Breast cancer is the most commonly diagnosed malignancy during or around the time of pregnancy. Due to diagnostic delay, potential promotional effects of gestational hormones on breast tumors, and perhaps other factors causing intrinsic biologic aggressiveness of breast tumors, pregnant women with breast cancer may present with more advanced stages of disease, and therefore have a poorer prognosis.1,2 Other studies have suggested that when controlled for stage of disease, pregnancy does not confer a worse prognosis on the woman with breast cancer.3–5
In this investigation, we reviewed a large population database of women in California diagnosed with breast cancer during the years 1991–1999 and compared a cohort of women with pregnancy-associated breast cancer to a cohort of women with breast cancer not associated with pregnancy. Our aims were to compare the stage distribution, tumor characteristics, and survival between the two cohorts and to evaluate whether pregnancy is an independent predictor of poor outcome.
MATERIALS AND METHODS
Breast cancer cases diagnosed between 1991and 1999 were identified using the California Cancer Registry, which documents 99% of all malignancies diagnosed in California. Pregnancies were identified through a data set available from the California Office of Statewide Health Planning and Development, which consists of vital statistics and birth, infant death, and neonatal and maternal hospital discharge records. This database contains all births in California hospitals beyond 20 weeks, including multiple births and stillbirths. Overall, 4,846,505 obstetric patients and 4,906,920 infants were reviewed through this database. The pregnancy information was then linked by a probabilistic data record linkage to data from the California Cancer Registry with Integrity 3.3 software (Vality Technologies, Inc., Boston, MA) by means of social security numbers, date of birth, and ZIP code of residence.6 Linked files were deidentified and contained coded identifiers to preserve patient anonymity.
Women were identified through the linked database as having pregnancy-associated breast cancer if their diagnoses of breast cancer occurred within 9 months before delivery date, at delivery, or within 1 year postpartum. Date of diagnosis was defined as the earlier of either the California Cancer Registry diagnosis date or any admission to the hospital with the coded diagnosis of breast cancer. Disease-specific data available from the cancer registry included stage assessment, histologic type, size of the primary tumor, hormone receptor status, lymph node status, and treatment information, as well as vital status. In this study, disease stage assessment was coded according to the Surveillance, Epidemiology, and End Results guidelines, in particular the Extent of Disease guideline, which was converted to American Joint Committee on Cancer staging, 5th edition.7,8 Only women with invasive breast cancer were included in the analysis. Lymph node information was available through the California Cancer Registry coded variables “number of lymph nodes assessed” and “number of lymph nodes positive,” therefore capturing all surgical lymph node assessments, including sentinel lymph node technique, biopsy, or lymphadenectomy. Women with breast cancer during the 9 months preceding obstetric delivery were termed “antepartum” cases, whereas women with date of diagnosis within the 12 months after delivery were termed “postpartum” cases. Cases were classified as “at delivery” if the date of diagnosis coincided with the delivery hospitalization.
Data for the women with pregnancy-associated breast cancer were compared against a group of age-matched, nonpregnant women with breast cancer during the same time period. The match was achieved by frequency matching: selecting the same proportion of women from each age group (younger than 30, 30–40, and older than 40 years). Overall, 4,177 control cases were used for the nonpregnant cohort.
Fisher exact tests and Student t tests were used to evaluate the differences in clinical outcomes between cohorts. Hazard ratios were calculated by the Cox proportional hazards model to measure the prognostic variables predictive of death from breast cancer.9 Kaplan-Meier survival distribution curves were created and log-rank tests were used to analyze the difference of survival functions.10 Survival information was obtained through December 31, 2004. We used SAS 9.1 software for all analyses (SAS Institute Inc., Cary, NC). The investigation was approved by the Office of Statewide Health Planning and Development, by the Human Subjects Committee of the University of California-Davis, and by the State of California Committee for the Protection of Human Subjects.
Pregnancy-associated breast cancer was identified in 797 women in California during 1991–1999, including 179 antepartum, 8 at delivery, and 610 postpartum cases. These cases were compared with 4,177 age-matched women with breast cancer who did not have associated pregnancies. Patients were followed through December 31, 2004. The median follow-up was 2,297 days for the pregnancy-associated breast cancer group and 2,671 days for the non–pregnancy-associated group.
Table 1 illustrates the demographic characteristics of the patient cohorts. Although matched for age, women with pregnancy-associated breast cancer were significantly different from the nonpregnant control group in most other variables studied. Women with pregnancy-associated breast cancer were less likely to be non-Hispanic White and more likely to have higher stage disease, larger primary tumor, mastectomy rather than other forms of surgical treatment, and infiltrating ductal histology. There was no statistically significant difference in the rates of lymph node assessment between the two groups. Women with pregnancy-associated breast cancer were significantly less likely to have estrogen receptor–positive or progesterone receptor–positive tumors. The percentage of women with more advanced disease (Stages II and higher) was higher for women with antepartum breast cancer (77.1%) and for women with pregnancy-associated breast cancer 0–12 months postpartum (72.3%) than for the nonpregnant control women (62.0%, P<.001). The death rate was higher in women with pregnancy-associated breast cancer.
There was no difference in the average time between diagnosis and surgical treatment between the nonpregnant control cohort (20.7 days) and the pregnancy-associated breast cancer group (24.2 days). Pregnancy-associated breast cancer antepartum and “at delivery” cases averaged 23.9 days to surgery, whereas postpartum cases averaged 24.3 days to surgery (P=.11, .34 and .15 respectively). Average time to start chemotherapy from diagnosis also was not significantly different for the pregnancy-associated breast cancer cases (67.4 days) compared with nonpregnant control cohort cases (65.0 days), even when analyzed by antepartum (75.3 days) compared with postpartum cases (65.0 days) (P=.68, .16, and 1.0, respectively). In women who received radiation, however, there was a significant difference in the pregnant group compared with the nonpregnant group: nonpregnant women with breast cancer averaged 146.2 days from diagnosis to the start of radiation, whereas pregnancy-associated cases averaged 183.3 days to radiation (P=.001). The difference was more marked for the antepartum or at-delivery cases, with an average time to radiation of 215.4 days (P=.001). Postpartum-associated cases averaged 174.0 days to radiation (P=.004). When comparing the various treatments received (Table 2), the only difference seen was the higher rate of mastectomy in the pregnancy-associated breast cancer group compared with other surgeries such as lumpectomy.
Table 3 shows the proportion of pregnancy-associated breast cancer and nonpregnant control cohort patients who died by the end of 2004 based on an analysis of tumor characteristics or treatment received. In this univariable analysis, mortality rates were not significantly different between the pregnancy-associated breast cancer and nonpregnant control cohort cohorts when looking at specific stage groupings or at outcome related to the primary tumor size. Of patients who received chemotherapy, the pregnancy-associated breast cancer group had a higher death rate (42.4% compared with 37.8%, P=.04); however, patients in the pregnancy-associated breast cancer group who did not receive chemotherapy also had lower survival rates than those in the nonpregnant control cohort group who did not receive chemotherapy (31.6% dead compared with 23.7%, P=.01). Also, pregnancy-associated breast cancer patients who did not receive radiation therapy had lower survival rates than the nonpregnant control cohort patients who did not receive radiation therapy (40.9% dead compared with 34.9%, P=.01).
Various prognostic factors were assessed for their effects on the survival of patients with breast cancer using a Cox proportional hazards model. These factors are illustrated in Table 4. More advanced stage was associated with worse survival. Women with pregnancy-associated breast cancer, when all the other variables such as stage, surgery, hormone receptor status, tumor size, and race are controlled, have a marginally significantly higher risk of dying (P=.046). Tumor size was not an independent predictor of poor outcome. African American women had a two-thirds increased risk of death from breast cancer compared with non-Hispanic white women (P<.001), whereas Asian women had a lower risk of death (P=.048). Modified radical mastectomy was associated with a slightly worse prognosis compared with conservative surgeries (P=.005). Hormone receptor-negative tumors had a 20% increased risk of death compared with positive tumors (either estrogen receptor or progesterone receptor) (P=.003). When tumors with unknown hormone receptor status were excluded from the Cox model, there were no significant changes in the hazard ratios for any of the other variables (data not shown). If only localized-stage disease is considered, pregnancy association was still a significant risk factor for worse survival even when controlling for surgical treatment, hormone receptor status, age, and race (P=.049).
Breast cancer is the most common invasive malignancy diagnosed during pregnancy.6 The overwhelming majority of case–control studies has found that the prognosis of breast cancer in pregnancy overall is worse, but when controlled for stage at presentation, prognosis is not significantly different from non–pregnancy-associated breast cancer.3–5 However, one study from Norway of 20 patients and one case–control study from France showed pregnancy to confer a worse prognosis.11,12 The Norwegian study controlled for stage but was a small study. The French study performed a multivariate analysis including primary tumor size (more than 3 cm), microscopic lymph node involvement (3 or more lymph nodes), age older than 33 years, and pregnancy and found pregnancy at diagnosis gave a relative risk of 1.76 (95% confidence interval [CI] 1.50–2.02) for death.11
Our study has both advantages and limitations in comparison with prior studies on pregnancy-associated breast cancer. Our study is a large, population-based retrospective evaluation including all of California over a defined period of time. Since 1988, all cancer cases in California have been mandated by law to be registered with the California Cancer Registry. Case-finding audits performed at the California Cancer Registry have shown a very high case reporting rate (approximately 99%), so the population of the state can be assumed to be accurately represented in the data.13 The large number of pregnancy-associated breast cancer cases we have been able to identify with this review minimizes the risk of a type II error in the analysis of results. The limitations of our study are those seen with other database analyses: some data points may be missing or erroneously entered into the system, some patients may have left the state and not been captured in follow-up, pathology review is not possible, and not all risk factors of interest (such as Her2 positivity) were available in the data. Since the database is more focused on hospital admissions, data from the California Cancer Registry on chemotherapy administration and radiation treatment may be missing.14 Endocrine therapy, an important component of breast cancer care, is not captured. Additionally, women who did not have pregnancies that continued beyond 20 weeks gestation (due to either spontaneous loss or therapeutic abortion) were not included in the analysis. The inability to capture these women in the data may have eliminated some patients with more advanced or metastatic tumors who chose to terminate or otherwise lost their pregnancies. However, our prognostic factor modeling and survival analysis controlled for the stage of disease and size of the primary tumor, therefore any effect of missing advanced stage patients would have been minimal.
There is a paucity of information in the literature about the demographic characteristics of women with pregnancy-associated breast cancer. Breast cancer in general is more common in non-Hispanic white women and women of higher socioeconomic status.15 In our study, we found a significantly higher proportion of the pregnancy-associated cases to be either of Hispanic ethnicity or Asian/ Pacific Islander (37.9% of cases) compared with age-matched nonpregnant control cohort cases (30.9%) and a lower proportion to be non-Hispanic white. Our finding of a higher proportion of cases being found in ethnic minorities may reflect the diversity of our state; however, our control women are also from California. Although it is not clear, the differences may therefore reflect differences in pregnancy rates among the different ethnic groups rather than true intrinsic differences in the risk of a pregnancy-associated breast cancer.16
In the multivariable analysis of risk factors for death in women diagnosed with breast cancer, African-American women had a 68% higher risk of death than non-Hispanic white women. Lower incidence of breast cancer but higher mortality has been demonstrated in a number of studies and may be due in part to a higher prevalence of basal-like histology in African-American women.17 Hispanic women showed no difference in overall survival from non-Hispanic whites, and Asian women showed improved prognosis over non-Hispanic whites.
Our data show a significant difference in the rate of estrogen and progesterone receptor positivity, with the pregnancy-associated tumors being significantly less likely to be hormone receptor positive. The multivariable analysis confirmed that estrogen receptor– and progesterone receptor–negative tumors conferred a worse prognosis (hazard ratio [HR] 1.20 for death, 95% CI 1.07–1.36). This finding was consistent even when only those cases with estrogen receptor or progesterone receptor information available were analyzed in a multivariable fashion. This improved outcome for women with hormone receptor–positive tumors is consistent with the nonpregnant breast cancer literature.18 Not many prior studies have analyzed the hormone receptor status of pregnancy-associated tumors; however, some studies have also reported a higher rate of hormone receptor–negative tumors in these women.11,19,20 Other studies, however, have not shown a difference in receptor status in pregnant patients compared with age-matched nonpregnant patients.21
The stage distribution of pregnancy-associated breast cancer in the current study confirms findings of other studies that a larger proportion of pregnancy-associated cases are more advanced stage at diagnosis. The pregnancy-associated breast cancer tumors were also more likely to be larger, with only 29.5% 2 cm or less compared with 39.6% of nonpregnant control cohort tumors. Theories behind the more advanced size of tumors and more advanced stage at presentation that are seen in pregnancy include diagnostic delay (both due to physiologic changes in the breast creating more benign changes as well as reluctance to perform mammography or biopsy during the pregnant or lactating state) and the creation of a more aggressive tumor biology with the high endogenous hormonal milieu of pregnancy (estrogen, progesterone, and potentially other hormones such as insulin-like growth factor or prolactin).2 In this investigation, we evaluated the stage distribution of the pregnancy-associated breast cancer cases diagnosed during the antepartum period (77.1% beyond stage I) and the postpartum period (72.3% beyond stage I) compared with nonpregnant control cohort control women (62.0% beyond stage I). Both antepartum and postpartum diagnoses were more likely to be more advanced stage. Although there could still be a component of diagnostic delay in the postpartum period as compared with the antepartum period (due to physiologic breast changes and low physician suspicion for malignancy), the postpartum hormonal milieu is quite different from the antepartum state, with postpartum women having low to normal levels of estradiol and progesterone and elevated levels of prolactin only if lactating.22 It is likely, however, that the tumors could have been initiated or accelerated during the high hormonal antepartum state and only diagnosed in the postpartum state. Other epidemiologic studies have shown that having given birth within 2 years of the diagnosis of breast cancer increases risk of mortality, suggesting that factors other than the high hormonal state may be implicated.23 Schedin2 provides an analysis of mammary gland involution (postpartum or postlactation) as a potentially important prooncogenic local environment in the breast that could explain the poor prognoses associated with these tumors.
We evaluated surgical treatment, chemotherapy, and radiation treatment to attempt to ascertain treatment delay after diagnosis in pregnancy-associated breast cancer. Although there were no differences between the pregnancy-associated breast cancer and nonpregnant control cohort cohorts with regard to average time to surgery or average time to start chemotherapy (including antepartum cases), there was a significant difference in time to radiation, particularly in the antepartum group. This finding is understandable given the effects of ionizing radiation on the developing fetus. Although there may be some limitations in our data with regard to initiation of treatment (such as potential for inaccuracy with regard to the reporting of these dates), overall, the finding that neither surgery nor chemotherapy are being delayed once the diagnosis is made is reassuring. We also saw that the type of surgical approach differed between the pregnancy-associated breast cancer cohort and the nonpregnant control cohort, with pregnancy-associated breast cancer cases more likely to undergo mastectomy. Although studies in nonpregnant women support the equivalency of modified radical mastectomy compared with breast-conserving lumpectomy with radiation,24 it is not clear that with the increased vascularity and proliferative changes of the breast during and after pregnancy that local control would be equivalent with the more conservative procedure. In addition, the use of ionizing radiation would be more limited or not available in cases of pregnancy-associated breast cancer, especially in antepartum cases. Therefore, mastectomy has been the mainstay of treatment of breast cancer in pregnancy-associated breast cancer and is recommended by the Society of Obstetricians and Gynaecologists of Canada clinical practice guidelines for breast cancer in pregnancy.25 Our data, however, do not support the idea that modified radical mastectomy is necessary for optimal treatment of pregnancy-associated breast cancer. A multivariable analysis for risk of death in the pregnancy-associated breast cancer cases only showed no difference between women treated with modified radical mastectomy compared with other surgical procedures, controlling for age, stage, tumor size, race, and hormone receptor status (HR 1.12, 95% CI 0.86–1.46) (data not shown). Lymph node evaluation rates in our study did not differ between pregnancy-associated breast cancer and nonpregnant control cohort groups. Sentinel lymph node evaluation, which has become standard in most cases of early breast cancer, is considered contraindicated in pregnancy.1 Unfortunately, our data do not allow us to determine whether any patients actually underwent sentinel lymph node evaluation, but our findings are reassuring that overall lymph node evaluation rates are not being compromised by the pregnant or postpartum state.
Our study shows a modest independent effect of pregnancy on outcome in breast cancer. Even when controlled for stage of disease, size of tumor, hormone receptor status, age, race, and type of surgery, survival is worse in pregnancy-associated breast cancer as compared with nonpregnant breast cancer cases (P=.046). Pregnancy association conferred only a 14% increase in the risk of death. Our multivariable analysis contradicts the multiple prior studies that relate the difference in survival to the difference in stage at presentation and is more in line with the French study11 that showed a worse prognosis in pregnancy even when controlled for tumor size and lymph node status. Although overall rates of treatment modalities such as radiation and chemotherapy are similar, it is not clear why there would be a difference in survival for pregnancy-associated cases. It is possible that the relative immune suppression of pregnancy may play a role. Another theory often cited is a possible difference in tumor biology caused by either the higher estrogen and progesterone levels or some other yet-to-be-clarified difference, such as differences in the growth hormone/insulin-like growth factor axis.26 The elevated hormone levels of the pregnant state could accelerate tumor growth or metastasis through either an intrinsic effect on the tumor cell biology or by increasing vascularity and subsequently metastasis out of the breast. The lower rate of hormone receptor–positive tumors associated with pregnancy is not consistent with an explanation involving high levels of estrogens or progesterone during pregnancy driving tumor progression. A question not addressed by our data is whether the worse prognosis of the pregnancy-associated cases could be changed in any way by the termination of pregnancy after the diagnosis of breast cancer (although the HRs are similar for antepartum disease and postpartum disease, therefore, it seems unlikely). Other studies do not support an improvement in prognosis with termination of pregnancy.1
Guidelines for management of breast cancer associated with pregnancy have previously been published by Psyrri and Burtness,1 Loibl et al,27 and the Society of Obstetricians and Gynaecologists of Canada (Helewa et al25), based on reviews of the literature. Our study gives additional information to help further clarify the counseling and recommendations for clinicians treating women in this difficult clinical situation.
1. Psyrri A, Burtness B. Pregnancy-associated breast cancer. Cancer J 2005;11:83–95.
2. Schedin P. Pregnancy-associated breast cancer and metastasis. 2006;6: 281–91.
3. Ishida T, Yokoe T, Kasumi F, Sakamoto G, Makita M, Tominaga T, et al. Clinicopathologic characteristics and prognosis of breast cancer patients associated with pregnancy and lactation: analysis of case-control study in Japan. Jpn J Cancer Res 1992;83:1143–9.
4. Petrek JA, Dukoff R, Rogatko A. Prognosis of pregnancy-associated breast cancer. Cancer 1991;67:869–72.
5. King RM, Welch JS, Martin JK Jr, Coulam CB. Carcinoma of the breast associated with pregnancy. Surg Gynecol Obstet 1985;160: 228–32.
6. Smith LH, Danielsen B, Allen ME, Cress R. Cancer associated with obstetric delivery: results of linkage with the California cancer registry. Am J Obstet Gynecol 2003;189:1128–35.
7. Shambaugh EM, Weiss MA, editors. Summary staging guide: cancer Surveillance, Epidemiology, and End Results reporting, April 1977 (reprinted July, 1998), NIH Publication no. 98- 2313. Bethesda (MD): National Institutes of Health; 1998.
8. Seiffert J,. SEER Program: comparative staging guide for cancer, version 1.1. NIH Publication No. 93-3640. Bethesda (MD): National Cancer Institute; 1993.
9. Cox DR, Oakes D. Analysis of survival data. London (UK): Chapman and Hall; 1984.
10. Peto R, Peto J. Asymptotically efficient rank invariant test procedures. J R Stat Assoc A. 1972;135:185–206.
11. Bonnier P, Romain S, Dilhuydy JM, Bonichon F, Julien JP, Charpin C, et al. Influence of pregnancy on the outcome of breast cancer: a case-control study. Societe Francaise de Senologie et de Pathologie Mammaire Study Group. Int J Cancer 1997;72:720–7.
12. Tretli S, Kvalheim G, Thoresen S, Høst H. Survival of breast cancer patients diagnosed during pregnancy or lactation. Br J Cancer 1988;58:382–4.
13. Kwong SL, Perkins CI, Morris CR, Cohen R, Allen M, Wright WE. Cancer in California: 1988-1999. Sacramento (CA): California Department of Health Services, Cancer Surveillance Section; 2001.
14. Cress RD, Zaslavsky AM, West DW, Wolf RE, Felter MC, Ayanian JZ. Completeness of information on adjuvant therapies for colorectal cancer in population-based registries [published erratum appears in Med Care 2005;43:423]. Med Care 2003;41:1006–12.
15. Ries LAG, Melbert D, Krapcho M, Stinchcomb DG, Howlader N, Horner MJ, et al. SEER cancer statistics review, 1975–2005. National Cancer Institute. Available at: http://seer.cancer.gov/csr/1975_2005/
. Retrieved May 19, 2008.
17. Carey LA, Perou CM, Livasy CA, Dressler LG, Cowan D, Conway K, et al. Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study. JAMA 2006;295:2492–502.
18. Dunnwald LK, Rossing MA, Li CI. Hormone receptor status, tumor characteristics, and prognosis: a prospective cohort of breast cancer patients. Breast Cancer Res 2007;9:R6.
19. Middleton LP, Amin M, Gwyn K, Theriault R, Sahin A. Breast carcinoma in pregnant women: assessment of clinicopathologic and immunohistochemical features. Cancer 2003;98:1055–60.
20. Aziz S, Pervez S, Khan S, et al. Case control study of novel prognostic markers and disease outcome in pregnancy/lactation-associated breast carcinoma. Pathol Res Pract 2003;199:15–21.
21. Elledge RM, Ciocca DR. Langone G, McGuire WL. Estrogen receptor, progesterone receptor, and HER-2/neu protein in breast cancers from pregnant patients. Cancer 1993;71: 2499–506.
22. Velasquez EV, Trigo RV, Creus S, Campo S, Croxatto HB. Pituitary-ovarian axis during lactational amenorrhoea. I. Longitudinal assessment of follicular growth, gonadotrophins, sex steroids and inhibin levels before and after recovery of menstrual cyclicity. Hum Reprod 2006;21:909–15.
23. Daling JR, Malone KE, Doody DR, Anderson BO, Porter PL. The relation of reproductive factors to mortality from breast cancer. Cancer Epidemiol Biomarkers Prev 2002;11:235–41.
24. Fisher B, Anderson S, Bryant J, Margolese RG, Deutsch M, Fisher ER, et al. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med 2002;347:1233–41.
25. Helewa M, Levesque P, Provencher D, Lea RH, Rosolowich V, Shapiro HM, et al. Breast Cancer, Pregnancy, and breastfeeding. J Obstet Gynaecol Can 2002;24: 164–80.
26. Thordarson G, Slusher N, Leong H, Ochoa D, Rajkumar L, Guzman R, et al. Insulin-like growth factor (IGF)-1 obliterates the pregnancy-associated protection against mammary carcinogenesis in rats: evidence that IGF-1 enhances cancer progression through estrogen receptor-alpha activation via the mitogen-activated protein kinase pathway. Breast Cancer Res 2004;6:R423–36.
27. Loibl S, von Minckwitz G, Gwyn K, Ellis P, Blohmer JU, Schlegelberger B, et al. Breast carcinoma during pregnancy. International recommendations from an expert meeting. Cancer 2006;106:237–46.