alternative explanation that the carcinogenesis of non-HPV-mediated OPSCC and OSCC is derived from the CK7+ salivary gland duct and duct epithelial cells, which may arise carcinogenesis even without HPV infection. Our previous study observed such SCJ structures in the vallate papilla of the base of the tongue (Chen et al., 2022). Mestre et al. (2020) also observed a structure similar to SCJ at the duct of the Von Ebner’s gland of HPV-16 transgenic mice. Therefore, this form of carcinogenesis might originate from the SCJ-like structures such as the Von Ebner’s gland at base of the tongue.

An interesting finding was that the CK19 expression level in OPSCC in the present study was higher in the non-HPV-mediated than that in OSCC (Table 3), similar to Woods et al.’s (2022) results. CK19 is normally expressed in the basal cells of nonkeratinized stratified squamous epithelium (Barrett et al., 2005; Moll, Divo & Langbein, 2008). We believe that the CK19 negative OSCC was derived from the keratinized squamous epithelium of the oral mucosa. This keratinized squamous epithelium usually lacks the expression of CK19, while the epithelium of the oropharynx is mostly nonkeratized stratified squamous epithelium with high expression of CK19. This result also further demonstrates the significance of the interaction between CK7 and CK19 in the development of HPV-mediated (p16+) OPSCC.

The findings of our study provided the possibility of “CK7→CK19→p16 expression axis” during the carcinogenesis of HPV-mediated OPSCC. However, one limitation of this study is that it is only a serial section study in which IHC was conducted, but no cytological or animal models experiments were performed. Currently, only cervical carcinoma conduct the cytologic study of these three proteins (Kanduc, 2002; Favia et al., 2004; Lucchese et al., 2010), while no in-depth cytological studies and even animal models have been performed on oropharyngeal carcinoma. The other limitation is that the present study lacks molecular studies to identify HPV status like HPV E7-HPV-mRNA in situ hybridization and HPV-DNA chips technology. Further research in molecular and cytologic experiments as well as animal models are needed in the future.