Case–Control Study of Cutaneous Human Papillomaviruses in Squamous Cell Carcinoma of the Skin Free

Squamous cell carcinoma (SCC) of the skin is the second most frequently occurring cancer among Caucasians in the United States, and the incidence continues to increase each year (1). Established risk factors for SCC include UV radiation exposure, older age, light skin, and immunosuppression (2). Emerging evidence suggests that cutaneous human papillomavirus (HPV) infection may also be a risk factor for SCC (3–8). HPV types that infect cutaneous epithelia have been identified from genera alpha, beta, gamma, mu, and nu (9). Presence of antibodies against one or more of the genus-beta HPV types as a group has been associated with SCC in several case–control studies (3–8) and type-specific associations with SCC have also been observed with HPV 8 (5, 8, 10), 15 (11), 17 (11), and 38 (5, 11). Estimates of HPV DNA prevalence in SCC tissues from immunocompetent individuals range from 20% to 48% (10, 12–16). To our knowledge, only one study has reported findings on the relationship between HPV seroreactivity and SCC in conjunction with detection of HPV DNA in the skin cancer tumor itself, and in that study, DNA-positive SCC cases were more likely to be seropositive for any HPV type than DNA-negative SCC cases (3).

This is the first case–control study in a U.S. population to investigate the association between SCC and seroreactivity to cutaneous HPV types belonging to 5 different genera. In addition, serologic associations between beta HPV types and SCC were further stratified by the presence of DNA of these HPV types in the tumor tissues.

A clinic-based case–control study was conducted in Tampa, FL to investigate the association between cutaneous HPV infection and SCC of the skin. The study design and population have been previously described in detail (17). Briefly, histologically confirmed SCC cases were recruited from the University of South Florida (USF; Tampa, FL) Dermatology clinic (n = 191). Control subjects were recruited from the USF Family Medicine and Moffitt's Lifetime Cancer Screening and Prevention (LCS) clinics. All control participants included in the current study underwent a full-body skin cancer screening exam, were negative for current signs of skin cancer, and had no history of any type of cancer (n = 281). If a patient had a suspicious lesion detected during the skin screening exam that was later determined to be benign based on pathology review, the patient was also included as a control (n = 77). If a patient's screen-detected lesion was histologically confirmed to be an SCC then that patient was included as a case (n = 6). Study participants were ages 18 to 80 and were eligible to participate regardless of immune status. Nine SCC cases (4.7%) reported a history of organ transplantation.

Participants completed a self-administered questionnaire on skin cancer risk factors, and blood samples were obtained from 174 (90.6%) cases and 340 (95.0%) controls. With the exception of 2 non-White controls, the current analysis was restricted to White participants. The final sample size for the analysis of cutaneous HPV seroreactivity was 173 SCC cases and 300 controls. From patients with SCC undergoing surgical excision, a 3-mm punch of the residual SCC tumor was obtained and flash frozen in liquid nitrogen. Analyses were restricted to tumor specimens that tested positive for β-globin, corresponding to 180 SCC tumors from 159 individual patients, including 19 patients who contributed tissues from distinct, concurrent tumors. The final sample size for analyses including cutaneous HPV seroreactivity and DNA status consisted of 146 SCC cases and 300 controls. Written informed consent was provided by all study participants and all study procedures were approved by the Institutional Review Board at the USF.

Sera were tested for antibodies to the major capsid protein L1 of cutaneous HPV type(s) within genera alpha (2, 3, 7, 10, 27, 57, and 77); beta (5, 8, 9, 15, 17, 20, 23, 24, 36, 38, 49, 75, 76, 92, 96, and 107); gamma (4, 48, 50, 65, 88, 95, 101, and 103); mu (1); and nu (41). Sera were also tested for antibodies to the VP1 capsid protein of 2 human polyomaviruses, JC virus (JCV) and KI virus (KIV), to test the specificity of associations observed between cutaneous HPV and SCC.

The antibody detection method used is based on glutathione S-transferase (GST) capture ELISA (18, 19) in combination with fluorescent bead technology (Luminex; refs. 20, 21), as previously described. Individual cutoff values to define HPV type–specific seropositivity were applied as described previously to allow for the direct comparison of cutaneous HPV seroprevalence across studies that used the same assay (4, 22).

DNA extraction from frozen SCC tumor tissues was carried out using the Qiagen BioRobot EZ1 with the EZ1 DNA tissue kit according to the manufacturer's instructions (Qiagen). Briefly, frozen tissues were incubated in proteinase K and a buffer G2 (Qiagen) at 56°C until the tissue was completely lysed. To monitor the possible occurrence of cross-contamination between the different specimens during DNA extraction, tubes containing buffer only were also included.

HPV DNA was measured in all samples using an assay that detects 25 genus-beta HPV types (5, 8, 9, 12, 14, 15, 17, 19, 20, 21, 22, 23, 24, 25, 36, 37, 38, 47, 49, 75, 76, 80, 92, 93, and 96), based on the same techniques used to measure mucosal HPV types (23, 24). The assay combines 2 different techniques: multiplex PCR using HPV type–specific primer for amplification of viral DNA and array primer extension (APEX) for typing with 2 type-specific probes each.

Skin cancer risk factors were compared between SCC cases and controls using the χ² test, and independent associations between these factors and case–control status were estimated by calculating ORs and 95% confidence intervals (CI) using conditional logistic regression with adjustment for all other factors. The same approach was used to compare factors between controls who were seropositive for at least one cutaneous HPV type and controls who were seronegative for all HPV types. HPV type–specific seroprevalence was calculated as the proportion of patients who tested positive for antibodies to a given HPV type. Genus-specific HPV seroprevalence was calculated as the proportion of patients who tested seropositive for at least one HPV type in a given genus. Cutaneous HPV seropositivity to any type was defined as testing seropositive to at least one of the 33 cutaneous HPV types tested across the 5 genera. Logistic regression was used to calculate ORs and 95% CIs to estimate the associations between SCC and (i) type-specific cutaneous HPV seropositivity, (ii) genus-specific seropositivity, and (iii) the number of types in a given genus for which an individual tested seropositive. Tests for trend in risk for SCC associated with seropositivity for increasing number of cutaneous HPV types within a given genus were conducted by assigning ordinal values to each category and including the ordinal variable in the logistic regression model. The Bonferroni method was used to account for multiple comparisons, reducing the statistical significance level for SCC associated with genus-specific HPV seropositivity to P < 0.01 and type-specific HPV seropositivity to P < 0.002.

Tumor tissues obtained from SCC cases were classified as positive or negative for the presence of HPV DNA for types in genus-beta. The 19 cases who contributed more than one SCC tumor tissue were considered HPV DNA–positive if at least one of the tumor tissues provided tested positive for genus-beta HPV DNA. Logistic regression was then used to calculate the OR and 95% CI for the association between seropositivity to any genus-beta HPV type and SCC, stratified by the presence or absence of DNA to any genus-beta HPV type in the tumor tissue (i.e., genus-beta HPV–positive vs. genus-beta HPV–negative). This analysis was restricted to the genus-beta HPV types included in both the antibody and PCR assays (5, 8, 20, 24, 36, 9, 15, 17, 23, 38, 75, 76, 92, and 96). Type-specific concordance was calculated among the SCC cases as the proportion that tested seropositive for a given HPV type and had DNA in their tumor tissue corresponding to the same HPV type. Using logistic regression, additional stratified analyses were then conducted to compare genus-beta type–specific HPV seropositivity between controls and SCC cases that had DNA in their tumors corresponding to the same HPV type for which antibodies were detected.

All factors listed in Table 1 were considered as potential confounders. With the exception of age and sex, none of the factors listed in Table 1 altered the ORs and corresponding 95% CIs for the associations between SCC and cutaneous HPV infection (measured by serology or DNA) by more than 10%. Therefore, all associations were adjusted for age (as a continuous variable) and sex only. To rule out the possibility of residual confounding by age, an age-restricted analysis was conducted by restricting both cases and controls to those aged 40 to 69. In addition, sex-stratified ORs and 95% CIs were estimated for the associations between genus-specific HPV seropositivity and SCC, also restricted to cases and controls aged 40 to 49. Analyses were conducted with and without the 9 SCC cases who reported a history of organ transplantation with similar results, thus, all cases were included. Analyses were conducted using the SAS statistical software package (version 9.2; SAS Institute).

The distribution of demographic, lifestyle, and skin cancer risk factors among SCC cases and controls is presented in Table 1. Compared with controls, SCC cases were significantly more likely to be older, male, and less educated, and to have light eye and hair color, past occupational sunlight exposure, history of blistering sunburn, cutaneous sensitivity to sunlight exposure, poor tanning ability, and a history of alcohol consumption and smoking (Table 1). Demographic, lifestyle, and skin cancer risk factors were not associated with cutaneous HPV serostatus for any HPV type among the controls (Table 2).

Associations between genus-specific HPV seroreactivity and SCC are presented in Table 3. SCC was significantly associated with seropositivity to any genus-beta HPV type (OR, 1.93; 95% CI, 1.23–3.02), with increasing risk observed with an increase in the number of genus-beta HPV types for which an individual tested seropositive up to 8 types; seropositivity for >8 types showed no additional risk for SCC. Similar associations with SCC were not observed for seropositivity to HPV type(s) in genus alpha, gamma, mu, or nu or the polyomaviruses JCV and KIV (Table 3).

Analyses restricted to cases and controls ages 40 to 69 years yielded associations between cutaneous HPV seroreactivity and SCC of similar magnitude as those observed for the total study population. Specifically, SCC was significantly associated with seropositivity for at least one HPV type in genus-beta (OR, 2.05; 95% CI, 1.21–3.46) but not with genus-alpha (OR, 1.15; 95% CI, 0.69–1.89), gamma (OR, 1.25; 95% CI, 0.77–2.02), mu (OR, 1.20; 95% CI, 0.74–1.97), or nu (OR, 1.06; 95% CI, 0.49–2.27). In addition, age-restricted, sex-stratified analyses showed similar associations between genus-specific cutaneous HPV seropositivity and SCC among males (genus-alpha: OR, 1.04; 95% CI, 0.54–2.01; beta: OR, 2.15; 95% CI, 1.07–4.29; gamma: OR, 1.02; 95% CI, 0.54–1.91; mu: OR, 1.05; 95% CI, 0.56–1.97; nu: OR, 1.10; 95% CI, 0.43–2.77) and females (genus-alpha: OR, 1.35; 95% CI, 0.62–2.94; beta: OR, 2.17; 95% CI, 0.93–5.07; gamma: OR, 1.69; 95% CI, 0.80–3.60; mu: OR, 1.59; 95% CI, 0.72–3.51; nu: OR, 1.03; 95% CI, 0.27–3.98).

Seroprevalence was greatest for HPV type 4 in genus-gamma for both SCC cases (43.4%) and controls (34.3%); however, seropositivity to HPV 4 was not significantly associated with SCC (Table 4). HPV 10 was the single type from genus-alpha significantly associated with SCC (OR, 2.24; 95% CI, 1.04–4.85; Table 4). Within genus-beta, seroprevalence was greatest for HPV types 49, 8, and 17 among cases and types 49, 17, 15, and 9 among controls with significant associations observed with SCC for types 8 (OR, 1.80; 95% CI, 1.14–2.84) and 17 (OR, 1.59; 95% CI, 1.02–2.49; Table 4). None of the type-specific associations corresponded to a P value of less than 0.002, the significance level adjusted for multiple comparisons using the Bonferroni method. At the species level, SCC was significantly positively associated with seropositivity for HPV types in genus-beta species 1 (OR, 1.86; 95% CI, 1.22–2.85; P = 0.004). No significant associations were observed for genus-beta species 2 to 5 (Table 4) and no significant associations with SCC were observed for individual HPV types in genera gamma, mu, or nu (Table 4).

Ninety-six SCC cases (66%) were DNA positive for at least one genus-beta HPV type tested. Seropositivity to at least one of the genus-beta HPV types tested was associated with overall DNA positivity of genus-beta HPV types in SCC (OR, 2.81; 95% CI, 1.53–5.16; data not shown). Seropositivity to any genus-beta HPV type was not associated with SCC among those cases with tumors negative for HPV DNA (OR, 0.98; 95% CI, 0.53–1.84; data not shown). On the basis of these observations, a decision was made to investigate genus-beta type–specific correlations between seroantibodies and DNA presence in the tumor for the same HPV types. Among the 96 SCC cases with HPV DNA–positive tumors, 41.7% tested seropositive for antibodies to the same HPV type detected in the tumor tissue (data not shown). Concordance between the specific HPV type for which DNA was observed in the tumor and the HPV type for which circulating antibodies were observed ranged from 12.5% for type 9 to 54.5% for type 17 (Table 5). Compared with controls, SCC cases who were DNA positive for the following HPV types had significantly greater seroprevalence for the same HPV types: HPV 5 (OR, 3.48; 95% CI, 1.27–9.59), 24 (OR, 3.79; 95% CI, 1.24–11.5), and 17 (OR, 3.36; 95% CI, 1.29–8.72; Table 5). SCC cases with tumors that tested DNA negative for all genus-beta HPV types were significantly less likely to be seropositive for HPV types 20, 36, 9, 15, 38, and 75 than controls (Table 5).

Observations from the current case–control study support the association between cutaneous HPV seropositivity and SCC. Seropositivity to HPV 10 was the single type from genus-alpha associated with SCC. Seropositivity for cutaneous HPV types in genus-beta overall, and for types 8 and 17 specifically, was significantly associated with SCC in this clinic-based case–control study. In addition, correlations between seropositivity and DNA positivity for the same genus-beta HPV type were observed for HPV types 5, 24, and 17. SCC cases with tumors negative for any genus-beta HPV type were less likely to be seroprevalent for HPV types 20, 36, 9, 15, 38, and 75 than controls.

The positive association observed between SCC and seropositivity to HPV types in genus-beta is consistent with 2 (7, 8) of 5 case–control studies (3, 5, 7, 8, 16) including one study from the United States (7). Type-specific associations observed with SCC for HPV types 8 and 17 in genus-beta agree with findings from case–control studies in the United States (7), the Netherlands (5), and Australia (8) for HPV type 8 and with findings from case–control studies in the United States (7) and Italy (25) for HPV type 17. In contrast to previous case–control studies, SCC was not associated with seropositivity to HPV types 5 (6), 15 (7, 11), and 38 (5, 11), as well as with types 20, 24, 9, 49, 75, 76, 92, and 96 (7). Neither of 2 published cohort studies (4, 26) observed statistically significant increased risks of SCC associated with HPV seropositivity for types in genera alpha (4), beta (4, 26), gamma (4), mu (4), or nu (4).

A positive dose–response between seropositivity to increasing numbers of HPV types in genus-beta and SCC was observed in the current study population, consistent with 3 (7, 8, 27) of 4 (7, 8, 25, 27) previous case–control studies. However, in a recent prospective cohort study (26), an increased risk of SCC was not observed among individuals who were seropositive to at least one genus-beta HPV type or multiple types. Furthermore, neither of the 2 prospective studies reporting increased risks of SCC associated with baseline cutaneous HPV seropositivity observed statistically significant type-specific associations with any individual type from genus-beta (4, 26).

The single type-specific association with SCC observed in this study for a cutaneous HPV type in a genus other than beta was with HPV type 10 in genus-alpha, consistent with the 2 studies in the literature previously reporting on HPV type–specific associations outside of genus-beta (4, 11). HPV 10 has been associated with benign skin lesions (28) but displays weak transforming activities by in vitro experimental models (29), thus, its direct involvement in skin carcinogenesis remains unclear.

Among SCC cases from Sweden and Austria (3), 58% tested DNA positive for at least 1 HPV type in genus-beta compared with 66% in SCC cases from the current study. In addition, a low overall concordance between HPV DNA and seropositivity for the same HPV type (19%) was observed (3) compared with the current study with an overall type-specific concordance of 41.7%. In addition, among participants in the current study, individual type-specific concordance was greater for genus-beta HPV types 5, 8, and 15 but lower for types 9 and 24 as compared with participants from Sweden/Austria (3). The observed difference in overall concordance between HPV types identified from the tumor and circulating antibodies may be explained by a difference in the number of genus-beta HPV types examined, (8 types in the Swedish/Austrian population compared with 13 types in the current study population). Individual type-specific variations in concordance may be explained by differences in the laboratory techniques used to test for the presence of HPV DNA in the tissue specimens. In addition, Andersson and colleagues obtained biopsies from the malignant lesion and adjacent healthy skin and subsequently classified SCC cases as DNA positive if either of the biopsies tested positive for beta HPV DNA (3) in contrast to the current study that defined DNA positivity based on detection in the SCC tumor only.

It was unexpected to observe statistically significant inverse associations between seropositivity and SCC among HPV DNA–negative cases. One explanation may simply be that SCC cases with tumors negative for HPV DNA and subsequently low antibody seroprevalence compared with controls may have a different risk profile for disease. In terms of previous findings, these data may explain the low-risk estimates observed in previous studies for the associations between cutaneous HPV seropositivity and SCC when HPV DNA status of the tumor has not been accounted for. In addition, the observations from the current study may underlie the discrepant findings presented across case–control studies. Further research is needed to delineate the role of cutaneous HPV antibodies as either a marker of prevalent and cleared infections or a marker of current active infections and their subsequent association with SCC risk.

The current study provides evidence for the association between genus-beta HPV and SCC, however, the exact mechanism by which cutaneous HPV is associated with SCC remains unclear. It has been hypothesized that the effects of genus-beta HPV on UV-induced DNA damage and apoptosis could lead to the accumulation of mutations which predispose to SCC formation (30). Evidence from in vitro and in vivo studies have shown that E6 and E7 proteins from certain cutaneous HPV types are capable of inhibiting UV-induced cell-cycle check points, apoptosis, and DNA repair machinery (31–36). Observations from epidemiologic studies have shown statistically significant interactions between sun-related factors and genus-beta HPV seropositivity in relation to SCC (27, 37, 38). In addition, HPV 38 E6 and E7 proteins have displayed transforming properties in primary human keratinocyte cells sufficient to deregulate cell-cycle control and growth arrest (39). Taken together, these findings support a role for cutaneous HPV as a cofactor in SCC carcinogenesis, in contrast to the direct role of mucosal HPV infections in the carcinogenesis of cervical cancer.

Clinic based case–control study populations are often not representative of the general population. However, cases and controls were recruited from clinics that serve the same underlying populations, thus preserving internal validity. Because a considerable proportion of the skin cancer screening patients were confirmed to be free of skin cancer only after completion of the follow-up visit with a dermatologist, there was at time lag from study consent to verification of case–control status. As such, matching controls to cases was logistically challenging and significant case–control differences in age and sex resulted. However, all analyses were adjusted for age and sex. Furthermore, the analysis restricted to those aged 40 to 69 yielded similar results as those obtained from the entire study population. In addition, similar associations between genus-specific HPV seropositivity and SCC were observed between males and females in the restricted age range. Therefore, the observed associations between cutaneous HPV seropositivity and SCC are not likely due to residual confounding by age and/or sex. Adjustment for multiple comparisons reduced the statistical significance of most of the observed associations, thus, chance cannot be ruled out as an explanation for these findings. However, multiple comparisons is a methodologic challenge faced by all epidemiologic studies of cutaneous HPV infections, given the sheer number of types potentially implicated in SCC. Benign lesions were not obtained for control subjects that completed dermatologic follow-up, as such, for comparison purposes it was not possible to test the benign lesions for the presence of HPV DNA. If in fact cutaneous HPV is associated with benign skin lesions then the observed results would be attenuated given that a subset of the controls were diagnosed with such lesions.

Several study strengths should also be noted. The current study is the first to present cutaneous HPV genus and type-specific associations outside of genus-alpha and -beta and to incorporate comparisons with genus-beta HPV DNA reactivity in SCC tumor tissues in a U.S. population. Only one study in the published literature has presented similar results among residents of Sweden and Austria (3). This is a major strength of the current study compared with previously published case–control studies of cutaneous HPV serology in SCC, allowing the unique opportunity to investigate specificity of serologic response and the presence of DNA for the same HPV type. In addition, the laboratory used to test for cutaneous HPV seropositivity has been used in most of the seroepidemiologic studies of cutaneous HPV published to date (4, 6, 7, 11, 22, 25, 40, 41), allowing for direct comparison across studies. Finally, the full-body skin examination conducted on all control subjects is another study strength, as it eliminated misclassification of SCC case–control status that can result by inclusion of apparently healthy individuals who in fact have a prevalent, undiagnosed SCC. Of note, 6 of the SCC cases included in the current study were screen-detected.

The current study confirms previous findings supporting the association between genus-beta seropositivity and SCC. In addition, genus-beta type–specific associations with SCC are supported by the observed correlation between HPV seropositivity and tumor DNA positivity for specific types in genus-beta. However, additional longitudinal studies need to be conducted to better understand the direction of the association between cutaneous HPV infection and SCC to rule out the possibility of reverse causality.

T. Waterboer is an employee for Boehringer Ingelheim. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A.R. Giuliano, M. Pawlita, D.E. Rollison

Development of methodology: T. Waterboer, J.L. Messina, K.M. Michael, M. Pawlita, D.E. Rollison

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.R. Iannacone, T. Waterboer, J.L. Messina, N.A. Fenske, B.S. Cherpelis, V.K. Sondak, R.G. Roetzheim, K.M. Michael, M. Pawlita, D.E. Rollison

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.R. Iannacone, A.R. Giuliano, V.K. Sondak, M. Tommasino, M. Pawlita, D.E. Rollison

Writing, review, and/or revision of the manuscript: M.R. Iannacone, T. Waterboer, A.R. Giuliano, J.L. Messina, N.A. Fenske, B.S. Cherpelis, V.K. Sondak, R.G. Roetzheim, K.M. Michael, M. Pawlita, D.E. Rollison

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.R. Iannacone, D.E. Rollison

Study supervision: V.K. Sondak, D.E. Rollison

Development of methods for the detection of papillomaviruses (ICB, IARC): T. Gheit

The authors thank the supporting staff at USF and LCS for their assistance with patient recruitment, especially Kristen Jonathan, Jill Weber, and Carolyn Gerow.

This work was supported by a James and Esther King New Investigator grant awarded to DER (06NIR-08) by the Florida Department of Health (FL DOH). This research was also supported in part by the Miles for Moffitt Foundation Funds.

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.