Introduction

Ocular surface squamous neoplasia (OSSN) is the most prevalent non-pigmented neoplasia affecting the ocular surface.¹ OSSN is a tumor that affects the squamous epithelial cells of the conjunctiva and the cornea, and most frequently originates in the interpalpebral limbal area of the ocular surface. It is believed that limbal stem cells in the ocular surface, having a high mitotic rate, are more prone to suffer DNA damage. Ultraviolet radiation has been the primary risk factor associated with OSSN development,^2,3 but also increased age, fair skin, tobacco consumption and oncogenic viruses, such as human immunodeficiency virus (HIV) and human papillomavirus (HPV),⁴ have been implicated.^2,5,6

The role of HPV persistent infection as a risk factor for the development of OSSN has been controversial. Some studies have reported HPV detection in 100% of OSSN samples,^6,7,8 whereas others have reported no detection at all.^9,10,11 The large variability in HPV prevalence in OSSN can be explained by differences in the molecular techniques employed. Also, to the quality and quantity of biopsy samples, their storage conditions, and to potential sample contamination. Furthermore, most studies investigating the presence of HPV in OSSN lacked a control group, with adequate histopathological analysis of the epithelial characteristics to rule out epithelial dysplasia.^{6,7,8,12,13,14,15}

We aimed to investigate the presence of HR-HPV in fresh frozen biopsies of proven OSSN lesions, and to compare it to the presence of HR-HPV in a group of epithelial dysplasia-free pterygium samples, using a highly sensitive molecular technique. We also investigated the potential association of OSSN with tobacco consumption, sun exposure and other sociodemographic variables.

Materials and methods

Results

Sociodemographic characteristics, sun exposure and tobacco smoking data are shown in table I. In the OSSN group we found a male to female ratio of 1.4:1, whereas in the pterygia group it was a 1.2:1 ratio. The average age in patients with OSSN was 66 ± 16.2 years, in comparison to 57.6 ± 12.6 years in the pterygia group. Education level was higher in the OSSN group, where 32% had high school or college level, whereas in the pterygia group only 13.6% had high school level. We observed no difference in tobacco smoking between groups. A higher proportion of patients in the pterygia group reported outdoor jobs, mainly farming and street trading; while OSSN patients reported more recreational outdoor activities. The OSSN group had lower accumulated sun exposure (7.3 ± 7.7 hours) than the pterygia group (11.3 ± 9.7 hours), yet this difference was not statistically significant. Sunburn was more common in the OSSN group, reporting more skin redness with sun exposure than the pterygia group (45.5% vs 13.6%, respectively; p=0.021).

Table II shows data on histopathological diagnosis and HPV genotyping results in both groups. HR-HPV proportion in OSSN samples was 40.9%, whereas in pterygia samples was 4.5% (data not shown in table). The most prevalent genotypes in the OSSN group were HPV 16 (32%), HPV 58 (23%), and HPV 66 (19%). In the OSSN group, six cases had mild dysplasia, eight cases had moderate dysplasia, six cases had in situ carcinoma, and two cases had invasive carcinoma. The latter cases were referred to the National Institute of Cancerology to get a comprehensive evaluation by an oncologist. In the pterygia group, only one case had histological features compatible with koilocytotic atypia without other findings of epithelial dysplasia. Furthermore, this was the only subject, in the control group, with a positive result for HPV infection, where multiple genotypes were involved.

As shown in table III, bivariate logistic regression analysis showed an association between HR-HPV and OSSN (OR=14.54, CI95%=1.65,128.44, p=0.016), which became stronger after covariate adjustment (OR=16.35, CI95%=1.23,218.11, p=0.03). We found no association between OSSN and sex, age, tobacco smoking, nor accumulated sun exposure. We observed a significant association between sunburn and OSSN (OR=5.28, CI95%=1.20,23.15, p=0.02), which increased after covariate adjustment (OR=10.77, CI95%=1.85,86.01, p=0.02).

Discussion

We found that OSSN patients had 16.3 times the odds of having HR-HPV, compared to patients with pterygia (OR=16.35, CI95%=1.23,218.11, p=0.03). When we compared our results with previous reports, we found that very few studies included a control group, and fewer conducted a statistical analysis to control for potential confounders. De Koning and colleagues reported a proportion of HR-HPV in OSSN samples of 38%, and also included low risk OSSN lesions.²² Nevertheless, even though they had a control group, they found no association between HR-HPV and OSSN (OR=1.0, CI95%=0.4,2.7). Several differences between our study and theirs can be noticed. They used a group of patients with different diseases as controls (pinguecula, chronic inflammation, pyogenic granuloma, cavernous angioma, and other non-disclosed diagnoses), which could imply different degrees of participation of HPV in these lesions, given that no histopathological analysis to discard epithelial dysplasia was performed. The study was done in Uganda, which is a country with a high prevalence of HIV seropositivity.²³ The mean age in their OSSN cases was 35 years, while the mean age in our study was 66 years, indicating that their OSSN cases occurred in younger adults. They did not exclude HIV seropositive participants. HIV infection could simultaneously increase the probability of OSSN development^23,24,25 and HPV persistent infection,²⁶ which is an important confounder.

Previous studies show a great variability in the molecular techniques employed. These differences could explain, in part, the inconsistency in the HR-HPV prevalence observed across studies. Many studies employed a single-round PCR method to generate relatively long fragments, that may not be sensitive enough in samples with a low copy number of viral DNA, or in samples with multiple HPV genotypes.²⁷ Previous reports using tissue derived from genital lesions, showed that short fragment amplimers, such as SPF10, allow ultrasensitive detection of a broad spectrum of HPVs under these circumstances.^19,28 We only found one previous study using the same molecular technique employed in our study, the SPF₁₀HPV-LiPA system.²² Their results, in terms of HR-HPV prevalence in OSSN, are comparable to ours. In Mexico, two previous studies have analyzed HPV DNA from paraffin embedded samples of squamous cell carcinoma.^14,15 Even though, both studies lacked a control group, did not include low risk OSSN lesions and had DNA fragmentation in several samples; they reported a 22 to 50% prevalence of HPV 16, which is also similar to our findings.

We did not find an association between OSSN and solar exposure. Sun exposure is a common risk factor for OSSN and pterygia; thus, our controls likely had a high level of solar exposure, which could explain why we did not observe an association with OSSN. However, we did observe a difference between people who experienced sunburn and people who did not. Our results showed that sunburn is associated with the development of dysplastic changes and carcinoma development in the conjunctival epithelium (OR=10.77). Only one previous study conducted in Australia,² have reported that fair skin and sunburn were associated with OSSN. This is interesting given that people with more pigmented skin also have more pigment in the conjunctiva.²⁹ Melanocytes in the limbal conjunctiva are interspersed in the epithelial basal cells within Vogt palisades. Higa and colleagues have shown that melanin granules within these melanocytes have a polarized location towards the apical area of the cell where they function as a “solar shade” for limbal stem cells.³⁰ Probably, our finding is explained by the fact that people with less pigmentation have a higher risk of suffering DNA damage secondary to UV light exposure. Unrepaired DNA damage, especially in the limbal stem cells, can trigger dysplastic changes in the epithelium. To our knowledge this is one of the first studies to describe this association, yet further studies are needed to support this finding.

Despite the fact that some authors have reported the presence of HR-HPV in pterygium,^31,32 a causal relationship of HR-HPV with pterygium seems improbable. First, pterygium consists of a fibrovascular abnormal growth with elastotic degeneration of the collagen fibers and fibroblastic hyperplasia in the substantia propria of the conjunctiva with epithelial hypertrophy or atrophy,³³ but without epithelial dysplastic changes. Given that HPV is an epitheliotropic virus, evidence has shown that the oncogenic changes induced by HR-HPV always happen in the epithelial cells of the affected tissues such as anogenital, laryngeal and esophageal,³⁴ which is not the case in pterygium. Second, OSSN and other neoplasms, such as primary acquired melanosis, may coexist with pterygium.³⁵ Unfortunately, the majority of studies have not performed histological analysis on the pterygia samples to rule out dysplastic changes in the epithelium. Several studies have found evidence of concurrence with OSSN in pterygia samples, with a frequency that ranges from 5 to 24%.^35,36 In a recent study from Mexico the frequency of OSSN in pterygia was 11.2%,³⁷ highlighting the importance of a careful histological evaluation to rule out the concurrence of epithelial dysplasia in these lesions.

To our knowledge this is the first case-control study where HIV seropositive patients were excluded and fresh-frozen tissue samples from OSSN and pterygium were used for detection of HR-HPV through a highly sensitive molecular technique. Even more, we also documented many variables associated with OSSN, which allowed us to employ inferential statistical analysis to adjust for confounding factors, increasing the validity of our results. We excluded HIV seropositive patients, given the potential confounding role of HIV infection in the association between OSSN and HPV persistent infection.⁴ One limitation of our study is that only a single experienced anatomical pathologist made the histologic diagnosis of our samples, which potentially could be a source of information bias. Another limitation in our study, is that HR-HPV prevalence in pterygia and in healthy conjunctiva is unknown in Mexico. Nevertheless, if HR-HPV had an etiological role in pterygium the HR-HPV prevalence in these patients would need to be higher than in healthy conjunctiva, making our estimates for the association between OSSN and HR-HPV conservative. A final limitation of our study is that given the low frequency of OSSN we did not reach the initial estimated sample number; however, our expected HR-HPV proportion in OSSN (33%) was lower than the observed proportion (40.9%), giving us sufficient statistical power to detect the association.

Our findings must also be interpreted from a public health perspective. Previous cancer studies in Mexico have highlighted the importance of early detection,³⁸ and that is also relevant for ocular surface neoplasms. In the case of different types of eye cancerous lesions, including ocular surface neoplasms, few epidemiological studies have been conducted in Mexico. The availability and use of a national cancer registry could improve the epidemiological characterization of ocular surface neoplasms.³⁹ Finally, if HPV were to be confirmed as an etiologic agent for OSSN, HPV vaccination could substantially reduce the incidence of the disease; unfortunately, recent analyses suggest that HPV vaccination is decreasing in Mexico, which could affect the incidence of all HPV-related cancers.⁴⁰