Abstract
Background:
Breastfeeding is associated with decreased breast cancer risk, yet associations with prognosis and survival by tumor subtype are largely unknown.
Methods:
We conducted a cohort study of 1636 women from two prospective breast cancer cohorts. Intrinsic tumor subtype (luminal A, luminal B, human epidermal growth factor receptor 2 [HER2]–enriched, basal-like) was determined by the PAM50 gene expression assay. Breastfeeding history was obtained from participant questionnaires. Questionnaires and medical record reviews documented 383 recurrences and 290 breast cancer deaths during a median follow-up of nine years. Multinomial logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) between breastfeeding and tumor subtype. Cox regression was used to estimate hazard ratios (HRs) for breast cancer recurrence or death. Statistical significance tests were two-sided.
Results:
Breast cancer patients with basal-like tumors were less likely to have previously breastfed than those with luminal A tumors (OR = 0.56, 95% CI = 0.39 to 0.80). Among all patients, ever breastfeeding was associated with decreased risk of recurrence (HR = 0.70, 95% CI = 0.53 to 0.93), especially breastfeeding for six months or more (HR = 0.63, 95% CI = 0.46 to 0.87, P trend = .01). Similar associations were observed for breast cancer death. Among women with luminal A subtype, ever breastfeeding was associated with decreased risks of recurrence (HR = 0.52, 95% CI = 0.31 to 0.89) and breast cancer death (HR = 0.52, 95% CI = 0.29 to 0.93), yet no statistically significant associations were observed among the other subtypes. Effects appeared to be limited to tumors with lower expression of proliferation genes.
Conclusions:
History of breastfeeding might affect prognosis and survival by establishing a luminal tumor environment with lower proliferative activity.
Numerous epidemiologic studies have reported associations between breastfeeding and breast cancer risk (1–9). Breastfeeding is hypothesized to decrease breast cancer risk by reducing a woman’s lifetime number of menstrual cycles, and thus her cumulative exposure to endogenous hormones (10), and increasing the differentiation of ductal cells, making them less susceptible to carcinogens (11). Additionally, breastfeeding may facilitate excretion of carcinogens via the milk ducts (12,13). History of breastfeeding could also have a beneficial effect on breast cancer prognosis and survival and might be an important prognostic factor to consider when evaluating a newly diagnosed patient. However, studies to date have reported mixed results for mortality (14–16), and to our knowledge no studies have examined recurrence.
Breast cancer is a heterogeneous disease with distinct molecular subtypes that appear to have different risk factor profiles, require different treatment regimens, and have considerable variability in prognosis. For example, history and duration of breastfeeding have been found to be associated with decreased risk of triple-negative or basal-like breast cancer, although not with ER+ or luminal A breast cancer (2,4,6,7,9,17).
Recently, the PAM50 gene expression assay has been used to subtype breast tumors into luminal A, luminal B, human epidermal growth factor receptor 2 (HER2)–enriched, basal-like, and normal-like intrinsic subtypes (18). The assay has been shown to add valuable clinical information beyond standard clinical and pathological factors (estrogen receptor [ER], progesterone receptor [PR], HER2) used for risk assessment and therapy decision-making in breast cancer (19,20).
Our study aims were to examine the associations of lifetime breastfeeding with breast cancer intrinsic subtype, as defined by the PAM50 assay, and the associations of lifetime breastfeeding with breast cancer recurrence and mortality overall and by subtype.
Methods
Study Population
The study population consisted of women previously diagnosed with breast cancer from the Life After Cancer Epidemiology (LACE) (21) and Pathways (22) prospective cohort studies of breast cancer survivors (Supplementary Table 1, available online). LACE participants were age 18 to 79 years when diagnosed with early-stage breast cancer between 1997 and 2000 (American Joint Committee on Cancer [AJCC] stage I with tumor size ≥1cm, stage II, or stage IIIA), primarily at Kaiser Permanente Northern California (KPNC) (83%) and the University of Utah (12%). At the time of study recruitment from 2000 to 2002, patients were within 39 months of diagnosis (mean = 23 months, 61% between 12 and 24 months), had completed chemotherapy or radiotherapy, and had no prior five-year history of breast cancer or other cancer.
The Pathways Study recruited women diagnosed with AJCC stage I to IV breast cancer from 2006 to 2013 at KPNC. The women had no previous diagnosis of other invasive cancer, were at least 21 years of age at diagnosis, and spoke English, Spanish, or Chinese. Most women were approached for enrollment within two months of diagnosis (mean = 1.8 months, minimum 0.3 months, maximum 7.2 months).
Clinicopathologic characteristics at the time of diagnosis, including disease stage, tumor size, nodal status, grade, estrogen receptor status, progesterone receptor status, and HER2 overexpression in the primary tumor, were obtained from the cancer registry and medical records. Treatment information (surgery, chemotherapy, radiotherapy, and hormonal therapy) and history of comorbidity were also collected from these data sources.
Participants provided informed consent under study protocols approved by institutional review boards at KPNC and the University of Utah.
Breastfeeding and Covariates
Breastfeeding and other relevant covariate information was collected at study enrollment using mailed (LACE) or in-person (Pathways) questionnaires. Demographic information included age at breast cancer diagnosis, race/ethnicity, and education. Breastfeeding data consisted of ever/never and cumulative duration over all births in months. While information on breastfeeding was from self-report, recall of breastfeeding history has been shown to be accurate and reliable (23,24). Other self-reported covariate information included menopausal status, parity, smoking, and family history of breast cancer.
Sampling Strategy
A total of 2135 LACE and 2172 Pathways women were potentially eligible for the PAM50 assay. A stratified case-cohort study design was used to select eligible women (25), with strata defined by clinical subtype based on immunohistochemistry (IHC) results for ER, PR, and HER2 (26). The case-cohort design is an efficient alternative to the nested case-control design in studies examining multiple outcomes (eg, recurrence and survival) (25). The subcohort consisted of a random sample of women with the most common IHC subtype (ER+ or PR+, HER2-) (sampling fraction = 18%), and all women with the remaining less common subtypes having worse prognosis (ER+ or PR+, HER2+; ER-, PR-, HER2-; and ER-, PR-, HER2+) (sampling fraction = 100%). Then, all those not selected for the subcohort but with outcomes of interest during follow-up were also included. Out of 2087 women selected for the case-cohort, 1691 had tumor tissue successfully assayed by the PAM50 (Supplementary Figure 1, available online).
Tissue Samples
For those selected into the case-cohort study, we contacted the hospital where the primary surgery was performed or the institution’s pathology storage facility to obtain formalin-fixed, paraffin-embedded (FFPE) tissue blocks and corresponding slides from that procedure. Slides were reviewed by one pathologist (REF), who identified and marked an area of representative tumor tissue on a slide. If the area of invasive tumor was smaller than 0.5cm in diameter, the case was classified as ineligible. Tissue punches 1mm in diameter were obtained from an area of the FFPE tissue block corresponding to the marked slide.
Clinical Tissue Markers
ER, PR, and HER2 expression were obtained from medical record review and the KPNC Cancer Registry (KPNC case patients) or Utah Cancer Registry (Utah case patients). For all breast surgical specimens at KPNC, ER, PR, and HER2 status were determined by IHC at the KPNC regional IHC lab; at the University of Utah, they were determined by hospital pathology departments or ARUP Laboratories, Inc. (Salt Lake City, UT). ER and PR were considered positive when 1% or more of tumor cells demonstrated nuclear staining (27). HER2 expression was considered positive with greater than 30% strong staining (28).
Gene Expression Assay and PAM50 Intrinsic Subtypes
The tissue punch was deparaffinized and digested for RNA extraction as described previously (19). The RNA was reverse transcribed, and quantitative polymerase chain reaction (RT-qPCR) was performed for 55 genes (PAM50 plus housekeepers) (18). Details of the PAM50 RT-qPCR methods have been provided elsewhere (19,29). Laboratory personnel (IJS) were blinded to patient and clinical information. Each batch of tissue samples included a mix of clinical phenotypes.
To determine intrinsic subtypes from the gene expression data, we applied centroid-based algorithms to the calibrated log-expression ratio for the 50 genes. For each sample, this process generated five continuous-scale normalized subtype scores representing degree of Spearman correlation of gene expression with that of prototype luminal A, luminal B, HER2-enriched, basal-like, and normal-like breast tumors (18). For each case patient, we assigned a subtype based on the highest intrinsic subtype score. Additionally, expression of 10 cell cycle regulation genes was averaged into a cell proliferation value (CENPF, ANLN, CDC20, CCNB1, CEP55, MYBL2, MKI67, UBE2C, RRM2, KIF2C).
We previously reported that the agreement between PAM50 subtypes (luminal A, luminal B, HER2-enriched, basal-like) and corresponding IHC subtypes across the four groups was kappa = 0.49. Sensitivity (0.83) and specificity (0.96) were highest for basal-like and triple-negative subtypes (30).
Outcomes
Outcomes were ascertained by self-report and additionally at KPNC by regular linkage to electronic medical records and mortality files. All outcomes were then verified by medical record review. Cause of death was determined from death certificates and supplemented by medical records as necessary.
Primary analytic outcomes were first breast cancer recurrence/metastasis and breast cancer–specific death. After excluding women with normal-like subtype (n = 53) because of insufficient power for analyses and missing breastfeeding data (n = 2), the final sample size was n = 1636 (Supplementary Figure 1, available online). Follow-up continued through April 2014. A total of 383 women had a recurrence, and 551 died of any cause, with 290 (52.6%) from breast cancer.
Statistical Analysis
All analyses incorporated the stratified sampling design for unbiased estimation of population parameters and valid estimates of standard errors. The ‘svy’ commands in Stata software (StataCorp, College Station, TX) use sampling weights to estimate frequency distributions of baseline characteristics.
A multinomial logistic regression model was used to estimate associations of breastfeeding with likelihood of having a PAM50 subtype. This approach is similar to the case-case analysis method widely used for dichotomous tumor characteristics (31) but extended to the four subtype categories via the multinomial model. Treating the most prevalent subtype, luminal A, as the base comparator outcome, we estimated odds ratios (ORs) and 95% confidence intervals (CIs) associated with the breastfeeding categories for each of the non-luminal A subtypes, adjusting for age at diagnosis, race/ethnicity, stage, and parity.
Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals for the associations of each PAM50 subtype with recurrence and breast cancer–specific mortality. Established prognostic factors and potential confounders were selected a priori and were included in the final multivariable models if one or more of the hazard ratios changed by 10% or more (32). Thus, models were adjusted for age at diagnosis, race/ethnicity, stage, parity, chemotherapy, radiotherapy, hormonal therapy, breast cancer surgery, and PAM50 subtype as specified in Table 1. Time since diagnosis was the time scale used in the regression models, allowing for delayed entry into the cohort (ie, left truncation, with study entry ranging from 0 to 3.2 years postdiagnosis). Assumptions of proportionality were verified with variable by time interactions. Point and interval estimation of regression parameters accounted for the case-cohort study design with stratified sampling of the subcohort using the methods of Borgan, Langholz, et al. (33), as implemented in SAS subroutines developed by Langholz and Jiao (34). P interaction effects were generated using the Wald test. Tests of statistical significance were two-sided. Statistical significance was considered P values of less than or equal to .05.
Table 1.
Demographic and clinical characteristics by PAM50 tumor subtype*
Characteristic | Luminal A, % | Luminal B, % | Basal-like, % | HER2-E, % | Total, % | P† |
---|---|---|---|---|---|---|
(n = 591) | (n = 365) | (n = 321) | (n = 359) | (n = 1636) | ||
Age at breast cancer diagnosis, y | <.001 | |||||
<50 | 17.8 | 29.9 | 36.8 | 21.7 | 23.0 | |
50–64 | 46.3 | 44.3 | 44.3 | 49.2 | 46.0 | |
≥65 | 35.9 | 25.8 | 19.0 | 29.1 | 31.0 | |
mean | 60.5 | 57.4 | 54.5 | 58.9 | 58.9 | |
Race/ethnicity | <.001 | |||||
White | 75.1 | 74.0 | 62.0 | 70.2 | 72.7 | |
African American | 4.8 | 3.6 | 19.7 | 6.4 | 6.4 | |
Hispanic | 7.5 | 9.8 | 10.9 | 9.9 | 8.7 | |
Asian | 10.6 | 10.3 | 4.7 | 9.2 | 9.7 | |
Other | 1.9 | 2.3 | 2.6 | 4.4 | 2.5 | |
Education | .44 | |||||
High school or less | 20.5 | 18.8 | 20.6 | 24.6 | 20.8 | |
Some college | 40.6 | 35.9 | 45.7 | 35.1 | 39.4 | |
College graduate | 20.7 | 24.0 | 15.6 | 19.8 | 20.7 | |
Postgraduate | 18.2 | 21.3 | 18.1 | 20.5 | 19.1 | |
Smoking history at breast cancer diagnosis | .06 | |||||
Never | 53.7 | 57.0 | 52.3 | 46.5 | 53.2 | |
Past | 41.0 | 39.1 | 40.6 | 42.5 | 40.8 | |
Current | 5.3 | 3.9 | 7.1 | 10.9 | 6.0 | |
Menopausal status at breast cancer diagnosis | <.001 | |||||
Postmenopausal | 73.4 | 63.9 | 54.4 | 69.9 | 68.7 | |
Premenopausal | 19.7 | 32.0 | 36.8 | 23.2 | 24.7 | |
Unknown | 6.9 | 4.1 | 8.8 | 6.8 | 6.5 | |
AJCC stage | <.001 | |||||
I | 57.0 | 36.8 | 43.0 | 38.7 | 48.5 | |
II | 38.0 | 49.7 | 50.5 | 53.1 | 44.0 | |
III | 4.9 | 12.6 | 5.3 | 7.1 | 6.9 | |
IV | 0.1 | 0.9 | 1.2 | 1.1 | 0.6 | |
Breast cancer surgery | .12 | |||||
None | 0.5 | 1.5 | 0.6 | 0.7 | 0.7 | |
Lumpectomy | 59.0 | 52.9 | 61.4 | 49.3 | 56.6 | |
Mastectomy | 40.5 | 45.6 | 38.0 | 50.0 | 42.7 | |
Chemotherapy | <.001 | |||||
No | 64.9 | 38.4 | 19.7 | 28.7 | 48.90 | |
Yes | 35.1 | 61.6 | 80.3 | 71.3 | 51.1 | |
Radiotherapy | .05 | |||||
No | 51.0 | 58.7 | 48.0 | 59.3 | 53.6 | |
Yes | 49.0 | 41.3 | 52.0 | 40.7 | 46.4 | |
Hormonal therapy | <.001 | |||||
No | 17.6 | 11.8 | 95.2 | 45.4 | 29.0 | |
Yes | 82.4 | 4.8 | 4.8 | 54.6 | 71.0 | |
Any comorbidity (Charlson comorbidity index) | .82 | |||||
No | 86.2 | 87.1 | 84.8 | 88.2 | 86.5 | |
Yes | 13.8 | 12.9 | 15.2 | 11.8 | 13.5 | |
Tumor subtype by immunohistochemistry | <.001 | |||||
ER+PR+HER2- | 93.4 | 76.3 | 9.7 | 34.6 | 71.9 | |
ER+PR+HER2+ | 5.7 | 19.4 | 1.2 | 24.1 | 10.8 | |
ER-PR-HER2- | 0.7 | 2.9 | 83.3 | 18.3 | 12.8 | |
ER-PR-HER2+ | 0.2 | 1.5 | 5.9 | 23.0 | 4.4 | |
Parity‡ and breastfeeding | .002 | |||||
Nulliparous | 15.6 | 25.8 | 17.0 | 13.9 | 17.7 | |
Parous, no | 26.7 | 18.4 | 35.4 | 28.4 | 26.1 | |
Parous, <6 mo BF | 22.5 | 19.4 | 17.4 | 25.6 | 21.7 | |
Parous, ≥6 mo BF | 35.2 | 36.4 | 30.2 | 32.1 | 34.5 |
* Percentages are weighted because of stratified case-cohort study design with strata defined as immunohistochemistry clinical subtype. AJCC = American Joint Committee on Cancer; BF = breastfeeding; ER = estrogen receptor; HER2 = human epidermal growth factor receptor 2; PR = progesterone receptor.
† From two-sided Pearson-chi-square test.
‡ Parous is at least one live or stillbirth (range 1–10); nulliparous is no live or stillbirth.
Results
In the case-cohort sample, demographic and clinical characteristics differed by PAM50 subtype (Table 1). Women with luminal A subtype were more likely to be older at breast cancer diagnosis (35.9% ≥65 years) compared with luminal B (25.8%), basal-like (19.0%), and HER2-enriched (29.1%) subtypes. Luminal A subtypes were also more commonly white (75.1%), whereas basal-like subtypes were more likely African American (19.7%) and Hispanic (10.9%). The proportion of ever breastfed was similar across luminal A (57.7%), luminal B (55.8%), and HER2-enriched (57.7%) subtypes, but lower among the basal-like subtype (47.6%).
Luminal A subtypes were most likely to be diagnosed at an early stage (AJCC stage I or II, 95.0%), followed by basal-like (93.5%), HER2-enriched (91.8%), and luminal B (86.5%) subtypes (Table 1). Women with basal-like (80.3%) and HER2-enriched (71.3%) subtypes were more likely to have chemotherapy compared with women with the luminal A (35.1%) and luminal B (61.6%) subtypes. As expected, most women with luminal A (82.4%) and luminal B (88.2%) subtypes were treated with hormonal therapy compared with the HER2-enriched (54.6%) and Basal-like (4.8%) subtypes.
By multinomial logistic regression adjusted for age at diagnosis, race/ethnicity, stage, and parity, women with the basal-like subtype were less likely to have a history of breastfeeding compared with women with the luminal A subtype (OR = 0.56, 95% CI = 0.39 to 0.80) (Table 2). However, women with the luminal B subtype (OR = 1.08, 95% CI = 0.74 to 1.56) and HER2-enriched subtype (OR = 0.94, 95% CI = 0.64 to 1.39) had no difference in breastfeeding history compared with those with the luminal A subtype. Furthermore, when restricted to parous women, all associations were similar (data not shown).
Table 2.
Association of breastfeeding with PAM50 tumor subtype*
Total
No. |
Luminal A | Luminal B | Basal-like | HER2-E | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|
% | % | OR | (95% CI) | % | OR | (95% CI) | % | OR | (95% CI) | ||
Breastfeeding | |||||||||||
Never | 754 | 42.3 | 44.2 | Ref. | 52.4 | Ref. | 42.2 | Ref. | |||
Ever | 882 | 57.7 | 55.8 | 1.08 | (0.74 to 1.56) | 47.6 | 0.56 | (0.39 to 0.80) | 57.8 | 0.94 | (0.64 to 1.39) |
* From multinomial logistic regression with comparison group = Luminal A, adjusted for age at diagnosis, race/ethnicity, American Joint Committee on Cancer tumor stage, and parity. CI = confidence interval; OR = odds ratio.
In Cox models adjusted for age at diagnosis, race/ethnicity, stage, parity, chemotherapy, radiotherapy, hormonal therapy, surgery, and PAM50 subtype, ever breastfeeding was associated with a reduced risk of breast cancer recurrence compared with never breastfeeding (HR = 0.70, 95% CI = 0.53 to 0.93), and it was slightly more protective among those who breastfed six months or more compared with those who never breastfed (HR = 0.63, 95% CI = 0.46 to 0.87, P trend = .01) (Table 3). Similar patterns of association were observed for breast cancer–specific mortality (ever breastfeeding: HR = 0.72, 95% = 0.53 to 0.98; ≥6 months breastfeeding: HR = 0.61, 95% CI = 0.43 to 0.88, P trend = .01).
Table 3.
Association of breastfeeding with breast cancer recurrence and death*
Recurrence† | No. events‡ | HR | (95% CI) |
---|---|---|---|
Breastfeeding | |||
Never | 190 | Ref. | |
Ever | 193 | 0.70 | (0.53 to 0.93) |
Lifetime breastfeeding | |||
Never | 192 | Ref. | |
<6 m | 77 | 0.81 | (0.58 to 1.14) |
≥6 mo | 116 | 0.63 | (0.46 to 0.87) |
P trend | .01 | ||
Breast cancer mortality† | |||
Breastfeeding | |||
Never | 144 | Ref. | |
Ever | 146 | 0.72 | (0.53 to 0.98) |
Lifetime breastfeeding | |||
Never | 144 | Ref. | |
<6 mo | 62 | 0.90 | (0.61 to 1.32) |
≥6 mo | 84 | 0.61 | (0.43 to 0.88) |
P trend | .01 |
* n = 27 women had missing data on parity, surgery, chemotherapy, and/or radiotherapy and were thus excluded from all models. CI = confidence interval; HR = hazard ratio.
† Adjusted for age at diagnosis, race/ethnicity, stage, chemotherapy, radiotherapy, hormonal therapy, surgery type, PAM50 subtype, and parity.
‡ Not population based given case-cohort study design.
When stratified by PAM50 subtype (Table 4), ever breastfeeding was associated with decreased risks of recurrence (HR = 0.52, 95% CI = 0.31 to 0.89) and breast cancer–specific mortality (HR = 0.52, 95% CI = 0.29 to 0.93), among those with the luminal A subtype. Breastfeeding six months or more compared with never breastfeeding again conferred slightly greater risk reductions for all outcomes in the luminal A group: recurrence (HR = 0.43, 95% CI = 0.23 to 0.79) and breast cancer–specific mortality (HR = 0.42, 95% CI = 0.22 to 0.83). While no statistically significant associations were observed in the luminal B or basal-like groups for ever breastfeeding and duration of breastfeeding, the hazard ratios were of similar magnitude and suggestive of reduced risks of these outcomes. However, breastfeeding was not associated with a reduced risk of recurrence among women with HER2-E tumors. The P values for interactions of breastfeeding with subtype were not statistically significant.
Table 4.
Association of breastfeeding with breast cancer recurrence and death, stratified by PAM50 intrinsic subtype*
Luminal A | Luminal B | Basal-like | HER2-E | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Recurrence† | No. events‡ | HR | (95% CI) | No. events‡ | HR | (95% CI) | No. events‡ | HR | (95% CI) | No. events‡ | HR | (95% CI) |
Breastfeeding | ||||||||||||
Never | 67 | Ref. | 61 | Ref. | 33 | Ref. | 29 | Ref. | ||||
Ever | 65 | 0.52 | (0.31 to 0.89) | 62 | 0.60 | (0.26 to 1.41) | 24 | 0.64 | (0.24 to 1.72) | 42 | 1.29 | (0.44 to 3.82) |
P interaction = .19 | ||||||||||||
Lifetime breastfeeding | ||||||||||||
Never | 67 | Ref. | 61 | Ref. | 35 | Ref. | 29 | Ref. | ||||
<6 mo | 29 | 0.67 | (0.34 to 1.30) | 21 | 0.70 | (0.24 to 2.00) | 12 | 1.34 | (0.41 to 4.37) | 15 | 1.01 | (0.26 to 3.91) |
≥6 mo | 36 | 0.43 | (0.23 to 0.79) | 41 | 0.55 | (0.21 to 1.44) | 12 | 0.39 | (0.14 to 1.07) | 27 | 1.58 | (0.51 to 4.92) |
P trend | .03 | .15 | .04 | .54 | ||||||||
P interaction = .11 | ||||||||||||
Breast cancer mortality† | ||||||||||||
Breastfeeding | ||||||||||||
Never | 53 | Ref. | 42 | Ref. | 28 | Ref. | 21 | Ref. | ||||
Ever | 47 | 0.52 | (0.29 to 0.93) | 49 | 0.61 | (0.24 to 1.56) | 20 | 0.70 | (0.18 to 2.79) | 30 | 1.29 | (0.35 to 4.77) |
P interaction = .21 | ||||||||||||
Lifetime breastfeeding | ||||||||||||
Never | 53 | Ref. | 42 | Ref. | 28 | Ref. | 21 | Ref. | ||||
<6 mo | 22 | 0.68 | (0.33 to 1.43) | 18 | 0.78 | (0.25 to 2.37) | 9 | 1.22 | (0.26 to 5.72) | 13 | 1.23 | (0.28 to 5.39) |
≥6 mo | 25 | 0.42 | (0.22 to 0.83) | 31 | 0.53 | (0.18 to 1.54) | 11 | 0.49 | (0.13 to 1.90) | 17 | 1.35 | (0.29 to 6.40) |
P trend | .08 | .12 | .17 | .92 | ||||||||
P interaction = .39 |
* n = 27 women had missing data on parity, surgery, chemotherapy, and/or radiotherapy and were thus excluded from all models.
† Adjusted for age at diagnosis, race/ethnicity, stage, chemotherapy, radiotherapy, hormonal therapy, surgery type, and parity.
‡ Not population based given case-cohort study design.
To explore the potential role of tumor proliferation in the breastfeeding associations, we examined breastfeeding and risks of recurrence and mortality stratified by the median split of the PAM50 gene expression level of proliferation (Table 5). Among women with lower (below median) proliferation gene expression, ever breastfeeding compared with never breastfeeding was associated with a decreased risk of breast cancer recurrence (HR = 0.58, 95% CI = 0.37 to 0.93) and a suggestive decreased risk of breast cancer–specific mortality (HR = 0.65, 95% CI = 0.39 to 1.09). No statistically significant associations were observed among women with higher (median or above) proliferation gene expression. Furthermore, breastfeeding six months or more conferred the greatest reductions in risk in the below median, but not above median, proliferation group. However, P interaction values were not statistically significant. We then further adjusted the main subtype models shown in Table 4 for proliferation gene expression, and risk reductions for all outcomes became even more protective (Supplementary Table 2, available online).
Table 5.
Association of breastfeeding with breast cancer recurrence and death, stratified by median expression level of proliferation*
<Median proliferation expression | ≥Median proliferation expression | |||||
---|---|---|---|---|---|---|
Recurrence† | No. events‡ | HR | (95% CI) | No. events‡ | HR | (95% CI) |
Breastfeeding | ||||||
Never | 93 | Ref. | 97 | Ref. | ||
Ever | 87 | 0.58 | (0.37 to 0.93) | 106 | 0.75 | (0.46 to 1.24) |
P interaction = .47 | ||||||
Lifetime breastfeeding | ||||||
Never | 93 | Ref. | 97 | Ref. | ||
<6 mo | 38 | 0.71 | (0.41 to 1.23) | 39 | 0.81 | (0.43 to 1.53) |
≥6 mo | 49 | 0.50 | (0.29 to 0.86) | 67 | 0.72 | (0.40 to 1.27) |
P trend | .03 | .22 | ||||
P interaction = .70 | ||||||
Breast cancer mortality† | ||||||
Breastfeeding | ||||||
Never | 63 | Ref. | 81 | Ref. | ||
Ever | 65 | 0.65 | (0.39 to 1.09) | 81 | 0.71 | (0.40 to 1.25) |
P interaction = .82 | ||||||
Lifetime breastfeeding | ||||||
Never | 63 | Ref. | 81 | Ref. | ||
<6 mo | 32 | 0.85 | (0.45 to 1.59) | 30 | 0.81 | (0.40 to 1.63) |
≥6 mo | 33 | 0.52 | (0.29 to 0.95) | 51 | 0.66 | (0.35 to 1.26) |
P trend | .08 | .20 | ||||
P interaction = .80 |
* n = 29 women had missing data on breastfeeding, parity, surgery, chemotherapy, and/or radiotherapy and were thus excluded from all models. CI = confidence interval; HR = hazard ratio.
† Adjusted for age at diagnosis, race/ethnicity, stage, chemotherapy, radiotherapy, hormonal therapy, surgery type, and parity.
‡ Not population based given case-cohort study design.
Discussion
In this prospective study of breast cancer prognosis, we found that women with basal-like tumors were less likely to have breastfed before breast cancer diagnosis, regardless of duration, than women with luminal A tumors at diagnosis, which is consistent with previous study findings (2,4,6,9,35). Furthermore, any prior breastfeeding was associated with a 30% decreased risk of recurrence, and breastfeeding at least six months or more conferred a slightly larger reduction in risk. Any prior breastfeeding was also associated with decreased risks of breast cancer –specific death. The protective associations were statistically significant among women with luminal A tumors, and while women with luminal B and basal-like tumors also had protective associations, they were not statistically significant. In contrast, among HER2-enriched tumors, the patterns of association were suggestive of increased risk.
To our knowledge, this is the first study to examine the influence of breastfeeding on breast cancer recurrence and by tumor subtype. Two prior studies reported no association between breastfeeding and mortality (14,16), and one study reported an association between shorter (≤12 months), but not longer, duration of breastfeeding and reduced mortality (15). Possible reasons for these inconsistencies include differences in patient populations by menopausal status, race/ethnicity, and limited sample size. In Trivers et al. (15), the study population consisted of 1264 women diagnosed before age 55 years (78% premenopausal), and limited statistical power could have explained the lack of observing a possible dose-response relationship with longer breastfeeding. However, in the two null studies, the populations included 3107 (14) and 2640 (16) older, exclusively white, postmenopausal women.
Our findings support the biologic mechanism of how breastfeeding years before breast cancer diagnosis may be associated with a better prognostic tumor subtype. During pregnancy, there is proliferative expansion of progenitor cells in the breasts, and subsequent breastfeeding leads to progressive loss of this expanded cell population through terminal differentiation of epithelial cells in the lobules and an accompanying reduction in the proliferative activity of the mammary epithelium (36,37). Thus, possible malignant transformation of these differentiated progenitor cells could lead to the development of a more differentiated (ER+ and/or PR+), rather than an undifferentiated (ER-/HER2-), breast cancer, which is consistent with our findings. In complement, lack of breastfeeding may preserve the progenitor cells in an undifferentiated state within the breast lobules, and malignant transformation of these progenitor cells could lead to the development of an undifferentiated (ER-/HER2-), as opposed to differentiated (ER+ and/or PR+), breast cancer (37). In fact, one study found that elevated expression of GABApi, which is coexpressed with progenitor cell genes within breast lobules, was associated with the ER-/HER2- phenotype (P < .0001) and shorter lifetime duration of breastfeeding (≤6 months) in parous women (P = .013) (38).
Regarding how prior breastfeeding might improve prognosis and survival in women with luminal A tumors, we conjecture that breastfeeding could drive terminal differentiation of breast ductal cells via increased production of the GATA family of transcription factors, specifically GATA-3, which has been shown to actively maintain the differentiated luminal epithelium (39). Additionally, it was reported in a study of 166 invasive breast cancers with 10 years of follow-up that GATA-3 makes ER+ tumors more responsive to anti-estrogen therapy and thus leads to reduced risks of recurrence and death (40). Consistent with our hypothesis, we observed that the protective effect of breastfeeding on breast cancer outcomes was possibly limited to women whose tumors had lower (below median) proliferation gene expression levels. When we further adjusted for proliferation levels in the luminal A group, the associations became more protective.
Interestingly, we also observed a suggestive association of breastfeeding with improved prognosis and survival among women with basal-like tumors, which usually have high proliferation and a worse prognosis. However, a larger cohort of women with basal-like breast cancer would be needed to potentially confirm this finding, as we had limited statistical power to examine this rarer subtype. Furthermore, our observation might be explained by the fact that LACE women were enrolled on average two years postdiagnosis. Thus, those basal-like tumors with better prognosis (ie, less proliferative) were more likely to be enrolled in the study, whereas those basal-like tumors with worse prognosis were more likely to be not enrolled.
In conclusion, breastfeeding was associated with better prognosis and survival among breast cancer patients, particularly those diagnosed with luminal A tumors. While better outcomes were also possible among those diagnosed with luminal B and basal-like tumors, the associations were not statistically significant. Further replication studies are needed to understand the underlying biological mechanisms by intrinsic subtype. However, it is both intriguing and plausible that prior breastfeeding leads to increased differentiation of breast luminal cells via GATA-3 expression, which can predetermine a luminal A tumor with better prognosis and improved response to anti-estrogen therapy. Our findings not only lend further support to the established benefits of breastfeeding (41,42) but also provide new insight into the mechanistic intricacies of lactation on breast cancer prognosis.
Funding
This work was supported by the US National Institutes of Health awards R01 CA129059 (B. J. Caan, PI) and R01 CA105274 (L. H. Kushi, PI). Additional support was provided by the Bioinformatics and Biostatistics core resources of the Huntsman Cancer Institute (P30 CA042014). The Utah Cancer Registry is funded by Contract No. HHSN261201000026C from the National Cancer Institute’s Surveillance, Epidemiology, and End Results Program, along with additional support from the Utah State Department of Health and the University of Utah.
Supplementary Material
We thank Ms. Carole Davis at the University of Utah for laboratory assay support.
PSB has an interest in Bioclassifier LLC and University Genomics. The other authors declare that they have no conflict of interest.
The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health.
References
- 1. Kwan ML, Kushi LH, Weltzien E, et al. Epidemiology of breast cancer subtypes in two prospective cohort studies of breast cancer survivors. Breast Cancer Res. 2009;11(3):R31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Li CI, Beaber EF, Tang MT, et al. Reproductive factors and risk of estrogen receptor positive, triple-negative, and HER2-neu overexpressing breast cancer among women 20–44 years of age. Breast Cancer Res Treat. 2013;137(2):579–587. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Ma H, Bernstein L, Pike MC, et al. Reproductive factors and breast cancer risk according to joint estrogen and progesterone receptor status: a meta-analysis of epidemiological studies. Breast Cancer Res. 2006;8(4):R43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Millikan RC, Newman B, Tse CK, et al. Epidemiology of basal-like breast cancer. Breast Cancer Res Treat. 2008;109(1):123–139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Phillips KA, Milne RL, Friedlander ML, et al. Prognosis of premenopausal breast cancer and childbirth prior to diagnosis. J Clin Oncol. 2004;22(4):699–705. [DOI] [PubMed] [Google Scholar]
- 6. Phipps AI, Malone KE, Porter PL, et al. Reproductive and hormonal risk factors for postmenopausal luminal, HER-2-overexpressing, and triple-negative breast cancer. Cancer. 2008;113(7):1521–1526. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Phipps AI, Chlebowski RT, Prentice R, et al. Reproductive history and oral contraceptive use in relation to risk of triple-negative breast cancer. J Natl Cancer Inst. 2011;103(6):470–477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Work ME, John EM, Andrulis IL, et al. Reproductive risk factors and oestrogen/progesterone receptor-negative breast cancer in the Breast Cancer Family Registry. Br J Cancer. 2014;110(5):1367–1377. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Anderson KN, Schwab RB, Martinez ME. Reproductive risk factors and breast cancer subtypes: a review of the literature. Breast Cancer Res Treat. 2014;144(1):1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Colditz GA, Baer HJ, Tamimi RM. Breast Cancer. In: Schottenfeld D, Fraumeni JF, (eds). Cancer Epidemiology and Prevention. New York City: Oxford University Press; 2006. [Google Scholar]
- 11. Russo J, Russo IH. Molecular Basis of Breast Cancer: Prevention and Treatment. Berlin, Heidelberg, New York: Springer-Verlag; 2004. [Google Scholar]
- 12. Russo J, Russo IH. Differentiation and breast cancer. Medicina. 1997;57(Suppl 2):81–91. [PubMed] [Google Scholar]
- 13. Lipworth L, Bailey LR, Trichopoulos D. History of breast-feeding in relation to breast cancer risk: a review of the epidemiologic literature. J Natl Cancer Inst. 2000;92(4):302–312. [DOI] [PubMed] [Google Scholar]
- 14. Phillips KA, Milne RL, West DW, et al. Prediagnosis reproductive factors and all-cause mortality for women with breast cancer in the breast cancer family registry. Cancer Epidemiol Biomarkers Prev. 2009;18(6):1792–1797. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Trivers KF, Gammon MD, Abrahamson PE, et al. Association between reproductive factors and breast cancer survival in younger women. Breast Cancer Res Treat. 2007;103(1):93–102. [DOI] [PubMed] [Google Scholar]
- 16. Alsaker MD, Opdahl S, Asvold BO, et al. The association of reproductive factors and breastfeeding with long term survival from breast cancer. Breast Cancer Res Treat. 2011;130(1):175–182. [DOI] [PubMed] [Google Scholar]
- 17. Palmer JR, Viscidi E, Troester MA, et al. Parity, Lactation, and Breast Cancer Subtypes in African American Women: Results from the AMBER Consortium. J Natl Cancer Inst. 2014;106(10): In press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Parker JS, Mullins M, Cheang MC, et al. Supervised risk predictor of breast cancer based on intrinsic subtypes. J Clin Oncol. 2009;27(8):1160–1167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bastien RR, Rodriguez-Lescure A, Ebbert MT, et al. PAM50 breast cancer subtyping by RT-qPCR and concordance with standard clinical molecular markers. BMC Med Genomics. 2012;5:44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Caan BJ, Sweeney C, Habel LA, et al. Intrinsic subtypes from the PAM50 gene expression assay in a population-based breast cancer survivor cohort: Prognostication of short and long term outcomes. Cancer Epidemiol Biomarkers Prev. 2014;23(5):725–734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Caan B, Sternfeld B, Gunderson E, et al. Life After Cancer Epidemiology (LACE) Study: a cohort of early stage breast cancer survivors (United States). Cancer Causes Control. 2005;16(5):545–556. [DOI] [PubMed] [Google Scholar]
- 22. Kwan ML, Ambrosone CB, Lee MM, et al. The Pathways Study: a prospective study of breast cancer survivorship within Kaiser Permanente Northern California. Cancer Causes Control. 2008;19(10):1065–1076. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal recall of breastfeeding practice. Nutr Rev. 2005;63(4):103–110. [DOI] [PubMed] [Google Scholar]
- 24. Natland ST, Andersen LF, Nilsen TI, et al. Maternal recall of breastfeeding duration twenty years after delivery. BMC Med Res Methodol. 2012;12:179. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Wacholder S. Practical considerations in choosing between the case-cohort and nested case-control designs. Epidemiology. 1991;2(2):155–158. [DOI] [PubMed] [Google Scholar]
- 26. Carey LA, Perou CM, Livasy CA, et al. Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study. JAMA. 2006;295(21):2492–2502. [DOI] [PubMed] [Google Scholar]
- 27. Hammond ME, Hayes DF, Dowsett M, et al. American Society of Clinical Oncology/College Of American Pathologists guideline recommendations for immunohistochemical testing of estrogen and progesterone receptors in breast cancer. J Clin Oncol. 2010;28(16):2784–2795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Wolff AC, Hammond ME, Schwartz JN, et al. American Society of Clinical Oncology/College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer. J Clin Oncol. 2007;25(1):118–145. [DOI] [PubMed] [Google Scholar]
- 29. Ebbert MT, Bastien RR, Boucher KM, et al. Characterization of uncertainty in the classification of multivariate assays: application to PAM50 centroid-based genomic predictors for breast cancer treatment plans. J Clin Bioinforma. 2011;1:37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Sweeney C, Bernard P, Factor RE, et al. Intrinsic subtypes from PAM50 gene expression assay in a population-based breast cancer cohort: Differences by age, race, and tumor characteristics. Cancer Epidemiol Biomarkers Prev. 2014;23(5)714–724. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Begg CB, Zhang ZF. Statistical analysis of molecular epidemiology studies employing case-series. Cancer Epidemiol Biomarkers Prev. 1994;3(2):173–175. [PubMed] [Google Scholar]
- 32. Mickey RM, Greenland S. The impact of confounder selection criteria on effect estimation. Am J Epidemiol. 1989;129(1):125–137. [DOI] [PubMed] [Google Scholar]
- 33. Borgan O, Langholz B, Samuelsen SO, et al. Exposure stratified case-cohort designs. Lifetime Data Anal. 2000;6(1):39–58. [DOI] [PubMed] [Google Scholar]
- 34. Langholz B, Jiao J. Computational methods for case-cohort studies. Computational Stat Data Anal. 2007;51:3737–3748. [Google Scholar]
- 35. Gaudet MM, Press MF, Haile RW, et al. Risk factors by molecular subtypes of breast cancer across a population-based study of women 56 years or younger. Breast Cancer Res Treat. 2011;130(2):587–597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Bocker W, Moll R, Poremba C, et al. Common adult stem cells in the human breast give rise to glandular and myoepithelial cell lineages: a new cell biological concept. Lab Invest. 2002;82(6):737–746. [DOI] [PubMed] [Google Scholar]
- 37. Russo J, Russo IH. Toward a physiological approach to breast cancer prevention. Cancer Epidemiol Biomarkers Prev. 1994;3(4):353–364. [PubMed] [Google Scholar]
- 38. Symmans WF, Fiterman DJ, Anderson SK, et al. A single-gene biomarker identifies breast cancers associated with immature cell type and short duration of prior breastfeeding. Endocr Relat Cancer. 2005;12(4):1059–1069. [DOI] [PubMed] [Google Scholar]
- 39. Kouros-Mehr H, Slorach EM, Sternlicht MD, et al. GATA-3 maintains the differentiation of the luminal cell fate in the mammary gland. Cell. 2006;127(5):1041–1055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Ciocca V, Daskalakis C, Ciocca RM, et al. The significance of GATA3 expression in breast cancer: a 10-year follow-up study. Human Pathol. 2009;40(4):489–495. [DOI] [PubMed] [Google Scholar]
- 41. Breastfeeding and the use of human milk. Pediatrics. 2012;129(3):e827– e841. [DOI] [PubMed] [Google Scholar]
- 42. Horta BL, Victora CG. Long-term effects of breastfeeding: a systematic review. In: Publications of the World Health Organization: World Health Organization; 2013. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.