Depot-medroxyprogesterone acetate (DMPA) and cancer: Memorandum from a WHO Meeting*

Bulletin of the World Health Organization, 64 (3): 375-382 (1986) © World Health Organization 1986 Depot-medroxyprogesterone acetate (DMPA) and cancer: Memorandum from a WHO Meeting* A review of all available data including thosefrom the WHO Collaborative Study of Neoplasia and Steroid Contraceptives has shown no increased risk ofcancers ofthe breast, endometrium, ovary or liver in women using depot-medroxyprogesterone acetate (DMPA). The issue of a causal association between DMPA use and cervical cancer is as yet unresolved and will require the accumulation of additional data. To date, in the WHO study only a small number of women have usedDMPA forprolonged periods or have had a long interval since first use. Information on cancer risk in these women can only be gained by continuing the present study or by initiating additional studiesfocused on these specific topics. INTRODUCTION For more than a decade WHO's Special Pro- gramme of Research, Development and Research Training in Human Reproduction has been regularly assessing the safety and side-effects of methods of fertility regulation used throughout the world. Specific issues of safety raised by Member govern- ments or by the scientific community are investigated through research activities undertaken by the Pro- gramme and also examined at meetings of experts convened to review the available data. One subject continuously under review by the Programme is whether the use of steroid contra- ceptives alters the risk of neoplasia. In 1977, the Pro- gramme convened a Scientific Group which con- cluded that "there are no adequate data from studies in women to assess whether progestogens used as contraceptives in the form of progestogen-only pills or as injections have any effect on the risk of neoplasia" (1). This conclusion and the finding of tumors in animal toxicology studies of progestogen- only contraceptives (2) emphasized the need for research in this area. As a result, WHO's Special Programme embarked upon the planning of a multi- national collaborative case-control study to examine the relationship between steroid contraceptives and the risk of selected neoplasms. In 1981, the Programme reviewed all data available on injectable contraceptives. At this meeting it was * This Memorandum was drafted by the signatories listed on page 381 on the occasion of aWHO meeting in Geneva in September 1985. Requests for reprints should be sent to Special Programme of Research, Development and Research Training in Human Reproduction, World Health Organization, 1211 Geneva 27, Switzerland. A French translation of this Memorandum will appear in a later issue of the Bulletin. concluded from both animal and human data on depot-medroxyprogesterone acetate (DMPA) and cancer that, "although epidemiological studies in women receiving DMPA have thus far demonstrated no increase in the risk of developing any type of cancer, ... because of the lack of well controlled trials and the long latency period of some cancers, it is important to continue to monitor the possible development of neoplasms among women who have used DMPA" (2); it was noted that WHO had recently initiated a collaborative study on this issue. Preliminary results from this study have been published concerning breast and cervical cancers (3, 4). Other epidemiological studies on DMPA and cancers of the breast and cervix have been initiated in New Zealand, Costa Rica and Jamaica; results of these studies are not yet available. Meanwhile, drug regulatory agencies in a number of countries, including Canada, Federal Republic of Germany, Sweden, the United Kingdom and USA, have reviewed DMPA with a view to its approval as a contraceptive. In these reviews neoplasia has con- sistently been raised as an issue of concern; the Pro- gramme has also received numerous inquiries on the subject from Member governments in both developed and developing countries. It was therefore considered opportune to convene a meeting in 1985 to review both published and unpublished epidemiological data from human studies only, since results from animal toxicology studies had been thoroughly reviewed pre- viously (2). Data available since the 1981 meeting were thus examined in depth, including the pre- liminary results from the WHO collaborative study. This Memorandum summarizes the data reviewed and the conclusions and presents recommendations for future research. 4675 -375 MEMORANDUM THE WHO COLLABORATIVE STUDY OF NEOPLASIA AND STEROID CONTRACEPTIVES Methods The WHO Collaborative Study of Neoplasia and Steroid Contraceptives was carried out in 14 collab- orating centres in 11 countries. The risk of cancer in DMPA users was examined using data collected in three centres in Thailand, one centre in Kenya and one centre in Mexico, which were the centres where DMPA use was appreciable. Recruitment of subjects began in October 1979 in the centres in Thailand and Mexico, and in June 1981 in Kenya. The most recent analysis of data on DMPA and cancer available for this review was based on subjects whose complete data were available at the coordinating centre at the Fred Hutchinson Cancer Research Center in Seattle, Washington, USA, on 14 June 1985. The study was restricted to women born after 1930 in all the centres except Chiang Mai, Thailand. In Chiang Mai, the subjects were required to have been born after 1925, which is due to the earlier availability of DMPA in this area of Thailand. The study was further restricted to women who had been resident for at least one year in a defined geographical area served by the participating hospitals. Cases of breast, cervical, endometrial, ovarian and liver cancers among women in the eligible birth years and residence areas were identified by monitoring all admissions to hospital wards where these cancers were treated, and by checking the records of out- patient clinics and pathology departments in each centre. Approximately two controls per case were selected from among women in the eligible birth years and residence areas who had been admitted to other than obstetric and gynaecological wards in the same hospitals for conditions that would not alter contra- ceptive practice-i.e., excluding circulatory or car- diovascular disease, diabetes, chronic renal disease, benign breast disease, previously diagnosed cancer, chronic liver disease, and obstetric or gynaecological disease. Neither cases nor controls were interviewed if they had been referred from a family planning or fertility clinic unless the visit that prompted the referral was the woman's first visit to that clinic. A standard questionnaire presented to cases and controls by a trained, female interviewer, was used to elicit information on known and suspected risk factors for the various cancers being studied, and on obstetric history and previous use of contraceptives. Recall of whether steroid contraceptives had been used was facilitated by showing samples of prepar- ations available in each country. Data collection forms underwent preliminary coding and editing at each study site. Further editing, including range and consistency checks performed by computer, was conducted at the coordinating centre in Seattle. Potential errors, not correctable at the coordinating centre, were referred to the centres for clarification. Provisional diagnosis of cancer was made by one pathologist in each participating centre. Stained and unstained histological slides for each case were reviewed again by a single reference pathologist for each cancer and classified using the WHO histo- logical classification of tumours. Analyses were based only on cases where the diagnosis of malig- nancy was confirmed by the reference pathologist, except for liver cancer, where clinical as well as patho- logical diagnoses were accepted. Analyses of breast and cervical cancers included all the control subjects from all four centres and used the unconditional logistic regression model to estimate risks, controlling for potential confounders by enter- ing them into the model as stratified variables. For the other three cancers, cases were matched with up to eight controls on the year of birth, centre, and year of entry to the study, and the conditional logistic regression model was used to estimate the relative risks. For all analyses, 95% confidence intervals for relative risk estimates were calculated using the normal approximation to the exact confidence limits (S). Bias and confounding There is at least one potentially important uncontrolled confounder in the WHO study and one potential source of bias. With regard to confounding, information on cigarette smoking was not collected in the early part of the study, but this has been initiated recently. Several studies have shown cigarette smoking to be associated with an increase in the risk of cervical neoplasia (6-8) and a decrease in the risk of endometrial cancer (9). The risk of breast cancer appears not to be related to cigarette smoking (9), while few data are available on the possible association of ovarian or liver cancer with cigarette smoking. If cigarette smoking is more prevalent among users of DMPA, as it is among users of other steroid contraceptives in some developed countries, then failure to collect information on smoking and to control for possible differences in it between users and non-users of DMPA could overestimate the risk of cervical cancer and exaggerate the apparent pro- tective effect of DMPA on endometrial cancer. The small amount of tobacco consumed by women in these countries suggests that confounding from this source will not affect relative risk estimates 376 DEPOT-MEDROXYPROGESTERONE ACETATE AND CANCER 377 appreciably. With regard to bias, cases and controls referred from fertility or family planning clinics were excluded unless the visit leading to hospital referral was the woman's first visit to the fertility or family planning clinic. This decision was made to prevent over- representation in the study of cases that had used steroid contraceptives. However, it has been pointed out that these exclusions could have the potential to create bias both ifwomen attending fertility or family planning clinics are more likely to be using DMPA and if the cancers of interest are more likely to be diagnosed in these clinics than other conditions lead- ing to hospitalization. It seems obvious that women attending family planning clinics are more likely to be using DMPA. Yet, it cannot be assumed that the cancers studied in the WHO study are more likely than other serious diseases to be diagnosed in family planning and fertility clinics in these countries, since cervical cytology screening and breast examinations are not necessarily done in such clinics. There are no data to rule out this possibility, however. If the cancers of interest are more likely than other hospital- ized conditions to be diagnosed in a family planning or fertility clinic, the exclusion of subjects referred from these clinics would lead to an underestimate of the risk for each cancer. If this difference were large, it would constitute a potentially important source of bias in the WHO study. Although data are incomplete, a number of facts mitigate against this possible bias as an explanation for the largely negative results of the WHO study to date. First, in centres where data are available on the number of cases of cancer excluded because of referral from family planning or fertility clinics, the number of cases excluded for this reason is small. At Chulalongkorn University in Bangkok, no subjects were excluded for this reason. In Chiang Mai, 16 subjects with cervical cancer, 3 with breast cancer, and zero subjects with the other three cancers were excluded on this basis. In Kenya, two subjects with cervical cancer were excluded for this reason. Data on exclusions in the other Thai centre and the centre in Mexico are pending. The small number of subjects excluded in the centres for which data are available makes it unlikely that the magnitude of relative risk estimates is seriously affected by these exclusions. Further reassurance comes from examination of preliminary data showing that relative risk estimates for cervical cancer and breast cancer are unchanged when current users of any form of contraception, who seem most likely to be affected by this potential bias, are excluded from the analyses. A final con- clusion about the effect of this bias must, however, await examination of the number of excluded cases and controls from all five centres. ENDOMETRIAL CANCER Limited experimental data in rhesus monkeys have raised the possibility that DMPA may increase the risk of endometrial cancer. The relevance of these findings in humans is controversial. At least 24 biopsy studies in humans indicate that use of DMPA reversibly changes proliferative endo- metrium to secretory or suppressed!' Published data from epidemiological studies are limited. Three descriptive studies of women developing or dying from endometrial cancer were undertaken in areas of DMPA use!' Although these studies showed no evidence of an adverse effect of DMPA, they pro- vided little information on this issue because of a variety of methodological limitations. Two record- linkage epidemiological studies were carried out in Atlanta, USA, using data from the Grady Memorial Hospital family planning clinic records of approxi- mately 5000 women who had used DMPA. One study utilized hospital admission records to estimate the incidence of cancer, which was compared with the ex- pected cancer incidence (11). One limitation of the study was an estimated 45% underascertainment of cancer, although the authors attempted to correct for that in the analyses. In these analyses, the expected number of cases of uterine cancer was 0.83. One case (leiomyosarcoma) was found. The other study used death certificates to estimate cause-specific mortality rates by the type of contraceptive used (12). Potential problems with this study include selection bias (the health of women possibly being related to the choice of contraceptive method), misclassification of con- traceptive exposure status with change of methods during the period of observation, and differential in- complete ascertainment of cancer deaths according to contraceptive method. Although neither of the two studies suggested an adverse effect ofDMPA with re- spect to endometrial cancer, no meaningful con- clusion on this issue can be drawn from the studies, primarily because of the small numbers involved. However, several studies have suggested that the increased risk of endometrial cancer associated with estrogen therapy for perimenopausal symptoms can be reduced or eliminated by the addition of cyclical progestogen therapy (13-17). Furthermore, at least 7 studies have demonstrated a negative association between use of combined oral contraceptives and endometrial cancer (16). This protective effect is thought to be due to the progestogen component of oral contraceptives (17). The data from the endometrial biopsy studies, perimenopausal therapy, and studies of combined a FOOD AND DRUG ADMINISTRATION. Report ofthe Public Board of Inquiry on Depo-Provera. 1981 (unpublished). 378 MEMORANDUM oral contraceptives, while indirect, suggest the hypo- thesis that DMPA is negatively associated with endo- metrial cancer. Resultsfrom the WHO study Complete data on 52 cases of endometrial cancer and 6012 controls were available for these interim analyses. A total of 316 controls were matched to individual cases on exact year of birth, year of entry into the study, and centre. Only one of the cases, and 30 of the 316 controls had ever used DMPA, giving an estimated relative risk in women who had ever used DMPA of 0.3 (Table 1). This estimate was not appreciably altered after controlling individually for total number of live births, total number of pregnancies, history of infertility, and use of exogenous estrogen. Although this low relative risk suggests that DMPA may protect against endometrial cancer, the 95% confidence limits of the estimate are wide (0.04-2.4) and the observed low relative risk could be due to chance. There is no indication from these findings that DMPA, as it has been used in the past in the three countries from where the data for this report were obtained, increases the risk of endometrial cancer. However, this study has Table 1. Relative risks of five neoplasms in women who have ever used DMPA: results of the WHO Collaborative Study of Neoplasia and Steroid Contraceptives No. of subjects in the analysis Adjusted Cancer site Cases Controls relative riska Endometrium 57 316 0.3 (0.04-2.4)b Ovary 105 637 0.7 (0.3-1.7)c Liver 57 290 1.0 (0.4-2.8)d Breast 427 5951 1.0 (0.7-1.5)' Cervix 920 5833 1.2 (0.9-1.5)f a Figures in parentheses are 95% confidence intervals. b Controls matched with cases by age, centre, and year of entry into the study. c Adjusted for total number of live births, history of infertility, oral contraceptive use and IUD use; controls matched with cases by age, centre, and year of entry into the study. d Adjusted for oral contraceptive use and IUD use; controls matched with cases by centre, age, and year of entry into the study. ' Adjusted for age, centre, age at first live birth, total number of live births, oral contraceptive use and IUD use. f Adjusted for age, centre, total number of pregnancies, history of vaginal discharge, age at first sexual relationship, number of sexual partners, number of Pap smears, oral contra- ceptive use and IUD use. accumulated insufficient data to assess the risk of endometrial cancer in long-term users or the risk long after initial exposure. OVARIAN CANCER The two studies using Grady Memorial Hospital data, which attempted to examine the relationship between DMPA use and the development of ovarian cancer, did not indicate any adverse effect of DMPA in this regard (11, 12). However, both had methodo- logical limitations as mentioned above, and produced insufficient data to allow meaningful conclusions. Nine studies have demonstrated a negative association between oral contraceptive use and ovarian cancer (16). This effect is thought to be mediated through ovulation suppression (18). Since DMPA usually prevents ovulation, the negative association observed with oral contraceptives, if due to ovulation suppression, suggests that DMPA is negatively associated with ovarian cancer. Results from the WHO study A total of 105 cases of ovarian cancer, largely epithelial carcinomas, and 6206 controls were available for the present analyses. From among the controls available, 637 were matched to individual cases on exact year of birth, year of entry into the study, and centre. Seven of the 105 cases and 74 of the 637 controls had been exposed to DMPA. The relative risk in women who had ever used DMPA was estimated to be 0.7 (95% confidence interval, 0.3-1.7) after controlling for the potentially con- founding effects of parity, history of infertility, use of oral contraceptives, and use of an IUD (Table 1). When the analyses were confined to women of proven fertility (at least one live birth), the comparable relative risk was found to be 1.0 (95% confidence interval, 0.4-2.6). Insufficient data were available for more detailed analyses. These preliminary findings provide no indication that the risk of ovarian cancer is altered in women who have ever used DMPA, as this contraceptive has so far been used in the three countries from where the present data were collected. Data are, however, insufficient to assess the influence on risk of ovarian cancer among long-term users or risk after long-term exposure. LIVER CANCER Resultsfrom the WHO study Complete data on 57 cases of primary liver cancer were available for analysis. Two hundred and ninety DEPOT-MEDROXYPROGESTERONE ACETATE AND CANCER controls were matched to individual cases on exact year of birth, year of entry into the study, and centre. Seven of the 57 cases, and 34 of the 290 controls had ever used DMPA, giving a relative risk adjusted for age and centre of 1.0 (Table 1) in women who had ever used DMPA (95% confidence interval, 0.37-2.85). This estimate was not altered by con- trolling for a variety of potentially confounding vari- ables, including history of jaundice, use of alcohol, and other factors. These results provide no evidence that DMPA alters the risk of liver cancer, but the power of this study to detect small alterations in risk, risk in long-term users, and risk long after the initial exposure is low. Table 2. Relative risk of breast cancer in relation to duration of DMPA use: results of the WHO Collaborative Study of Neoplasia and Steroid Contraceptives No. of subjects Months of use Cases Controls Relative risk' None 385 5290 1.0 1-12 17 220 1.1 (0.7-1.9) 13-36 13 156 1.2 (0.7-2.2) 37 9 157 0.8 (0.4-1.7) a Adjusted for age, centre, year of entry, total number of live births, age at first live birth, oral contraceptive use and IUD use. Figures in parentheses are 95% confidence intervals. BREAST CANCER Limited experimental data in beagle bitches have raised the possibility that injectable progestogens, such as DMPA, may increase the risk of breast cancer. The relevance of these findings to women is controversial, but they underline the need to evaluate breast cancer risk in epidemiological studies. Until the preliminary results from the WHO study were published (3), there were no publications with adequate data to determine whether DMPA in- creases, decreases, or has no effect on the risk of breast cancer. In a comparison of 30 cases of breast cancer and 179 controls derived from the Grady Memorial Hospital in Atlanta (19), the frequencies of DMPA exposure were 5 (17%) and 32 (18%), respectively (relative risk estimate 1.0; no confidence interval given). For oral contraceptive use, the relative risk estimate was 0.3 (95% confidence interval, 0.1-0.7). This study had little statistical power and almost no information on long-duration exposure. In addition, a significant reduction in breast cancer for oral contraceptive users has not been found in any other study, raising the possibility of bias in the Grady Hospital data. In all other studies (11, 12, 20-22), the findings were difficult to interpret because of methodological problems. Among others, these included inadequate statistical power, short duration of exposure, absence of appropriate comparison groups, and sparse descriptive data. In one study, among 19 875 women who received DMPA, no breast abnormalities of any type were diagnosed in any of the women, suggesting obvious underascertainment of breast pathology (23). Resultsfrom the WHO study This report is based on interim analyses of data available for 427 cases and 5951 controls, ofwhom 39 cases and 557 controls had ever used DMPA. The relative risk in women who had ever used DMPA was estimated to be 1.0 (95% confidence interval, 0.7- 1.5), after controlling for the possible confounding effects of age, centre, age at first live birth, total number of live births, use of oral contraceptives, and use of an IUD (Table 1). Risk was not found to change appreciably with duration of use (Table 2) and was not altered either in women who were first ex- posed before age 30, or in women initially exposed at an older age. Too few women used DMPA before the birth of their first child to assess the influence of such use on risk. These findings suggest that DMPA, as it has been used to date in the three countries from where the present data were collected, has not altered the risk of breast cancer. However, the data currently available are insufficient to assess the influence of DMPA on the risk of breast cancer in long-term users, and the risk long after initial exposure. CERVICAL CANCER It is difficult to carry out satisfactory epidemio- logical studies of the relationship between steroidal contraceptives and cancer of the cervix. First, the information about potentially confounding sexual variables is likely to be inadequate, especially about age at first sexual intercourse and the number of sexual partners. Furthermore, the sexual histories of the male partners may also be important. Secondly, it is known that occlusive methods of contraception offer some protection against cervical neoplasia, so the use of these methods should be taken into account. Thirdly, almost all pre-invasive lesions (and some invasive ones) are detected by cervical cytologi- cal screening. Furthermore, treatment of pre-invasive lesions is believed to reduce the risk of invasive disease. Thus, any substantial difference in the 379 MEMORANDUM pattern of cytological screening between groups being compared could lead to incorrect conclusions. Finally, histopathologists vary greatly in their inter- pretation and classification of pre-invasive lesions of the cervix. This could lead to bias if any one patholo- gist reviews a disproportionate amount of material from women using a particular contraceptive method. Very few published studies have examined the relationship between use of DMPA and cervical cancer; only two, other than the WHO study (4), are mentioned here. Powell & Seymour followed up 1123 women in Texas who used DMPA for a total of 14 000 woman-months (24); only 51 of these women accumulated more than 4 years of use. Both the abnormal cytology rate and the rate of biopsy-proven carcinoma in-situ were higher in the DMPA users than was expected from past experience in the same centre, but no allowance was made for the effect of confounding factors such as age, social class, race, or sexual behaviour. A study conducted by Dabancens et al. (10) in 1974 in Santiago, Chile, included 2409 IUD acceptors and 2684 comparable women who accepted injectable contraceptives (2234 with DMPA, 445 with chlormadinone). Of the women using injectables, 331 accumulated more than 4 years of exposure. In total, 9 women in the injectable groups (1.26/1000 woman-years) and 6 in the IUD group (1.82/1000 woman-years) developed pre- invasive or (in 2 cases) invasive lesions of the cervix. Although reassuring, this study is handicapped by its small size. Resultsfrom the WHO study These interim analyses are based on 920 cases of invasive cervical cancer and 5833 controls with complete data at the coordinating centre as of 14 June 1985. One hundred and twenty six cases and 545 controls had ever used DMPA, giving an estimated relative risk of 1.4 after adjusting for only age and centre. However, after adjusting also for the potential confounding effects of total number of pregnancies, history of vaginal discharge, age at first sexual relationship, number of sexual partners, number ofPap smears, use of an IUD, and use of oral contraceptives, the relative risk of invasive cervical cancer in women who had ever used DMPA was re- duced to a non-significant 1.2 (95% confidence inter- val, 0.9-1.5) (Table 1). This small elevation in risk could be a result of residual confounding due to im- precise information on sexual behaviour of the study subject, or to confounding by other variables not considered, such as smoking or sexual practices of the subjects' husbands. As shown in Table 3, the risk did not increase with duration of use, which also suggests that non-causal factors may be responsible for the ob- Table 3. Relative risk of cervical cancer in relation to duration of use of DMPA: results of the WHO Collabor- ative Study of Neoplasia and Steroid Contraceptives No. of subjects Months of use Cases Controls Relative risk' None 782 5184 1.0 1-12 58 216 1.4 (1.0-2.0) 13-24 20 92 1.2 (0.7-2.0) 25-60 17 127 0.6 (0.4-1.1) 61 26 86 1.4 (0.9-2.2) a Adjusted for age, centre, year of entry, total number of pregnancies, vaginal discharge, age at first intercourse, number of sexual partners, number of Pap smears, oral contraceptive use and IUD use. Figures in parentheses are 95% confidence intervals. served small increase in relative risk. Although the risk of cervical cancer among long- term users (use for more than 4 years) was no greater than among short-term users (use for one year or less), the subgroup of long-term users was examined further because earlier, published analyses had indicated an increased risk among this subgroup (4). Among women who had used DMPA for more than 4 years, the relative risk of developing cervical cancer before the age of 36 was estimated as 2.3 (95% con- fidence limits, 1.5-3.6). Among women aged 36 to 45 and 46 to 60 years the risk was estimated as 1.7 (95% confidence limits, 1.4-2.1) and 0.4 (95% confidence limits, 0.2-0.8), respectively. A relative risk estimate of 2.4 (95% confidence limits, 1.9-3.0) was observed among women who had first used DMPA before the age of 30 and who had used it for more than 4 years, whereas the risk for women who had first used DMPA after the age of 30 and had used it for more than 4 years was 1.3 (95% confidence limits, 1.0-1.7). However, it should be noted that these are prelimi- nary data, the number of subjects in the subgroups is small, and the significant findings are a result of mul- tiple subgroup analyses. The magnitude of the risk elevations among young, long-term users of DMPA is similar to that reported for cervical cancer in young, long-term users of oral contraceptives. These findings might indicate an effect of the steroids on risk; an alternative explan- ation is incomplete adjustment for confounding factors. CONCLUSIONS Until recently, there were no adequate studies of the effects of DMPA on cancer incidence. The 380 DEPOT-MEDROXYPROGESTERONE ACETATE AND CANCER 381 published data concerning cancers of the endo- metrium, ovary, breast, and cervix did not demon- strate any increased risk associated with DMPA but, because of methodological limitations, these data provide only limited reassurance. The multinational case-control study being con- ducted by WHO provides the first reliable infor- mation on DMPA and neoplasia. The preliminary data concerning cancers of the endometrium, ovary, liver, and breast suggest that there is no risk in users of DMPA. The relative risk estimates for these sites are 0.3 for endometrial cancer, 0.7 for ovarian cancer, 1.0 for liver cancer, and 1.0 for breast cancer. There is still only limited information about long- term users of DMPA, but these interim results are reassuring. For cervical cancer, the adjusted relative risk in women who have ever used DMPA is 1.2 (95% confidence interval, 0.9-1.5). There is no consistent trend with duration of use of DMPA, although certain subgroups of women show an increased risk. An elevation of risk has also been observed in users of oral contraceptives in the WHO study. In both instances, it is possible that this is due to incomplete adjustment for sexual risk factors. To date, in theWHO study only a small number of women have used DMPA for prolonged periods or have had a long interval since first use. Information on cancer risk in these women can only be gained by continuing the present study or by initiating addi- tional studies focused on these specific topics. Since any effect of DMPA on cancer incidence might not appear until after a delay of many years, further studies will need to be carried out in the future. Supawat Chutivongse, Chulalongkorn Hospital Medical School, Bangkok, Thailand R. Gray, The Johns Hopkins University, Baltimore, MD, USA C. Hill, Institut Gustave Roussy, Villejuif, France B. Hulka, School of Public Health, Chapel Hill, NC, USA (Chairman) P. Kenya, Medical Research Centre, Nairobi, Kenya V. Odlind, University Hospital, Uppsala, Sweden Tieng Pardthaisong, Chiang Mai University, Chiang Mai, Thailand D. Petitti, University of California, San Francisco, CA, USA G. Rubin, Centers for Disease Control, Atlanta, GA, USA S. Shapiro, Boston University School of Medicine, Brookline, MA, USA Siporn Silpisornokosol, Chiang Mai University, Chiang Mai, Thailand D. Skegg, University of Otago, Dunedin, New Zealand Suwanee Srisupandit, Siriraj Hospital, Bangkok, Thailand B. Stadel, National Institute of Child Health and Human Development, Bethesda, MD, USA D. Thomas, The Fred Hutchinson Cancer Research Center, Seattle, WA, USA M. Vessey, University of Oxford, Oxford, England WHO Secretariat P. Corfman, Special Programme of Research, Development and Research Training in Human Reproduction, WHO, Geneva, Switzerland E. Diczfalusy, Special Programme of Research, Development and Research Training in Human Reproduction, WHO, Geneva, Switzerland (Con- sultant) P. Hall, Special Programme of Research, Develop- ment and Research Training in Human Repro- duction, WHO, Geneva, Switzerland S. Harlap, Special Programme of Research, Develop- ment and Research Training in Human Repro- duction, WHO, Geneva, Switzerland (Consul- tant) S. Holck, Special Programme of Research, Develop- ment and Research Training in Human Repro- duction, WHO, Geneva, Switzerland (Secretary) L. Mehra, Maternal and Child Health, WHO, Geneva, Switzerland M. Ten Ham, Pharmaceuticals, WHO, Geneva, Switzerland REFERENCES 1. WHO Technical Report Series, No. 619, 1978 (Steroid contraception and the risk of neoplasia: report of a WHO Scientific Group). 2. Facts about injectable contraceptives: Memorandum from a WHO Meeting. Bulletin of the World Health Organization, 60: 199-210 (1982). 382 MEMORANDUM 3. WHO COLLABORATIVE STUDY OF NEOPLASIA AND STEROID CONTRACEPTIVES. Breast cancer and depot- medroxyprogesterone acetate. Bulletin of the World Health Organization, 63: 513-519 (1985). 4. WHO COLLABORATIVE STUDY OF NEOPLASIA AND STEROID CONTRACEPTIVES. Invasive cervical cancer and depot-medroxyprogesterone acetate. Bulletin of the World Health Organization, 63: 505-511 (1985). 5. BRESLOW, N. E. & DAY, N. E. Statistical methods in cancer research. Volume L - The analysis of case-control studies. Lyon, International Agency for Research on Cancer, 1980 (Scientific Publications No. 32). 6. WINKELSTEIN, W. JR. Smoking and cancer of the uterine cervix hypothesis. American journal of epidemiology, 106: 257-259 (1977). 7. TRAVANTHAN, E. ET AL. Cigarette smoking and dysplasia and carcinoma in situ of the uterine cervix. Journal of the American Medical Association, 250: 499-502 (1983). 8. LYON, J. L. ET AL. Smoking and carcinoma in situ of the uterine cervix. American journal ofpublic health, 73: 558-562 (1983). 9. BARON, J. A. Smoking and estrogen-related disease. American journal of epidemiology, 119: 9-22 (1984). 10. 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). 11. LIANG, A. P. ET AL. Risk of breast, uterine corpus, and ovarian cancer in women receiving medroxy- progesterone injections. Journal of the American Medical Association. 249: 2909-2912 (1983). 12. ORY, H. W. ET AL. Mortality among young black women using contraceptives. Journal of the American Medical Association, 251: 1044-1048 (1984). 13. GAMBRELL, R. D. ET AL. Role of estrogens and pro- gesterone in the etiology and prevention of endometrial cancer: Review. American journal of obstetrics and gynecology. 146: 696-707 (1983). 14. NACHTIGALL, L. E. ET AL. Estrogen replacement therapy. II: A prospective study in the relationship to carcinoma and cardiovascular and metabolic problems. Obstetrics and gynecology, 54: 74-79 (1979). 15. HAMMOND, C. B. ET AL. Effects of long-term estrogen replacement therapy. II. Neoplasia. American journal of obstetrics and gynecology, 133: 537-547 (1979). 16 VESSEY, M. P. Exogenous hormones in the etiology of cancer in women. Journal of the Royal Society of Medicine, 77: 541-549 (1984). 17. CENTERS FOR DISEASE CONTROL CANCER AND STEROID HORMONE STUDY. Oral contraceptive use and the risk of endometrial cancer. Journal of the American Medical Association, 249: 1600-1604 (1983). 18. CENTERS FOR DISEASE CONTROL CANCER AND STEROID HORMONE STUDY. Oral contraceptive use and risk of ovarian cancer. Journal of the American Medical Association, 249: 1596-1599 (1983). 19. GREENSPAN, A. R. ET AL. The association of depo- medroxyprogesterone acetate and breast cancer. Contraception, 21: 563-568 (1980). 20. SCHWALLIE, P. C. Experience with depo-provera as an injectable contraceptive. Journal of reproductive medicine, 13: 113-117 (1974). 21. MCDANIEL, E. B. & PARDTHAISONG, T. Incidence of breast nodules in women receiving multiple doses of medroxyprogesterone acetate. Journal of biosocial science, 5: 83-88 (1973). 22. ZANARTU, J. ET AL. Mammary gland nodules in women under continuous exposure to progestogens. Contra- ception, 7: 203-210 (1973). 23. RALL, H. J. S. ET AL. Comparative contraceptive experience with three and six month medroxypro- gesterone acetate regimens. Journal of reproductive medicine, 18: 55-60 (1977). 24. POWELL, L. C. & SEYMOUR, R. G. Effects of depo- medroxyprogesterone acetate as a contraceptive agent. American journal of obstetrics and gynecology, 110: 36-41 (1971).