Invasive cervical cancer and depot-medroxyprogesterone acetate

Bulletin of the WorldHealth Organization, 63 (3): 505-5t 1 (1985) © World Health Organization 1985 Invasive cervical cancer and depot- medroxyprogesterone acetate WHO COLLABORATIVE STUDY OF NEOPLASIA AND STEROID CONTRACEPTIVES' Preliminary results of a study of the possible relationship of depot-medroxy- progesterone acetate (DMPA) to invasive cervical cancer are presented. The findings are based on datafrom three participating centres in Thailand and one in Mexico. A relative riskfor cervical cancer of 1.2 was observed in women who had ever used DMPA; this was not statistically significant. No consistent increase in risk with duration of use was observed, although a relative risk of2 wasfound in women who had usedDMPA for more than 5 years. This observed increase in risk was confined to women who were aged under 46 years or who had first been exposed to DMPA before 30 years of age. These findings are based on small numbers of subjects, and may not represent a causal relationship. Depot-medroxyprogesterone acetate (DMPA) is a long-acting steroid preparation that has been approved for use as a contraceptive in over 80 countries (1). It is highly effective and acceptable to women in many different cultures. Approximately 1.5 million wonmen were using DMPA in 1980 (2). However, in part because of concern over possible adverse health effects, DMPA has not yet been approved for use in the United States of America. This concern stems largely from the results of animal toxicology studies, which have raised the possibility that DMPA may increase the risks of neoplasms of the breast, endometrium, and cervix. Evidence that DMPA increases the risk of cervical cancer in women is meagre (2). Two studies conducted in the USA have reported the prevalence of squamous dysplasia and carcinoma in situ to be This report was prepared by David B. Thomas, Liza Noonan, Anne Whitehead, and Diane Roseman, on behalf of the collaborating investigators listed in Annex 1. All correspondence concerning this report, including requests for reprints, should be addressed to either: Dr David B. Thomas, Fred Hutchinson Cancer Research Center, 1124 Columbia Street, Seattle, WA 98104, USA; or Dr Susan Holck, Special Programme of Research, Development, and Research Training in Human Reproduction, World Health Organization, 1121 Geneva 27, Switzerland. elevated in users of DMPA (3, 4). In one of these studies, the expected prevalence was based on that in a general hospital population, and in the other study, on national statistics. In neither case were the known risk factors for cervical cancer taken into account. The possibility that women at high risk of cervical neoplasia may have more frequently chosen to use DMPA than other women was not ruled out; this possibility was actually investigated in a study in Chile (5). It was found that the incidence of cervical neoplasia after normal Pap smears was no different in users of progestational agents (DMPA or chlormadi- none acetate) than in women with an intrauterine device (IUD). A small study in Belgium showed a lower rate of abnormal cervical cytology in women who used progestational contraceptives than in women who used non-hormonal methods (6). However, all these studies were small, none has assessed the risk of cervical neoplasia in long-term users of DMPA, or risk after a long latency period, and none has included invasive cervical cancer cases. In view of the limitations of these studies, plus recent findings suggesting that combined oral contra- ceptives may increase the risk of cervical cancer (7, 8), 4547 505- WHO COLLABORATIVE S[UDY additional information on the possible relationship of DMPA to cancer of the uterine cervix is needed. To assess the influence of DMPA and other steroid contraceptives on risks of gynaecological, mammary, and hepatic malignancies, a collaborative, multi- national, hospital-based, case-control study is currently being conducted under the auspices of the World Health Organization (8). This report presents preliminary findings of the study on the relationship of DMPA to invasive cervical cancer. It is based on analyses of data from three participating centres in Thailand and one in Mexico. At the other partici- pating centres, insufficient numbers of women had used DMPA to warrant their inclusion in these analyses. METHODS The background of this study, and the methods utilized, have been described in a previous publication (8). In Thailand, the study is conducted at Siriraj and Chulalongkorn Hospitals in Bangkok, and at the Maharaj Nakorn Chiang Mai Hospital in Chiang Mai. In Mexico it is conducted at Mexico General Hospital in Mexico City. In each hospital, cases are detected by monitoring all new admissions to wards where cervical cancer is treated. These activities are supplemented by checking outpatient gynaecological or tumour clinics and pathology reports. Cases are restricted to women born after 1930 in Bangkok and Mexico, and after 1925 in Chiang Mai because DMPA became available earlier there. In all areas, cases must have been resident for at least one year in a defined geographic area served by the hospital. In addition, women whose neoplasm was initially detected at a family planning clinic, at any time other than the first visit to the clinic, are ineligible for study, in order to avoid the potential bias that women attending family planning clinics (and thus using contraception) would be more likely to be diagnosed as having cancer. Controls are selected from among hospitalized women admitted for conditions other than obstetric or gynaecological, and thought not to be associated with use of steroid contraceptives. Like the cases, the controls must be residents of a defined area and born after 1925 or 1930, according to the study area. Controls are not matched to individual cases, but are selected from specific hospital wards listed in a specified order. At the beginning of each week, all women who were admitted within the previous 48 hours to a particular ward and who meet the eligibility criteria are asked to participate. The same procedure is repeated on succeeding wards until two controls per case have been selected for the week. This procedure is continued on the following week beginning with the ward listed after the last one visited. Because this procedure initially resulted in the cases tending to be older than the controls, during the second and subsequent years of the study, on alternate weeks only women over 40 years of age were selected. Since this study considers other neoplasms in addition to those of the cervix, there are more than two controls per case of cervical cancer. A standard questionnaire is used to obtain information on the known and suspected risk factors for the neoplasms under study, as well as a complete obstetric and contraceptive history. The question- naire was initially written in English. Spanish and Thai translations are used during interviews, and the results are subsequently transcribed on to the English version. A calendar is incorporated into the question- naire to aid in the establishment of dates of use of steroid contraceptives. Attempts are made to validate selected items on the questionnaire, including use of DMPA, by review of hospital and clinic records. In Mexico, where injectable contraceptives other than DMPA are also used by appreciable numbers of women, users of an unknown type of injectable contraceptive were assumed to have used DMPA if injections had been received every three months. All cases are provisionally diagnosed by a local pathologist. To ensure that all eligible women with invasive disease are included, those diagnosed as having carcinoma in situ or severe dysplasia are tenta- tively included. Slides from all participating centres are sent to a single reference pathologist in Geneva for review and classification according to the Inter- national Histological Classification of Tumours (9). All data are coded and sent to a coordinating centre in Seattle, WA, USA, where they are monitored for quality, processed, and analysed. For this report, the unconditional logistic regres- sion model for large strata (10) was used to calculate estimates of relative risks adjusted for various potentially confounding variables. The General Linear Interactive Modelling (GLIM) system (11) was used for this purpose. All potentially confounding variables were entered into the regression models as stratified variables. RESULTS This report is based on data on controls and those cases considered by the reference pathologist to have invasive cervical cancer. Table 1 shows the cases and controls who were excluded from analysis because (a) they had not been interviewed, (b) they had had a hysterectomy, or (c) they had used an injectable 506 INVASIVE CERVICAI. CANCER AND D)Ml3A Table 1. Number of invasive cervical cancer cases and controls excluded from analyses Cases Controls No. % No. % Total accrued 512 100.0 2947 100.0 No. excluded from analysis Not interviewed 32 6.2 96 3.3 Prior hysterectomy 1 0.2 89 3.0 Used other injectable contraceptive 10 2.0 58 2.0 No. included in analyses 469 91.6 2704 91.7 contraceptive other than DMPA. All but one of the cases and two of the controls that were excluded because they had used other injectable contraceptives were from Mexico. The proportions not included for other reasons were similar in all four centres. Of the 469 cases included in the analysis, 400 (85.37o) were classified by the reference pathologist as squamous cell carcinoma, 41 (8.7%) as adeno- carcinoma, 15 (3.2/o) as adenosquamous carcinoma, and 13 (2.87o) as other histological types. There were too few types other than squamous cell to analyse separately, and all analyses are therefore based on all histological types combined. In all four participating centres, the cases tended to be older than the controls, and all estimates of relative risk are therefore adjusted for age. The number of cases and controls from each centre and the numbers with a history of DMPA use are shown in Table 2. The percentage of controls who had used DMPA varied from 3.7% in Mexico to 22.27o in Chiang Mai and, since the numbers of controls per case also varied among the centres (from 4.8 in Table 2. Numbers of cases and controls in each centre and numbers who had ever used DMPA Cases Controls DMPA users DMPA users Centre Total No. % Total No. % Mexico 57 5 8.8 493 18 3.7 Siriraj 142 1 5 10.6 760 40 5.3 Chulalongkorn 119 7 5.9 729 51 7.0 Chiang Mai 151 40 26.5 722 160 22.2 Total 469 67 14.3 2704 269 9.9 Chiang Mai to 8.6 in Mexico), all estimates of relative risk based on data from more than one centre are also adjusted for centre. Twenty-three potentially confounding variables were considered. These were: numbers of preg- nancies, stillbirths, miscarriages, induced abortions, total abortions, sexual relationships, residences, chest X-rays, and prior Pap smears; history of dilatation and curettage, vaginal discharge, sterilization, and tubal ligation; ages at menarche, first pregnancy, and first sexual relationship; marital status; place of residence; number of years at current residence; and use of oral contraceptives, a condom, withdrawal, and IUD for contraception. Relative risks of invasive cervical cancer in relation to each of these variables were estimated separately from the data from each centre. Also, the relative risks in women who had ever used DMPA were estimated separately from the data from each centre, controlling for age alone and age plus each of these 23 variables. Use of oral contraceptives, number of Pap smears, and number of pregnancies were the variables most strongly and consistently related to cervical cancer, or with the greatest influence on the estimates of relative risks for users of DMPA. These variables were therefore selected for inclusion in more detailed analyses. In addition, age at first sexual relationship, number of sexual partners, and a history of vaginal discharge were considered further because they had previously been shown to influence estimates of relative risks of cervical cancer in relation to oral contraceptive use. For reasons of efficiency, the four cases and 482 controls aged under 26 years were excluded from further analyses. In addition, because only 14 cases had no known pregnancies, these cases, and the 336 controls with such a history, were also eliminated from further consideration. A further 4 (0.9%) of the remaining 451 cases, and 15 (0.8%) of the remaining 1886 controls were eliminated, because of missing information on one or more of the above variables of interest, leaving 447 cases and 1871 controls to be included in the further analyses. Table 3 shows estimates of relative risk controlled for various combinations of variables that were added stepwise into logistic regression models. When only age and centre were controlled for, the relative risk was estimated to be 1.31, which was not significantly greater than 1.0. The estimate was reduced to 1.28 when prior Pap smears were controlled for, to 1.19 when controlled also for prior use of oral contra- ceptives, and to 1.13 when the number of pregnancies was also taken into account. When history of vaginal discharge, age at first sexual relationship, and number of sexual partners were also controlled fur, the estimate was increased slightly, but not to a statistically significant level. 507 508 WHO COLLABORATIVE SIUDY Table 3. Relative risks of invasive cervical cancer in users of DMPA adjusted for various potentially confounding variables Relative 95% confidence Variables controlled for risk interval A. Age; centre 1.31 0.95-1.80 B. Age; centre; no. of Pap smears 1.28 0.92-1.80 C. As B + oral contraceptives 1.19 0.86-1.65 D.AsC + no. of pregnancies 1.13 0.82-1.58 E. As D + vaginal discharge, age at first sexual relationship, and no. of partners 1.24 0.88-1.75 Among the variables considered in Table 3, signifi- cant interactions were observed between age and number of Pap smears, and between age and total number of pregnancies. However, including terms for these interactions in the logistic models did not appreciably alter the estimates of the relative risk (data not shown). Unlike the case with oral contraceptive use (8), there was no clear trend of increasing risk of cervical cancer with duration of use of DMPA (Table 4). However, the highest relative risk (2.20) was seen in the longest-term users. There was a significant interaction between duration of DMPA use and age, and Table 5 shows the increased risk in long-term users to be confined to women under 46 years of age. Broader and fewer categories of months of use of DMPA are considered here, because of the small number of cases that were long-term users. The apparent reduced risk in older women is based on a small number of cases and is not statistically signifi- cant. Table 4. Relative risks of invasive cervical cancer in relation to duration of use of DMPA Months of use No. of No. of Relative 95% confidence of DMPA cases' controls" risk' interval None 381 1642 1.00 1-12 29 87 1;30 0.81-2.10 13-24 8 40 0.87 0.39-1.96 25-60 11 55 0.71 0.36-1.42 61 16 33 2.20 1.15-4.21 Excluding 2 with unknown duration of use of DMPA. hExcluding 14 with unknown duration of use of DMPA. ' Estimate controlled for age, centre, number of Pap smears, use of oral contraceptives, number of pregnancies, history of vaginal discharge, age at first sexual relationship, and number of partners. Since young women tended to have begun using DMPA at an earlier age than older women (data not shown), the relationship of cervical cancer risk to duration of DMPA use by age at first use was con- sidered. As shown in Table 6, the increased risk in long-term users appears to be confined to women who began using DMPA before 30 years of age, although the interaction between duration of DMPA use and age at first use was not statistically significant. Estimates of the relative risks in women who had ever used DMPA did not vary significantly with the passage of time since first or most recent exposure. The increased risk in long-term users was evident in both recent users and in women who had last used DMPA in the more distant past (data not shown). Table 5. Relative risks" of invasive cervical cancer in relation to duration of use of DMPA and age Months Age (years) of use of DMPA 26-35 36-45 46-60 None 1.00 1.00 1.00 1-12 1.47 1.74 0.45 13-48 0.78 0.93 0.39 >48 2.19 2.64 0.26 Controlled for variables as listed in footnote c to Table 4. DISCUSSION The observed relative risk of about 1.2 in women who had ever taken DMPA could be due to chance or to incomplete control for the confounding effect of sexual variables, on which somewhat inaccurate information might have been obtained. No infor- mation on smoking was collected, and our inability to control for this potentially confounding variable could also explain the small observed increase in risk; however, this is unlikely because the prevalence of smoking is low among women in the areas where this study was conducted. The absence of a trend of increasing risk with duration of use, or with time since initial or most recent use does not support a causal relationship between DMPA and cervical cancer. However, the doubling of risk observed in women who had used DMPA for five or more years is of potential concern. This increase persisted after the estimate was controlled for a variety of potentially confounding variables. In addition, the possibility that women who had received DMPA were more INVASIVE CERVICAL CANCER AND DMPA Table 6. Relative risksa of invasive cervical cancer in relation to duration of use and age at first use of DMPA Age at first use of DMPA Months (years) of use of DMPA < 30 > 30 None 1.00 1.00 1-12 1.62 1.13 13-48 0.74 0.77 > 48 2.94 1.05 a Controlled for variables as listed in footnote c to Table 4. likely to have their disease diagnosed, and hence be included in the study, is not a likely explanation for the finding for two reasons: the analyses were restricted to cases with invasive disease; and the findings persisted after controlling for differences between cases and controls with respect to their prior history of Pap smears. A number of studies have shown early age at first sexual intercourse to be a risk factor for cervical cancer (12, 13). While some reports have disputed this (14), if true it suggests that the young cervix is more susceptible than the older cervix to the carcinogenic effect of whatever sexually transmitted agent causes cervical cancer. The observed increase in risk among long-term users of DMPA was confined to women who were under 46 years of age or who began using DMPA before 30 years of age. Although these observations could be due to chance, they are also consistent with the idea that the young cervix is more vulnerable to potentially carcinogenic substances. The small number of subjects precluded more detailed analyses of the data on which this interim report is based, and the results should therefore be considered preliminary. ;UME CANCER INVASIF DU COL ET ACETATE DE MEDROXYPROGESTERONE-RETARD Ce rapport presente les resultats preliminaires d'une etude cas-temoins collective multinationale effectuee en milieu hospitalier sur les relations entre l'acetate de medroxy- progesterone-retard (DMPA) et le cancer du col. 11 est base sur l'analyse de donnees provenant de quatre centres participants, trois en Thailande et un au Mexique. Dans chaque h6pital participant, on depiste les cas en surveillant les nouvelles admissions dans les salles reservees au traitement du cancer du col uterin. On complete ces activites par des enquetes dans les dispensaires de soins gynecologiques ou cancerologiques ambulatoires et par les rapports d'anatomo-pathologie. Sont uniquement prises en consideration les femmes nees apres 1930 a Bangkok et a Mexico et apres 1925 a Chiang Mai, car le DMPA y a e disponible plus t6t. Dans toutes ces regions, les interessees doivent avoir reside au moins un an dans une zone geographique precise desservie par l'hopital. En outre, sont exclues les femmes dont la tumeur avait et decouverte a l'origine dans un dispensaire de planification familiale (sauf au cours de la premiere visite) de faion a eviter un biais eventuel, car les femmes qui frequentent ces dispensaires (et qui par consequent utilisent des contraceptifs) ont plus de chances de presenter un cancer. On choisit les temoins parmi les femmes hospitalisees pour des raisons autres qu'obstetriques ou gynecologiques et normalement sans rapport avec l'utilisation de contra- ceptifs steroidiens. Un questionnaire type permet de recueillir des renseigne- ments sur les facteurs de risque, connus et presumes, associes aux neoplasmes etudies, ainsi que sur tous les antecedents en matiere d'obstetrique et dc contraception. Un diagnostic provisoire est systematiquement pose par un anatomo-pathologiste local. Pour etre certain que toutes les femmes atteintes d'une maladie invasive et repondant aux criteres de l'etude y soient bien admises, on retient provisoirement toutes celles pour lesquelles on a diagnosti- que un epithelioma in situ ou une dysplasie grave. Les frottis de tous les centres participants sont ensuite envoyes a Geneve ou un seul et meme anatomo-pathologiste en effectue l'examen et la classification selon la Classification histologique internationale des tumeurs. Toutes ces donnees sont codees puis envoyees a un centre coordinateur a Seattle oui, apres un controle de qualite, elles sont traitees et analysees. On a utilise dans ce rapport le modee de regression logistique inconditionnelle applicable aux strates d'effectif eleve pour calculer des risques relatifs estimatifs (corriges pour differentes variables parasites eventuelles). Toutes ces variables ont et incorporees au modele de regression o0 elles servent de base a la stratification. Le present rapport repose sur l'interrogatoire de 2704 temoins ayant un uterus non lese et de 469 femmes dont l'anatomo-pathologiste centralisateur a considere qu'elles presentaient un cancer invasif du col. Soixante-sept cas (14,3%) et 269 temoins (9,907o) n'avaient jamais utilise de DMPA. On a estime que le risque relatif pour les femmes ayant pris du DMPA a une periode quelconque de leur vie etait de 1,2 (intervalle de confiance a 95%/o: 0,88-1,75) une fois eliminee l'influence globale de divers parametres: I'age, le centre, le nombre de tests de Papanicolaou pratiques, l'utilisation de contraceptifs oraux, le nombre de grosses- ses, l'existence anterieure de pertes vaginales, I'age lors des 509 510 WHO COLLABORATIVE STFUDY premiers rapports sexuels et le nombre de partenaires sexuels. Une correction effectuee pour 17 autres variables n'a pas modifie cette estimation de facon appreciable. Le risque ne semble pas augmenter systematiquement avec la duree d'utilisation du DMPA, mais sa valeur la plus elevee (2,20, intervalle de confiance a 9507/o: 1,15-4,21) a et trouvee chez les femmes qui en avaient pris le plus longtemps (plus de 5 ans). On n'a observe cette augmentation apparente du risque que chez les femmes de moins de 46 ans ou chez celles qui avaient pris du DMPA pour la premiere fois avant 30 ans. Aucun indice d'une evolution du risque avec le temps n'a pu etre mis en evidence, que l'on considere la premiere utilisation du DMPA ou la derniere en date. Ces resultats sont bases sur un petit nombre d'utilisatrices au long cours, et sont susceptibles de ne pas representer une relation de cause a effet. La collecte des donnees se poursuit dans tous les centres dont on a parle dans ce rapport, de facon a rassembler un nombre suffisant de cas et de temoins pour permettre des analyses plus fines, de puissance statis- tique convenable. REFERENCES 1. Facts about injectable contraceptives: Memorandum from a WHO meeting. Bulletin of the World Health Organization, 60: 199-210 (1982). 2. FRASER, 1. S. & WEISBERG, E. A comprehensive review of injectable contraception with special emphasis on depot medroxyprogesterone acetate. Medicaljournal of Australia, 1 (Suppl.): 1-19 (1981). 3. POWELL, L. C. & SEYMOUR, R. J. Effects of depot medroxyprogesterone acetate as a contraceptive agent. American journal of obstetrics and gynecology, 110: 36-41 (1971). 4. LITT, B. P. Statistical review of carcinoma in situ reported among contraceptive users of Depoprovera. Memorandum of 17 June 1974 to United States Food and Drug Administration hearing on Depoprovera. 7 May, 1975. 5. DABANCENS, A. ET AL. Intraepithelial cervical neoplasia in women using intrauterine devices and long-acting injectable progestogens as contraceptives. American journal of obstetrics and gynecology, 119: 1052-1056 (1974). 6. IDE, P. ET AL. Observations cytologiques de frottis cervicovaginaux sous contraception hormonale. Revue de cytologie clinique, 5: 105-112 (1972). 7. VESSEY, M. P. ET AL. Neoplasia of the cervix uteri and contraception: a possible adverse effect of the pill. Lancet, 2: 930-934 (1983). 8. WHO Collaborative Study of Neoplasia and Steroid Contraceptives. Invasive cervical cancer and combined oral contraceptives. British medical journal, 290: 961- 965 (1985). 9. SOBIN, L. H. ET AL., ed. A coded compendium of the International Histological Classification of Tumours. Geneva, World Health Organization, 1978. 10. BRESLOW, N. E. & DAY, N. E. Statistical methods in cancer research. Vol. 1. The analysis of case-control studies. Lyon, International Agency for Research on Cancer, 1980 (IARC Scientific Publications, No. 32). 11. BAKER, R. J. & NELDER, J. A. The GLIM system, Release 3, Oxford, Numerical Algorithms Group, 1978. 12. CRAMER, D. W. Uterine cervix. In: Schottenfeld, D. & Fraumeni, J. F., ed., Cancer epidemiology and prevention, Philadelphia, W. B. Saunders Co., 1982. 13. AURELIAN, L. ET AL. Viruses and gynecologic cancers: herpesvirus protein (ICP 10/AG-4), a cervical tumor antigen that fulfills the criteria for a marker of carcinogenicity. Cancer, 48: 455-471 (1981). 14. HARRIS, R. W. C. ET AL. Characteristics of women with dysplasia or carcinoma in situ of the cervix uteri. British journal of cancer, 42: 359-369 (1980). INVASIVE CERVICAL CANCER AND DMPA 511 Annex I PARTICIPATING CENTRES AND INVESTIGATORS The data collection centres and investigators were as follows: Mexico General Hospital, Mexico City, Mexico Hector Rodriguez Cuevas, a Socorro Benavidei Salazar,b Jorges Albores Saavedra," Patricia Ontiveros. ' Chiang Mai University, Faculty ofMedicine, Chiang Mai, Thailand Suporn Silpisornkosol,a Tieng Pardthaisong, b Virote Sahapong, b Viruch Charoenium, b Choti Theetramont.' Chulalongkorn University, Faculty of Medicine, Department ofObstetrics and Gynaecology, WHO Collaborating Centre for Research in Human Reproduction, Bangkok, Thailand Banpot Boonsiri,a Pramuan Virutamasen, b Chan- suda Wongsrichanalai, b Sermsri Sindhvananda. c a Principal investigator. bCo-investigator. ' Pathologist. Mahidol University, Faculty of Medicine, Siriraj Hospital, Department of Obstetrics and Gynaecology, Family Planning Research Unit, Bangkok, Thailand Suporn Koetsawang,a Duangdao Rachawat,h Amorn Koetsawang.' Reference Pathologist Gustave Riotton, Centre de Cytologie et de Depistage du Cancer, Geneva, Switzerland. Coordinating Centre Fred Hutchinson Cancer Research Center, Division of Public Health Sciences, Seattle, WA, USA David B. Thomas (Study Coordinator), Diane Roseman (Epidemiologist), Anne Whitehead (Statistician), Liza Noonan (Statistician). World Health Organization Susan Holck, Special Programme of Research, Development and Research Training in Human Reproduction, World Health Organization, Geneva, Switzerland.