Fax: (206) 667-5948
Article first published online: 11 AUG 2003
Copyright © 2003 American Cancer Society
Volume 98, Issue 6, pages 1131–1140, 15 September 2003
How to Cite
Mueller, B. A., Simon, M. S., Deapen, D., Kamineni, A., Malone, K. E. and Daling, J. R. (2003), Childbearing and survival after breast carcinoma in young women. Cancer, 98: 1131–1140. doi: 10.1002/cncr.11634
Cancer data were supported by the Cancer Surveillance System of the Fred Hutchinson Cancer Research Center (funded by Contract No. N01-CN-05230 from the Surveillance, Epidemiology, and End Results [SEER] Program of the National Cancer Institute [NCI] with additional funding from the Fred Hutchinson Cancer Research Center) in the Seattle region. In Los Angeles, cancer incidence data were collected under a subcontract with the Public Health Institute. The subcontract is supported by the California Department of Health Services within the statewide cancer reporting program, mandated by Health and Safety Code Section 103875 and 103885. The University of Southern California Los Angeles County Cancer Surveillance Program is funded by Contract No. NO1-PC-67010 from the SEER Program of the NCI. Cancer incidence data for Detroit were collected by the Metropolitan Detroit Cancer Surveillance System of Wayne State University/Karmanos Cancer Institute and funded by Contract No. NO1-CN-65064 from the SEER Program of the NCI. Vital statistics data were provided by the Washington State Department of Health Center for Health Statistics; the California Department of Health Services, Center for Health Statistics, Vital Statistics Section; and the Vital and Health Record Section, Department of Community Health, Community Public Health Agency of the State of Michigan.
Ideas and opinions expressed herein are those of the authors, and no endorsement by the State of California, Department of Health Services or the Public Health Institute is intended or should be inferred.
- Issue published online: 5 SEP 2003
- Article first published online: 11 AUG 2003
- Manuscript Accepted: 16 APR 2003
- Manuscript Revised: 11 APR 2003
- Manuscript Received: 27 FEB 2003
- U.S. Army Medical Research and Materiel Command, U.S. Department of Defense. Grant Number: DAMD17-94-J-4262
- breast carcinoma;
Many young patients with breast carcinoma have not started, or completed, their desired families. How childbearing after a diagnosis of breast carcinoma affects survival is of importance to these women and their families. The authors measured relative mortality among young patients with breast carcinoma with and without births occurring after diagnosis.
The authors conducted a cohort study using data from three population-based cancer registries in the U.S. (Seattle, Detroit, and Los Angeles), linked to birth certificate data in each state. Four hundred thirty-eight women younger than 45 years of age with primary invasive breast carcinoma were identified as having births after diagnosis. In addition, 2775 comparison women, matched on the basis of age at the time of diagnosis, race/ethnicity, diagnosis year, disease stage, and presence of previous nonbreast primary tumors, were identified among those with breast carcinoma without births after diagnosis. Relative mortality was assessed using multivariable statistical methods.
After adjustment for stage of disease, age at diagnosis, study region, diagnosis year, and race/ethnicity, women with births occurring 10 months or more after diagnosis had a significantly decreased risk of dying (relative Risk [RR] = 0.54, 95% confidence interval [CI], 0.41–0.71) compared to women without subsequent births. Women pregnant at the time of diagnosis had a mortality rate similar to those who did not give birth (RR = 1.10, 95% CI, 0.80–1.60).
The results of the current study, in light of growing evidence from other studies using various methods, may provide some reassurance to young women with breast carcinoma that subsequent childbearing is unlikely to increase their risk of mortality. Cancer 2003;98:1131–40. © 2003 American Cancer Society.
The incidence of breast carcinoma among women younger than 45 years of age is increasing1, 2 and young women appear to have relatively poor survival.3, 4 The prognosis may be particularly poor for young women who are diagnosed within a relatively short period after a birth.5–7 Although recent epidemiologic studies generally suggest no increased risk of mortality associated with childbearing after a breast carcinoma diagnosis,8–10 evaluation of relative mortality among women with different characteristics such as age at diagnosis, first course of therapy, race/ethnicity, lymph node status, and tumor size may provide a more complete understanding of the relation among groups of women with different disease or other characteristics. As many young breast carcinoma survivors may not yet have started, or completed, their desired families, the decision whether to become pregnant after treatment is of great importance to these women, their families, and physicians. Using data from three large population-based cancer registries, we evaluated the effects of childbearing after diagnosis on survival among young women with breast carcinoma overall and among selected subgroups.
MATERIALS AND METHODS
Human subject protection committee approval was received by the appropriate institutions and from each state's Department of Health before the conduct of the current study. Data from three population-based cancer registries participating in the Surveillance, Epidemiology and End Results (SEER) program of the National Cancer Institute were used to identify study subjects. These registries, located in Seattle, Detroit, and Los Angeles, perform active surveillance and lifelong follow-up of incident cancer cases occurring in each region. Registry activities in Seattle cover more than 3.3 million residents living in the 13 counties of western Washington State. The Los Angeles County Registry covers 9.2 million residents in California and registry activities in Detroit provide surveillance for a three-county area surrounding the city, with a population base of nearly 4.1 million residents. Routine monitoring of registry data for accuracy and completeness of coverage is conducted by the SEER program. Data available for each subject included demographic information (date of birth, age at diagnosis, race/ethnicity), tumor characteristics (date of diagnosis, primary site, histologic type, SEER summary stage [coded as local, regional, distant, and unknown in the SEER Coding Manual, 1997]), and initial course of treatment. The registries routinely obtain vital status information for cases in their areas by linkage with state death certificates and drivers' license data tapes and via contact with patients' physicians of record and/or hospital tumor registrars.
All women younger than 45 years of age at the time of diagnosis with primary invasive breast carcinoma, diagnosed during the years 1980–1993 (in Detroit and Los Angeles) and 1980–1994 (in Seattle) were identified. This included 3925 women in Seattle, 4496 women in Detroit, and 6962 women in Los Angeles. Identification of women with births occurring after their breast carcinoma diagnoses was conducted by linkage of registry and vital records data. Birth certificate records for the years 1982–1993 and 1980–1994 were obtained from the Departments of Health in California and Washington states, respectively. (Before 1982, California birth certificates did not contain the parents' names. Therefore, records linkage was not possible for these years. Because of this, 222 women diagnosed during 1980–1981 in the Los Angeles registry were excluded from further analyses.) In California and Washington, the linkage of registry to birth certificate data was conducted by registry-affiliated study staff. Michigan registry data were linked to birth data for 1980–1994 by the Michigan State Department of Health. For each eligible woman with breast carcinoma, the diagnosis year and subsequent years were screened for her live births. If a woman was known to have died based on current registry data, the birth certificates were screened from the year of diagnosis to the date of death. In this manner, women with live births after their diagnoses of breast carcinoma were identified in each region.
For each breast carcinoma case with a subsequent live birth, we sought to identify 12 comparison subjects from among women without births after diagnosis in each region, matched based on age at diagnosis, race/ethnicity, diagnosis year (1980–1984, 1985–1989, 1990–1994), disease stage, and the presence of a previous nonbreast primary tumor. All potential comparison candidates with a specific set of criteria matching a profile for a set of women with births were identified. A comparison subject was selected randomly from this group for each “exposed” woman matching these characteristics. Random selection rotated among all women within each set of characteristics and continued until the pool of comparison candidates was depleted, or women with births had 12 matched comparison subjects, whichever occurred first. To be eligible, comparison women must have been known not to have died by the date their matching subjects gave birth. For 39 women for whom no potential comparison candidate was found using this procedure, a comparison subject was identified on a case-by-case basis by reviewing each woman's demographic and disease characteristics and relaxing a single matching criterion (usually race/ethnicity).
Women pregnant at diagnosis may have relatively poor survival compared with women who are not pregnant. For this reason, subjects were categorized on the basis of whether they were pregnant at the time of their diagnosis with breast carcinoma. Women with live births occurring 10 or more months after diagnosis were categorized as having pregnancies conceived after their diagnosis with breast carcinoma. Data for these women and their comparison subjects were analyzed separately from those for women with births occurring within 10 months of their diagnoses.
Cox proportional hazards modeling11 was used to estimate the relative risk (RR) of dying for women who had children after diagnosis compared with women who did not. Follow-up was accrued for all subjects from the date of diagnosis, with left truncation of subjects' survival times occurring at the date of a woman's first live birth after diagnosis, and at a similar date for her comparison subjects in the calculation of RR estimates. Follow-up information from each registry through March 1999 was obtained for each subject. The date of death or last follow-up by the registry was used to count follow-up time in months since diagnosis. The median follow-up times for women with and without births less than 10 months after diagnosis were 55 months (range, 2–199 months) and 62 months (range, 6–107 months), respectively. The median follow-up times for women with and without births 10 months or more after diagnosis were 105 months (range, 16–203 months) and 99 months (range, 12–214 months), respectively. Risk-set stratification was used to account for the matched set number and adjustment for potentially confounding variables was conducted by their inclusion in the regression model. The potentially confounding variables were age at diagnosis (continuous), year of diagnosis (1980–1984, 1985–1989, 1990–1994), stage at diagnosis (local, regional, distant, unknown), presence of a previous nonbreast primary tumor (no other primary tumor/other nonbreast primary tumor present before breast carcinoma diagnosis/other nonbreast tumor diagnosed in the interval between diagnosis and birth/other nonbreast tumor diagnosed after birth), and race/ethnicity. Lymph node status and first course of therapy were also evaluated for their possible effects.
Of the 15,161 women younger than 45 years of age at diagnosis with primary invasive breast carcinoma identified during the study period, 438 (3%) subsequently gave birth to a live born infant. The proportion of women with births did not vary by region. Of those with births in each region, 73–79% gave birth at least 10 months after their diagnosis with breast carcinoma, indicating pregnancies conceived after the diagnosis date. The mean age at diagnosis of patients with breast carcinoma in each region who subsequently gave birth ranged from 31.1 years (± a standard error [SE] of 4.7) in Detroit to 32.3 years (± SE 4.0) in Los Angeles (data not shown). The majority of women (56–63%) had local stage disease and few women (ranging from 2% in Seattle to 5% in Los Angeles) had distant metastases. About 16% of all subjects were African-American (4% in Seattle, 33% in Detroit, and 13% in Los Angeles). Five percent were of Asian ethnicity or countries of origin (Japanese, Chinese, Filipino, Korean, Vietnamese).
A mean of seven comparison subjects were identified for each woman with a birth occurring 10 or more months after diagnosis and a mean of six comparison subjects were identified for women with births occurring less than 10 months after diagnosis. About one-third of women in each group had 12 comparison subjects. Although the majority of comparison subjects (90% and 82% of those for women with births 10 or more months and less than 10 months after diagnosis, respectively) were matched within 1 year of age at diagnosis, the varying number of comparison subjects for each exposed woman gives the appearance of disparate ages at diagnosis (Table 1). Women with live births occurring 10 or more months after diagnosis were somewhat younger, in general, than women with births occurring less than 10 months after diagnosis and they had less severe disease. Distant metastases were present in less than 1% of women with births occurring 10 months or more after diagnosis compared with 12% of women with births occurring less than 10 months after diagnosis. Only two women with births had other, nonbreast primary tumors before their diagnosis with breast carcinoma. One had a birth occurring less than 10 months after diagnosis and the other had a birth occurring 10 or more months after diagnosis. Five women with births occurring 10 months or more after diagnosis had other, nonbreast primary tumors diagnosed after their diagnoses with breast carcinoma, but before giving birth.
|Characteristics||Live birth occurring < 10 mos after diagnosis||Live birth occurring ≥ 10 mos after diagnosis|
|Live birth (n = 109)||No live birth (n = 687)||Live birth (n = 329)||No live birth (n = 2088)|
|Age at diagnosis (yr.)|
|Tumor size (cm)|
|Lymph node status|
Risk of Mortality Associated with Childbirth 10 Months or More after Diagnosis
After adjustment for stage of disease and age at diagnosis, study region, diagnosis year, and race/ethnicity, women with births occurring 10 months or more after a breast carcinoma diagnosis had a decreased risk of dying (RR = 0.54; 95% confidence interval [95% CI], 0.41–0.71), relative to women without subsequent births (Table 2). (Unless otherwise indicated, all estimates are adjusted for age at diagnosis, diagnosis year, race/ethnicity, study region, and stage of disease.) A decreased risk was observed both among women with local disease at diagnosis (RR = 0.59; 95% CI, 0.40–0.89) and among women with regional disease at diagnosis (RR = 0.54; 95% CI, 0.37–0.78). The risk of mortality associated with a live birth occurring 10 months or more after diagnosis for women younger than 35 years of age at diagnosis was 0.55 (95% CI, 0.40–0.74), but did not differ significantly from unity among women 35 years or older at diagnosis (RR = 0.83; 95% CI, 0.47–1.5). Among white women, the risk of dying associated with having a live birth occurring 10 months or more after diagnosis was 0.48 (95% CI, 0.34–0.67). Our ability to evaluate the relation within other race/ethnicity groups was limited by small numbers of subjects. However, among African-American women, the risk was 0.77 (95 % CI, 0.45–1.3). Among Asian women, all three deaths occurred among women without births.
|Characteristics||Women at risk||Women-years at riska||Total deaths (%)||Relative risk of mortality (95% CI)|
|All births (≥ 10 mos after diagnosis)|
|Women without births||2002||10,485||532 (26.6)||1.0 (referent)|
|Women with births||328||1851||62 (18.9)||0.54 (0.41–0.71)b|
|Local disease at diagnosis|
|Women without births||1278||7160||231 (18.1)||1.0 (referent)|
|Women with births||208||1190||28 (13.5)||0.59 (0.40–0.89)b|
|Regional disease at diagnosis|
|Women without births||697||3214||288 (41.3)||1.0 (referent)|
|Women with births||106||602||31 (29.3)||0.54 (0.37–0.78)b|
|Age < 35 yrs at diagnosis|
|Women without births||1239||5990||368 (29.7)||1.0 (referent)|
|Women with births||256||1441||49 (19.1)||0.55 (0.40–0.74)b|
|Age ≥ 35 yrs at diagnosis|
|Women without births||763||4495||164 (21.5)||1.0 (referent)|
|Women with births||72||410||13 (18.1)||0.83 (0.47–1.5)b|
|Women without births||1689||8917||440 (26.1)||1.0 (referent)|
|Women with births||242||1381||39 (16.1)||0.48 (0.34–0.67)b|
|Women without births||252||1231||78 (31.0)||1.0 (referent)|
|Women with births||55||302||18 (32.7)||0.77 (0.45–1.3)b|
|Women without births||29||124||3 (10.3)||1.0 (referent)|
|Women with births||16||84||0 (0.0)||0 (0–3.6)c|
|Lymph node positive|
|Women without births||550||2404||220 (40.0)||1.0 (referent)|
|Women with births||77||368||25 (32.5)||0.65 (0.42–0.99)b|
|Lymph node negative|
|Women without births||901||4671||152 (16.9)||1.0 (referent)|
|Women with births||136||748||16 (11.8)||0.56 (0.33–0.95)b|
|Tumor size < 2 cm|
|Women without births||481||2407||86 (17.9)||1.0 (referent)|
|Women with births||76||414||11 (14.5)||0.70 (0.36–1.4)b|
|Tumor size ≥ 2 cm|
|Women without births||846||3889||267 (31.6)||1.0 (referent)|
|Women with births||127||625||28 (22.1)||0.52 (0.35–0.77)b|
|Women without births||652||2704||203 (31.1)||1.0 (referent)|
|Women with births||85||435||21 (24.7)||0.54 (0.34–0.86)b|
|Women without births||829||4396||197 (23.8)||1.0 (referent)|
|Women with births||154||832||26 (16.9)||0.56 (0.37–0.86)b|
|Women without births||569||2750||143 (25.1)||1.0 (referent)|
|Women with births||93||490||14 (15.1)||0.44 (0.25–0.77)b|
|Women without births||947||4531||266 (28.1)||1.0 (referent)|
|Women with births||152||802||33 (21.7)||0.59 (0.41–0.86)b|
|Women without births||220||941||83 (37.7)||1.0 (referent)|
|Women with births||24||133||6 (25.0)||0.36 (0.15–0.88)b|
|No other treatment|
|Women without births||1335||6530||332 (24.9)||1.0 (referent)|
|Women with births||227||1188||42 (18.5)||0.57 (0.41–0.79)b|
A similar decrease in the RR of mortality occurred among women who were lymph node positive (RR = 0.65; 95% CI, 0.42–0.99) and lymph node negative (RR = 0.56; 95% CI, 0.33–0.95). A somewhat similar pattern was observed for the risk estimates for women with tumor size less than 2 cm (RR = 0.70; 95% CI, 0.36–1.4) and for women with tumor size 2 cm or larger (RR = 0.52; 95% CI, 0.35–0.77). The RRs of mortality also decreased and ranged from 0.44–0.59 among women who had and had not received chemotherapy or radiotherapy as part of their first course of treatments. Risk estimates were also decreased among women who had and had not received other treatment modalities such as hormonal or immunotherapy.
Risk of Mortality Associated with Childbirth Less Than 10 Months after Diagnosis
Similar decreases in the risk of mortality were not associated with having a live birth less than 10 months after a diagnosis of breast carcinoma. Forty-five percent of women with births and 34% of women without births had died by the end of the follow-up period, a difference that was not statistically significant (RR = 1.1; 95% CI, 0.8–1.6) (Table 3). No increased risk of mortality was observed for women with local disease at diagnosis (RR = 0.79; 95% CI, 0.37–1.7). Among women with regional disease at diagnosis, there was a nearly twofold increased risk associated with a live birth less than 10 months after diagnosis (RR = 1.7; 95% CI, 1.1–2.6). Women 35 years or older at diagnosis had an increased risk of mortality associated with a live birth less than 10 months after diagnosis (RR = 1.8; 95% CI, 1.1–2.9), an increase not observed among women younger than 35 years of age (RR = 0.99; 95% CI, 0.63–1.6). Significant increases in risks were not observed when the analysis was restricted to white women (RR = 1.1; 95% CI, 0.77–1.6) or to African-American women (RR = 1.6; 95% CI, 0.55–4.6). Unlike the other two groups, a slightly greater proportion of Asian women without births (25%) than with births (17%) had died. An increased risk was observed for women who were lymph node positive (RR = 1.8; 95% CI, 1.1–2.8), but not for women who were lymph node negative (RR = 1.1; 95% CI, 0.44–2.6). Among women with tumor size less than 2 cm, the RR was 0.83 (95% CI, 0.21–3.2) and among women with a tumor size 2 cm or larger, the RR was 1.4 (95% CI, 0.9–2.3). Among women who had received chemotherapy or radiation as part of their first course of treatment, the risks of mortality associated with a birth less than 10 months from diagnosis were elevated (RR = 1.7; 95% CI, 1.1–2.7 and RR = 3.7; 95% CI, 2.0–7.0, respectively). Women who received other treatments (hormonal or immunotherapy) were at increased risk as well (RR = 3.7; 95% CI, 1.6–8.9). Significantly increased risks were not observed among women who had not received these treatment modalities.
|Characteristics||No. of women at risk||Woman-years at riska||Total deaths (%)||Relative risk of mortality (95% CI)|
|All births (< 10 mos after diagnosis)|
|Women without births||677||3755||230 (34.0)||1.0 (referent)|
|Women with births||109||567||49 (45.0)||1.1 (0.80–1.6)b|
|Local disease at diagnosis|
|Women without births||331||2052||59 (17.8)||1.0 (referent)|
|Women with births||49||339||9 (18.4)||0.79 (0.37–1.7)b|
|Regional disease at diagnosis|
|Women without births||312||1614||142 (45.5)||1.0 (referent)|
|Women with births||44||171||27 (61.4)||1.7 (1.1–2.6)b|
|Age < 35 yrs at diagnosis|
|Women without births||253||1386||103 (40.7)||1.0 (referent)|
|Women with births||65||372||28 (43.1)||0.99 (0.63–1.6)b|
|Age ≥ 35 yrs at diagnosis|
|Women without births||424||2369||127 (30.0)||1.0 (referent)|
|Women with births||44||195||21 (47.7)||1.8 (1.1–2.9)b|
|Women without births||613||3368||210 (34.3)||1.0 (referent)|
|Women with births||85||427||38 (44.7)||1.1 (0.77–1.6)b|
|Women without births||54||328||15 (27.8)||1.0 (referent)|
|Women with births||15||82||8 (53.3)||1.6 (0.55–4.6)b|
|Women without births||4||23||1 (25.0)||1.0 (referent)|
|Women with births||6||45||1 (16.7)||0.52 (0.03–8.4)c|
|Lymph node positive|
|Women without births||232||1100||94 (40.5)||1.0 (referent)|
|Women with births||43||137||27 (62.8)||1.8 (1.1–2.8)b|
|Lymph node negative|
|Women without births||281||1643||46 (16.4)||1.0 (referent)|
|Women with births||37||244||6 (16.2)||1.1 (0.44–2.6)b|
|Tumor size < 2 cm|
|Women without births||187||1033||30 (16.0)||1.0 (referent)|
|Women with births||12||62||3 (25.0)||0.83 (0.21–3.2)b|
|Tumor size ≥ 2 cm|
|Women without births||301||1486||108 (35.9)||1.0 (referent)|
|Women with births||60||284||23 (38.3)||1.4 (0.90–2.3)b|
|Women without births||310||1437||106 (34.2)||1.0 (referent)|
|Women with births||50||189||28 (56.0)||1.7 (1.1–2.7)b|
|Women without births||226||1381||49 (21.7)||1.0 (referent)|
|Women with births||30||172||8 (26.7)||1.2 (0.53–2.5)b|
|Women without births||218||1138||61 (28.0)||1.0 (referent)|
|Women with births||22||77||15 (68.2)||3.7 (2.0–7.0)b|
|Women without births||329||1770||99 (30.1)||1.0 (referent)|
|Women with births||59||306||20 (33.9)||1.1 (0.61–1.9)b|
|Women without births||111||514||46 (41.4)||1.0 (referent)|
|Women with births||15||58||10 (66.8)||3.7 (1.6–8.9)b|
|No other treatment|
|Women without births||447||2424||119 (26.6)||1.0 (referent)|
|Women with births||73||358||26 (35.6)||1.2 (0.75–1.9)b|
When evaluated within elapsed time intervals in years since diagnosis, the risk of mortality increased significantly only for women with live births occurring 3 months or less after diagnosis (RR = 1.7; 95% CI, 1.2–2.6; data not shown). Relative to women without births, the risk estimates for women with births occurring 4–6 months, 7–9 months, and 10 to less than 12 months after diagnosis were 1.0 (95% CI, 0.55–1.9), 0.38 (95% CI, 0.12–1.2), and 1.0 (95% CI, 0.23–4.8), respectively. For women with births occurring 2–5 years after their diagnosis, the risks decreased significantly (RR = 0.49; 95% CI, 0.27–0.86 for 2 to less than 3 years; RR = 0.30; 95% CI, 0.12–0.71 for 3 to less than 4 years; RR = 0.19; 95% CI, 0.05–0.81 for 4 to less than 5 years).
Childbearing options and outcomes may not be an initial concern for many young women with breast carcinoma facing immediate issues of therapy and survival. However, for many patients with breast carcinoma, childbearing eventually is relevant. Although our results suggest that, at least for women delivering more than 3 months after diagnosis, there is no associated increased risk of mortality, the lack of complete pregnancy history after a diagnosis of breast carcinoma, such as would be obtained by interview, is a limitation of the current study. We were unable to evaluate survival related to pregnancy in general or to evaluate other pregnancy outcomes that did not result in a live birth. If, due to an inability to identify birth certificates (e.g., for births occurring out of the state in which their cancer was diagnosed or for women who may have changed their names), women with births occurring after diagnosis were included erroneously in the comparison group, our results would be biased toward the null. In an earlier, population-based linkage of Finnish cancer registry and birth certificate data, about 4% of women younger than 40 years of age at diagnosis had a term delivery 10 months or more after diagnosis,8 a level similar to that observed in the current study. More recent data from a similar linkage study with Danish registry data indicated that 3% of women 45 years or younger became pregnant after diagnosis.9 A recent study based on self-reported pregnancy history indicated that 36 of 520 (6.9%) women younger than 40 years of age with breast carcinoma in the Seattle region had a subsequent live birth,10 a greater proportion than the 3% observed in the current study using birth records alone. As the earlier study included only premenopausal women with local or regional disease, this may overestimate the proportion of all cancer survivors with live births. Nevertheless, it is likely that our ascertainment of subsequent births is, to some extent, incomplete. If having a live birth is associated with either an increased or decreased risk of mortality, then misclassification of women incorrectly identified as not having births among the comparison group would bias our results toward the null. We were also unable to evaluate the potential effects of other factors possibly associated with risk, and which may also affect survival, including reproductive history prior to diagnosis, family history, and body mass index.12–18
These data suggest that, among young women with breast carcinoma, with the exception of women delivering infants within 3 months of diagnosis, having a subsequent live birth is not associated with an increased risk of dying. These results are generally similar to previous studies.8–10, 19–21 In our data, except for a subgroup with births 3 months or less after diagnosis, women with a live birth less than 10 months after their diagnosis overall did not have worse survival than women without subsequent births. Increased risk estimates were noted, however, for women with regional disease, women 35 years of age or older, women who were lymph node positive, and for women who received chemotherapy, radiation, or other treatments indicated in the registry records as their first course of therapy. For women who were pregnant at the time of diagnosis, decisions regarding whether to become pregnant were only influenced by health status to the extent that reproductive choices in any woman's life are similarly affected. Because of this, results related to live births occurring less than 10 months after diagnosis can be considered unbiased with respect to a cancer-related, “healthy mother” effect, or the likelihood that disease-free, relatively healthier women are more likely to become pregnant than women with adverse disease sequelae or poor prognosis. Although the increased risks of mortality observed for women who had undergone chemotherapy, radiation, or other treatments for their first course of therapy, or among women with regional disease, might be due to an adverse effect of pregnancy on women with more advanced disease, the increased risk may also be related to the use of less aggressive treatments or to a delay in initiation of therapy among pregnant women. Surgery, radiation, and chemotherapy are used in pregnant as well as nonpregnant patients.22 However, some modifications of standard therapies or delay in their initiation may occur in pregnant patients. Because the risk of spontaneous abortion due to mastectomy is quite low, surgery is delayed rarely.23 Recommendations regarding the use of chemotherapy or radiation to treat breast carcinoma in pregnant patients vary in relation to the trimester during which the diagnosis is made, as well as the stage of disease. Pregnant patients with local disease who are in their first trimester may be treated similarly to nonpregnant patients.24 Treatment decisions for later stage disease, or for early or local disease in women who desire adjuvant therapies in addition to surgery, become more complex. As chemotherapeutic agents administered during the first trimester may be associated with the occurrence of malformations in the offspring, recommendations for their use range from delaying any use in lymph node-positive patients until after delivery23 to delaying chemotherapy only until after the first trimester has ended.22 External beam radiation is also a cause of congenital defects and, therefore, is generally delayed until at least 20 weeks of gestation25 or delivery.22 As a delay in adjuvant therapy is associated with relatively poor survival, it is possible that in our data, the increased risk of dying associated with childbearing less than 10 months after diagnosis among women who are lymph node positive, or among women who received chemotherapy or radiation, may be due to a pregnancy-related delay in treatment for women with more advanced disease.
As measured with these data, the risk of dying associated with giving birth at 10 months or more after diagnosis with breast carcinoma was significantly decreased. With few exceptions, this risk decreased significantly in all subgroups explored and for no groups were increased risks observed. Women age 35 years or older at diagnosis had neither an increased nor a decreased risk, unlike younger women, who had a nearly 50% reduction in risk. Unlike the results we observed for women who were pregnant at diagnosis, significant decreases were observed regardless of whether a woman's disease was local or regional or whether she had received radiation or chemotherapy as her first course of therapy. It is possible that the first course of treatment for these women was not inhibited or delayed as a result of pregnancy at the time of diagnosis, resulting in earlier, more aggressive therapy that may have improved survival.
Although African-American and Asian women also had decreased risks for mortality associated with births occurring 10 months or more after diagnosis, the 95% CIs in these subanalyses included one. Among Asian women, a greater proportion of women with births survived relative to women without births, whether or not they were pregnant at diagnosis. Although African-American women did not have significantly increased or decreased risks of mortality associated with births occurring after diagnosis, the difference in mortality among women pregnant at diagnosis is more marked (53% of women with births died vs. 28% of women without births) than for whites. Similarly, the significantly decreased risk of mortality associated with births occurring 10 or more months after diagnosis that was observed for white women was not observed among African-American women. Although this could be due to a lack of statistical power in the analyses restricted to African-American women, the trend of relatively less favorable cancer outcomes among African-American women was reported in previous studies26–28 and may also partly reflect poorer survival in this group, possibly due to access to care or delay in treatment.
The finding in the current study of a decreased risk of mortality associated with a birth occurring 10 months or more after diagnosis is similar to findings of record linkage studies conducted in Finland8 and Denmark.9 Although one could speculate that this may be the result of a hormonal “treatment” effect of pregnancy after diagnosis on breast carcinoma recurrence and mortality, it has been postulated that the apparent protective effect may be due to a healthy mother bias.8 A more recent study using data for 94 women with early-stage breast carcinoma with subsequent pregnancies (as noted in physician records at institutions associated with the International Breast Cancer Study Group) also reported a significantly decreased risk of mortality (RR = 0.44; 95%CI, 0.21–0.96) relative to a comparison group of nonpregnant patients with breast carcinoma with similar characteristics.29 In a related study, the risk of recurrence among patients with breast carcinoma with a subsequent pregnancy, relative to patients without a pregnancy, was of similar magnitude (relative hazard = 0.48; P = 0.14).30 As in the earlier Finnish study, our lack of specific information concerning a woman's health status at the time her pregnancy began is a limitation, and our results may be similarly affected. The extent to which this bias operated in the analysis is unknown. In a previous analysis conducted with a series of women for whom self-reported information on disease recurrence was available, the risk of mortality associated with a subsequent live birth, 1.1 (95% CI, .04–3.7), was neither increased nor decreased.10 Although we lack similar health information for the women in our study cohorts, we attempted to better understand the possible effects of a healthy mother bias within our data by restricting analysis to only women known not to have died 1 year or less after delivery. The resultant risk of mortality associated with a subsequent live birth was 0.70 (95% CI, 0.52–0.94, adjusted for stage, age at diagnosis, study region, diagnosis year category, and race). Lacking further health information for the women in our study cohorts, it is not possible to assess health status at delivery to minimize this potential bias. For this reason, the results related to women with births 10 months or more after diagnosis should not be taken as indicative of a true pregnancy- or birth-associated protective affect on survival. However, the current study results, particularly in light of a growing body of evidence from other studies using various methods, may provide some reassurance to young women with breast carcinoma that subsequent childbearing is unlikely to increase their risk of mortality.
- 11Statistical methods in cancer research. Volume 2. The design and analysis of cohort studies. IARC Scientific Pub. No. 82. Lyon, France: IARC, 1987., .
- 19How subsequent pregnancy affects outcome in women with a prior breast cancer. Oncology. 1991; 5: 23–35..
- 21Pregnancy after breast-conserving surgery and radiation therapy for breast cancer. J Natl Cancer Inst Monogr. 1994; 16: 131–137., , .
- 23Carcinoma of the breast in pregnancy and lactation. In: BlandKI, CopelandEM, editors. The breast: comprehensive management of benign and malignant diseases. Philadelphia: WB Saunders, 1991: 1034–1040..
- 24Breast carcinoma and pregnancy. In: DoneganWL, SprattJS, editors. Cancer of the breast, 4th edition. Philadelphia: WB Saunders, 1995: 732–741..
- 25Special situations in the treatment of breast cancer. In: LippmanME, LichterAS, DanforthDN, editors. Diagnosis and management of breast cancer. Philadelphia: WB Saunders, 1988: 407–438., .