Detection of Human Papillomavirus in Squamous Lesions of the Conjunctiva Using RNA and DNA In-Situ Hybridization

In-situ hybridization provides a convenient and reliable method to detect human papillomavirus (HPV) infection in formalin-fixed paraffin-embedded tissue. Cases of conjunctival papillomas, conjunctival intraepithelial neoplasia (CIN), conjunctival carcinoma in situ (cCIS), and invasive squamous cell carcinoma (SCC), in which low-risk (LR) and/or high-risk (HR) HPV types were evaluated by RNA or DNA in-situ hybridization, were retrospectively identified. LR HPV types were frequently detected in conjunctival papillomas (25/30, 83%), including 17/18 (94%) with RNA probes, compared to 8/12 (75%) with DNA probes. None of the CIN/cCIS or SCC cases were positive for LR HPV by either method. HR HPV was detected by RNA in-situ hybridization in 1/16 (6%) of CIN/cCIS cases and 2/4 (50%) of SCC cases, while DNA in-situ hybridization failed to detect HPV infection in any of the CIN/cCIS lesions. Reactive atypia and dysplasia observed in papillomas was generally associated with the detection of LR HPV types. Collectively, our findings indicate RNA in-situ hybridization may provide a high-sensitivity approach for identifying HPV infection in squamous lesions of the conjunctiva and facilitate the distinction between reactive atypia and true dysplasia. There was no clear association between HPV infection and atopy in papillomas or dysplastic lesions.

HPV signal presented in epithelial cells as dot-like foci visible at 100× magnification in the nucleus. Multiple dots in at least a subset of cells were required to call a case positive. In some cases, the small intracellular signals were quite numerous, and could become almost confluent within a single nucleus. Staining was restricted to epithelial cells, and stromal elements, including blood vessels and mesenchymal cells, were negative in all cases examined. In cases containing normal conjunctiva adjacent to a lesion, the non-neoplastic epithelium was also always negative.
With respect to the extent of staining within individual lesions, HPV was present diffusely in most papillomas. Focal HPV positivity was observed in 8/25 (32%) LR HPVpositive papillomas and 1/2 (50%) HR HPV-positive papillomas, while diffuse staining was noted in 17/25 (68%) and 1/2 (50%) LR and HR HPV-positive papillomas, respectively. In contrast, HPV expression was more localized in the OSSN cases that contained virus. Focal HPV positivity was observed in the single HR HPV-positive CIN (Case 39) and both HR HPV-positive invasive SCC lesions. Further details of HPV in-situ hybridization are summarized in Table 2, and representative images for positive LR and HR HPV in-situ hybridization are presented in Figures 2 and 3.

Correlation of HPV with Histologic Features
Conjunctival papillomas can become quite inflamed, leading to reactive epithelial atypia, in which epithelial cells show generally mild nuclear enlargement and pleo- with arrows pointing to the epithelial base and asterisks in the underlying substantia propria. Increased nuclear immunolabeling for Ki67 within the neoplastic epithelium (E,H) and the corresponding RNA in-situ hybridization, which were both negative for high-risk HPV (F,I).

Correlation of HPV with Histologic Features
Conjunctival papillomas can become quite inflamed, leading to reactive epithelial atypia, in which epithelial cells show generally mild nuclear enlargement and pleomorphism, along with some chromatin changes and more prominent nucleoli. Reactive atypia can sometimes be difficult to distinguish from dysplasia, although the latter process is often less uniform and more severe. Detection of LR HPV could potentially be helpful in such cases and support a less aggressive biological potential if positive. In 18 of 33 (55%) papilloma cases, intraepithelial inflammation and epithelial atypia diagnosed as reactive were detected. Dysplasia was rarer, and noted in 4/25 (16%) LR HPV-positive papillomas, 1/2 (50%) HR-positive papillomas (Case 29), and none of the HPV-negative papillomas. In three LR HPV-positive papillomas diagnosed with dysplasia, the focal in-situ hybridization signal corresponded to the foci of dysplastic epithelium, while in Case 33 and Case 39, discrete foci of epithelial dysplasia were observed, but more diffuse LR and HR HPV signal was noted. An inflamed papilloma with reactive atypia and LR HPV-positive epithelium is demonstrated in Figure 4A Koilocytic cells were identified in all lesion types but were relatively rare, with focal change observed in 2/6 (33%) of HPV-negative papillomas, 10/25 (40%) LR HPV-positive papillomas, 1/15 (7%) HPV-negative CIN/cCIS, and 1/2 (50%) HR HPV-positive invasive SCCs. More diffuse koilocytic-appearing cells were observed in 1/25 (4%) LR HPV-positive papillomas, as shown in Figure 4F. Immunolabeling for p16 was performed on seven papillomas, five CIN/cCIS, and three invasive SCCs at the time of initial case evaluation. Generally, p16 labeling corresponded to HPV detection. However, p16 labeling was observed in two CIN/cCIS and one SCC in which HPV was not detected.  Koilocytic cells were identified in all lesion types but were relatively rare, with focal change observed in 2/6 (33%) of HPV-negative papillomas, 10/25 (40%) LR HPV-positive papillomas, 1/15 (7%) HPV-negative CIN/cCIS, and 1/2 (50%) HR HPV-positive invasive SCCs. More diffuse koilocytic-appearing cells were observed in 1/25 (4%) LR HPV-positive papillomas, as shown in Figure 4F. Immunolabeling for p16 was performed on seven papillomas, five CIN/cCIS, and three invasive SCCs at the time of initial case evaluation. Generally, p16 labeling corresponded to HPV detection. However, p16 labeling was observed in two CIN/cCIS and one SCC in which HPV was not detected.

Patient Follow-Up and Atopic History
Patients returned for follow-up care in 17/33 (52%) papilloma cases, 8/16 (50%) CIN/cCIS cases, and none of the invasive SCCs. The length of follow-up (mean ± SD) was 605 ± 1219 days and 887 ± 1534 days for papilloma and CIN/cCIS cases, respectively. Recurrence was noted in only one case (Case 14) following an initial diagnosis of an HPV-negative papilloma. The recurrent lesion was originally treated with excision and subsequently observed.
History of atopic disease, including asthma, eczema, or allergic rhinitis or conjunctivitis, was also recorded for patients when possible. Electronic medical records were available for review for 18/33 (56%) of all papillomas, including 3 HPV-negative lesions, 14 LR HPV-positive lesions, and 1 HR HPV-positive lesion. Atopic disease, including asthma (1/3), rhinitis/conjunctivitis (1/3), and a combination of the two (1/3), was reported in patients with HPV-negative papillomas. A history of asthma was present in 9/14 (64%) patients with LR HPV-positive papillomas, 2 of whom also had additional atopic disease, including eczema and rhinitis/conjunctivitis, respectively. Two additional patients with LR HPV-positive papillomas had histories of eczema and rhinitis/conjunctivitis, alone. There was no reported history of atopic disease in the single HR HPV-positive papilloma. For cases of CIN/cCIS, there was no history of atopic disease in the HR HPV-positive lesion that had medical records available for review.

Discussion
The prevalence of HPV infection in conjunctival epithelial proliferations has been difficult to establish, due, at least in part, to the range of detection methods used [35,72]. To some degree, this has been dependent on the type of lesion, and findings in conjunctival papillomas have been more uniform over time [14,31,36]. Our results using DNA and RNA in-situ hybridization were similar to previous studies demonstrating LR HPV infection in the majority of conjunctival papillomas using PCR, as shown in Table 3 [12][13][14]18,31,36,37,44]. In the current study, LR HPV was detected in 94% of papillomas using RNA probes, while only 75% were positive for LR HPV when evaluated by DNA probes. Interestingly, prior studies using DNA probes had lower rates of HPV detected, perhaps reflecting the age of the studies [73]. While some single cases reports examined papillomas using RNA in-situ hybridization, this study represents the first series of cases, and the larger percentage of papillomas found to be positive using RNA in-situ hybridization could potentially reflect the increased sensitivity of this approach [74,75]. However, given the relatively small number of cases in each group, the differences between DNA and RNA in-situ approaches could also be due to normal variation. The cases that were negative for HPV by DNA in-situ hybridization were all several years old, and because of the known potential for RNA degradation, these were not retested using RNA in-situ hybridization. Conjunctival papillomas can occasionally become inflamed and show prominent reactive atypia, which can be difficult to distinguish from clinically worrisome dysplasia. Among the 33 papillomas we examined, generally mild dysplasia was identified in 5 cases (15%). This number may be elevated due to selection bias, as papillomas with concerning microscopic features were more likely to have had HPV status analyzed during the study period. LR HPV was detected in four of five potentially dysplastic papillomas, while HR HPV was only detected in one. This suggests that the majority of conjunctival papillomas with atypia or dysplasia are associated with LR HPV, and that in-situ analysis could be useful in supporting limited malignant potential in many of these cases. Similar to these findings at the ocular surface, LR HPV was detected in the majority of papillomas, including those with atypical features of the nasopharynx, oral cavity, and larynx, while HR types are more commonly detected in invasive SCCs of these tissues [78][79][80][81][82][83][84].
The reported prevalence of HPV in CIN and other more aggressive OSSN is more variable, with positive cases almost always containing high-risk types [31,34,35,85]. In prior studies, 40-100% of CIN/cCIS were HPV-positive by DNA in-situ hybridization and 0-31% positive by RNA in-situ hybridization [22,34,46,86]. Prevalence was higher in SCC (30-36% positive by DNA in-situ hybridization and 16-26% positive by RNA in-situ hybridization) [34,46,85,86]. In our cohort, 1 CIN/cCIS out of 12 (8%) was positive for HR HPV using RNA in-situ hybridization, while none of the 4 cases evaluated by DNA in-situ hybridization were positive, suggesting that RNA approaches may provide higher sensitivity for detection. The overall positive rate for the 16 cases examined was, thus, 6%.
Many demographic and clinical associations in our study were similar to prior reports. Papillomas of the conjunctiva reportedly occur more commonly in younger patients, while more dysplastic OSSN lesions tend to present in older ones, and our cases had a mean age of 43, 68, and 74 years at the time of presentation for papillomas, CIN/cCIS, and invasive SCCs, respectively [3,40,77,87]. Prior studies suggested that conjunctival papillomas and OSSN are more common in male patients, and this was true overall in our group as well, except for the relatively small set of SCC [3,7,21,40,46]. Papillomas in our cohort presented more frequently on palpebral conjunctiva (78%), while CIN/cCIS cases were more common for bulbar conjunctiva (82%). However, while a recent report suggested correlation of HR HPV-positive OSSN with atopy, our study did not identify a history of atopic disease in patients with HR HPV lesions [22]. Another group reported increased keratinization in HPV-negative OSSN, but we did not identify a similar association in our cases [46].
Limitations of the current study include the relatively small number of invasive SCC cases, precluding the possibility of statistical comparisons between groups. In addition, detailed review of electronic medical records was not possible for all cases in this series, as many were older pathology consultations submitted by outside physicians. While RNA in archived formalin-fixed paraffin-embedded tissue is labile and sensitive to degradation, potentially affecting the sensitivity of in-situ hybridization, we do not believe this is of concern in our study, as all in-situ hybridization assays were performed around the time of initial histologic review [63,65].

Study Design
A retrospective review of the Ophthalmic Pathology Archives of The Johns Hopkins Hospital for cases matching keyword searches including "conjunctiva," "in-situ hybridization," and "human papillomavirus" between 2007-2021 was conducted. Electronic medical records were reviewed for the following clinical and pathologic information: age at presentation, gender, lesion type and location, clinical treatment modalities, length of clinical follow-up, history of recurrence, history of atopic disease, HPV in-situ hybridization assay details, and details regarding any additional immunohistochemical stains. Any case for which in-situ hybridization was performed to detect LR and/or HR HPV in a conjunctival papilloma, CIN/cCIS, or invasive squamous cell carcinoma was included. Cases were reviewed to confirm the initial diagnosis, and intralesional koilocytic or dysplastic features were noted.

Conclusions
We confirmed that LR HPV is frequently detected in conjunctival papillomas while HR HPV is rarely detected in these lesions. The fact that most papillomas with histopathological dysplasia were positive for LR HPV suggests that their malignant potential is limited, and that in-situ hybridization may help with prognostic assessment, although analysis of more cases with long-term follow-up will be needed to confirm this. HR HPV was detected by RNA in-situ hybridization in a relatively small number of CIN and SCC, but was not clearly linked to atopy or other clinical and histopathological features.

Funding:
The authors received no financial support for authorship and/or publication of this article. C.P. was supported by NIH T32 OD011089 (PI: Mankowski).

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Johns Hopkins Hospital (NA_00011133, approved 10/9/2009) for studies involving humans.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.