Infection by papillomaviruses is essential for the development of cervical cancer (reviewed by zur Hausen1). A number of the papillomavirus types (high-risk) have been associated with and demonstrated in premalignant and malignant lesions of the genital tract. The infection per se is, however, not sufficient to induce malignant conversion. A complex interplay between viral and cellular genes is required to disrupt the cell-cycle control and induce immortalization, followed by progression to malignant conversion. The molecular mechanism through which papillomaviruses, in particular HPV-16 and HPV-18, participate to achieve this condition has been well established. The low-risk mucosal papillomaviruses, causing benign lesions mainly of the genital tract, share several functional characteristics with the high-risk mucosal papillomaviruses but lack others necessary for the immortalization of normal keratinocytes (reviewed by zur Hausen2). The maintenance of a high level of certain viral proteins within the cell is necessary for the virus to induce and regulate distinct cellular pathways. Several exogenous and endogenous factors controlling or influencing viral transcription and the expression of the viral oncogenes, E6 and E7 in particular, are known. Very few data are available on the in vitro and in vivo mechanisms through which steroid hormones, in particular, influence the carcinogenic process induced by high-risk papillomavirus infections.
Steroid hormones exert a broad range of biologic functions. The synthesis, transport and metabolism of the naturally occurring hormones are tightly regulated in various organs of the body (reviewed by Clemens and Goss3 and Gruber et al.4). A delicate balance exists during the premenstrual, menstrual and postmenopausal phases of the female body, as well as during pregnancy. This balance is greatly influenced by the intake of exogenous steroid hormones in the form of OCs in young women or HRT in postmenopausal women. Reports in which epidemiologic data taken from 8 case-control studies have been pooled and analyzed indicate an increased risk for developing cervical carcinoma in women with high parity5 or after the long-term use of OCs.6 Odds ratios were 3.8 for cervical carcinoma after 7 full-term pregnancies, and 2.3 for women with 1 or 2 full-term pregnancies, respectively. Odds ratios for use of OCs were 2.82 for 5- to 9-year use and 4.03 for 10-year or longer use. Other epidemiologic studies have also reported an increased risk of developing cervical adenocarcinoma among long-term users of OCs.7, 8, 9
Regulation of viral transcription occurs mainly within the control region of the genome (LCR). Cellular as well as viral factors bind to this region, either to stimulate or to suppress the activity of the individual genes within the viral genome. A large number of cellular transcription factors binding to specific binding sites in the LCR have been identified.10 At least 3 GREs respond to both progesterones and glucocorticoids by increasing the transcription of the E6 and E7 genes of the high-risk types HPV-16 and HPV-18 and the low-risk type HPV-11.11, 12, 13, 14 This response depends on the activity of other binding sites, e.g., NF-1 and AP1, positioned in the same region of the LCR as the GRE;15, 16 and a complex interaction of c-jun and c-fos and the glucocorticoid receptor with the HPV-16 GRE has been demonstrated.17 Interestingly, not all HPV GREs respond to hormones to the same degree. The 3 GREs present in the HPV-31 LCR are only weakly inducible by dexamethasone.18
The differential effects of steroid analogs on the regulation of the HPV LCR have been investigated.19 The interaction of the steroid hormones with the cognate receptor is of importance. The transcriptional activation functions of the A and B isoforms of the human progesterone receptor result in different molecular events for association with the GREs or progesterone- or estrogen-responsive elements within the target genes. This is mirrored in the differential activation of the HPV-16 promoter by a series of progestins and estrogens and influenced by the structures of these molecules. These hormones have a lesser effect on the HPV-18 LCR.
The transformation of rodent cells20, 21, 22 and human keratinocytes23 by HPV-16 in combination with activated ras and fos has been demonstrated to be hormone-dependent and could be inhibited by the hormone antagonist RU486.24 The continued growth of these transformed cells was, however, not hormone-dependent.25, 26 This loss of hormone-dependent growth was correlated with altered transcription patterns of the HPV E6 and E7 genes.14, 26–28 The glucocorticoid requirement during the early passages of HPV-16-immortalized human keratinocytes decreases during in vitro progression of cells. Cervical carcinoma cell lines, however, express sufficient endogenous glucocorticoid receptors to allow for a hormonal response, indicating additional indirect hormone-mediated mechanisms.29 These data are mirrored in vivo by the loss of expression of the estrogen receptor in CIN lesions and cervical carcinoma.30 The progesterone receptor is expressed at high levels in low- and high-grade CIN lesions, whereas it is expressed at very low levels in biopsies taken from cervical carcinomas.31
Estrogens are metabolized through several pathways. The hydroxylation of estrogens yields catechol estrogens, which include 2-, 4- and 16α-hydroxyestrogens. The increased 16α-hydroxylation of estradiol to 16α-hydroxyestrone and estriol has been regarded as a risk factor for cancer (reviewed by Gruber et al.4), whereas 2-hydroxyestrogens have been associated with a decreased risk as well as an antiproliferative effect. 16α-Hydroxyestrone covalently binds to the estrogen receptor, thereby prolonging the effect of estrogen. This specific conversion, in contrast to the other pathways, is greatly enhanced in HPV-16-immortalized endocervical cells, as well as cervical carcinoma cells in comparison to normal keratinocytes, suggesting that 16α-hydroxylation and high-risk HPV types increase the effects of each other in promoting cell proliferation.32 The interaction of HPV with estrogen metabolism is schematically presented in Figure 1. Increased 16α-hydroxylation of estradiol to estrone in the presence of HPV infection was also evidenced in samples taken from a population of normal women in comparison to premenopausal women with CIN.33
The HPV genome is mainly present in an episomal form in benign and premalignant lesions. Progesterone and glucocorticoid hormones increase viral mRNA34 and significantly stimulate viral replication.35 The frequency of integration of the viral genome increases with progression to high-grade lesions and invasive carcinoma, during which the genome is interrupted and deletions occur, mainly within the E2 ORF (reviewed by zur Hausen36). High levels of E7 protein of high-risk papillomaviruses lead to apoptosis (reviewed by zur Hausen2). The E7 protein binds to the Rb–E2F complex, releasing E2F, which in turn upregulates genes that induce apoptosis.37, 38, 39 The E2 protein, when overexpressed in in vitro systems, interacts with p53 and induces p53-dependent apoptosis.40, 41 E2- and E7-induced apoptosis is increased by estrogen and progesterone, as well as by 16α-hydroxyestrone. The apoptotic effect of E2 and E7 is overridden by the overexpression of E6 protein, which binds to p53, leading to degradation of the latter. The estrogen receptor antagonist 3-hydroxytamoxifen also blocks apoptosis induced by E2 and E7, whereas the progesterone antagonist RU486 blocks the effects of both estrogen and progesterone.42
The dietary component I3C blocks the effect of estrogen by reducing the formation of 16α-hydroxyestrone42 as well as the ability of estradiol to support anchorage-independent growth of HPV-immortalized keratinocytes.43 Conflicting results were reported from in vitro experiments in cervical carcinoma cell lines, where I3C induced apoptosis.44 These results were confirmed in vivo in the cervical epithelium of HPV-16 transgenic mice exposed to 17β-estradiol and fed I3C. Oral treatment with I3C in patients with CIN led to significant regression of the lesions.45
The transformation zone of the cervix is highly sensitive to estrogens in comparison to other reproductive system sites. Introduction of HPV-16 DNA into these cells led to an 8-fold increase in the already existing high level of estradiol conversion to 16α-hydroxyestrone.32, 46 An in vivo model was established by xenografting human tissue harboring HPV-16 DNA into SCID mice. The transplanted human squamocolumnar junction of normal and dysplastic tissue could be maintained under these conditions, whereas application of high doses of estrogen induced cervical hyperplasia and metaplasia in normal tissue transplants.47 Chronic estrogen exposure in HPV-16 E6/E7 transgenic mice resulted in multistage neoplastic progression in the squamous cell epithelium of the cervix and vagina.48 The carcinogenic process was, however, restricted to the cervical transformation zone in a follow-up study in which a 5-fold reduction in estrogen dose was used,46 as confirmed in another study.49 Elson et al.46 speculated that the intracellular acidosis brought about by the binding of HPV E7 to M2 pyruvate kinase,50 which results in aerobic glycolysis of intracellular glucose metabolism, contributes to a squamous, rather than a columnar, fate decision in glandular reserve cells in transgenic mice treated with low doses of estrogen. An I3C-supplemented diet for mice prevented the development of cervical cancer and reduced the estrogen-induced effect to dysplasia and hyperplasia.51 The tissue-specific induction of the HPV-18 early promoter in the genital epithelium of transgenic mice was suppressed by ovariectomy but restored by administration of estrogen alone or in combination with progesterone. Tamoxifen and RU486 reversed this latter effect.52 Squamous metaplasia appears to represent the first stage of transformation.46 A prospective study in young women showed an increase in the prevalence of ectopy among OC users. This exposes endocervical cells to HPV infection.53 Endocervical cells express the E6/E7 HPV genes when implanted in nude mice or in in vitro organotypic raft cultures and resemble immature squamous epithelia.54
The in vitro and in vivo models are helpful in analyzing the role of hormones in papillomavirus carcinogenesis. Unfortunately they cannot mirror the exact situation in humans: in the majority of in vitro systems, the HPV genes involved were overexpressed, whereas it may be difficult to mimic human physiologic levels of hormone dosages in mice. Epidemiologic data may provide more insight into the human situation, though other factors may complicate the interpretation of these data. The use of OCs may influence sexual behavior and thereby increase the risk for acquisition of papillomavirus infections. These individuals are usually subjected to increased screening procedures, which in turn leads to a possible earlier diagnosis.55
Hormone use could also have a protective effect, depending on age: in a young age group,56 frequent acquisition of infections by different types of papillomavirus may stimulate immunologic control. The continuous stimulatory effect of the hormones on expression of the papillomavirus genes present in cervical tissue could explain the reported epidemiologic data on an increased risk of cervical cancer through long-term use of OCs and high parity. This continued expression leads to an increase in viral replication and, in particular, in expression of the HPV E6 and E7 proteins, thereby revealing their transforming potential. Long-term use of OCs may increase the risk of cervical carcinoma, but the risk of unwanted pregnancies may be even higher.
In conclusion, all available data reveal an effect of 16α-hydroxyestrone in enhancing high-risk HPV gene activity. This corresponds to epidemiologic studies showing an increased risk for long-term OC-using, HPV-infected women. No evidence exists for an increase in cervical cancer among HPV-negative OC users.