Seroreactivity to Cutaneous Human Papillomaviruses among Patients with Nonmelanoma Skin Cancer or Benign Skin Lesions Free

Cutaneous human papillomaviruses (HPV) are common in nonmelanoma skin cancers, benign skin lesions, and healthy skin. Increased seroprevalences for cutaneous HPV among nonmelanoma skin cancer patients have been described. To determine whether antibodies to cutaneous HPV are related to presence of the virus and/or to skin disease, we collected serum and biopsies from both lesions and healthy skin from 434 nonimmunosuppressed patients (72 squamous cell carcinomas, 160 basal cell carcinomas, 81 actinic keratoses, and 121 benign lesions). Biopsies were analyzed for HPV DNA by PCR, cloning, and sequencing. Serum antibodies to the major capsid protein L1 of HPV 1, 5, 6, 8, 9, 10, 15, 16, 20, 24, 32, 36, 38, and 57 as well as to the oncoproteins E6 and E7 of HPV 8 and 38 were detected using a multiplexed fluorescent bead-based assay. Type-specific seroprevalence among patients with the same type of HPV DNA (sensitivity of serology) varied from 0% to at most 28%. Presence of HPV DNA and antibodies to the same HPV type was not significantly correlated. However, seropositivity to any HPV type was significantly more common among patients positive for HPV DNA of any HPV type (odds ratio, 1.90; 95% confidence interval, 1.55-2.34). Seroprevalences were similar among the different patient groups but was, for most HPV types, somewhat higher among squamous cell carcinoma patients than among basal cell carcinoma patients (P < 0.01). In conclusion, additional studies are required to clarify the biological meaning of seropositivity as a marker of cutaneous HPV infection and skin disease. (Cancer Epidemiol Biomarkers Prev 2008;17(1):189–95)

Human papillomaviruses (HPV) are associated with proliferative lesions of the squamous epithelium, such as skin warts (HPV 1, 2, 27, and 57), laryngeal papillomas and genital condylomas (HPV 6 and 11), and cervical carcinomas (HPV 16, 18, etc.; ref. 1). The HPV with cutaneous tropism is found in the β-papillomavirus genus (earlier designated as epidermodysplasia verruciformis types), which is divided into five distinct species (β1: HPV 5, 8, 12, 14, 19, 20, 21, 24, 25, 36, 47, and 93; β2: HPV 9, 15, 17, 22, 23, 37, 38, and 80; β3: HPV 49, 75, and 76; β4: HPV 92; and β5: HPV 96); the γ-papillomavirus genus (HPV 4, 48, 50, 60, 65, 88, and 95); the μ-papillomavirus genus (HPV 1 and 63); and the υ-papillomavirus genus (HPV 41; ref. 2). The majority of viruses in the α-papillomavirus genus are mucosal HPV types, but this genus also contains cutaneous HPV types in species α2 (HPV 3, 10, 28, 29, 77, 78, and 94), α4 (HPV 2, 27, and 57), and α8 (HPV 7, 40, 43, and c91; ref. 2).

The β-papillomaviruses HPV 5 and 8 associate with squamous cell carcinoma (SCC) on sun-exposed sites of individuals with the rare diagnosis epidermodysplasia verruciformis, characterized by disseminated wart-like or pityriasis versicolor–like lesions (3). The HPV types of the β-papillomavirus group are not restricted to epidermodysplasia verruciformis patients and have frequently been detected in premalignant lesions and skin tumors from immunosuppressed renal transplant recipients, who also commonly develop warts and cutaneous cancers on sun-exposed sites (4-9). Also in the general population, these HPV types are occasionally detected in various skin tumors (10-15). However, β-papillomaviruses are also commonly found on the healthy human skin (9, 15-21).

Increased seroprevalences of HPV 8 have been reported among immunocompetent patients with SCC compared with patients with basaliomas or healthy controls (22-24). Antibodies to any of HPV 5, 8, 15, 20, 24, or 38 were more common among patients with a history of SCC (19.9%) than among controls (12.5%; P < 0.05; ref. 25). However, antibodies were more frequent among patients with fair skin; in a multivariate model, adjusting for skin type, antibodies did not associate with SCC (25). A recent study found seropositivity to these cutaneous types to be more prevalent in patients with SCC (37%) than in controls (13%), with HPV 8 antibodies strongly associating with SCC [odds ratio (OR), 9.3; 95% confidence interval (95% CI), 1.9-45.6; ref. 26]. Antibodies against any of the β-papillomaviruses HPV 5, 8, 9, 15, 20, 24, 36, and 38 were more common in SCC (61.9%; OR, 1.6; 95% CI, 1.2-2.3 compared with controls) than in basal cell carcinoma (BCC; 48.8%) or among controls (53.2%; ref. 27). Increased seroprevalences of HPV 5 have been detected among patients with psoriasis and among patients with autoimmune bullous diseases (3, 28). Apparently, skin burns can also be associated with HPV 5 seroconversion (28).

Although there are many seroepidemiologic studies of cutaneous HPV types, few studies have investigated the sensitivity, specificity, or reproducibility of the serology for cutaneous HPV. A study comparing seropositivity with HPV DNA positivity in plucked eyebrow hairs found only 33% of the HPV DNA-positive individuals to be seropositive, and of the 33%, only 17% were DNA positive and seropositive for the same type (22-24).

Thus, although several studies have found antibodies to at least some HPV types to associate with skin cancers, a better understanding of the meaning of these antibodies in relation to infection with these viruses and more large-scale studies with comprehensive testing are necessary to confirm or refute the possibility that these HPV types may be associated with skin cancer.

We did a comprehensive testing for several different type-specific HPV antibodies in a large hospital-based case-control study that also included comprehensive genotyping, with the aims to investigate (a) whether serology for cutaneous HPV was associated with presence of HPV DNA and (b) whether seropositivity to cutaneous HPV is associated with SCC.

The study was designed as a hospital-based case-control study of nonmelanoma skin cancer (29). Eligible patients were those who sought medical care at participating dermatology clinics where surgical removal of a skin lesion was medically indicated and those who gave informed consent to participate. Initially, 443 patients were enrolled. Following the review of the study file, 9 patients were excluded from serology testing. The exclusion criteria were decided on by the entire study group and without knowledge of any analysis results. The study was restricted to immunocompetent subjects (3 subjects excluded), nongenital lesions (1 subject excluded), and nonmelanoma skin cancer and benign diagnoses (3 cases of melanoma were excluded). A patient with fungal infection was excluded because of not meeting the criterion on surgical removal of a skin lesion being medically indicated. Finally, a patient with simultaneous presence of two different types of tumors was difficult to classify according to the study protocol and was excluded.

The 434 immunocompetent patients included in the study attended dermatology clinics in Sweden (Stockholm, 170 patients; Gothenburg, 118 patients; Malmö, 112 patients) or Austria (Vienna, 34 patients). Cases were defined as having a histologically confirmed diagnosis of SCC (n = 72; mean age, 80 years; range, 50-94) or BCC (n = 160; mean age, 73 years; range, 34-93). For the case group with premalignant actinic keratoses (n = 81; mean age, 75 years; range, 53-95), clinical diagnosis was sufficient. The hospital-based control patients had a variety of benign skin lesions (n = 121; mean age, 71 years; range, 29-97), the most common being seborrheic keratosis (n = 62), and the other benign lesions were benign epidermal dysplasia (n = 1), unspecified benign lesion (n = 1), benign squamous papilloma (n = 2), cornu cutaneous (cutaneous horn; n = 1), epidermoid cyst (n = 1), fibroma (n = 2), healthy skin (n = 1), inflammatory lesion (n = 1), keratoacanthoma (n = 9), large cell acanthoma (n = 1), neurofibroma (n = 1), nevus (n = 7), no rest of tumor (n = 2), pilaracanthoma (n = 1), pyogenic granuloma (n = 1), scar tissue (n = 9), squamous atypia (n = 3), squamous dysplasia (n = 1), squamous hyperplasia (n = 1), sun-damaged skin (n = 1), tag skin (n = 1), trichoepithelioma (n = 1), verruca (n = 1), verruca seborrhoica (n = 1), and verruca vulgaris (n = 3).

The age distributions in the study are in accordance with the age distribution of these diseases in the general population (30, 31).

At the dermatology clinic, patients were interviewed according to a standardized questionnaire, including history of profession, sun-reactive skin types 1 to 4 according to the Fitzpatrick classification (32), where 1 is fair skin, history of sunburns, natural hair color, eye color, smoking habits, history of skin cancer, and family history of skin cancer (Table 1).

A serum sample was obtained from each subject and stored at −20°C until shipped frozen to the laboratory, except samples from Gothenburg that were delivered unfrozen by overnight express mail. At the same visit as the serum sampling, two biopsies were collected from each patient. One biopsy was collected from the lesion and another was collected from healthy skin of the same patient. The control biopsy was most often collected 10 to 15 cm from the lesion at sites such as the neck, the shoulder, proximal to the ear, and the back. Before taking biopsies from the lesion and from healthy skin, the skin was anesthetized (without cleaning with ethanol) and stripped with tape (Art 1527-1, Transpore, 3M Health Care) that was attached five times; then new tape was attached for another five times. Thereafter, one 2-mm-diameter punch biopsy each was taken from the tumor and from healthy skin. The remaining tumor was sent for histopathologic diagnosis as per normal procedures. Actinic keratosis, seborrheic keratosis, and other benign lesions were in most cases only diagnosed clinically. The DNA from each punch biopsy was extracted and dissolved in 200 μL TE buffer using a simple phenol-free method (33). Seven individuals were excluded from HPV DNA testing (but were included in serology testing) because the biopsies had inadvertently not been stripped (SCC, n = 1; BCC, n = 2; actinic keratosis, n = 1; benign, n = 3).

The level of sun exposure of the location of the tumor was classified by a dermatologist into three categories: extensive (that is, head and neck and dorsal side of the hands), moderate (that is, trunk and extremities), and low (that is, buttocks and genital area).

The serum samples were tested for antibodies to the major capsid protein L1 of HPV 1, 5, 6, 8, 9, 10, 15, 16, 20, 24, 32, 36, 38, and 57 and to the early proteins E6 and E7 of HPV 8 and 38. We used multiplex serology, an antibody detection method based on a glutathione S-transferase capture ELISA, in combination with fluorescent technology (34-36).

The DNA from each punch biopsy was extracted using a simple phenol-free method (33). Adequacy of the samples was tested by PCR amplification of the human β-globin gene (37) and analyzed by PCR-enzyme immunoassay in one laboratory (38).

For testing of presence of HPV DNA, all samples were tested with four different PCR methods in three different laboratories: a single-round FAP-PCR (33), a nested FAP-PCR (9), and CP primers as described by Berkhout et al. (4) as single-round and nested PCR but using modified PCR conditions (6). All primers were located in the L1 open reading frame. PCR-positive samples were cloned. Between three and nine clones per positive sample were sequenced. Patient samples were scored positive for a HPV type if the full length of the amplicon sequences showed at least 90% similarity to a HPV sequence in the GenBank database by the Blast program.⁶

HPV DNA typing results were only considered for the types detectable in the serology testing. Cloning and sequencing were considered unambiguous demonstrations of virus, even if not detected in all laboratories. Results from all three laboratories were therefore pooled for the analysis, and if positive in at least one HPV test in at least one laboratory, the sample was scored HPV DNA positive.

The statistical analyses were done using LogXact version 6 (Cytel Software), applying exact methods when possible. Logistic regression was used to estimate ORs, 95% CIs, and P values. P < 0.05 was considered significant. The multivariate analysis of factors associated with skin lesions initially included all measured variables, but variables that did not associate with any of the investigated diagnoses (that is, eye and hair color) were then left out of the final multivariate model.

The study adheres to the Declaration of Helsinki guidelines and was approved by the Ethics Committees of Karolinska Institute and Lund University (Sweden) and Medical University of Vienna (Austria).

The characteristics of the study population are shown in Table 1. Both actinic keratoses and SCC lesions were more common in sun-exposed sites (OR, 5.94; 95% CI, 2.98-11.8 for actinic keratosis; OR, 2.57; 95% CI, 1.30-5.08 for SCC), whereas BCC lesions were not more common on sun-exposed sites (OR, 1.39; 95% CI, 0.82-2.37). SCC patients were often more than 75 years old (OR, 5.38; 95% CI, 1.66-17.5; Table 1) and BCC was associated with skin type 2 (OR, 3.30; 95% CI, 1.09-9.96). History of sunburns tended to be negatively associated with BCC (Table 1). Finally, BCC tended to be more common among males (OR, 1.77; 95% CI, 1.02-3.08; Table 1).

There was little difference in type-specific HPV antibody prevalences between the different diagnosis groups (Table 2). However, for almost all HPV types, the seroprevalences among SCC patients were slightly higher than among BCC patients (P < 0.01; Table 2).

There were no differences in seroprevalences between the sexes (Table 3). Individuals below 60 years old had higher seroprevalence for any HPV genus α (mainly HPV 16; data not shown) than did the older age groups (Table 3). Seroprevalences for other HPV types or genera did not differ between the age groups.

Seropositivity for any β type or multiple positivities for β types were not significantly different compared with the group of patients with benign skin lesions. However, there was a nonsignificant tendency for positivity for at least three β HPV types to be more common among SCC patients than among BCC patients (Table 2).

The prevalence of HPV antibodies was not significantly related to skin type or previous sunburns (Table 3).

Biopsies from tumor and healthy skin from 427 patients were tested for HPV DNA using PCR. Overall, 58% of SCC patients were HPV positive in at least one sample in at least one HPV assay (OR, 2.08; 95% CI, 1.03-4.22 compared with HPV presence in benign tissue). HPV DNA positivity in the other patient groups was 39% in BCC, 38% in actinic keratosis, and 39% among the patients with benign lesion (Table 1). More detailed results on HPV DNA results by type of tissue have been reported elsewhere (29). Seropositivity was almost twice as common among individuals positive for any HPV DNA (64%) than among HPV DNA-negative individuals (34%; OR, 1.90; 95% CI, 1.55-2.34; Table 4). No biopsies were HPV DNA positive for HPV 1, 6, 10, 32, or 57 and only one patient had HPV 16 DNA in one biopsy. Overall, 19% (52 of 274) of the HPV DNA-positive samples were seropositive for the same HPV type. For each individual HPV type, between 0% and 28% of the HPV DNA-positive patients were seropositive for the same type (Table 4).

Antibodies to the oncoproteins E6 and E7 of HPV 8 and 38 were rare in this study, and no significant determinants of oncoprotein seropositivity could be found (data not shown).

We found that HPV seroprevalences were consistently higher among SCC patients than among BCC patients, suggesting that several HPV types may have a slight association with SCC. The reason why the BCC patients seem to have lower seroprevalences than the control group of patients with benign diagnoses is not entirely clear. A previous serologic study using the same serology methods also reported lower HPV antibody prevalences in BCC patients than that of controls and SCC cases (3). Possibly, some of the benign conditions in the current control group may also be associated with infection with skin types of HPV. The control group was not sufficiently large to allow meaningful exploratory analysis of HPV seroprevalences among the different benign conditions that were included. Lower ability to mount an adequate HPV antibody response among subjects with BCC is another possibility.

None of the known risk factors for nonmelanoma skin cancer, such as age, sex, skin type, or previous sunburns, was associated with HPV seroprevalences. For example, sun exposure of tumor site was significantly correlated to actinic keratosis and SCC but was not associated with seropositivity. Associations of HPV antibodies with SCC or BCC are thus not likely to be due to (residual) confounding. Favre et al. (28) found that the epidermal repair process in second-degree burns was associated with the generation of HPV 5 antibodies but also found that epidermal proliferation limited to small areas (such as nonmelanoma skin cancer) was not sufficient to evoke antibody formation.

The self-reported information about skin type, previous sunburns, and smoking was not associated with SCC or BCC, even if some of them are known risk factors. This is probably because of the low validity of self-reported information.

The proportion of patients who were seropositive for the same type among those who had HPV DNA detected was only 20%, suggesting that serology has a very low sensitivity for the cutaneous HPV types. This confirms the finding from a previous study where seroprevalences were compared with detection of HPV DNA in plucked eyebrows and only 5.2% of the individuals were seropositive for the same HPV type as detected in the plucked eyebrow hair (26). Weissenborn et al. (39) found very low HPV DNA copy numbers (median, 1 HPV DNA copy per 344 cells) in biopsies of cutaneous SCC and perilesional tissues. The low viral loads of β HPV may perhaps sometimes be too small to induce specific antibody responses.

On the other hand, most antibody reactions found were against skin HPV types not found by DNA analysis of the biopsy tissues. These antibodies may reflect past infections, current infections at other skin sites, and current infections in the biopsy tissues but missed by the DNA analysis technique (cloning and sequencing of PCR products) that may select for the most abundant and/or best amplifiable type(s).

There has thus far not been any report of different serotypes within a genotype of genital HPV. However, the cutaneous HPV 5 is known to consist of at least three different serotypes (HPV 5a, 5b, and 5c; ref. 3), suggesting that serologic heterogeneity could be an explanation for low sensitivity of cutaneous HPV serology. Also, it should be pointed out that the typical duration of cutaneous HPV infections is not known, although it seems that persistent infection over many years is common on healthy skin (40). For genital HPV, seroconversion is less common among women with transient infection than among women with persistent infection (41). In addition, the natural history of cutaneous HPV antibodies in terms of acquisition and loss is not known. For genital HPV, seroconversion is typically delayed after infection with 0.5 year or more, whereas antibodies wane only slowly after viral clearance (41).

We did not detect any increased HPV seroprevalence among subjects positive for the same type of HPV DNA, suggesting that the type specificity of the antibody response to cutaneous HPV is very low or that sampling only two biopsy sites is not sufficiently sensitive to detect all skin HPV infections in an individual. Albeit we found no evidence to suggest the existence of a type-specific antibody response to cutaneous HPV, the strongly increased seropositivity for any HPV among subjects that were positive for HPV DNA of any type suggests that seropositivity for cutaneous HPV indeed has at least some specificity for cutaneous HPV infection. Larger studies with large numbers of observations for each HPV DNA type are required to study the issue. It should also be pointed out that cutaneous HPV serology is still in its infancy and the issue of whether assay improvements may increase type specificity has not been investigated in detail.

In conclusion, further studies on the natural history of cutaneous HPV infection as well as of the sensitivity, specificity, and natural history of antibodies to cutaneous HPV are required to allow reliable use of HPV serology in elucidation of the possible role of HPV in nonmelanoma skin cancer. In spite of the limited sensitivity and specificity, the present study suggests that several HPV types may have some modest associations with SCC of the skin when contrasted to BCC.