Abstract
Background. Estrogen is an established growth factor in breast cancer and it has been hypothesized that pregnancy associated estrogens may increase the risk of recurrence of breast cancer. In 1997 we published a population-based Danish study indicating no negative prognostic effect of pregnancy after breast cancer treatment. The present study is a ten-year update. Material and Methods. Danish Breast Cancer Cooperative Group has since 1977 collected population-based data on tumour characteristics, treatment regimes, and follow-up status on Danish women with breast cancer. Pregnancy history was added from the Danish Civil Registration System, the National Birth Registry, and the National Induced Abortion registry. Cox regression was used to estimate the risk ratio of dying among women with a pregnancy after breast cancer treatment compared with women without such experience. Results. In all, 10 236 women with primary breast cancer aged 45 years or less at the time of diagnosis were followed for 95 616 person years. Among these, 371 women experienced pregnancy after treatment of breast cancer. In a multivariate analysis that included age at diagnosis, stage of disease, and pregnancy history prior to diagnosis, women who had a full-term pregnancy subsequent to breast cancer treatment were found to have a reduced risk of dying (relative risk: 0.73; 95% confidence interval: 0.54–0.99) compared with other women with breast cancer. The effect was not significantly modified by age at diagnosis, tumour size, nodal status, or pregnancy history before diagnosis of breast cancer. Neither spontaneous abortions nor induced abortions subsequent to breast cancer treatment had a negative impact on prognosis. Conclusion. In line with our previous study, but based on more than twice the patient material, we found no evidence that a pregnancy after treatment of breast cancer has a negative influence the prognosis.
Approximately 25% of breast cancers are diagnosed in women under the age of 50, and 15% under the age of 45. The incidence of breast cancer has increased by 2–3% over a time span of more than 40 years, while a decrease in mortality rate has been observed during the last decade Citation[1]. The decline in mortality has been most notable in women 50 years or younger Citation[2]. With a trend towards postponing childbearing in the western world alongside increased possibilities to preserve fertility during chemotherapy Citation[3] an increasing number of women are expected to conceive a child after treatment of breast cancer. Since estrogen is an established growth factor in breast cancer, it has been discussed whether women should be advised against pregnancy subsequent to breast cancer treatment because of the fear of a negative prognostic effect of the high estrogen levels associated with pregnancy.
The majority of studies on the prognostic impact of pregnancies succeeding treatment of breast cancer have only included a small number of patients (n = 7–50) Citation[4–12]. All these studies, with limited statistical power, conclude that women with subsequent pregnancy do equally as good as or even better than women who do not experience pregnancy after breast cancer treatment.
Larger case-control studies comprising 89 to 329 women with a full-term pregnancy after treatment of breast cancer confirms an overall good prognosis for these women Citation[13–16].
The above-mentioned studies have used a case-control design, which is prone to selection bias. Only women at good health are expected to get pregnant after cancer treatment, and if the registration is not complete, women who have died may have been lost in the follow-up.
Ten years ago we performed a Danish population-based cohort study addressing the question of the prognostic influence of pregnancy subsequent to breast cancer treatment Citation[17]. We found that subsequent pregnancy conferred no negative impact on prognosis. We here present an update on this study with additional ten years of inclusion and observation.
Patients and methods
The design of the original study and the registries has been described in detail elsewhere Citation[17]. Briefly, breast cancer patients aged 45 years or younger were identified through the DBCG Registry and information on vital status, childbirths, stillbirths and abortions were retrieved through linkage to the Danish Civil Registration System (CRS), the National Birth Registry, and the National Induced Abortion Registry.
DBCG founded a nation-wide registry in 1977 and diagnostic, therapeutic and follow-up data has since been collected by the use of standardized forms from women with breast cancer in Denmark. A continuum of randomized trials has been organized by the DBCG and coherent treatment guidelines have for each program been provided for patients treated outside of the trials.
Patients included in the treatment programs had a completely resected unilateral invasive carcinoma of the breast and no signs of distant metastasis as determined by routine examination. Axillary sampling (sentinel node) or clearance (level I and II) in combination with mastectomy or lumpectomy followed by irradiation was required.
Patients
Permission to perform the study was obtained in advance from the Danish National Committee on Biomedical Research Ethics and the Danish Data Protection Agency. Information on patients in the DBCG registry was linked with the other national registries to obtain information on pregnancy history and vital status. As women born before 1935 have no systematic link to all their children in the CRS registry, we restricted our study group to women born since April 1, 1935. Since the aim was to identify women with pregnancies, we further restricted the study group to women aged 45 years or less at the time of diagnosis. All women diagnosed before June 1, 2005, were included and followed until that day with respect to vital status.
Statistical analysis
The associations between the study variables and survival were investigated using the Cox Proportional Hazards method. Multivariate analyses included tumour characteristics (tumour size, nodal status, and histological type and grade), time between diagnosis and most recent previous childbirth (with nulliparous in a separate category), age at diagnosis, year of treatment, protocol allocation, full-term pregnancy after diagnosis, induced abortion after diagnosis, and spontaneous abortion after diagnosis. The three last variables were included in the analysis as time-dependent variables.
The adequacies of the proportional hazard assumptions for the included variables were assessed by Schoenfeld residuals and by including a time-dependent component for each covariate in the model. The hazard rates of histological type and grade were not proportional and stratification was used, and a time-dependent component was kept for each of the variables tumour size, nodal status and protocol allocation. Estimation was done using the SAS procedure PROC PHREG.
Results
Overall, 10 295 women aged 45 years or less with primary breast cancer were identified. Since the specific aim of the study was to evaluate the prognostic effect of having a pregnancy subsequent to breast cancer treatment, we excluded 59 women who might have been pregnant at the time of diagnosis, i. e. women who had a childbirth less than 10 months after the breast cancer diagnosis, or women who had an abortion with a gestational age indicating that they might have been pregnant at the time of diagnosis of breast cancer, leaving 10 236 patients for further investigation.
Among these, 371 women (3.6%) experienced a total of 465 pregnancies (236 full-term pregnancies, 36 spontaneous abortions, and 193 induced abortions). Seventy-four women had more than one pregnancy after breast cancer diagnosis. The median time between breast cancer diagnosis and birth/abortion was a) full-term pregnancy 39 months (range 10–228 months, n = 236), b) spontaneous abortion 26 months (range 5–204 months, n = 36), and c) induced abortion 23 months (range 3–200 months, n = 193).
We have extended our earlier nation-wide cohort of breast cancer patients aged 45 years or younger from 5 752 to 10 236 women, and provided an additional follow-up of 10 years for the original cohort representing in all 95 616 person-years of follow-up.
Distribution of patients according to histo-pathological tumour criteria, protocol allocation, and pregnancy history after diagnosis of breast cancer is shown in . Significantly fewer women with subsequent pregnancy were node positive (p < 0.01), and the group tended to have smaller tumours (p = 0.09). Adjusted relative risk of dying according to pregnancy history after treatment of breast cancer is given in .
Table I. Distribution of 10 236 breast cancer patients, diagnosed 1978–2005, according to age at diagnosis, tumour characteristics, protocol allocation, and pregnancy history subsequent to their diagnosis. Danish women born after April 1, 1935 and up to 45 years of age at diagnosis.
Table II. Adjusted relative risk (aRR) of dying according to pregnancy history after diagnosis of breast cancer among 10 236 women.
Women with a full-term pregnancy after treatment of breast cancer had significantly reduced risk of dying (RR: 0.73; 95% CI: 0.54-0.99, p = 0.04) compared to other women with breast cancer. Women experiencing spontaneous abortion were also found to have a significant reduced risk of death whereas women having induced abortion experienced no significant risk alteration.
Further multivariate analysis showed that the effect of subsequent pregnancy was not significantly modified by age at diagnosis, tumour size, nodal status, status as parous/nulliparous before diagnosis, time since most recent previous pregnancy before breast cancer diagnosis, age at subsequent pregnancy, or time to subsequent pregnancy (data not shown).
We subsequently performed a restricted analysis including only women who were classified as having a low-risk tumour not receiving adjuvant systemic treatment (n = 2 901). Also in this group of breast cancer patients, the survival was favourable for women with a full-term pregnancy subsequent to breast cancer treatment (RR: 0.56; 95% CI: 0.29–1.11) compared to other women with low-risk breast cancer.
Women with relapse are not expected to get pregnant deliberately. However, restricting the analysis to women with available information on relapse, and introducing relapse in the multivariate analysis did not change the overall results.
Calculated on the basis of age and period specific incidence of pregnancies among Danish women the expected number of full term pregnancies was 696 compared with the observed 236 (34%) and the expected number of induced abortions was 205 compared with the observed 193.
Discussion
Breast cancer diagnosed during pregnancy or in the first years afterwards has been associated with a poor outcome Citation[18]. Based on anecdotal reports the primary concern regarding pregnancy following breast cancer has been a possible negative impact on the patients prognosis. Furthermore, the high serum levels of pregnancy related hormones are well known for their induction of breast tissue growth. The incidence of breast cancer is increased in the first years after childbirth indicating a pregnancy related growth induction Citation[19], Citation[20]. Findings such as these have led to considerable discussion on the possible negative prognostic effect of a pregnancy that follows breast cancer treatment.
This nation-wide populations based cohort study extends the results of our previous publication, and the present update of the data from DBCG confirms that pregnancy after breast cancer has no negative impact on prognosis. Among the strengts of our study are the population based and nation-wide design, the high completeness of the registries used, the comprehensive data on diagnosis and treatment of breast cancer, essentially complete long-term follow-up and the relative large sample size. This update includes more than twice as many women experiencing pregnancy after breast cancer treatment as the primary publication Citation[17], and results are based on more than two times the number of person years of follow-up. Recently an Australian study of similar design has shown result in line with the present study Citation[21].
About 2% of breast cancer cases are diagnosed in women aged less than 35 years. Since the average age of first birth in Denmark today is around 30 years and the overall survival in breast cancer patients has increased over the years the number of women seeking advise concerning pregnancy after treatment of breast cancer is expected to rise Citation[22], Citation[23].
The present study documented that a pregnancy subsequent to treatment of breast cancer conferred no negative effect on the prognosis. Because women with a poor prognosis are believed to avoid pregnancies, there is a potential problem of the exposed group being selected. This problem is not easy to overcome and has been the main concern regarding the interpretation of results from previous studies on this subject. The present investigation took advantage of the clinical population-based DBCG database that over more than three decades has recorded detailed information on breast tumour characteristics. Also, in the present study the group of women with subsequent pregnancy tended to earlier stage disease regarding tumours size and a significantly lover risk of nodal involvement. Though, we were able to perform a detailed adjustment for the influence of important prognostic factors and thus to minimize selection bias. Furthermore, the use of time-dependent variables in a cohort design enabled us to adequately adjust for the important influence of time from breast cancer diagnosis to birth/abortion. Thus, the length of the relapse free period is believed to significantly influence the woman's decision regarding pregnancy. Restricting the analysis to women with low-risk disease or introducing information on relapse in the analysis did not change the results. We acknowledge that despite these efforts, there are likely to be other selection mechanisms for which we were unable to adequately adjust with the available prognostic factors. This may explain why women with a full-term pregnancy subsequent to breast cancer treatment, even after adjustment for established prognostic factors, tended to have a better outcome than women without a subsequent pregnancy.
The fertility rate, measured as number of full term pregnancies, was reduced to one third of the expected level in the group of treated breast cancer patients. This is a consequence of both an overall lower number of pregnancies and an increased incidence of induced abortions. The expected number of induced abortions constituted 29% of the expected number of full-term pregnancies, whereas induced abortions among breast cancer patients constituted 82% of full-term pregnancies. Unplanned pregnancy when a woman is seriously ill most likely leads to a higher rate of induced abortions. It is obvious that many women have avoided getting pregnant after a diagnosis of breast cancer. Furthermore, it is possible that some women have chosen induced abortion due to lack of knowledge of the influence a pregnancy might have on the course of their treated breast cancer. However, women with a history of induced abortion after breast cancer treatment did not have a different profile of prognostic factors than other women, which suggests that induced abortion was not chosen primarily among patients with a poor prognosis. This finding further supports the credibility of the overall result.
We conclude that there is no reason to consider a pregnancy subsequent to breast cancer treatment to have a negative effect on the woman's survival.
References
- Ewertz M, Duffy SW. Incidence of female breast cancer in relation to prevalence of risk factors in Denmark. Int J Cancer 1994; 56: 783–7
- Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: An overview of the randomised trials. Lancet 2005;365(9472):1687–717.
- Sonmezer M, Oktay K. Fertility preservation in young women undergoing breast cancer therapy. Oncologist 2006; 11: 422–34
- Cooper DR, Butterfield J. Pregnancy subsequent to mastectomy for cancer of the breast. Ann Surg 1970; 171: 429–33
- Blakely LJ, Buzdar AU, Lozada JA, Shullaih SA, Hoy E, Smith TL, Hortobagyi GN. Effects of pregnancy after treatment for breast carcinoma on survival and risk of recurrence. Cancer 2004; 100: 465–9
- Applewhite RR, Smith LR, DiVincenti F. Carcinoma of the breast associated with pregnancy and lactation. Am Surg 1973; 39: 101–4
- Harvey JC, Rosen PP, Ashikari R, Robbins GF, Kinne DW. The effect of pregnancy on the prognosis of carcinoma of the breast following radical mastectomy. Surg Gynecol Obstet 1981; 153: 723–5
- Nugent P, O'Connell TX. Breast cancer and pregnancy. Arch Surg 1985; 120: 1221–4
- King RM, Welch JS, Martin JK, Jr, Coulam CB. Carcinoma of the breast associated with pregnancy. Surg Gynecol Obstet 1985; 160: 228–32
- von Schoultz E, Johansson H, Wilking N, Rutqvist LE. Influence of prior and subsequent pregnancy on breast cancer prognosis. J Clin Oncol 1995; 13: 430–4
- Lethaby AE, O'Neill MA, Mason BH, Holdaway IM, Harvey VJ. Overall survival from breast cancer in women pregnant or lactating at or after diagnosis. Auckland Breast Cancer Study Group. Int J Cancer 1996; 67: 751–5
- Malamos NA, Stathopoulos GP, Keramopoulos A, Papadiamantis J, Vassilaros S. Pregnancy and offspring after the appearance of breast cancer. Oncology 1996; 53: 471–5
- Sankila R, Heinavaara S, Hakulinen T. Survival of breast cancer patients after subsequent term pregnancy: “Healthy mother effect”. Am J Obstet Gynecol 1994; 170: 818–23
- Velentgas P, Daling JR, Malone KE, Weiss NS, Williams MA, Self SG, et al. Pregnancy after breast carcinoma. Cancer 1999; 85: 2424–32
- Gelber S, Coates AS, Goldhirsch A, Castiglione-Gertsch M, Marini G, Lindtner J, et al. Effect of pregnancy on overall survival after the diagnosis of early-stage breast cancer. J Clin Oncol 2001; 19: 1671–5
- Mueller BA, Simon MS, Deapen D, Kamineni A, Malone KE, Daling JR. Childbearing and survival after breast carcinoma in young women. Cancer 2003; 98: 1131–40
- Kroman N, Jensen MB, Melbye M, Wohlfahrt J, Mouridsen HT. Should women be advised against pregnancy after breast-cancer treatment?. Lancet 1997; 350(9074)319–22
- Kroman N, Mouridsen HT. Prognostic influence of pregnancy before, around, and after diagnosis of breast cancer. Breast 2003; 12: 516–21
- Lambe M, Hsieh C, Trichopoulos D, Ekbom A, Pavia M, Adami HO. Transient increase in the risk of breast cancer after giving birth. N Engl J Med 1994; 331: 5–9
- Wohlfahrt J, Andersen PK, Mouridsen HT, Melbye M. Risk of late-stage breast cancer after a childbirth. Am J Epidemiol 2001; 153: 1079–84
- Ives A, Saunders C, Bulsara M, Semmens J. Pregnancy after breast cancer: Population based study. BMJ 2007; 334(7586)194
- Braun M, Hasson-Ohayon I, Perry S, Kaufman B, Uziely B. Motivation for giving birth after breast cancer. Psychooncology 2005; 14: 282–96
- Partridge AH, Gelber S, Peppercorn J, Sampson E, Knudsen K, Laufer M, et al. Web-based survey of fertility issues in young women with breast cancer. J Clin Oncol 2004; 22: 4174–83