age-specific differences in the relationship between oral contraceptive use and breast cancer

1506

Age-Specific Differences in the Relationship between Oral Contraceptive Use and Breast Cancer Phyllis A. Wingo, M.S.,* Nancy C. Lee, M.D.,* Howard W. Ory, M.D.,t Valerie Beral, M.B.B.S.,S Herbert B. Peterson, M.D.,* and Philip Rhodes, M.S.5

Background. Nearly all studies have suggested that the use of oral contraceptives (OC) is not associated with the aggregate risk of breast cancer diagnosed in women aged 20-54 years. Because of age-specific differences in the breast cancer-parity relationship and because of age- specific differences in other breast cancer risk factors, the Centers for Disease Control reexamined data from the Cancer and Steroid Hormone Study (CASH) to assess whether OC use has different effects on the risk of breast cancer at different ages of diagnosis.

Methods. This population-based case-control study was designed to examine the relationship between the use of OC and the risk of breast, ovarian, and endometrial cancer. CASH was conducted in eight geographic areas in the United States during 1980-1982. All participants were interviewed at home with a pretested standardized questionnaire including a calendar of life events and a photograph book of all pills marketed in the United States.

We found that the relationship between the risk of breast cancer and OC use appeared to vary by the age at diagnosis. Among women aged 20-34 years at

Presented at the National Conference on Gynecologic Cancers, Orlando, vorida, April 2-4, 1992.

From the *Division of Reproductive Health, Center for Chronic Disease Prevention and Health Promotion; the tInformation Re- sources Management Office; and §the Division of Injury Control, Center for Environmental Health and Injury Control, Centers for Disease Control, Department of Health and Human Services, At- lanta, Georgia, and the $Imperial Cancer Research Fund, Cancer Epi- demiology Unit, Radcliffe Infirmary, Oxford, England.

Supported by interagency agreement 3-Y01-HD-8-1037 between the Centers for Disease Control (Atlanta, Georgia) and the National Institute of Child Health and Human Development, with additional support from the National Cancer Institute (Bethesda, Maryland).

The authors thank James J. Schlesselman, David C. G. Skegg, George L. Rubin, and Peter M. Layde for comments and suggestions on the manuscript; Karen S. Colberg for her computer assistance; and Beverly H. Malone for her word-processing help. The authors ac- knowledge the contributors to the Cancer and Steroid Hormone Study: study design and coordination, The Division of Reproductive Health, Center for Chronic Disease Prevention and Health Promo- tion, Centers for Disease Control; principal investigator, George L.

Results.

diagnosis or interview, those who had ever used OC had a slightly increased risk of breast cancer (odds ratio [OR], 1.4; 95% confidence interval [CI], 1.0-2.1) compared with women of the same ages who had never used OC. Among these women, there were no trends of increasing or decreasing risk with any measure of OC use. Among women aged 35-44 years, there was no association between OC use and breast cancer. Among women aged 45- 54 years, those who used OC had a slightly decreased risk of breast cancer (OR, 0.9; 95% CI, 0.8-1.0). Among these women, risk estimates decreased significantly with increasing time since first and last use.

Conclusions. Although the slightly increased risk estimates for the youngest women were compatible with findings by other investigators, the decreased risk estimates for the oldest women have not been described in as many studies. Available data provide no reasons to change prescribing practices or the use of OC that are related to the breast cancer risk. Cancer 1993; 71:

Key words: oral contraceptives, breast cancer, case- control studies, epidemiology.

1506-17.

~~~

Rubin,M.B.,F.R.A.C.P.;projectdirector,PhyllisA. Wingo,M.S.;project associates, Nancy C. Lee, M.D., Michele G. Mandel, B.A., Herbert B Peterson, M.D.; data collection centers and principal investigators, Atlanta: Raymond Greenberg, M.D.; Connecticut: J. Wister Meigs, M.D., and W. Douglas Thompson, Ph.D.; Detroit, Michigan: G. Mane Swanson, Ph.D.; Iowa: Elaine Smith, Ph.D.; New Mexico: Charles Key, M.D., and Dorothy Pathak, Ph.D.; San Francisco: Donald Aus- tin, M.D.; Seattle: David Thomas, M.D.; Utah: Joseph Lyon, M.D., and Dee West, Ph.D.; pathology review principal investigators, Fred Gorstein, M.D., Robert McDivitt, M.D., and Stanley J . Robboy, M.D.; project consultants, Lonnie Bumett, M.D., Robert Hoover, M.D., Peter M. Layde, M.D., M.Sc., Howard W. Ory, M.D., M.Sc., James J . Schlesselman, Ph.D., David Schottenfeld, M.D., Bruce Stadel, M.D., Linda A. Webster, M.S.P.H., and Colin White, M.B.B.S.; and pathology consultants, Walter Bauer, M.D., William Christopherson, M.D., Deborah Gersell, M.D., Robert Kurman, M.D., Allen Paris, M.D., and Frank Vellios, M.D.

Address for correspondence: Phyllis A. Wingo, M.S., Centers for Disease Control, 1600 Clifton Road NE, Mail-stop K 34, Atlanta, GA 30333.

Accepted for publication September 2, 1992.

Oral Contraceptives and Breast Cancer/Wingo et al . 1507

Nearly all epidemiologic reports on the subject confirm that the risk of breast cancer diagnosed in women at ages 20-54 years is the same for those who have ever used oral contraceptives (OC) as for women who have never used them.'-3 In 1983, two reports suggested that using OC before the first term pregnancy or using "high progestogen" OC before age 25 years increased the risk of breast cancer diagnosed at young age^.^,^ More recent publications suggest possible increased risks of breast cancer associated with OC use among specific groups of users, including premenopausal women, nulliparous women, and women with long durations of use, espe- cially before the first term pregnancy.'p6-" Analyses of premenopausal women in the Cancer and Steroid Hor- mone Study data suggested a possible increased risk of breast cancer among nulliparous women aged 20-44 years and a decreased risk among nulliparous women aged 45-54 years.' Although the subgroups of users with increased risks varied across studies, most reports with positive findings involved breast cancer diagnosed at younger ages, usually younger than 35 years.

Several risk factors for breast cancer, including marital status, age at menarche, age at first birth, race, and birthplace alter the breast cancer risk in opposite directions, depending on the age at diagnosi~. '~-'~ The breast cancer-parity relationship varies by age.I3-l7 Women with high parity who are younger than 35 years of age have an increased risk of breast cancer compared with women with low parity of the same age. After a woman reaches approximately 35-40 years of age, parity confers the well-described protection against breast cancer. However, because the incidence of breast cancer before age 35 years is low and because most reports only describe the parity effect for all ages combined, high parity generally is considered to be protective against breast cancer.

One author suggested that the breast cancer-OC relationship may parallel the breast cancer-parity relationship and may vary by age, with an increased risk for women diagnosed at young ages and no risk or a decreased risk for women diagnosed at older To assess possible age-specific differences in the breast cancer-OC relationship and to pursue the possible anal- ogy with the breast cancer-parity relationship, we reexamined data from the Cancer and Steroid Hormone Study. 19,'0

Materials and Methods

The methods for the Cancer and Steroid Hormone Study have been described in detail previously.20,21 The study was a population-based case-control study conducted in eight geographic locations in the United States. The cases were 471 1 women aged 20-54 years

who were living in the study areas and had histologi- cally confirmed breast cancer, initially diagnosed between December 1, 1980 and December 31, 1982. The control subjects were 4676 women aged 20-54 years who were identified by Waksberg's method of tele- phoning randomly selected residential telephone num- bers in the same locations where the women with breast cancer resided." Control subjects were matched to the expected 5-year geographic-specific age distribution of the patients with breast cancer. All study participants were interviewed in person at home with a pretested standardized questionnaire. The interview lasted approximately 50 minutes and focused on reproductive and contraceptive histories, family histories of cancers, and personal characteristics and behaviors.

We used unconditional logistic-regression methods to adjust the estimates of relative risk within three age groups (ages 20-34, 35-44, and 45-54 years) for vari- ables considered a priori as potential confounder^.'^ The risk factors that were considered as potential con- founders included age at diagnosis or interview, parity, age at menarche, age at first term pregnancy, Quetelet's index of adiposity, menopausal status, family history of breast cancer, history of biopsy for benign breast disease before use of OC, regularity of menses during ado- lescence, frequency of breast examinations, breast feeding, a history of spontaneous or induced abortions, and geographic location. The age at diagnosis or interview was assessed in different logistic-regression models both as a continuous variable and as a matched variable. Because the results from both approaches were similar, we included age in the final models as a continuous variable. We also included those risk factors that altered the crude relative risk estimates by more than 5%. Models that included all confounding factors from any of the three age-specific models yielded compara- ble results.

We used likelihood ratio tests to assess multiplica- tive interactions between OC use and the risk factors of intere~t.'~ When we tested for the statistical significance of trends for the various measures of OC use, we excluded women who had never used OC and included months of OC use in the model as a continuous variable.

Results

To assess whether the relationship between OC use and the risk of breast cancer varied by age, we considered several logistic-regression models, Regardless of the age or OC groupings, the interaction between age and OC use was statistically significant (P c 0.05, by likelihood ratio tests).23 Because age seemed to modify the effect of OC use on the risk of breast cancer, we presented all

1508 CANCER Supplement February 25, 2993, Volume 71, No. 4

findings stratified by age at diagnosis or interview: 20- 34 years, 35-44 years, and 45-54 years.

Among women aged 20-34 years, those with breast cancer were more likely than women who were control subjects to be black, to have had their first menstrual period before age 12 years, to have higher panty,

and to be obese. By contrast, among women aged 45-54 years, women with breast cancer were more likely to be premenopausal and nulliparous than women of the same ages who were control subjects (Table 1).

When we examined patterns of OC use between cases and controls by age and the characteristics listed

Table 1. Age-Specific Percent Distributions of Characteristics for Breast Cancer Cases and Controls: Cancer and Steroid Hormone Study, 1980-1982

Age at diagnosis or interview (vr)

Characteristic

20-34 35-44 45-54

Cases Controls Cases Controls Cases Controls

No. Race

White Black Other/unknown

< High school High school College Unknown

Income ($)

< 10,000

10,000-19,999 20,000-29,999 30,000-39,999

40,000-49,999

Education

50,000+ Refused/unknown

< 12 yr 12-14 yr 14+ yr Other/unknown

Menopausal status Premenopausal Perimenopausal Natural menopause Surgical menopause Other/unknown

Parity 0 1-2

3-4 4+ Unknown

Age first menstrual period

Age first term pregnancy Nulliparous women c 20 yr 20-24 yr 25+ yr Unknown

Breast-feeding YeS

No Unknown

524

77.5%

18.3 4.2

8.8 28.4

62.8 -

14.1 21.2

29.2 14.3 9.7

8.4 3.1

24.2 69.3 6.1

0.4

91.0 - -

4.6

4.4

27.5%

53.6 17.4 1.1

0.4

27.5

17.9 32.1

21.6 0.9

31.5

68.3 0.2

704

84.9%

11.4 3.7

7.4 29.6

62.9 0.1

11.9 24.4

26.7 18.3 7.7

6.4 4.6

18.5 71.3 10.1

0.1

93.2 - -

2.8

4.0

32.2%

50.0 16.8

0.9 0.1

32.2

18.7 30.1

18.0 1.0

36.1

63.8 0.1

1565

85.4% 11.0

3.6

10.3

29.8 59.9 -

8.2 16.6 24.7

19.7 11.5

14.4 4.9

25.9 65.0 8.9

0.2

69.0 9.8

0.6 15.5

5.1

15.2% 45.1

32.3 7.0

0.4

15.2

19.4 37.3 27.0

1.1

40.1

58.9 1 .o

1361

87.2% 9.9

2.9

10.6 30.9

58.5 -

8.2 16.9

23.4 19.6

14.5 13.2 4.2

25.4

62.6 11.5 0.5

60.6 13.5 0.7 19.2

6.0

11.2% 39.4 38.0

11.0 0.4

11.2 23.9 40.9

22.2 1.8

45.8 53.3 0.9

2622

88.5%

8.7 2.8

16.4 36.2

47.3 0.1

10.2 18.8

22.9 16.6

11.0 12.9 7.6

22.4

65.5 11.8 0.3

21.7 25.0 23.0 28.7

1.6

14.9% 33.3 35.9

15.0 0.9

14.9 16.8 39.8

25.6 2.9

42.9 56.4 0.7

261 1

87.0%

10.5 2.5

19.2 36.5

44.2 0.1

10.4 19.0

22.8 18.5

11.4 12.0 5.9

20.2

67.7 11.7 0.4

12.2 25.4 26.0

35.5 0.9

8.7% 29.3 40.4

20.8 0.8

8.7 22.1 42.9

23.3 3.0

49.1 49.7

1.2

Oral Contraceptives and Breast Cancer/ Wingo et al. 1509

Table 1. (Continued)

Age at diagnosis or interview (yr)

20-34 35-44 45-54

Characteristic Cases Controls Cases Controls Cases Controls

Spontaneous and induced abortions None 72.7% 72.5% 70.8% 70.2% 71.3% 69.0% 1 17.7 19.0 19.0 18.5 18.4 19.5

1+ 9.2 7.8 9.4 10.5 8.8 10.3

Unknown 0.4 0.7 0.8 0.8 1.5 1.2

Family history of breast cancer None 57.6 74.3 57.2 66.0 53.7 61.8 First-degree 9.7 2.6 11.4 5.2 11.7 7.0 Second-degreen 21.4 16.3 22.2 20.1 20.5 18.5 Unknown 11.3 6.8 9.2 8.7 14.1 12.7

Biopsy for benign breast disease before oral contraceptive use

Yes 3.4 1.8 4.9 2.5 13.3 8.6

No 90.5 92.8 86.1 91.2 77.0 82.8 Unknown 6.1 5.4 9.0 6.3 9.7 8.6

in Table 1, we found that cases and controls of the same ages reported similar percentages of use (Table 2). How- ever, across age groups, the patterns of use demon- strated birth cohort effects. Older women were approximately half as likely to have ever used OC as younger women. Among women aged 45-54 years, fewer than 0.1 % first used OC before age 20 years compared with more than 35% of women younger than 35 years of age at diagnosis or interview. The mean ages at first use of OC were 20 years for the women who were youngest at diagnosis or interview, 24 years for women aged 35-44 years, and 32 years for women who were 45-54 years. Among the oldest OC users, approximately 54% of the cases and controls had ever used OC containing more than 50 pg of estrogen compared with 42% of cases and controls who used OC and were younger than 35 years of age at diagnosis or interview. Among parous women, 3% of the oldest cases and controls used OC before the first term pregnancy compared with approximately 5 1 YO of youngest cases and controls.

Among the women who were youngest at diagnosis or interview, OC users had an increased risk of breast cancer, which was of borderline statistical significance (odds ratio [OR], 1.4; 95% confidence interval [CI], 1.0-2.1; Table 3). Among the oldest women, OC users had a decreased risk of breast cancer, which was of borderline statistical significance (OR, 0.9; 95% CI, 0.8-1.0). There were no trends of increasing or decreasing risk with increasing duration of use for any age group. When we excluded women who had ever used OC containing more than 50 pg of estrogen to examine the effect of OC formulations similar to currently available pills, the risk estimate for women younger than 45 years of age were unaffected, and the risk estimate for older women who had ever used OC was 1.0.

When we examined months since first and last use of OC, several dose-response patterns emerged (Table 4). Among women aged 45-54 years, those who last used OC more than 20 years (240 months) earlier had a statistically significant reduced risk of breast cancer (OR, 0.6; 95% CI, 0.3-0.8). The trend of decreasing risk with increasing time since last use was significant. There was also a significant trend among women aged 35-44 years at diagnosis or interview. When we excluded women who had used OC containing more than 50 pg of estrogen, neither trend of decreasing risk with increasing time since last use remained significant.

Consistent with our findings for time since last use of OC were statistically significant trends of decreasing risk with increasing months since first use among the two older age groups (Table 4). Excluding women who had ever used OC containing more than 50 pg of estrogen moderated the trends of decreasing risk with increasing time since first use, although the trend in the oldest group remained significant (P c 0.01).

We also examined the age-specific risks of breast cancer associated with the exclusive use of specific formulations of OC. Other than the findings of slightly increased risks for the youngest women and slightly decreased risks for the oldest women shown for other measures of OC use, we were unable to identify any specific OC formulations with obvious patterns of effect. When we grouped OC by estrogen dose, the risk estimates for women who exclusively used OC with 50 pg or less of estrogen were 1.3, 1.1, and 1.0 for women aged 20-34 years, 35-44 years, and 45-54 years, respec- tively. For women who exclusively used pills containing more than 50 pg of estrogen, the risk estimates were 1.5, 0.9, and 0.8. None of these estimates was statistically significant.


Table 2. Percent Ever Use of Oral Contraceptives Use By Age and Selected Characteristics of Breast Cancer Cases and Controls: Cancer and Steroid Hormone Study, 1980-1982

Characteristic

Age at diagnosis or interview (yr) 20-34 35-44 45-54

Cases Controls Cases Controls Cases Controls

No. Total Race

White Black

Education < High school High school College

Income ($) < 10,000 10,000-19,999 20,000-29,999 30,000-39,999 40,000-49,999 50,000+

< 12 12-14

Age first menstrual period (yr)

14+ Menopausal status

Premenopausal Perimenopausal Natural menopause Surgical menopause

Parity 0 1-2 3-4 4+

< 20 yr 20-24 yr 25+ yr

Breast-feeding Yes No

None 1 1+

Age first term pregnancy

Spontaneous and induced abortions

Biopsy for benign breast disease before oral contraceptive use

Yes No

None First-degree

Family history of breast cancer

524 88.9%

91.6 88.2

89.1 89.3 88.8

87.8 91.0 86.3 86.7 92.2 95.4

89.8 88.2 93.8

88.9 - -

91.7

84.0% 91.8 90.1 83.3

98.9 86.9 91.2

92.1 87.4

87.4 95.7 91.7

50.0 89.9

89.4 90.2

Second-degree 90.2

704 85.240

87.6 78.8

76.9 89.4 84.4

77.4 86.0 86.7 90.7 83.3 86.7

82.3 86.1 84.5

85.2 - -

85.0

77.1% 90.9 83.9 83.3

88.6 88.2 91.3

89.0 83.1

83.1 93.3 85.4

38.5 85.3

84.7 94.4 87.8

1565 79.046

79.5 80.0

61.1 76.6 83.2

72.1 72.6 78.3 82.8 87.2 85.8

76.9 79.8 79.9

79.4 70.8 66.7 81.5

66.8% 84.1 78.4 73.6

79.3 80.3 83.2

84.2 75.4

76.2 86.6 84.4

43.4 79.0

80.0 76.4 78.7

1361 78.0%

80.4 66.9

64.8 76.2 81.4

60.7 79.6 79.3 80.1 83.7 77.1

76.6 79.6 73.1

78.6 71.2 66.7 80.5

65.4% 80.8 81.2 69.1

77.8 81.1 79.8

80.0 76.8

76.1 82.5 81.1

44.1 77.4

79.6 70.4 78.5

2622 43.5%

44.0 40.3

34.8 40.5 49.0

32.2 38.2 45.3 46.1 48.8 54.1

44.5 43.5 42.6

49.0 44.1 36.3 45.0

28.4% 42.7 48.2 48.4

44.2 45.7 47.8

46.8 40.9

40.8 49.2 53.2

15.7 42.6

43.8 43.1 47.3

261 1 47.5%

48.9 41.7

35.4 46.6 53.5

36.5 46.4 47.6 48.4 54.5 53.7

44.0 47.7 52.5

52.4 52.0 42.2 46.6

30.4% 46.6 51.0 48.8

43.1 50.5 51.6

50.2 44.5

44.9 53.0 53.7

23.2 44.5

49.0 45.6

., 48.8

Oral Contraceptives and Breast Cancer/Winp et ul. 1511

Table 3. Age-Specific Relative Risk of Breast Cancer by Duration of Oral Contraceptive Use: Cancer and Steroid Hormone Study, 1980-1982

Months of oral contraceptive use Cases Controls RR* 95% CI Cases Controls RRt 95% CI Cases Controls RR+ 95% CI

Age at diagnosis or interview (yr) 20-34 35-44 45-54

Never use Ever use < 6 6-1 1 12-23 24-47 48-71 72-95 96-119 120-143 144-167 168+ Test for linear

trend P

57 425 13 32 66 105 85 52 35 37

104 Ref 328 298 Ref 547 1.4 (1.0, 2.1) 1190 1031 1.1 (0.9, 1.3) 29 0.9 (0.4, 1.9) 183 127 1.3 (1.0, 1.7) 55 1.2 (0.7, 2.2) 96 97 0.9 (0.7, 1.2) 86 1.5 (0.9, 2.4) 145 147 0.9 (0.7, 1.2) 137 1.5 (1.0, 2.4) 215 197 1.0 (0.8, 1.3)

’ 90 1.9 (1.2, 3.0) 162 153 1.0 (0.7, 1.3) 68 1.3 (0.8, 2.3) 125 111 1.0 (0.8, 1.4) 46 1.3 (0.7, 2.3) 97 61 1.5 (1.0, 2.1) 36 1.46 (0.8, 2.6) 83 60 1.3 (0.9, 1.8)

56 46 1.1 (0.7, 1.7) - 28 32 0.8 (0.5, 1.4) -

0.7 0.2

1453 888 96 90 128 135 124 85 69 69 43 49

1371 Ref 991 0.9 (0.8, 1.0) 103 0.9 (0.7, 1.2) 119 0.7 (0.6, 1.0) 151 0.8 (0.7, 1.1) 156 0.8 (0.7, 1.1) 120 1.0 (0.8, 1.3) 87 1.0 (0.7, 1.3) 84 0.8 (0.6, 1.1) 61 1.1 (0.8, 1.6) 46 0.9 (0.6, 1.4) 64 0.8 (0.5, 1.1)

0.9

RR: relative risk; CI: confidence interval. Adjusted for age at diagnosis or interview, family history of breast cancer, history of benign breast disease before oral contraceptive use, Quetelet’s index, frequency

of breast examinations, and breast-feeding; 42 cases and 53 controls with unknown covariates were excluded. t Adjusted for age at diagnosis or interview and frequency of breast examinations; 47 cases and 32 controls with unknown covariates were excluded.

were excluded. 5 Last category with estimate includes all women with longer durations of use.

Adjusted for age at diagnosis or interview and history of benign breast disease before oral contraceptive use; 281 cases and 249 controls with unknown covariates

Table 5 presents the risk of breast cancer diagnosed at different ages by the age when the women first used OC. Among women aged 45-54 years at diagnosis or interview, there was a statistically significant trend of decreasing breast cancer risk with decreasing age at first use of OC. This protective effect was most pronounced among the women who first used OC before age 25 years (OR, 0.5; 95% CI, 0.3-0.8). This pattern was not apparent for women with younger ages of diagnosis. Because age at first exposure and months since first exposure were correlated highly, we conducted analyses to disentangle the effects among the women 45-54 years old. Our results suggested that the decreased risks appeared more likely to be related to the effects of months since first use than to age at first use.

In Table 6 , we present age-specific risks of breast cancer by duration of OC use for nulliparous and parous women. For nulliparous women, the duration includes the total months of OC use. For parous women, the duration is categorized in relation to their first term pregnancy. For parous women who first used OC before the first term pregnancy, the duration includes only the months of use before the first term pregnancy, regardless of the subsequent use. For parous women who first used OC after the first term pregnancy, the duration includes the total months of OC use. In general, regardless of the duration or timing of OC use with

respect to the first term pregnancy, women who used OC and were younger than age 45 years had higher risks of breast cancer than women of similar age and timing who had never used OC. The only exception appeared to be women aged 35-44 years who first used OC after their first term pregnancy; estimates of risk for these women varied around 1.0. Among nulliparous women, there were two trends related to increasing duration of OC use that were of borderline statistical significance: a trend of increasing risk among women aged 35-44 years and a trend of decreasing risk among the oldest women. Stratification by menopausal status suggested that the these trends were confined to premenopausal women.’

Discussion

The main new finding from the current analysis is the suggestion that the breast cancer-OC relationship varies by age at cancer diagnosis. In general, the risk of breast cancer decreased across the three age groups for most categories of duration, the time since first or last use, and the age at first use of OC. For women aged 20-34 years of age, OC use appeared to increase slightly the risk of breast cancer but did not appear to affect the risk of breast cancer for women aged 35-44


Table 4. Age-Specific Relative Risk of Breast Cancer by Timing of Oral Contraceptive Use: Cancer and Steroid Hormone Study, 1980-1982

Age at diagnosis or interview (yr) 20-34 35-44 45-54 Timing of oral

contraceptive use RR 95% CI RR 95% CI RR 95% CI

Never use Ref Ref t Ref Months since last use

< 12 1.7 (1.1, 2.6) 1.2 (0.8, 1.8) 0.8 (0.4, 1.5) 12-23 1.1 (0.6, 2.1) 1.2 (0.7, 2.3) 0.8 (0.3, 2.2) 24-47 1.2 (0.7, 1.9) 1.5 (1.0, 2.2) 1.0 (0.7, 1.5) 48-71 1.8 (1.1, 3.0) 1.2 (0.9, 1.6) 1.1 (0.8, 1.4) 72-95 1.5 (0.9, 2.5) 1.1 (0.8, 1.5) 1.0 (0.8, 1.3) 96-119 1.5 (0.9, 2.6) 1.0 (0.7, 1.3) 1.3 (0.9, 1.7) 120-143 1.3 (0.7, 2.4) 1.0 (0.7, 1.3) 0.9 (0.7, 1.2) 144-167 1.05 (0.5, 1.8) 0.9 (0.7, 1.2) 0.8 (0.6, 1.1) 168-191 - 1.0 (0.7, 1.3) 0.8 (0.6, 1.1) 192-21 5 - 1.2 (0.8, 1.7) 0.6 (0.5, 0.8) 216-239 - 0.9 (0.5, 1.5) 1.0 (0.7, 1.5) 240+ - 0.6 (0.3, 1.5) 0.6 (0.4, 0.8) Test for linear trend P 0.5 < 0.01 < 0.01

< 48 1.3 (0.6, 2.6) - - 48-71 0.7 (0.4, 1.5) - -

Months since first use

72-95 1.5 (0.8, 2.6) 1.45 (0.8, 2.4) 1.05 (0.6, 1.7) 96-119 1.7 (1.0, 2.7) 1.5 (0.9, 2.5) 1.9 (1.1, 3.2) 120-143 1.8 (1.2, 2.9) 1.1 (0.8, 1.7) 1.1 (0.7, 1.6) 144-167 1.2 (0.8, 2.0) 1.2 (0.9, 1.6) 0.9 (0.7, 1.2) 168-191 1.55 (0.9, 2.4) 1.0 (0.8, 1.3) 1.0 (0.8, 1.3) 192-215 - 1.0 (0.8, 1.2) 0.9 (0.7, 1.1) 216-239 - 1.0 (0.7, 1.2) 0.9 (0.7, 1.2) 240-263 - 1.5 (1.0, 2.2) 0.6 (0.5, 0.9) 264-287 - 0.8 (0.4, 1.7) 0.8 (0.5, 1.2) 288+ - 0.4 (0.1, 1.2) 0.5 (0.3, 0.8) Test for trend P§ 0.4 0.04 < 0.01

RR: relative risk; CI: confidence interval. Adjusted for age at diagnosis or interview, family history of breast cancer, history of benign breast disease before

oral contraceptive use, Quetelet's index, frequency of breast examinations, and breast-feeding. Cases and controls with unknown covariates were excluded. t Adjusted for age at diagnosis or interview and frequency of breast examinations. Cases and controls with unknown covariates were excluded. $ Adjusted for age at diagnosis or interview and history of benign breast disease before oral contraceptive use. Cases and controls with unknown covariates were excluded. lj Last (first) category with estimate includes all women with a longer (shorter) time interval since first or last use.

years. It appeared to confer a slightly decreased risk of breast cancer diagnosis for women aged 45-54 years.

We strongly caution against the overinterpretation of these findings. The Cancer and Steroid Hormone Study was large and had substantial statistical power. Although the age-specific differences in risk were consistent for the various measures of OC use, the magni- tudes of the risk estimates were close to 1.0. Increased risks less than 1.5 and decreased risks greater than 0.8 should be interpreted with circum~pection.~~ Further- more, the age groups presented in this analysis reflect

different birth cohorts, that varied in the formulations, timing, and duration of OC used; these differences may have affected our findings.

Multiple statistical tests and chance also must be considered as possible explanations for the current and other results from the Cancer and Steroid Hormone Study. There have been seven analyses of the relationship between breast cancer and the use of OC from these data:7,19.20.25-28 two planned before data collec- t i ~ n , ' ~ , ~ ' and five conceived after 1982 and based on data from subgroups of women (women younger than

Oral Contraceptives and Breast Cancer/Wingo et al . 1513

Table 5. Age-Specific Relative Risk of Breat Cancer by Age at First Use of Oral Contraceptives: Cancer and Steroid Hormone Study, 1980-1982

Age at diagnosis or interview (yr) 20-34 35-44 45-54 Age at first oral

Never used Ref Ref t Ref $ Cases/controls 57/104 328/298 1453/1371 Ever use Case/controls 432/556 11 17/976 915/1015

20-24 1.5 (1.0, 2.3) 1.0 (0.8, 1.3) 0.55 (0.3, 0.8) 25-29 1.18 (0.6, 2.0) 1.0 (0.8, 1.3) 0.8 (0.7, 1.0) 30-34 - 1.4 (1.0, 1.9) 0.9 (0.7, 1.0) 35+ - - 1.0 (0.9, 1.2) Test for linear trend P 0.4 0.8 0.0001

contraceptive use (yr) RR 95% cr RR 95% CI RR 95% CI

< 20 1.4 (0.9, 2.1) 1.0 (0.7, 1.3) -

RR: relative risk; CI: confidence interval. Adjusted for age at diagnosis or interview, family history of breast cancer, history of benign breast disease before

oral contraceptive use, Quetelet's index, frequency of breast examinations, and breast-feeding; 35 cases and 44 controls with unknown covariates were excluded. t Adjusted for age at diagnosis or interview and frequency of breast examinations; 120 cases and 87 controls with unknown covariates were excluded.

Adjusted for age at diagnosis or interview and history of benign breast disease before oral contraceptive use; 254 cases and 225 controls with unknown covariates were excluded. 5 Last (first) category with an estimate includes all women with older (younger) age at first use of oral contraceptives.

45 years of age, women with a family history of breast cancer, or premenopausal w ~ m e n ) . ' , ~ ~ - ~ ~ In consider- ation of the large number of comparisons done to in- vestigate the relationship between OC and breast cancer, the probability of identifying subgroups of women with statistically significant increased or decreased risks would be likely.

Subgroup analyses from a study as large as the Cancer and Steroid Hormone Study are difficult to interpret when the overall risk estimate is 1.0 and the subgroup dose-response relationships are inconsistent and comparatively weak.24 Although we believe, in general, that results previously published from the Cancer and Steroid Hormone Study cannot be explained by biases, we cannot claim, nor can any epidemiologic study claim, that bias does not explain the weak effects found in the current analysis.

To address continuing controversies, we separately analyzed the younger and older women. In our data, the protection that OC use appeared to confer against breast cancer among the oldest women was a more con- vincing epidemiologic pattern than the increased risk related to OC use among the youngest women. The protection evidenced for the oldest women was supported by several dose-response rela tionships; the risks decreased significantly with decreasing age when they first used OC and with increasing time since first and last use. The consistently elevated risks of breast cancer

among the youngest women, without any supporting dose-response patterns, are more difficult to interpret.

Using the Cancer and Steroid Hormone Study data, we and other investigators have reported results regarding the use of OC before the first term pregnancy. Each analysis focused on different groups of women: all women (ages 20-54 years), women younger than age 45 years, and premenopausal women; the latter group also included perimenopausal In addition to the current analysis, there are two reports con- cerning all w ~ m e n . ' ~ * ~ ~ In the 1983 Centers for Disease Control report based on a preliminary analysis and the 1986 Centers for Disease Control report based on the complete data set, women who used OC before the first term pregnancy had relative risk estimates of 1.3 and 1 .O, r espec t i~e ly . '~~~~ The analysis of women younger than age 45 years found slightly elevated relative risks of 1.3, 1.1, and 1.2 for durations of 12 months or less, 13-48 months, and more than 48 months of use before first term pregnan~y.'~ Finally, the analysis of the use of OC before first term pregnancy among premenopausal women identified two trends among nulliparous women: a trend of increasing risk among women aged 20-44 years and a trend of decreasing risk among older women.' These trends were related to the borderline trends reported in the current analysis.

Acceptance of new epidemiologic findings by the scientific community depends on consistency with and

1514 CANCER Supplement February 25, 2993, Volume 71, NO. 4

Table 6. Age-Specific Relative Risk of Breast Cancer Among Nulliparous and Parous Women by Duration of Oral Contraceptive Use: Cancer and Steroid Hormone Study, 1980-1982

Aee at diaenosis or interview (vr) 20-34 35-44 45-54 Duration of oral

contraceutive use (mo) RR 95% CI RR 95% CI RR 95% CI

Nulliparous women Never use

Case/controls

Ever use Case/controls < 24 24-47 48-71 72-95 96+

Test for linear trend P Parous women

Never use

Duration of use before Cases/controls

first term pregnancy

Ever use Cases/controlsll < 24 24-47 48-71 72+

Test for linear trend P Duration of first use

after first term pregnancy7

Ever use Case/controls < 24 24-47 48-71 72-95 96-119 120-143 144-167 168+

Ref * 22/52

1.5 (0.8, 2.7) 107/161

1.3 (0.6, 2.9) 1.2 (0.5, 2.9) 2.0 (0.9, 4.7) 1.3 (0.5, 3.4) 1.4 (0.6, 3.2) 0.5

Ref * 33/52

1.4 (0.9, 2.4) 180/225

1.6 (0.9, 2.8) 1.7 (0.9, 3.2) 1.5 (0.7, 3.2) 1.3 (0.5, 3.1) 0.5

1.4 (0.8, 2.4) 137/158

1.2 (0.6, 2.2) 1.4 (0.7, 2.8) 1.8 (0.8, 3.9) 1.3 (0.5, 3.0) 1.65 (0.7, 3.3) - - - 0.4

Reft 79/52

1.0 (0.7, 1.6) 147/96

0.9 (0.5, 1.7) 0.8 (0.4, 1.6) 1.2 (0.6, 2.5) 1.7 (0.6, 4.5) 1.7 (0.8, 3.5) 0.1

Reft 249/246

1.4 (1.1, 1.8) 318/247

1.3 (1.0, 1.9) 1.4 (1.0, 2.1) 1.6 (0.9, 2.8) 1.4 (0.8, 2.6) 0.5

1.0 (0.8, 1.3) 631/611

1.0 (0.7, 1.2) 1.0 (0.7, 1.4) 0.9 (0.7, 1.3) 0.9 (0.6, 1.4) 1.3 (0.9, 2.0) 1.4 (0.9, 2.1) 1.0 (0.6, 1.6) 0.8 (0.4, 1.6) 0.3

Ref$ 274/158

1.0 (0.7, 1.5) 80/49

1.2 (0.7, 2.1) 1.0 (0.4, 2.2) 0.85 (0.5, 1.5)

- -

0.1

Ref $ 1170/1203

(0.7, 1.8) 1.1 30/30

1.2 (0.7, 2.1) 0.95 (0.3, 2.7)

- -

0.3

(0.8, 1.0) 0.9 753/879

0.8 (0.7, 1.0) 0.9 (0.7, 1.1) 1.1 (0.8, 1.4) 1.1 (0.8, 1.5) 1.0 (0.7, 1.3) 1.2 (0.8, 1.7) 1.0 (0.7, 1.6) 0.8 (0.5, 1.2) 0.4 Test for linear trend P ~~

R R relative risk, CI: confidence interval. Adjusted for age at diagnosis or interview, family history of breast cancer, history of benign breast disease before

oral contraceptive use, Quetelet's index, frequency of breast examinations; 45 cases and 56 controls had unknown covariates and were excluded. t Adjusted for age at diagnosis or interview and frequency of breast examinations; 141 cases and 109 controls had unknown covariates and were excluded. $ Adjusted for age at diagnosis or interview and history of benign breast disease before oral contraceptive use; 315 cases and 292 controls had unknown covariates and were excluded. 4 Last category with an estimate includes all women with longer durations of use. (1 Restricted to women who first used oral contraceptives before their first term pregnancy, regardless of oral contraceptive use subsequent to fiat term pregnancy. ll Restricted to women who first used oral contraceptives after their fiat term pregnancy.

Oral Contraceptives and Breast Cancer/Wingo et al. 1515

confirmation by other studies. Unfortunately, our abil- ity to compare our age-specific findings with the exist- ing literature is limited. First, many recent investiga- tions were restricted to the relationship between OC and breast cancer diagnosed before age 45 years and were therefore unable to address the relationship between OC and breast cancer diagnosed at older ages.' Second, most remaining studies, like our earlier reports, did not provide age-specific estimates by different measures of OC use, such as duration and time since first and last use.'

Among the studies that presented age-specific results, three included estimates for ever use of OC that appeared to vary by age.3*6*29 In women aged 35-54 years, a relative risk of 0.9 was reported to be associated with ever use of OC, whereas women younger than 35 years had a 2.2-fold increased risk of breast cancer.29 Others detected slightly reduced risks for women older than 45 years of age and slightly elevated risks for women younger than 45 years.6 In a cohort study, among older women, those who used OC had a lower risk of breast cancer than women who never used OC.3 However, among women younger than 45 years of age, there was no relationship between breast cancer and OC. The remaining studies that presented age-specific results did not support a hypothesis of differences in the age-specific breast cancer risks related to OC

Results from this and other studies indicate that several risk factors for breast cancer vary by the age at

The suggestion in our analysis that the

use~2,8,11,30-35

breast cancer-OC relationship may vary by age at diagnosis could be explained by hypothesizing a biologic mechanism, currently unknown, similar to the one that produces the age-specific differences in the breast cancer-parity relationship. Such a hypothesis would be consistent with the suggestion that OC may have health effects that are similar to pregnancy." Both high parity and OC use, for example, protect against ovarian and endometrial cancers.','' In addition, analyses of the relationship between parity and cause of death in the United Kingdom revealed that both parity and OC use had similar effects on mortality."

We recognize that the findings of a possible increased risk of breast cancer in patients younger than 35 years of age and a possible decreased risk in older women require confirmation by other studies. Subse- quent investigation into these issues should consider the entire age range of women who may have taken OC. The proposed upper age limit for future studies of the relationship between OC and breast cancer should continue to increase as the women who first used OC in the 1960s attain the ages of high risk for breast cancer. Currently, the upper age limit should be 60-64 years of age. In the 1990s, OC will have been available in the United States for 30 years. Because exposure to OC has been so common, prudent public health thinking de- mands that researchers vigorously pursue the long- term effects of OC use on breast cancer risk. The literature currently indicates that OC use has no effect on the aggregate lifetime risk of developing breast cancer. However, the oldest women studied have been

Table 7. Hypothetical Annual Age-Specific Breast Cancer Incidence Rates: U.S. Women 20 to 54 Years of Age in 1982

Rate Percent oral Incidence per difference

Age Oral contraceptive contraceptive R R t 100,000 women per 100,000 (yr) use history use* (95% CI) per yead women ~~~~

20-34 Never Ever

24.0 Ref 8.5 76.0 1.4 (1.0-2.1) 11.9 3.4

All women 100.0 11.1 35-44 Never 28.6 Ref 74.8

Ever 71.4 1.1 (0.9-1.3) 82.2 7.4 All women 100.0 80.1

45-54 Never 53.1 Ref 177.5

All women 100.0 169.2 Ever 46.9 0.9 (0.8-1.0) 159.8 -17.7

The age-specific estimates of percentages of never and ever use of oral contraceptives for controls younger than age 45 at interview in the Cancer and Steroid Hormone Study data were weighted to estimates from the 1982 National Survey of Family Growth (unpublished data). Estimates for older controls are unweighted. t The age-specific relative risk estimates are from Table 3. The breast cancer incidence data are annual age- and sex-spedfic rates from SEER for 1978-1981." The estimates

of the incidence of breast cancer per 100,000 women per year by oral contraceptive use are calculated using percentage oral contraceptive use, RR, and the age- and sex-specific SEER rates."


younger than age 60 years. Future studies that include older women who used OC will permit a more accurate assessment of the relationship between OC use and lifetime risk.

In the meantime, controversies regarding the im- pact of OC use on breast cancer risk are likely to continue, pending additional biologic and epidemiologic data. To help women and their clinicians make deci- sions about the breast cancer-OC relationship, we estimated the possible effects of OC use on breast cancer incidence rates (Table 7). The percentages of ever use of OC were estimated from the control women interviewed for this study; for women younger than 45 years, the percentages were weighted to estimates from the 1982 National Survey of Family The age- specific relative risks of breast cancer related to OC use were estimated from the current analysis. Lastly, the age-specific rates of breast cancer for all women were obtained from the Surveillance, Epidemiology, and End Results Program for the years 1978-1981.37 We combined these data to compute estimates of the incidence of breast cancer per 100,000 per year women according to age and ever use of OC.23 Our estimates suggest that, before age 45 years when the incidence of breast cancer is low, OC use could be associated with an increase of fewer than 11 cases per 100,000 women, annually. For ages 45-54 years, when the incidence of breast cancer begins to increase substantially, our estimates suggest that OC use could be associated with a decrease of 17.8 cases per 100,000 women. These estimates should be interpreted with caution because they are sensitive to small changes in the magnitude of the relative risk estimates, particularly in the older ages in which breast cancer incidence rates are high. In general, these data are reassuring and provide no reasons for changes in prescribing practices or in the use of OC as related to breast cancer risk.

References

1.

2.

3.

4.

5.

Schlesselman JJ. Cancer of the breast and reproductive tract. Contraception 1989; 4O:l-38. Romieu I, Willette WC, Colditz GA, Stampfer MJ, Rosner B, Hennekens CH, et al. Prospective study of oral contraceptive use and risk of breast cancer in women. ] Natl Cancer Inst 1989;

Vessey MP, McPherson K, Villard-Mackintosh L, Yeates D. Oral contraceptives and breast cancer: latest findings in a large cohort study. Br ] Cancer 1989; 59:613-7. McPherson K, Neil A, Vessey MP, Doll R. Oral contraceptives and breast cancer. Lancet 1983; 2:1414-5. Pike MC, Henderson BE, Krailo MD, Duke A, Roy S. Breast cancer in young women and use of oral contraceptives: possible modifying effects of formulation and age at use. Lancet 1983;

81 : 1313-21.

2:926-30.

6.

7.

8.

9.

10.

11 .

12.

13.

14

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

McPherson K, Vessey MP, Neil A, Doll R, Jones L, Roberts M. Early oral contraceptive use and breast cancer: results of another case-control study. Br ] Cancer 1987; 56:653-60. Stadel BV, Schlesselman JJ, Murray PA. Oral contraceptives and breast cancer. Lancet 1989; 1:1257-8. Ravnihar 8, Zakelj MP, Kosmelj K, Stare J. A case-control study of breast cancer in relation to oral contraceptive use in Slovenia. Neoplasma 1988; 35:109-21. Chilvers C, McPherson K, Peto J, Pike MC, Vessey MP. Oral contraceptive use and breast cancer risk in young women. Lan- cet 1989; 1:973-82. Meirik 0, Farley TMM, Lund E, Adami H-0, Christoffersen T, Bergsjo P. Breast cancer and oral contraceptives: patterns of risk among parous and nulliparous women: further analysis of the Swedish-Norwegian material. Contraception 1989; 39:471-5. The WHO Collaborative Study of Neoplasia and Steroid Con- traceptives, Thomas DB, Noonan EA. Breast cancer and combined oral contraceptives: results from a multinational study. Br

Janerich DT, Hoff MB. Evidence for a crossover in breast cancer risk factors. A m ] Epidemiol 1982; 116:737-42. Ron E, Lubin F, Wax Y. Re: evidence for a crossover in breast cancer risk factors. A m ] Epidemiol 1984; 119:139-41. Layde PM, Webster LA, Baughman AL, Wingo PA, Rubin GL, Ory HW. The independent associations of parity, age at first full term pregnancy, and duration of breastfeeding with the risk of breast cancer. ] Clin Epidemiol 1989; 42:963-73. Bruzzi P, Negri E, La Vecchia C, Decarli A, Palli D, Parazzini F, et al. Short-term increase in risk of breast cancer after full-term pregnancy. BM] 1988; 297:1096-8. Pathak DR, Speizer FE, Willette WC, Rosner B, Lipnick RJ. Par- ity and breast cancer risk possible effect on age at diagnosis. Znt ]Cancer 1986; 37:21-5. Beral V. European organization for cooperation in cancer prevention studies: workshop on optimization of influence of ovarian steroid consumption on cancer risk. Eur ] Obstet Gynecol Reprod Biol 1987; 26:l-6. Beral V. Long term effects of childbearing on health. Epidemiol Community Health 1985; 39:343-6. The Centers for Disease Control and National Institute of Child Health and Human Development Cancer and Steroid Hormone Study. Oral contraceptive use and the risk of breast cancer.

The Centers for Disease Control and National Institute of Child Health and Human Development Cancer and Steroid Hormone Study. Oral contraceptive use and the risk of breast cancer. N Engl ] Med 1986; 315:405-11. Wingo PA, Ory HW, Layde PM, Lee NC. The evaluation of the data collection process for a multicenter, population-based, case-control design. A m ] Epidemiol 1988; 128:206-17. Waksberg J . Sampling methods for random digit dialing. ] A m

Schlesselman JJ. Case control studies: design, conduct, analysis. New York Oxford University Press, 1982. Wynder EL, Schlesselman JJ, Wald N, Lilienfeld A, Stolley PD, Higgins I T , et al. Conference report: weak associations in epidemiology and their interpretation. Prev Med 1982; 11:464-76. Stadel BV, Rubin GL, Webster LA, Schlesselman JJ, Wingo P. Oral contraceptives and breast cancer in young women. Lancet

Schlesselman JJ, Stadel BV, Murray P, Shenghan L. Breast cancer risk in relation to type of estrogen contained in oral contraceptives. Contraception 1987; 36:595-6 13,

Cancer 1990; 61:llO-9.

]AMA 1983; 249~1591-5.

Stat ASOC 1978; 73~40-6.

1985; 2~970-3.

Oral Contraceptives and Breast Cancer/Wingo et al. 1517

27.

28.

29.

30.

31.

32.

33.

Schlesselman JJ, Stadel BV, Murray P, Shenghan L. Breast cancer risk in relation to early use of oral contraceptives: no evidence of a latent effect. IAMA 1988; 259:1828-33. Murray P, Schlesselman JJ, Stadel BV, Shenghan L. Oral contraceptives and breast cancer risk in women with a family history of breast cancer. Ani 1 O b s f e f Gyriecol 1989; 73:977-83. Paul C, Skegg DCG, Spears GFS, Kaldor JM. Oral contraceptives and breast cancer: a national study. BMI 1986; 293:723-6. Vessey M, Baron J, Doll R, McPherson K, Yeates D. Oral contraceptives and breast cancer: final report of an epidemiologic study. Br Cancer 1983; 47:455-62. Rosenberg L, Miller DR, Kaufman DW. Breast cancer and oral contraceptiye use. A m I Epidemiol 1984; 119:167-76. La Vecchia C, Decarli A, Fasoli M, Franceschi S, Gentile A, Negri E, et al. Oral contraceptives and cancers of the breast and of the female genital tract: interim results from a case-control study. Br J Cancer 1986; 54:311-7. Stanford JL, Brinton LA, Hoover RN. Oral contraceptives and

breast cancer: results from an expanded case-control study. Br 1 Cancer 1989; 60:375-81.

34. Lee NC, Rosero-Bixby L, Oberle MW, Grimaldo C, Whatley AS, Rovira EZ. A case-control study of breast cancer and hormonal contraception in Costa Rica. 1 N a f l Cancer lnst 1987; 79:1247- 54.

35. Paffenbarger RS, Fasal E, Simmons ME, Kampert JB. Cancer risk as related use of oral contraceptives during fertile years. Cancer

36. Mosher WD, Bachrach CA. Contraceptive use, United States: 1982 data from the National Survey of Family Growth. Hyatts- ville (MD): Vital Health Statistics National Center for Health Statistics; 1986; Publication no. (PHS) 86-1988.

37. Horm JW, Asire AJ, Young JL, Pollack ES. SEER program: cancer incidence and mortality in the United States, 1973-1981. Be- thesda (MD): National Institutes of Health, National Cancer In- stitute; 1984; Publication no. (NIH) 85-1837.

1977; 393887-91.

age-specific differences in the relationship between oral contraceptive use and breast cancer

Documents