Ocular surface squamous neoplasia (OSSN) is an epithelial tumour of the conjunctiva (Figure 1).¹ It starts with dysplastic cells in the basal epithelium as conjunctival intra-epithelial neoplasia (CIN), progressing to full-thickness dysplasia with carcinoma in situ (CiS).² Once the tumour breaks through the conjunctival basement membrane, it is classified as squamous cell carcinoma (SCC).³ It is the most common ocular surface tumour with incidence rates ranging from 0.03 to 3.4 per 100 000 persons/year, with lower- and middle-income countries disproportionately affected.^4,5

The three commonly described risk factors for developing OSSN are ultraviolet-B (UVB) exposure, human immunodeficiency virus (HIV) and human papillomavirus (HPV) infection.⁴ Ultraviolet-B causes DNA mutations, impairs the natural cancer immunosurveillance mechanisms and can activate a latent HPV infection.⁶ HIV infection similarly reduces the body’s natural immunosurveillance and is associated with an increased risk of concomitant oncogenic viral infections such as HPV.^4,7 Human papillomavirus infection produces oncoproteins that stop the arrest of mutated cells and result in uncontrolled cellular proliferation.⁶

Although HPV infection is commonly cited as a risk factor, the evidence in the literature remains inconsistent.⁴ Several reasons for this have been hypothesised, including sample handling, testing methodology and the geographic variation in HPV prevalence. Human papillomavirus is predominantly detected using three diagnostic modalities: immunohistochemistry (IHC), in situ hybridisation (ISH), and polymerase chain reaction (PCR).⁸ Immunohistochemistry is an indirect test that measures the expression of p16. This is overexpressed with an HPV infection, but can be elevated in other situations. It is therefore useful as a screening test with high sensitivity, but lacking in specificity. In situ hybridisation tests for the HPV DNA/RNA within cells and is therefore very specific. Polymerase chain reaction is also very specific and identifies the types of HPV present.^8,9

A recent study in South Africa found HPV in 66.7% of OSSN histology specimens with PCR testing.¹⁰ The prevalence of HPV in South African females undergoing pap smear has been reported at 32.7%.¹¹ There are no data for the prevalence in conjunctiva in South Africa.

The aim of this study is to describe the prevalence of HPV infection in OSSN in an urban South African context.

A prospective cross-sectional case-control study of HPV in ocular surface squamous neoplasia was conducted at a tertiary academic Ophthalmology unit in Johannesburg, South Africa.

Informed consent was obtained, and data were entered into a database designed for the study. An interview was conducted to determine the history of presentation and associated risk factors for OSSN. An anterior segment examination with anterior segment photos was performed for all participants. Cases were recruited from patients undergoing biopsy for the diagnosis of suspected OSSN. Ocular surface squamous neoplasia was suspected in the presence of an elevated conjunctival lesion with feeder vessels, leukoplakia, pigmentation or a diffuse pearly corneal lesion. Controls were recruited from patients undergoing surgery for indications other than OSSN or pterygium excision. The conjunctiva was examined as normal in these participants in the preoperative assessment.

Tissue was collected during surgery. For participants with suspected OSSN, this was taken from the corneal margin in cases undergoing excision biopsy and from any free edge of the biopsy in cases undergoing incision biopsy. For the controls, the conjunctiva was taken at the edge of a conjunctival incision (glaucoma or extracapsular cataract surgery) or of the bulbar conjunctiva when no routine conjunctival incision was made.

Tissue was stored in DNA/RNA-shield (R1100; Zymo Research) and frozen until it was processed for PCR. Samples were processed using the Quick-DNA Miniprep Plus kit (D4068; Zymo Research) using the manufacturer’s protocol for DNA isolation from solid tissues. Nested PCR was used to detect cutaneous cancer using published primers.¹² Amplification was performed with NEB OneTaq 2X Master Mix (M0482; NEB) under standard cycling conditions, and PCR products were run on a 1% agarose gel. PCR products were purified using the ExoSAP protocol (M0293 and M0293L; NEB), sequenced with the Nimagen Brilliant Dye Terminator kit, cleaned with the ZR-96 DNA sequencing Clean-up kit (D4053; Zymo) and analysed on ABI 350XL Genetic Analyser. The detection of 19 high-risk HPV and nine low-risk HPV types was performed using AnyplexTM II HPV28 Detection (HP7S00X, Seegene) according to the manufacturer’s instructions using 5 µL of DNA, utilising five fluorescent dyes on the CFX96 real-time PCR instrument (Bio-Rad). Distinction of HPV genotypes was performed using melting curve analysis and semiquantitative assessment of viral load was categorised as low (+; > 40 cycles), intermediate (++; 31–39 cycles), or high (+++; < 31 cycles). For all PCR steps, a no-template control is used to ensure that there is no signal or bank for the control. The laboratory personnel were blinded to the study data.

Data analyses were performed in Stata 18.5 (StataCorp LLC, Texas, USA). The required sample size was calculated at 22 participants to detect the expected effect with 80% power at a two-sided significance level of 0.05. Continuous variables were summarised using mean and standard deviation when approximately normally distributed, or median and interquartile range when non-normally distributed. Normality was assessed with the Shapiro–Wilk test; a P-value > 0.20 was used to indicate a normal distribution. Comparisons of continuous data that did not meet normality assumptions were conducted using the Mann–Whitney U test. Categorical variables were compared using Fisher’s exact test. A two-sided P-value < 0.05 was considered statistically significant.

Sixty-two participants were included in the study, 32 with OSSN (cases) and 30 with normal conjunctiva (controls) (see Figure 2). The median age of the cases and controls was 43 years with an equal distribution of males and females (Table 1).

The OSSN lesions were divided by histology into CIN (n = 7, 22%), CiS (n = 14, 44%), and SCC (n = 11, 34%). Morphologically, 28 (87.5%) had placoid, 6 (18.8%) diffuse and 1 (3%) nodular features.

Human papillomavirus was found in 21.8% of OSSN biopsies, where none of the controls had HPV (P = 0.011). Cutaneous HPV types accounted for 57% of the positive cases (Table 2). The most common HPV type identified was HPV14D (n = 3, 43%).

Sun exposure was greater in the cases (p = 0.007), and ocular trauma was greater in the controls (p = 0.02). There was no statistically significant difference between the cases and controls regarding tobacco use and petroleum product exposure.

HPV infection was only found in the HIV positive cases, with none of the HIV negative cases having a concomitant HPV infection (Table 3). This was not found to be statistically significant (p = 0.296).

The cluster of differentiation 4 (CD4) count, viral load and HAART status were compared between the cases with HPV infection and those without (Table 4). No statistically significant correlation was found.

HPV infection was correlated with the histology of the OSSN lesions. No statistically significant association was found (p = 0.67), see Table 5.

Ocular surface squamous neoplasia is the most common ocular surface tumour with three main risk factors: HPV infection, UVB exposure and HIV infection. Our study investigated the association between HPV infection and OSSN. We found that 21.8% of the OSSN cases had an associated HPV infection, whereas none of the controls was positive for HPV. Cutaneous HPV had a higher prevalence in our study than the mucosal types.

Human papillomavirus was first described in OSSN in 1989 by McDonnel et al.¹³ Following this, numerous studies were conducted to investigate this association. The initial studies focused on the mucosal HPV types, based on the association with cervical cancer.^{14,15,16,17,18,19} Ateenyi-Agaba et al.²⁰ were the first to investigate the association with cutaneous HPV types, finding a strong association between OSSN and cutaneous HPV and no association with mucosal HPV. Since then, more studies have looked at both the mucosal and cutaneous types, with a stronger association with the cutaneous subtypes.^21,22,23 Our study investigated the association between the mucosal and cutaneous types with a higher prevalence of cutaneous subtypes found. The most common type found was HPV14D. HPV14 was similarly described in other studies on the African continent.^21,23

Human immunodeficiency virus is a well-established association with OSSN.^5,24 The pathogenesis of this association is underpinned by reduced immunosurveillance and an increased prevalence of oncogenic viruses in these patients.⁴ Simbiri et al.⁷ described a high prevalence of concomitant infection with oncogenic viruses in OSSN samples. Tornesello et al.²¹ investigated the prevalence of HPV in OSSN and found that 25% of patients living with HIV with OSSN tested positive for HPV, compared to 10% of the patients living without HIV, with OSSN (p = 0.03). Our study looked at this and found that HPV was only found in the patients living with HIV. Our sample size was small in this pilot, and so significance was not demonstrated with statistical analysis.

The interplay between UVB, HIV and HPV as risk factors for OSSN is not well understood. UV exposure is an independent risk factor with decreasing risk as one moves away from the equator. Newton et al.²⁵ described a 49% decrease in the incidence of SCC of the conjunctiva for every 10° increase in latitude. Conversely, a recent meta-analysis found that there is a lower association between OSSN and HPV in countries closer to the equator.²⁶ This contradicts one of the proposed mechanisms of UVB-induced risk, which poses that UV exposure increases the activation of latent HPV.⁴ South Africa straddles the 30° latitude line and has a high prevalence of HIV. A study conducted in Cape Town (−34° latitude) found a prevalence of 66.7% for HPV in OSSN with a study population that had a prevalence of HIV of 88.9%.¹⁰ Our study in Johannesburg (−26° latitude) had a prevalence of 21.8% for HPV, with a prevalence of HIV of 81.3%. Future studies could investigate the association between HPV infection and latitude.

Tornesello et al.²¹ investigated the association between HPV in OSSN and the role that HIV plays in this association. They found that the rate of HPV infection decreased with increasing stage of histology.²¹ In mucosal tumours, the higher the histologic grade of a lesion, the more likely the presence of HPV.²⁷ We reviewed this in our study and did not find a statistically significant association with histological grade and HPV infection.

We found that trauma was more common in the control group. This is partly explained by the recruitment process for the controls, which included patients undergoing surgery following ocular trauma. This was the case for three of the five control participants with a history of ocular trauma.

This was a pilot study with a small sample size, which limited further statistical analysis with multivariate regression analysis. No HIV testing was routinely offered in the control group, which prevented comparison of this important risk factor between the two groups. As this is a cross-sectional study, we are not able to infer direct causality but rather just highlight an association. No in situ hybridisation or immunohistochemistry was performed, which limits the ability to comment on the direct pathogenesis of HPV on OSSN. The results have provided a springboard for further research in this area.

Our pilot study investigated the association of HPV with OSSN and found that 21.8% of OSSN cases had an HPV infection compared to no infections in the control group. This serves as a foundation for further research on the topic in South Africa.