IJC
International Journal of Cancer
Reproductive factors and risk of breast cancer by tumor
subtypes among Ghanaian women: A population-based
case–control study
Jonine D. Figueroa 1,2, Brittny C. Davis Lynn 1, Lawrence Edusei3, Nicholas Titiloye4, Ernest Adjei4,
Joe-Nat Clegg-Lamptey3, Joel Yarney3, Beatrice Wiafe-Addai5, Baffour Awuah4, Maire A. Duggan6, Seth Wiafe7, Kofi Nyarko8,
Francis Aitpillah4, Daniel Ansong9, Stephen M. Hewitt10, Thomas Ahearn1, Montserrat Garcia-Closas1, and
Louise A. Brinton1, on behalf of the Ghana Breast Health Study Team†
1Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD
2Usher Institute and CRUK Edinburgh Centre, University of Edinburgh, Edinburgh, United Kingdom
3Korle Bu Teaching Hospital, Accra, Ghana
4Komfo Anokye Teaching Hospital, Kumasi, Ghana
5Peace and Love Hospital, Kumasi, Ghana
6Department of Pathology and Laboratory Medicine, University of Calgary, Calgary, AB, Canada
7Loma Linda University, School of Public Health, Loma Linda, CA
8University of Ghana, Accra, Ghana
9Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
10Center for Cancer Research, National Cancer Institute, Bethesda, MD
Higher proportions of early-onset and estrogen receptor (ER) negative cancers are observed in women of African ancestry than
in women of European ancestry. Differences in risk factor distributions and associations by age at diagnosis and ER status may
explain this disparity. We analyzed data from 1,126 cases (aged 18–74 years) with invasive breast cancer and 2,106 controls
recruited from a population-based case–control study in Ghana. Odds ratios (OR) and 95% confidence intervals (CI) were
estimated for menstrual and reproductive factors using polytomous logistic regression models adjusted for potential
confounders. Among controls, medians for age at menarche, parity, age at first birth, and breastfeeding/pregnancy were
15 years, 4 births, 20 years and 18 months, respectively. For women ≥50 years, parity and extended breastfeeding were
associated with decreased risks: >5 births vs. nulliparous, OR 0.40 (95% CI 0.20–0.83) and 0.71 (95% CI 0.51–0.98) for ≥19 vs.
<13 breastfeeding months/pregnancy, which did not differ by ER. In contrast, for earlier onset cases (<50 years) parity was
associated with increased risk for ER-negative tumors (p-heterogeneity by ER = 0.02), which was offset by extended
breastfeeding. Similar associations were observed by intrinsic-like subtypes. Less consistent relationships were observed with
ages at menarche and first birth. Reproductive risk factor distributions are different from European populations but exhibited
etiologic heterogeneity by age at diagnosis and ER status similar to other populations. Differences in reproductive patterns and
subtype heterogeneity are consistent with racial disparities in subtype distributions.
Introduction ages at menarche, nulliparity, late ages at first birth and lim-
Reproductive factors have been well documented as key breast ited breastfeeding. Breast cancer is a heterogeneous disease,
cancer risk factors with direct associations observed with early with differential etiologic associations for tumor subtypes
J.D.F., B.C.D.L. M.G.C. and L.A.B. contributed equally to this work
Additional Supporting Information may be found in the online version of this article.
Key words: reproductive risk factors, subtype heterogeneity, racial disparities, breast cancer
Abbreviations: ABCS: African Breast Cancer Study; AMBER: African American Breast Cancer Epidemiology and Risk; CI: confidence interval;
ER: estrogen receptor; HER2: human epidermal growth factor receptor 2; IHC: immunohistochemical; NCI: National Cancer Institute; OR:
odds ratios; PR: progesterone receptor
†Ghana Breast Health Study Team members listed in the Appendix
DOI: 10.1002/ijc.32929
History: Received 2 Oct 2019; Accepted 29 Jan 2020; Online 18 Feb 2020
Correspondence to: Jonine D. Figueroa, E-mail: jonine.figueroa@ed.ac.uk or Montserrat Garcia-Closas, E-mail: montserrat.garcia-
closas@nih.gov
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
2 Breast cancer risk by tumor subtypes in Ghana
What’s new?
Breast-cancer risks differ between women of African ancestry and women of European ancestry, especially for aggressive,
hormone-negative tumors. In this large African study, the authors found that, while increased parity reduced the risk of all
breast-cancer subtypes after age 50, the opposite was true for risk of developing estrogen-receptor (ER) negative tumors at a
younger age. However, longer breastfeeding appeared to reduce the risk both for early-onset, ER-negative tumors and for all
tumor subtypes with later onset. Efforts to promote extended breastfeeding may thus help reduce many breast cancers in this
population.
defined by estrogen receptor (ER), progesterone receptor described.17,18 Our primary analyses focused on ER status
(PR) and human epidermal growth factor receptor 2 (HER2) because this was the key marker of etiological heterogeneity
status.1 Most of these results derive from studies on European demonstrated in previous studies.19–21
ancestry populations. Similar investigations among African The study was approved by the Special Studies Institutional
ancestry populations are crucial given the differences in Review Board of the National Cancer Institute (Rockville,
demographic and risk factor distributions and their dispropor- MD), the Ghana Health Service Ethical Review Committee
tionately high incidence of early-onset breast cancer and ER- and institutional review boards at the Noguchi Memorial
negative aggressive subtypes.1–4 Institute for Medical Research (Accra, Ghana), the Kwame
Analyses of risk factors by the African American Breast Nkrumah University of Science and Technology (Kumasi,
Cancer Epidemiology and Risk (AMBER) consortium have Ghana), the School of Medical Sciences at Komfo Anokye
revealed differential risk factor associations by tumor subtypes Teaching Hospital (Kumasi, Ghana) and Westat (Rockville,
de 5,6fined by ER, PR and HER2 status. Parity was associated MD). All participants provided written informed consent.
with a decreased risk for ER-positive cancers but an increased
risk for triple-negative breast tumors; furthermore, ever Risk factor information
breastfeeding in parous women was strongly inversely related Subjects were asked about their pregnancies and outcomes,
to the risk of triple-negative tumors.6 Accumulating data sup- the month and year that each pregnancy was completed,
port similar observations in other studies on women of whether the baby (or babies) was breastfed, and for how many
African American and European ancestry, although distribu- months they breastfed. Women were also asked questions on
tions of risk factors differ.1,7–11 age at first menstruation, whether they were still menstruat-
With substantially increasing rates of breast cancer in sub- ing, and if no longer menstruating, the age at which menstrual
Saharan Africa, identifying risk factors and strategies for periods stopped and the reason for stopping.
reducing incidence are essential.12,13 A population-based case–
control study of breast cancer in Ghana aimed to overcome Tumor characteristics
challenges of previous African studies that were unable to Prior to treatment, 4–8 core-needle biopsies (14-gage) were
select population-based controls and properly classify hor- fixed in 10% neutral buffered formalin for 24–72 hr and then
mone receptor-negative cases.3,12,14,15 Using a census-based processed into formalin-fixed paraffin-embedded blocks for
sampling of controls16 and standardized protocols for collect- diagnosis using standardized protocols.17 Blocks that were not
ing tumor biopsy samples for immunohistochemical (IHC) required for diagnosis were sent to the National Cancer Insti-
staining from cases prior to treatment,17 we sought to deter- tute (NCI) for additional pathological review (80% of the
mine the associations between menstrual and reproductive 1,126 invasive cases). Because organized mammography
risk factors and breast cancer subtypes. screening is not routine in Ghana, 96% of tumors presented
as lumps >2 cm based on clinical examination.18 We obtained
Materials and Methods information on key IHC ER, PR and HER2 markers from
Study population pathology departments in Ghana for 776 cases (69%). ER and
In brief, cases were women presenting with lumps suspected PR status were considered positive if ≥10% of tumor cells sta-
to be breast cancer at three hospitals [Korle Bu Teaching Hos- ined positive. The proportion of cases that were classified as
pital (Accra), Komfo Anokye Teaching Hospital (Kumasi), 1–9% ER-positive cells was minimal (1.8%). For HER2,
and Peace and Love Hospital (Kumasi)] were recruited from tumors were considered positive if they demonstrated a
2013 to 2015 and controls frequency matched to cases by age homogeneous, dark pattern of staining in ≥10% of tumor
and districts of residence. Recent data cleaning efforts identi- cells. Indeterminate and negative cases were combined and
fied some duplicate subjects, leading to a few changes in eligi- considered HER2 negative.
bility status. Supporting Information Fig. S1 details the 1,126 We assessed the agreement of IHC assays performed in
invasive breast cancer cases and 2,106 controls included in the pathology departments in Ghana with those performed at an
present analysis. Details of the multidisciplinary population- NCI laboratory in 87 cases, using two tumor tissue samples
based case–control study in Ghana have been previously from the same patient. We observed good agreement for ER
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
Figueroa et al. 3
Table 1. Demographic and reproductive characteristics of 1,126 diagnosed invasive breast cancer cases and 2,106 controls from the Ghana
Breast Health Study
Controls (n = 2,106) Cases (n = 1,126)
Study population characteristics n % n %
Age (years)
<35 435 21 114 10
35–44 561 27 277 25
45–55 554 26 330 29
≥55 546 26 401 36
Unknown 10 4
Study site
Accra 736 35 384 34
Kumasi 1,370 65 742 66
Education
No formal education 498 24 254 24
Primary school 369 18 153 15
Junior secondary school 654 32 260 25
>Senior secondary school 512 25 387 37
Unknown 73 72
Family history of breast cancer
No 2,036 98 1,034 93
Yes 46 2 78 7
Unknown 24 14 1
Body size
Slight 585 29 253 24
Average 827 40 434 41
Slightly heavy 470 23 261 25
Heavy 163 8 104 10
Unknown 61 74
Age at menarche (years)
Median age at menarche (IQR) 15 (15–15) 15 (15–15)
<15 568 30 266 27
15 548 29 255 26
16 383 20 223 23
≥17 395 21 228 23
Unknown 212 154
Parity
Median parity (IQR) 4 (2–5) 3 (2–5)
Nulliparous 228 11 107 10
1–2 533 25 319 28
3–4 685 33 365 33
≥5 652 31 331 30
Unknown 8 4
Age at first birth (years)
Median age (IQR) 20 (18–24) 21 (19–25)
<19 555 31 235 25
19–21 510 28 265 28
22–25 412 23 260 27
≥26 322 18 197 21
Unknown 79 62
(Continues)
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
4 Breast cancer risk by tumor subtypes in Ghana
Table 1. Demographic and reproductive characteristics of 1,126 diagnosed invasive breast cancer cases and 2,106 controls from the Ghana
Breast Health Study (Continued)
Controls (n = 2,106) Cases (n = 1,126)
Study population characteristics n % n %
Median breastfeeding per pregnancy (months)
Median months(IQR) 18 (15–24) 18 (12–24)
<13 352 20 239 26
13–18 692 39 341 37
≥19 747 42 347 37
Unknown 87 92
Menopausal status
Premenopausal 1,276 61 495 44
Postmenopausal 816 39 629 56
Unknown 14 2
Age at menopause
Median years (IQR) 49 (45–51) 49 (45–51)
<45 119 18 86 17
45–49 222 33 162 33
50–54 267 39 192 39
≥55 68 10 54 11
Unknown 147 137
ER status
Positive 393 50
Negative 387 50
Unknown 346
PR status
Positive 402 52
Negative 374 48
Unknown 350
HER2 status
Positive 181 23
Negative 544 70
Inconclusive 54 7
Unknown 347
Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; IQR, interquartile range; PR, progesterone receptor.
and HER2 (79% for ER, n = 87, p < 0.0001 and 78% for HER2, breast cancer subtype (comparing case IHC-defined subtypes
n = 76, p < 0.0001). PR showed a 65% agreement (n = 86, with controls). Heterogeneity between menstrual and repro-
p = 0.002). To determine if associations differed by proxies for ductive risk factors was assessed using polytomous logistic
intrinsic subtypes based on IHC data, we further classified regression analyses restricted to cases (case-only analyses)
tumors as luminal A-like (ER+ or PR+ and HER2−), luminal B- with tumor characteristics and IHC as the outcome variable.
like (ER+ or PR+ and HER2+), HER2-enriched-like (ER−, PR− To test differences in ORs by age, a likelihood-ratio test was
and HER2+) or triple-negative/basal-like (ER−, PR− and performed by fitting the logistic regression models with and
HER2−). without interaction terms. Further, stratified analyses were
performed to determine risk associations according to older
Statistical analysis and younger women.
We observed a high correlation between total breastfeeding Odds ratios (ORs) and 95% confidence intervals (95% CIs)
years and number of births (rho = 0.87 among controls) and were estimated to determine menstrual and reproductive fac-
a lower correlation with median breastfeeding months per tors using polytomous logistic models adjusted for study site
pregnancy (rho = 0.15 among controls), the latter of which and age (as a categorical variable) as well as key risk factors,
was used to avoid collinearity in the models. Polytomous including education, a family history of breast cancer, self-
logistic regression estimated the OR and 95% CI for each reported body size based on pictograms17 and menopausal
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
Figueroa et al. 5
status or age at menopause. Trend tests were based on ordinal A total of 50, 52 and 23% of cases were ER-positive, PR-
categories of variables and a missing category was used in positive and HER2-positive, respectively (Table 1). Luminal A-
models to retain all women in the models. All the statistical like breast cancer was the most common subtype (49%)
tests were two-sided. Analyses were performed using STATA/ followed by triple-negative/basal-like (28%), HER2-enriched
MP 14.2 (StataCorp, College Station, TX). Plots on the means (15%) and luminal B-like breast cancers (8%; Supporting Infor-
and standard deviations of a 3-point running average for mation Fig. S1). There were no significant differences in cases
reproductive factors stratified by case/control status were pres- missing ER, PR and HER2 status by risk factor data (data not
ented to illustrate how these reproductive exposures have shown). We did not find significant differences in distributions
changed over time and were performed using R version 3.4.4. of molecular subtypes between women <50 years compared to
≥50 years of age (Supporting Information Fig. S1, χ2 p = 0.24).
Data availability We assessed descriptively if reproductive factors varied by
The datasets generated or analyzed during the current study age which is highly correlated with birth cohort (1945–1975,
are not publicly available due to data privacy of patients but are Fig. 1). Number of births was lower in older cases (correlated
available from the corresponding author on reasonable request. with older birth cohorts) compared to younger ages born in
more recent birth cohorts, with cases having fewer births on
Results average compared to the controls (among women 70 years of
Descriptive characteristics of cases and controls age, mean 4.1 for cases and 6.4 for controls; among women
Cases were slightly older than controls reflecting that the con- 40 years of age, mean 2.5 for cases and 3.3 for controls). Age
trols were initially frequency matched to all women with a at first birth increased by approximately 1 year in the later
suspicion of breast cancer prior to diagnosis confirmation. compared to earlier birth cohorts for both cases and controls
Approximately half of the cases had nonmalignant breast dis- (21.7 at ages 70 and 22.3 at ages 40 among controls). Age at
eases and tended to be younger than those with malignant menarche showed no apparent trends, hovering around
breast disease.17 The cases more often than the controls 15 years across the birth cohorts. Breastfeeding months per
reported late ages at menarche, few births, late ages at first pregnancy among controls declined until the 1960s and
birth and low median breastfeeding months (Table 1). steadily increased until 1975. Among the cases, breastfeeding
Figure 1. Temporal trends of menstrual and reproductive risk factors for cases and controls in the Ghana Breast Health Study by birth cohorts
from 1945 to 1975. (a) Age at menarche, (b) parity, (c) age at first birth and (d) median breastfeeding months per pregnancy. The means and
standard deviations plotted are the results of a three-point running average. Gray indicates standard deviation. [Color figure can be viewed at
wileyonlinelibrary.com]
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
6 Breast cancer risk by tumor subtypes in Ghana
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
Table 2. ORs and 95% CIs for select reproductive risk factors and overall breast cancer risk in women younger and older than 50 years in 1,122 cases and 2,096 controls
All women Women < 50 years old (n = 564) Women ≥ 50 years old (n = 558)
OR 95% CI p p-trend Controls Cases OR 95% CI p p-trend Controls Cases OR 95% CI p p-trend p-int (LRT)
Age at menarche (years)
<15 1.00 391 148 1.00 174 118 1.00
15 0.88 0.70 1.10 0.25 323 121 0.88 0.65 1.20 0.42 225 134 0.81 0.57 1.14 0.22
16 1.13 0.89 1.44 0.31 238 114 1.17 0.86 1.61 0.31 143 108 0.94 0.64 1.38 0.75
≥17 1.08 0.85 1.37 0.53 0.30 249 117 1.08 0.79 1.48 0.63 0.46 146 111 0.97 0.66 1.42 0.87 0.82 0.34
Parity
Nulliparous 1.00 209 73 1.00 19 34 1.00
1–2 1.04 0.72 1.51 0.83 406 204 0.93 0.58 1.50 0.78 122 115 0.59 0.28 1.24 0.16
3–4 0.80 0.55 1.15 0.23 429 187 0.79 0.49 1.27 0.32 254 176 0.41 0.20 0.84 0.01
≥5 0.73 0.50 1.07 0.10 0.005 246 99 0.70 0.42 1.18 0.18 0.06 403 230 0.40 0.20 0.83 0.01 0.01 0.02
Age at first birth (years)
<19 1.00 310 103 1.00 242 132 1.00
19–21 1.14 0.90 1.43 0.28 290 121 1.18 0.85 1.64 0.32 219 143 1.15 0.82 1.60 0.42
22–25 1.27 1.00 1.62 0.05 247 127 1.42 1.01 2.00 0.04 164 133 1.15 0.81 1.65 0.43
≥26 1.18 0.91 1.54 0.22 0.135 204 121 1.40 0.97 2.01 0.07 0.05 118 76 1.03 0.68 1.56 0.88 0.74 0.28
Median months breastfeeding per pregnancy (among parous women)
<13 1.00 189 89 1.00 163 150 1.00
13–18 0.85 0.68 1.06 0.16 416 182 0.98 0.71 1.36 0.92 272 158 0.74 0.54 1.03 0.07
≥19 0.84 0.67 1.05 0.12 0.159 434 184 1.04 0.75 1.44 0.82 0.77 308 160 0.71 0.51 0.98 0.04 0.05 0.01
Logistic regression models were adjusted for age, education, study site, body size, family history of breast cancer, menopausal status, age at menopause and all reproductive factors listed above.
Abbreviations: CI, confidence interval; LRT, likelihood ratio test for interaction term; OR, odds ratio.
Figueroa et al. 7
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Table 3. Association between reproductive risk factors and breast cancer risk in women <50 years of age in 378 cases and 1,294 controls stratified by ER status
ER-positive (n = 185) ER-negative (n = 193)
ER-negative/ER-positive
Controls ER-positive ER-negative OR 95% CI p-trend OR 95% CI p-trend p-het
Age at menarche (years)
<15 391 40 54 1.00 1.00
15 323 44 37 1.21 0.76 1.93 0.69 0.43 1.09 0.08
16 238 37 43 1.38 0.85 2.25 1.21 0.77 1.90 0.79
≥17 249 47 38 1.61 1.00 2.58 0.05 0.97 0.61 1.55 0.81 0.09
Parity
Nulliparous 209 23 19 1.00 1.00
1–2 406 62 64 0.57 0.26 1.22 1.70 0.82 3.51 0.06
3–4 429 66 67 0.51 0.24 1.09 1.62 0.78 3.36 0.04
≥5 246 34 43 0.46 0.20 1.06 0.19 1.80 0.82 3.95 0.32 0.02
Age at first birth (years)
<19 310 31 40 1.00 1.00
19–21 290 45 40 1.43 0.86 2.36 1.08 0.66 1.76 0.54
22–25 247 37 52 1.34 0.78 2.30 1.64 1.01 2.65 0.49
≥26 204 45 32 1.72 0.99 2.97 0.08 1.15 0.66 1.99 0.30 0.29
Median months breastfeeding per pregnancy (among parous women)
<13 189 24 39 1.00 1.00
13–18 416 65 57 1.37 0.82 2.29 0.67 0.42 1.05 0.02
≥19 434 64 61 1.39 0.83 2.34 0.29 0.71 0.45 1.12 0.25 0.04
Polytomous logistic regression models were adjusted for age, education, study site, body size, family history of breast cancer, menopausal status and all reproductive factors listed above.
Abbreviations: CI, confidence interval; ER, estrogen receptor; OR, odds ratio; p-het, p-heterogeneity test.
Cancer Epidemiology
8 Breast cancer risk by tumor subtypes in Ghana
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
Table 4. Reproductive risk factors in women ≥50 years of age in 398 cases and 802 controls stratified by ER status
ER-positive (n = 205) ER-negative (n = 193)
ER-negative/ER-positive
Controls ER-positive ER-negative OR 95% CI p-trend OR 95% CI p-trend p-het
Age at menarche (years)
<15 177 51 39 1.00 1.00
15 225 50 46 0.69 0.43 0.85 0.85 0.51 1.39 0.60
16 145 37 27 0.73 0.43 0.69 0.69 0.39 1.22 0.84
≥17 146 42 43 0.84 0.50 1.13 0.74 1.13 0.67 1.92 0.77 0.36
Parity
Nulliparous 19 12 14 1.00 1.00
1–2 127 48 36 0.82 0.31 2.16 0.34 0.13 0.88 0.13
3–4 256 72 53 0.58 0.23 1.49 0.23 0.09 0.57 0.83
≥5 406 76 91 0.49 0.19 1.26 0.24 0.28 0.11 0.70 0.004 0.33
Age at first birth (years)
<19 245 46 50 1.00 1.00
19–21 220 49 56 1.10 0.67 1.78 1.26 0.80 2.00 0.55
22–25 165 48 42 1.10 0.66 1.83 1.06 0.64 1.75 1.00
≥26 118 32 24 1.09 0.61 1.93 0.72 1.02 0.57 1.84 0.95 0.84
Median months breastfeeding per pregnancy (among parous women)
<13 163 68 42 1.00 1.00
13–18 276 54 53 0.61 0.39 0.95 0.81 0.50 1.30 0.35
≥19 313 51 66 0.54 0.34 0.85 0.01 0.89 0.56 1.42 0.82 0.07
Polytomous logistic regression models were adjusted for age, education, study site, body size, family history of breast cancer, menopausal status, age at menopause and all reproductive factors
listed above.
Abbreviations: CI, confidence interval; ER, estrogen receptor; OR, odds ratio; p-het, p-heterogeneity test.
Figueroa et al. 9
months per pregnancy increased over time by 1 month per <19 years: OR 1.40, 95% CI 0.97–2.01, p-trend = 0.05). In more
pregnancy from 17 to 18 months. discrete categories of age, we observed a significant trend
(p = 0.01) with advancing age at first birth among women aged
Associations with reproductive factors overall and strati ed <40 years (Supporting Information Table S1). Age at menarchefi
by age and median breastfeeding months were not significantly associ-
Associations for age at menarche, number of births, age at rst ated with breast cancer risk among younger women. Amongfi
birth and median breastfeeding months per pregnancy overall women aged ≥50 years, a strong inverse association was observed
and strati ed by age are shown in Table 2. Analyses of all cases with parity (≥5 vs. 0 births: OR 0.40, 95% CI 0.20–0.83); a testfi
combined showed number of births as the only risk factor with a for interaction with age was significant (p = 0.02). Similarly,
statistically signi cant risk association (p-trend = 0.005). Among median breastfeeding months among older women werefi
women aged <50 years, we observed an inverse association with inversely associated with risk (≥19 vs. <13 months: OR 0.71, 95%
parity (≥5 vs. 0 births: OR 0.70, 95% CI 0.42–1.18, p-trend = 0.06) CI 0.51–0.98) and demonstrated a significant interaction with
and an increased risk with older ages at rst birth (≥26 vs. age (p = 0.01). Age at menarche was unrelated to risk amongfi
Figure 2. ORs and 95% CIs for joint effects of parity and breastfeeding (vs nulliparous) by ER status and age of onset. Polytomous logistic
regression models were used to calculate ORs and 95% CI, adjusted for age, education, study site, body size, family history of breast cancer,
age at menarche, age at first birth, menopausal status and age at menopause. Bars indicate standard deviations. Abbreviations:
BF, breastfeeding; ER, Estrogen receptor. Details of sample sizes, effect estimates and p values are presented in Supporting Information
Table S3.
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
10 Breast cancer risk by tumor subtypes in Ghana
older women (Table 2). Evaluation of these associations with those with ≥3 births who breastfed <13 months/pregnancy
more detailed categories of age revealed a significant interaction (OR 1.91, 95% CI 0.89–4.10). Women with ≥3 births who
by age for parity and median breastfeeding months per preg- breastfed, on average, ≥13 months per pregnancy were not at
nancy, with the strongest inverse associations of parity and increased risk (OR 1.09, 95% CI 0.56–2.10), due to the multi-
extended breastfeeding among women aged ≥60 years plicative joint association of two factors associated with risk in
(Supporting Information Table S1). opposite directions.
Associations with reproductive factors by ER and stratified Associations with reproductive factors by ER, PR, HER2
by age status and stratified by age
Analyses for all cases combined did not show statistically sig- We evaluated if associations with parity and breastfeeding dif-
nificant differences for ER-negative compared to ER-positive fered using the IHC proxy for intrinsic subtypes. We focused
cases (Supporting Information Table S2). When we evaluated our analyses on triple-negative compared to luminal A-like
the associations by ER status among women aged <50 years cases because previous studies have shown differences between
(Table 3), we observed a strong inverse association with parity these two groups6–9,20,21 and these were also the two most
for ER-positive tumors and a positive association for ER- common tumor subtypes (Supporting Information Tables S4
negative tumors, with the test for heterogeneity being statisti- and S5). Parity was inversely related to the risk of luminal A-
cally significant (p-het = 0.02). Among women <50 years, like tumors regardless of age, as well as with risk of triple-
older ages at first birth showed a slightly stronger direct asso- negative tumors among women aged ≥50 years (Supporting
ciation for ER-positive than ER-negative breast tumors, but Information Tables S4 and S5). In contrast, a positive associa-
the test for heterogeneity was not statistically significant. tion was observed for triple-negative tumors among women
Extended breastfeeding only showed an inverse association aged <50 years (p-het = 0.03). Among younger women with
among ER-negative tumors, with evidence of significant het- triple-negative tumors, extended breastfeeding was inversely
erogeneity compared to ER-positive tumors (≥19 vs. associated with risk, a relationship not observed for luminal
<13 months: ER-positive tumors OR 1.39, 95% CI 0.82–2.34; A-like tumors. In contrast, among older women, we observed
ER-negative tumors 0.71, 0.45–1.12; p-het = 0.04). There was a strong inverse association of breastfeeding with luminal A-
no additional relationship for ≥19 breastfeeding months; when like tumors (OR 0.52, 95% CI 0.33–0.82) that was not
we compared women with ≥13 breastfeeding months per observed for triple-negative tumors (p-het = 0.04; Supporting
pregnancy to <13 months, the resultant OR for ER-negative Information Table S5).
tumors was OR 0.69 (95% CI 0.45–1.03). There was a sugges-
tion of a positive association with older ages at menarche for Discussion
ER-positive breast tumors that was not apparent for ER- Among Ghanaian women, we observed substantial heteroge-
negative breast tumors. neity of the parity association with breast cancer risk by age at
Among the women aged ≥50 years (Table 4), parity was diagnosis and ER status, with strong inverse associations for
inversely associated with risk for both ER-negative and ER- all tumor subtypes in older (≥50 years) women and for
positive tumors (although there were few nulliparous women, younger-onset ER-positive tumors, but an opposite association
p-het = 0.33). Although extended breastfeeding showed an for younger-onset ER-negative tumors (i.e., increased risk with
inverse association regardless of ER status, a stronger associa- increasing birth numbers). Higher median breastfeeding
tion was observed among ER-positive tumors (p-het = 0.07). months per pregnancy were strongly inversely associated with
Age at first birth did not demonstrate any consistent associa- later-onset breast tumor risk (particularly ER-positive or lumi-
tions with risk. nal A-like tumors); among younger women, it was an appar-
We further assessed the joint effects of parity and ent protective factor for ER-negative tumors. Similar to
breastfeeding per pregnancy (Fig. 2 and Supporting Informa- previous reports,6–9,20,21 our study population allowed an eval-
tion Table S3). Among women aged ≥50 years, increasing par- uation of associations for a wide range of number of births
ity, and breastfeeding were associated with reduced risks for and breastfeeding months per pregnancy.
both ER-negative and ER-positive tumors, with the lowest Few studies have addressed the relation of reproductive
risks observed among women with ≥3 births who breastfed risk factors in women of African ancestry. The largest dataset
for ≥13 months/pregnancy compared to nulliparous women derives from the African Breast Cancer Study (ABCS),22 a
(ER-negative cases: OR 0.45, 95% CI 0.21–0.95; ER-positive hospital-based case–control study in Nigeria, Cameroon and
cases: OR 0.31, 95% CI 0.13–0.75). This trend was less appar- Uganda, comprising 1,995 cases and 2,631 controls (with 81%
ent among women aged <50 years with ER-positive tumors of the cases from Nigeria). Analyses from our study showed
[≥3 births who breastfed for ≥13 months/pregnancy compared changing reproductive patterns over time (particularly num-
to nulliparous women (OR 0.69, 95% CI 0.36–1.30)]. In con- ber of births) and an inverse association of risk with parity;
trast, among women aged <50 years with ER-negative tumors, however, it did not show statistically significant heterogeneity
compared to nulliparous women, the highest risk was for of risk associations by menopausal status or age at
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
Figueroa et al. 11
diagnosis.22,23 Notably, in contrast to our study, ABCS was increased risks for older ages at first birth for ER-positive but
not population-based and lacked information on hormone not for ER-negative tumors. Our data were consistent with
receptor status of the tumors, thereby limiting the comparabil- these findings, suggesting that this association may be stronger
ity of the findings. Data from the AMBER consortium, a or limited to early-onset ER-positive breast cancer cases.6
pooled analysis of four studies of African American women However, in African populations, this is a difficult exposure to
with available tumor IHC data found that among 1,252 ER- assess given that few women actually delay their first births
negative breast tumors parous women were at elevated risk until truly late ages. With the increasing adoption of western-
compared to nulliparous women, increasing to 1.60 among ized lifestyles and access to birth control, continued monitor-
those aged <40 years.6 Our data are consistent with AMBER ing of maternity data are needed to determine if ages at first
and other recent studies,9,11 supporting a cross-over associa- birth continue to increase.
tion between parity on breast cancer risk that is dependent on Despite the observed trends in reproductive patterns
age at onset and ER status. toward westernization, our study population still maintained
In our Ghanaian population, number of births and higher parity and breastfeeding frequencies compared to
breastfeeding years were highly correlated. Our data showed a other populations. The reproductive patterns in our study
significant inverse risk relationship with median breastfeeding are consistent with recent nationally representative sur-
months per pregnancy, with a 15% reduced risk for those with veys.29,30 For example, the decline in fertility rate from 6.4 in
13–18 vs. <13 months/pregnancy. In pooled analyses of 1988 to 3.9 in 2017 reported in surveys by the Ghana Mater-
populations of European ancestry, breastfeeding has been nal Health Survey ages 15–49 years is similar to the decline
shown to have a weak inverse association with breast cancer in average number of live births in our control population
risk. However, recent data that includes molecular subtyping from 6.4 to 3.3 for women born in 1945 (i.e., 43 years old in
information provides evidence of a possible stronger inverse 1988) and 1975 (i.e., 42 years old in 2017).29 Median
association for hormone-negative breast tumors.6–9,20,21 In the breastfeeding months per pregnancy were 17 to 18 months
AMBER study, the inverse association of breastfeeding was in our study controls and in a 2011 survey median months
most pronounced for younger-onset ER-negative and triple- breastfeeding were 17.4 and 17.9 months for Greater Accra
negative breast tumors. In fact, for such tumors, analyses and Ashanti regions, respectively.30 The strong inverse asso-
demonstrated that extended breastfeeding could reduce the ciations of these factors with late-onset, mostly ER-positive
adverse risks associated with parity, which has also been seen tumors, together with a lack of population-based screening,
in other studies that included African American women.9,11 are likely important factors contributing to historically low
Our results revealed similar associations given that extended incidence of late-onset ER-positive breast cancers. In con-
breastfeeding appeared to largely counteract the adverse rela- trast, for early-onset cancers, higher parity was directly asso-
tionship with multiparity among younger women with ER- ciated with ER-negative disease in our study. It is doubtful,
negative tumors. however, that high parity explains the higher incidence of
Recent studies assessing associations by molecular subtypes ER-negative early-onset cancers in our population given the
using IHC and mRNA expression profiling have shown high prevalence of breastfeeding, which appeared to offset
increased risk with parity that may predominate for triple- the higher risk from multiparity. Instead, the younger demo-
negative or basal-like breast tumors.20,24 In our study, the modifi- graphics in Ghana and other sub-Saharan African countries
cations in risk associations between parity and breastfeeding by probably explain the higher proportion of these early-onset
age reflected different temporal trend patterns by birth cohorts cancers compared to populations of European ancestry.3 It
in cases and controls: the rate of decrease in number of births may be that rather than a population with an “excess” of
was faster for controls than cases in early birth cohorts early-onset ER-negative cancers that there could be fewer
(i.e., older women); a decreasing trend of breastfeeding months diagnoses of late-onset ER-positive breast cancer compared
per pregnancy in early birth cohorts was seen in controls but not to other populations, as suggested in other studies.31 To spe-
in cases. Given that multiparity and increased breastfeeding are cifically address this, further studies comparing age-
inversely associated with later-onset breast cancers (with some- incidence rates of breast cancer subtypes in Africa are
what stronger associations with ER-positive tumors), if the needed, similar to U.S. studies that have addressed racial dif-
observed temporal trends of decreasing parity and breastfeeding ferences by age.32
continue, they are likely to result in an increased incidence of Age at menarche has been inversely associated with risk in
later-onset breast cancer.13 This indicates the importance of pub- European ancestry populations.33 In the studies of African
lic health measures to maintain high rates of breastfeeding,25 American women, later ages at menarche were inversely asso-
which could potentially attenuate the projected increase in risk ciated with breast cancer regardless of hormone receptor sta-
due to changes in reproductive patterns and demographics.13,26 tus.5,9 In contrast, we observed no such relationship. The
Older age at first birth has been associated with increases median age of menarche of 15 years in Ghanaian women is
in breast cancer risk in numerous studies, particularly for ER- quite different from the reported age of 12 years among
positive tumors.8,10,20,27,28 The AMBER consortium also found African American women, with our study having limited
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
12 Breast cancer risk by tumor subtypes in Ghana
variation in ages at menarche. Increased nutrition has been to African American or European ancestry populations, their
suggested to lower the age at menarche; this variable could associations with breast cancer risk were generally consistent
reflect early exposures that may differ between populations with those observed in these populations. Our data support the
(e.g., early adolescent weight).34 In addition, a substantial importance of breastfeeding to prevent early-onset ER-negative
number of women in our study could not recall their ages at breast cancer associated with multiparty and the longer-term
menarche, suggesting that measurement error could have protection of parity and breastfeeding for later-onset breast
impacted our ability to assess relationships reliably. tumors, irrespective of their ER status. Further studies including
Strengths of our study are the population-based design, more detailed molecular characterization of tumors and addi-
detailed risk factor assessment, and tissue collection for qual- tional risk factors may provide additional insights into breast
ity assessment of IHC markers to examine etiologic heteroge- cancer etiology in sub-Saharan Africa.
neity in sub-Saharan Africa. A limitation is that although IHC
data can be used as a proxy for molecular subtypes, mRNA
expression assays are required to classify previously described Acknowledgements
intrinsic molecular subtypes, especially HER2-enriched and We are grateful to all women who agreed to participate in the study and
luminal B subtypes. Further, although our study is one of the provided information and biospecimens. Dr. Figueroa would also
acknowledge personal funding on molecular subtypes of breast cancer
largest breast cancer epidemiological studies conducted in from Wellcome Trust 207800/Z/17/Z and MRC MR/S015027/1.
sub-Saharan Africa, analyses by age and subtypes resulted in
small numbers within strata of these critical factors.
Our study indicates that while reproductive factors showed Conflict of interest
important temporal trends and distinct distributions compared None declared.
References
1. Anderson KN, Schwab RB, Martinez ME. Repro- systematic review and meta-analysis of epidemio- 21. Work ME, John EM, Andrulis IL, et al. Reproduc-
ductive risk factors and breast cancer subtypes: a logical studies. Cancer Treat Rev 2016;49:65–76. tive risk factors and oestrogen/progesterone
review of the literature. Breast Cancer Res Treat 11. John EM, Hines LM, Phipps AI, et al. Reproduc- receptor-negative breast cancer in the breast can-
2014;144:1–10. tive history, breast-feeding and risk of triple nega- cer family registry. Br J Cancer 2014;110:1367–77.
2. Ademuyiwa FO, Tao Y, Luo J, et al. Differences in tive breast cancer: the breast cancer etiology in 22. Sighoko D, Ogundiran T, Ademola A, et al. Breast
the mutational landscape of triple-negative breast minorities (BEM) study. Int J Cancer 2018;142: cancer risk after full-term pregnancies among
cancer in African Americans and Caucasians. 2273–85. African women from Nigeria, Cameroon, and
Breast Cancer Res Treat 2017;161:491–9. 12. Akarolo-Anthony SN, Ogundiran TO, Uganda. Cancer 2015;121:2237–43.
3. Brewster AM, Chavez-MacGregor M, Brown P. Adebamowo CA. Emerging breast cancer epi- 23. Huo D, Adebamowo CA, Ogundiran TO, et al.
Epidemiology, biology, and treatment of triple- demic: evidence from Africa. Breast Cancer Res Parity and breastfeeding are protective against
negative breast cancer in women of African ances- 2010;12:S8. breast cancer in Nigerian women. Br J Cancer
try. Lancet Oncol 2014;15:e625–34. 13. Torre LA, Bray F, Siegel RL, et al. Global cancer 2008;98:992–6.
4. Newman LA, Kaljee LM. Health disparities and statistics, 2012. CA Cancer J Clin 2015;65:87–108. 24. Sun X, Nichols HB, Tse CK, et al. Association of
triple-negative breast cancer in African American 14. Bird PA, Hill AG, Houssami N. Poor hormone parity and time since last birth with breast cancer
women: a review. JAMA Surg 2017;152:485–93. receptor expression in east African breast cancer: prognosis by intrinsic subtype. Cancer Epidemiol
5. Ambrosone CB, Zirpoli G, Hong CC, et al. evidence of a biologically different disease? Ann Biomarkers Prev 2016;25:60–7.
Important role of menarche in development of Surg Oncol 2008;15:1983–8. 25. Victora CG, Bahl R, Barros AJ, et al. Breastfeeding
estrogen receptor-negative breast cancer in Afri- 15. Brinton LA, Figueroa JD, Awuah B, et al. Breast in the 21st century: epidemiology, mechanisms,
can American women. J Natl Cancer Inst 2015; cancer in sub-Saharan Africa: opportunities for and lifelong effect. Lancet 2016;387:475–90.
107:djv172. prevention. Breast Cancer Res Treat 2014;144: 26. Tamimi RM, Spiegelman D, Smith-Warner SA,
6. Palmer JR, Viscidi E, Troester MA, et al. Parity, 467–78. et al. Population attributable risk of modifiable
lactation, and breast cancer subtypes in African 16. Nyante SJ, Biritwum R, Figueroa J, et al. Rec- and nonmodifiable breast cancer risk factors in
American women: results from the AMBER con- ruiting population controls for case-control stud- postmenopausal breast cancer. Am J Epidemiol
sortium. J Natl Cancer Inst 2014;106:dju237. ies in sub-Saharan Africa: the Ghana breast health 2016;84:884–93.
7. Fortner RT, Sisti J, Chai B, et al. Parity, study. PLoS One 2019;14:e0215347. 27. Ma H, Wang Y, Sullivan-Halley J, et al. Use of
breastfeeding, and breast cancer risk by hormone 17. Brinton LA, Awuah B, Nat Clegg-Lamptey J, et al. four biomarkers to evaluate the risk of breast can-
receptor status and molecular phenotype: results Design considerations for identifying breast can- cer subtypes in the women’s contraceptive and
from the Nurses’ health studies. Breast Cancer Res cer risk factors in a population-based study in reproductive experiences study. Cancer Res 2010;
2019;21:40. Africa. Int J Cancer 2017;140:2667–77. 70:575–87.
8. Gaudet MM, Gierach GL, Carter BD, et al. Pooled 18. Brinton L, Figueroa J, Adjei E, et al. Factors con- 28. Tamimi RM, Colditz GA, Hazra A, et al. Tradi-
analysis of nine cohorts reveals breast cancer risk tributing to delays in diagnosis of breast cancers tional breast cancer risk factors in relation to
factors by tumor molecular subtype. Cancer Res in Ghana, West Africa. Breast Cancer Res Treat molecular subtypes of breast cancer. Breast Cancer
2018;78:6011–21. 2017;162:105–14. Res Treat 2012;131:159–67.
9. Ma H, Ursin G, Xu X, et al. Reproductive factors 19. Allred DC, Brown P, Medina D. The origins of 29. Ghana Statistical Service (GSS) GHSG, and ICF.
and the risk of triple-negative breast cancer in estrogen receptor alpha-positive and estrogen Ghana Maternal Health Survey. Rockville, MD:
white women and African-American women: a receptor alpha-negative human breast cancer. Department of Health Services, ICF, 2017. 2018.
pooled analysis. Breast Cancer Res 2017;19:6. Breast Cancer Res 2004;6:240–5. 30. Ghana Statistical Service (GSS). Ghana multiple
10. Lambertini M, Santoro L, Del Mastro L, et al. 20. Holm J, Eriksson L, Ploner A, et al. Assessment of indicator cluster survey with an enhanced malaria
Reproductive behaviors and risk of developing breast cancer risk factors reveals subtype hetero- module and biomarker, 2011, final report. Ghana:
breast cancer according to tumor subtype: a geneity. Cancer Res 2017;77:3708–17. Accra, 2011.
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology
Figueroa et al. 13
31. Dickens C, Duarte R, Zietsman A, et al. 32. Clarke CA, Keegan TH, Yang J, et al. Age-specific analysis, including 118 964 women with breast
Racial comparison of receptor-defined incidence of breast cancer subtypes: understanding cancer from 117 epidemiological studies. Lancet
breast cancer in southern African women: the black-white crossover. J Natl Cancer Inst 2012; Oncol 2012;13:1141–51.
subtype prevalence and age-incidence 104:1094–101. 34. Cheng G, Buyken AE, Shi L, et al. Beyond over-
analysis of nationwide cancer registry data. 33. Collaborative Group on Hormonal Factors in weight: nutrition as an important lifestyle factor
Cancer Epidemiol Biomarkers Prev 2014;23: Breast Consortium. Menarche, menopause, and influencing timing of puberty. Nutr Rev 2012;70:
2311–21. breast cancer risk: individual participant meta- 133–52.
APPENDIX: GHANA BREAST HEALTH STUDY TEAM
Ghana Statistical Service, Accra, Ghana: Dr Robertson Adjei Frempong, Emma Brew Abaidoo, Bridget Nortey Mensah,
and Dr Lucy Afriyie. Korle Bu Teaching Hospital, Accra, Samuel Amanama, Prince Agyapong, Debora Boateng,
Ghana: Dr Anthony Adjei, Dr Florence Dedey, Dr Verna Ansong Thomas Agyei, Richard Opoku and Kofi Owusu
Vanderpuye, Victoria Okyne, Naomi Ohene Oti, Evelyn Tay, Gyimah. Memorial Sloan Kettering Cancer Center, NY, USA:
Dr Adu-Aryee, Angela Kenu and Obed Ekpedzor. Komfo Dr Lisa Newman. National Cancer Institute, Bethesda, MD,
Anoyke Teaching Hospital, Kumasi, Ghana: Marion Alcpaloo, USA: Maya Palakal and Jake Thistle. Westat, Inc.: Michelle
Isaac Boakye, Bernard Arhin, Emmanuel Assimah, Samuel Brotzman, Shelley Niwa, Usha Singh and Ann Truelove.
Ka-chungu, Dr Joseph Oppong and Dr Ernest Osei-Bonsu. University of Ghana: Prof Richard Biritwum.
Peace and Love Hospital, Kumasi, Ghana: Prof Margaret
Int. J. Cancer: 00, 00–00 (2020) Published 2020. This article is a U.S. Government work and is in the public domain in the USA.
Cancer Epidemiology