For the past two decades, more than 75% of U.S. women have used oral contraceptives (OCs) at some time during their life.1 Breast cancer is the leading cause of cancer in women, and among women aged 15 to 54 years, it is the leading cause of death from cancer.2 The high prevalence of OC use and the high incidence of breast cancer mandate continued analysis of possible relationships.
Two large population-based case-control studies in the United States (the Cancer and Steroid Hormone [CASH] study in 1986 and the Women's Contraceptive and Reproductive Experiences Study in 2002) provided strong evidence that pill use did not increase the risk of breast cancer.3,4 In contrast, a collaborative analysis of data from 54 epidemiologic studies worldwide in 1996 showed that women who were currently using OCs and women who had used them during the 10 years before diagnosis had a slightly increased risk of having breast cancer diagnosed compared with women who had never used OCs.5 Women who stopped using OCs more than 10 years before diagnosis had the same risk of developing breast cancer as women who had never used OCs. However, the collaborative analysis was published 10 years ago and did not include recent studies,4,6 the largest of which did not find an effect of recent use.4
Amid the controversy about the relationship between OCs and breast cancer, a key question remains. Once a woman has a new diagnosis of breast cancer, does prior pill use, particularly recent use, influence long-term survival? If using OCs results in earlier detection of existing tumors, then recent users would be expected to have improved survival. Alternatively, if using OCs promotes the growth of existing tumors, then mortality could be either increased or decreased, depending on how the pill alters tumor biology. Among pill users, better survival might indicate the development of less aggressive tumors, or shorter survival might indicate the development of more aggressive tumors.7
We examined the 15-year survival among women diagnosed with breast cancer during the CASH study. We investigated the effects of OC use on death from breast cancer by linking in-person interview data to follow-up data routinely collected by the Surveillance, Epidemiology, and End Results Program.
MATERIALS AND METHODS
The CASH study was a large, population-based, case-control study designed to assess the relationship between OCs and the risk of breast, ovarian, and endometrial cancer.3 The methods have been described previously.8 Briefly, cases were women aged 20 to 54 years who resided in Atlanta, Georgia; Detroit, Michigan; San Francisco, California; Seattle, Washington; Connecticut; Iowa; New Mexico; or four urban counties of Utah and who were diagnosed with histologically confirmed, primary breast cancer between December 1, 1980, and December 31, 1982. The study did not include women older than 54 years during 1980 through 1982 because they were unlikely to have used OCs. The study received institutional review board approval through the Centers for Disease Control and Prevention.
Of the 5,884 women with breast cancer who met the inclusion criteria for enrollment in the study, 4,730 (80.4%) agreed to a personal interview. Reasons for nonparticipation included inability to locate the woman or failure to conduct the interview within 6 months of diagnosis (8.1%), refusal (4.1%), illness (3.6%), physician refusal (2.9%), and death (0.9%). Each respondent was interviewed in person according to a pretested standardized questionnaire which focused on reproductive, contraceptive, and disease histories, family histories of cancer, use of medical services, and personal characteristics and behaviors. To enhance the women's recall and collect detailed information about the dates and formulations of OCs taken, interviewers used two memory aids: 1) a calendar containing graduations, marriages, pregnancies, and other life events that might be associated with contraceptive use and 2) a color photograph book of all OC pills marketed to date in the United States. The mean time from diagnosis to interview was about 12 weeks with a range of less than 1 week to 31 weeks.
We linked the CASH interview data to Surveillance, Epidemiology, and End Results Program cancer registry data to study tumor characteristics, first course of treatment, length of survival, and vital status.9 More than 95% of the completed interviews were successfully matched to Surveillance, Epidemiology, and End Results Program data (4,536 women) by gender, cancer site, geographic location, Surveillance, Epidemiology, and End Results Program identification number, year of birth, and date of diagnosis (within 2 months). Follow-up was available through December 31, 1997, and length of survival was defined as time in months from date of diagnosis to either date of death or date last known to be alive. Cases that did not match to Surveillance, Epidemiology, and End Results Program data were excluded from the analysis.
For the analysis, we excluded 17 women who did not know if they had ever used OCs. Because this was a study of breast cancer mortality, we also excluded 227 women diagnosed with in situ disease. We examined the relationship between survival and various measures of OC use, including ever use, duration, time since first use (latency) and last use (recency), and age at first use among 4,292 women diagnosed with breast cancer.
A total of 1,845 women died: 1,473 (79.8%) from breast cancer, 103 (5.6%) from other cancers, and 269 (14.6%) from other causes. For the survival analyses, events were defined as deaths due to breast cancer, coded as International Classification of Diseases, Ninth Revision 174.0–174.9.10 Follow-up for women with deaths due to other causes and for women whose death certificates could not be located was censored at date of death.
We used Cox proportional hazards models to estimate the relative rate of dying from breast cancer and 95% confidence limits (CLs).11 Confounding was assessed univariately for 22 variables for each measure of OC exposure, and adjustment for these variables did not appreciably alter the risk estimates. A priori models included age, race, education, income, time since last birth,12 menopausal status, body mass index,13–14 time since last use of hormone therapy, and radiation therapy. The univariate and a priori models yielded similar findings, and the a priori models are presented in the Results. Tests for trends in the measures of exposure to OCs were assessed among users, and the measure was included in the model as an ordinal variable; the trend was considered statistically significant if P<.05 for the ordinal variable. To determine whether recency of OC use influenced the risk of death from breast cancer in different subgroups of women, we used likelihood ratio tests; two-way interaction terms between the recency of OC use and the risk factor of interest were included in the models; none was statistically significant.
Of the 4,292 women diagnosed with breast cancer during 1980 through 1982, 1,473 died from their disease during the follow-up period. About 80% of the women survived 5 years, 70% survived 10 years, and 64%, 15 years. Demographic, reproductive, behavioral, treatment, and tumor characteristics varied significantly by time since last use of OCs (Table 1).
Overall, use of OCs did not have an adverse effect on breast cancer mortality (Table 2). Neither duration of use nor age at first use was related to the risk of death from breast cancer. Compared with women who had never used OCs, the risk of death from breast cancer was less than 1.0 for all duration and age at first use categories, and no dose-response relationships were apparent.
We also examined the relationships between breast cancer mortality and time since first and last use of OCs (Table 2). For time since first use, there were no trends of increasing or decreasing risk, and no risk estimate was elevated. For time since last use of OCs, the risk of death from breast cancer decreased significantly, but a consistent gradient effect was not observed (Table 2). Adjusted hazard ratios ranged from 0.86 to 1.41 and were 1.00 or less for all recency categories except 13 to 24 months before diagnosis (1.41 [95% CL 0.91, 2.17]); none was statistically significant. Women who were currently using OCs had an adjusted hazard ratio of 0.90 (95% CL 0.68, 1.19).
Because stage of disease at diagnosis is a strong prognostic factor for survival and potentially can act as a confounder, effect modifier, or intervening variable, the effects of stage on breast cancer mortality and time since last pill use were examined further. When stage was treated as a potential confounder by adding it to the other adjustment factors in the model, the hazard ratios (current use 1.00, 13–24 months 1.30, 25–60 months 1.03, 61–120 months 1.03, 121–180 months 0.91, 181 months or more 0.93) were similar to those in the recency model without stage, and the trend was no longer statistically significant (P=.12). When stage was treated as an effect modifier, the interaction between stage and recency was not statistically significant (P=.26).
Finally, we studied ever use and use within 25 months of diagnosis for 11 OC formulations, low- and high-dose estrogen content, and specific progestins (Table 3). None of these pill potency measures was related to the risk of death from breast cancer.
Overall, this study did not find evidence of either a beneficial or harmful effect of prior OC use on long-term survival after breast cancer diagnosis. Duration and latency of pill use, age at first use, and specific formulations were not related to long-term prognosis. Adjusted hazard ratios for time since last use were 1.0 or less for all recency categories except for women who last used OCs 13 to 24 months before diagnosis, and no estimate was statistically significant. Women who were currently using OCs had an adjusted hazard ratio of 0.90 (0.68, 1.19). The differences between pill users and nonusers were slight, and the hazard ratios were usually reduced with CLs that nearly always included 1.0.
Several investigators have examined the relationship between breast cancer survival and prior use of OCs.15–26 Most studies were not population-based15–20,22,24,25 and analyzed all cause mortality, or the cause of death as an endpoint was not specified.15–19,21–23,25,26 Others were limited by small numbers of patients who used OCs,15,16 and still others could not assess multiple measures of pill use15–17,19,21,25,26—usually reporting survival according to ever-never use. Like our study, most studies did not provide evidence of a relationship between contraceptive hormones and mortality.15–19,21–26 The World Health Organization Collaborative Study of Neoplasia and Steroid Contraceptives showed a protective effect of OC use; 79.5% of users survived 5 years compared with 70.3% of never users.22 Finally, although Lees and colleagues,21 whose study had the longest follow-up period (10 years), found that women who had ever used OCs had better survival than women who had never used them; adjustment by stage reversed the relationship.
The recency findings from the present analysis were similar to those from the combined case-control studies conducted in Sweden and Norway23 and in England.26 The latter study found a statistically significant trend of decreasing risk of death with increasing time since last use, the only measure of OC use presented. However, like our study, no estimates of risk were statistically significant, and there was not a consistent gradient effect for the trend. Other Swedish data exhibited adverse effects on survival among women who first used OCs at young ages.20
Several investigators have examined the influence of hormonal contraceptives on cell proliferation in the normal human breast and in breast cancers with varying results.27–33 Some studies identified different effects of specific progestins on cell proliferation,27,28,32,33 and others explored the relationship with hormone receptor status,34 HER-2/neu overexpression,35 and histology.36 A limitation of the current study is that it did not include information about estrogen and progesterone receptor status, BRCA1, BRCA2, HER-2/neu, measures of cell proliferation, and other biologic factors. In the 1980s, the tumor data from the Surveillance, Epidemiology, and End Results Program that were matched to the CASH study data did not include these factors because they were less well understood or not yet discovered. At that time, hormone receptor status was just beginning to be used to establish the course of treatment for breast cancer, and HER-2/neu was not used at all. If new treatment protocols were related to OC use, then the inability to adjust for such factors could bias the relationship between OC use and breast cancer mortality.
Another potential limitation of our study is the lack of data about risk factors and exposures, particularly OC use, which occurred after breast cancer diagnosis. The study findings are based on risk factors reported during interviews conducted on average 12 weeks after diagnosis. Women would likely have been advised to discontinue or not to start OC use after their breast cancer diagnosis. The World Health Organization medical eligibility criteria for using OCs state that women who currently have breast cancer should not use OCs, and that pill use is not usually recommended even for women with no evidence of disease for 5 years.37 We did not have information to adjust for confounding by factors that may have changed after diagnosis such as subsequent use of hormone therapy, changes in body mass index, and physical activity, or the effects of radiation and chemotherapy on menopausal status. Information bias in the determination of exposure to OCs was probably reduced by the use of the photobook and life calendar during the interview. Moreover, a previous study using these memory aids showed that the contraceptive histories collected were in good agreement with physician records.38
The strengths of the present study include the use of death from breast cancer as an endpoint (analyses of these data using all cause mortality as the endpoint yielded findings similar to the breast cancer mortality findings), long follow-up (at least 15 years), low loss to follow-up (the Surveillance, Epidemiology, and End Results Program estimates loss to follow-up at less than 3%39 after use of vital records, voter registration lists, driver license registration files, telephone directories, Social Security files, etc.), the population-based design, detailed measures of exposure to OCs, and large sample size. Based on the 1,457 women in Table 2 who died from breast cancer, we estimated post hoc that this study had 80% power to detect a 17% increased risk for breast cancer mortality associated with ever use of OCs against a two-tailed test with alpha=.05.40 Thus, the study had adequate power to detect rather small relative effects of OC use on breast cancer mortality. Finally, an extensive set of risk factors were available to assess confounding.
Subsequent investigations should take into account the biology of the breast and the effects of OCs on cell proliferation in the normal breast and in breast cancers. Additional information about the effects of OCs on prognostic biomarkers, such as hormone receptor status, genetic factors, and measures of tumor aggression, is also needed. In the meantime, the findings of this study, combined with the existing body of evidence about the relationship between OCs and breast cancer incidence and mortality, are reassuring—OC use did not affect the risk of dying from breast cancer.
1. Mosher WD, Martinez GM, Chandra A, Abma JC, Willson SJ. Use of contraception and use of family planning services in the United States: 1982–2002. Adv Data Vital Health Stat 2004;350:1–36.
2. United States Department of Health and Human Services, National Program of Cancer Registries. Cancer Registry Public Information Data: 1999–2002, WONDER On-line Database. Atlanta (GA): Centers for Disease Control and Prevention; November 2005. Available at http://wonder.cdc.gov/cancer.html
. Retrieved April 10, 2006.
3. Sattin RW, Rubin GL, Wingo PA, Webster LA, Ory HW. Oral contraceptive use and the risk of breast cancer. N Engl J Med 1986;315:405–11.
4. Marchbanks PA, McDonald JA, Wilson HG, Folger SG, Mandel MG, Daling JR, et al. Oral contraceptive use and the risk of breast cancer. N Engl J Med 2002;346:2025–32.
5. Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and hormonal contraceptives: collaborative reanalysis of individual data on 53,297 women with breast cancer and 100,239 women without breast cancer from 54 epidemiological studies. Lancet 1996;347:1713–27.
6. Bernstein L. The risk of breast, endometrial and ovarian cancer in users of hormonal preparations. Basic Clin Pharmacol Toxicol 2006;98:288–96.
7. Thomas DB. Oral contraceptives and breast cancer: review of the epidemiologic literature. In: Institute of Medicine, Division of Health Promotion and Disease Prevention, Committee on the Relationship Between Oral Contraceptives and Breast Cancer. Washington, DC: National Academy Press; 1991. p 102–35.
8. Wingo PA, Ory HW, Layde PM, Lee NC. The evaluation of the data collection process for a multicenter, population based, case-control design. Am J Epidemiol 1988;128:206–17.
9. National Cancer Institute, DCCPS, Surveillance Research Program. Surveillance, Epidemiology, and End Results (SEER) Program Public Use Data (1973–1997), Cancer Statistics Branch, released April 2000, based on the August 1999 submission. Available at www.seer.concer.gov
. Retrieved July 30, 2007.
10. World Health Organization. Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death, Based on the Recommendations of the Ninth Revision, Conference, 1975. Geneva, Switzerland: World Health Organization; 1977.
11. Cox DR. Regression models and life tables. J R Stat Soc Ser B 1972;34:187–220.
12. Whiteman MK, Hillis SD, Curtis KM, McDonald JA, Wingo PA, Marchbanks PA. Reproductive history and mortality after breast cancer diagnosis. Obstet Gynecol 2004;104:146–54.
13. Whiteman MK, Hillis SD, Curtis KM, McDonald JA, Wingo PA, Marchbanks PA. Body mass and mortality after breast cancer diagnosis. Cancer Epidemiol Biomarkers Prev 2005;14:2009–14.
14. World Health Organization. Obesity: preventing and managing the global epidemic. Report of a WHO consultation of obesity. Geneva, Switzerland: World Health Organization; 1997.
15. Spencer JD, Millis RR, Hayward JL. Contraceptive steroids and breast cancer. Br Med J 1978;1:1024–6.
16. Matthews PN, Millis RR, Hayward JL. Breast cancer in women who have taken contraceptive steroids. Br Med J (Clin Res Ed) 1981;282:774–6.
17. Greenberg ER, Vessey MP, McPherson K, Doll R, Yeates D. Body size and survival in premenopausal breast cancer. Br J Cancer 1985;51:691–7.
18. Rosner D, Lane WW. Oral contraceptive use has no adverse effect on the prognosis of breast cancer. Cancer 1986;57:591–6.
19. Millard FC, Bliss JM, Chilvers CED, Gazet JC. Oral contraceptives and survival in breast cancer. Br J Cancer 1987;56:377–8.
20. Olsson H, Moller TR, Ranstam J, Borg A, Ferno M. Early oral contraceptive use as a prognostic factor in breast cancer. Anticancer Res 1988;8:29–32.
21. Lees AW, Jenkins HJ, May CL, Cherian G, Lam EWH, Hanson J. Risk factors and 10-year breast cancer survival in northern Alberta. Breast Cancer Res Treat 1989;13:143–51.
22. Schönborn I, Nischan P, Ebeling K. Oral contraceptive use and the prognosis of breast cancer. Breast Cancer Res Treat 1994;30:283–92.
23. Holmberg L, Lund E, Bergström R, Adami H-O, Meirik O. Oral contraceptives and prognosis in breast cancer: effects of duration, latency, recency, age at first use, and relation to parity and body mass index in young women with breast cancer. Eur J Cancer 1994;30A:351–4.
24. Schouten LJ, Hupperets PS, Jager JJ, Volovics L, Wils JA, Verbeek AL, et al. Prognostic significance of etiological risk factors in early breast cancer. Breast Cancer Res Treat 1997;43:217–23.
25. Sauerbrei W, Blettner M, Schmoor C, Bojar H, Schumacher M. The effect of oral contraceptive use on the prognosis of node positive breast cancer patients. German Breast Cancer Group. Eur J Cancer 1998;34:1348–51.
26. Reeves GK, Patterson J, Vessey MP, Yeates D, Jones L. Hormonal and other factors in relation to survival among breast cancer patients. Int J Cancer 2000;89:293–9.
27. Seeger H, Rakov V, Mueck AO. Dose-dependent changes of the ratio of apoptosis to proliferation by norethisterone and medroxyprogesterone acetate in human breast epithelial cells. Horm Metab Res 2005;37:468–73.
28. Isaksson E, von Schoultz E, Odlind V, Soderqvist G, Csemiczky G, Carlstrom K, et al. Effects of oral contraceptives on breast epithelial proliferation. Breast Cancer Res Treat 2001;65:163–9.
29. Laidlaw IJ, Clarke RB, Howell A, Owen AW, Potten CS, Anderson E. The proliferation of normal human breast tissue implanted into athymic nude mice is stimulated by estrogen but not progesterone. Endocrinology 1995;136:164–71.
30. Williams G, Anderson E, Howell A, Watson R, Coyne J, Roberts SA, et al. Oral contraceptive (OCP) use increases proliferation and decreases oestrogen receptor content of epithelial cells in the normal human breast. Int J Cancer 1991;48:206–10.
31. Anderson TJ, Battersby S, King RJB, McPherson K, Going JJ. Oral contraceptive use influences resting breast proliferation. Hum Pathol 1989;20:1139–44.
32. Menendez JA, Oza BP, Colomer R, Lupu R. The estrogenic activity of synthetic progestins used in oral contraceptives enhances fatty acid synthase-dependent breast cancer cell proliferation and survival. Int J Oncol 2005;26:1507–15.
33. Jeng MH, Parker CJ, Jordan VC. Estrogenic potential of progestins in oral contraceptives to stimulate human breast cancer cell proliferation. Cancer Res 1992;52:6539–46.
34. Althius MD, Fergenbaum JH, Garcia-Closas M, Brinton LA, Madigan MP, Sherman ME. Etiology of hormone receptor-defined breast cancer: a systematic review of the literature. Cancer Epidemiol Biomarkers Prev 2004;13:1558–68.
35. Gammon MD, Hibshoosh H, Terry MB, Bose S, Schoenberg JB, Brinton LA, et al. Oral contraceptive use and other risk factors in relation to HER-2/neu
overexpression in breast cancer among young women. Cancer Epidemiol Biomarkers Prev 1999;8:413–9.
36. Newcomer LM, Newcomb PA, Trentham-Dietz A, Longnecker MP, Greenberg ER. Oral contraceptive use and risk of breast cancer by histologic type. Int J Cancer 2003;106:961–4.
38. Coulter A, Vessey M, McPherson K, Crossley B. The ability of women to recall their oral contraceptive histories. Contraception 1986;33:127–37.
39. Ries LAG, Harkins D, Krapcho M, editors. Contents of the SEER Cancer Statistics Review, 1975–2003. Bethesda (MD): National Cancer Institute; 2006. Available at http://seer.cancer.gov/csr/1975_2003/
. Retrieved July 30, 2007.
40. Rosner B. Fundamentals of biostatistics. 6th ed. Belmont (CA): Duxbury, 2006.