Table of Contents Author Guidelines Submit a Manuscript
BioMed Research International
Volume 2018 (2018), Article ID 6026315, 10 pages
Research Article

Mammographic Breast Density and Breast Cancer Molecular Subtypes: The Kenyan-African Aspect

1Section of Medical Oncology, Department of Medicine, The Aga Khan University Hospital, Nairobi, Kenya
2Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA
3Cancer Prevention Fellowship Program, Division of Cancer Prevention, National Cancer Institute, Bethesda, MD, USA
4Department of Pathology, The Aga Khan University Hospital, Nairobi, Kenya
5Department of Radiology, The Aga Khan University Hospital, Nairobi, Kenya
6Department of Medical Education, The Aga Khan University Hospital, Nairobi, Kenya
7Department of Statistics and Epidemiology, The Aga Khan University Hospital, Nairobi, Kenya
8Section of Breast Surgery, Department of Surgery, The Aga Khan University Hospital, Nairobi, Kenya

Correspondence should be addressed to Asim Jamal Shaikh

Received 29 August 2017; Accepted 5 December 2017; Published 22 January 2018

Academic Editor: Neeta Kumar

Copyright © 2018 Asim Jamal Shaikh et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


Introduction. Data examining mammographic breast density (MBD) among patients in Sub-Saharan Africa are sparse. We evaluated how MBD relates to breast cancer characteristics in Kenyan women undergoing diagnostic mammography. Methods. This cross-sectional study included women with pathologically confirmed breast cancers (). Pretreatment mammograms of the unaffected breast were assessed to estimate absolute dense area (cm2), nondense area (cm2), and percent density (PD). Relationships between density measurements and clinical characteristics were evaluated using analysis of covariance. Results. Median PD and dense area were 24.9% and 85.3 cm2. Higher PD and dense area were observed in younger women (). Higher dense and nondense areas were observed in obese women (-trend < 0.01). Estrogen receptor (ER) positive patients (73%) had higher PD and dense area than ER-negative patients (). Triple negative breast cancer (TNBC) patients (17%) had lower PD and dense area () compared with non-TNBCs. No associations were observed between MBD and tumor size and grade. Conclusions. Our findings show discordant relationships between MBD and molecular tumor subtypes to those previously observed in Western populations. The relatively low breast density observed at diagnosis may have important implications for cancer prevention initiatives in Kenya. Subsequent larger studies are needed to confirm these findings.

1. Introduction

Mammographic breast density (MBD) is defined as the relative amounts of radio dense stromal and epithelial tissue compared with radiolucent adipose tissue. MBD is recognized as an established independent breast cancer risk factor [1]. It is a heritable trait [2, 3] and is thought to reflect cumulative exposure to established breast cancer risk factors, including parity, age at first birth, age at menopause, and endogenous and exogenous hormonal influences [47]. Studies of primarily Western populations have suggested that elevated MBD is a general marker of breast cancer risk, irrespective of molecular tumor subtype [8]. In addition, elevated MBD has been associated with adverse breast tumor clinical characteristics including larger tumor size, nodal involvement, and advanced stage at diagnosis [1, 6, 9]. Relationships between MBD and adverse prognostic features may reflect decreased mammographic sensitivity in women with dense breasts, resulting in delayed diagnosis in screened populations [6]. Furthermore, among women diagnosed with breast cancer, MBD may play a role in breast cancer progression [10]. However, much of the MBD evidence published to date is largely based on studies carried out among Western populations and it is currently unclear whether these findings are similar among African women.

Little is known about MBD within Sub-Saharan Africa (SSA), particularly among indigenous women. One study of 190 Ugandan women without breast cancer, referred for screening, with an average age of 46 years (standard deviation (SD) 13 years) at the time of mammography, indicated possible differences in patterns of MBD from those observed in the United States (US) or Europe, with a higher proportion of study participants having low MBD [11], as defined visually in categories using the American College of Radiology’s Breast Imaging Reporting and Data System (BI-RADS) [12]. The proportion of Ugandan women with low MBD (BI-RADS category 1) was 41% [11], which is greater than the proportion of US women in this lowest BI-RADS density category, which ranges from 0.6 to 9.4%, depending on race, age, and body mass index (BMI) [13, 14]; BMI was not assessed in the Ugandan report [11]. Whether differential patterns of MBD exist and how these MBD patterns relate to breast tumor characteristics in other regions of SSA is unknown.

In particular, the role of MBD in relation to breast cancer clinical characteristics within Kenya is currently unclear. To investigate this, we performed a cross-sectional analysis, examining MBD among indigenous Kenyan women with breast cancer utilizing quantitative software for MBD assessment. We examined relationships between MBD with features of breast cancer aggressiveness, including breast molecular tumor subtype and clinicopathological breast cancer characteristics.

2. Methods

2.1. Study Population

This study was a cross-sectional study of indigenous Kenyan women with pathologically confirmed breast cancer, diagnosed between February 2014 and May 2015 at the Aga Khan University Hospital, Nairobi. Diagnoses were confirmed by a breast pathologist, according to the National Comprehensive Cancer Network® (NCCN®) guidelines, which are followed as institutional policy for classification and treatment related decisions. Overall, 123 women out of 148 who were diagnosed with breast cancer during this time-period had pretreatment, diagnostic mammograms available for assessment. Participant information was abstracted from medical records and a computerized database included patient age, menopausal status, and BMI. Upon diagnosis, all breast cancer biopsies underwent a central pathology review with double reporting of clinical and tumor characteristics, including tumor size, tumor type, grade, and lymph node metastases. Breast cancer molecular subtype was defined according to estrogen receptor (ER)/progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2) status, which was evaluated using immunohistochemistry on the Dako Automated platform followed by pathology review (ZM/SS). Interpretation of ER/PR was done according to the Allred scoring system [15]. HER2 scoring was done according to the American Society of Clinical Oncology (ASCO)/College of American Pathologists (CAP) guidelines. Samples that scored as HER2 2+ (equivocal) were subjected to fluorescence in situ hybridization (FISH) testing to assess gene amplification, at the Aga Khan University Hospital reference laboratory in Karachi Pakistan. A FISH test result was considered positive if ratio of the HER2/neu/CEP17 signals was >2.2 [16].

All patients were consented for utilization of their information in this study. Approvals to conduct this analysis were obtained from University Research Ethics Committee prior to the study commencement.

2.2. Assessment of Mammographic Breast Density

MBD was assessed by a single radiologist (RN) from diagnostic digital mammograms (mediolateral view) of the unaffected breast, taken prior to any treatment was instituted, and using the semiautomated Cumulus 4 Software [17], a well-studied quantitative method of MBD assessment [6]. Cumulus is a computer-assisted thresholding program in which the user defines a threshold for the skin edge and dense area. Three independent measures of MBD were determined including percentage density (PD), absolute dense area (cm2), and total breast area (cm2). PD was calculated by dividing the dense breast area by the total breast area and multiplying by 100. Nondense area (cm2) was calculated by subtracting absolute dense area from the total breast area. The radiologist was masked to demographic and pathologic data. Quality assurance and reliability assessment of MBD measurements were performed internally for the first 25 cases using visual assessment by a second radiologist at the Aga Khan University Hospital, Nairobi. This independent assessment showed 97% concordance for PD > 50%.

2.3. Statistical Analysis

To examine the relationships between quantitative MBD measurements and patient and clinical characteristics, we used analysis of covariance (ANCOVA) models. Characteristics investigated included age (<50 years, ≥50 years), menopausal status, BMI (kg/m2 classified according to the World Health Organization as lean, overweight and obese) [18], histology (invasive ductal vs. other), tumor size (<3 cm, ≥3 cm), tumor grade (grade 1, grade 2, grade 3, and unknown), nodal status (negative, positive or unknown), ER status (negative, positive, unknown), PR status (negative, positive, and unknown), HER2 (negative, positive, and equivocal/ductal carcinoma in situ (DCIS)), and triple negative breast cancer (TNBC) status [no, yes (i.e., negative for ER, PR, and HER2), and unknown]. MBD measurements were square root transformed to better approximate normal distributions. For ease of interpretation, the least square means and standard errors from ANCOVA models were back transformed, and corresponding 95% confidence intervals were calculated and are presented in the tables. All models are presented as unadjusted and adjusted for age and BMI. The results of adjusted models are discussed below unless otherwise mentioned. In sensitivity analyses, investigations of relationships between MBD measurements and molecular tumor subtypes were also stratified by menopausal status. values of <0.05 were considered statistically significant and tests of statistical significance were two-tailed. Reported values were not corrected for multiple testing. All analyses were performed using SAS software v9.3 (SAS Institute Inc., Cary, NC).

3. Results

3.1. Overview of Study Population

Patient demographic and breast tumor clinical characteristics for the overall study population are presented in Table 1. Participant characteristics stratified by ER status are also shown in Table 1. The median (range) age at diagnosis was 54 (27–83) years, with most women being postmenopausal (66%) and overweight/obese (79%) at diagnosis. Most breast tumors were invasive ductal (85%), ER-positive (73%), PR-positive (67%), and HER2-negative (76%). The median (range) tumor size was 3 cm (0.5–10 cm). A higher proportion of participants had grade 2 tumors (46%) and was positive for lymph node metastasis (49%). Compared with ER-positive breast cancers, ER-negative tumors tended to be larger, higher grade, and PR-negative (Table 1). No significant differences were observed between ER-positive and ER-negative breast tumors for patient age, menopausal status, BMI, nodal status, and HER2 status (Table 1).

Table 1: Select characteristics of breast cancer patients with pretreatment diagnostic mammograms, Aga Khan University Hospital, Kenya, 2014-2015.
3.2. Relationships between MBD Measurements and Patient and Clinical Breast Tumor Characteristics

The distributions of MBD measurements by age are shown in Figure 1 and Supplementary Figure 1. The median (range) PD was 24.9% (2.1–76.9%) and median (range) dense area was 85.3 cm2 (1.5–355.6 cm2). Relationships between MBD measurements and breast tumor clinical characteristics are shown in Table 2. On average, significantly higher MBD was observed in younger (<50 years) versus older (≥50 years) women (adjusted mean PD: 31.8 versus 19.8%, ; adjusted mean dense area: 110.1 versus 72.1 cm2, ). Significantly lower nondense area was also observed in younger than older women (232.5 versus 306.9 cm2; ). In unadjusted models, higher PD and lower nondense area were observed in pre- versus postmenopausal women (); however, these findings were attenuated after adjusting for age and BMI. Significantly higher dense area and nondense area, reflecting higher total breast area, were observed in women who were obese versus overweight/lean ( values for trend < 0.01 for each); however, relationships between PD and BMI were not statistically significant. No relationships were observed between the MBD measurements examined (PD, dense area and nondense area) and breast tumor, size, grade, and nodal status (Table 2).

Table 2: Relationships between patient and clinical characteristics of women with breast cancer and mammographic density measurements (unadjusted and adjusted analysis).
Figure 1: Distribution of mammographic percent breast density (PD) measurements by age.
3.3. Relationships between MBD Measurements and Breast Molecular Tumor Subtype

Unadjusted and adjusted relationships between MBD measurements and breast molecular tumor subtype are shown in Table 3. In addition, because prior studies have shown that the prevalence of dense breasts as well as breast cancer molecular subtypes vary by menopausal status, we carried out sensitivity analyses investigating relationships between MBD measurements and subtypes also stratified by menopausal status. Compared with women diagnosed with ER-negative tumors, women with ER-positive tumors had significantly higher PD and dense area (for PD: 25.8% versus 19.3%, ; and for dense area: 94.4 versus 63.2 cm2, ) (Table 3). These patterns were consistent albeit attenuated when analyses were stratified by menopausal status. No significant relationships were observed between MBD measurements and either PR or HER2 status. In contrast, TNBC breast cancer cases had significantly lower PD and dense area as compared with non-TNBC subtypes (for PD: 17.4 versus 26.1%, , and for dense area: 56.3 versus 96.6 cm2, ), findings that were most apparent, albeit attenuated, among premenopausal women. Overall, no relationships were observed between nondense area and any of the breast cancer molecular subtypes examined.

Table 3: Relationship between MD and tumor subtype (overall and stratified by menopausal status) among women included in this study.

4. Discussion

This study is the first to present data from indigenous Kenyan women showing relationships between breast cancer clinical characteristics and MBD. To date, much of the literature surrounding MBD has focused on Western populations. In this analysis, we show inverse relationships between MBD and patient age, which are consistent with Western trends [7], with higher PD and dense area observed in younger than older women. Similar inverse associations were found for MBD and menopausal status, with higher MBD observed among premenopausal compared with postmenopausal women. In contrast to Western populations of screened women [10, 19, 20], we did not identify relationships between MBD measurements and breast cancer prognostic features such as tumor size, grade, and nodal status. However, we did find significant differences in MBD measures by molecular tumor subtype, with higher PD and absolute dense area for women who were positive for ER expression and lower PD and absolute dense area for women with TNBC.

As highlighted in the Introduction section, a prior analysis of MBD among Ugandan women referred for mammographic screening and without breast cancer found that a large proportion of study participants had low breast density (BI-RADS category 1 and 2; 67.9%) [11]. This proportion is higher than studies among Western populations, including a study of MBD among patients undergoing a breast biopsy following an abnormal mammogram which found that about 40% of African-American patients had MBD of either BI-RADs category 1 or 2, which roughly reflects PD < 50% [14]. Within our study population, the mean PD was 26.6%, with most (89%) women having PD measurements less than 50%. These distributions are comparable to those found among the screening population in Uganda, although BI-RADs measurements were utilized in Uganda. Studies that have examined the relationship between MBD and race have found mixed results to date and often yielded differing findings following the inclusion of age, BMI, and breast size as adjustments factors, all of which vary between populations [13]. The high proportion of lower breast density observed in SSA could be partly accounted for by elevated BMI, as an increase in breast size associated with a higher BMI may result in lower percentage breast density. However, BMI was not reported in the prior Ugandan study and thus a strength of this study was our ability to account for this influence by including BMI as an adjustment factor within our analysis.

Among breast cancer patients within SSA, it is recognized that there are differences in breast cancer characteristics compared with Western populations. Findings of a recent meta-analysis by Jedy-Agba and colleagues, which included 83 studies from 17 countries in SSA and reviewed breast cancer diagnosis stage in this setting, found that 77% of breast cancer patients were diagnosed with stage 3/4 disease and 10–15 years younger than in developed countries [21, 22]. In agreement with that meta-analysis, our study population had a high proportion of tumors that were larger, higher grade, and positive for lymph node metastasis. Our study showed no association between these prognostic characteristics and MBD measurements. While to our knowledge these relationships have not been previously investigated in Kenya, these findings contrast with a larger case-control study nested within the Nurses’ Health Study prospective cohort, which found that elevated MBD was associated with larger, higher grade tumors [20].

The literature investigating relationships between MBD and breast cancer molecular subtypes defined according to hormone receptor status have also been inconsistent to date. A meta-analysis conducted by Antoni and colleagues that is consisted of over 24,000 breast cancer cases from 19 studies found that the magnitudes of the relationships between MBD and risk of ER-positive and ER-negative breast cancer were similar, highlighting that MBD is a strong marker of breast cancer risk irrespective of clinical subtype [8]. However, it is important to note that the majority of the studies included in that meta-analysis were based within US and European populations [8]. We showed higher MBD measurements among women with ER-positive than ER-negative breast cancers, findings that were attenuated when stratified by menopausal status. While these findings could be an artifact of a smaller sample size of ER-negative breast cancers, these relationships are likely influenced by multiple factors that could affect both MBD and ER-status, including menopausal status [23]. Of note in this study, a higher proportion of ER-positive compared to ER-negative breast cancer cases was premenopausal at diagnosis. MBD is inversely associated with menopausal status and is higher in younger premenopausal women in this and other populations [23]. In analyses stratified by menopausal status, differences in MBD by ER status were attenuated among premenopausal women. Future studies with larger numbers of ER-positive and ER-negative breast cancers from pre- and postmenopausal women are needed to clarify these relationships among African women. The findings observed for lower PD and absolute dense area among women with TNBC are consistent with our findings of higher MBD measurements observed for women positive for ER expression.

A high proportion of low MBD in SSA, particularly among indigenous Kenyan women may have important implications for the application of cancer prevention initiatives. For example, lower resource settings, where it may not be feasible to establish large population-based screening programs for breast cancer, may obtain greater benefits from inexpensive and relatively safe breast cancer screening tools, such as mammography or ultrasound. This is particularly relevant in a resource challenged setting like Kenya, where breast cancer remains the most common cancer of women, with an age-standardized incidence rate of 51.7 per 100,000 [24].

This analysis is one of the first to date to investigate the role of MBD among indigenous Kenyan women with breast cancer. In addition, the use of computer-assisted software for breast density assessment allows for quantitative interpretation of MBD measures, reducing the potential biases associated with visual assessment. The population included in this analysis was not enrolled in a breast cancer screening program and thus their diagnoses were based upon presentation of symptoms or evaluation of breast lumps. Therefore, this provides a unique opportunity to examine relationships between MBD and breast tumor molecular subtypes without the influence of screening. Limited data, however, were available on traditional breast cancer risk factors including family history and reproductive factors such as parity. Therefore, future studies will be needed to determine whether the observed relationships between MBD and ER-positive versus ER-negative breast cancer are influenced by these factors. Expanded ongoing efforts in Kenya to understand the role of MBD in breast cancer risk and progression will further complement this analysis. Kenya’s participation in the International Consortium of Mammographic Density, a consortium developed by the International Agency for Research on Cancer (IARC), whose goal is to pool international MBD data on women without breast cancer, will allow for increased understanding of the role of MBD in breast cancer etiology within this population and will also allow in-depth investigation of MBD in Kenya and its comparability with other countries [25].

It is acknowledged that the landscape of breast cancer in Africa is changing with increased incidence and mortality being observed over the last 20 years. As the drivers of these trends are not fully elucidated, it is important to better understand factors that may influence breast tumor development and progression. This study is a first step in understanding MBD from a Kenyan and African perspective in a clinical setting. The differential patterns of association observed within this population for MBD by ER status suggest that MBD contributes to ER-positive and ER-negative breast cancers potentially through differential mechanisms. Identifying factors specific to breast cancer molecular subtypes will help improve our understanding of ER-positive versus ER-negative tumor etiology and progression.

In conclusion, we present results from indigenous Kenyan breast cancer patients that offer further insights into the role of MBD in breast cancer. This study provides a foundation for expanding efforts to investigate the role of MBD in breast cancer risk and prognosis in Kenya and to understand geographical variation in MBD between Kenya and other countries worldwide. Additional larger multicenter studies in independent validation populations are needed to confirm these findings.

Conflicts of Interest

The authors have no conflicts of interest to declare.

Authors’ Contributions

Asim Jamal Shaikh and Maeve Mullooly contributed equally as first authors.


This study was supported by the University Research Support Unit of The Aga Khan University Hospital, Nairobi. This work was also supported in part by the Intramural Research Program of the US National Cancer Institute.

Supplementary Materials

Supplementary Figure  1. Distribution of mammographic breast density (MBD) measurements by age including (A) dense area, (B) nondense area, and (C) breast area. (Supplementary Materials)


  1. V. A. McCormack and I. dos Santos Silva, “Breast density and parenchymal patterns as markers of breast cancer risk: a meta-analysis,” Cancer Epidemiology Biomarkers & Prevention, vol. 15, no. 6, pp. 1159–1169, 2006. View at Publisher · View at Google Scholar · View at Scopus
  2. J. Stone, G. S. Dite, A. Gunasekara et al., “The heritability of mammographically dense and nondense breast tissue,” Cancer Epidemiology, Biomarkers & Prevention, vol. 15, no. 4, pp. 612–617, 2006. View at Publisher · View at Google Scholar · View at Scopus
  3. N. F. Boyd, G. S. Dite, J. Stone et al., “Heritability of mammographic density, a risk factor for breast cancer,” The New England Journal of Medicine, vol. 347, no. 12, pp. 886–894, 2002. View at Publisher · View at Google Scholar · View at Scopus
  4. N. F. Boyd, L. J. Martin, M. J. Yaffe, and S. Minkin, “Mammographic density and breast cancer risk: current understanding and future prospects,” Breast Cancer Research, vol. 13, no. 6, article 223, 2011. View at Publisher · View at Google Scholar · View at Scopus
  5. G. Ursin, E. O. Lillie, E. Lee et al., “The relative importance of genetics and environment on mammographic density,” Cancer Epidemiology, Biomarkers & Prevention, vol. 18, no. 1, pp. 102–112, 2009. View at Publisher · View at Google Scholar · View at Scopus
  6. N. F. Boyd, H. Guo, L. J. Martin et al., “Mammographic density and the risk and detection of breast cancer,” The New England Journal of Medicine, vol. 356, no. 3, pp. 227–236, 2007. View at Publisher · View at Google Scholar · View at Scopus
  7. L. J. Martin and N. F. Boyd, “Mammographic density. Potential mechanisms of breast cancer risk associated with mammographic density: Hypotheses based on epidemiological evidence,” Breast Cancer Research, vol. 10, no. 1, article no. 201, 2008. View at Publisher · View at Google Scholar · View at Scopus
  8. S. Antoni, A. J. Sasco, I. Dos Santos Silva, and V. McCormack, “Is mammographic density differentially associated with breast cancer according to receptor status? A meta-analysis,” Breast Cancer Research and Treatment, vol. 137, no. 2, pp. 337–347, 2013. View at Publisher · View at Google Scholar · View at Scopus
  9. J. N. Wolfe, “Breast patterns as an index of risk for developing breast cancer,” American Journal of Roentgenology, vol. 126, no. 6, pp. 1130–1139, 1976. View at Publisher · View at Google Scholar · View at Scopus
  10. K. Kerlikowske and A. I. Phipps, “Breast density influences tumor subtypes and tumor aggressiveness,” Journal of the National Cancer Institute, vol. 103, no. 15, pp. 1143–1145, 2011. View at Publisher · View at Google Scholar · View at Scopus
  11. M. Galukande and E. Kiguli-Malwadde, “Mammographic breast density patterns among a group of women in sub Saharan Africa,” African Health Sciences, vol. 12, no. 4, pp. 422–425, 2012. View at Publisher · View at Google Scholar · View at Scopus
  12. American College of Radiology, Breast Imaging Reporting and Data System® (BI-RADS®) 4, American College of Radiology, Virginia, Va, USA, 2003.
  13. M. G. Del Carmen, E. F. Halpern, D. B. Kopans et al., “Mammographic breast density and race,” American Journal of Roentgenology, vol. 188, no. 4, pp. 1147–1150, 2007. View at Publisher · View at Google Scholar · View at Scopus
  14. M. G. Del Carmen, K. S. Hughes, E. Halpern et al., “Racial differences in mammographic breast density,” Cancer, vol. 98, no. 3, pp. 590–596, 2003. View at Publisher · View at Google Scholar · View at Scopus
  15. R. W. Carlson, D. C. Allred, B. O. Anderson, H. J. Burstein, W. B. Carter, S. B. Edge et al., “Breast cancer. Clinical practice guidelines in oncology,” Journal of the National Comprehensive Cancer Network, vol. 7, no. 2, pp. 122–192, 2009. View at Google Scholar
  16. A. C. Wolff, M. E. H. Hammond, D. G. Hicks et al., “Recommendations for human epidermal growth factor receptor 2 testing in breast cancer: American society of clinical oncology/college of American pathologists clinical practice guideline update,” Archives of Pathology & Laboratory Medicine, vol. 138, no. 2, pp. 241–256, 2014. View at Publisher · View at Google Scholar · View at Scopus
  17. J. W. Byng, N. F. Boyd, E. Fishell, R. A. Jong, and M. J. Yaffe, “The quantitative analysis of mammographic densities,” Physics in Medicine and Biology, vol. 39, no. 10, article no. 008, pp. 1629–1638, 1994. View at Publisher · View at Google Scholar · View at Scopus
  18. J. C. Seidell and K. M. Flegal, “Assessing obesity: Classification and epidemiology,” British Medical Bulletin, vol. 53, no. 2, pp. 238–252, 1997. View at Publisher · View at Google Scholar · View at Scopus
  19. G. L. Gierach, L. Ichikawa, K. Kerlikowske et al., “Relationship between mammographic density and breast cancer death in the breast cancer surveillance consortium,” Journal of the National Cancer Institute, vol. 104, no. 16, pp. 1218–1227, 2012. View at Publisher · View at Google Scholar · View at Scopus
  20. L. Yaghjyan, G. A. Colditz, L. C. Collins et al., “Mammographic breast density and subsequent risk of breast cancer in postmenopausal women according to tumor characteristics,” Journal of the National Cancer Institute, vol. 103, no. 15, pp. 1179–1189, 2011. View at Publisher · View at Google Scholar · View at Scopus
  21. E. Jedy-Agba, V. McCormack, C. Adebamowo, and I. dos-Santos-Silva, “Stage at diagnosis of breast cancer in sub-Saharan Africa: a systematic review and meta-analysis,” The Lancet Global Health, vol. 4, no. 12, pp. e923–e935, 2016. View at Publisher · View at Google Scholar · View at Scopus
  22. S. Sayed, Z. Moloo, P. Bird et al., “Breast cancer diagnosis in a resource poor environment through a collaborative multidisciplinary approach: The Kenyan experience,” Journal of Clinical Pathology, vol. 66, no. 4, pp. 307–311, 2013. View at Publisher · View at Google Scholar · View at Scopus
  23. N. Boyd, L. Martin, J. Stone, L. Little, S. Minkin, and M. Yaffe, “A longitudinal study of the effects of menopause on mammographic features,” Cancer Epidemiology, Biomarkers & Prevention, vol. 11, no. 10 I, pp. 1048–1053, 2002. View at Google Scholar · View at Scopus
  24. A. Korir, N. Okerosi, V. Ronoh, G. Mutuma, and M. Parkin, “Incidence of cancer in Nairobi, Kenya (2004–2008),” International Journal of Cancer, vol. 137, no. 9, pp. 2053–2059, 2015. View at Publisher · View at Google Scholar · View at Scopus
  25. V. A. McCormack, A. Burton, I. dos-Santos-Silva et al., “International Consortium on Mammographic Density: Methodology and population diversity captured across 22 countries,” Cancer Epidemiology, vol. 40, pp. 141–151, 2016. View at Publisher · View at Google Scholar · View at Scopus