Association of the Timing of Pregnancy With Survival in Women With Breast Cancer

Key Points

Question  Does pregnancy increase the risk of death in women with breast cancer?

Findings  In this cohort study of 7553 women with breast cancer, the 5-year overall survival was 87.5% for women with no pregnancy, 82.1% for women with pregnancy-associated breast cancer, and 96.7% for women who had pregnancy 6 months or more after diagnosis of breast cancer. The difference in survival between pregnancy groups was not statistically significant.

Meaning  Pregnancy around the time of, or after, diagnosis of breast cancer, does not worsen survival.

Importance  Increasing numbers of women experience pregnancy around the time of, or after, a diagnosis of breast cancer. Understanding the effect of pregnancy on survival in women with breast cancer will help in the counseling and treatment of these women.

Objective  To compare the overall survival of women diagnosed with breast cancer during pregnancy or in the postpartum period with that of women who had breast cancer but did not become pregnant.

Design, Setting, and Participants  This population-based, retrospective cohort study linked health administrative databases in Ontario, Canada, comprising 7553 women aged 20 to 45 years at the time of diagnosis with invasive breast cancer, from January 1, 2003, to December 31, 2014.

Exposures  Any pregnancy in the period from 5 years before, until 5 years after, the index date of the diagnosis of breast cancer. Women were classified into the following 4 exposure groups: no pregnancy (the referent), pregnancy before breast cancer, pregnancy-associated breast cancer, and pregnancy following breast cancer.

Main Outcomes and Measures  Five-year actuarial survival rates for all exposure groups, age-adjusted and multivariable hazard ratios [HRs] of pregnancy for overall survival for all exposure groups, and time-dependent hazard ratios for women with pregnancy following breast cancer.

Results  Among the 7553 women in the study (mean age at diagnosis, 39.1 years; median, 40 years; range, 20-44 years) the 5-year actuarial survival rate was 87.5% (95% CI, 86.5%-88.4%) for women with no pregnancy, 85.3% (95% CI, 82.8%-87.8%) for women with pregnancy before breast cancer (age-adjusted hazard ratio, 1.03; 95% CI, 0.85-1.27; P = .73), and 82.1% (95% CI, 78.3%-85.9%) for women with pregnancy-associated breast cancer (age-adjusted hazard ratio, 1.18; 95% CI, 0.91-1.53; P = .20). The 5-year actuarial survival rate was 96.7% (95% CI, 94.1%-99.3%) for women who had pregnancy 6 months or more after diagnosis of breast cancer, vs 87.5% (95% CI, 86.5%-88.4%) for women with no pregnancy) (age-adjusted HR, 0.22; 95% CI, 0.10-0.49; P < .001).

Conclusions and Relevance  Pregnancy did not adversely affect survival in women with breast cancer. For breast cancer survivors who wish to conceive, the risk of death is lowest if pregnancy occurs 6 months or more after diagnosis.

In the United States, approximately 247 000 women received a diagnosis of breast cancer in 2016, including 26 392 women younger than 45 years (10.7% of all cases).¹ In recent decades, maternal age at first childbirth has risen by approximately 5 years in North America, and the fertility rate among women older than 35 years has increased.² Given the trend of delaying motherhood, the proportion of women who develop breast cancer while pregnant (or wish to become pregnant after undergoing breast cancer treatment) is increasing.

Breast cancers diagnosed during pregnancy (or up to 1 year post partum) have unfavorable clinicopathologic features, such as high tumor grade and low expression of estrogen receptor (ER) and progesterone receptor (PR), compared with breast cancers diagnosed in nonpregnant women of the same age.^3-6 However, whether or not pregnancy compromises survival remains a matter of debate.^7-16 The treatment of these women is challenging because both the health of the mother and of the fetus need to be considered. Chemotherapy is not advisable during the first trimester of pregnancy; radiotherapy and hormonal therapies are not recommended at any time during pregnancy.¹⁷

Pregnancy that occurs more than 1 year after a diagnosis of breast cancer does not appear to affect survival.^18-22 It is not clear how long a woman should wait after undergoing breast cancer treatment before attempting a pregnancy.^23-25 Most women are advised to postpone pregnancy for at least 2 years after treatment. The main reason for this recommendation is a concern that gestational hormones, in particular, estrogen, might stimulate dormant micrometastases, and thereby worsen survival.²⁶ This study was conducted to compare the overall survival of women diagnosed with breast cancer during pregnancy or in the postpartum period, with that of women who had breast cancer but did not become pregnant.

We performed a population-based, retrospective cohort study in Ontario, Canada, linking a cancer database, a natality database, and a mortality database, located at the Institute for Clinical Evaluative Sciences, Toronto (eTable 1 in the Supplement). We linked these databases using unique encoded identifiers and analyzed them at the Women’s College Hospital, Toronto.

The research protocol was approved by the research ethics boards of the Sunnybrook Health Sciences Centre and the Women’s College Hospital in Toronto. Under Section 45 of the Personal Health Information Protection Act, the Institute for Clinical Evaluative Sciences can receive and use health information without consent for purposes of analysis and compiling statistical information about the Canadian health care system.

From January 1, 2003, to December 31, 2014, we identified 96 676 patients who were residents of Ontario and had a diagnosis of breast cancer in the Ontario Cancer Registry (eTable 2 in the Supplement). We excluded 837 cases (0.9%) that were coded as nonfemale or unknown sex, or where the death date was earlier than diagnosis date. Of the remaining cases, we excluded the following: women younger than 20 years and 45 years or older at the index date of diagnosis (82 744 [85.6%]), women with a history of hysterectomy or oophorectomy (3164 [3.3%]), women with a history of previous cancer (except skin cancer) (396 [0.4%]), and women with stage 0 or unknown stage breast cancer (1982 [2.1%]), leaving 7553 women eligible for the study. For each woman in the study population, we retrieved information on the following: age at diagnosis, date of diagnosis, American Joint Commission on Cancer pathologic stage at diagnosis, ER, PR, ERBB2/HER2 (OMIM 164870) status, chemotherapy (yes or no), and radiotherapy (yes or no).

The exposure was any pregnancy in the period from 5 years before, until 5 years after, the index date of diagnosis of breast cancer (eTable 3 in the Supplement). For each woman, we constructed a pregnancy history for the period of 5 years before, until 5 years after, diagnosis of breast cancer (eFigure 1 in the Supplement). For women who had a live birth (or a stillbirth), we calculated the date of conception by subtracting 9 months from the date of delivery, assuming a span of 40 weeks for a full-term pregnancy. For a woman who had an abortion, the date of conception was calculated by subtracting 3 months from the date of the procedure code.

We classified women into 4 exposure groups (eFigure 2 in the Supplement). The first group, no pregnancy, was defined as women who had no conception from 5 years before, until 5 years after, the index date of their diagnosis. The second group, pregnancy before breast cancer, was defined as women who had conception from 5 years to 1 year before the index date of their diagnosis of breast cancer. The third group, pregnancy-associated breast cancer, was defined as women who had conception from 11 months before, until 21 months after, the index date of their diagnosis. The fourth group, pregnancy following breast cancer, was defined as women who had conception from 22 to 60 months after the index date of their diagnosis.

As of the end of study follow-up, we classified each woman as alive, dead from breast cancer, dead from another cause, or dead from an unknown cause. The primary outcome was 5-year overall survival.

The baseline characteristics of women according to exposure group were compared using descriptive statistics. We used the χ² test to compare frequency distributions for categorical variables, and a 2-tailed t test or the Wilcoxon rank sum test to compare continuous variables, where appropriate.

We defined overall survival as the period from the date of diagnosis of breast cancer to the date of death from any cause. We used the Kaplan-Meier method to estimate actuarial survival by exposure group. The log-rank test was used to compare survival rates between strata. We used the Cox proportional hazards regression model to estimate the hazard ratio (HR) for overall survival, comparing women with no pregnancy with women with pregnancy before breast cancer and women with pregnancy-associated breast cancer. We performed separate Cox proportional hazards regression models for women with abortion only and for women with live birth only (or stillbirth). We performed age-adjusted and multivariable models. The multivariable models included the year of diagnosis, age at diagnosis, cancer stage at diagnosis, ER, PR, and ERBB2/HER2 status, radiotherapy, chemotherapy, and pregnancy exposure group.

We performed a time-dependent analysis for women with pregnancy following breast cancer.^27-29 In the model, pregnancy was considered as a time-dependent covariate whereby exposure status was transferred from nonexposed to exposed at the date of conception. The immortal time (from diagnosis to conception) was combined with person-years in the unexposed group (nonpregnant women).

For all estimates, P ≤ .05 was considered statistically significant. All P values were 2-tailed. SAS, version 9.2 (SAS Institute Inc), was used for all statistical analyses.

We identified 7553 women with invasive breast cancer who were eligible for study entry; eTable 4 in the Supplement presents the baseline characteristics of the participants. The mean age at diagnosis of breast cancer was 39.1 years (median, 40 years; range, 20-44 years). The mean follow-up time was 5.2 years (median, 5.0 years; range, 0-11.2 years). Of all women, 5832 (77.2%) had no pregnancy, 1202 (15.9%) had 1 pregnancy, 398 (5.3%) had 2 pregnancies, and 121 (1.6%) had 3 pregnancies or more from 5 years before, until 5 years after, the index date of diagnosis of breast cancer. eTable 5 in the Supplement presents details on parity; 1302 women (17.2%) had a live birth, 14 (0.2%) had a stillbirth, and 700 (9.3%) had an abortion. eFigures 3 to 9 in the Supplement show the prevalence of pregnancy, abortion, and live birth in the 7553 women in the study from 5 years before, until 5 years after, the index date of diagnosis according to different age groups. A total of 700 women (9.3%) had an abortion from 5 years prior, to 5 years after, diagnosis of breast cancer. Most abortions (94%) occurred before the diagnosis (eFigure 4 in the Supplement).

eTable 6 in the Supplement presents baseline characteristics of breast cancers according to pregnancy group. Of 7553 women, 5832 (77.2%) had no pregnancy, 1108 (14.7%) had pregnancy before breast cancer, 501 (6.6%) had pregnancy-associated breast cancer, and 112 (1.5%) had pregnancy following breast cancer. Women with pregnancy-associated breast cancer were more likely than women with no pregnancy to be diagnosed with stage II to IV breast cancer (77.8% vs 71.5%; P < .001), to have node-positive tumors (52.1% vs 47.7%; P = .02), to have ER-negative tumors (36.5% vs 23.2%; P < .001), and to have triple-negative tumors (27.3% vs 16.8%; P = .001).

A total of 975 (12.9%) women had died at the end of study follow-up. eFigure 10 in the Supplement shows the number of women at risk of death from the time of diagnosis of breast cancer. The actuarial 5-year overall survival rate was 87.5% (95% CI, 86.5%-88.4%) for women with no pregnancy, 85.3% (95% CI, 82.8%-87.8%) for women with pregnancy before breast cancer, and 82.1% (95% CI, 78.3%-85.9%) for women with pregnancy-associated breast cancer (log-rank P = .003) (Figure). Fewer than 6 women died in the pregnancy following breast cancer group (actual number not shown owing to privacy protection).

The age-adjusted HR of all-cause mortality for women with pregnancy-associated breast cancer (vs women with no pregnancy) was 1.18 (95% CI, 0.91-1.53; P = .20) (Table 1). The multivariable HR was 1.11 (95% CI 0.86-1.45; P = .41). The multivariable HR of all-cause mortality for women with pregnancy-associated breast cancer (vs women with no pregnancy) remained unchanged in models without chemotherapy (1.12; 95% CI, 0.86-1.45; P = .40), radiotherapy (1.11; 95% CI, 0.86-1.44; P = .42), or both (1.12; 0.86-1.45; P = .40) (eTables 7 to 9 in the Supplement).

Table 2 presents the HR of all-cause mortality for women who had live birth or stillbirth. In these women, the age-adjusted HR for women with pregnancy-associated breast cancer (vs women with no pregnancy) was 1.42 (95% CI, 1.06-1.90; P = .02). The multivariable HR fell to 1.24 (95% CI, 0.92-1.66; P = .15). eTable 10 in the Supplement presents the HR of all-cause mortality for women who had an abortion.

In this model, pregnancy was incorporated as a time-dependent variable. Women with pregnancy following breast cancer were considered to be unexposed from the date of diagnosis to the date of conception and exposed from the date of conception. The 5-year actuarial survival rate was 96.7% (95% CI, 94.1%-99.3%) for women who had a live birth or stillbirth 6 months or more after diagnosis of breast cancer (vs 87.5% [95% CI, 86.5%-88.4%] for women with no pregnancy) (age-adjusted HR, 0.22; 95% CI, 0.10-0.49; P < .001) (eTable 11 in the Supplement). The annual absolute mortality rate for women who had live birth or stillbirth 6 months or more after diagnosis was 0.5%.

Among women with pregnancy-associated breast cancer who had a live birth or stillbirth, we examined features that affect survival (eTable 12 in the Supplement). Compared with women aged 40 to 44 years, the multivariable HR for all-cause mortality was poorer for women aged 20 to 29 years (2.06; 95% CI, 0.68-6.28; P = .20) and for women aged 30 to 34 years (1.71; 95% CI, 0.61-4.83; P = .31). The multivariable HR was lower for women with ER-positive tumors vs ER-negative tumors (0.37; 95% CI, 0.04-3.37; P = .38). eFigure 11 in the Supplement shows the unadjusted Kaplan-Meier survival curves in women with pregnancy-associated breast cancer according to age groups.

We conducted this study to determine whether pregnancy has an adverse effect on survival in women with breast cancer and found that a pregnancy simultaneous with breast cancer was not detrimental to survival. The 5-year age-adjusted HR of all-cause mortality was marginally higher for women with pregnancy-associated breast cancer vs those with no pregnancy (1.18; P = .20). The age-adjusted HR did not change substantially in the multivariable model (1.11; P = .41). We demonstrated that the strong age effect observed here may explain the survival disadvantage in this group reported in the past. We also demonstrated that a pregnancy following diagnosis of breast cancer was not detrimental to survival. In particular, compared with nonpregnant women, the risk of death significantly dropped for women who delivered 6 months or more after diagnosis of breast cancer (multivariable HR, 0.25 [eTable 11 in the Supplement]).

Our estimates for pregnancy-associated breast cancer are similar to those of 2 previous population-based cohort studies that adjusted their hazard models for tumor characteristics^13,14 and chemotherapy¹⁴ but not for radiotherapy. In a European collaborative study, Amant and colleagues¹⁴ compared 447 cases of pregnancy-associated breast cancer with 865 cases of breast cancer in women who were not pregnant. The 5-year adjusted HR of overall mortality for women with pregnancy-associated breast cancer vs those who were not pregnant was 1.19 (95% CI, 0.73-1.93; P = .51). Rodriguez and colleagues¹³ using the California Cancer Registry reported a similar estimate of the adjusted HR for women with pregnancy-associated breast cancer (1.14; 95% CI, 1.00-1.29; P = .05).

Many have proposed that pregnancy may adversely affect survival in women with breast cancer when the 2 conditions occur together.^30-35 Our study questions this position. We demonstrated a strong effect of age on survival in all hazard estimates. For pregnant women vs those who were not pregnant, the unadjusted HR was 1.50, which fell to 1.18 when we adjusted for age alone. Adjusting for factors other than age did not substantially change the HR estimates. In a population-based study of the Swedish Cancer Registry, Johansson and colleagues also reported a strong effect of age on HR estimates; in their study, unadjusted HR fell from 1.47 to an age-adjusted HR of 1.18.¹² It has been proposed that breast cancer during pregnancy has a poorer prognosis because of a delayed diagnosis leading to advanced disease at presentation, or because of a tendency for these tumors to have aggressive biology, or from a delay (or modification) in cancer treatment to avoid harm to the fetus.^36,37 We show that the principal underlying factor for poor prognosis is early age of diagnosis of breast cancer.

We identified that, among women with pregnancy-associated breast cancer, younger women had relatively poor overall survival. Compared with the nonpregnant women, the age-adjusted HR of all-cause mortality was 1.93 for women younger than 30 years, and 1.75 for women aged 30 to 34 years (eTable 12 in the Supplement). The HRs remained high in these age groups after adjusting for tumor characteristics and treatment, at 2.06 and 1.71, respectively. Amant and colleagues¹⁴ reported similar estimates. It is well documented that young age is independently associated with poor breast cancer survival.^38-41 In our study, among women with pregnancy-associated breast cancer, the proportion with more advanced (stages II-IV) breast cancer was 87.9% for women younger than 30 years, 83.3% for women aged 30 to 34 years, and 76.1% for women aged 35 to 39 years, compared with 62.6% for women 40 years and older (P < .001). The corresponding proportions of ER-negative tumors for these age groups were 22.7%, 20.2%, 16.8%, and 10.8%, respectively (P = .02).

Women with pregnancy-associated breast cancers were more likely than nonpregnant women to have ER-negative tumors (36.5% vs 23.2%; P < .001). The age-adjusted HR for ER-negative tumors (vs ER-positive) was 3.30 (95% CI, 1.14-9.55; P = .02). The mechanism by which the hormonal milieu of pregnancy could give rise to a higher proportion of ER-negative breast cancers is unclear. Possible explanations are that pregnancy might prevent ER-positive breast cancers, pregnancy might increase the risk of ER-negative breast cancers, and pregnancy might switch an ER-positive breast cancer to ER-negative cancer.

Our finding that pregnancy 6 months after diagnosis of breast cancer is safe is consistent with results of 3 previous cohort studies.^19,20,42 Mueller and colleagues¹⁹ reported an adjusted relative risk of 1.7 (95% CI, 1.2-2.6) for death for women who had a live birth within 3 months after diagnosis of breast cancer (vs no pregnancy); the relative risk dropped to 0.38 (95% CI, 0.12-1.2) for women who delivered within 7 to 9 months after diagnosis of breast cancer. In an Australian study, Ives and colleagues⁴² reported an adjusted HR of 2.20 for women who had a pregnancy less than 6 months after diagnosis of breast cancer, 0.45 for women who had a pregnancy 6 to 24 months after diagnosis, and 0.48 for women who had a pregnancy more than 24 months after diagnosis, compared with nonpregnant women. In our time-dependent analysis, the age-adjusted HR of all-cause mortality was 0.22 (P < .001) for women who had live birth or stillbirth 6 months after the index date of diagnosis of breast cancer. The HR remained unchanged in the multivariable model. The apparent survival benefit in women with pregnancy following breast cancer (vs nonpregnant women) might be owing to some degree of the so-called healthy mother effect, in which women who conceive are most likely to be a self-selected group of healthy women with better prognoses.

Our study has some limitations. The Ontario Cancer Registry has no information on the tumor size, nodal status, ER or PR status, and ERBB2/HER2 status for cases of breast cancer that were diagnosed before 2010. For women who had a live or stillbirth, the information on gestational weeks was available only for deliveries after April 2002. Gestational weeks were not available for most women who had an abortion. We calculated an approximate date of conception, which could underestimate or overestimate, the actual time of conception. Finally, we were unable to obtain information on the use of hormonal treatment (eg, tamoxifen). Despite these limitations, to our knowledge, our study is one of the largest to date on this subject. The major strength of our study was our ability to link the Ontario Cancer Registry database to other health administrative databases of Ontario. Therefore, our cohort was representative of the entire population of Ontario. Moreover, we adjusted the hazard model for both chemotherapy and radiotherapy. Furthermore, we performed a time-dependent analysis in which we calculated the survival from the date of conception, rather than the date of diagnosis.

Pregnancy did not adversely affect survival in women with breast cancer. The ER status is the strongest factor that discriminates between pregnancy-associated breast cancer and breast cancer in nonpregnant women. However, adjusting for the ER status did not influence the risk of death. For breast cancer survivors who wish to conceive, the risk of death is lowest if pregnancy occurs 6 months or more after diagnosis.

Accepted for Publication: January 23, 2017.

Corresponding Author: Steven A. Narod, MD, FRCPC, Women’s College Research Institute, Women’s College Hospital, 76 Grenville St, Toronto, ON M5S 1B2, Canada (steven.narod@wchospital.ca).

Published Online: March 9, 2017. doi:10.1001/jamaoncol.2017.0248

Author Contributions: Dr Iqbal had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Iqbal, Rochon, Giannakeas, Sun, Narod.

Acquisition, analysis, or interpretation of data: Iqbal, Amir, Giannakeas, Narod.

Drafting of the manuscript: Iqbal, Amir, Narod.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Iqbal, Giannakeas, Sun.

Obtained funding: Iqbal.

Administrative, technical, or material support: Iqbal, Amir, Rochon, Giannakeas.

Study supervision: Amir, Rochon, Narod.

Conflict of Interest Disclosures: Dr Narod reported holding a Canada Research Chair in Breast Cancer. No other disclosures were reported.

Funding/Support: This study was supported in part by a Canada Graduate Scholarship (Master’s) from the Canadian Institute of Health Research (Dr Iqbal), a Helen Marion Walker Award and an Enid Walker Graduate Student Award from the Women’s College Hospital (Dr Iqbal), and an Open Fellowship Award from the Institute of Medical Science, University of Torontol (Dr Iqbal).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

Disclaimer: This study made use of deidentified data from the Institute for Clinical Evaluative Sciences (ICES) Data Repository, which is managed by the Institute for Clinical Evaluative Sciences with support from its funders and partners: Canada’s Strategy for Patient-Oriented Research (SPOR), the Ontario SPOR Support Unit, the Canadian Institutes of Health Research and the Government of Ontario. The opinions, results and conclusions reported are those of the authors. No endorsement by ICES or any of its funders or partners is intended or should be inferred. Parts of this material are based on data and information compiled and provided by the Canadian Institute for Health Information. However, the analyses, conclusions, opinions, and statements expressed herein are those of the authors, and not necessarily those of CIHI.

Additional Contributions: Joel G. Ray, MD, MSc, FRCPC, St. Michael’s Hospital; Lorraine Lipscombe, MD, MSc, FRCPC, Women’s College Research Institute; and Susan Bronskill, PhD, ICES Central, reviewed the dataset creation plan during the program advisory committee meetings. Alison Park, MSc, ICES, provided technical help in regards to the coding of exposure. They were not compensated for their contributions.