Established breast cancer risk factors and risk of intrinsic tumor subtypes

Biochimica et Biophysica Acta 1856 (2015) 73–85

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbacan

Review

Established breast cancer risk factors and risk of intrinsic tumor subtypes Mollie E. Barnard ⁎, Caroline E. Boeke, Rulla M. Tamimi Department of Epidemiology, Harvard T.H. Chan School of Public Health, 677 Huntington Avenue, Boston, MA 02115, USA Channing Division of Network Medicine, Brigham and Women's Hospital, 181 Longwood Avenue, Boston, MA 02115, USA

a r t i c l e

i n f o

Article history: Received 20 January 2015 Received in revised form 7 June 2015 Accepted 8 June 2015 Available online 10 June 2015 Keywords: Breast cancer Subtype Risk factor Luminal HER2 Triple negative

a b s t r a c t Breast cancer is a heterogeneous disease with multiple intrinsic tumor subtypes. These subtypes vary in tumor gene expression and phenotype, and are most commonly grouped into four major subtypes: luminal A, luminal B, HER2-overexpressing and triple negative (or basal-like). A growing number of studies have evaluated the relationship between established breast cancer risk factors and risk of one or more intrinsic tumor subtypes. We conducted a systematic review of 38 studies to synthesize their results and identify areas requiring more research. Taken together, published studies suggest that most established breast cancer risk factors reflect risk factors for the luminal A subtype of breast cancer, and some breast cancer risk factors may be differentially associated with other intrinsic tumor subtypes. Future breast cancer research will need to consider etiologic differences across subtypes and design studies focused on understanding the etiology and prevention of less common tumor subtypes. © 2015 Elsevier B.V. All rights reserved.

Contents 1. 2. 3.

Introduction . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . 3.1. Age at menarche . . . . . . . . . 3.2. Parity . . . . . . . . . . . . . . 3.3. Age at first birth . . . . . . . . . . 3.4. Lifetime duration of breastfeeding . . 3.5. Age at menopause . . . . . . . . . 3.6. BMI . . . . . . . . . . . . . . . 3.7. Family history of breast cancer . . . 3.8. Alcohol use . . . . . . . . . . . . 3.9. Use of oral contraceptives . . . . . 3.10. Use of menopausal hormone therapy 3.11. Summary . . . . . . . . . . . . 4. Discussion . . . . . . . . . . . . . . . . Transparency Documents . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . Appendix A. Supplementary data . . . . . . References . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

73 74 75 75 75 75 75 75 75 75 75 75 75 75 75 84 84 84 84

1. Introduction Abbreviations: ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2; CK, cytokeratin; OC, oral contraceptive; HT, postmenopausal hormone therapy ⁎ Corresponding author at: Department of Epidemiology, Harvard T.H. Chan School of Public Health, 677 Huntington Avenue, Boston, MA 02115, USA. E-mail address: [email protected] (M.E. Barnard).

http://dx.doi.org/10.1016/j.bbcan.2015.06.002 0304-419X/© 2015 Elsevier B.V. All rights reserved.

Breast cancer is the second most common cancer in women, with more than 1.5 million incident cases diagnosed globally each year [1]. Advances in breast cancer screening and treatment have decreased breast cancer mortality for luminal and HER2-overexpressing cancers;

74

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

however, triple negative cancers continue to have high mortality rates [2]. Understanding breast cancer heterogeneity and improving subtype-specific prevention, diagnosis and survival are current priorities in both clinical and epidemiologic research [2,3]. For more than fifty years, histopathology has recognized breast cancer as a heterogeneous disease with both inter- and intra-tumor variability [4]. More recently, advances in molecular analysis and genetic sequencing have further described the diversity of breast tumors using a combination of microarray analyses and next-generation DNA sequencing [5,6]. Current methods of breast cancer classification group tumors into genetically- and molecularly-defined intrinsic subtypes including, but not limited to: luminal A, luminal B, HER2overexpressing, and triple negative cancers [3]. Surrogate definitions of these subtypes are typically based on the results of immunohistochemical (IHC) staining for estrogen, progesterone, and HER2 receptors as well as confirmation of HER2 status using fluorescent in situ hybridization. More recent tumor classification systems also consider markers of tumor proliferation and aggressiveness, such as histologic grade or Ki-67 status, to differentiate between luminal A and luminal B cancers [3]. It is well established that breast cancer subtypes have unique prognoses and differ in their responsiveness to chemoprevention and chemotherapy [2,3]. The distinct natural history of each subtype suggests that breast cancer subtypes may also have unique risk profiles. Elucidating which established breast cancer risk factors are pervasive across all subtypes and which are subtype-specific could be important to understanding tumor etiology and essential to developing prevention strategies. More specifically, if established breast cancer risk factors reflect only the risk factors and etiology of luminal A cancer, the most common subtype, alternate or additional strategies may be needed to reduce the risk of luminal B, HER2-overexpressing or triple negative cancer. A growing number of small- to mid-size studies have started to explore whether breast cancer subtypes have unique risk profiles; however, the comparisons presented in these studies are highly variable and difficult to synthesize. We conducted a systematic review to illustrate what is currently known about the associations between established breast cancer risk factors and risk of specific tumor subtypes. 2. Methods On January 22, 2014 we conducted a PubMed literature search to identify original, peer-reviewed research related to breast cancer risk factors and risk of intrinsic tumor subtypes. We considered all types of established risk factors, including family history of breast cancer, reproductive factors, lifestyle factors, and exogenous exposures [7–10]. Our final search included the following terms: “breast cancer” AND (“luminal” OR “basal-like” OR “HER2” OR “triple-negative”) AND (“menarche” OR “menopause” OR “menopausal” OR “age at first birth” OR “parity” OR “nulliparous” OR “parous” OR “breastfeeding” OR “lactation” OR “hormone replacement therapy” OR “HRT” OR “postmenopausal hormone” OR “hormone therapy” OR “estrogen therapy” OR “oral contraceptive” OR “oral contraceptives” OR “family history” OR “benign breast disease” OR “BMI” OR “weight” OR “obese” OR “obesity” OR “alcohol” OR “mammographic density” OR “breast density”). We eliminated by title all duplicate articles and any publications that clearly were not on the topic of breast cancer, were not original research, were conducted in animal models, focused on cancer clinical outcomes, or addressed topics in pharmacology, genetics, or epigenetics. Two independent reviewers (MB and CB) considered the content of the remaining articles and selected for inclusion every article that studied one or more established breast cancer risk factors as an exposure, identified at least one breast cancer subtype as an outcome, adjusted for confounding by age or menopausal status, and included at least 200 cases with known subtypes. Our initial search resulted in 708 articles. Four hundred and ninety-four articles were eliminated by title, and another 176 were eliminated after assessing

the content of each paper. The remaining 38 original research articles are included in this review. The studies included in this review are presented in Supplementary Table 1. They include nine cohort studies [11–19], twelve populationbased case–control studies [20–31], four hospital-based case–control studies [32–35], twelve case-only studies [36–47], and one pooled analysis that included cohort studies, case–control studies and studies of mixed design [48]. The majority of studies were conducted using unique case populations, and all studies with similar or identical case populations are identified. For each study, cases with known ER, PR, and HER2 status were regarded as cases with known subtypes irrespective of whether their designated subtypes were based solely on ER/PR/HER2 status or also based on expression of auxiliary tumor markers such as HER1 or CK5/ 6 (see table footnotes for study specific definitions of subtypes). Briefly, for the majority of studies, cases described as ER + and/or PR + and HER2− were labeled luminal A and cases described as ER and/or PR+ and HER2+ were labeled luminal B. For a small subset of studies, additional information on tumor proliferation was incorporated to differentiate between luminal A (low grade or Ki-67 absent or low) and luminal B (high grade or Ki-67 positive) subtypes [18,38,43]. Cases described as ER−, PR− and HER2+ were labeled HER2-overexpressing, and cases described as ER −, PR − and HER2 − were labeled triple negative when information on HER1 and CK5/6 status was not available, basallike when positive for HER1 or CK5/6, or unclassified when negative for both HER1 and CK5/6. Classifying tumors by ER/PR/HER2 status acknowledged that, while the exact biomarkers used to define each intrinsic subtype have varied over time, the subtype names and the tumors they encompass have been relatively consistent [3,6]. We methodically summarized the associations between established breast cancer risk factors and risk of each tumor subtype for every risk factor considered in five or more studies. These risk factors include age at menarche [14,18,20,21,25–30,33,35,39,41,42,46, 48], parity [14,15,18,20,21,23,25–30,33,35–37,39,41,42,46–48], age at first birth [14,15,18,20,21,23–30,33,35,37,39,42,46,48], breastfeeding duration [14,18,20,21,23,25–27,29,30,33,35–37,39, 41,42,46], age at menopause [14,18,21,24,28,33,39,41,42,46], BMI [13,19,20,22,26–30,34,35,37–41,43–45,48], family history of breast cancer [11,16,18,24,26–28,30,32,33,35,37,39,41,46], alcohol use [17,18,27,29,35,37], use of oral contraceptives [14,23,26,27,30, 37,39,42], and use of menopausal hormone therapy [12,18,21,30, 31,37,39,42,44]. Measures of association reflecting the most extreme exposure comparisons from each study were outlined in risk factor-specific tables. For example, when a study compared nulliparous women to parous women with 1, 2, 3, or 4 + children, we presented the association comparing women with 4 or more children to nulliparous women. Risk factors considered in ten or more studies with non-case reference groups were summarized in the main tables, while risk factors considered in more than five but fewer than nine studies with non-case reference groups were included in the supplementary tables. Estimates from studies with the most informative comparisons (e.g. case–control and cohort studies that compared women with breast cancer to those without breast cancer) were used to create summary descriptions of the results. Associations were classified as “consistent” when the direction of association was consistent across studies and at least three studies reported a statistically significant association. We described associations as “probable” when the direction of association was consistent across most studies and at least two studies reported a statistically significant association. We indicated a “possible” association when most studies reported results that were either consistent in direction or null and one or two studies reported a statistically significant association. Associations were classified as “null or inconsistent” when fewer than two significant results were reported and any additional evidence was either sparse or inconsistent in direction.

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

3. Results 3.1. Age at menarche Older age at menarche was consistently associated with a moderately decreased risk of triple negative breast cancer and showed a probable association with reduced risk of luminal A cancer (Table 1). Odds ratios from two case–control studies suggested a possible relationship between older age at menarche and a lower risk of luminal B cancer [26,33], but data from two additional case–control studies and one cohort study supported a null association [18,28,29]. The relationship between age at menarche and risk of HER2-overexpressing breast cancer was ambiguous, likely due to the small number of HER2-overexpressing cases included in each study.

75

between BMI and risk of luminal A, luminal B or triple negative breast cancer, though it is important to note that the absence of an association between BMI and risk of luminal A cancer may be an artifact of study design. Many studies used luminal A breast cancers as the referent group, making it impossible to estimate an association between post-menopausal BMI and risk of luminal A breast cancer. 3.7. Family history of breast cancer Family history of breast cancer, especially in one or more firstdegree relatives, was a strong risk factor for risk of luminal A, luminal B, HER2-overexpressing and triple negative breast cancers (Supplementary Table 3). Across all subtypes, the hazard of breast cancer among those with a family history was consistently one and a half to two times the hazard of breast cancer among those without a family history.

3.2. Parity 3.8. Alcohol use Compared to nulliparity, greater parity was associated with a substantially lower risk of luminal A breast cancer; the odds of luminal A breast cancer in women with multiple lifetime pregnancies was consistently reported as roughly half the odds of luminal A breast cancer in nulliparous women (Table 2). In contrast, many of the studies with the largest number of triple negative cases and with sound control for confounding suggested that parity may be associated with an increased risk of triple negative tumors [14,18,20,33]. No clear patterns were evident in the reported associations between parity and risk of luminal B or HER2-overexpressing breast cancer. 3.3. Age at first birth The association between greater parity and decreased risk of luminal A breast cancer was strongest among women with younger compared to older age at first birth (Table 3). The effects of age at first birth on risk of luminal B, HER2-overexpressing and triple negative breast cancer were unclear. 3.4. Lifetime duration of breastfeeding Longer duration of breastfeeding was consistently associated with decreased risk of luminal A, luminal B and triple negative breast cancers (Table 4). There was no clear pattern in how duration of breastfeeding related to risk of HER2-overexpressing breast cancer. 3.5. Age at menopause Four studies with non-case reference groups considered the association between age at menopause and risk of luminal A cancer (Table 5). All four observed an increased risk of luminal A cancer with older age at menopause [18,24,28,33]. There was also evidence of a possible positive association between older age at menopause and triple negative cancer, though more research is needed to fully understand the magnitude of the relationship. There were no obvious patterns in the reported associations between age at menopause and risk of either luminal B or HER2overexpressing breast cancer.

We observed a possible association of drinking alcohol and increased risk of HER2-overexpressing breast cancer but saw no obvious association with risk of triple negative cancer. Two studies with non-case reference groups addressed the associations between alcohol intake and risk of luminal breast cancers (Supplementary Table 4). Trivers et al. reported an increased odds (OR = 1.86) of luminal A breast cancer among those who had more than seven drinks per week compared to those who did not drink [29]. Tamimi et al. observed a weaker association (OR = 1.3) in the same direction as well as a statistically significant trend in increasing risk of luminal A breast cancer with increased alcohol intake [18]. Neither study showed a significant association between alcohol use and risk of the luminal B subtype. 3.9. Use of oral contraceptives Use of oral contraceptives was consistently associated with an increased risk of triple negative breast cancer (Supplementary Table 5). However, for the luminal A subtype, two out of the three studies with non-case reference groups showed a decreased risk of breast cancer with use of oral contraceptives [26,30]. There was insufficient research on luminal B and HER2-overexpressing cancers to understand the relationships between oral contraceptive use and risk of these subtypes. 3.10. Use of menopausal hormone therapy Use of menopausal hormone therapy was consistently associated with a dramatic increase in risk of luminal A breast cancer (Supplementary Table 6). Reported hazard ratios were as high as 2.07 for fifteen years of use compared to never use [12]. The two studies that compared luminal B breast cancer cases to non-cases suggested that use of menopausal hormone therapy might also be associated with an increased risk of luminal B breast cancer [12,18], though more research is needed. We were unable to elucidate any information about the relationship between menopausal hormone use and risk of HER2-overexpressing or triple negative breast cancers. The studies that estimated these associations considered relatively few relevant cases, and were limited in power [18,21].

3.6. BMI 3.11. Summary Among pre-menopausal women, higher BMI showed a possible association with decreased risk of luminal A breast cancer and a possible association with increased risk of triple negative cancer (Table 6, Supplementary Table 2). The associations between BMI and risk of luminal-B cancer were inconsistent, as were the associations between BMI and risk of HER2-overexpressing breast cancer. In post-menopausal women, the association between BMI and risk of HER2-overexpressing breast cancer appeared null (Table 7, Supplementary Table 2). No clear patterns of association emerged

The results for each subtype have been summarized in Table 8. 4. Discussion This systematic review presents evidence that established breast cancer risk factors are differentially associated with each breast cancer subtype. Family history of breast cancer was the only established risk factor consistently associated with an increased risk of all

76

Table 1 Age at menarche and risk of breast cancer subtypes. First author year Cohort and case–control Comparison: older v. younger Dolle 2009 Phipps 2011b Tamimi 2012

Li 2013 Razzaghi 2013

Yang 2007

Comparison: younger v. older Trivers 2009 Millikan 2008

Xing 2010 Phipps 2008a Islam 2012 Case only Comparison: older v. younger Redondo 2012 Martinez 2013 Comparison: younger v. older Devi 2012 Millikan 2008

Turkoz 2013 Yang 2011

Exposure comparison (age in years)

Luminal A

Odds ratio (ref: controls) Hazard ratio (non-cases) Hazard ratio (non-cases)

≥15 v. b12⁎ ≥14 v. b12⁎ ≥14 v. b12⁎

– – 0.7 (0.5, 0.9)⁎⁎

– – 1.0 (0.6, 1.6)

– – 1.1 (0.5, 2.3)

Odds ratio (ref: controls) Odds ratio (ref: controls)

≥14 v. b12⁎ ≥13 v. b13

– 0.76 (0.53, 1.09)

– –

0.5 (0.2, 1.4) –

Odds ratio (ref: controls)

Per two-year increase

0.87 (0.75, 1.01)

0.60 (0.42, 0.85)

0.82 (0.63, 1.07)

Odds ratio (ref: controls)

Per two-year increase

0.99 (0.90, 1.08)

0.98 (0.75, 1.28)

1.14 (0.86, 1.50)

Odds ratio (ref: controls)

b12 v. ≥12

1.09 (0.84, 1.42)

0.93 (0.50, 1.73)

1.63 (0.92, 2.88)

Odds ratio (ref: controls)

b13 v. ≥13

1.1 (0.9, 1.3)

–

–

Odds ratio (ref: controls)

b13 v. ≥13

2.35 (1.45, 3.81)

1.61 (0.70, 3.71)

2.21 (0.90, 5.45)

1.55 (0.79, 3.03)

Odds ratio (ref: controls) Odds ratio (ref: controls)

b13 v. ≥13 ≤12 v. ≥15⁎

– –

– –

2.7 (1.4, 5.5) 0.63 (0.31, 1.28)

1.1 (0.7, 1.7) 2.04 (0.95, 4.35)

Luminal A (287), luminal B (58), HER2-overexpressing (21), triple negative (39) Luminal A (571), triple negative (159)

Odds ratio (ref: luminal A)

≥15 v. ≤13

(ref)

–

–

0.65 (0.21, 1.68)

Odds ratio (ref: luminal A)

≥13 v. b13

(ref)

–

–

1.27 (0.88, 1.82)

Luminal A (493), luminal B (125), HER2-overexpressing (117), triple negative (299) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)c, unclassified (150)d Luminal A (1249), luminal B (234), HER2-overexpressing (169), triple negative (232) Luminal A (9534), luminal B (1657), HER2-overexpressing (953), triple negative (1997)

Odds ratio (ref: luminal A)

b13 v. ≥13

(ref)

1.11 (0.71, 1.75)

0.81 (0.50, 1.30)

0.81 (0.57, 1.14)

Odds ratio (ref: luminal A)

b13 v. ≥13

(ref)

0.9 (0.7, 1.4)

1.0 (0.7, 1.5)

Odds ratio (ref: all other cases)

b12 v. ≥12

1.09 (0.88, 1.36)

0.97 (0.71, 1.34)

0.85 (0.57, 1.25)

0.99 (0.72, 1.36)

Odds ratio (ref: luminal A)

≤12 v. ≥15

(ref)

1.01 (0.84, 1.22)

0.85 (0.67, 1.07)

1.08 (0.92, 1.28)

Triple negative (187) Triple negative (307) Luminal A (1267)a, luminal B (321)b, HER2-overexpressing (113), basal-like (226)c, unclassified (95)d HER2-overexpressing (60), triple negative (184) Luminal A (181), luminal B (17), HER2-overexpressing (29), basal-like (48)c, unclassified (41)d Luminal A (455), luminal B (72), HER2-overexpressing (117), triple negative (246) Luminal A (552), luminal B (48), HER2-overexpressing (61), basal-like (95)e, unclassified (48)f Luminal A (272), luminal B (36), HER2-overexpressing (33), triple negative (135) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)c, unclassified (150)d Luminal A (722), luminal B (214), HER2-overexpressing (154), triple negative (327) HER2-overexpressing (39), triple negative (78) HER2-overexpressing (84), triple negative (68)

⁎ Includes test of trend across exposure categories. ⁎⁎ p-Value for test of trend was statistically significant at α = 0.05. a Luminal A tumors with histologic grade 1 or 2. b Luminal B tumors with histologic grade 3. c Triple negative tumor positive for at least one of CK5/6 or HER1. d Triple negative tumor negative for CK5/6 and HER1. e Triple negative tumor positive for CK5 and/or HER1. f Triple negative tumor negative for CK5 and HER1.

Measure of association

Luminal B

HER2

Triple negative Basal-like

Unclassified

0.4 (0.2, 1.0) 0.96 (0.67, 1.39) 0.8 (0.5, 1.3) 0.8 (0.4, 1.7)

0.7 (0.5, 1.2) 0.37 (0.19, 0.70)

–

0.98 (0.81, 1.18) 0.78 (0.68, 0.89)

0.92 (0.72, 1.18)

1.60 (1.17, 2.19) –

1.4 (1.1, 1.9)

1.3 (0.9, 1.7)

1.5 (1.0, 2.1)

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Gaudet 2011

Cases with subtyping (n=)

Table 2 Parity and risk of breast cancer subtypes. First author year Cohort and case–control Comparison: parous v. nulliparous Dolle 2009 Ma 2010

Islam 2012 Millikan 2008

Tamimi 2012

Xing 2010

Yang 2007

Li 2013 Phipps 2008a Razzaghi 2013

Comparison: nulliparous v. parous Phipps 2011b Gaudet 2011

Triple negative (187) Luminal A (645), luminal B (120), HER2-overexpressing (97), triple negative (335) Luminal A (272), luminal B (36), HER2-overexpressing (33), triple negative (135) HER2-overexpressing (84), triple negative (68) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)a, unclassified (150)b Luminal A (1267)c, luminal B (321)d, HER2-overexpressing (113), basal-like (226)a, unclassified (95)b Luminal A (722), luminal B (214), HER2-overexpressing (154), triple negative (327) Luminal A (552), luminal B (48), HER2-overexpressing (61), basal-like (95)e, unclassified (48)f HER2-overexpressing (60), triple negative (184) HER2-overexpressing (39), triple negative (78) Luminal A (181), luminal B (17), HER2-overexpressing (29), basal-like (48)a, unclassified (41)b

Triple negative (307) Luminal A (455), luminal B (72), HER2-overexpressing (117), triple negative (246)

Measure of association

Exposure comparison

Luminal A

Luminal B

HER2

Triple negative Basal-like

Unclassified

Odds ratio (ref: controls) Odds ratio (ref: controls)

≥4 v. nulliparous⁎ ≥4 v. nulliparous⁎

– 0.55 (0.38, 0.79)⁎⁎

– 0.55 (0.26, 1.20)⁎⁎

– 0.47 (0.19, 1.15)⁎⁎

0.6 (0.2, 1.9) 1.00 (0.60, 1.67)

Odds ratio (ref: controls)

≥4 v. nulliparous

0.28 (0.17, 0.46)

0.53 (0.20, 1.41)

0.90 (0.25, 3.23)

0.62 (0.36, 1.07)

Odds ratio (ref: controls)

≥3 v. nulliparous⁎

–

–

1.15 (0.49, 2.71)

0.62 (0.27, 1.47)

Odds ratio (ref: controls)

≥3 v. nulliparous⁎

0.7 (0.5, 0.9)

–

–

1.9 (1.1, 3.3)⁎⁎

–

Hazard ratio (ref: non-cases)

≥3 v. nulliparous⁎

0.7 (0.5, 1.0)⁎⁎

1.1 (0.5, 2.6)⁎⁎

0.7 (0.2, 2.3)

1.1 (0.4, 2.9)⁎⁎

1.4 (0.3, 6.0)

Odds ratio (ref: controls)

≥3 v. nulliparous

1.64 (0.74, 3.62)

1.51 (0.55, 4.12)

2.07 (0.59, 7.33)

3.02 (0.99, 9.17)

Odds ratio (ref: controls)

≥2 v. nulliparous

0.42 (0.21, 0.84)

0.56 (0.07, 4.56)

1.15 (0.17, 7.63)

1.80 (0.37, 8.85) 0.71 (0.09, 5.71)

Odds ratio (ref: controls)

Parous v. nulliparous

–

–

1.1 (0.6, 2.2)

0.7 (0.5, 1.0)

Odds ratio (ref: controls)

Parous v. nulliparous

–

–

1.2 (0.4, 3.9)

1.1 (0.5, 2.3)

Parous v. nulliparous (age b26) Parous v. nulliparous (age ≥26)

0.93 (0.64, 1.34)

–

–

2.07 (1.04, 4.15)

–

0.96 (0.63, 1.47)

–

–

0.43 (0.18, 1.06)

–

Nulliparous v. ≥1 Nulliparous v. parous

– 3.02 (1.47, 6.23)

– 4.11 (0.80, 21.07)

– 2.21 (0.60, 8.16)

Odds ratio (ref: controls)

Hazard ratio (ref: non-cases) Odds ratio (ref: controls)

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Trivers 2009

Cases with subtyping (n=)

0.61 (0.37, 0.97) 1.08 (0.41, 2.81)

(continued on next page)

77

78

Table 2 (continued) First author year Phipps 2011c

Case only Comparison: parous v. nulliparous Devi 2012

Millikan 2008

Redondo 2012

Shinde 2010 Turkoz 2013

Pilewskie 2012

Comparison: nulliparous v. parous Martinez 2013 Yang 2011

HER2-overexpressing (288), triple negative (645)

Luminal A (493), luminal B (125), HER2-overexpressing (117), triple negative (299) Luminal A (1868), luminal B (294), HER2-overexpressing (94), triple negative (288) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)a, unclassified (150)b Luminal A (287), luminal B (58), HER2-overexpressing (21), triple negative (39) Luminal A (1579), triple negative (468) Luminal A (1249), luminal B (234), HER2-overexpressing (169), triple negative (232) Triple negative (47)

Luminal A (571), triple negative (159) Luminal A (9534), luminal B (1657), HER2-overexpressing (953), triple negative (1997)

⁎ Includes test of trend across exposure categories. ⁎⁎ p-Value for test of trend was statistically significant at α = 0.05. a Triple negative tumor positive for at least one of CK5/6 or HER1. b Triple negative tumor negative for CK5/6 and HER1. c Luminal A tumors with histologic grade 1 or 2. d Luminal B tumors with histologic grade 3. e Triple negative tumor positive for CK5 and/or HER1. f Triple negative tumor negative for CK5 and HER1.

Measure of association

Exposure comparison

Luminal A

Luminal B

HER2

Triple negative Basal-like

Unclassified

Hazard ratio (ref: non-cases)

Nulliparous v. parous

–

–

1.15 (0.84, 1.58)

1.07 (0.87, 1.33)

Odds ratio (ref: luminal A)

≥3 v. nulliparous

(ref)

0.98 (0.57, 1.65)

1.97 (1.07, 3.63)

1.29 (0.87, 1.91)

Odds ratio (ref: luminal A)

≥3 v. nulliparous

(ref)

1.03 (0.70, 1.50)

1.82 (0.94, 3.53)

1.18 (0.81, 1.72)

Odds ratio (ref: luminal A)

≥3 v. nulliparous

(ref)

1.3 (0.7, 2.3)

1.0 (0.5, 1.7)

Odds ratio (ref: luminal A)

≥3 v. nulliparous

(ref)

–

–

1.84 (0.56, 7.23)

Odds ratio (ref: non-triple negative cases) Odds ratio (ref: all other cases)

≥3 v. nulliparous

–

–

–

2.76 (1.86, 4.08)

≥2 v. nulliparous

0.59 (0.38, 0.71)

0.66 (0.45, 0.97)

1.46 (0.85, 2.50)

1.37 (0.87, 2.15)

Parous b5 years ago v. nulliparous Parous 5–15 years ago v. nulliparous

–

–

–

1.7 (0.7, 3.9)

–

–

–

1.8 (0.8, 4.1)

(ref) (ref)

– 1.00 (0.85, 1.18)

0.98 (0.81, 1.20)

0.71 (0.32, 1.57) 0.69 (0.59, 0.81)

Odds ratio (ref: non-triple negative)

Odds ratio (ref: luminal A) Odds ratio (ref: luminal A)

Nulliparous v. 1–2 Nulliparous v. parous

1.7 (1.0, 2.9)

1.7 (1.0, 3.0)

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Kwan 2009

Cases with subtyping (n=)

Table 3 Age at first birth and risk of breast cancer subtypes. First author year Cohort and case–control Comparison: older v. younger Ma 2010 Phipps 2011b Phipps 2008a Xing 2010 Islam 2012 Phipps 2011c Li 2013 Gaudet 2011

Tamimi 2012

Comparison: age group at first birth v. nulliparous Dolle 2009 Millikan 2008

Razzaghi 2013 Trivers 2009

Case only Comparison: older v. younger Martinez 2013 Redondo 2012 Turkoz 2013 Yang 2011 Comparison: age group at first birth v. nulliparous Kwan 2009 Millikan 2008

Luminal A (645), luminal B (120), HER2-overexpressing (97), triple negative (335) Triple negative (307) HER2-overexpressing (39), triple negative (78) Luminal A (722), luminal B (214), HER2-overexpressing (154), triple negative (327) HER2-overexpressing (84), triple negative (68) HER2-overexpressing (288), triple negative (645) HER2-overexpressing (60), triple negative (184) Luminal A (455), luminal B (72), HER2-overexpressing (117), triple negative (246) Luminal A (651), triple negative (72) Luminal A (552), luminal B (48), HER2-overexpressing (61), basal-like (95)a, unclassified (48)b Luminal A (1267)c, luminal B (321)d, HER2-overexpressing (113), basal-like (226)e, unclassified (95)f

Triple negative (187) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)e, unclassified (150)f Luminal A (181), luminal B (17), HER2-overexpressing (29), basal-like (48)e, unclassified (41)f Luminal A (272), luminal B (36), HER2-overexpressing (33), triple negative (135)

Luminal A (571), triple negative (159) Luminal A (287), luminal B (58), HER2-overexpressing (21), triple negative (39) Luminal A (1249), luminal B (234), HER2-overexpressing (169), triple negative (232) Luminal A (9534), luminal B (1657), HER2-overexpressing (953), triple negative (1997)

Luminal A (1868), luminal B (294), HER2-overexpressing (94), triple negative (288) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)e, unclassified (150)f

Measure of association

Exposure comparison (age in years)

Luminal A

Odds ratio (ref: controls)

≥30 v. ≤19⁎

1.03 (0.67, 1.57)

0.78 (0.33, 1.85)

0.91 (0.35, 2.36)

1.32 (0.80, 2.17)

Hazard ratio (ref: non-cases) Odds ratio (ref: controls) Odds ratio (ref: controls)

≥30 v. b20⁎ ≥30 v. b20 ≥30 v. ≤24

– – 0.96 (0.67, 1.40)

– – 0.51 (0.24, 1.10)

– 0.7 (0.1, 4.4) 0.86 (0.39, 1.89)

1.05 (0.53, 2.06) 0.7 (0.2, 2.3) 0.74 (0.42, 1.29)

Odds ratio (ref: controls) Hazard ratio (ref: non-cases) Odds ratio (ref: controls) Odds ratio (ref: controls)

≥30 v. b25⁎ ≥30 v. b30 Per 5-year increase Per 5-year increase

– – – 1.16 (1.01, 1.33)

– – – 1.31 (0.95, 1.80)

0.99 (0.46, 2.11) 1.83 (1.31, 2.56) 0.85 (0.65, 1.10) 1.19 (0.93, 1.53)

0.67 (0.25, 1.82) 1.18 (0.93, 1.51) 0.77 (0.65, 0.91) 1.08 (0.90, 1.31)

Odds ratio (ref: controls) Odds ratio (ref: controls) Hazard ratio (ref: non-cases)

Per 5-year increase Per 5-year increase

1.15 (1.02, 1.28) 1.08 (0.95, 1.23)

– 1.04 (0.70, 1.55)

– 0.87 (0.60, 1.05)

0.72 (0.49, 1.06) 0.95 (0.71, 1.27) 1.06 (0.71, 1.58)

Per 1-year increase

1.018 (1.007, 1.030)

0.994 (0.964, 1.024)

1.024 (0.991, 1.059)

Odds ratio (ref: controls) Odds ratio (ref: controls)

≥30 v. nulliparous⁎ b26 v. nulliparous ≥26 v. nulliparous

– 0.7 (0.5, 0.9) 0.9 (0.6, 1.2)

– – –

– – –

Odds ratio (ref: controls)

b26 v. nulliparous ≥26 v. nulliparous b18 v. nulliparous ≥18 v. nulliparous

0.93 (0.64, 1.34) 0.96 (0.63, 1.47) 0.23 (0.12, 0.44) 0.75 (0.57, 1.00)

– – 0.32 (0.09, 1.13) 0.53 (0.30, 0.95)

– – 2.53 (0.84, 7.68) 1.85 (0.75, 4.55)

Odds ratio (ref: luminal A) Odds ratio (ref: luminal A)

≥25 v. b21 ≥28 v. ≤22

(ref) (ref)

– –

– –

0.61 (0.39, 0.95) 1.38 (0.48, 3.93)

Odds ratio (ref: all other cases) Odds ratio (ref: luminal A)

≥30 v. b30

1.14 (1.08, 1.30)

1.57 (1.08, 2.30)

0.96 (0.55, 1.69)

0.71 (0.41, 1.22)

Per 5-year increase

(ref)

0.99 (0.92, 1.07)

1.01 (0.92, 1.10)

0.89 (0.83, 0.95)

b26 v. nulliparous ≥26 v. nulliparous b26 v. nulliparous ≥26 v. nulliparous

(ref) (ref) (ref) (ref)

1.25 (0.88, 1.78) 1.04 (0.72, 1.52) 1.2 (0.7, 2.1) 1.6 (0.9, 2.8)

2.02 (1.07, 3.80) 0.86 (0.42, 1.79) 0.8 (0.5, 1.5) 0.7 (0.4, 1.3)

1.28 (0.90, 1.82) 0.93 (0.63, 1.38) 1.9 (1.2, 3.0) 1.5 (0.9, 2.5) 1.2 (0.7, 2.1) 1.5 (0.8, 2.7)

Odds ratio (ref: controls)

Odds ratio (ref: luminal A) Odds ratio (ref: luminal A)

Luminal B

HER2

Triple negative Basal-like

1.002 (0.968, 1.037)

Unclassified

1.006 (0.959, 1.056)

1.2 (0.5, 3.0)⁎⁎ 1.9 (1.2, 3.2) 1.5 (0.8, 2.8)

– –

2.07 (1.04, 4.15) – 0.43 (0.18, 1.06) – 0.67 (0.39, 1.17) 0.79 (0.55, 1.14)

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Pollan 2013 Yang 2007

Cases with subtyping (n=)

⁎ Includes test of trend across exposure categories. ⁎⁎ p-Value for test of trend was statistically significant at α = 0.05. a Triple negative tumor positive for CK5 and/or HER1. b Triple negative tumor negative for CK5 and HER1. c Luminal A tumors with histologic grade 1 or 2. d Luminal B tumors with histologic grade 3. e Triple negative tumor positive for at least one of CK5/6 or HER1. f Triple negative tumor negative for CK5/6 and HER1. 79

80

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Table 4 Breastfeeding and risk of breast cancer subtypes. First author year

Cases with subtyping (n=)

Cohort and case–control Ma 2010 Luminal A (645), luminal B (120), HER2-overexpressing (97), triple negative (335) Phipps 2011b Triple negative (307) Li 2013 Dolle 2009

HER2-overexpressing (60), triple negative (184) Triple negative (187)

Trivers 2009

Luminal A (272), luminal B (36), HER2-overexpressing (33), triple negative (135) Phipps 2008a HER2-overexpressing (39), triple negative (78) Islam 2012 HER2-overexpressing (84), triple negative (68) Gaudet 2011 Luminal A (455), luminal B (72), HER2-overexpressing (117), triple negative (246) Tamimi 2012 Luminal A (1267)a, luminal B (321)b, HER2-overexpressing (113), basal-like (226)c, unclassified (95)d Millikan 2008 Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)c, unclassified (150)d Razzaghi 2013 Luminal A (181), luminal B (17), HER2-overexpressing (29), basal-like (48)c, unclassified (41)d Xing 2010 Luminal A (722), luminal B (214), HER2-overexpressing (154), triple negative (327) Case only Martinez Luminal A (571), triple negative 2013 (159) Redondo 2012 Luminal A (287), luminal B (58), HER2-overexpressing (21), triple negative (39) Devi 2012 Luminal A (493), luminal B (125), HER2-overexpressing (117), triple negative (299) Kwan 2009 Luminal A (1868), luminal B (294), HER2-overexpressing (94), triple negative (288) Shinde 2010 Luminal A (1579), triple negative (468) Turkoz 2013

Millikan 2008

Luminal A (1249), luminal B (234), HER2-overexpressing (169), triple negative (232) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)c,unclassified (150)d

Measure of association

Exposure comparison

Odds ratio (ref: controls)

≥24 months v. never⁎

Hazard ratio (non-cases) Odds ratio (ref: controls) Odds ratio (ref: controls) Odds ratio (ref: controls)

N12 months v. never⁎ N12 months v. never⁎

Odds ratio (ref: controls) Odds ratio (ref: controls) Odds ratio (ref: controls) Hazard ratio (non-cases)

≥6 months v. never ≥6 months v. never⁎ Per 6-month increase in duration ≥4 months v. never

Odds ratio (ref: controls)

≥12 months v. never⁎ ≥12 months v. never

Luminal A

Luminal B

HER2

Triple negative Basal-like

0.55 (0.34, 0.88)⁎⁎ 0.91 (0.38, 2.20) 1.31 (0.53, 3.21)

Unclassified

0.80 (0.47, 1.36)⁎⁎

–

–

–

0.81 (0.53, 1.26)

–

–

1.1 (0.3, 3.6)

0.5 (0.3, 0.9)⁎⁎

–

–

–

1.0 (0.6, 1.7)

0.68 (0.48, 0.96)

0.12 (0.03, 0.56) 0.37 (0.13, 1.08)

0.53 (0.32, 0.85)

–

–

1.5 (0.7, 3.3)

0.5 (0.3, 0.9)

–

–

1.09 (0.60, 1.98)

0.91 (0.48, 1.73)

0.70 (0.48, 1.01) 0.85 (0.70, 1.04)

0.76 (0.64, 0.90)

0.94 (0.86, 1.02)

0.8 (0.7, 1.0)

0.8 (0.6, 1.1)

0.9 (0.6, 1.5)

0.6 (0.4, 0.9)

0.6 (0.4, 1.1)

≥4 months v. never⁎

0.9 (0.7, 1.1)

–

–

0.7 (0.4, 0.9)⁎⁎

–

Odds ratio (ref: controls)

≥4 months v. never

1.02 (0.67, 1.55)

–

–

0.50 (0.22, 1.16)

–

Odds ratio (ref: controls)

Ever v. never

0.44 (0.31, 0.63)

Odds ratio (ref: luminal A) Odds ratio (ref: luminal A)

N12 months v. never ≥7 months v. never

(ref)

–

–

2.14 (1.24, 3.68)

(ref)

–

–

0.25 (0.08, 0.68)

Odds ratio (ref: luminal A)

≥6 months v. never

(ref)

0.90 (0.44, 1.81) 1.84 (0.84, 3.40)

0.79 (0.47, 1.31)

Odds ratio (ref: luminal A)

≥4 months v. never

(ref)

0.86 (0.65, 1.14) 0.86 (0.54, 1.38)

0.78 (0.59, 1.03)

Odds ratio (ref: non-triple negative cases) Odds ratio (ref: all other cases)

N2 months/child v. never

–

Ever v. never

0.71 (0.56, 0.98)

Odds ratio (ref: luminal A)

≥4 months v. never

(ref)

0.54 (0.31, 0.95) 0.33 (0.18, 0.61)

–

–

0.69 (0.48, 0.96) 1.34 (0.81, 2.21)

1.0 (0.6, 1.5)

0.9 (0.6, 1.5)

0.48 (0.30, 0.77)

0.54 (0.40, 0.73)

1.26 (0.83, 1.92)

0.7 (0.5, 1.1)

1.1 (0.7, 1.7)

⁎ Includes test of trend across exposure categories. ⁎⁎ p-Value for test of trend was statistically significant at α = 0.05. a Luminal A tumors with histologic grade 1 or 2. b Luminal B tumors with histologic grade 3. c Triple negative tumor positive for at least one of CK5/6 or HER1. d Triple negative tumor negative for CK5/6 and HER1.

subtypes. Breastfeeding was associated with a decreased risk of most subtypes, though the relationship between breastfeeding and risk of HER2-overexpressing cancer was unclear. Reproductive factors and exogenous hormones varied in their associations with risk of luminal A, luminal B and triple negative cancer, and research on alcohol use and BMI also demonstrated possible variation across subtypes.

Established breast cancer risk factors closely reflected risk factors for luminal A breast cancer, the most common subtype. An elevated risk of luminal A cancer was seen among women with risk factors that conferred a greater number of lifetime hormone cycles, such as earlier age at menarche, not using oral contraceptives, fewer pregnancies, later age at menopause, and use of menopausal hormones. Additionally, risk of luminal A cancer was positively associated with family history

Table 5 Age at menopause and risk of breast cancer subtypes. First author year

Cases with subtyping (n=)

Case only Devi 2012

Luminal A (493), luminal B (125), HER2-overexpressing (117), triple negative (299) Luminal A (571), triple negative (159)

Martinez 2013 Redondo 2012 Luminal A (287), luminal B (58), HER2-overexpressing (21), triple negative (39) Turkoz 2013 Luminal A (1249), luminal B (234), HER2-overexpressing (169), triple negative (232) ⁎ Includes test of trend across exposure categories. a Triple negative tumor positive for CK5 and/or HER1. b Triple negative tumor negative for CK5 and HER1. c Luminal A tumors with histologic grade 1 or 2. d Luminal B tumors with histologic grade 3. e Triple negative tumor positive for at least one of CK5/6 or HER1. f Triple negative tumor negative for CK5/6 and HER1.

Luminal A

Luminal B

HER2

Triple negative

Odds ratio (ref: controls) Hazard ratio (ref: non-cases) Odds ratio (ref: controls)

≥55 v. b45⁎ ≥55 v. 45–54⁎ ≥55 v. ≤44

– – 1.76 (0.83, 3.75)

– – 0.64 (0.21, 1.90)

2.9 (0.8, 10.7) – 0.72 (0.24, 2.20)

1.2 (0.5, 3.0) 1.02 (0.68, 1.52) 2.24 (0.77, 6.54)

Odds ratio (ref: controls) Odds ratio (ref: controls)

Per 5-year increase Per 5-year increase

1.11 (0.98, 1.27) 1.13 (1.01, 1.28)

– 1.10 (0.78, 1.57)

1.09 (0.81, 1.46)

1.01 (0.68, 1.51) 1.02 (0.82, 1.28) 1.46 (0.99, 2.15)

Hazard ratio (ref: non-cases)

Per 1-year increase

Odds ratio (ref: luminal A)

N50 v. ≤50

(ref)

1.34 (0.72, 2.51)

1.75 (0.93, 3.28)

1.59 (1.01, 2.51)

Odds ratio (ref: luminal A)

≥50 v. b50

(ref)

–

–

0.59 (0.33, 1.04)

Odds ratio (ref: luminal A)

≥50 v. b50

(ref)

–

–

1.28 (0.63, 2.61)

Odds ratio (ref: all other cases)

≥55 v. b55

0.97 (0.70, 1.35)

0.95 (0.56, 1.61)

0.94 (0.79, 1.11)

1.04 (0.90, 1.21)

Basal-like

Unclassified

1.039 (1.021, 1.058) 1.067 (1.024, 1.111) 1.075 (1.006, 1.150) 1.012 (0.972, 1.053) 1.089 (1.006, 1.179)

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Cohort and case–control Phipps 2008a HER2-overexpressing (39), triple negative (78) Phipps 2011b Triple negative (307) Xing 2010 Luminal A (722), luminal B (214), HER2-overexpressing (154), triple negative (327) Pollan 2013 Luminal A (651), triple negative (72) Yang 2007 Luminal A (552), luminal B (48), HER2-overexpressing (61), basal-like (95)a, unclassified (48)b Tamimi 2012 Luminal A (1267)c, luminal B (321)d, HER2-overexpressing (113), basal-like (226)e, unclassified (95)f

Exposure comparison (age in years)

Measure of association

81

82

Table 6 Adult premenopausal BMI and risk of breast cancer subtypes. First author year

Cases with subtyping (n=)

Cohort and case–control Sueta 2012 Luminal A (455), luminal B (108), HER2-overexpressing (84), triple negative (68) Phipps HER2-overexpressing (308), triple negative (705) 2012⁎⁎⁎ Millikan 2008

Gaudet 2011

Case only Gillespie 2010 Chen 2013

Yang 2011⁎⁎⁎ Kwan 2009 Millikan 2008

Devi 2012 Kimura 2012 ⁎ ⁎⁎ ⁎⁎⁎ ⁎⁎⁎⁎ a b c d e f g h

Triple negative (418) Luminal A (1127)e, luminal B (227)f, HER2-overexpressing (119), triple negative (477) Luminal A (9534), luminal B (1657), HER2-overexpressing (953), triple negative (1997) Luminal A (1868), luminal B (294), HER2-overexpressing (94), triple negative (288) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)a, unclassified (150)b Luminal A (493), luminal B (125), HER2-overexpressing (117), triple negative (299) Luminal A (333)g, luminal B (85)h, HER2-overexpressing (43), triple negative (70)

Exposure comparison

Luminal A

Luminal B

HER2

Triple negative Basal-like

Unclassified

Odds ratio (ref: controls)

≥25 v. 18.5–21.9⁎

0.96 (0.55, 1.68)

0.17 (0.03, 1.03)

2.66 (0.13, 52.8)

–

Hazard ratio (ref: non-cases)

≥30 v. b25

–

–

0.94 (0.61, 1.44)

1.05 (0.81, 1.36)

Odds ratio (ref: controls)

≥30 v. b25⁎

0.7 (0.5, 1.0)

–

–

Odds ratio (ref: controls)

Per 5-unit increase

0.71 (0.57, 0.88)

0.88 (0.48, 1.60)

0.53 (0.25, 1.12) 1.18 (0.86, 1.64) 1.17 (0.68, 2.00)

Per 1 unit increase in WHO 1.11 (0.84, 1.48) BMI category ⁎⁎⁎⁎

1.73 (1.07, 2.77)

0.67 (0.32, 1.41)

1.67 (1.22, 2.28)

Odds ratio (ref: controls)

–

1.0 (0.6, 1.8)

Odds ratio (ref: non-triple negative cases) Odds ratio (ref: luminal A)

≥35 v. b25⁎ ≥30 v. b18.5

– (ref)

– 0.7 (0.1, 6.5)

– –

0.09 (0.01, 0.97)⁎⁎ 3.7 (1.2, 12.1)

Odds ratio (ref: luminal A)

≥30 v. b25

(ref)

0.95 (0.71, 1.28)

1.25 (0.87, 1.80)

1.80 (1.42, 2.29)

Odds ratio (ref: luminal A)

≥30 v. b25

(ref)

1.68 (0.92, 3.07)

2.51 (0.74, 8.51)

1.97 (1.03, 3.77)

Odds ratio (ref: luminal A)

≥30 v. b25

(ref)

0.7 (0.3, 1.4)

0.6 (0.3, 1.5)

Odds ratio (ref: luminal A)

≥30 v. b25

(ref)

0.77 (0.23, 2.54)

1.12 (0.39, 3.19)

1.46 (0.66, 3.21)

Odds ratio (ref: luminal A)

≥25.0 v. 18.5–24.9

(ref)

3.32 (0.98, 10.81) 1.23 (0.06, 9.76)

2.94 (0.80, 10.04)

Includes test of trend across exposure categories. p-Value for test of trend was statistically significant at α = 0.05. Under age 50. WHO BMI categories: underweight (b18.5), normal weight (18.5 to b25), overweight (25 to b30), obese (30+). Triple negative tumor positive for at least one of CK5/6 or HER1. Triple negative tumor negative for CK5/6 and HER1. Triple negative tumor positive for CK5 and/or HER1. Triple negative tumor negative for CK5 and HER1. Luminal A tumors with Ki-67 low (b14%). Luminal B tumors with Ki-67 high (≥14%). Luminal A tumors with histologic grade 1 or 2. Luminal B tumors with histologic grade 3.

1.6 (0.9, 2.7)

1.1 (0.6, 2.0)

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Yang 2007

Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)a, unclassified (150)b Luminal A (552), luminal B (48), HER2-overexpressing (61), basal-like (95)c, unclassified (48)d Luminal A (455), luminal B (72), HER2-overexpressing (117), triple negative (246)

Measure of association

Table 7 Adult postmenopausal BMI and risk of breast cancer subtypes. First author year

Cases with subtyping (n=)

Cohort and case–control Phipps 2012⁎⁎⁎ HER2-overexpressing (308), triple negative (705)

Millikan 2008

Yang 2007

Gaudet 2011

Case only Chen 2013 Yang 2011⁎⁎⁎ Kwan 2009 Millikan 2008

Devi 2012 Kimura 2012 ⁎ ⁎⁎ ⁎⁎⁎ ⁎⁎⁎⁎ a b c d e f g h

Luminal A (1127)e, luminal B (227)f, HER2-overexpressing (119), triple negative (477) Luminal A (9534), luminal B (1657), HER2-overexpressing (953), triple negative (1997) Luminal A (1868), luminal B (294), HER2-overexpressing (94), triple negative (288) Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)a, unclassified (150)b Luminal A (493), luminal B (125), HER2-overexpressing (117), triple negative (299) Luminal A (333)g, luminal B (85)h, HER2-overexpressing (43), triple negative (70)

Exposure comparison

Luminal A

Luminal B

HER2

Triple negative Basal-like

Unclassified

Hazard ratio (ref: non-cases)

Odds ratio (ref: controls)

≥30 v. b25 (age ≥50, no HT) ≥30 v. b25 (age ≥50, HT) ≥30 v. b25⁎

– – 0.8 (0.6, 1.1)

– – –

1.11 (0.82, 1.51) 1.03 (0.58, 1.84) –

Odds ratio (ref: controls)

≥25 v. 18.5–21.9⁎

2.13 (1.33, 3.42)⁎⁎

2.38 (0.75, 7.58)

1.04 (0.33, 3.30)

Odds ratio (ref: controls)

Per 5-unit increase

1.00 (0.90, 1.12)

1.02 (0.74, 1.42)

0.97 (0.73, 1.28)

Odds ratio (ref: controls)

Per 1 unit increase in WHO BMI category ⁎⁎⁎⁎

1.16 (0.87, 1.54)

0.83 (0.36, 1.93)

0.93 (0.57, 1.52)

1.02 (0.70, 1.48)

Odds ratio (ref: luminal A)

≥30 v. b18.5

(ref)

0.7 (0.1, 3.8)

1.0 (0.1, 13.3)

0.3 (0.1, 1.0)

Odds ratio (ref: luminal A)

≥30 v. b25

(ref)

0.93 (0.76, 1.14)

0.84 (0.65, 1.08)

1.09 (0.91, 1.29)

Odds ratio (ref: luminal A)

≥30 v. b25

(ref)

0.99 (0.66, 1.47)

0.76 (0.38, 1.51)

0.76 (0.49, 1.17)

Odds ratio (ref: luminal A)

≥30 v. b25

(ref)

0.7 (0.4, 1.3)

1.1 (0.6, 2.2)

Odds ratio (ref: luminal A)

≥30 v. b25

(ref)

1.36 (0.59, 3.11)

0.43 (0.15, 1.18)

0.63 (0.34, 1.16)

Odds ratio (ref: luminal A)

≥25.0 v. 18.5–24.9

(ref)

0.66 (0.33, 1.26)

0.81 (0.34, 1.80)

0.88 (0.43, 1.73)

1.32 (1.06, 1.64) 1.09 (0.79, 1.49) 0.6 (0.3, 1.1) – 7.51 (1.84, 9.22)⁎⁎ 0.87 (0.66, 1.14)

1.0 (0.5, 1.7)

1.08 (0.78, 1.48)

0.6 (0.3, 1.1)

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Sueta 2012

Luminal A (796), luminal B (137), HER2-overexpressing (116), basal-like (225)a, unclassified (150)b Luminal A (455), luminal B (108), HER2-overexpressing (84), triple negative (68) Luminal A (552), luminal B (48), HER2-overexpressing (61), basal-like (95)c, unclassified (48)d Luminal A (455), luminal B (72), HER2-overexpressing (117), triple negative (246)

Measure of association

Includes test of trend across exposure categories. p-Value for test of trend was statistically significant at α = 0.05. Over age 50. WHO BMI categories: underweight (b18.5), normal weight (18.5 to b25), overweight (25 to b30), obese (30+). Triple negative tumor positive for at least one of CK5/6 or HER1. Triple negative tumor negative for CK5/6 and HER1. Triple negative tumor positive for CK5 and/or HER1. Triple negative tumor negative for CK5 and HER1. Luminal A tumors with Ki-67 low (b14%). Luminal B tumors with Ki-67 high (≥14%). Luminal A tumors with histologic grade 1 or 2. Luminal B tumors with histologic grade 3.

83

84

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85

Table 8 Summary of risk factors by subtype. Risk factor

Luminal A Luminal B HER2-overexpressing Triple negative

Younger age at menarche Greater parity Older age at first birth Breastfeeding Older age at menopause Greater BMI (premenopausal) Greater BMI (postmenopausal) Family history Alcohol use Use of oral contraceptives Menopausal HT use

++

+

unk

+++

−−− ++ −− ++ −

unk unk −− unk unk

unk unk unk unk unk

++ unk −−− + +

unk

unk

unk

unk

+++ + −

+ unk unk

+++ + unk

+++ unk +

++

unk

unk

unk

+++: indicates consistent evidence of a positive association between the risk factor and the subtype. − − −: indicates consistent evidence of a negative association between the risk factor and the subtype. ++: indicates a probable positive association between the risk factor and the subtype. − −: indicates a probable negative association between the risk factor and the subtype. +: indicates a possible positive association between the risk factor and the subtype. −: indicates a possible negative association between the risk factor and the subtype. unk: indicates insufficient or inconsistent evidence.

of breast cancer, the strongest and most consistent established risk factor, and negatively associated with breastfeeding duration. The associations between established breast cancer risk factors and risk of luminal B breast cancer were not as well understood. The luminal B subtype was similar to the luminal A subtype in its association with family history of breast cancer and lifetime duration of breastfeeding. Of the remaining established risk factors, age at menarche was possibly associated with luminal B cancer. No other reproductive risk factors demonstrated an obvious association with luminal B cancer, and the relationships between other lifestyle risk factors and luminal B cancer remain to be determined. Triple negative cancer was similar to luminal A breast cancer in its strong relationship with both family history of breast cancer and lifetime duration of breastfeeding; however, this subtype differed from the luminal A subtype in its relationship to hormonal and lifestyle risk factors. While luminal A breast cancer was associated with every risk factor that increases one's lifetime number of hormone cycles, triple negative cancer was associated only with younger age at menarche, use of oral contraceptives, and greater parity. The association between triple negative cancer and lifestyle risk factors requires more research, though we did see some evidence of a difference in the relationships between BMI and risk of luminal A versus triple negative cancer. The risk profile of HER2-overexpressing cancer remains poorly understood, suggesting a need for expanded research on this subtype. The number of HER2-overexpressing cases included in each study was low, with most studies including fewer than 120 cases. Future studies including a greater number of HER2-overexpressing cases could clarify the relationship between HER2-overexpressing tumors and established breast cancer risk factors, and may also uncover risk factors unique to HER2overexpressing tumors. To our knowledge, this is the first comprehensive review addressing the associations between all types of established breast cancer risk factors and risk of luminal A, luminal B, HER2-overexpressing and triple negative breast cancers. We have identified the relationships between risk factors and breast cancer subtypes that are already well understood, and highlighted several areas that require further research. The greatest challenge in conducting this review arose because each study considered a unique set of established breast cancer risk factors and presented associations with study-specific comparison and reference groups. We addressed this challenge by presenting, for each risk factor, a table of

quantitative data from all studies as well as a carefully defined summary of the highest quality studies. As with any review, our study was also limited by the available research. The studies we included were primarily observational and may have been affected by biases, measurement error, misclassification, or low statistical power. To address this limitation, we placed interpretive emphasis on several well-designed cohort and case–control studies with sound methodology. This review illustrates the heterogeneity in risk factors among breast cancer subtypes, and emphasizes the need for additional research on subtype-specific breast cancer etiology and risk. It is evident that prior studies have focused on the comparison of luminal A and basal-like cancers, the most common tumor subtypes. Future studies will need to move beyond this binary contrast and devote adequate resources to the study of less-common tumor subtypes. Additionally, previous subtype-specific research has largely focused on hormonal risk factors. The studies we reviewed and those recently reviewed by Anderson et al. (2014) showed clear associations between hormonal risk factors and risk of luminal A cancer, as well as some probable associations with triple negative cancer [49]. Future research will need to consider a wider array of non-hormonal breast cancer risk factors, including race, ethnicity, lifestyle factors, and exogenous exposures. Research on non-hormonal risk factors may be particularly important in understanding the etiology of ER-negative subtypes. Emerging data suggest that there are other risk factors for breast cancer beyond those considered established and included in this review. Proposed models of cancer development suggest that breast tumorigenesis begins early in life and is shaped by the number of cells at risk [10, 50] the integrity of those cells [51–54], and the environment to which they are exposed [10]. Accordingly, next steps in breast cancer prevention research could include: 1) conducting additional studies to identify factors that determine the number of breast tissue cells, perhaps by focusing on mammographic density, 2) exploring the relationship between the expression of DNA repair genes and risk of breast cancer subtypes, 3) investigating the link between molecular pathways activated during chronic inflammation and risk of breast cancer subtypes, or 4) considering how early, life-long and late exposures to known carcinogens work in concert to impact breast cancer risk. Understanding how each of these mechanisms differentially affects subtype-specific breast cancer risk could illuminate additional modifiable risk factors and inform the development of new chemopreventive agents and risk reduction strategies. Transparency Documents The Transparency documents associated with this article can be found, in the online version. Acknowledgments Caroline Boeke was funded by the National Institutes of Health Cancer Epidemiology Training grant T32 CA 09001. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bbcan.2015.06.002. References [1] International Agency for Research on Cancer, GLOBOCAN 2012: estimated cancer incidence, mortality and prevalence worldwide in 2012Updated 9/10/2014 [cited 9/15/2014]; Available from: http://globocan.iarc.fr/Pages/fact_sheets_cancer.aspx. [2] K. Polyak, Heterogeneity in breast cancer, J. Clin. Investig. 121 (10) (2011) 3786–3788. [3] A. Goldhirsch, et al., Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013, Ann. Oncol. 24 (9) (2013) 2206–2223.

M.E. Barnard et al. / Biochimica et Biophysica Acta 1856 (2015) 73–85 [4] H.J. Bloom, W.W. Richardson, Histological grading and prognosis in breast cancer; a study of 1409 cases of which 359 have been followed for 15 years, Br. J. Cancer 11 (3) (1957) 359–377. [5] H.G. Russnes, et al., Insight into the heterogeneity of breast cancer through nextgeneration sequencing, J. Clin. Investig. 121 (10) (2011) 3810–3818. [6] C.M. Perou, et al., Molecular portraits of human breast tumours, Nature 406 (6797) (2000) 747–752. [7] S. Hankinson, R. Tamimi, D. Hunter, Chapter 16: breast cancer, in: H.O. Adami, D. Hunter, D. Tricopoulos (Eds.), Textbook of Cancer Epidemiology, 5th editionOxford University Press 2008, pp. 403–436. [8] W.C. Willett, R. Tamimi, S.E. Hankinson, A. Hazra, A.H. Eliassen, G.A. Colditz, Chapter 18: nongenetic factors in the causation of breast cancer, in: J.R. Harris, M.E. Lippman, M. Morrow, C.K. Osborne (Eds.), Diseases of the Breast, 5th editionLippincott Williams & Wilkins 2014, pp. 211–267. [9] Susan G. Komen, understanding breast cancer: risk factors and risk reductionAvailable from: http://ww5.komen.org/BreastCancer/LowerYourRisk.htmlJanuary 2 2014. [10] D. Trichopoulos, et al., Early life events and conditions and breast cancer risk: from epidemiology to etiology, Int. J. Cancer 122 (3) (2008) 481–485. [11] M.L. Welsh, et al., Population-based estimates of the relation between breast cancer risk, tumor subtype, and family history, Breast Cancer Res. Treat. 114 (3) (2009) 549–558. [12] T. Saxena, et al., Menopausal hormone therapy and subsequent risk of specific invasive breast cancer subtypes in the California Teachers Study, Cancer Epidemiol. Biomark. Prev. 19 (9) (2010) 2366–2378. [13] A.I. Phipps, et al., Body size, physical activity, and risk of triple-negative and estrogen receptor-positive breast cancer, Cancer Epidemiol. Biomark. Prev. 20 (3) (2011) 454–463. [14] A.I. Phipps, et al., Reproductive history and oral contraceptive use in relation to risk of triple-negative breast cancer, J. Natl. Cancer Inst. 103 (6) (2011) 470–477. [15] A.I. Phipps, et al., Reproductive history and risk of three breast cancer subtypes defined by three biomarkers, Cancer Causes Control 22 (3) (2011) 399–405. [16] A.I. Phipps, et al., Family history of breast cancer in first-degree relatives and triplenegative breast cancer risk, Breast Cancer Res. Treat. 126 (3) (2011) 671–678. [17] G.C. Kabat, et al., Smoking and alcohol consumption in relation to risk of triplenegative breast cancer in a cohort of postmenopausal women, Cancer Causes Control 22 (5) (2011) 775–783. [18] R.M. Tamimi, et al., Traditional breast cancer risk factors in relation to molecular subtypes of breast cancer, Breast Cancer Res. Treat. 131 (1) (2012) 159–167. [19] A.I. Phipps, et al., Breast density, body mass index, and risk of tumor marker-defined subtypes of breast cancer, Ann. Epidemiol. 22 (5) (2012) 340–348. [20] R.C. Millikan, et al., Epidemiology of basal-like breast cancer, Breast Cancer Res. Treat. 109 (1) (2008) 123–139. [21] A.I. Phipps, et al., Reproductive and hormonal risk factors for postmenopausal luminal, HER-2-overexpressing, and triple-negative breast cancer, Cancer 113 (7) (2008) 1521–1526. [22] A.I. Phipps, et al., Body size and risk of luminal, HER2-overexpressing, and triplenegative breast cancer in postmenopausal women, Cancer Epidemiol. Biomark. Prev. 17 (8) (2008) 2078–2086. [23] H. Ma, et al., Use of four biomarkers to evaluate the risk of breast cancer subtypes in the Women's Contraceptive and Reproductive Experiences Study, Cancer Res. 70 (2) (2010) 575–587. [24] M. Pollán, et al., Mammographic density and risk of breast cancer according to tumor characteristics and mode of detection: a Spanish population-based case–control study, Breast Cancer Res. 15 (1) (2013) R9. [25] C.I. Li, et al., Reproductive factors and risk of estrogen receptor positive, triplenegative, and HER2-neu overexpressing breast cancer among women 20–44 years of age, Breast Cancer Res. Treat. 137 (2) (2013) 579–587. [26] M.M. Gaudet, 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. 130 (2) (2011) 587–597. [27] J.M. Dolle, et al., Risk factors for triple-negative breast cancer in women under the age of 45 years, Cancer Epidemiol. Biomark. Prev. 18 (4) (2009) 1157–1166.

85

[28] X.R. Yang, et al., Differences in risk factors for breast cancer molecular subtypes in a population-based study, Cancer Epidemiol. Biomark. Prev. 16 (3) (2007) 439–443. [29] K.F. Trivers, et al., The epidemiology of triple-negative breast cancer, including race, Cancer Causes Control 20 (7) (2009) 1071–1082. [30] H. Razzaghi, et al., Association between mammographic density and basal-like and luminal A breast cancer subtypes, Breast Cancer Res. 15 (5) (2013) R76. [31] Y. Cui, et al., Interactions of hormone replacement therapy, body weight, and bilateral oophorectomy in breast cancer risk, Clin. Cancer Res. 20 (5) (2014) 1169–1178. [32] W. Zhou, et al., Family history and risk of ductal carcinoma in situ and triple negative breast cancer in a Han Chinese population: a case–control study, World J. Surg. Oncol. 11 (2013) 248. [33] P. Xing, J. Li, F. Jin, A case–control study of reproductive factors associated with subtypes of breast cancer in Northeast China, Med. Oncol. 27 (3) (2010) 926–931. [34] A. Sueta, et al., Differential impact of body mass index and its change on the risk of breast cancer by molecular subtype: a case–control study in Japanese women, Springerplus 1 (1) (2012) 39. [35] T. Islam, et al., Reproductive and hormonal risk factors for luminal, HER2overexpressing, and triple-negative breast cancer in Japanese women, Ann. Oncol. 23 (9) (2012) 2435–2441. [36] S.S. Shinde, et al., Higher parity and shorter breastfeeding duration, Cancer 116 (21) (2010) 4933–4943. [37] M.L. Kwan, et al., Epidemiology of breast cancer subtypes in two prospective cohort studies of breast cancer survivors, Breast Cancer Res. 11 (3) (2009) R31. [38] F.-Y. Chen, et al., Associations between body mass index and molecular subtypes as well as other clinical characteristics of breast cancer in Chinese women, Ther. Clin. Risk Manag. (2013) 131. [39] F.P. Turkoz, et al., Association between common risk factors and molecular subtypes in breast cancer patients, Breast 22 (3) (2013) 344–350. [40] A. Stark, et al., Obesity and risk of the less commonly diagnosed subtypes of breast cancer, Eur. J. Surg. Oncol. 35 (9) (2009) 928–935. [41] C.R.B. Devi, T.S. Tang, M. Corbex, Incidence and risk factors for breast cancer subtypes in three distinct South-East Asian ethnic groups: Chinese, Malay and natives of Sarawak, Malaysia, Int. J. Cancer 131 (12) (2012) 2869–2877. [42] M.E. Martinez, et al., Reproductive factors, heterogeneity, and breast tumor subtypes in women of Mexican descent, Cancer Epidemiol. Biomark. Prev. 22 (10) (2013) 1853–1861. [43] K. Kimura, Association between body mass index and breast cancer intrinsic subtypes in Japanese women, Exp. Ther. Med. 22 (4) (2012) 391–396. [44] E.F. Gillespie, et al., Obesity and angiolymphatic invasion in primary breast cancer, Ann. Surg. Oncol. 17 (3) (2010) 752–759. [45] L.A. Stead, et al., Triple-negative breast cancers are increased in black women regardless of age or body mass index, Breast Cancer Res. 11 (2) (2009) R18. [46] C.M. Redondo, et al., Breast feeding, parity and breast cancer subtypes in a Spanish cohort, PLoS ONE 7 (7) (2012) e40543. [47] M. Pilewskie, et al., Association between recency of last pregnancy and biologic subtype of breast cancer, Ann. Surg. Oncol. 19 (4) (2012) 1167–1173. [48] X.R. Yang, et al., Associations of breast cancer risk factors with tumor subtypes: a pooled analysis from the Breast Cancer Association Consortium studies, J. Natl. Cancer Inst. 103 (3) (2011) 250–263. [49] K.N. Anderson, R.B. Schwab, M.E. Martinez, Reproductive risk factors and breast cancer subtypes: a review of the literature, Breast Cancer Res. Treat. 144 (1) (2014) 1–10. [50] N.F. Boyd, et al., Mammographic density: a heritable risk factor for breast cancer, Methods Mol. Biol. 472 (2009) 343–360. [51] X. Jiang, D.J. Shapiro, The immune system and inflammation in breast cancer, Mol. Cell. Endocrinol. 382 (1) (2014) 673–682. [52] S.I. Grivennikov, F.R. Greten, M. Karin, Immunity, inflammation, and cancer, Cell 140 (6) (2010) 883–899. [53] C.E. Caldon, Estrogen signaling and the DNA damage response in hormone dependent breast cancers, Front. Oncol. 4 (2014) 106. [54] T. Helleday, et al., DNA repair pathways as targets for cancer therapy, Nat. Rev. Cancer 8 (3) (2008) 193–204.