Mammo JNMA

R E S E A R C H P A P E R
Relationship between Mammographic

Findings and Breast Abnormalities in a Nigerian
Population
Joseph A. Adedigba, B.A., Bukunmi M. Idowu, M.D., F.W.A.C.S., F.M.C.R., Sarah P. Hermans, B.A.,
Bolanle O. Ibitoye, M.D., F.W.A.C.S., Shivani Pahwa, M.D.
has resulted in a decreased incidence of locally advanced

Conflicts of Interest: None declared. breast cancer.1 While overall breast cancer incidence rates
Declaration of interests: The authors hereby declare that they have no known are lower compared to those in North America (50.5/
competing financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
100,000 vs 56.8/100,000), women in Nigeria currently
have the highest age-standardized breast cancer mortality
Abstract: Purpose: This study aims to describe the mammographic findings in a
population of Nigerian women and to explore the relationships between
rate internationally (25.9/100,000) compared to women in
abnormal mammographic findings, breast malignancy, and breast North America (13.1/100,000).2 This is due to low
composition.
participation of Nigerian women in mammography
Methodology: This was a retrospective study of consecutive mammograms
carried out at Union Diagnostics and Clinical Services in Lagos, Nigeria from 2016 screening, which can be attributed to reasons such as the
to 2018. Demographic information, indications for and findings on
mammographic evaluation were obtained. A logistic regression fit model was
cost of obtaining a mammogram and the lack of quality
used to establish the correlation between mammographic findings, breast assurance programs.3
density, and suspicion for breast malignancy (higher BIRADS scores). P 0.05
represented a statistically significant result. Women with breast cancer in Nigeria are often diag-
Results: A total of 304 patients were involved in this study (age range 20-80 years, nosed at a more clinically advanced stages, with greater
mean age 49.0 ± 10.5 years). The patients between 40 and 49 years formed the
largest age group with 128 patients (42.4%). Most patients were referred for a
than 70% of patients presenting at stages 3 or 4.4 This has
breast mass/lump (115/304-38.6%); 56 patients (18.8%) presenting for routine been attributed to a dearth of early detection programs, as
screening. The most common finding on the mammograms was BIRADS 4 in both
breasts in 96 patients (31.6%). Most patients had heterogeneous breast density well as diagnostic and treatment facilities.5 Women with
(195 patients - 64.1%). Multivariate logistic regression analysis showed a
significant correlation between history of mass, poorly defined margins, and
breast cancer in sub-Saharan Africa present with more
suspicion of malignancy. There was no statistically significant association aggressive tumors and are diagnosed at a younger
between abnormal mammographic findings and higher breast density.
average age of 46 years, compared to an average age of
Conclusion: Poorly defined margins were positively correlated with BIRADS ratings
suspicious for malignancy. The presence of a breast mass was positively correlated 67 years in Caucasian women in the United Kingdom.2
with a higher BIRADS score when other possible cofounding variables were not
accounted for. Patient age did not correlate with breast density in this study.
The high breast cancer mortality rate in Nigeria
indicates the need for development and implementation
Keywords: Screening mammography-Awareness-Nigerian women-Women’s
health-Sub-Saharan Africa
of appropriate screening and treatment programs.
Mammography, clinical breast exams, and self-exams are
some of the options for screening. Mammogram sensi-
Author affiliations: Joseph A. Adedigba, Tulane University School of Medicine, New
Orleans, LA 70112, USA; Bukunmi M. Idowu, Union Diagnostics and Clinical Services, Lagos
tivity varies with the patient age and breast density but
220005, Nigeria; Sarah P. Hermans, Tulane University School of Medicine, New Orleans, LA has been estimated to range between 60 and 90%, with
70112, USA; Bolanle O. Ibitoye, Department of Radiology, Obafemi Awolowo University, Ile-
Ife 220005, Nigeria; Shivani Pahwa, Department of Radiology, University Hospital specificity ranging from 55% to 95%.6,7
Cleveland Medical Center, Cleveland, OH, USA Nigeria is of interest both due to the presence of many
Correspondence: Joseph A. Adedigba, B.A., Tulane University School of Medicine, New Nigerian immigrant communities throughout the Western
Orleans, LA 70112, USA. Fax: 504-988-6705., email: [email protected]
world and the fact that Nigerian residents represent a
ª 2020 Published by Elsevier Inc. on behalf of the National Medical Association.
https://doi.org/10.1016/j.jnma.2020.07.012
significant portion of the global Black population. Re-
searchers in Nigeria and across sub-Saharan Africa have
documented various mammographic findings seen among
INTRODUCTION women in the region. While these descriptive studies have
M
ammography is a diagnostic and screening tool been beneficial, there is a need for inferential studies that
for early detection of breast cancer. In devel- explore how these mammographic findings correlate with
oped countries, screening with mammograms breast malignancy and breast density. This study aims to
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION VOL -, NO -, - 2020 1

Downloaded for Anonymous User (n/a) at Tulane University from ClinicalKey.com by Elsevier on August 05, 2020.
For personal use only. No other uses without permission. Copyright ©2020. Elsevier Inc. All rights reserved.
MAMMOGRAPHIC FINDINGS IN NIGERIAN WOMEN
describe the pattern of mammographic findings in a pop- Statistical analysis

ulation of Nigerian women and to explore relationships
Descriptive characteristics were categorical and were
between abnormal mammographic findings, breast
described by frequencies and proportions. Univariate lo-
composition, and likelihood of breast malignancy.
gistic regression evaluated individual associations between
the presence of a BIRADS score of 4 or 5 as an outcome
PATIENTS AND METHODS variable and the following explanatory variables: breast
density, margin definition, density of the mass (radiopaque
This retrospective study was carried out at Union
or radiolucent), architectural distortion, skin thickening,
Diagnostics and Clinical Services, a private radiological
nipple retraction, calcifications, axillary lymphadenopathy,
center in Lagos, Nigeria. Consecutive mammograms from
patient history, and age. We additionally performed uni-
2016 to 2018 were included in this study. This study
variate logistic regression on the presence of these
protocol complies with the 1964 Declaration of Helsinki
explanatory variables as related to breast composition.
and its subsequent revisions. Standard ethical committee
To account for possible confounding variables, we
review and informed consent were waived because retro-
performed multivariate logistic regression to explore the
spective, de-identified data was used. Demographic
relationship between a BIRADS score of 4 or 5 and all the
information and indications for mammographic evaluation
covariates estimated in univariate logistic regression ana-
were obtained. The patients were referred for mammo-
lyses listed above. BIRADS scores of 4 or 5 were used as
graphic evaluation on account of routine examination,
outcome variables in order to ascertain whether there is
breast pain, breast discharge, breast pain/discharge, and
any relationship between the explanatory variables and the
breast lumps/masses. Mammographic images of the
likelihood of a breast mass/lesion suspicious for malig-
breasts were acquired with a General Electric (GE) digital
nancy. We also explored multivariate logistic regression
mammogram machine (General Electric, GE, Healthcare
analyses of the relationship between these factors and a
Systems, Chicago, Illinois, USA) using two standard
breast composition of C or D. Breast composition cate-
views: CranioCaudal (CC) and Mediolateral Oblique
gories C and D were used as outcome variables in order to
(MLO). Spot compression and magnification views were
ascertain the relationship between the explanatory vari-
added, as additional views, whenever necessary. The same
ables and the likelihood of increased breast density. Ana-
staff radiologist interpreted all the mammograms.
lyses were conducted using SAS 9.4 software (SAS
The Breast Imaging Reporting Data System (BIRADS)
Institute, Cary, North Carolina, USA) via a logistic
was developed by the American College of Radiology
regression fit model. P 0.05 represented a statistically
(ACR) to standardize the reporting of mammograms.8
significant result.
When the right and left breast had the same BIRADS
score, a single score was recorded. When the score differed
for each breast, the BIRADS score for each breast was RESULTS
specified. BIRADS categories include 0-6, with 0 indica- A total of 304 mammograms were analyzed. The patient
tive of an incomplete assessment, 1 as negative, 2 as age ranged from 20 to 80 years with a mean age of
benign, 3 as probably benign with a malignancy likelihood 49.0 ± 10.5 years. The 40-49-year-old age group had the
<2%, 4a indicative of malignancy likelihood of 2-9%, 4b highest frequency at 128 patients (42.4%), followed by the
a likelihood of 10-49%, 4c a likelihood of 50-94%, and 5 50-59-year-old age group at 80 patients (26.5%)
of 95-100%. A BIRADS category of 4 or 5 is often fol- (Figure 1). There were 2 patients (0.7%) included in this
lowed by a referral for biopsy. A BIRADS score of 6 study whose ages were not recorded. There were 268 pa-
corresponds with a biopsy-proven malignancy. tients who came in for their first mammographic exami-
Mammographic breast density patterns were classified nation. The indications for mammography are shown in
using the ACR Breast Composition Categories.8 Category Table 1. The BIRADS scores are shown in Table 2 and the
A: breasts that are almost entirely fatty; Category B: number of patients with BIRADS scores of 4 or 5 by age
scattered areas of fibroglandular density; Category C: group is shown in Table 3.
heterogeneously dense breast tissue which may obscure Regarding breast composition, Category C had the
small masses; and Category D: extremely dense breast highest frequency at 195 patients (64.1%), followed by
tissue which lowers the sensitivity of mammography. category D at 78 patients (25.7%). Category B was seen in
Mammographic abnormalities such as opacities, lucency, 30 patients (9.9%), while category A had the lowest fre-
calcifications, architectural distortion, axillary lymphade- quency with just one patient (0.3%).
nopathy, skin thickening, nipple retraction and margins Radiologic findings included 55 patients (18.1%) with
were assessed and recorded for all patients. masses that had poorly-defined margins and 15 patients
2 VOL. -, NO -, - 2020 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION

Fig. 1. Showing the frequency distribution of patients in each age group.
(4.9%) with masses that had well-defined margins. The 243 patients. with 22 patients (9.0%) demonstrating nipple
density of the mass could be characterized in 79 patients. retraction. Axillary lymphadenopathy was also evaluated
The mass was radiopaque (Figures 2 and 3) in 64 patients in 293 patients - 268 patients (91.5%) were found to have
(81.0%) and radiolucent in 12 patients (15.2%). There axillary lymphadenopathy.
were 2 patients (2.5%) with multiple radiopaque masses Univariate and multivariate regression was utilized to
and 1 patient (1.3%) with multiple radiolucent masses. evaluate associations between a BIRADS score of 4 or 5
Calcifications were assessed in 296 patients (Figure 4). and radiologic findings, age, and patient history. In the
Calcifications were seen in 4 patients (1.4%) while calci- univariate model, a BIRADS score of 4 or 5 had signifi-
fications were not seen in 292 patients (98.7%). Archi- cant associations with radiologic findings of a mass,
tectural distortion, defined as the distortion of architecture
in the absence of a visible mass, was assessed in 222
patients. There were 9 patients (4.1%) with architectural Table 2. Distribution of the mammographic BIRADS score in the patients
distortion while 213 patients (96.0%) did not have archi- studied.
tectural distortion (Figure 5). Skin thickening was assessed

in 301 patients (Figure 5). Skin thickening was seen in 32 Breast Assessment-BIRADS N (%)
patients (10.6%). Nipple retraction was evaluated among BIRADS 1 62 (20.4%)
BIRADS 2 20 (6.6%)
BIRADS 3 17 (5.6%)
Table 1. Distribution of the presenting complaints in the patients studied.
BIRADS 4 96 (31.6%)
Respondent’s history N (%) BIRADS 5 8 (2.6%)
Routine screen 73 (24.5%) BIRADS 0 54 (17.8%)
Pain 74 (24.8%) BIRADS 0 (Left Breast) BIRADS 1 (Right 4 (1.3%)

Breast)
Discharge 20 (6.7%)
BIRADS 2 (Left Breast) BIRADS 1 (Right 17 (5.6%)
Pain/Discharge 4 (1.3%) Breast)
Breast Mass/Lump 115 (38.6%) BIRADS 3 (Left Breast) BIRADS 1 (Right 26 (8.6%)
Breast cancer 12 (4.3%) Breast)
Total ¼ 298 Total ¼
304
Table 1 shows the number of patients who presented for
mammographic evaluation for each listed indications or Table 2 shows the number of patients who were ascribed each
reasons. BIRADS score after mammographic evaluation.

A similar model was used to evaluate these character-

Table 3. Distribution of the mammographic BIRADS score in various age
groups studied in the patient population. istics in association with breast composition (Table 6).
Univariate and multivariate analysis of exploratory vari-
Age group BIRADS score 4 or 5
ables and breast compositions C and D found no statisti-
cally significant associations between increased breast
20-29 0
density and margin definition, radiological findings (radi-
30-39 14 (4.6%) opaque mass and radiolucent mass), architectural distor-
40-49 38 (12.6%) tion, skin thickening, nipple retraction, calcifications,
50-59 29 (9.6%) axillary lymphadenopathy, patient history, or age.
60-69 15 (5.0%)
70-79 7 (2.3%) DISCUSSION
80-89 0 Mammography is a useful tool for both screening and
Total 302 diagnostic evaluation of various breast lesions and
assessing breast density.9 The age range of the participants
Table 3 shows the number of patients in each age group who
were ascribed a BIRADS score of 4 or 5 after mammographic in our study was wide, but the 40-49-year-old age group
evaluation. was the most predominant cohort at 128 patients (42.4%).
Similar dispersion has been reported in a retrospective
study in the Greater Accra region of Ghana in which the
41-50 years age group had the highest number of cases at
specifically a radiopaque mass, as well as skin thickening, 52.2%.10 While research concerning Western countries has
nipple retraction, axillary lymphadenopathy, and age contested the efficacy of screening of the 40-49 age group,
greater than 50 (Table 4). Significant associations in the divergent presentation of malignancy documented in
patient history found using multivariate analysis included prior research in our population may represent differing
presentation for screening due to a breast mass or lump. screening needs. Arguments against screening in this age
Univariate regression also found a negative association group in Western populations include a 12% reduction in
between women presenting for routine screening and a mortality in the 39-49-year-old age group as compared to a
BIRADS score of 4 or 5. 14% reduction in mortality in the 50-59-year-old age
In the multivariate analysis, the only significant associa- group, and of 33% in the 60 to 69 age group.11
tion was with poorly defined margins (Table 5). The differ- In our patient population, the largest number of patients
ence in significance between the univariate and multivariate with a BIRADS score of 4 or 5 was seen in the 40-49 year
model are likely due to collinearity and correlation between age group at 12.6%. However, the 70-79 and 60-69 age
variables within the multivariate model. Since poorly defined group had the highest percentage of patients with BIRADS
margins were significant in multivariate and not in the uni- scores of 4 and 5 at 53.9% and 46.2%, respectively. Uni-
variate analysis, we can assume that it is statistically signifi- variate analysis showed that patients with lesions suspi-
cant only when other covariates are controlled for. cious for malignancy were 1.6 times more likely to be
Fig. 2. (A, B &C) (Left-Right): Mediolateral Oblique (MLO) and Craniocaudal (CC) Mammograms (A & B) showing a lobulated radiopaque lesion (arrows)
which was confirmed to be a multiloculated simple cyst on ultrasound(C).

Fig. 3. (A&B) (Left-Right): Mediolateral Oblique (MLO) and Craniocaudal (CC) Mammograms (A & B) showing a large radiopaque lesion (arrows)
interspersed with radiolucent foci, which was suspected to be a Phyllodes tumor.
greater than 50 years old than those with benign lesions also found breast pain to be the second most common
(P ¼ 0.04; OR 1.64; 95% CI 1.01-2.67). We found no indication for mammographic evaluation.14-17 Another
statistically significant difference when looking at women Nigerian study found pain to be the most common pre-
above and below 40 years old, which might suggest senting symptom in 47% of patients.18 These studies
screening should begin at 50 years in our patient popula- suggest an underlying theme, which is that patients in the
tion. However, we believe our study is limited due to region of west Africa are more likely to present for
sample size and that further research involving signifi- symptomatic reasons than for routine screening.
cantly larger populations should be conducted regarding Women in sub-Saharan Africa face many unique bar-
the risks and benefits of early screening specific to women riers to breast cancer screening. A previous study con-
living in sub-Saharan Africa. Going forward, women in ducted on breast cancer awareness reported that one in four
Nigeria and sub-Saharan Africa at large would be better women in Nigeria and Uganda had no breast cancer
served with screening guidelines that are based on research awareness.19 A meta-analysis of African studies regarding
conducted in such populations rather than applying delayed presentation and diagnosis of breast cancer found
guidelines from the Western world, where breast cancer that late presentation to a person’s first health care provider
may present and progress differently. visit was associated with socioeconomic factors, lack of
Breast lump/mass was the most cited reason given by awareness, poor knowledge of early detection methods,
patients who visited our facility for mammographic eval- not taking initial symptoms seriously, fear of disease and
uation. Breast lump/mass was also a frequent indication fear of treatment such as mastectomy, preference for
for mammography in other studies conducted in Port spiritual/traditional cures, and poor access.20 This study of
Harcourt, Nigeria and at a multicenter study in the United 2788 patients reported an estimated 4-15 month delay
States.12,13 The high frequency of patients presenting for between recognition of symptoms and diagnosis in women
mammographic evaluation due to breast mass/lump across in North Africa and sub-Saharan Africa, compared to an
these various studies can be attributed to the fear patients average delay of 48 days in the United States.20
have when they discover a breast mass/lump during self- A BIRADS score of 4 in both breasts was the most
examination and also the lack of screening programs prevalent in our study at 96 patients (31.6%) followed by a
where patients with early cancers could be detected.12 This BIRADS score of 1 at 62 patients (20.4%). This means
fear of potential breast cancer makes them determined to that our study involved many patients whose findings were
seek timely mammographic evaluation. suspicious for malignancy. Other studies in Nigeria have
Breast pain was the second most common indication for reported different distributions of mammograms with
mammographic evaluation among our patient cohort. BIRADS classifications at risk of malignancy (BIRADS
Studies in Southwestern, Northern and Eastern Nigeria 4þ), including reports of 16% and 10.9% for studies that

Fig. 4. (A&B) (Left-Right): Mediolateral Oblique (MLO) and Craniocaudal (CC) Mammograms showing linear vascular (arrows) and popcorn (arrowheads)
calcifications. These are BIRADS 2 (Benign) lesions.
included both patients with clinician referrals and patients towards more benign findings such as BIRADS 1. The
presenting for routine screening.14,16 In these studies, difference compared to the retrospective cohort study in
54.7% and 32.3% of total patients had normal findings on Enugu, Nigeria is probably due to different patient pop-
mammographic evaluation.14,16 A retrospective cohort ulations. The patient population in Enugu, Nigeria is likely
study in Enugu, Nigeria showed that a BIRADS score of to have involved mostly natives of the region whereas our
1 at 151 patients (36.3%) was the most prevalent in their patient population in the cosmopolitan megacity of Lagos,
study.21 Another retrospective study in Benin City and Nigeria is likely to be more heterogeneous with patients
Lagos, Nigeria also found BIRADS 1 to be the most coming from all over Nigeria.
prevalent.22 The significant difference in their results Univariate logistic regression showed that those pa-
compared to ours is due to various reasons. Firstly, the tients who had skin thickening and nipple retraction were
study in Benin City and Lagos involved only routine statistically more likely to have lesions suspicious of ma-
screening mammography which likely skewed their results lignancy. This correlation was also found in a similar study

Fig. 5. (A&B) (Left-Right): Mediolateral Oblique (MLO) and Craniocaudal (CC) Mammograms (A & B) showing a BIRADS 4c spiculated radiopaque lesion
(arrows) in the inferomedial portion of the left breast with associated skin retraction and overlying skin thickening (arrowheads).
in Lucknow, India which found that nipple retraction and poorly-defined margins and irregular outlines were three
skin thickening were statistically associated with malig- times more likely to have lesions assessed as suspicious of
nancy in a patient cohort of Indian women.23 Univariate malignancy based on multivariate logistic regression
logistic regression analysis also showed that women with a (P ¼ 0.013; OR 3.36, 95% CI 1.30-8.69). Univariate anal-
breast mass were four times more likely to have lesions ysis showed that poorly defined margins had no correlation
suspicious for malignancy (BIRADS 4/5) (P ¼ 0.0003; with a BIRADS 4/5 rating. This difference between the two
OR 3.86; 95% CI 1.85-8.06). A multi-center study in the models may be due to the variability seen in the model when
United States also concluded that women presenting with poorly defined margins are paired with the other covariates
breast masses were more likely to be assessed with a versus by itself. The correlation found in our multivariate
higher BIRADS score and increased suspicion for malig- model is supported by the findings of similar studies. In a
nancy.13 Multivariate logistic regression did not show a similar study in India, poorly defined margins were seen in
significant correlation between breast mass/lump and a the most malignant cases (55.3%) compared to the most
higher BIRADS score and this finding is probably due to benign cases which were more likely to have well-defined
our smaller sample size derived from a single center. We margins (70.6%).23 This difference between malignant and
agree with other studies that breast mass/lump is a sig- benign cases was found to be statistically significant with
nificant factor in having a higher BIRADS score respect to the pattern of margins (P < 0.001).23
assessment. Radiological findings such as opacity and lucency of
Univariate analysis showed that patients who presented the breast mass had a correlation with whether the masses
for routine mammographic screening were statistically less were more likely to be assessed as benign or malignant. In
likely to have lesions suspicious for malignancy our study univariate regression demonstrated that more
(P ¼ 0.004; OR 0.40, 95% CI 0.15-0.58). This inverse dense or radiopaque masses had a significant association
correlation should be used by healthcare stakeholders in with a BIRADS score of 4 or 5 (P 0.0001; OR ¼ 2.77,
Nigeria and the wider region of sub-Saharan Africa to 95% CI 1.65-5.14), compared to less dense radiolucent
invest resources into ensuring that more women visit masses (P ¼ 0.06; OR 0.39, 95% CI 0.15-3.0). Masses
radiodiagnostic centers for routine screening instead of shown to be high density on imaging have been previously
visiting only after they have concerning breast symptoms. correlated with a 70% rate of malignancy, while those with
Previous research has reported that noncalcified cancers low density resulted in 22% occurrence of malignancy.25
present as spiculated masses 33% of the time, irregularly Heterogeneously dense breast tissue (breast composi-
outlined masses 25% of the time, and 25% as less defined tion C) at 64.1% was the most predominant density in our
round masses.24 In our study, patients assessed to have study. This finding is in concordance with a similar study

Table 4. Univariate logistic regression for explanatory variables and BIRADS rating 4/5, Odds Ratio and 95% CI.
Odds ratio 95% CI P-value

Breast composition C or D 0.81 0.37-1.17 0.58
Poorly defined margins 3.00 0.90-10.0 0.073
Radiological findings (RF) 2.23 1.32-3.78 0.003
Radiopaque mass 2.77 1.65-5.14 <.0001
Radiolucent mass 0.06 0.15-3.0 0.39
Architectural distortion 0.80 0.16-3.97 0.79
Skin thickening 3.28 1.55-6.94 0.002
Nipple retraction 2.76 1.14-6.69 0.025
Calcifications 0.64 0.07-6.23 0.70
Axillary lymphadenopathy 0.43 0.19-0.98 0.044
Age > 50 1.64 1.01-2.67 0.044
Patient history
Routine screen 0.30 0.15-0.58 0.0004
Pain 1.18 0.51-2.70 0.70
Discharge 1.57 0.50-4.96 0.44
Pain and discharge 1.22 0.12-12.83 0.87
Breast mass/lump 3.86 1.85-8.06 0.0003
Breast cancer 2.62 0.70-9.74 0.15
Table 4 shows the direct relationship between each of the explanatory variables in the first column and a BIRADS score of 4 or 5. The
significance or non-significance of the relationship was determined using a univariate logistic regression analysis with P 0.05
representing a significant relationship/association.
carried out in Ado-Ekiti, Nigeria which found heteroge- Various other factors have been reported to influence
neously dense breast tissue to be the most predominant breast density. None of the explanatory variables in our
breast density pattern in their study.26 Setting breast multivariate logistic regression model had a statistically
composition categories C and D (higher breast densities) significant correlation with breast composition C and D
as explanatory variables in our multivariate and univariate (higher breast density). Age is a factor that has been shown
models, we found that higher breast density did not have a to have correlation with breast density in prior research
statistically significant correlation with having a breast studies. A similar study at the State University of New
mass suspicious for malignancy. The inference from our York at Buffalo found that parenchymal density on
statistical analysis is different from the established positive mammograms decreased progressively in their patient
correlation between higher density and malignancy found cohorts as the patients’ ages increased with a Spearman’s
in prior literature. Breast density has been related to breast correlation P < 0.010.31 In our patient population, 91.8%
malignancy as higher breast density has been shown to be of women less than 50 were found to have a breast density
a strong risk factor for breast cancer.27 The higher fre- of C or D, compared to 87.0% of women above 50. We
quency of breast malignancy in women with higher breast found no statistical significance in differences regarding
density can be explained by molecular differences between breast density based on age, or in regards to BIRADS
dense and non-dense breast tissues.28 Previous studies scores of 4 or 5. While this study did not find a correlation
have suggested that women of African descent have higher between age and breast density, it would be imperative for
breast area density compared to women of European future multicenter studies across Nigeria and sub-Saharan
descent, however, studies controlling for confounders such Africa to examine this relationship. The evaluation of
as age, BMI, and reproductive factors failed to find a breast density patterns in our patient cohort is important as
statistically significant difference.29,30 it can help health policy makers in the region in making

Table 5. Multivariate logistic regression for explanatory variables and BIRADS rating 4/5, Odds Ratio and 95% CI.
Odds Ratio 95% CI P-Value

Breast composition C or D 0.55 0.23-1.31 0.18
RF: Radiopaque mass 1.67 0.74-3.77 0.22
RF: Radiolucent mass 0.36 0.081-1.62 0.18
Architectural distortion 0.075 0.005-1.16 0.064
Skin thickening 2.054 0.45-9.38 0.35
Calcifications 0.84 0.053-13.37 0.90
Age > 50 1.41 0.79-2.52 0.25
Reason: routine screen 0.27 0.054-1.34 0.11
Reason: mass 1.30 0.28-6.04 0.731
Reason: pain 0.73 0.143-3.70 0.14
Reason: discharge 0.73 0.14-3.69 0.70
Reason: breast cancer likelihood 0.53 0.063-4.50 0.56
Table 5 shows the direct relationship between each of the explanatory variables in the first column and a BIRADS score of 4 or 5 when
accounting for possible cofounding variables. The significance or non-significance of the relationship was determined using a
multivariate logistic regression analysis with P 0.05 representing a significant relationship/association.
Table 6. Multivariate logistic regression for explanatory variables and breast composition C/D Odds Ratio and 95% CI.
Odds Ratio 95% CI P-Value

BIRADS 4 or 5 0.53 0.22-1.26 0.15
RF: radiopaque mass 0.79 0.23-2.68 0.71
RF: radiolucent mass 0.56 0.061-5.16 0.61
Skin thickening 0.63 0.075-5.28 0.67
Age > 50 0.60 0.27-1.36 0.22
Reason: routine screen 1.086 0.22-5.36 0.92
Reason: mass 1.48 0.33-6.60 0.61
Reason: pain 0.53 0.15-1.96 0.34
Reason: discharge 0.42 0.10-1.76 0.24
Table 6 shows the direct relationship between each of the explanatory variables in the first column and a Breast Composition C or D
when accounting for possible cofounding variables. The significance or non-significance of the relationship was determined using a
multivariate logistic regression analysis with P 0.05 representing a significant relationship/association.

guidelines for mammographic exams that are tailored to- potential multi-center assessment can help in formulating
wards Nigerian women and the sub-Saharan African new screening guidelines that are specific to this patient
population at large. population. Breast lump/mass had was the most common
presenting complaint among the patients visiting our fa-
Strengths and limitations cility. Our study found that women coming in for routine
To date, this is the largest study exploring mammographic screenings were less likely to present with lesions suspi-
findings and breast abnormalities in the cosmopolitan cious for malignancy. This suggests that more awareness
environment of Lagos, Nigeria. The patient population in initiatives are needed to encourage routine mammographic
Lagos, Nigeria is a better representation of Nigerian screening in Nigerian women. Heterogeneous breast
women nationally compared to other previous regional density was the most common breast density pattern in our
studies. Lagos is the main business hub of Nigeria and it is study, but age was not significantly correlated with
inhabited by Nigerians of various ethnic groups. While this increased breast density. While the patient population in
study involves a population of Nigerian women undergo- this study was comprised of Nigerian women, under-
ing screening and diagnostic mammography, the findings standing the mammographic patterns and correlations with
in this study may be helpful to clinicians as they encounter breast abnormalities can potentially help medical practi-
Nigerian patients or patients of sub-Saharan African tioners across the world when evaluating breast abnor-
descent, wherever they practice internationally. malities in patients of a similar national or ethnic
There were several limitations to our study. Although background as our patient population.
Lagos is a major cosmopolitan African city, the patient
cohort in this study is still not a full representation of
REFERENCES
Nigerian women. Therefore, it would be imperative to
1. Arowolo, O. A., Akinkuolie, A. A., Lawal, O. O., Alatise, O. I.,
have future multicenter studies across Nigeria and sub-
Salako, A. A., & Adisa, A. O. (2010). The impact of neoadjuvant
Saharan Africa. This would allow more robust results
chemotherapy on patients with locally advanced breast
that can help in formulating guidelines regarding
cancer in a Nigerian Semiurban teaching hospital: a single-
mammographic evaluations in women across the region. center descriptive study. World J Surg, 34(8), 1771e1778.
We also did not evaluate for more specific descriptions of
some abnormal findings in our study. For example, future 2. Azubuike, S. O., Muirhead, C., Hayes, L., & McNally, R. (2018).
Rising global burden of breast cancer: the case of Sub-Saharan
studies could evaluate how different types of margins
Africa (with emphasis on Nigeria) and implications for regional
(angular, spiculated, microlobulated), calcifications
development: a review. World J Surg Oncol, 16(1), 63.
(comedo, nonlinear irregular, vascular, large rod-like,
round, eggshell/rim), location of mass (subareolar, cen- 3. Lawal, O., et al. (2015). “Mammography screening in Nigeria e
a critical comparison to other countries. Radiography, 21(4),
tral, axillary), depth of mass, and shape (oval, round,
348e351.
irregular) correlate with breast malignancy or higher
BIRADS scores in a similar population. Furthermore, 4. Brinton, L. A., Figueroa, J. D., Awuah, B., et al. (2014). Breast
future studies could explore how these findings correlate cancer in Sub-Saharan Africa: opportunities for prevention.
with breast malignancy confirmed via biopsy and more Breast Cancer Res Treat, 144(3), 467e478.
specific types of breast cancer since we did not have access 5. Omidiji, O. A., Campbell, P. C., Irurhe, N. K., Atalabi, O. M., &
to the patients’ histopathological reports. Toyobo, O. O. (2017). Breast cancer screening in a resource
poor country: ultrasound versus mammography. Ghana Med J,
51(1), 6e12.
CONCLUSION
6. Armstrong, K., Moye, E., Williams, S., Berlin, J. A., & Reynolds, E. E.
This study has demonstrated the importance of the utili-
(2007). Screening mammography in women 40 to 49 years of
zation of mammography for screening and diagnosing age: a systematic review for the American College of physi-
breast abnormalities in sub-Saharan Africa. Women 41-50 cians. Ann Intern Med, 146(7), 516.
years of age group was the most common group visiting
7. Mushlin, A. I., Kouides, R. W., & Shapiro, D. E. (1998). Estimating
our facility for screening and diagnostic mammography
the accuracy of screening mammography. Am J Prev Med,
and had the largest number of patients with BIRADS
14(2), 143e153.
scores of 4 or 5. Although there was a positive correlation
between patients with ages >50 years and BIRADS 4/5 8. Spak, D. A., et al. (2017). BI-RADS ® fifth edition: a summary of
on univariate analysis, more multicenter studies with changes. Diagn Interv Imaging, 98(3), 179e190.
larger sample sizes are needed in the region to further 9. Akinola, R. A., Akinola, O. I., Shittu, L. A. J., Balogun, B. O., &
assess this relationship on a multivariate level. This Tayo, A. O. (2007). Appraisal of mammography in Nigerian

women in a new teaching hospital. Sci Res Essays, 2(8), 20. Espina, C., McKenzie, F., & Dos-Santos-Silva, I. (2017). Delayed
325e329. presentation and diagnosis of breast cancer in African women:
a systematic review. Ann Epidemiol, 27(10), 659e671.e7.
10. Brakohiapa, E. K., Armah, G. E., Clegg-Lamptey, J. N. A., & Bra-
kohiapa, W. O. (2013). Pattern of breast diseases in Accra: review 21. Nwadike, U. I., Eze, C. U., Agwuna, K., & Mouka, C. (2018).
of mammography reports. Ghana Med J, 47(3), 101e106. Mammographic classification of breast lesions amongst
women in Enugu, South East Nigeria. Afr Health Sci, 17(4),
11. Nelson, H. D., Fu, R., Cantor, A., et al. (2016). Effectiveness of
1044.
breast cancer screening: systematic review and meta-analysis
to update the 2009 U.S. Preventive Services task force recom- 22. Akhigbe, A. O., & Omuemu, V. O. (2009). Knowledge, attitudes
mendation. Ann Intern Med, 164, 244. and practice of breast cancer screening among female
health workers in a Nigerian Urban city. BMC Cancer, 9, 203.
12. Onwuchekwa, C. R., & Alazigha, N. S. (2018). Mammographic
findings among women in Port Harcourt: a multicentre study. 23. Khanduri, S., Chaudhary, M., Sabharwal, T., et al. (2017). A low-
Niger Health J, 18(1), 1e7. cost, low-skill model for efficient breast cancer screening in low
resource rural Settings of a developing country. Cureus, 9(8),
13. Berta, M. G., William, E. B., Rachel, B. B., et al. (2002). Use of the
e1571.
American College of Radiology BI - RADS to report on the
mammographic evaluation of women with signs and symp- 24. Stomper, P. C. (2000). Breast imaging. In D. F. Hayes (Ed.), Atlas
toms of breast disease. Radiology, 222, 536e542. of Breast Cancer (54). Philadelphia: Mosby.
14. Akande, H. J., Olafimihan, B. B., & Oyinloye, O. I. (2015). A five 25. Woods, R. W., Sisney, G. S., Salkowski, L. R., et al. (2011). The
year audit of mammography in a tertiary hospital, North Cen- mammographic density of a mass is a significant predictor of
tral Nigeria. Niger Med J, 56(3), 213e217. breast cancer. Radiology, 258, 417.
15. Danfulani, M. (2014). Pattern of mammographic findings in 26. Aduayi, O. S., Akanbi, G. O., & Aduayi, V. A. (2016). Introducing
Sokoto, Nigeria. Asian J Med Sci, 5(4), 79e83. digital mammography in a resource constrained setting:
16. Akinola, R., Akinola, O., & Tayo, A. (2007). Appraisal of spectrum of imaging findings and diagnostic value in Ado-ekiti,
mammography in Nigerian women in a new teaching hospital. South Western Nigeria. Int J Med Imag, 4(2), 7e11.
Sci Res Essays, 2(8), 325e329. http://www.academicjournals.org/ 27. Kang, Y. J., Ahn, S. K., Kim, S. J., Oh, H., Han, J., & Ko, E. (2019).
app/webroot/article/article1380272448_Akinola%20et%20al. Relationship between mammographic density and age in the
pdf. Accessed December 18, 2019. United Arab Emirates population. J Oncol, 2019, 7351350.
17. Ebubedike, U. R., Umeh, E. O., Anyanwu, S. N., Ukah, C. O., & 28. Nazari, S. S., & Mukherjee, P. (2018). An overview of mammo-
Ikegwuonu, N. C. (2016). Pattern of mammography findings graphic density and its association with breast cancer. Breast
among symptomatic females referred for diagnostic Cancer, 25(3), 259e267.
mammography at a Tertiary Center in South-East Nigeria. W Afr
29. El-Bastawissi, A. (2001). Variation in mammographic breast
J Radiol, 23(1), 23.
density by race. Ann Epidemiol, 11(4), 257e263.
18. Adesunkanmi, A. R. K., Lawal, O. O., Adelusola, K. A., & Dur-
30. McCarthy, A. M., Keller, B. M., Pantalone, L. M., et al. (2016).
osimi, M. A. (2006). The severity, outcome and challenges of
Racial differences in quantitative measures of area and volu-
breast cancer in Nigeria. Breast, 15(3), 399e409.
metric breast density. J Natl Cancer Inst, 108(10), djw104.
19. McKenzie, F., Zietsman, A., Galukande, M., et al. (2018). Breast
cancer awareness in the Sub-Saharan African ABC-DO cohort: 31. Stomper, P. C., D’souza, D. J., DiNitto, P. A., et al. (1996).
African breast cancerddisparities in outcomes study. Cancer Analysis of parenchymal density on mammograms in 1353
Causes Control, 29(8), 721e730. women 25-79 years old. Am J Roentgenol, 167(5), 1261e1265.


Mammo JNMA

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mammo JNMA

Uploaded by

Copyright:

Available Formats

R E S E A R C H P A P E R

Relationship between Mammographic

has resulted in a decreased incidence of locally advanced

JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION VOL -, NO -, - 2020 1

describe the pattern of mammographic ﬁndings in a pop- Statistical analysis

2 VOL. -, NO -, - 2020 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION

Fig. 1. Showing the frequency distribution of patients in each age group.

tectural distortion (Figure 5). Skin thickening was assessed

Respondent’s history N (%) BIRADS 5 8 (2.6%)

Routine screen 73 (24.5%) BIRADS 0 54 (17.8%)

Pain 74 (24.8%) BIRADS 0 (Left Breast) BIRADS 1 (Right 4 (1.3%)

JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION VOL -, NO -, - 2020 3

A similar model was used to evaluate these character-

4 VOL. -, NO -, - 2020 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION

JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION VOL -, NO -, - 2020 5

6 VOL. -, NO -, - 2020 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION

JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION VOL -, NO -, - 2020 7

Odds ratio 95% CI P-value

8 VOL. -, NO -, - 2020 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION

Odds Ratio 95% CI P-Value

Odds Ratio 95% CI P-Value

JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION VOL -, NO -, - 2020 9

10 VOL. -, NO -, - 2020 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION

JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION VOL -, NO -, - 2020 11

You might also like