CancerCausesand Control,3, 433 - 439
Type of postmenopausal hormone use and risk of breast cancer: 12-year follow-up from the Nurses' Health Study Graham A. Colditz, Meir J. Stampfer, Walter C. WiUett, David J. Hunter, JoAnn E. Manson, Charles H. Hennekens, Bernard A. Rosner, and Frank E. Speizer (Received 9 April 1992;accepted in revisedform 16June 1992) We prospectively examined the use of hormone replacement therapy in relation to breast cancer incidence in a cohort of women 30 to 55 years of age in 1976. During 12 years of follow-up (480,665 person-years) among postmenopausal women, 1,050 incident cases of breast cancer were documented. Overall, past users of replacement estrogen were not at increased risk. After adjustment for established risk factors, type of menopause, age at menopause, and current age, the rate ratio (RR) was 0.91, 95 percent confidence interval ( C I ) = 0.78-1.07. The risk of breast cancer was elevated significantly among current users (RR -- 1.33, CI -- 1.12-1.57); after adjusting for age, we observed no evidence of increasing risk with increasing duration of use among current users (P trend = 0.41), or among past users (P trend = 0.46). Women currently using unopposed estrogen ( R R = 1.42, C I = 1.19-1.70), estrogen and progesterone (RR = 1.54, CI = 0.99-2.39), or progesterone alone ( R R = 2.52, C I = 0.66-9.63), were all at increased risk of breast cancer compared with never users. These data suggest that long-term past use of estrogen replacement therapy is not related to risk, that current estrogen use increases risk of breast cancer to a modest degree, and that the addition of progesterone does not remove the increased risk observed with current use of unopposed estrogen.
Key words: Breast cancer, cohort study, estrogens, progestins, Nurses' Health Study, USA.
Introduction
Methods
We have reported data previously on the relationship between use of postmenopausal hormones and risk of breast cancer in a prospective cohort study of 121,700 female registered nurses? We now update these data with two additional years of follow-up, adding 398 incident cases of breast cancer, and report the associations with different hormone types,
The Nurses' Health Study cohort was established in 1976, when 121,700 female registered nurses 30 to 55 years of age completed a mailed questionnaire that included items about known or suspected risk factors for cancer and cardiovascular diseases. Baseline information included details of breast cancer risk factors,2,3 postmenopausal use of hormones, and history of use of
The authors are with the Nurses'Health Study, Channing Laboratory, Department of Medicine, Brigham and Women's Hospital, Boston, MA; and Harvard Medical School, Boston, MA, USA. Address correspondenceto Dr Colditz, Channing Laboratory, 180 Longwood Ave., Boston, MA 02115-5899, USA. Supported by research grant CA40356 from the National Cancer Institute, NIH, Department of Health and Human Services. Dr Colditz is supported by an American Cancer Society Faculty Research Award FRA-398. © 1992 Rapid Communications of Oxford Ltd
Cancer Causes and Control. Vol 3. 1 9 9 2
433
G. A. Colditz et al oral contraceptive agents. Every two years, follow-up questionnaires have been mailed to the women to bring up to date the information on risk factors, including details of current hormone use and duration of use, and to ascertain whether major medical events have occurred. We used the response on each questionnaire to define current-use status for the subsequent twoyear interval. In biennial questionnaires sent in 1978 and subsequently, we asked each woman if she was currently using postmenopausal hormones and, if so, the type of hormones. The daily dose of conjugated estrogen (by far the most common form) was ascertained from 1980 on and was classified as 0.3 mg, 0.625 mg, 1.25 mg, and more than 1.25 mg.
In 1976, 23,965 women met the criteria for classification as postmenopausal; these women entered the follow-up from 1976 to 1978. The report of menopause was updated every two years, and the population was expanded to include women who became postmenopausal and had remained free from cancer. During the 12 years of follow-up from the return of the 1976 questionnaire to 1 June 1988, we accrued 480,665 person-years of follow-up among postmenopausal women; 1,050 incident cases of breast cancer were identified among these postmenopausal women. Follow-up of the cohort for the identification of nonfatal breast cancer, by questionnaire, telephone interview, and certified mail was 95 percent complete. For fatal breast cancer, follow-up was more than 98 percent complete?
Identification of breast cancer cases On each questionnaire, we inquired whether breast cancer had been diagnosed and, if so, the date of diagnosis. All women who reported breast cancer (or the next-of-kin for decedents) were asked for permission to review the relevant hospital records and confirm the self-reported diagnosis. Pathology reports were obtained for 93 percent of the cases, and information on histologic tumor type, tumor size, and node involvement was extracted by physicians blinded to data on hormone-supplement use reported by the participants. Although permission to review pathology reports and hospital records could not be obtained for seven percent of cases, we based our analysis on all incident breast cancers because the rate of accuracy of selfreporting was extremely high (99.6 percent) among those for whom records were obtained. We omitted the small number of carcinomas in situ from analysis.
Population for analysis The endpoint for the primary analysis was the occurrence of incident breast cancer. We therefore excluded from the analysis all women who reported breast or other cancer (excluding nonmelanoma skin cancer) on the 1976 questionnaire. This left a total cohort of 118,300 women eligible for follow-up. The present analysis was limited to postmenopausal women. We classified a woman as postmenopausal from the time she returned a questionnaire on which she reported natural menopause or hysterectomy with bilateral oophorectomy. Self-report of menopause is highly reproducible in this cohort and is valid with regard to details of the extent of ovarian surgery.4In addition, we classified women reporting hysterectomy without bilateral oophorectomy as postmenopausal at the age when natural menopause had occurred in 90 percent of the cohort (54 years for current cigarette smokers and 56 years for nonsmokers). 434
Cancer Causes and Control. Vol 3. 1992
Statistical analysis The primary analysis used incidence rates with personmonths of follow-up as the denominator. For each participant, person-months were allocated according to the 1976 exposure variables and were updated according to information on follow-up questionnaires. For women who developed breast cancer or died, personmonths were assigned according to the covariate status reported on the most recent completed questionnaire, but follow-up terminated with the diagnosis of breast cancer or death. Women who reported a diagnosis of cancer other than nonmelanoma skin cancer on any questionnaire were excluded from subsequent follow-up. If no questionnaire was returned for a follow-up cycle, the most recently recorded covariate data were used for the subsequent follow-up interval. If the previous status was 'current' postmenopausal hormone use, however, the updated status of use was classified as missing. For past use and never use, the status of use was classified according to the most recent questionnaire. We used the rate ratio (RR) as a measure of association, defined as the incidence of breast cancer among women who had used postmenopausat hormones divided by the corresponding rate among women who had never used these drugs. Additional stratified analyses were carried out to control for risk factors and to explore the possibility that the effect of current use of postmenopausal hormones might be modified by these factors. We used the Mantel test for linear trend to examine the relation between duration of current use of postmenopausal hormones and risk of breast cancer,6 and proportional-hazards models to adjust for multiple risk factors simultaneously.7 We also used proportional hazards models to analyze survival status among patients with breast cancer according to the history of use of postmenopausal hor-
Estrogens, progestins, and breast cancer Table 1. Age-specific incidence rates of breast cancer by never use, ever use, current use, and past use of replacement hormone therapy, 1976-88' Age
Never use Cases
Personyears
30-39 40-44 45-49 50-54 55-59 60-66
0 5 36 163 221 104
1,039 4,147 23,907 84,587 95,366 35,560
Total
529
244,606
Ever use Rate b Cases
-120 151 193 232 292
Current use
Past use
Personyears
Rate
RR '
Cases
Personyears
Rate
RR
Cases
Personyears
Rate
RR
3 11 40 114 195 139
4,516 11,766 26,948 60,303 78,005 37,804
66 93 148 189 250 368
-0.78 0.98 0.98 1.06 1.24
3 7 26 64 92 62
3,471 8,526 16,652 29,982 28,377 11,377
86 82 156 213 324 545
-0.68 1.03 1.10 1.40 1.87
0 4 14 50 97 77
1,045 3,240 10,295 30,110 48,867 26,370
-123 136 166 198 292
1.03 0.90 0.86 0.85 1.00
502
219,342
1.08
254
98,385
1.34
242
119,927
0.90
Missing data on h o r m o n e use for 19 cases during 16,717 person-years; and on status (current or past) for six cases. b Rate per 100,000 person-years. ° RR = risk ratio.
mones prior to diagnosis. Survival status for the 722 incident cases of breast cancer diagnosed between 1976 and June 1986 were ascertained as of 1 June 1990 for 702 women; the remaining 20 women were censored when they were last known to be alive between June 1988 and June 1990. Overall, these women were censored for a total of 22.8 years of potential follow-up.
Results In 1976, 33 percent of postmenopausal women were current users of postmenopausal hormones, 18 percent were past users, 3.6 percent had missing information on use of postmenopausal hormones, and the remainder had never used these agents. From 1976 to 1988, current use of postmenopausal hormones accounted for 21 percent of follow-up time; past use, 25 percent; missing information on hormone use, four percent; and never use, 50 percent. As previously reported, women who had never used postmenopausal hormones and current and past users had similar distributions for age at menarche and family history of breast cancer.' Current users were slightly more likely to have reported a history of benign breast disease, to be lean, to consume alcohol, to be nulliparous, or to have given birth only once or twice. We examined the relation between ever-use, current, and past use of postmenopausal hormones and risk of breast cancer within five-year age groups (Table 1). Ever-use was not associated with increased risk of breast cancer, age-adjusted RR = 1.08 (95 percent confidence interval [CI] = 0.96-1.22). Examining current and past use separately, we observed an increased risk of breast cancer among current, but not among past users. The increased risk associated with current use of estrogen replacement therapy (ERT) was most evident
among women 55 years of age or older. Overall, the age-adjusted RR of breast cancer among current users was 1.29 (CI-- 1.11-1.50). After adjusting for age in single years, age at menopause, and type of menopause, the RR increased slightly to 1.33 (CI = 1.12-1.57). Past use was not associated with increased risk of breast cancer. The RR, adjusted for age, was 0.90 (CI = 0.77-1.04). We examined the association between current use of estrogen replacement and risk of breast cancer separately among women reporting natural menopause, bilateral oophorectomy, and hysterectomy without removal of ovaries. After age-standardizing to the distribution of women with a natural menopause who never took replacement hormone therapy, the RR for current use among the women with natural menopause was 1.33 (103 cases, 33,899 person-years); among women with a bilateral oophorectomy, the RR was 1.35 (114 cases, 53,865 person-years); and among women with hysterectomy without oophorectomy, the RR was 1.59 (35 cases, 7,321 person-years). To determine whether the association between current use and risk of breast cancer varied according to other risk factors, we examined the age-adjusted RR within strata of parity, age at first birth, family history of breast cancer in mother or sister, personal history of benign breast disease, age at menarche, and body mass index. The RRs varied little within strata of these risk factors providing little evidence for effect modification (Table 2). The trend for increasing risk among current users with later ages at first birth, and the lack of association among women with a personal history of benign breast disease are noteworthy. After multivariate adjustment for all the established risk factors, age in single years, age at menopause in years, and type of menopause, the RR was essentially unchanged from that adjusted for age alone for current use (RR = 1.33, Cancer Causes and Control. Vol3. 1992
435
G. A. Colditz et al CI=1.12-1.57), and for past use (RR=0.92, CI = 0.79-1.09). Addition of alcohol to models for follow-up from 1980-88 did not materially alter the results. We next examined duration of use and risk of breast
cancer among current and former users (Tables 3 and 4). We observed no evidence of increasing risk with increasing duration of use among either current or past users. Among current users, the test for trend in risk with increasing duration of use was not significant (trend, P = 0.41). Among past users, women who had used replacement estrogens for 10 or more years in the past had an age-adjusted RR of 0.75 (CI -- 0.47-1.21) compared with women who had never used replacement hormones. Further adjustment for age at menopause and type of menopause did not materially alter these results. Among past users, the test for trend in risk with increasing duration of use was not significant, P = 0.46. In a previous study, Gambrell 8 suggested that the addition of progestin was associated with a reduced risk of breast cancer of 0.3 (CI = 0.1-0.8). Therefore, we examined this question using follow-up data from 1978 when type of hormone replacement was first asked. During this 10-year follow-up period, we observed 972 incident cases of breast cancer. Compared with never-users, women who were current users of conjugated estrogen were at increased risk of breast cancer, as observed in the overall cohort analysis (the age-adjusted RR = 1.42, CI = 1.19-1.70) (Table 5). Current use of estrogen and progestin was associated with an age-adjusted RR of 1.54 (0.99-2.39). Women who used estrogen and testosterone were also at increased risk ( R R = 2.45, CI = 0.95-6.35). We had insufficient data to provide informative risk estimates for other forms of replacement therapy. Adjustment for type of menopause in addition to age did not materially alter these RR estimates, nor did multivariate analysis control for established breast-cancer risk factors, although confidence intervals broadened. Prior use of oral contraceptives (OC) before ERT has been raised as a possible modifier of breast cancer risk associated with estrogen use. We specifically
Table 2. Age-adjusted rate ratio (RR) and 95% confidence
interval (CI) for current and past use of estrogen replacement therapy within strata of established risk factors for breast cancer, 1976-88 Current use
Past use
RR
CI
RR
CI
Family history Yes No
1.33 1.30
(0.8%1.98) (1.10-1.52)
0.77 0.92
(0.51-1.17) (0.78-1.08)
History of BBD" Yes No
0.97 1.42
(0.75-1.26) (1.18-1.71)
0.81 0.91
(0.62-1.05) (0.75-1.10)
Menarche (age) 13
1.37 1.36 0.96
(1.10-1.70) (1.04-1.77) (0.67-1.38)
0.85 0.96 0.90
(0.68-1.07) (0.73-1.26) (0.65-1.26)
Parity Nulliparous 1-2 3+
0.94 1.58 1.17
(0.55-1.60) (1.23 -2.03) (0.95-1.44)
1.21 0.86 0.86
(0.77-1.91) (0.66-1.14) (0.70-1.06)
Age at first birth 20-24 25-29 30-34 35-39 40+
1.23 1.32 1.43 2.08 2.79
(0.97-1.63) (1.03-1.71) (0.94-1.69) (1.11-3.29) (0.85-9.19)
0.91 0.87 0.78 1.0 1.47
(0.69-1.19) (0.67-1.13) (0.50-1.20) (0.51-1.97) (0.40-5.40)
Body mass index (kg/m 2)
Type of postmenopausal hormone use and risk of breast cancer: 12-year follow-up from the Nurses' Health Study Graham A. Colditz, Meir J. Stampfer, Walter C. WiUett, David J. Hunter, JoAnn E. Manson, Charles H. Hennekens, Bernard A. Rosner, and Frank E. Speizer (Received 9 April 1992;accepted in revisedform 16June 1992) We prospectively examined the use of hormone replacement therapy in relation to breast cancer incidence in a cohort of women 30 to 55 years of age in 1976. During 12 years of follow-up (480,665 person-years) among postmenopausal women, 1,050 incident cases of breast cancer were documented. Overall, past users of replacement estrogen were not at increased risk. After adjustment for established risk factors, type of menopause, age at menopause, and current age, the rate ratio (RR) was 0.91, 95 percent confidence interval ( C I ) = 0.78-1.07. The risk of breast cancer was elevated significantly among current users (RR -- 1.33, CI -- 1.12-1.57); after adjusting for age, we observed no evidence of increasing risk with increasing duration of use among current users (P trend = 0.41), or among past users (P trend = 0.46). Women currently using unopposed estrogen ( R R = 1.42, C I = 1.19-1.70), estrogen and progesterone (RR = 1.54, CI = 0.99-2.39), or progesterone alone ( R R = 2.52, C I = 0.66-9.63), were all at increased risk of breast cancer compared with never users. These data suggest that long-term past use of estrogen replacement therapy is not related to risk, that current estrogen use increases risk of breast cancer to a modest degree, and that the addition of progesterone does not remove the increased risk observed with current use of unopposed estrogen.
Key words: Breast cancer, cohort study, estrogens, progestins, Nurses' Health Study, USA.
Introduction
Methods
We have reported data previously on the relationship between use of postmenopausal hormones and risk of breast cancer in a prospective cohort study of 121,700 female registered nurses? We now update these data with two additional years of follow-up, adding 398 incident cases of breast cancer, and report the associations with different hormone types,
The Nurses' Health Study cohort was established in 1976, when 121,700 female registered nurses 30 to 55 years of age completed a mailed questionnaire that included items about known or suspected risk factors for cancer and cardiovascular diseases. Baseline information included details of breast cancer risk factors,2,3 postmenopausal use of hormones, and history of use of
The authors are with the Nurses'Health Study, Channing Laboratory, Department of Medicine, Brigham and Women's Hospital, Boston, MA; and Harvard Medical School, Boston, MA, USA. Address correspondenceto Dr Colditz, Channing Laboratory, 180 Longwood Ave., Boston, MA 02115-5899, USA. Supported by research grant CA40356 from the National Cancer Institute, NIH, Department of Health and Human Services. Dr Colditz is supported by an American Cancer Society Faculty Research Award FRA-398. © 1992 Rapid Communications of Oxford Ltd
Cancer Causes and Control. Vol 3. 1 9 9 2
433
G. A. Colditz et al oral contraceptive agents. Every two years, follow-up questionnaires have been mailed to the women to bring up to date the information on risk factors, including details of current hormone use and duration of use, and to ascertain whether major medical events have occurred. We used the response on each questionnaire to define current-use status for the subsequent twoyear interval. In biennial questionnaires sent in 1978 and subsequently, we asked each woman if she was currently using postmenopausal hormones and, if so, the type of hormones. The daily dose of conjugated estrogen (by far the most common form) was ascertained from 1980 on and was classified as 0.3 mg, 0.625 mg, 1.25 mg, and more than 1.25 mg.
In 1976, 23,965 women met the criteria for classification as postmenopausal; these women entered the follow-up from 1976 to 1978. The report of menopause was updated every two years, and the population was expanded to include women who became postmenopausal and had remained free from cancer. During the 12 years of follow-up from the return of the 1976 questionnaire to 1 June 1988, we accrued 480,665 person-years of follow-up among postmenopausal women; 1,050 incident cases of breast cancer were identified among these postmenopausal women. Follow-up of the cohort for the identification of nonfatal breast cancer, by questionnaire, telephone interview, and certified mail was 95 percent complete. For fatal breast cancer, follow-up was more than 98 percent complete?
Identification of breast cancer cases On each questionnaire, we inquired whether breast cancer had been diagnosed and, if so, the date of diagnosis. All women who reported breast cancer (or the next-of-kin for decedents) were asked for permission to review the relevant hospital records and confirm the self-reported diagnosis. Pathology reports were obtained for 93 percent of the cases, and information on histologic tumor type, tumor size, and node involvement was extracted by physicians blinded to data on hormone-supplement use reported by the participants. Although permission to review pathology reports and hospital records could not be obtained for seven percent of cases, we based our analysis on all incident breast cancers because the rate of accuracy of selfreporting was extremely high (99.6 percent) among those for whom records were obtained. We omitted the small number of carcinomas in situ from analysis.
Population for analysis The endpoint for the primary analysis was the occurrence of incident breast cancer. We therefore excluded from the analysis all women who reported breast or other cancer (excluding nonmelanoma skin cancer) on the 1976 questionnaire. This left a total cohort of 118,300 women eligible for follow-up. The present analysis was limited to postmenopausal women. We classified a woman as postmenopausal from the time she returned a questionnaire on which she reported natural menopause or hysterectomy with bilateral oophorectomy. Self-report of menopause is highly reproducible in this cohort and is valid with regard to details of the extent of ovarian surgery.4In addition, we classified women reporting hysterectomy without bilateral oophorectomy as postmenopausal at the age when natural menopause had occurred in 90 percent of the cohort (54 years for current cigarette smokers and 56 years for nonsmokers). 434
Cancer Causes and Control. Vol 3. 1992
Statistical analysis The primary analysis used incidence rates with personmonths of follow-up as the denominator. For each participant, person-months were allocated according to the 1976 exposure variables and were updated according to information on follow-up questionnaires. For women who developed breast cancer or died, personmonths were assigned according to the covariate status reported on the most recent completed questionnaire, but follow-up terminated with the diagnosis of breast cancer or death. Women who reported a diagnosis of cancer other than nonmelanoma skin cancer on any questionnaire were excluded from subsequent follow-up. If no questionnaire was returned for a follow-up cycle, the most recently recorded covariate data were used for the subsequent follow-up interval. If the previous status was 'current' postmenopausal hormone use, however, the updated status of use was classified as missing. For past use and never use, the status of use was classified according to the most recent questionnaire. We used the rate ratio (RR) as a measure of association, defined as the incidence of breast cancer among women who had used postmenopausat hormones divided by the corresponding rate among women who had never used these drugs. Additional stratified analyses were carried out to control for risk factors and to explore the possibility that the effect of current use of postmenopausal hormones might be modified by these factors. We used the Mantel test for linear trend to examine the relation between duration of current use of postmenopausal hormones and risk of breast cancer,6 and proportional-hazards models to adjust for multiple risk factors simultaneously.7 We also used proportional hazards models to analyze survival status among patients with breast cancer according to the history of use of postmenopausal hor-
Estrogens, progestins, and breast cancer Table 1. Age-specific incidence rates of breast cancer by never use, ever use, current use, and past use of replacement hormone therapy, 1976-88' Age
Never use Cases
Personyears
30-39 40-44 45-49 50-54 55-59 60-66
0 5 36 163 221 104
1,039 4,147 23,907 84,587 95,366 35,560
Total
529
244,606
Ever use Rate b Cases
-120 151 193 232 292
Current use
Past use
Personyears
Rate
RR '
Cases
Personyears
Rate
RR
Cases
Personyears
Rate
RR
3 11 40 114 195 139
4,516 11,766 26,948 60,303 78,005 37,804
66 93 148 189 250 368
-0.78 0.98 0.98 1.06 1.24
3 7 26 64 92 62
3,471 8,526 16,652 29,982 28,377 11,377
86 82 156 213 324 545
-0.68 1.03 1.10 1.40 1.87
0 4 14 50 97 77
1,045 3,240 10,295 30,110 48,867 26,370
-123 136 166 198 292
1.03 0.90 0.86 0.85 1.00
502
219,342
1.08
254
98,385
1.34
242
119,927
0.90
Missing data on h o r m o n e use for 19 cases during 16,717 person-years; and on status (current or past) for six cases. b Rate per 100,000 person-years. ° RR = risk ratio.
mones prior to diagnosis. Survival status for the 722 incident cases of breast cancer diagnosed between 1976 and June 1986 were ascertained as of 1 June 1990 for 702 women; the remaining 20 women were censored when they were last known to be alive between June 1988 and June 1990. Overall, these women were censored for a total of 22.8 years of potential follow-up.
Results In 1976, 33 percent of postmenopausal women were current users of postmenopausal hormones, 18 percent were past users, 3.6 percent had missing information on use of postmenopausal hormones, and the remainder had never used these agents. From 1976 to 1988, current use of postmenopausal hormones accounted for 21 percent of follow-up time; past use, 25 percent; missing information on hormone use, four percent; and never use, 50 percent. As previously reported, women who had never used postmenopausal hormones and current and past users had similar distributions for age at menarche and family history of breast cancer.' Current users were slightly more likely to have reported a history of benign breast disease, to be lean, to consume alcohol, to be nulliparous, or to have given birth only once or twice. We examined the relation between ever-use, current, and past use of postmenopausal hormones and risk of breast cancer within five-year age groups (Table 1). Ever-use was not associated with increased risk of breast cancer, age-adjusted RR = 1.08 (95 percent confidence interval [CI] = 0.96-1.22). Examining current and past use separately, we observed an increased risk of breast cancer among current, but not among past users. The increased risk associated with current use of estrogen replacement therapy (ERT) was most evident
among women 55 years of age or older. Overall, the age-adjusted RR of breast cancer among current users was 1.29 (CI-- 1.11-1.50). After adjusting for age in single years, age at menopause, and type of menopause, the RR increased slightly to 1.33 (CI = 1.12-1.57). Past use was not associated with increased risk of breast cancer. The RR, adjusted for age, was 0.90 (CI = 0.77-1.04). We examined the association between current use of estrogen replacement and risk of breast cancer separately among women reporting natural menopause, bilateral oophorectomy, and hysterectomy without removal of ovaries. After age-standardizing to the distribution of women with a natural menopause who never took replacement hormone therapy, the RR for current use among the women with natural menopause was 1.33 (103 cases, 33,899 person-years); among women with a bilateral oophorectomy, the RR was 1.35 (114 cases, 53,865 person-years); and among women with hysterectomy without oophorectomy, the RR was 1.59 (35 cases, 7,321 person-years). To determine whether the association between current use and risk of breast cancer varied according to other risk factors, we examined the age-adjusted RR within strata of parity, age at first birth, family history of breast cancer in mother or sister, personal history of benign breast disease, age at menarche, and body mass index. The RRs varied little within strata of these risk factors providing little evidence for effect modification (Table 2). The trend for increasing risk among current users with later ages at first birth, and the lack of association among women with a personal history of benign breast disease are noteworthy. After multivariate adjustment for all the established risk factors, age in single years, age at menopause in years, and type of menopause, the RR was essentially unchanged from that adjusted for age alone for current use (RR = 1.33, Cancer Causes and Control. Vol3. 1992
435
G. A. Colditz et al CI=1.12-1.57), and for past use (RR=0.92, CI = 0.79-1.09). Addition of alcohol to models for follow-up from 1980-88 did not materially alter the results. We next examined duration of use and risk of breast
cancer among current and former users (Tables 3 and 4). We observed no evidence of increasing risk with increasing duration of use among either current or past users. Among current users, the test for trend in risk with increasing duration of use was not significant (trend, P = 0.41). Among past users, women who had used replacement estrogens for 10 or more years in the past had an age-adjusted RR of 0.75 (CI -- 0.47-1.21) compared with women who had never used replacement hormones. Further adjustment for age at menopause and type of menopause did not materially alter these results. Among past users, the test for trend in risk with increasing duration of use was not significant, P = 0.46. In a previous study, Gambrell 8 suggested that the addition of progestin was associated with a reduced risk of breast cancer of 0.3 (CI = 0.1-0.8). Therefore, we examined this question using follow-up data from 1978 when type of hormone replacement was first asked. During this 10-year follow-up period, we observed 972 incident cases of breast cancer. Compared with never-users, women who were current users of conjugated estrogen were at increased risk of breast cancer, as observed in the overall cohort analysis (the age-adjusted RR = 1.42, CI = 1.19-1.70) (Table 5). Current use of estrogen and progestin was associated with an age-adjusted RR of 1.54 (0.99-2.39). Women who used estrogen and testosterone were also at increased risk ( R R = 2.45, CI = 0.95-6.35). We had insufficient data to provide informative risk estimates for other forms of replacement therapy. Adjustment for type of menopause in addition to age did not materially alter these RR estimates, nor did multivariate analysis control for established breast-cancer risk factors, although confidence intervals broadened. Prior use of oral contraceptives (OC) before ERT has been raised as a possible modifier of breast cancer risk associated with estrogen use. We specifically
Table 2. Age-adjusted rate ratio (RR) and 95% confidence
interval (CI) for current and past use of estrogen replacement therapy within strata of established risk factors for breast cancer, 1976-88 Current use
Past use
RR
CI
RR
CI
Family history Yes No
1.33 1.30
(0.8%1.98) (1.10-1.52)
0.77 0.92
(0.51-1.17) (0.78-1.08)
History of BBD" Yes No
0.97 1.42
(0.75-1.26) (1.18-1.71)
0.81 0.91
(0.62-1.05) (0.75-1.10)
Menarche (age) 13
1.37 1.36 0.96
(1.10-1.70) (1.04-1.77) (0.67-1.38)
0.85 0.96 0.90
(0.68-1.07) (0.73-1.26) (0.65-1.26)
Parity Nulliparous 1-2 3+
0.94 1.58 1.17
(0.55-1.60) (1.23 -2.03) (0.95-1.44)
1.21 0.86 0.86
(0.77-1.91) (0.66-1.14) (0.70-1.06)
Age at first birth 20-24 25-29 30-34 35-39 40+
1.23 1.32 1.43 2.08 2.79
(0.97-1.63) (1.03-1.71) (0.94-1.69) (1.11-3.29) (0.85-9.19)
0.91 0.87 0.78 1.0 1.47
(0.69-1.19) (0.67-1.13) (0.50-1.20) (0.51-1.97) (0.40-5.40)
Body mass index (kg/m 2)
