Human papillomavirus (HPV) genotyping and prognostic value of HPV E4 protein and transcription factors NANOG and SOX11 in atypical p16 patchy squamous epithelium of cervix

Cervical tissue samples fixed in 10% buffered formalin and embedded in paraffin, from Institute of Pathology, Faculty of Medicine, University of Ljubljana (IP FM UL) archives were used for this retrospective study. Using the laboratory information system, we identified cases labeled “p16 neg” from gynecological biopsies between January 1, 2016, and December 31, 2018. The study focused on that period because it provided sufficient time for complete follow-up. Of the 320 matches, we excluded glandular changes, LSIL cases, and cases of cervical abrasion.

We selected 20 unequivocally p16-negative i.e. p16 null atypical squamous epithelium cases and 56 with patchy p16 expression. After review, cases with sufficient tissue were retained, 11 p16-negative (null) cases and 24 with patchy nuclear and cytoplasmic positivity (Figure 1).

FIGURE 1.

Images of atypical squamous epithelium cases with patchy p16 reaction: (A–F). The same sample stained with HE and p16 at 200× magnification. Panels (A), (C), and (E) show HE staining, while panels (B), (D), and (F) show corresponding p16 staining.

Atypical squamous epithelium was morphologically classified as such when the proliferating squamous or metaplastic epithelium exhibited nuclear atypia, characterized by enlarged nuclei and irregular nuclear membranes. Cytoplasmic differentiation was minimal or absent in the in the middle and superficial thirds of the epithelium.

Control group included 9 condylomas, 12 CIN1, 11 CIN2, and 9 CIN3, matched for age (±5 years) with atypical squamous epithelium cases with patchy p16 reaction, diagnosed at IP FM UL between January 1, 2015, and December 31, 2019. All slides in the study were independently reviewed by two pathologists. Eligible tissue samples were identified through our laboratory information system. After the review, samples with adequate material for additional testing were selected. The exclusion criteria specified that no lesions classified as higher grade than the defining group were present in the slides from these women.

The final study group, in which all immunohistochemical (IHC) stainings and HPV genotyping were performed, included 17 excision biopsies and 7 conizations in the group with patchy p16 reaction, 9 excisions and 2 conizations in the group with null p16 reaction, 6 excisions and 3 conizations in the group with condylomas, 5 excisions and 7 conizations in the CIN1 group, 5 excisions and 6 conizations in the CIN2 group and 2 excisions and 7 conizations in the CIN3 group.

HPV genotyping was performed at the Institute of Microbiology and Immunology (IMI) FM UL. The samples were prepared according to previously described in-house protocol (5). Briefly, we cleaned the microtome with xylene and a DNA decontamination solution before cutting 5 sections of 10 μm from the paraffin block, with first two sections being discarded. A negative control (leiomyoma tissue) was cut between consecutive samples. A new blade was used for each sample. The last tissue section was designated for HE to assess the presence of changes in the remaining tissue block. In addition, corresponding cervical smears of the same patients that were taken up to two years before the tissue biopsy were tested as well.

For the DNA isolation, commercially available QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) was used and HPV genotyping was performed utilizing highly sensitive Allplex HPV28 Detection Kit (Seegene, Seoul, South Korea), both following the manufacturer’s instructions. Latter enables individual detection of 28 HPV genotypes: HPV6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 42, 43, 44, 45, 51, 52, 53, 54, 56, 58, 59, 61, 66, 68, 69, 70, 73, 82. For further analysis, the latest IARC classification was followed, where the following 12 HPV genotypes were considered as high-risk: 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59 (41) and the remaining 16 were categorized as low-risk (lrHPV) genotypes.

Immunohistochemical analyses for p16, Ki67, E4 HPV, NANOG and SOX11 were conducted on 3-4 μm sections of formalin-fixed, paraffin-embedded tissue. We performed the immunohistochemical reactions which are regularly performed in routine practice without controls, namely Ki67 (monoclonal antibody, clone MIB-1, Dako, Agilent Technologies, Santa Clara, California, USA), p16 (antibody against p16INK4a protein) (clone E6H4, Ventana/Roche), SOX11 (Mouse monoclonal antibody, clone MRQ-58, Cell Marque, Rocklin, California, USA).

Immunohistochemical reactions for E4 HPV were performed using commercially available monoclonal antibodies against E4 (XR-E4-1) (IVD, Labo Bio-medical Products, Rijswijk, Netherlands), which react with the E4 protein from at least the following HPV strains: 6, 11, 16, 18, 27, 31, 33, 35, 39, 42, 43, 44, 45, 51, 52, 53, 56, 57, 58, 59, 66, 67, 70, and 74. Both negative and positive controls were included on each slide. For the negative control, we used leiomyoma tissue, and for the positive control, a case of CIN1.

Immunohistochemical reactions for NANOG were performed using commercially available monoclonal antibodies against NANOG (Cell Signaling, dilution 1:200) (Merck, Kenilworth, New Jersey, USA). One negative and three positive controls were included on each slide. For the negative controls, we used normal endocervical tissue and for the positive controls a non-keratinizing squamous cell carcinoma of the oropharynx, a seminoma of the testis, and CIN3..

The staining process for all immunohistochemical reactions was performed automatically using the BenchMark XT apparatus (Ventana, Tucson, Arizona, USA), with the ultraVIEW detection system and/or OptiVIEW DAB Detection Kit (Roche, Basel, Switzerland).

The criteria for the evaluation were adjusted to the different expressions of the markers. Ki-67 was evaluated as a positive reaction with nuclear staining. Parabasal reaction in the squamous epithelium was considered normal (0); otherwise, it was assessed by thirds of the thickness of squamous epithelium, namely 1 (predominantly lower 1/3), 2 (predominantly lower 2/3), 3 (full thickness) (42).For p16, a positive tissue reaction was assessed as diffuse, strong staining of cytoplasm and nuclei in at least the lower third of the epithelium (en block) (2). The reaction was evaluated with 3 levels: 0 for completely negative reaction, 1 for reaction with isolated stained nuclei and cytoplasm (patchy), 2 for positive reaction en block as described above. For HPVE4, a positive tissue reaction was assessed as cytoplasmic staining of at least one squamous epithelial cell (22, 27). For NANOG, a positive reaction was assessed as staining of the cytoplasm or nucleus (43, 44). The reaction was evaluated in three levels: 1 for weak reaction stronger than endocervical cells in the control, 2 for moderate reaction similar to CIN3, and 3 for strong reaction as seen in the nuclei of control seminoma cells. SOX11 was evaluated as a positive reaction with nuclear staining of epithelial cells (38, 39).

Follow-up data on cytology, HPV testing and tissue biopsy diagnosis from the date of sample collection until May 30, 2024 were obtained from National Cervical Cancer Screening Registry ZORA (45).

Statistical analysis was conducted using IBM SPSS Statistics 27. The Spearman correlation test was applied due to the data’s non-parametric nature, with p < 0.05 as the significance threshold.

For all cases, a moderate correlation between HPV genotypes detected in cervical smears and in histological sample was observed, with a special emphasis that cervical smears were taken up to 2 years prior to tissue biopsy samples (r=0.526, p=0.001). The correlation was statistically significant for hrHPVs (r=0.633, p=0.000) but not significant for lrHPVs (r=0.261, p=0.114).

In squamous epithelial atypia with null p16 expression the following HPVs were detected in tissue as a single infection: 53, 82.

In squamous epithelial atypia with patchy p16 expression the following HPVs were detected in tissue as either single/co-infection: 6, 16, 31, 39, 44, 52, 53, 56, 59, 66, 68, 58, 73.

In condylomas the following HPVs were detected in tissue as either single/co-infection: 6, 11, 16, 18, 44, 51.

In CIN1, the following HPVs were detected in tissue as either single/co-infection: 6, 16, 31, 40, 51, 52, 53, 56, 61, 66, 68, 70.

In CIN2, the following HPVs were detected in tissue as single infections: 16, 31, 58, 73, while coinfections included 18 and 58, 26 and 53, and 58 and 59.

In CIN3, the following HPVs were detected in tissue as single infections: 16, 18, 31, 51, while coinfections included 16 and 52.

Regarding various hrHPVs, HPV16, either single or as a coinfection, was present in none of the cases from the group with a completely negative reaction to p16, in 1 case from the group with a patchy reaction to p16, in 2 cases in the CIN1 group and the condyloma group, in 2 cases in the CIN2 group, and in 4 cases in the CIN3 group.

E4 HPV was most frequently expressed in thickened epithelium with hyper- and parakeratosis of CIN1 and CIN2 than in other groups, including majority of atypical squamous epithelia with patchy or null p16 expression.

E4 HPV was negative in all cases with negative tissue HPV genotyping and all 6 cases with confirmed single HPV6 infection, including five condyloma cases, one CIN1 case and one case in the patchy p16 group.

An exception was a case with positive E4 HPV in the atypical squamous epithelium with patchy p16 expression group and with HPV16 and 31, where NANOG was 1+ and Ki67 was expressed throughout the epithelium (Figure 2).

FIGURE 2.

Only case in the group with atypical squamous epithelium with patchy p16 expression and positive reaction for E4 human papillomavirus (HPV). The tissue was positive for HPV16 and 31 and the follow-up was negative. The same section at 100x magnification: (A) HE staining; (B) p16 staining, (C) Ki67 staining, (D) E4 HPV staining, (E) Nanog staining, F. SOX11 staining.

FIGURE 3.

One of the three cases in the group with atypical squamous epithelium with patchy p16 with CIN3 on follow-up. The tissue was positive for human papillomavirus (HPV) 52, 53,56, 66, 73. The same section at 100x magnification: (A) HE. (B) p16 staining. (C) Ki67 staining. (D) E4 HPV staining. (E) Nanog staining. (F) SOX11 staining.

Another exception were two cases of negative E4 HPV in the CIN1 group were, namely one conization case with HPV16 detected on genotyping and in another biopsy case with HPV6 infection, where NANOG was 2+ and Ki67 was expressed throughout the epithelium.

In the CIN2 group, three cases were negative for E4 HPV. One was a biopsy positive for HPV73, another was a conization case with negative HPV genotyping, and the third was a conization case with HPV16 infection. All three cases had weak NANOG expression (1+).

In the CIN3 group, E4 HPV was positive in two cases. One case involved a combined (CIN2 and CIN3) lesion with surface koilocytic changes and HPV16, while the other had no koilocytic changes but was positive for HPV31.

The reaction to NANOG was predominantly cytoplasmic, perinuclear, and occasionally nuclear. There was a weak (1+) positive reaction to NANOG in morphologically normal squamous epithelium outside the lesion. Normal glandular epithelium was negative for NANOG, as well as the normal metaplastic squamous epithelium. Atypical squamous epithelium with patchy p16 expression showed 1+ NANOG expression in 11 out of 24 cases, while the remaining cases were negative. Among the 11 NANOG-positive cases, 2 were HPV-negative. In contrast, among the 13 NANOG-negative cases, 7 were HPV-negative.

In a minority of CIN cases, NANOG expression was uneven, sometimes present only in the lower layers of dysplastic epithelium.

Considering a reaction of 2+ or greater (strong reaction), NANOG was positive in 9/20 cases of HSIL, in 1/12 cases of CIN1 (the one with HPV16 infection), but no strong reaction was observed in cases with p16 patchy/null atypical squamous epithelium.

The immunohistochemical marker SOX11 was not included in the tables because it was negative everywhere except in one case. The only positive reaction for SOX11 was nuclear, in the basal part of CIN3 with HPV16 infection (Figure 4).

FIGURE 4.

Immunohistochemical reactions in the only cervical intraepithelial neoplasia 3 (CIN3) case with a positive SOX11 reaction in study. The same section at 100x magnification: (A) HE. (B) p16 staining. (C) Ki67 staining. (D) E4 HPV staining. (E) Nanog staining. (F) SOX11 staining.

Our study showed that atypical squamous epithelium with patchy p16 expression, which should be considered negative for HSIL according to the LAST guidelines, do not exclude high-risk HPV infection or HSIL in the follow-up, which is consistent with finding of previous studies (11, 23, 46). In particular, one third of cases with atypical squamous epithelium with patchy p16 expression had hrHPV, one forth had lrHPV, comperable to previous studies (11).

Our study group consisting of cases with atypical squamous epithelium with patchy p16 expression closely resembles CIN1 and condylomas in p16 expression, with no statistically significant difference in hrHPV positivity between them (Table 4). Notably, lrHPV are significantly more prevalent in CIN1 and condylomas than in atypical squamous epithelium with patchy p16 expression. In comparison, CIN2 had statistically more hrHPV and less lrHPV than atypical squamous epithelium with patchy p16 expression group. Conversely, there is a statistically significant difference between groups of atypical squamous epithelium with patchy p16 compared to atypical squamous epithelium with null expression. The latter contained no hrHPV, indicating a much lower risk for HSIL, in line with other studies (11).

We observed that patchy p16 staining, besides in the group with atypical squamous epithelium and patchy p16 expression, is frequently seen in LSIL with lrHPV infections. As expected, en bloc p16 staining appeared exclusively in HSIL and two CIN1 cases containing hrHPV (2). The only exception with positive p16 and negative HPV genotyping in tissue was a CIN2 case with initially positive high-risk HPV genotyping in cervical smear a month before our biopsy. The literature suggests that “HPV-negative” CIN may result from false negatives due to rare or latent infections, p53-related oncogenesis, CIN2 regression, or reactive changes mimicking CIN2, beside the technical issues (loss of tissue) (47, 48).

Our study confirmed that p16 strongly correlates with hrHPV infection, as no cases of completely p16-negative atypia tested positive for hrHPV. Clinically, this suggests that such patients are unlikely to develop high-risk precancerous changes. Notably, none of the CIN1 cases in our study exhibited a completely negative p16 reaction, which aligns with findings from other studies suggesting that p16 is not a surrogate marker for just transformative infection (2, 23, 49).

In the group with atypical squamous epithelium and patchy p16 expression, Ki67 showed fullthickness positivity in two of three cases that progressed to CIN3 during follow-up, making it the most reliable predictive marker for future HSIL in this group, consistent with the literature (8, 50). However, Ki67 alone would not be sufficient for risk stratification, as there is no clear distinction in its expression between groups with atypical squamous epithelium with patchy versus null p16 expression.

E4 HPV expression provided the clearest distinction between atypical squamous epithelium with patchy p16 and CIN1/2 groups, being significantly more frequent in CIN1/2 (Table 4). That is in line with another study, where E4 positivity increased with positivity of p16 reaction when p16 expression was limited to the lower two third of the epithelium, since two cases from CIN1 group were p16 positive (Table 1) (23). In the CIN3 group, E4 HPV was positive in one conventional CIN3 case and another with combined lesions (23, 50).

As a potential marker of productive HPV infection, E4 HPV was detected in only one case of squamous condyloma. This may not indicate a lack of productive HPV infection but rather that immunohistochemical staining for E4 HPV does not detect lesional cells related to HPV6, a common cause of condylomas (26). Notably, recent literature and manufacturer specifications no longer mention that this staining is unvalidated for HPV6, suggesting a need for reevaluation (21). Among other unvalidated HPV types in single infections, immunohistochemical staining reacted with HPV68-related lesion (one CIN2 case) but not with HPV73 (one CIN2 case). This selective reaction may complicate the clinical management E4 HPV negative CIN2, as an undetected lrHPV73 along with an en block p16 reaction could be mistaken for a transforming infection, leading to overtreatment.

NANOG expression was less frequently observed in atypical squamous epithelium with null or patchy p16 expression compared to normal squamous epithelium, where it typically exhibited a weak 1+ staining pattern. Notably, squamous metaplastic and glandular epithelium demonstrated a completely negative reactions to NANOG, consistent with findings from previous studies (30, 32). This observation supports the final histopathological diagnoses of immature or reactive squamous metaplasia in these lesions at least at some cases with patchy p16 staining.

No strong (2+) NANOG reaction was observed in the group with atypical squamous epithelium and was present in only one case within the LSIL group. This suggests a lower malignant potential in these lesions compared to HSIL. Similar findings have been reported in laryngeal dysplasia, where a strong cytoplasmic NANOG reaction was identified as an independent predictor of carcinoma, whereas weak reactions were not (29). Additionally, the same study noted “negligible” NANOG staining in normal laryngeal squamous epithelium (29). Potential explanations for the negligible NANOG expression in laryngeal epithelium compared to the consistently weak reactions observed in cervical epithelium include selective NANOG binding to high-risk HPV, though this does not fully explain its presence in not HPV related laryngeal dysplasia (36). We hypothesize that the absence of NANOG expression in squamous metaplastic and glandular epithelium may be just a characteristic of the original glandular epithelium, from which metaplastic squamous epithelium arises.

In our study, SOX11 expression in basal cells of the normal cervix, atypical squamous epithelium and LSIL was null, contrary to the literature (38). SOX11 was expressed in one case of CIN3 (Figure 4). Previous studies reported increased SOX11 expression in cervical squamous cell carcinoma and adenocarcinoma (40). Despite deeper tissue sectioning, no local invasion was observed in our case, leaving SOX11 expression in the group of HSIL.

During follow-up, three incident cases of CIN3 were identified in the group of atypical squamous epithelium with patchy p16 expression in the initial biopsy. None of these three cases had expressed E4 HPV. All three had confirmed HPV infections, two with lrHPV and one with hrHPV. In contrast, one case in the group of atypical squamous epithelium with patchy p16 expression did express E4 HPV and had a hrHPV, and no precancerous changes were found upon further monitoring (Figure 2). These results support the hypothesis that E4 HPV may be an indicator of productive infection that could regress in the group with patchy p16 expression.

The main limitation of our study is small sample size. Thus, caution is needed in interpreting the results of our study. This underscores the need for additional studies with larger sample size, some of which are already in progress (25).”