High Ambient Solar UV Correlates with Greater Beta HPV Seropositivity in New South Wales, Australia

Human papillomavirus (HPV) infection is highly prevalent worldwide and may contribute, among other things, to the pathogenesis of cutaneous squamous cell carcinoma (SCC; ref. 1), a cancer for which sun exposure is the main cause (2).

HPVs are nonenveloped, double-stranded DNA viruses, which infect skin and mucosa, where they can persist asymptomatically or cause neoplasia. More than 200 different HPV types have been fully characterized to date (3, 4). Genus beta contains about 50 HPV types in five species (beta 1–5; ref. 4) and is the most detected genus in cancerous, precancerous, and normal keratinocytes (5).

Cutaneous HPVs are of interest because of a postulated joint role of UV and cutaneous HPVs in causing SCC, based on evidence from experimental systems (1–3, 6–10), although direct evidence in humans is lacking (11). Very little is known, however, about the relationship between UV exposure and the seroprevalence of cutaneous HPV types (11). To address this deficiency, we report here on the relationship of UV exposure with seroprevalence of cutaneous HPVs in a sample of the population of New South Wales, Australia. The people in this sample were control participants, free of skin cancer, in a case–control study of cutaneous keratinocyte cancers [SCC and basal cell carcinoma (BCC)] nested in the New South Wales 45 and Up Study (12).

We also report seroprevalence of HPVs in genera beta and gamma, the major cutaneous genera, and, for comparison and as positive controls, several other antigens: HPVs of genera mu and nu, which are associated with warts (13), genus alpha HPV16, which mainly infects genital mucosae, the human polyomaviruses PyV7 and MCV, the latter playing a role in causing Merkel cell carcinoma (14), and p53.

Study subjects were participants in the Skin Health Study (SHS), a population-based study of people with a first BCC or SCC diagnosed in New South Wales in 2005 to 2009 (cases) and no history of melanoma, and controls, a random sample of people with no history of any skin cancer who were frequency-matched to the age (45–84 years) and sex distributions of the cases (12). The SHS was nested in a larger NSW cohort, the Sax Institute's 45 and Up Study, a population-based prospective cohort study comprising 267,153 middle aged and older participants randomly selected from the Department of Human Services (formerly Medicare Australia) enrollment database. Participants joined the study by completing a postal questionnaire in 2006 to 2009; all gave their consent to be approached for future research. Details of the 45 and Up Study including the baseline questionnaire and sampling frame are described elsewhere (15). We extracted participants' histories of cutaneous keratinocyte cancer from the 45 and Up Study baseline self-completed questionnaire and confirmed a history of BCC or SCC in medical records (see ref. 12).

Ambient UV at residential locations is a surrogate for personal UV exposure (16). Adult UV exposure is largely from ambient UV, individually determined by time spent outdoors, and totals around 2% to 10% of annual ambient UV, a percentage that varies with latitude (17–19). SHS participants recorded locations for their personal residence history in a calendar format against each year of age and calendar year from birth to 74 years of age, from which we constructed individual life histories of annual ambient UV irradiance (UVR).

Worldwide mean daily ambient UVR levels were downloaded from the Total Ozone Mapping Spectrometer (TOMS) database (https://science.nasa.gov/missions/toms/; accessed June 2011) maintained by the National Aeronautics and Space Administration (NASA), which provides estimates of daily noon-time ambient UV exposure levels on a 1-degree latitude by 1-degree longitude grid. Each recorded residential location was geocoded, assigned to a point on the location grid, and the relevant UV value extracted. Daily ambient erythemal UVR was calculated (see ref. 12); this estimate, expressed in mJ/cm², is the basis of “ambient UV” in this report.

We estimated annual average daily ambient solar UV (mJ/cm) in the birth year and for the 10 years before completing the questionnaire, in age intervals and over the lifetime to the date of SHS data collection (12). Totals for all except birth year were calculated as the estimated individual annual average ambient UV in each residence location within each exposure period multiplied by duration of residence and summed over relevant residence locations; for analysis, we used tertiles of ambient UV values. Residents of the Sydney metropolitan region (∼65% of the 2016 New South Wales population; ambient UV 49.99 mJ/cm², latitude 33°S) are mainly in the intermediate tertile.

A brief job history assessed total years of outdoor occupational exposure by asking for age at first regular full or part-time job since age 15 years, at retirement from a regular job, when first working in “any job for 1 year or more in which you usually worked outdoors for more than 1 hour between 9 am and 5 pm,” and how many years in total for all such jobs.

The 45 and Up Study and SHS self-completed questionnaires included items to collect data on these demographic, lifestyle or personal and sunlight-related characteristics (see ref. 12): age, sex, country of birth, age at arrival in Australia, household income, education, a census-based index of relative socioeconomic disadvantage, retirement status, skin color, ability to tan, years of outdoor work, times in the last week spent walking, in moderate physical activity, and in vigorous physical activity, mostly done outdoors in Australia (20). The SHS collected additional relevant data: hair color, eye color, density of freckling, density of moles, solar keratoses (as “told by a doctor”), skin screening, immunosuppressive therapy, lifetime number of sex partners (LSP), smoking, and drinking alcohol.

We collected venous blood samples (6 mL) and separated plasma, white blood cells, and red blood cells by centrifugation at 2,500 rpm for 10 minutes at 4°C. One aliquot of ACD plasma was shipped on dry ice to the German Cancer Research Center (DKFZ) for serologic analysis. We analyzed plasma samples for antibodies to the major capsid protein L1 of 30 genus beta and gamma HPV types: 23 beta HPVs (types 5, 8, 9, 12, 14, 15, 17, 20–24, 36, 38, 47, 49, 75, 76, 80, 92, 93, 96, 107); 7 gamma HPVs (4, 48, 50, 60, 65, 88, 95). Additional antigens assayed were mu HPV1 and 63, nu HPV41, alpha HPV16 L1, and the human polyomaviruses HPyV7 and MCV VP1 antigens, and p53 (21). All antigens included were tested in a 1:100 serum dilution and were expressed as GST fusion proteins (11, 22). In addition, GST was included as an antigen for background determination.

Seroreactivity against all antigens was measured by fluorescent bead-based multiplex serology (23). In brief, each antigen was loaded onto one distinct polystyrene bead set (COOH-beads xMAP Technology Microspheres; Luminex Corp.). Loaded bead sets were mixed and simultaneously presented to primary serum antibodies. Immunocomplexes formed were detected using a biotinylated goat-α-human IgG/IgM/IgA secondary antibody and streptavidin-R-phycoerythrin as reporter dye. Median fluorescence intensities (MFI) of ≥100 beads of the same bead set were measured by a Luminex flow cytometer (Luminex 200). MFI values reflect antibody affinity, titer, and viral load (11).

Serologic analysis was performed blind to case–control status and all sera were tested over 3 consecutive days. For quality control, a subset of study sera supplemented with sera of known reactivity was tested daily and additionally, a standard serum with known reactivity pattern to a subset of the antigens was analyzed on each 96-well plate. Loading of the different beads with antigens was monitored by a mAb against a peptide fused to the C-terminus of all expressed antigens (22).

Cutoff points for seropositivity for all antigens were selected by visual inspection of the distributions of HPV, HPyV, MCV, and p53 MFI values in participants' serum samples (see refs. 13, 14, 24, 25). For the beta, gamma, mu and nu HPVs, and alpha HPV16, the cutoff point for antibody positivity was MFI >300; for HPyV7 and MCV, it was MFI >400 and for p53 it was >200.

Overall and sex- and age-specific prevalence of antibody seropositivity was calculated for each of the beta and gamma HPV types: for each genus (beta, gamma), each species (beta-1 to -3; gamma-1), and by individual types, including HPV92 (beta-4), HPV96 (beta-5), HPV48 (gamma-2), HPV50 (gamma-3), HPV60 (gamma-4), and HPV88 (gamma-5), and for the other antigens.

We tested the associations of seroprevalence of beta and gamma HPV types and other antigens in categories of solar UV exposure and sun sensitivity variables by estimating prevalence ratios (PR) and their 95% confidence intervals (CI) adjusted for sex and age (45–54, 55–64, 65–84 years). In addition, we examined the associations between these antigens and demographic and lifestyle characteristics. We used binomial generalized linear models (GENLIN in SPSS) with a log-link function and including as covariates in each model age (45–54, 55–64, 65–84 years), sex, and birthplace (Australia or elsewhere). The P values calculated were two-sided. The Benjamini–Hochberg (BH; ref. 26) method was used to account for multiple testing, that is, for guidance to significance of P values in Table 1 (sex differences in seroprevalence), Supplementary Tables S3 (PRs and CIs for 37 beta and gamma HPV types and other antigens in categories of solar UV exposure) and S4 (hair color). Estimates of PRs for UV exposure or hair color by antigen were considered to be statistically significant if the BH-adjusted P value was ≤0.01. All analyses were performed in SPSS versions 19 or 24.

All participants gave written informed consent. The studies were conducted in accordance with the ethical guidelines of the National Health and Medical Research Council of Australia; the University of New South Wales Human Research Ethics Committee approved the 45 and Up Study and the Cancer Council New South Wales Ethics Committee approved the Skin Health Study.

The 1,161 participants were ages 45 to 54 (30%), 55 to 64 (34%), and 65 to 84 years (36%), included more women than men (57% female, 43% male), were mostly born in Australia (75%) and were current alcohol drinkers (69%), infrequently current smokers (4%), and half (51.4%) were educated to degree or diploma level (Supplementary Table S1).

Venous blood samples were evaluated in the serologic analyses for 1,171 of 1,174 participants, of whom 1,161 (99%) had valid results, almost all of whom (1,087 of 1,161, 94%) had a positive serology result for one and up to 32 of the 33 cutaneous HPV types (genera beta, gamma, mu, or nu). Most (1,027 of 1,161, 88%) were seropositive for at least one of the 23 genus beta or 7 genus gamma HPVs, 81% were positive for 1 to 19 types and 7% for 20 to 29 types, 55% were seropositive to at least one beta and one gamma type, and 34% were positive to a single beta or gamma HPV type. For additional data on seropositivity patterns and comparisons with other reports based on similar serologic methods, see Supplementary Table S2 and associated text.

Seroprevalence was 73.8% for any beta and 70% for any gamma HPV type and generally lower in women than men with PRs for women around 0.9 (Table 1). Women were more often positive to less than 5 types (1–4 types 54% female, 48% male) and men were more often positive to 5 or more types (5–29 types 34% female, 42% male; P < 0.05); women were less often seropositive than men to both beta and gamma HPVs (51% female, 60% male; P = 0.03). PRs for women were 0.9 for genus gamma, around 0.7–0.8 for beta-1 HPV5, beta-2 HPVs 23, 38 and 80, and gamma HPVs 65 and 48, and 0.43 for beta-2 HPV22 (all P_adj ≤ 0.01). All individual beta and gamma HPV types except gamma-5 HPV88 (4% in each of male and female) had PRs <1.0 for women relative to men, although with P_adj > 0.01.

There was no significant association between increasing age within the 45 to 84 years age range of this study and seropositivity for genus beta or genus gamma, by species or individual type (Supplementary Table S1).

Seroprevalence for other antigens was high for the polyomaviruses HPyV7 and MCV (74%–75%) and any genus mu HPV (54%) and lower in women than men for HPyV7 (PR = 0.91; 95% CI, 0.86–0.97), but similar in both sexes for mu HPV1, the human polyomavirus MCV and p53 autoantibodies (Table 1). Women were more often seropositive than men for genus alpha HPV16 (PR = 1.67; 95% CI, 1.20–2.32), which is well known (27). There was no strong evidence of any age trend, but seroprevalence at 65 to 84 years was higher for HPyV7 (PR = 1.18; 95% CI, 1.09–1.28) and at 55 to 65 years for p53 (PR = 1.33; 95% CI, 1.0–1.75) and alpha HPV16 (PR = 1.45; 95% CI, 1.0–2.11).

We present the main results below. Given the paucity of present knowledge in this area, we also present, in Supplementary Table S3, all PRs, 95% CIs, and P values for the associations of UV exposure variables close to the time of enrollment in the SHS (ambient UV in the 10 years before enrollment, physical activities in the past week) for the 30 individual beta and gamma serotypes and 7 other antigens. Likewise, Supplementary Table S4 displays associations for hair color.

Higher ambient UV in the 10 years before SHS enrollment was associated with elevated seropositivity for genus beta (PR = 1.17; 95% CI, 1.07–1.28) and for any beta-1, beta-2, and beta-3 (Table 2). Four individual beta-2 HPV types (HPVs 9, 17, 38, 107) had PRs between 1.4 and 1.7 for higher ambient UV in this 10-year interval, and beta-1 HPV36, beta-4 HPV92, and beta-5 HPV96 had PRs 1.4 to 1.5 (all P_adj < 0.01; Supplementary Table S3). There were no such associations for genus gamma. There were also no such associations for ambient UV in the birth year, ambient UV over the lifetime, or in people born outside Australia relative to those born in Australia for genus beta or genus gamma (Table 2). The associations with ambient UV were similar in Australian born participants (N = 877) and participants born elsewhere (N = 276; genus beta: for overseas born PR = 1.21; 95% CI, 1.01–1.46 intermediate; PR = 1.19; 95% CI, 0.97–1.45 high UV in the prior 10 years), although for the latter the highest PR was commonly in the intermediate UV tertile (e.g., beta-3: overseas born PR = 1.61, 95% CI; 1.07–2.44 for UV in the prior 10 years), not the highest (overseas born PR = 1.15; 95% CI, 0.71–1.88). There was no material difference between the sexes in analyses that compared seropositivity prevalence ratios for genus beta, any beta-1, beta-2, and beta-3 in women and men for the sun exposure variables of Table 2 (highest UV level for genus beta: PR for men = 1.13; 95% CI, 0.99–1.29; for women PR = 1.22; 95% CI, 1.08–1.38).

We explored issues of dose response for ambient UV in the 10 years before SHS enrollment. Higher ambient UV in the prior 10 years was associated with a greater number of genus beta, but not genus gamma, HPV types. PRs were increased for 10 to 22 beta HPV types relative to none for each of intermediate (PR = 1.52; 95% CI, 1.21–1.91) and high (PR = 1.50; 95% CI, 1.22–1.84) ambient UV (Table 3). There were similarly elevated associations for beta-1, beta-2, and beta-3 species for which PRs were between 1.2 and 1.4 for higher counts of positive HPV types.

The seroprevalence for self-reported solar keratoses and years of outdoor work showed no evidence of an association at the genus or species level (Table 2). Physical activity categorized as high frequency in the week before completion of the 45 and Up Study baseline questionnaire was associated with higher seroprevalence for the individual beta-1 HPVs 21 and 24 (PRs = 1.6–1.7; P adjusted < 0.01; Supplementary Table S3), but not at the genus or species level (Table 2).

The PR for red hair was high for beta-3 HPV75: PR = 2.26 (95% CI, 1.28–3.97) with P_adj < 0.01 for hair color (Supplementary Table S4). A number of beta and gamma HPV types had PRs above unity for red hair and also for light brown hair, relative to black or dark brown hair but all P_adj values were >0.01. Our results for hair color otherwise were null.

No other personal or lifestyle characteristics measured had a strong or consistent association with seroprevalence of beta or gamma HPV types in this study (Supplementary Table S1).

Alpha HPV16 was more common in participants reporting two or more lifetime sex partners (PR = 1.62; 95% CI, 1.11–2.35; P = 0.01), with similarly increased PRs in both sexes. There was no strong or consistent association of any other antigens with ambient UV, physical activity, or other personal characteristics measured.

Neither genus, individual HPV type, nor other antigen had a potentially significant association with current or former smoking.

We examined the prevalence of serum antibodies against 30 beta and gamma HPVs and 7 other antigens in a large sample of Australian adults. HPV seropositivity was high: 74% for any beta, 70% for any gamma, 88% for any beta or gamma HPV type. We found strong evidence that UV exposure, as measured by ambient UV in the 10 years before enrollment, is associated with an increased beta HPV seroprevalence. There was no evidence of an association between UV exposure and gamma HPV types.

We discuss below potentially informative associations of multiple individual beta and gamma serotypes with sun-related variables, partly in acknowledgment of the uncertainty about whether the diversity of HPV types included in the genera beta and gamma share similar biological and pathologic characteristics (13, 28). The papillomavirus classification system relies heavily on sequence data (28). Our observational data, sourced from controls in a case–control study who were free of a history of skin cancer, could mean selective factors are operating. PRs, though, are based on internal comparisons and remain valid within the source population. Given the almost complete lack of evidence on UV exposure and cutaneous HPV seropositivity in populations, the detailed information we provide may assist in formulating new research hypotheses in this area.

Only three previous studies of cutaneous HPV seroprevalence have used multiplex serology, examined individual characteristics as possible determinants of seropositivity, and reported ORs; two were in Australia and are referred to here as Queensland I (11) and Queensland II (29). A third, in Florida (30), reported “Florida residency” as the one measure of sun exposure and, additionally, included people with past skin cancers and, probably, current skin cancers given their recruitment through a skin cancer screening clinic (30). Although Queensland I reported a positive association for “weekend UV index hours” with beta but not with gamma HPV types in Australian participants, the authors concluded that there was no “strong role for UV exposure in the development of HPV antibodies” (11). Our finding that seroprevalence is increased with greater residential UV in the past 10 years for beta HPV types is consistent with the Queensland I result. Queensland II did not include a sun exposure variable (29) and the Florida study reported no association between one or more of 17 beta HPVs and Florida residency (30). The increased seroprevalence in our study for ambient UV in the past 10 years for genus beta, including the beta-1 and beta-2 species, and for each of ambient UV in the past 10 years and more frequent physical activity in the past week for the two beta-1 HPVs 21 and 24 favors an association between sunlight exposure and cutaneous beta HPV prevalence, more strongly for more recent UV exposure. Genus gamma HPV were generally not associated with sun exposure.

The evidence in our study of a positive dose response for beta HPVs, that seropositivity to more beta HPV types is associated with greater residential UV in the past 10 years, is concordant with Waterboer and colleagues' finding of increasing seropositivity to multiple beta, but not gamma, HPV types with increasing country-specific UV intensity (11). Our observation supports continuing exploration of the correlates of multiple beta HPV types in immunocompetent populations and the possible role of multiple beta seropositives in keratinocyte skin cancers (31–33).

Some current interest in understanding the role of UV exposure in the natural history of HPV infection relates to the proposal, based on animal experimental evidence, that cutaneous beta HPV infection and UV radiation cooperate in SCC development (1, 3, 9, 10). While the association between sun exposure and beta HPV seropositivity we have observed is consistent with this hypothesis, being based exclusively on people without a history of skin cancer, it cannot address it directly.

Seroprevalence for beta-3 HPV75 was elevated in people with red hair in this study (P_adj < 0.01), and an increased OR was reported with red hair for grouped beta and grouped gamma HPVs in Queensland I, although nonsignificant (11). Results in Queensland II are unclear, probably null (29). None of the 7 other antigens we examined was positively associated with red hair. The elevated seroprevalence for light brown hair for beta-3 HPV75 could not be directly compared with Queensland I [(dark) blond; ref. 11] or Queensland II (blonde/light brown) colors (29).

We found no relationship of any beta or gamma HPV type with lifestyle characteristics other than UV exposure, including smoking, a finding consistent with Queensland I (11) and possibly Queensland II (29).

The lower seroprevalence, around 10% lower, for almost all beta and gamma HPV types in women than in men in our study is consistent with others (11, 29, 32). The Queensland studies reported much lower seroprevalences in women, for example, any beta HPV 36% (11) and 58% (29), than the 72% we found. The relationship with age has been unclear. Our finding that seroprevalence for beta or gamma genera was relatively constant with increasing age after 45 years, the youngest in SHS, supports other studies (11, 32, 34). Increases in seroprevalence for these HPV types occur mainly before age 45 (13). Michael and colleagues (13) speculate that high seroprevalence at older ages might be due to an environmental factor like sun exposure influencing the extent of HPV replication or the function of the immune system (13).

Ours is the only purpose-designed seroprevalence study of primarily cutaneous HPVs and their correlates. Major strengths are the base in one large, population-based study and that serology was done in one laboratory preeminent in HPV research and had comprehensive coverage of 30 cutaneous beta and gamma HPV types and measurement of 7 other antigens. The specificity of our observations for the cutaneous HPVs is underscored by the coherence of our findings for other antigens with reports by others: the higher alpha HPV16 seroprevalence in women and for lifetime number of sex partners (LSP), which is in contrast with the cutaneous HPVs (lower seroprevalence for women, lack of association with LSP); and the high seroprevalence for HPV4, the main cause of cutaneous warts.

We covered a broad range of personal and sun-sensitivity characteristics and our UV exposure measurements were more detailed and individual than in previous studies (11, 29). Although we adjusted for multiple possible confounding factors in the relationship of HPV serotypes to UV exposure, other, as yet unidentified, confounding variables might explain the weak associations. Such unknown confounders, however, would need to be biologically and quantitatively plausible. Although we cannot exclude antibody cross-reactivity between closely related HPV types, our substantive conclusions are at HPV genus level and not at the individual HPV type level. The fine-grained approach of our study has clearly shown that solar UV exposure is associated with an increased seroprevalence of beta, but probably not gamma, HPV types. The weakness of the associations we observed, and the number examined, however, should be borne in mind when interpreting these results, and our findings replicated, if replicable, before they are considered to be at all certain.

We found a weak association for genus beta, for which there was strong statistical evidence, that seroprevalence of cutaneous beta HPVs was positively related to exposure to solar UV in the 10 years before its measurement, a finding that may progress our understanding of the relationship between UV exposure and the seroprevalence of cutaneous HPV types. Associations of sun exposure with genus gamma were null. Further studies could consider examining the time course of HPV infections in populations, specifically whether serologic responses relate to past or present infection, and whether UV plays a role in sustaining infection with cutaneous beta HPVs. Understanding these issues will provide important input to the model of cooperation between cutaneous beta HPV infection and UV radiation in the development of cutaneous SCC, if such cooperation exists.

In addition to the above, our study corroborates evidence for the identities of high and low seroprevalence beta and gamma HPV types and highlights the similarity of HPV seroprevalence patterns, when based on similar serology analysis, in several different populations.

No potential conflicts of interest were disclosed.

Conception and design: A. Kricker, M.F. Weber, E. Banks, M. Pawlita, F. Sitas, C.H. van Kemenade, B.K. Armstrong, T. Waterboer

Development of methodology: A. Kricker, M. Pawlita, F. Sitas, V.S. Hodgkinson, C.H. van Kemenade, B.K. Armstrong, T. Waterboer

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.F. Weber, N. Brenner, M. Pawlita, V.S. Hodgkinson, C.H. van Kemenade, T. Waterboer

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Kricker, E. Banks, M. Pawlita, F. Sitas, B. Rahman, B.K. Armstrong, T. Waterboer

Writing, review, and/or revision of the manuscript: A. Kricker, M.F. Weber, N. Brenner, E. Banks, M. Pawlita, F. Sitas, V.S. Hodgkinson, B. Rahman, B.K. Armstrong, T. Waterboer

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V.S. Hodgkinson, C.H. van Kemenade

Study supervision: A. Kricker, M.F. Weber, F. Sitas, B.K. Armstrong

Other (Secured funding for the study): F. Sitas

This work was supported in part by the National Health and Medical Research Council of Australia (NHMRC) grant no. 550001 from 2009 to 2011. E. Banks is supported by an NHMRC Principal Research Fellowship (#1136128). This research was completed using data collected through the 45 and Up Study (www.saxinstitute.org.au). The 45 and Up Study is managed by the Sax Institute in collaboration with major partner Cancer Council NSW, and partners: the National Heart Foundation of Australia (NSW Division); NSW Ministry of Health; NSW Government Family & Community Services – Ageing, Carers and the Disability Council NSW; and the Australian Red Cross Blood Service. The authors thank the staff of the Skin Health Study at the Cancer Council NSW and the many thousands of people participating in the 45 and Up Study.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.