Do Epidermodysplasia Verruciformis Human Papillomaviruses Contribute to Malignant and Benign Epidermal Proliferations? FREE

In an earlier article¹ on epidermodysplasia verruciformis (EV) as a model of genetic human papillomavirus (HPV)-associated skin cancer, we presented clinical, pathologic, virologic, and immunologic characteristics of this rare genetic disease.

In brief, the main characteristics of EV are as follows. In most EV cases, the transmission is autosomal recessive, but in single families, an X-linked mode of inheritance has been reported. The first cutaneous lesions usually appear at age 4 to 8 years, and the infection is lifelong. Involvement is limited to the skin; the internal organs, mucous membranes, hair, and lymph nodes are not affected. Cutaneous lesions in EV are heterogeneous. Benign lesions are of plane wart type, macular (red plaques), brownish (brown plaques), and pityriasis versicolor–like. Proliferative benign lesions are papilloma- and verruca seborrheica–like.

The first malignancies—actinic keratoses, early microinvasive squamous cell carcinomas of the Bowen type, and slowly progressing to invasive tumors—usually start to appear in the fourth decade of life.² However, metastases are rare if no radiation therapy is applied.³ Ultraviolet light, like x-rays, is a cocarcinogen, and, therefore, malignant tumors develop mostly on the sun-exposed parts of the body, frequently on the forehead, originating from hair follicles (Figure 1 and Figure 2).⁴ A characteristic feature of EV is substantially decreased cell-mediated immunity, specifically, delayed-type hypersensitivity toward EV HPVs.^{5 - 6} Thus, patients' immune system cells do not recognize and do not reject EV HPV–harboring keratinocytes.

Transmission of genital HPVs from mother to infant usually occurs by passage of the neonate through an in fected birth canal⁷ ; however, HPV-16 DNA has also been found in infants born by cesarean section, suggestive of intrauterine infection, presumably via transplacental transmission.^{8 - 9}

Favre et al¹⁰ reported possible vertical transmission in a patient with HPV-5– and HPV-8–associated EV. She delivered a healthy infant by cesarean section, and the same EV HPV types as in the mother were in the amniotic fluid, the placenta, and oral and genital scrape specimens.¹⁰ The mechanism of the vertical transmission is not known; however, detection of the same EV HPVs in cervical scrape specimens may suggest an ascending infection from the genitalia.

Why do benign EV lesions start at age 4 to 8 years, and the malignant conversion starts mostly in the fourth decade? This question cannot be addressed in epidemiologic studies of this rare genetic disease. Based on the results of a large epidemiologic survey of women,¹¹ a high incidence of cervical intraepithelial neoplasia was found in those aged 25 to 35 years, and a peak incidence of cervical cancer was found in those aged 55 to 65 years. Thus, it was shown that progression toward invasive cervical cancer requires 20 to 30 years.^{12 - 13} To find out why the benign lesions in EV start in late infancy and carcinomas develop after the third decade of life,^{1 ,14} we¹⁵ observed skin autografted on the forehead of patients with EV after surgical removal of multiple carcinomas. We found that around grafted skin obtained from the unexposed part of the interior aspect of the arm, multiple premalignant lesions started to develop within several months to a few years. In only about 6 years, single benign lesions appeared within the graft, but no carcinoma developed until 22 years of graft life, although several premalignant and early malignant lesions were present in the surrounding skin. This could be explained by the higher cumulative dose of UV radiation in the skin surrounding the graft than in the graft itself. These findings also provide some explanation as to why the first benign lesions in patients with EV start to appear at age 4 to 6 years. In addition, the study showed that malignant conversion in EV is a similarly long-lasting process as progression toward invasive cancer in high-risk HPV-associated genital lesions.

Benign and malignant EV lesions harbor a high number of copies of various EV HPV DNA. In EV cancers, HPV-5 is a predominant type, and this virus and HPV8 were found in more than 90% of EV cancers.¹⁴ However, EV HPVs, although infectious, do not evoke cutaneous lesions in the general population, which strongly suggests the role of genetic factors predisposing to EV HPV productive infection. Although the specific gene for EV has not been disclosed, and no specific haplotype associated with EV has been characterized until now, there is an important new finding on the susceptibility locus for EV (EV-1) on chromosome arm 17qter in a region containing a psoriasis locus (PSORS2).¹⁶ The second susceptibility locus for EV is also found in a region containing another psoriasis susceptibility locus on chromosome 2.¹⁷ It is possible that other susceptibility loci for EV will be mapped to other chromosomes, as shown for psoriasis, because both diseases are polygenic and highly heterogenous.

Since the HPVs responsible for EV were found in cutaneous cancers in the general population, EV has received special attention. Previously, EV HPVs were believed to be specific to the disease because, using routine DNA hybridization techniques (Southern blot and in situ hybridization), they could not be detected in patients without EV, except immunosuppressed populations. By using a new, extremely sensitive technique of nested polymerase chain reaction (PCR) and degenerate primers, EV HPV DNA was detected in most nonmelanoma skin cancers (NMSCs) in immunosuppressed and immunocompetent populations.¹⁸ Minute amounts of EV HPV DNA were also detected in uninvolved skin and plucked eyebrow hairs of healthy people.^{19 - 20}

Studies on HPV involvement in cutaneous oncogenesis in the general population, based mostly on small series of patients, showed the presence of various genital^{21 - 22} or cutaneous HPVs or both (for reviews see Pfister and ter Schegget²³ and Harwood et al²⁴ ). The prevalence of HPV types and the frequency of detection depend on the technique and the primers used. By using nested PCR¹⁸ and a highly sensitive modification,²⁴ EV HPVs or EV-related HPVs were found to be highly prevalent in cutaneous cancers, especially in immunosuppressed populations.^{25 - 30} The viral load, however, was extremely low, even when primers were used in combination for detection.^{24 ,31 - 32} There was no difference in viral load and HPV types among squamous cell carcinomas, basal cell carcinomas, and actinic keratoses.²⁴ Moreover, in a recent study,³³ no significant association between the presence of EV HPV DNA in plucked eyebrow hairs and various NMSCs was found in the general population. Thus, a causative role of EV HPVs in skin cancers in immunocompetent individuals is not established.

De Villiers²⁹ and colleagues²⁶ detected various EV HPVs in 90% of NMSCs of organ transplant recipients and suggested that infection probably occurred in early infancy and remained latent until activated by some cocarcinogens (eg, UV radiation). However, there is not yet direct proof for this hypothesis. Epidermodysplasia verruciformis HPVs were found not only in tumors in immunosuppressed populations but also in warts, dysplastic keratoses, and unusual neoplasias originating from hair follicles.³⁰ Harwood et al^{27 ,32} disclosed EV HPV DNA in 80.4% of the warts in the immunosuppressed population and genital HPV DNA in 27.4%. Two or more distinct HPV types could be codetected in 94% of lesions with no predominance of any distinct HPV type.^{28 ,32} In severe cases in highly immunosuppressed populations, EV HPVs present in warts together with viruses responsible for flat warts in the general population^{34 - 35} could lead to the appearance of clinically typical EV phenotypes.^{36 - 37} Although the involvement of EV HPV DNA in oncogenesis is not clear, the association between the number of keratotic and dysplastic lesions and the development of NMSCs in immunosuppressed populations favors some role for HPVs in addition to some factors other than HPV, eg, sun exposure.

In most studies, the EV HPV types in NMSC were multiple and divergent, whereas in basal cell carcinoma, with 43.5% prevalence of EV HPVs in specimens, mostly one HPV type has been detected; however, none of the types were predominant.³⁸ Only in a single study³⁹ was oncogenic EV HPV-5 detected in skin tumors developing mainly in patients with psoriasis treated long term with large doses of psoralen–UV-A (PUVA) (>500 J/cm²). In this study, 75.0% of NMSCs and 41.2% of dysplastic keratoses disclosed EV HPV-5, HPV20, HPV21, HPV23, and HPV-24 and, in some cases, additional infection with non-EV cutaneous and mucosal HPV types.

The role of early oncoproteins (E6 and E7) of high-risk genital HPVs is well established for genital cancers, and their transforming activity is related mainly to their capability to interfere with cellular antioncogenes, mainly p53 and pRB.^{12 - 13} The mechanism of action of E6 and E7 of oncogenic EV HPVs (especially HPV-5) is not known. It was shown that E6 oncoprotein does not degrade p53 protein and that E7 oncoprotein does not transform keratinocytes.^{40 - 41} Both E6 and E7 are expressed in EV cancers, although integration of HPV-5 DNA into the host DNA is an exception and occurs only in metastases.¹⁴ Most important, HPV-5 shows great genomic heterogeneity of E6 and L1 proteins compared with other HPV types, and this may reflect selective immune mechanisms related to formation of specific antibodies against these proteins.^{42 - 43}

Despite the inability of HPV-5 and HPV-8 E6 proteins to degrade p53,^{40 ,44} some abnormalities of this antioncogene were described in skin tumors in patients with EV and the general population. The p53 mutations are the most common genetic abnormalities in cutaneous cancers of the general population and play an important role in tumor progression.⁴⁵ The most frequent mutations in NMSCs were specific for UV-B: C→T substitutions and less frequently CC→TT double base mutations.⁴⁶ In a recent study on p53 DNA using PCR and sequencing of the amplification products, we found UV-related (C→T and CC→TT) and non–UV-related (G→C and G→T) mutations.⁴⁷ In cutaneous viral oncogenesis, stabilization and accumulation of inactivated p53 may occur through interaction with viral or cellular proteins, thus preventing growth arrest and apoptosis. The high prevalence of p53 immunoreactivity due to accumulation of inactivated or mutated p53 in tumors of allograft recipients favors a role for the p53 gene in cutaneous carcinogenesis.⁴⁸ In contrast, no positive immunostaining for p53 was detected in benign viral warts from organ transplant recipients and the general population.⁴⁹ In EV, expression of p53 was found in most benign lesions (warts) and malignant tumors,⁵⁰ whereas in patients without EV and allograft recipients, there was no expression of this protein.^{51 - 53}

Although cutaneous cancers in the general population and in patients with EV occur most frequently in sun-exposed areas, in our study, 35 specimens from EV tumors positive for p53 were obtained from nonexposed skin, suggesting that some other factors, besides UV light, could up-regulate p53 expression. In a recent study⁵⁴ it was shown for a unique HPV-77 type detected mainly in immunosuppressed populations that UV light may act through its p53-dependent positive response element, indicating that UV light not only has a DNA-damaging and immunosuppressive effect but might be a cofactor with specific HPV types in cutaneous carcinogenesis.

Contrary to a positive correlation between apoptosis and degree of malignancy in cervical cancers,^{55 - 56} no such correlation has been established in EV. Apoptosis, studied using DNA fragmentation assay, was also detected in benign EV lesions, in the areas of cytopathic effect (unpublished observations). Thus, EV oncogenesis differs from HPV-associated genital and UV light–related cutaneous carcinogenesis.

What is the role of EV HPVs in cutaneous carcinogenesis in non-EV populations? A high frequency of EV HPVs and EV-related HPV DNA in cutaneous premalignant and malignant lesions does not provide direct evidence for their causative role because various EV HPVs, but usually nononcogenic types, are detected. Contrary to the association of EV cancers with oncogenic EV HPVs, mainly HPV-5 or HPV-8, these viruses were not found in malignant cutaneous tumors in the general population. Moreover, the amounts of HPV DNA in the tumors are extremely low because for their detection, a highly sensitive method (nested PCR) must be used. Most important, no messenger RNA or transforming proteins E6 and E7 have been found in the tumors. Without this, the role of HPVs in cutaneous oncogenesis cannot be confirmed because EV HPVs, shown to be ubiquitous HPVs, are present in malignant lesions and in normal skin.⁵⁷ Thus, although no direct proof for their causative role in cutaneous oncogenesis in the general population is available, an important mechanism by which cutaneous HPVs may contribute to skin carcinogenesis could depend on a recently established ability of E6 proteins of cutaneous HPVs to inhibit UV light–induced apoptosis.⁵⁸ This finding indicates that some viral proteins could contribute to the early stages of tumor development.

An important finding was the detection of EV HPVs in approximately 35% of probands in nonaffected skin from patients with cutaneous malignancies and healthy individuals.⁵⁷ Almost the same percentage of positive results (33%) was reported in unaffected skin from PUVA-exposed patients,³⁹ strongly suggesting the presence of a latent EV HPV infection in the skin. On activation, EV HPV DNA might become expressed in proliferating keratinocytes. The reservoir of EV HPV could be the bulge area of the outer sheath of hair follicle, found in approximately 45% to 62% of probands to harbor various EV HPV types and, inter alia, HPV-5, in immunocompetent and immunosuppressed populations.^{19 - 20} In cases of genital infection, the eyebrow hair harbored not only HPV-6/-11 but also EV HPVs.⁵⁹ This is suggestive of a latent EV HPV infection and would explain development of EV HPV–associated warts, dysplastic keratoses, and cancers in immunosuppressed populations owing to a dramatic decrease in cell-mediated immunity, resulting in alleviation of immunosurveillance and activation of HPV infection.⁶⁰ The latent infection of hair follicles could also explain why the premalignant and malignant changes develop frequently from the hair follicles in patients with EV.²

Because EV HPVs were detected in malignant cutaneous tumors, we were interested in knowing whether these viruses could also be present in benign epidermal proliferations. Psoriasis, a common inflammatory, genetically determined skin disorder, served as a model for such an epidermal hyperproliferation.

By using highly sensitive nested PCR and type-specific primers, we detected HPV-5 DNA in approximately 80% of psoriatic lesions.⁶¹ By sequencing PCR products, additional EV HPVs—mainly HPV-36 and some other EV HPVs or EV-related HPVs—were disclosed. In a few cases, the presence of EV HPV DNA has been confirmed by Southern blot testing; thus, the amount of viral DNA might be higher than suggested by nested PCR. Most important, in a large series of patients with psoriasis, using enzyme-linked immunosorbent assay and HPV-5 L1 viruslike particles bearing conformational epitopes, specific antibodies to HPV-5 were detected in 24.5% of patients; in controls (including patients with various warts, patients with atopic dermatitis, immunosuppressed patients, and healthy individuals), the results were positive in only 2% to 5%.⁶¹ Similar to EV, there was remarkable heterogeneity in EV HPVs. After sequencing of amplification products, 27 variants of HPV-5 and 13 variants of HPV-36 DNA were disclosed in psoriatic plaques, which indicates that the results of nested PCR are not owing to contamination.

The finding of EV HPV DNA in psoriatic skin has been confirmed by other researchers,⁶² who found EV HPV DNA sequences in 83% of psoriatic lesions, with a higher prevalence of HPV-36 (63%) than HPV-5 (38%) (these EV HPV types are closely related). The differences in the frequency of HPV-5 detection might be due to technical reasons, especially the sensitivity of the primer sets for nested PCR and type-specific PCR tests.

What is the importance of these virologic findings for the pathogenesis of psoriasis? Psoriasis is a T-cell–mediated disease, and polyclonal activation of T lymphocytes (CD4) by superantigens, trauma, or other factors seems to be an important first step in the development of psoriatic lesions.⁶³ However, intraepidermal, oligoclonally expanded CD8⁺ T cells prevail in plaque psoriasis, strongly suggesting a classical pathway of antigen activation.^{64 - 65} We hypothesized that the antigen could be a viral epitope or some self-peptide modified by the viral proteins within upper layers of the epidermis. The L1 EV HPV (capsid protein of EV HPV) could also serve as a target for specific humoral autoimmune reactions, resulting in complement activation and chemoattraction of polymorphonuclear leukocytes with formation of psoriasis Munro abscesses.⁶⁶ T-cell activation promotes keratinocyte proliferation, and this leads to replication of HPV DNA in differentiating keratinocytes, and, in turn, established HPV infection enhances epidermal proliferation. In contrast to herpes simplex virus, HPV is a proliferative and not a lytic virus and thus replicates in concert with cell proliferation. In psoriasis, keratinocyte hyperproliferation seems to result from self-perpetuation of the autoimmune process related to T-cell activation and cytokine production.

Epidermodysplasia verruciformis HPVs are also associated with keratinocyte hyperproliferation other than psoriasis (eg, epidermal regeneration in wound healing). We also detected EV HPV DNA in the skin of patients with burns and bullous autoimmune disorders.⁶⁷ However, in contrast to psoriasis, in patients with burns, the antibodies to L1 HPV-5 are transient and appear several weeks after burns, which is suggestive of a link between epidermal proliferation in healing, expression of HPV-5 proteins, and generation of antibodies. Several weeks after complete healing, antibodies disappear. In autoimmune bullous diseases, EV HPV-5 DNA also may be expressed and induce HPV-5 antibody generation, and this process seems to be enhanced by autoimmunity. However, it remains to be determined whether EV HPVs directly contribute to hyperproliferation of keratinocytes in epidermal repair.

In psoriasis, which could be regarded as a reservoir of HPV-5,⁶¹ the process of epidermal hyperproliferation is perpetual, probably linked to HPV-5 expression. It is not clear why potentially oncogenic HPV-5 and HPV-8, present in benign psoriatic hyperproliferative plaques, do not lead to malignant conversion, as seen in cancers in patients with EV, and why HPV-5 DNA is detected rarely, if at all, in NMSCs in the general population and in immunosuppressed individuals, except for cancers developed after prolonged PUVA therapy in patients with psoriasis.^{39 ,68} In a single study,⁶² HPV-5 was found mostly in patients after prolonged and high-dose PUVA treatment. However, small doses of PUVA applied in psoriasis is one of the best treatment modalities because it decreases keratinocyte proliferation (a basic factor promoting HPV replication), inflammatory changes, cytokine trafficking, etc. Thus, there is a strong rationale for this therapy. High doses (>500 J/cm²), because of the mutagenic effect of UV light, might eventually lead to cutaneous carcinogenesis.^{39 ,69}

The role of HPV-5 in psoriasis must be confirmed by detection of E6/E7 transcripts or oncoproteins in psoriatic plaques. Although the genes responsible for EV and psoriasis are not identified, the co-localization of the susceptibility loci for psoriasis (PSORS2) and EV (EV-1) to the same region on chromosome arm 17qter could indicate that distinct defects (mutations) of the same gene might lead to various manifestations of EV HPV infection in such different diseases as EV and psoriasis.¹⁶ In EV, the genetic restriction toward HPV-5, present in the general population, is abrogated, whereas in psoriasis, it is partially alleviated, probably allowing for persistence of EV HPV infection and establishment of an HPV-5 reservoir.

In conclusion, various HPV types might be present as a normal skin flora in immunocompetent populations. Epidermodysplasia verruciformis HPVs were proved to be of special interest not only as a causative factor of EV, a rare genetic disease/genetic cancer, but as ubiquitous viruses associated with keratinocyte proliferation and epidermal repair processes. Epidermodysplasia verruciformis HPV-5, the most common oncogenic virus in patients with EV, was disclosed in psoriasis, together with other EV HPVs, and conceivably contributes to the pathogenesis of this hyperproliferative epidermal disorder. Detection of the main susceptibility loci for psoriasis (PSORS2) and for EV (EV-1), mapped to the same region of 17qter, strongly suggests a genetic link between these 2 different genetically determined diseases. Thus, our understanding of EV HPVs as specific viruses of EV had to be revised, and the question of their possible contribution to the malignant and benign epidermal proliferations should be addressed.

Accepted for publication December 19, 2000.

This study was partially supported by grants 4P05B 08819 and 4P05A 01719 from the Polish Committee for Scientific Research, Warsaw.

We thank Gerard Orth, VD, and Michel Favre, PhD, of Institut Pasteur, Paris, France, for their long-lasting and fruitful cooperation.

Corresponding author and reprints: Stefania Jablonska, MD, Department of Dermatology and Venereology, Warsaw School of Medicine, Koszykowa 82A, 02-008 Warsaw, Poland (e-mail: sjablonska@bibl.amwaw.edu.pl).