Int. J . Cancer: 45, 860-864 (1990) 0 1990 Wiley-Liss, Inc.
Publication of the International Union Against Cancer Publication de I'Union lnternationale Contre le Cancer
ASSOCIATION OF ORAL CONTRACEPTIVE USE AND HUMAN PAPILLOMAVIRUSES IN INVASIVE CERVICAL CANCERS Allan HILDESHEIM', William c. REEVES~, Louise A. BRINTON',carol LAVERY3, Maria BRENES4, Maria Elena DE LA GUARDIA~, Julio GODOY'and William E. RAWLS~ 'Environmental Epidemiology Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD; 2Division of Viral Diseases, Center for Infectious Diseases, Center for Disease Control, Atlanta, GA, USA; 3Molecular Virology and Immunology Program, Dept. of Pathology, McMaster University, Hamilton, Ontario, Canada; 4Division of Epidemiology, Gorgas Memorial Laboratoiy, Panama City; 51nstituto Oncologico Nacional, Panama City, Republic of Panama. In a study of 197 cases of histologically confirmed invasive cervical cancer, 61% of biopsies were positive for human papillomavirus (HPV) D N A by Southern or dot-blot hybridization. An association between detection of HPV D N A and oral contraceptiveuse was observed when HPV-positive and -negative cases were compared. Women reporting recent or longterm (>4 yrs) oral contraceptive use were at 2.3 and 2.9-fold increased risks of HPV positivity, respectively. An increased risk of HPV positivity was also associated with formal education and with urban residence, while long-term smoking was negatively associated with HPV detection. A non-significant trend of increasing risk of HPV positivity with increasing number of sexual partners of the women and of the male partners of monogamous women was observed. Detection of HPV DNA was not associated with other cervical cancer risk factors examined, including age at first coitus, number of pregnancies, and Pap smear screening history. Our findings suggest either an interaction between HPV infection and oral contraceptive use in the genesis of cervical cancer or an increased expression of HPV genome in neoplasms of oral contraceptive users. These observations also support a multifactorial model of cervical cancer causation.
There is an expanding body of evidence linking human papillomaviruses (HPV) to cervical neoplasms (Reeves et al., 1989b; Fuchs et al., 1988). DNA from HPV types 16 and 18 have been found in invasive cancers or cell lines derived from these lesions (Pater and Pater, 1985) and in some instances the viral DNA was shown to be integrated into cellular DNA (Durst et al., 1987; Spence et al., 1988). In addition, experimental evidence suggests that HPV-16 is capable of transforming cells in vitro (Phelps et a/., 1988; Matlashewski et al., 1987). HPV DNA sequences, however, have not been detected in all cases of invasive cervical cancer, and the proportions of HPV-positive cases vary in different reports (Reeves et al., 19896). This suggests that infection with the viruses is not a necessary cause for the disease and that other variables may be involved in the disease process. A number of other risk factors for cervical cancer have been identified. Some of these factors, such as number of sexual partners, age at first intercourse, and male partner sexual promiscuity, most likely represent traits of sexual behavior associated with acquiring venereally transmitted agents including HPVs. Other risk factors for cervical cancer, such as smoking (Brinton et al., 1986b; Winkelstein, 1986), oral contraceptive use (Brinton et al., 1986a; WHO, 1985; Beral et al., 1988), and number of pregnancies (Brinton et al., 1987; Brinton et al., 1989) could promote carcinogenesis by increasing the probability of oncogenic mutations or by facilitating the growth of cancer cells through an altered physiological environment. This latter mechanism might be operative in an interaction between oral contraceptives and HPVs in cervical cancer development since the transcriptional regulatory regions of HPV- 16 DNA contain hormone-recognition elements (Gloss et al., 1987). In addition, transformation of cells in vitro with the viral DNA is enhanced by hormones (Pater et al., 1988). In the present study we examined the
interaction between oral contraceptive use, as well as other risk factors, and HPV in invasive cervical cancer by recruiting cases, testing biopsies of the cancers for HPV DNA, and comparing virus-positive and -negative cases with respect to oral contraceptive use and other cancer-associated risk factors. SUBJECTS, MATERIAL AND METHODS
Subject selection Cases consisted of all women with newly diagnosed microinvasive or invasive cervical cancer diagnosed at the Panamanian National Oncology Institute during the period July 1985 to June 1987. To this end, a rapid ascertainment system was developed whereby physicians reported any new diagnosis of cervical cancer to the study personnel. This was complemented by daily review of all hospital admissions, screening of pathology departments, and review of monthly tumor registry records to assure completeness of ascertainment. Approximately 90% of all invasive cervical cancer cases in the country are referred to the Oncology Institute (Reeves et al., 1984). All cases were histologically confirmed. Patients who were not newly diagnosed, had received prior treatment for cervical cancer, were older than 69 years, or had not resided in Panama for 6 months or longer were ineligible. As part of this study, male partners of sexually monogamous women were also recruited. Sexually monogamous women were requested to provide location information for their partners. These male partners were then visited at their home or work place and asked to participate in the study. Information collected All subjects were administered a standardized interview (average length = 60 min) by trained personnel with prior interviewing experience. Questions were asked using simple and colloquial terminology which reflected the language of the respondents to obtain information regarding demographic characteristics, sexual and reproductive behavior, smoking behavior, and family history of cancer. In addition, a biopsy of the cervical lesion was obtained. A total of 220 subjects were eligible for the study. Interviews were obtained from 217 (99%) of them. Three were not interviewed due to hearing/language problems. Additionally, 14 (6%)subjects had no biopsy taken and 6 (3%) had insufficient DNA material obtained from the samples to permit testing. Thus, 197 subjects were available for analysis. Of these 197 subjects, 59 reported being monogamous. The male partners of 45 (76%) of the 59 monogamous women were located and agreed to participate in the study. Each was administered a 25-minute questionnaire which obtained information on their sexual behavior and other factors.
Received December 9, 1989 and in revised form January 29, 1990
861
HPV, OC AND CERVICAL CANCER
Processing of biopsies Biopsy specimens were frozen in liquid nitrogen and transported to the Gorgas Memorial Laboratory where they were stored at - 70°C until shipment on dry ice to McMaster University. DNA was extracted from the biopsies as detailed elsewhere (Prakash et al., 1985). To detect HPV-16 and 18,20 Fg of DNA were digested with BamHI, electrophoresed in 1% agarose gels, and transferred to nitrocellulose membrane filters using a modification of the Southern blot method (Wahl et al., 1979). The DNA on the filters was then hybridized (at -40°C) with plasmid DNA containing HPV-16 DNA (Diirst et al., 1983) which had been radiolabelled by the random primer method (Feinberg and Vogelstein, 1983). After hybridization, the filters were washed at - 20°C. After autoradiography, the nitrocellulose filters were dehybridized at 70°C for 2 hr in 5 m~ Tris HCL, PH 8.0, 0.5 mM EDTA and 0.05% sodium pyrophosphate. The filters were then hybridized at - 40°C with plasmid DNA containing HPV-18 sequences (Boshart et al., 1984) with subsequent washing at -20°C. Samples producing characteristic bands on the autoradiographs were considered positive for virus. To test specimens for HPV types other than HPV-16 and -18, and to obtain estimates of the amount of viral DNA in the samples, 2 Fg of denatured biopsy DNA were placed on nitrocellulose filter using a slot-blot apparatus (Minifold 11, Schleicher and Schuell, Keene, NH). After fixing the DNA to the filter by baking for 2 hr at 8O"C, the DNA was hybridized and washed at - 40°C with a radiolabelled 637 base-pair fragment of the L1 open reading frame of HPV-16. This region of the genome shares considerable sequence homology with other HPV types and was purified from agarose gels after electrophoretic separation of HPV-16 DNA fragments generated by digestion with BamHI and PstI (Broker, 1987). Following autoradiography, the filters were dehybridized and rehybridized, under the same conditions, with plasmid DNA containing HPV-33 sequences (Beaudenon et al., 1986). The autoradiographs of the slot blots were scanned on a BioRad (Richmond, CA) model 620 densitometer with a model 3392A integrator. The mean density value (expressed to lo6) and standard deviation of 14 slots loaded with placental DNA and probed with the L1 fragment were 2.69 and 1.0, respectively. Control values were subtracted from values of the test specimens and the resulting values that exceeded 3 standard deviations ( i . e . , 3.0 or above) were considered positive. Density values of 2.0 or above, similarly derived, were considered positive for samples probed with HPV-33 DNA. For purposes of analysis, cases were considered positive if their biopsy specimen was positive in any of these hybridization tests for HPV DNA.
Statistical methods The Pearson product-moment correlation coefficient was used to determine the association between various exposures (Zar, 1974). The relative risk (RR), as estimated by the odds ratio, was the measure of association used to determine the relationship between risk factors and HPV detection. Unconditional logistic regression (Breslow and Day, 1980) was performed to obtain maximum likelihood estimates of the RRs, while adjusting for confounding factors. To determine the statistical significance of these RRs, 95% confidence intervals (CI) were calculated, with a lower level of 1.0 representing significance at the p < 0.05 level. The significance of trends across multiple levels of an exposure was assessed by categorizing the variables of interest and treating the scored variable as a continuous variable in the regression model. RESULTS
Sixty-one percent of cases had HPV DNA detected by at
least one assay system. Given the interrelatedness of the risk factors for cervical cancer, we determined the risk of HPV positivity by each exposure variable while simultaneously adjusting for the others. Adjustment was made for age, education, number of sexual partners, age at first intercourse, number of pregnancies, Pap smear screening history, duration of smoking, and duration of oral contraceptive use. There was no need to control for barrier methods of contraception or vaginal spermicide use given the very low lifetime prevalence of use of these methods in our study population (0, 3, and 6 women reported diaphragm, condom, and spermicide use, respectively). The relative risk of HPV positivity for various exposures are shown in Table I. Risk of positivity increased significantly with age. Those older than 50 years were 4.6 times more likely to be HPV-positive when compared with women less than 35 years old. Risk of HPV detection was also higher among women residing in urban areas (RR = 2.8; 95% CI = 1.3,6.3) and those with primary education (1-6 years) compared with women with no formal education (RR = 3.1; 95% CI = 1.3,7.7). Smokers of more than 20 years were at one-third the risk of HPV positivity compared to non-smokers, an effect which was only marginally significant (95% CI = 0.1,l.O). Positive associations with HPV were detected for number of sexual partners of the women and of the male partners of sexually monogamous women, but these effects were not statistically significant at the 5% level. Age at first intercourse, number of pregnancies, and history of Pap smear screening were not related to HPV positivity in this population of cases. Oral contraceptive use was associated with an increased risk of positivity to HPV (Table 11). When calculating the RRs of HPV detection among oral contraceptive users, adjustment was made for all exposures found to be associated with HPV positivity in Table I. This included age at diagnosis, education, residency, number of sexual partners, and duration of smoking. Women who were recent oral contraceptive users were at a 2.3-fold increased risk of positivity compared with nonusers, and long-term users were at a 2.9-fold increased risk of positivity compared with non-users (p for trend = 0.05). Given the high correlation (r = 0.95) between these two variables, it was not possible to determine the independent effects of these factors. Ninety-three percent of recent users (defined as women who used oral contraceptives at the time of interview or had ceased using oral contraceptives within a year) were also long-term users (more than 4 years of use), and 95% of shortterm users were also ex-users. When latency of oral contraceptive use was examined, no strong pattern was observed, although women with less than 10 years since first use were at a non-significant 1.9-fold increased risk of HPV detection compared with women who reported never using oral contraceptives. The association between histological class or clinical stage of the tumors and HPV positivity was also examined. No differences in distribution of cases by HPV status were noted with regard to these two variables. DISCUSSION
The present analysis compared risk-factor profiles among invasive cancer cases which were positive and negative for HPV DNA. Both Southern and slot-blot hybridization techniques under relaxed stringency and probes for HPV-16, HPV18, and HPV-33 were used, theoretically enabling us to detect infection with a wide range of HPV types. HPV DNA was detected in 61% of the cases, which is similar to the occur-
862
HILDESHEILM ET AL.
TABLE I - RISK OF HPV DETECTION BY SELECTED RISK FACTORS HPV + /HPV -
Crude RR
Age at diagnosis: 1 .o =18yrs 16, 17 yrs 25118 0.8 54/34 1 .o < = 1 5 yrs Trend p = 0.95 Number of pregnancies: 0-2 1319 1.0 3-8 60144 0.9 29 47124 1.4 Trend p = 0.34 History of pap smear: Yes 67/44 1 .o No 4913 1 1 .o Missing 412
TABLE ll - RISK OF HF'V DETECTION BY ORAL CONTRACEPTIVE USE
HPV+/HPV-
ADJ RR1
1 .o 1.4 4.64
p = 0.003 1 .o 2.84 1 .o
3.14 1.5 p = 0.32 1 .o 0.9 0.33
Recency of OC use: Never Ex-user User Missing Trend Duration of OC use: Never 1 4 yrs >4 vrs Misiing Trend Time since first use: Never 1-10 yrs >lo yrs Missing Trend
90160 18/12 1114 111
90160 1217 1715 115
90160 1316 16/10 111
CrudeRR
ADJRR1
1 .o 1 .o 1.8
1 .o 1.2 2.3
p = 0.37
p = 0.24
1 .o 1.1 2.3
1 .o 1.7 2.9
p = 0.11
p = 0.05
1 .o 1.4 1.1
1 .o 1.9 1.3
p = 0.68
p = 0.51
'Adjusted for age, education, residency, number of sexual'partners of the women
and smoking.
p = 0.06
HPV. Hormones enhance the transcription of the HPV-16 genome (Gloss et al., 1987). Increased expression of the HPV genome resident in SiHa cells has been noted after treatment with estrogen (Mitrani-Rosenbaum et al., 1988); thus the observed interaction might result from the increased ability of the p = 0.25 test to detect HPV DNA in oral contraceptive users. However, it has also previously been shown that the frequency of in vitro 1 .o transformation of cells by HPV-16 DNA can be enhanced by 1.1 hormones (Pater et al., 1988). This indicates that oral contra1.9 ceptive use may also be a causal risk factor for cervical cancer, simultaneously enhancing the expression of the virus and inp = 0.82 creasing the likelihood that HPV will induce transformation of cervical epithelial cells. In addition, oral contraceptive use 1 .o might increase the risk of disease by hormonally increasing the 0.7 susceptibility of epithelial cells to HPV infection. Finally, oral 1 .o contraceptive users could be at higher risk of infection given p = 0.94 their sexual practices. Although we have taken sexual behavior into account when computing the relative risk of infection by 1 .o oral contraceptive use, it is possible that some aspect of sexual 0.9 behavior has not been appropriately controlled for and that oral 1.3 contraceptive use functions as a proxy for this behavior. As an p = 0.48 example, we were unable to adjust for the sexual behavior of the male partner in the main analysis and it is possible that the 1 .o observed oral contraceptive effect reflects confounding by this 1.2 factor. Also, women might have under-reported their number of sexual partners, thus biasing the risk estimate for number of 'Adjusted for all exposures listed above plus duration of OC use.-*Includes only sexual partners downward. This in turn might have resulted in male partners of sexually monogamous ~ o m e n . - ~ = p 0.05.-4p < 0.05. incomplete adjustment for sexual behavior when number of sexual partners was included in analysis as a potential confounder. rences noted in other studies (Meanwell et al., 1987; Fuchs et In a previous publication from our study (Reeves et al., al., 1988). 1989a), cases and controls who had been tested for HPV DNA When comparing HPV-positive and -negative cases, no dif- by a filter in situ hybridization (FISH) technique using scrape ferences were detected with respect to age at first coitus, num- material were compared. No interaction between HPV and oral ber of pregnancies, and Pap smear screening history. These contraceptive use was detected in that analysis, and no assoresults are similar to those observed by Meanwell et al. (1987), ciation between HPV and oral contraceptive use was found and indicate that the above factors, if associated with cervical among cases. This apparent contradiction indicates that there cancer, might be independent from HPV infection. are differences in the various testing techniques and/or collecAn interesting finding was an association of oral contracep- tion methods. Hallam et al. (1989) recently reported the results tive use with HPV positivity. Women who used oral contra- of dot-blot DNA hybridization for HPV performed on paired ceptives in this study were found to be more likely to be pos- scrapes and biopsies of 100 women. Although a similar proitive for HPV DNA. Although based on small numbers, this portion of the women were positive, regardless of the sampling finding suggests that oral contraceptives might interact with method used (51% of scrapes were positive compared with 1 .o 0.9 1.8
863
HPV. OC A N D CERVICAL CANCER
50% of biopsies), only 34% of women were positive by both methods, whereas 67% were positive by one or the other sampling technique. The fact that in our study an association between HPV and oral contraceptive use is detected when one assay system is used (biopsies + Southern) but not another (scrapes FISH), indicates that the HPV/oral contraceptive association found in this analysis is a reflection of increased detectability of HPV DNA among oral contraceptive users. If this were true, one would expect recency of use to be more strongly associated with HPV positivity than duration of use. No support for this was found, although it was difficult to distinguish independent effects of recency and duration of use given their high correlation (r = 0.95). As previously reported (Acs et al., 1989), women from predominantly urban areas of Panama are more likely to be positive for HPV DNA than women from rural areas. It is interesting to note, however, that the incidence of cervical cancer in Panama is lowest in these urban areas, indicating that factors other than HPV might account for the higher incidence of cervical cancer in rural Panama. Differential screening might also partially explain the higher incidence rate in rural areas. Indeed, we observed in our analysis that 66%of subjects from urban areas reported ever having had a Pap smear, compared with 52% of subjects from rural areas. In past studies, education has consistently been inversely associated with cervical cancer (Brinton and Fraumeni, 1986). Interestingly, when examining the risk of infection with HPV by educational status among cases, we observed a positive association between the two variables which persisted after adjustment for age, residence, sexual and reproductive behavior, smoking, and Pap smear screening history. Even though a clear dose-response pattern was not observed (p for trend = 0.32), both women with 1-6 and those with 7 or more years of education were at increased risk of being HPVpositive compared with women reporting no education. This suggests that low education is not a risk factor for HPV infec-
+
tion, but may possibly represent a risk for other carcinogens likely to be operating in the genesis of cervical neoplasia. An unexpected finding of this analysis was the marginally significant negative association between smoking and detectability of HPV DNA. Women who reported having smoked for more than 20 years were less likely than others to be positive for HPV. This finding suggests that smoking, if indeed a causal factor in the genesis of cervical cancer, acts through a causal pathway different from that of HPV. Given the low prevalence of smoking among Panamanian women and the small number of long-term smokers in our study, it is also possible that the observed association is due to chance. The results presented above add evidence in support of a multifactorial model of cervical cancer causation. Most notably, an association between HPV and oral contraceptive use has been observed. Although based on small numbers of extensive users, this association suggests an interaction between these two variables. Whether the interaction is due to increased detectability of HPV DNA among oral contraceptive users or an interaction between HPV, oral contraceptives, and cervical cancer is not known. Future investigations should address this important issue. ACKNOWLEDGEMENTS
The authors thank Mr. J . Friedbauer for his programming assistance. This study was supported by a contract and grant from the National Cancer Institute, National Institutes of Health and by a grant from the National Cancer Institute of Canada. Note: The protocol used in this study was reviewed and approved by the Gorgas Memorial Laboratory Human Subjects Committee and the Office of Protection of Research Subjects, National Institutes of Health, US. All participants were volunteers who gave informed consent.
REFERENCES Acs, J., HILDESHEIM, A,, REEVES,W.C., BRENES,M.M., BRINTON, L.A., LAVERY, C., DE LA GUARDIA, M.E., GODOY,J . and RAWLS,W.E., Regional distribution of human papillomavirus DNA and other risk factors for invasive cervical cancer in Panama. Cancer Res., 49, 5725-5729 (1989). BEAUDENON, S., KREMSDORF, D., CROISSANT, 0.. JABLONSKA, S., WAIN-HOBSON, S. and ORTH,G., A novel type of human papillomavirus associated with genital neoplasias. Nature (Lond.), 321, 24G249 (1986). BERAL,V., HANNAFORD, P. and KAY, C., Oral contraceptive use and malignancies of the genital tract. Results from the Royal College of General Practitioners’ oral contraceptive study. Lancet, 11, 1331-1335 (1988). BOSHART,M., GISSMANN,L . , IKENBERG,H., KLEINHEINZ,A , , SCHEURLEN, W. and ZUR HAUSEN,H., A new type of papillomavirus DNA, its presence in genital cancer biopsies and in cell lines derived from cervical cancer. EMBO J., 3, 1151-1157 (1984). BRESLOW,N.E. and DAY,N.E., Statistical methods in cancer research, Vol. I. The analysis of case-control studies, IARC Scientific Publication 82, IARC, Lyon (1980). BRINTON,L.A. and FRAUMENI, J.F., JR., Epidemiology of uterine cervical cancer. J. chron. Dis., 39, 1051-1065 (1986). BRINTON,L.A., HAMMAN,R.F., HUGGINS,G.R., LEHMAN,H.F., J.F., JR., Sexual and reproLEVINE,R.S., MALLIN,K. and FRAUMENI, ductive risk factors for invasive squamous-cell cervical cancer. J. nut. Cancer Insf., 79, 23-30 (1987). BRINTON, L.A., HUGGINS,G.R., LEHMAN,H.F., MALLIN,K., SAVITZ, J. and HOOVER,R., Long-term use of D.A., TRAPIDO,E., ROSENTHAL, oral contraceptives and risk of invasive cervical cancer. Znr. J . Cancer, 38, 339-344 (1986~). BRINTON,L.A., REEVES,W.C., BRENES,M.M., HERRERO, R., DE BRITTON,R.C., GAITAN,E., TENORIO,F., GARCIA,M. and RAWLS,W.E., Parity as a risk factor for cervical cancer. Amer. J. Epidemiol., (1989) (In press).
BRINTON, L.A., SCHAIRER, C., HAENSZEL, W., STOLLEY,P., LEHMAN, H.F., LEVINE,R. and SAVITZ,D.A., Cigarette smoking and invasive cervical cancer. J. Amer. med. Assoc., 255, 3265-3269 (19866). BROKER,T.R., Structure and gene expression of papillomaviruses. Obster. Gynecol. Clin. North Amer., 14, 329-348 (1987). DURST, M., GISSMANN, L. and IKENBERG,H., A papillomavirus DNA from a cervical carcinoma and its prevalence in cancer biopsies from different geographic areas. Proc. nut. Acad. Sci. (Wash.),80, 3812-3815 (1983). DURST,M., CROCE,C.M., GISSMANN, L., SCHWARZ, E. and HUEBNER, K . , Papillomavirus sequences integrate near cellular oncogenes in some cervical carcinomas. Proc. nut. Acad. Sci. (Wash.), 84, 1070-1074 (1987). FEINBERG, A.P. and VOGELSTEIN, B . , A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity. Anal. Biochem., 123, 6-13 (1983). FUCHS,P.G., GIRARDI,F. and PFISTER,H., Human papillomavirus DNA in normal, metaplastic, preneoplastic and neoplastic epithelia of the cervix uteri. Znt. J. Cancer, 41, 41-45 (1988). GLOSS,B., BERNARD, H.U., SEEDORF,K. and KLOCK,G., The upstream regulatory region of the human-papilloma-virus- 16 contains an E2protein-independent enhancer which is specific for cervical carcinoma cells and regulated by glucocorticoid hormones. EMBO J., 6, 3735-3743 (1987). M., Detection and HALLAM, N., GIBSON,P., GREEN,J. and CHARNOCK, typing of human papillomavirus infection of the uterine cervix by dot-blot hybridisation: comparison of scrapes and biopsies. J. med. Virol., 27, 317-321 (1989). MATLASHEWSKI, G., SCHNEIDER, J., BANKS,L., JONES,N., MURRAY, A. and CRAWFORD,L., Human-papillomavirus-type- 16 DNA cooperates with activated ras in transforming primary cells. EMBO J., 6, 1741-1746 ( 1987).
864
HILDESHEIM ET AL.
MEANWELL, C.A., Cox, M.F., BLACKLEDGE, G. and MAITLAND, N.J., HPV-16 DNA in normal and malignant cervical epithelium: implications for the aetiology and behaviour of cervical neoplasia. Lancer, I, 703-707 ( 1987). MITRANI-ROSENBAUM, S., TSVIELI,R. and TUR-KASPA, R. Estrogen but not glucocorticoids stimulate HPV-16 transcription in SIHA cells, Seventh International Papillomavirus Workshop, p. 82, Nice, France (1988). PATER,M.M. and PATER,A,, Human-papillomavirus-types-16 and -18 sequences in carcinoma cell lines of the cervix. Virology, 145, 313-318 (1985). PATER,M.M., HUGHES,G.A., HYSLOP,D.E., NAKSHATRI, H. and PATER,A., Glucocorticoid-dependent oncogenic transformation by type 16 but not type 11 human papilloma virus DNA. Nature fL0nd.L 335, 832835 (1988). ~K. and H ~ ~~P.M., ne ~~ human~~ pHELPS,w.c., yEE,c.L., M papillomavirus-type-E7 gene encodes transactivation and transformation functions similar to those of adenovirus E1A. Cell, 53, 539-547 (1988). S . S . , REEVES,W.C., SISSON,G.R., BRENES,M., GODOY,J., PRAKASH, BACCHETTI, S., DE BRITTON,R.C. and RAWLS,W E . , Herpes simplex virus type 2 and human papillomavirus type 16 in cervicitis, dysplasia and invasive cervical carcinoma. Inr. J. Cancer, 35, 51-57 (1985). REEVES,W.C., BRENES,M.M., DE BRITTON,R.C., VALDES,P.F. and
JOPLIN,C.F.B., Cervical cancer in the Republic of Panama. Amer. J. Epidemiol., 119, 714-724 (1984). REEVES,w , c , , BRINTON,L.A., GARCIA,M,, BRENES,M,M,, HERRERO, R., GAITAN,E., TENORIO,F., DE BRITTON,R.C. and RAWLS,W.E., Human papillomavirus infection and cervical cancer in Latin America. New Engl. J. Med., 320, 1437-1441 (1989~). REEVES,W.C., RAWLS,W.E. and BRINTON,L.A., Epidemiology of genital papillomaviruses and cervical cancer. Rev. infect. Dis., 11, 426-439 (1989b). SPENCE,R.P., MURRY,A., BANKS,L., KELLAND, L.R. and CRAWFORD, L.,,AnalYSiS of human papillomavirus sequences in cell lines recently derived from cervical cancers. cheer Res., 48, 324-328 (1988). WAHL,G.M., STERN,M. and STARK,G.R., Efficient transfer of large DNA fragments from agarose gels to diazobenzyloxymethyl-paper and rapid hybridization ~ , using dextran sulfate. Proc. nut. Acad. Sci. (Wash.), ' ,6~ 3683-3687 WHO, Invasive cervical cancer and combined oral contraceptives. Brit. med. J . s 2909 961-965 (1985). WINKELSTEIN, W., Cigarette smoking and cancer of the uterine cervix. Banbury Rep., 23, 329-341 (1986). ZAR,J.H., Biostaristical analysis. Prentice-Hall, Englewood Cliffs, New Jersey (1974).
Publication of the International Union Against Cancer Publication de I'Union lnternationale Contre le Cancer
ASSOCIATION OF ORAL CONTRACEPTIVE USE AND HUMAN PAPILLOMAVIRUSES IN INVASIVE CERVICAL CANCERS Allan HILDESHEIM', William c. REEVES~, Louise A. BRINTON',carol LAVERY3, Maria BRENES4, Maria Elena DE LA GUARDIA~, Julio GODOY'and William E. RAWLS~ 'Environmental Epidemiology Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD; 2Division of Viral Diseases, Center for Infectious Diseases, Center for Disease Control, Atlanta, GA, USA; 3Molecular Virology and Immunology Program, Dept. of Pathology, McMaster University, Hamilton, Ontario, Canada; 4Division of Epidemiology, Gorgas Memorial Laboratoiy, Panama City; 51nstituto Oncologico Nacional, Panama City, Republic of Panama. In a study of 197 cases of histologically confirmed invasive cervical cancer, 61% of biopsies were positive for human papillomavirus (HPV) D N A by Southern or dot-blot hybridization. An association between detection of HPV D N A and oral contraceptiveuse was observed when HPV-positive and -negative cases were compared. Women reporting recent or longterm (>4 yrs) oral contraceptive use were at 2.3 and 2.9-fold increased risks of HPV positivity, respectively. An increased risk of HPV positivity was also associated with formal education and with urban residence, while long-term smoking was negatively associated with HPV detection. A non-significant trend of increasing risk of HPV positivity with increasing number of sexual partners of the women and of the male partners of monogamous women was observed. Detection of HPV DNA was not associated with other cervical cancer risk factors examined, including age at first coitus, number of pregnancies, and Pap smear screening history. Our findings suggest either an interaction between HPV infection and oral contraceptive use in the genesis of cervical cancer or an increased expression of HPV genome in neoplasms of oral contraceptive users. These observations also support a multifactorial model of cervical cancer causation.
There is an expanding body of evidence linking human papillomaviruses (HPV) to cervical neoplasms (Reeves et al., 1989b; Fuchs et al., 1988). DNA from HPV types 16 and 18 have been found in invasive cancers or cell lines derived from these lesions (Pater and Pater, 1985) and in some instances the viral DNA was shown to be integrated into cellular DNA (Durst et al., 1987; Spence et al., 1988). In addition, experimental evidence suggests that HPV-16 is capable of transforming cells in vitro (Phelps et a/., 1988; Matlashewski et al., 1987). HPV DNA sequences, however, have not been detected in all cases of invasive cervical cancer, and the proportions of HPV-positive cases vary in different reports (Reeves et al., 19896). This suggests that infection with the viruses is not a necessary cause for the disease and that other variables may be involved in the disease process. A number of other risk factors for cervical cancer have been identified. Some of these factors, such as number of sexual partners, age at first intercourse, and male partner sexual promiscuity, most likely represent traits of sexual behavior associated with acquiring venereally transmitted agents including HPVs. Other risk factors for cervical cancer, such as smoking (Brinton et al., 1986b; Winkelstein, 1986), oral contraceptive use (Brinton et al., 1986a; WHO, 1985; Beral et al., 1988), and number of pregnancies (Brinton et al., 1987; Brinton et al., 1989) could promote carcinogenesis by increasing the probability of oncogenic mutations or by facilitating the growth of cancer cells through an altered physiological environment. This latter mechanism might be operative in an interaction between oral contraceptives and HPVs in cervical cancer development since the transcriptional regulatory regions of HPV- 16 DNA contain hormone-recognition elements (Gloss et al., 1987). In addition, transformation of cells in vitro with the viral DNA is enhanced by hormones (Pater et al., 1988). In the present study we examined the
interaction between oral contraceptive use, as well as other risk factors, and HPV in invasive cervical cancer by recruiting cases, testing biopsies of the cancers for HPV DNA, and comparing virus-positive and -negative cases with respect to oral contraceptive use and other cancer-associated risk factors. SUBJECTS, MATERIAL AND METHODS
Subject selection Cases consisted of all women with newly diagnosed microinvasive or invasive cervical cancer diagnosed at the Panamanian National Oncology Institute during the period July 1985 to June 1987. To this end, a rapid ascertainment system was developed whereby physicians reported any new diagnosis of cervical cancer to the study personnel. This was complemented by daily review of all hospital admissions, screening of pathology departments, and review of monthly tumor registry records to assure completeness of ascertainment. Approximately 90% of all invasive cervical cancer cases in the country are referred to the Oncology Institute (Reeves et al., 1984). All cases were histologically confirmed. Patients who were not newly diagnosed, had received prior treatment for cervical cancer, were older than 69 years, or had not resided in Panama for 6 months or longer were ineligible. As part of this study, male partners of sexually monogamous women were also recruited. Sexually monogamous women were requested to provide location information for their partners. These male partners were then visited at their home or work place and asked to participate in the study. Information collected All subjects were administered a standardized interview (average length = 60 min) by trained personnel with prior interviewing experience. Questions were asked using simple and colloquial terminology which reflected the language of the respondents to obtain information regarding demographic characteristics, sexual and reproductive behavior, smoking behavior, and family history of cancer. In addition, a biopsy of the cervical lesion was obtained. A total of 220 subjects were eligible for the study. Interviews were obtained from 217 (99%) of them. Three were not interviewed due to hearing/language problems. Additionally, 14 (6%)subjects had no biopsy taken and 6 (3%) had insufficient DNA material obtained from the samples to permit testing. Thus, 197 subjects were available for analysis. Of these 197 subjects, 59 reported being monogamous. The male partners of 45 (76%) of the 59 monogamous women were located and agreed to participate in the study. Each was administered a 25-minute questionnaire which obtained information on their sexual behavior and other factors.
Received December 9, 1989 and in revised form January 29, 1990
861
HPV, OC AND CERVICAL CANCER
Processing of biopsies Biopsy specimens were frozen in liquid nitrogen and transported to the Gorgas Memorial Laboratory where they were stored at - 70°C until shipment on dry ice to McMaster University. DNA was extracted from the biopsies as detailed elsewhere (Prakash et al., 1985). To detect HPV-16 and 18,20 Fg of DNA were digested with BamHI, electrophoresed in 1% agarose gels, and transferred to nitrocellulose membrane filters using a modification of the Southern blot method (Wahl et al., 1979). The DNA on the filters was then hybridized (at -40°C) with plasmid DNA containing HPV-16 DNA (Diirst et al., 1983) which had been radiolabelled by the random primer method (Feinberg and Vogelstein, 1983). After hybridization, the filters were washed at - 20°C. After autoradiography, the nitrocellulose filters were dehybridized at 70°C for 2 hr in 5 m~ Tris HCL, PH 8.0, 0.5 mM EDTA and 0.05% sodium pyrophosphate. The filters were then hybridized at - 40°C with plasmid DNA containing HPV-18 sequences (Boshart et al., 1984) with subsequent washing at -20°C. Samples producing characteristic bands on the autoradiographs were considered positive for virus. To test specimens for HPV types other than HPV-16 and -18, and to obtain estimates of the amount of viral DNA in the samples, 2 Fg of denatured biopsy DNA were placed on nitrocellulose filter using a slot-blot apparatus (Minifold 11, Schleicher and Schuell, Keene, NH). After fixing the DNA to the filter by baking for 2 hr at 8O"C, the DNA was hybridized and washed at - 40°C with a radiolabelled 637 base-pair fragment of the L1 open reading frame of HPV-16. This region of the genome shares considerable sequence homology with other HPV types and was purified from agarose gels after electrophoretic separation of HPV-16 DNA fragments generated by digestion with BamHI and PstI (Broker, 1987). Following autoradiography, the filters were dehybridized and rehybridized, under the same conditions, with plasmid DNA containing HPV-33 sequences (Beaudenon et al., 1986). The autoradiographs of the slot blots were scanned on a BioRad (Richmond, CA) model 620 densitometer with a model 3392A integrator. The mean density value (expressed to lo6) and standard deviation of 14 slots loaded with placental DNA and probed with the L1 fragment were 2.69 and 1.0, respectively. Control values were subtracted from values of the test specimens and the resulting values that exceeded 3 standard deviations ( i . e . , 3.0 or above) were considered positive. Density values of 2.0 or above, similarly derived, were considered positive for samples probed with HPV-33 DNA. For purposes of analysis, cases were considered positive if their biopsy specimen was positive in any of these hybridization tests for HPV DNA.
Statistical methods The Pearson product-moment correlation coefficient was used to determine the association between various exposures (Zar, 1974). The relative risk (RR), as estimated by the odds ratio, was the measure of association used to determine the relationship between risk factors and HPV detection. Unconditional logistic regression (Breslow and Day, 1980) was performed to obtain maximum likelihood estimates of the RRs, while adjusting for confounding factors. To determine the statistical significance of these RRs, 95% confidence intervals (CI) were calculated, with a lower level of 1.0 representing significance at the p < 0.05 level. The significance of trends across multiple levels of an exposure was assessed by categorizing the variables of interest and treating the scored variable as a continuous variable in the regression model. RESULTS
Sixty-one percent of cases had HPV DNA detected by at
least one assay system. Given the interrelatedness of the risk factors for cervical cancer, we determined the risk of HPV positivity by each exposure variable while simultaneously adjusting for the others. Adjustment was made for age, education, number of sexual partners, age at first intercourse, number of pregnancies, Pap smear screening history, duration of smoking, and duration of oral contraceptive use. There was no need to control for barrier methods of contraception or vaginal spermicide use given the very low lifetime prevalence of use of these methods in our study population (0, 3, and 6 women reported diaphragm, condom, and spermicide use, respectively). The relative risk of HPV positivity for various exposures are shown in Table I. Risk of positivity increased significantly with age. Those older than 50 years were 4.6 times more likely to be HPV-positive when compared with women less than 35 years old. Risk of HPV detection was also higher among women residing in urban areas (RR = 2.8; 95% CI = 1.3,6.3) and those with primary education (1-6 years) compared with women with no formal education (RR = 3.1; 95% CI = 1.3,7.7). Smokers of more than 20 years were at one-third the risk of HPV positivity compared to non-smokers, an effect which was only marginally significant (95% CI = 0.1,l.O). Positive associations with HPV were detected for number of sexual partners of the women and of the male partners of sexually monogamous women, but these effects were not statistically significant at the 5% level. Age at first intercourse, number of pregnancies, and history of Pap smear screening were not related to HPV positivity in this population of cases. Oral contraceptive use was associated with an increased risk of positivity to HPV (Table 11). When calculating the RRs of HPV detection among oral contraceptive users, adjustment was made for all exposures found to be associated with HPV positivity in Table I. This included age at diagnosis, education, residency, number of sexual partners, and duration of smoking. Women who were recent oral contraceptive users were at a 2.3-fold increased risk of positivity compared with nonusers, and long-term users were at a 2.9-fold increased risk of positivity compared with non-users (p for trend = 0.05). Given the high correlation (r = 0.95) between these two variables, it was not possible to determine the independent effects of these factors. Ninety-three percent of recent users (defined as women who used oral contraceptives at the time of interview or had ceased using oral contraceptives within a year) were also long-term users (more than 4 years of use), and 95% of shortterm users were also ex-users. When latency of oral contraceptive use was examined, no strong pattern was observed, although women with less than 10 years since first use were at a non-significant 1.9-fold increased risk of HPV detection compared with women who reported never using oral contraceptives. The association between histological class or clinical stage of the tumors and HPV positivity was also examined. No differences in distribution of cases by HPV status were noted with regard to these two variables. DISCUSSION
The present analysis compared risk-factor profiles among invasive cancer cases which were positive and negative for HPV DNA. Both Southern and slot-blot hybridization techniques under relaxed stringency and probes for HPV-16, HPV18, and HPV-33 were used, theoretically enabling us to detect infection with a wide range of HPV types. HPV DNA was detected in 61% of the cases, which is similar to the occur-
862
HILDESHEILM ET AL.
TABLE I - RISK OF HPV DETECTION BY SELECTED RISK FACTORS HPV + /HPV -
Crude RR
Age at diagnosis: 1 .o =18yrs 16, 17 yrs 25118 0.8 54/34 1 .o < = 1 5 yrs Trend p = 0.95 Number of pregnancies: 0-2 1319 1.0 3-8 60144 0.9 29 47124 1.4 Trend p = 0.34 History of pap smear: Yes 67/44 1 .o No 4913 1 1 .o Missing 412
TABLE ll - RISK OF HF'V DETECTION BY ORAL CONTRACEPTIVE USE
HPV+/HPV-
ADJ RR1
1 .o 1.4 4.64
p = 0.003 1 .o 2.84 1 .o
3.14 1.5 p = 0.32 1 .o 0.9 0.33
Recency of OC use: Never Ex-user User Missing Trend Duration of OC use: Never 1 4 yrs >4 vrs Misiing Trend Time since first use: Never 1-10 yrs >lo yrs Missing Trend
90160 18/12 1114 111
90160 1217 1715 115
90160 1316 16/10 111
CrudeRR
ADJRR1
1 .o 1 .o 1.8
1 .o 1.2 2.3
p = 0.37
p = 0.24
1 .o 1.1 2.3
1 .o 1.7 2.9
p = 0.11
p = 0.05
1 .o 1.4 1.1
1 .o 1.9 1.3
p = 0.68
p = 0.51
'Adjusted for age, education, residency, number of sexual'partners of the women
and smoking.
p = 0.06
HPV. Hormones enhance the transcription of the HPV-16 genome (Gloss et al., 1987). Increased expression of the HPV genome resident in SiHa cells has been noted after treatment with estrogen (Mitrani-Rosenbaum et al., 1988); thus the observed interaction might result from the increased ability of the p = 0.25 test to detect HPV DNA in oral contraceptive users. However, it has also previously been shown that the frequency of in vitro 1 .o transformation of cells by HPV-16 DNA can be enhanced by 1.1 hormones (Pater et al., 1988). This indicates that oral contra1.9 ceptive use may also be a causal risk factor for cervical cancer, simultaneously enhancing the expression of the virus and inp = 0.82 creasing the likelihood that HPV will induce transformation of cervical epithelial cells. In addition, oral contraceptive use 1 .o might increase the risk of disease by hormonally increasing the 0.7 susceptibility of epithelial cells to HPV infection. Finally, oral 1 .o contraceptive users could be at higher risk of infection given p = 0.94 their sexual practices. Although we have taken sexual behavior into account when computing the relative risk of infection by 1 .o oral contraceptive use, it is possible that some aspect of sexual 0.9 behavior has not been appropriately controlled for and that oral 1.3 contraceptive use functions as a proxy for this behavior. As an p = 0.48 example, we were unable to adjust for the sexual behavior of the male partner in the main analysis and it is possible that the 1 .o observed oral contraceptive effect reflects confounding by this 1.2 factor. Also, women might have under-reported their number of sexual partners, thus biasing the risk estimate for number of 'Adjusted for all exposures listed above plus duration of OC use.-*Includes only sexual partners downward. This in turn might have resulted in male partners of sexually monogamous ~ o m e n . - ~ = p 0.05.-4p < 0.05. incomplete adjustment for sexual behavior when number of sexual partners was included in analysis as a potential confounder. rences noted in other studies (Meanwell et al., 1987; Fuchs et In a previous publication from our study (Reeves et al., al., 1988). 1989a), cases and controls who had been tested for HPV DNA When comparing HPV-positive and -negative cases, no dif- by a filter in situ hybridization (FISH) technique using scrape ferences were detected with respect to age at first coitus, num- material were compared. No interaction between HPV and oral ber of pregnancies, and Pap smear screening history. These contraceptive use was detected in that analysis, and no assoresults are similar to those observed by Meanwell et al. (1987), ciation between HPV and oral contraceptive use was found and indicate that the above factors, if associated with cervical among cases. This apparent contradiction indicates that there cancer, might be independent from HPV infection. are differences in the various testing techniques and/or collecAn interesting finding was an association of oral contracep- tion methods. Hallam et al. (1989) recently reported the results tive use with HPV positivity. Women who used oral contra- of dot-blot DNA hybridization for HPV performed on paired ceptives in this study were found to be more likely to be pos- scrapes and biopsies of 100 women. Although a similar proitive for HPV DNA. Although based on small numbers, this portion of the women were positive, regardless of the sampling finding suggests that oral contraceptives might interact with method used (51% of scrapes were positive compared with 1 .o 0.9 1.8
863
HPV. OC A N D CERVICAL CANCER
50% of biopsies), only 34% of women were positive by both methods, whereas 67% were positive by one or the other sampling technique. The fact that in our study an association between HPV and oral contraceptive use is detected when one assay system is used (biopsies + Southern) but not another (scrapes FISH), indicates that the HPV/oral contraceptive association found in this analysis is a reflection of increased detectability of HPV DNA among oral contraceptive users. If this were true, one would expect recency of use to be more strongly associated with HPV positivity than duration of use. No support for this was found, although it was difficult to distinguish independent effects of recency and duration of use given their high correlation (r = 0.95). As previously reported (Acs et al., 1989), women from predominantly urban areas of Panama are more likely to be positive for HPV DNA than women from rural areas. It is interesting to note, however, that the incidence of cervical cancer in Panama is lowest in these urban areas, indicating that factors other than HPV might account for the higher incidence of cervical cancer in rural Panama. Differential screening might also partially explain the higher incidence rate in rural areas. Indeed, we observed in our analysis that 66%of subjects from urban areas reported ever having had a Pap smear, compared with 52% of subjects from rural areas. In past studies, education has consistently been inversely associated with cervical cancer (Brinton and Fraumeni, 1986). Interestingly, when examining the risk of infection with HPV by educational status among cases, we observed a positive association between the two variables which persisted after adjustment for age, residence, sexual and reproductive behavior, smoking, and Pap smear screening history. Even though a clear dose-response pattern was not observed (p for trend = 0.32), both women with 1-6 and those with 7 or more years of education were at increased risk of being HPVpositive compared with women reporting no education. This suggests that low education is not a risk factor for HPV infec-
+
tion, but may possibly represent a risk for other carcinogens likely to be operating in the genesis of cervical neoplasia. An unexpected finding of this analysis was the marginally significant negative association between smoking and detectability of HPV DNA. Women who reported having smoked for more than 20 years were less likely than others to be positive for HPV. This finding suggests that smoking, if indeed a causal factor in the genesis of cervical cancer, acts through a causal pathway different from that of HPV. Given the low prevalence of smoking among Panamanian women and the small number of long-term smokers in our study, it is also possible that the observed association is due to chance. The results presented above add evidence in support of a multifactorial model of cervical cancer causation. Most notably, an association between HPV and oral contraceptive use has been observed. Although based on small numbers of extensive users, this association suggests an interaction between these two variables. Whether the interaction is due to increased detectability of HPV DNA among oral contraceptive users or an interaction between HPV, oral contraceptives, and cervical cancer is not known. Future investigations should address this important issue. ACKNOWLEDGEMENTS
The authors thank Mr. J . Friedbauer for his programming assistance. This study was supported by a contract and grant from the National Cancer Institute, National Institutes of Health and by a grant from the National Cancer Institute of Canada. Note: The protocol used in this study was reviewed and approved by the Gorgas Memorial Laboratory Human Subjects Committee and the Office of Protection of Research Subjects, National Institutes of Health, US. All participants were volunteers who gave informed consent.
REFERENCES Acs, J., HILDESHEIM, A,, REEVES,W.C., BRENES,M.M., BRINTON, L.A., LAVERY, C., DE LA GUARDIA, M.E., GODOY,J . and RAWLS,W.E., Regional distribution of human papillomavirus DNA and other risk factors for invasive cervical cancer in Panama. Cancer Res., 49, 5725-5729 (1989). BEAUDENON, S., KREMSDORF, D., CROISSANT, 0.. JABLONSKA, S., WAIN-HOBSON, S. and ORTH,G., A novel type of human papillomavirus associated with genital neoplasias. Nature (Lond.), 321, 24G249 (1986). BERAL,V., HANNAFORD, P. and KAY, C., Oral contraceptive use and malignancies of the genital tract. Results from the Royal College of General Practitioners’ oral contraceptive study. Lancet, 11, 1331-1335 (1988). BOSHART,M., GISSMANN,L . , IKENBERG,H., KLEINHEINZ,A , , SCHEURLEN, W. and ZUR HAUSEN,H., A new type of papillomavirus DNA, its presence in genital cancer biopsies and in cell lines derived from cervical cancer. EMBO J., 3, 1151-1157 (1984). BRESLOW,N.E. and DAY,N.E., Statistical methods in cancer research, Vol. I. The analysis of case-control studies, IARC Scientific Publication 82, IARC, Lyon (1980). BRINTON,L.A. and FRAUMENI, J.F., JR., Epidemiology of uterine cervical cancer. J. chron. Dis., 39, 1051-1065 (1986). BRINTON,L.A., HAMMAN,R.F., HUGGINS,G.R., LEHMAN,H.F., J.F., JR., Sexual and reproLEVINE,R.S., MALLIN,K. and FRAUMENI, ductive risk factors for invasive squamous-cell cervical cancer. J. nut. Cancer Insf., 79, 23-30 (1987). BRINTON, L.A., HUGGINS,G.R., LEHMAN,H.F., MALLIN,K., SAVITZ, J. and HOOVER,R., Long-term use of D.A., TRAPIDO,E., ROSENTHAL, oral contraceptives and risk of invasive cervical cancer. Znr. J . Cancer, 38, 339-344 (1986~). BRINTON,L.A., REEVES,W.C., BRENES,M.M., HERRERO, R., DE BRITTON,R.C., GAITAN,E., TENORIO,F., GARCIA,M. and RAWLS,W.E., Parity as a risk factor for cervical cancer. Amer. J. Epidemiol., (1989) (In press).
BRINTON, L.A., SCHAIRER, C., HAENSZEL, W., STOLLEY,P., LEHMAN, H.F., LEVINE,R. and SAVITZ,D.A., Cigarette smoking and invasive cervical cancer. J. Amer. med. Assoc., 255, 3265-3269 (19866). BROKER,T.R., Structure and gene expression of papillomaviruses. Obster. Gynecol. Clin. North Amer., 14, 329-348 (1987). DURST, M., GISSMANN, L. and IKENBERG,H., A papillomavirus DNA from a cervical carcinoma and its prevalence in cancer biopsies from different geographic areas. Proc. nut. Acad. Sci. (Wash.),80, 3812-3815 (1983). DURST,M., CROCE,C.M., GISSMANN, L., SCHWARZ, E. and HUEBNER, K . , Papillomavirus sequences integrate near cellular oncogenes in some cervical carcinomas. Proc. nut. Acad. Sci. (Wash.), 84, 1070-1074 (1987). FEINBERG, A.P. and VOGELSTEIN, B . , A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity. Anal. Biochem., 123, 6-13 (1983). FUCHS,P.G., GIRARDI,F. and PFISTER,H., Human papillomavirus DNA in normal, metaplastic, preneoplastic and neoplastic epithelia of the cervix uteri. Znt. J. Cancer, 41, 41-45 (1988). GLOSS,B., BERNARD, H.U., SEEDORF,K. and KLOCK,G., The upstream regulatory region of the human-papilloma-virus- 16 contains an E2protein-independent enhancer which is specific for cervical carcinoma cells and regulated by glucocorticoid hormones. EMBO J., 6, 3735-3743 (1987). M., Detection and HALLAM, N., GIBSON,P., GREEN,J. and CHARNOCK, typing of human papillomavirus infection of the uterine cervix by dot-blot hybridisation: comparison of scrapes and biopsies. J. med. Virol., 27, 317-321 (1989). MATLASHEWSKI, G., SCHNEIDER, J., BANKS,L., JONES,N., MURRAY, A. and CRAWFORD,L., Human-papillomavirus-type- 16 DNA cooperates with activated ras in transforming primary cells. EMBO J., 6, 1741-1746 ( 1987).
864
HILDESHEIM ET AL.
MEANWELL, C.A., Cox, M.F., BLACKLEDGE, G. and MAITLAND, N.J., HPV-16 DNA in normal and malignant cervical epithelium: implications for the aetiology and behaviour of cervical neoplasia. Lancer, I, 703-707 ( 1987). MITRANI-ROSENBAUM, S., TSVIELI,R. and TUR-KASPA, R. Estrogen but not glucocorticoids stimulate HPV-16 transcription in SIHA cells, Seventh International Papillomavirus Workshop, p. 82, Nice, France (1988). PATER,M.M. and PATER,A,, Human-papillomavirus-types-16 and -18 sequences in carcinoma cell lines of the cervix. Virology, 145, 313-318 (1985). PATER,M.M., HUGHES,G.A., HYSLOP,D.E., NAKSHATRI, H. and PATER,A., Glucocorticoid-dependent oncogenic transformation by type 16 but not type 11 human papilloma virus DNA. Nature fL0nd.L 335, 832835 (1988). ~K. and H ~ ~~P.M., ne ~~ human~~ pHELPS,w.c., yEE,c.L., M papillomavirus-type-E7 gene encodes transactivation and transformation functions similar to those of adenovirus E1A. Cell, 53, 539-547 (1988). S . S . , REEVES,W.C., SISSON,G.R., BRENES,M., GODOY,J., PRAKASH, BACCHETTI, S., DE BRITTON,R.C. and RAWLS,W E . , Herpes simplex virus type 2 and human papillomavirus type 16 in cervicitis, dysplasia and invasive cervical carcinoma. Inr. J. Cancer, 35, 51-57 (1985). REEVES,W.C., BRENES,M.M., DE BRITTON,R.C., VALDES,P.F. and
JOPLIN,C.F.B., Cervical cancer in the Republic of Panama. Amer. J. Epidemiol., 119, 714-724 (1984). REEVES,w , c , , BRINTON,L.A., GARCIA,M,, BRENES,M,M,, HERRERO, R., GAITAN,E., TENORIO,F., DE BRITTON,R.C. and RAWLS,W.E., Human papillomavirus infection and cervical cancer in Latin America. New Engl. J. Med., 320, 1437-1441 (1989~). REEVES,W.C., RAWLS,W.E. and BRINTON,L.A., Epidemiology of genital papillomaviruses and cervical cancer. Rev. infect. Dis., 11, 426-439 (1989b). SPENCE,R.P., MURRY,A., BANKS,L., KELLAND, L.R. and CRAWFORD, L.,,AnalYSiS of human papillomavirus sequences in cell lines recently derived from cervical cancers. cheer Res., 48, 324-328 (1988). WAHL,G.M., STERN,M. and STARK,G.R., Efficient transfer of large DNA fragments from agarose gels to diazobenzyloxymethyl-paper and rapid hybridization ~ , using dextran sulfate. Proc. nut. Acad. Sci. (Wash.), ' ,6~ 3683-3687 WHO, Invasive cervical cancer and combined oral contraceptives. Brit. med. J . s 2909 961-965 (1985). WINKELSTEIN, W., Cigarette smoking and cancer of the uterine cervix. Banbury Rep., 23, 329-341 (1986). ZAR,J.H., Biostaristical analysis. Prentice-Hall, Englewood Cliffs, New Jersey (1974).
