Abstract
Breast cancer incidence is rising in Africa, but there are scare data regarding risk factors in this region. We assessed the relation between risk factors and the occurrence of breast cancer, overall and by tumor subtype in women from Mozambique.
The associations between education, number of births, height, weight, body mass index (BMI), and breast cancer risk among 138 cases (participants from the Moza-BC cohort) and 638 controls from the general population (from a World Health Organization stepwise approach to surveillance survey), recruited during 2014 to 2017, were investigated. Adjusted ORs (aOR) and 95% confidence intervals (CI) were estimated using multivariable logistic regression.
Multiparity (≥6 vs. 0–1 live births) was a protective factor for the development of hormone receptor (HR)–positive (aOR = 0.22; 95% CI, 0.08–0.64) and HR-positive/HER2-negative tumors (aOR = 0.20; 95% CI, 0.06–0.68), whereas a higher educational level (≥8 vs. 0 schooling years) increased breast cancer risk across all subtypes (overall aOR = 1.98; 95% CI, 1.04–3.80). Higher weight and BMI were associated with a higher breast cancer risk among postmenopausal women (per 1-kg increase: aOR = 1.05; 95% CI, 1.02–1.08; per 1-kg/m2 increase: aOR = 1.11; 95% CI, 1.04–1.18, respectively), but were protective in premenopausal women (aOR = 0.98; 95% CI, 0.96–0.99; aOR = 0.95; 95% CI, 0.91–0.99, respectively), regardless of subtype. Higher height increased the risk of HR-negative tumors in postmenopause (per 10-cm increase: aOR = 2.81; 95% CI, 1.41–6.03).
These results demonstrate the etiological heterogeneity of breast cancer among native African women, namely regarding the differential effect of multiparity, education, and body parameters in breast cancer risk.
As the prevalence of obesity grows, these findings are important to inform public health policies on cancer prevention, by highlighting obesity as a modifiable risk factor for breast cancer among African women.
Introduction
The understanding of risk factors for the development of breast cancer has evolved over the past two decades, along with the discovery of the different breast cancer subtypes (1, 2). The effect of reproductive behaviors (e.g., parity, age at first birth, breastfeeding) and age at menarche/menopause may vary among these subtypes, being more strongly associated with hormone receptor (HR)–positive tumors (3, 4). The effects of body parameters on breast cancer risk may be modified by menopausal status, with higher weight and body mass index (BMI) appearing to be associated with a lower risk among premenopausal women and with an increased risk after menopause (5). Yet, their effect across breast cancer subtypes is still debated.
There are important differences in the overall incidence of breast cancer between Western and Sub-Saharan African countries (age-standardized incidence rate of >80.0 vs. <50.0/100,000 women, respectively; ref. 6) and in the distribution of tumor subtypes [e.g., 10%–15% vs. >20% triple-negative breast cancer (TNBC), respectively; refs. 7, 8]. Therefore, as the incidence and mortality from breast cancer are rising in Africa (6, 9), understanding the risk factors for the overall development of breast cancer in this region is of particularly importance.
Compared with Western countries, African patients have a different expression of risk factors, such as higher parity, longer breastfeeding patterns, or a lower prevalence of obesity, which could partly explain these differences (10). Multiple Sub-Saharan African reports have shown that multiparity protects from breast cancer, overall, but without an assessment by subtype (10–13). Some studies demonstrated that a higher educational level increased breast cancer risk (11, 13), whereas others suggested it to be a protective factor (12, 14). Data are also contradictory regarding body parameters: a report found that a higher BMI was a risk factor among postmenopausal women (14), whereas in others, it appeared to have no effect (15, 16). The specificities of Sub-Saharan African populations in terms of the distribution of potential risk factors and subtypes may allow for a better understanding of the etiology of each breast cancer subtype; however, this has never been explored, probably due to the difficulties in routinely determining breast cancer subtypes in this region (17). This case–control study aimed to assess the relation between factors, such as education, parity, BMI, height and weight, and the overall development of breast cancer as well as of each tumor subtype in African women from Mozambique.
Mozambique is a low-income country located in Southeast Sub-Saharan Africa, with a population of 30 million people (18) and a fertility rate close to five births/woman (19). Education is mandatory until the ninth grade, but even nowadays, only 45% of students complete the first four years of basic education (20). One in eight adults ages 15 to 49 years is infected by HIV (21), and noncommunicable diseases are also on the rise (22, 23). For instance, between 2005 and 2014/2015, the prevalence of overweight/obesity increased from 18.3% to 30.5% among adult women (24). Breast cancer is now the most common non-AIDS defining malignancy among Mozambican women, with an incidence rate of 8.6/100,000 women in Maputo City (9, 25). National guidelines for early breast cancer detection based on breast self-exam and clinical exam were implemented in 2010 (26), but there are no organized mammography screening programs, and women are usually diagnosed with advanced disease (stage III/IV; ref. 27).
Materials and Methods
This case–control analysis included patients with breast cancer recruited for the Mozambican Breast Cancer (Moza-BC) cohort and controls from the general population selected among the participants of the “World Health Organization stepwise approach to surveillance of noncommunicable disease risk factor” (WHO-STEPS) Survey.
Source of breast cancer cases: Moza-BC cohort
Cases were selected among women with breast cancer enrolled in the prospective Moza-BC cohort, previously described elsewhere (27). Briefly, it included a consecutive series of women with newly diagnosed primary breast cancer, with a pathologic diagnosis performed at one of the three Central Hospitals in Mozambique, from January 2015 to August 2017, which were the only hospitals with a Pathology Department at the time. Breast tissue samples were collected by fine-needle aspiration cytology (FNAC), surgical biopsy, breast surgery, or core-needle biopsy. Aspirated material from FNAC was used to create cell blocks, and histologic samples were handled according to the recommendations of the American Society of Clinical Oncology (ASCO)/College of American Pathologists (CAP; ref. 28). Immunostaining of histologic samples and cell blocks for estrogen receptor (ER), progesterone receptor (PR), and HER2 was carried and assessed as described in the literature (28, 29). If the IHC score was 0 or 1+, the tumor was considered to be “HER2-negative”; if it was 3+, it was classified as “HER2-positive.” In the case of an HER2 IHC score of 2+ (HER2-equivocal), samples were submitted to in situ hybridization.
Tumors were grouped according to their HR status as HR-positive (ER-positive or PR-positive, defined as an expression of ER or PR ≥ 1%; ref. 28) or HR-negative (ER-negative and PR-negative), regardless of the tumor's HER2 status; and then according to the classic tumor subtypes classified into HR-positive/HER2-negative, HER2-positive (HR-positive/HER2-positive and HR-negative/HER2-positive), and TNBC (HR-negative/HER2-negative).
Patients subsequently followed at the Maputo Central Hospital were interviewed regarding their place of residence, education, age of menarche/menopause and menopausal status, age at first live birth, number of live births, and family history of breast cancer. Study investigators assessed height and weight before the start of chemotherapy, to the nearest 0.1 kg or 0.1 cm, respectively, with participants in a standing position and wearing light clothing and no footwear.
Source of controls: WHO-STEPS survey
Controls were selected from among participants of the WHO-STEPS survey, conducted in Mozambique between December 2014 and February 2015, as previously described (23, 24, 30). This population-based survey assessed a representative sample of the Mozambican population aged 15 to 64 years (n = 3119) for risk factors for noncommunicable diseases. Participants were evaluated using WHO-STEPS–standardized methods. Data regarding place of residence, age, education, menopausal status, and number of live births were collected by face-to-face interviews. Anthropometric measurements were obtained by investigators, with the participants wearing light clothing and no footwear. Body weight was measured to the nearest 0.1 kg and height to the nearest 0.1 cm in the standing position.
Case–control design and statistical analysis
Sociodemographic and clinical characteristics of the participants with breast cancer were described for a total of 205 cases. From these, 138 were cases younger than 65 years (the WHO-STEPS survey only included potential controls younger than 65 years) who were dwellers in Maputo (City or Province) or in the south and center provinces closest to Maputo that had at least five cases (Gaza, Inhambane, and Sofala), and thus were included in the main case–control analyses. A total of 118 of these cases also had information on the breast cancer subtype and were therefore considered for the analyses by subtype.
A total of 638 participants in the WHO-STEPS survey were selected as controls, corresponding to all nonpregnant women who lived in the same provinces of the cases considered for the case–control analyses (Fig. 1).
Education, number of live births, height, weight, and BMI were selected as the potential risk factors to be analyzed in this study, given their availability in both the Moza-BC cohort and the WHO-STEPS survey. Height and weight were used to compute BMI, and participants were classified as obese if their BMI was ≥30 kg/m2. Cases were compared across tumor subtype subgroups, namely HR status and classic subtypes, using ANOVA for continuous variables, and χ2 or Fisher exact tests for categorical variables. Case–control analyses were conducted for all cases and according to subtype. Adjusted ORs (aOR) and corresponding 95% confidence intervals (CI) were estimated using multivariable logistic regression. Interactions were tested by including the interaction terms between menopausal status and body parameters in the logistic regression models, and stratum-specific estimates were obtained through the linear combination of coefficients. All tests were two-sided, and a P value of <0.05 was considered significant. Analyses were carried out using STATA version 15 (StataCorp). This study is reported according to the STROBE statement (31).
Ethics
The National Health Bioethical Committee of Mozambique approved the Moza-BC cohort study (226/CNBS/15) and the WHO-STEPS survey (98/CNBS/14). All participants provided written informed consent, and the study was conducted according to the Declaration of Helsinki recommendations.
Results
Cases
Overall, 64% of cases had HR-positive tumors; regarding classic subtypes, 52% had HR-positive/HER2-negative tumors, 25% HER2-positive, and 23% TNBC. Women with breast cancer had a mean age of 49.3 years (range, 21–87) and were mostly premenopausal (53.7%), parous (93.6%), black (98.0%), and dwellers in the South of the country (90.7%). Approximately a third of the cases were obese, and about a quarter had no formal education (Table 1).
. | . | HR status . | Classic tumor subtypes . | |||
---|---|---|---|---|---|---|
. | All cases (n = 205) . | HR− (n = 63) . | HR+ (n = 112) . | HR+/HER2− (n = 91) . | HER2+ (n = 44) . | TNBC (n = 40) . |
. | n (%)a . | n (%)a . | n (%)a . | n (%)a . | n (%)a . | n (%)a . |
Age (years) | ||||||
21–39 | 54 (26.3) | 16 (25.4) | 28 (25.0) | 21 (23.1) | 12 (27.3) | 11 (27.5) |
40–49 | 57 (27.8) | 15 (23.8) | 34 (30.4) | 28 (30.8) | 11 (25.0) | 10 (25.0) |
50–64 | 66 (32.2) | 25 (39.7) | 35 (32.2) | 28 (30.8) | 17 (38.6) | 15 (37.5) |
65–87 | 28 (13.7) | 7 (11.1) | 15 (13.4) | 14 (15.4) | 4 (9.1) | 4 (10.0) |
Race (black) | 201 (98.0) | 201 (100) | 109 (97.3) | 89 (97.8) | 43 (97.7) | 40 (100) |
Place of residence | ||||||
Maputo (city and province) | 153 (74.6) | 51 (81.0) | 80 (71.4) | 68 (74.7) | 30 (68.2) | 33 (82.5) |
Provinces closest to Maputob | 33 (16.1) | 6 (9.5) | 23 (20.5) | 15 (16.5) | 10 (22.7) | 4 (10.0) |
Provinces farthest from Maputoc | 19 (9.3) | 6 (9.5) | 9 (8.0) | 8 (8.8) | 4 (9.1) | 3 (7.5) |
Education, complete school years | ||||||
0 | 41 (23.7) | 8 (14.6) | 24 (26.4) | 20 (27.8) | 7 (18.0) | 5 (14.3) |
1–7 | 65 (37.6) | 23 (41.8) | 29 (31.9) | 24 (33.3) | 11 (28.2) | 17 (48.6) |
≥8 | 67 (38.7) | 24 (43.6) | 38 (41.8) | 28 (38.9) | 21 (53.8) | 13 (37.1) |
Missing | 32 | 8 | 21 | 19 | 5 | 5 |
Family history of breast cancer (yes) | 8 (3.9) | 1 (1.6) | 5 (4.5) | 4 (4.4) | 2 (4.5) | 0 |
Number of live births | ||||||
0 | 13 (6.4) | 2 (3.3) | 7 (6.3) | 7 (7.7) | 1 (2.4) | 1 (2.6) |
1 | 16 (7.9) | 3 (4.9) | 11 (9.9) | 8 (8.8) | 3 (7.1) | 3 (7.7) |
2–3 | 73 (36.1) | 23 (37.7) | 45 (40.5) | 35 (38.5) | 19 (45.2) | 14 (35.9) |
4–5 | 53 (26.2) | 14 (23.0) | 27 (24.3) | 24 (26.4) | 8 (19.0) | 9 (23.1) |
≥6 | 47 (23.3) | 19 (31.2) | 21 (18.9) | 17 (18.7) | 11 (26.2) | 12 (30.8) |
Missing | 3 | 2 | 1 | 0 | 2 | 1 |
Age at first live birth, yearsd | 21.5 (4.6) | 21.1 (4.7) | 21.8 (4.6) | 21.8 (4.9) | 21.2 (3.5) | 21.5 (5.2) |
Not applicable | 13 | 2 | 7 | 7 | 1 | 1 |
Missing | 42 | 14 | 17 | 13 | 10 | 8 |
Age at menarche, yearsd | 15.1 (2.2) | 14.8 (2.1) | 15.1 (2.3) | 14.9 (2.1) | 14.9 (2.8) | 15.3 (1.7) |
Missing | 53 | 18 | 29 | 25 | 10 | 12 |
Menopause (yes) | 95 (46.3) | 32 (50.8) | 52 (46.4) | 44 (48.4) | 20 (45.4) | 20 (50.0) |
Age at menopause, yearsd | 49.3 (4.5) | 48.7 (5.0) | 49.9 (4.1) | 49.8 (4.4) | 49.9 (4.2) | 48.1 (5.0) |
Not applicable | 110 | 31 | 60 | 47 | 24 | 20 |
Missing | 22 | 6 | 13 | 12 | 3 | 4 |
Weight, kgd | 70.1 (17.7) | 73.8 (16.1) | 69.4 (19.0) | 68.3 (18.2) | 74.0 (19.1) | 65.5 (14.7) |
Missing | 17 | 6 | 9 | 7 | 5 | 3 |
Height, cmd | 158.0 (13.8) | 158.4 (10.0) | 158.6 (6.5) | 158.0 (6.6) | 161.0 (7.0) | 157.3 (10.7) |
Missing | 16 | 5 | 9 | 7 | 4 | 3 |
BMI, kg/m2 d | 27.8 (6.8) | 29.5 (6.5) | 27.5 (7.1) | 27.3 (6.9) | 28.5 (7.1) | 29.9 (6.8) |
Missing | 17 | 6 | 9 | 7 | 5 | 3 |
Stage at diagnosis | ||||||
0–IIe | 53 (25.9) | 22 (34.9) | 22 (19.6) | 15 (16.5) | 16 (36.4) | 13 (32.5) |
III | 110 (53.7) | 30 (47.6) | 69 (61.6) | 58 (63.7) | 23 (52.3) | 18 (45.0) |
IV | 42 (20.5) | 11 (17.5) | 21 (18.8) | 18 (19.8) | 5 (11.4) | 9 (22.5) |
. | . | HR status . | Classic tumor subtypes . | |||
---|---|---|---|---|---|---|
. | All cases (n = 205) . | HR− (n = 63) . | HR+ (n = 112) . | HR+/HER2− (n = 91) . | HER2+ (n = 44) . | TNBC (n = 40) . |
. | n (%)a . | n (%)a . | n (%)a . | n (%)a . | n (%)a . | n (%)a . |
Age (years) | ||||||
21–39 | 54 (26.3) | 16 (25.4) | 28 (25.0) | 21 (23.1) | 12 (27.3) | 11 (27.5) |
40–49 | 57 (27.8) | 15 (23.8) | 34 (30.4) | 28 (30.8) | 11 (25.0) | 10 (25.0) |
50–64 | 66 (32.2) | 25 (39.7) | 35 (32.2) | 28 (30.8) | 17 (38.6) | 15 (37.5) |
65–87 | 28 (13.7) | 7 (11.1) | 15 (13.4) | 14 (15.4) | 4 (9.1) | 4 (10.0) |
Race (black) | 201 (98.0) | 201 (100) | 109 (97.3) | 89 (97.8) | 43 (97.7) | 40 (100) |
Place of residence | ||||||
Maputo (city and province) | 153 (74.6) | 51 (81.0) | 80 (71.4) | 68 (74.7) | 30 (68.2) | 33 (82.5) |
Provinces closest to Maputob | 33 (16.1) | 6 (9.5) | 23 (20.5) | 15 (16.5) | 10 (22.7) | 4 (10.0) |
Provinces farthest from Maputoc | 19 (9.3) | 6 (9.5) | 9 (8.0) | 8 (8.8) | 4 (9.1) | 3 (7.5) |
Education, complete school years | ||||||
0 | 41 (23.7) | 8 (14.6) | 24 (26.4) | 20 (27.8) | 7 (18.0) | 5 (14.3) |
1–7 | 65 (37.6) | 23 (41.8) | 29 (31.9) | 24 (33.3) | 11 (28.2) | 17 (48.6) |
≥8 | 67 (38.7) | 24 (43.6) | 38 (41.8) | 28 (38.9) | 21 (53.8) | 13 (37.1) |
Missing | 32 | 8 | 21 | 19 | 5 | 5 |
Family history of breast cancer (yes) | 8 (3.9) | 1 (1.6) | 5 (4.5) | 4 (4.4) | 2 (4.5) | 0 |
Number of live births | ||||||
0 | 13 (6.4) | 2 (3.3) | 7 (6.3) | 7 (7.7) | 1 (2.4) | 1 (2.6) |
1 | 16 (7.9) | 3 (4.9) | 11 (9.9) | 8 (8.8) | 3 (7.1) | 3 (7.7) |
2–3 | 73 (36.1) | 23 (37.7) | 45 (40.5) | 35 (38.5) | 19 (45.2) | 14 (35.9) |
4–5 | 53 (26.2) | 14 (23.0) | 27 (24.3) | 24 (26.4) | 8 (19.0) | 9 (23.1) |
≥6 | 47 (23.3) | 19 (31.2) | 21 (18.9) | 17 (18.7) | 11 (26.2) | 12 (30.8) |
Missing | 3 | 2 | 1 | 0 | 2 | 1 |
Age at first live birth, yearsd | 21.5 (4.6) | 21.1 (4.7) | 21.8 (4.6) | 21.8 (4.9) | 21.2 (3.5) | 21.5 (5.2) |
Not applicable | 13 | 2 | 7 | 7 | 1 | 1 |
Missing | 42 | 14 | 17 | 13 | 10 | 8 |
Age at menarche, yearsd | 15.1 (2.2) | 14.8 (2.1) | 15.1 (2.3) | 14.9 (2.1) | 14.9 (2.8) | 15.3 (1.7) |
Missing | 53 | 18 | 29 | 25 | 10 | 12 |
Menopause (yes) | 95 (46.3) | 32 (50.8) | 52 (46.4) | 44 (48.4) | 20 (45.4) | 20 (50.0) |
Age at menopause, yearsd | 49.3 (4.5) | 48.7 (5.0) | 49.9 (4.1) | 49.8 (4.4) | 49.9 (4.2) | 48.1 (5.0) |
Not applicable | 110 | 31 | 60 | 47 | 24 | 20 |
Missing | 22 | 6 | 13 | 12 | 3 | 4 |
Weight, kgd | 70.1 (17.7) | 73.8 (16.1) | 69.4 (19.0) | 68.3 (18.2) | 74.0 (19.1) | 65.5 (14.7) |
Missing | 17 | 6 | 9 | 7 | 5 | 3 |
Height, cmd | 158.0 (13.8) | 158.4 (10.0) | 158.6 (6.5) | 158.0 (6.6) | 161.0 (7.0) | 157.3 (10.7) |
Missing | 16 | 5 | 9 | 7 | 4 | 3 |
BMI, kg/m2 d | 27.8 (6.8) | 29.5 (6.5) | 27.5 (7.1) | 27.3 (6.9) | 28.5 (7.1) | 29.9 (6.8) |
Missing | 17 | 6 | 9 | 7 | 5 | 3 |
Stage at diagnosis | ||||||
0–IIe | 53 (25.9) | 22 (34.9) | 22 (19.6) | 15 (16.5) | 16 (36.4) | 13 (32.5) |
III | 110 (53.7) | 30 (47.6) | 69 (61.6) | 58 (63.7) | 23 (52.3) | 18 (45.0) |
IV | 42 (20.5) | 11 (17.5) | 21 (18.8) | 18 (19.8) | 5 (11.4) | 9 (22.5) |
Abbreviations: BMI, body mass index; HR, hormone receptor; SD, standard deviation; TNBC, triple-negative breast cancer.
aN (%), except if otherwise specified.
bIncludes the south and center provinces of Inhambane, Gaza, and Sofala.
cIncludes Manica, Tete, Zambézia, Nampula, Niassa, and Cabo Delgado.
dMean (SD).
eAmong cases within the category 0–II stage, one patient had stage 0 and four patients had stage I.
Risk factors for breast cancer, overall
Women with 8 or more years of education had a higher risk of developing breast cancer compared with women without formal education (aOR 1.98; 95% CI, 1.04–3.80; Table 2). Women reporting six or more live births were less likely to develop breast cancer (aOR 0.41; 95% CI, 0.18–0.91) compared with women with none or one live births—who were combined to define the reference category, given the very low number of nulliparous women.
. | Controls (n = 638) . | Cases (n = 138) . | . | . | . |
---|---|---|---|---|---|
. | n (%)a . | n (%)a . | Adjusted OR . | (95% CI) . | Pb . |
Age (years)c | |||||
21–39 | 343 (53.8) | 41 (29.7) | 1 | (ref.) | NA |
40–49 | 126 (19.8) | 46 (33.3) | 3.49 | (2.12–5.72) | |
50–64 | 169 (26.5) | 51 (37.0) | 2.83 | (1.76–4.54) | |
Education (complete school years)d | |||||
0 | 164 (25.7) | 28 (20.3) | 1 | (ref.) | NA |
1–7 | 305 (47.8) | 54 (39.1) | 1.04 | (0.58–1.85) | |
≥8 | 169 (26.5) | 56 (40.6) | 1.98 | (1.04–3.80) | |
P for trend | 0.008 | ||||
Paritye | |||||
Nulliparous (0 live births) | 50 (7.8) | 10 (7.2) | 1 | (ref.) | NA |
Parous (≥1 live births) | 588 (92.2) | 128 (92.8) | 0.68 | (0.30–1.51) | |
Number of live birthse | |||||
0–1 | 128 (20.1) | 22 (15.9) | 1 | (ref.) | NA |
2–3 | 207 (32.4) | 55 (39.9) | 0.90 | (0.48–1.70) | |
4–5 | 158 (24.8) | 36 (26.1) | 0.66 | (0.32–1.35) | |
≥6 | 145 (22.7) | 25 (18.1) | 0.41 | (0.18–0.91) | |
P for trend | 0.015 | ||||
Height (per 10-cm increase)f—overallg | 159.0 (7.4) | 159.5 (7.0) | 1.33 | (0.99–1.79) | NA |
Among premenopausal women | 158.4 (7.4) | 159.5 (6.8) | 1.11 | (0.76–1.60) | 0.101 |
Among postmenopausal women | 157.1 (7.4) | 159.6 (7.5) | 1.87 | (1.13–3.10) | |
Weight (per 1-kg increase)f—overallh | 64.6 (15.4) | 71.9 (18.2) | 1.00 | (0.99–1.02) | NA |
Among premenopausal women | 64.4 (15.5) | 67.9 (17.8) | 0.98 | (0.96–0.99) | <0.001 |
Among postmenopausal women | 65.0 (15.0) | 78.5 (17.2) | 1.05 | (1.02–1.08) | |
BMI (per 1-kg/m2 increase)f—overallg | 25.8 (5.8) | 28.2 (6.8) | 1.00 | (0.97–1.03) | NA |
Among premenopausal women | 25.7 (5.9) | 26.6 (6.4) | 0.95 | (0.91–0.99) | <0.001 |
Among postmenopausal women | 26.3 (5.6) | 30.0 (6.7) | 1.11 | (1.04–1.18) | |
Obesity (BMI ≥ 30 kg/m2)—overallg | 137 (21.5) | 51 (37.0) | 1.00 | (0.63–1.61) | NA |
Among premenopausal women | 89 (18.6) | 25 (29.1) | 0.69 | (0.37–1.28) | 0.055 |
Among postmenopausal women | 48 (30.0) | 26 (50.0) | 1.73 | (0.84–3.60) |
. | Controls (n = 638) . | Cases (n = 138) . | . | . | . |
---|---|---|---|---|---|
. | n (%)a . | n (%)a . | Adjusted OR . | (95% CI) . | Pb . |
Age (years)c | |||||
21–39 | 343 (53.8) | 41 (29.7) | 1 | (ref.) | NA |
40–49 | 126 (19.8) | 46 (33.3) | 3.49 | (2.12–5.72) | |
50–64 | 169 (26.5) | 51 (37.0) | 2.83 | (1.76–4.54) | |
Education (complete school years)d | |||||
0 | 164 (25.7) | 28 (20.3) | 1 | (ref.) | NA |
1–7 | 305 (47.8) | 54 (39.1) | 1.04 | (0.58–1.85) | |
≥8 | 169 (26.5) | 56 (40.6) | 1.98 | (1.04–3.80) | |
P for trend | 0.008 | ||||
Paritye | |||||
Nulliparous (0 live births) | 50 (7.8) | 10 (7.2) | 1 | (ref.) | NA |
Parous (≥1 live births) | 588 (92.2) | 128 (92.8) | 0.68 | (0.30–1.51) | |
Number of live birthse | |||||
0–1 | 128 (20.1) | 22 (15.9) | 1 | (ref.) | NA |
2–3 | 207 (32.4) | 55 (39.9) | 0.90 | (0.48–1.70) | |
4–5 | 158 (24.8) | 36 (26.1) | 0.66 | (0.32–1.35) | |
≥6 | 145 (22.7) | 25 (18.1) | 0.41 | (0.18–0.91) | |
P for trend | 0.015 | ||||
Height (per 10-cm increase)f—overallg | 159.0 (7.4) | 159.5 (7.0) | 1.33 | (0.99–1.79) | NA |
Among premenopausal women | 158.4 (7.4) | 159.5 (6.8) | 1.11 | (0.76–1.60) | 0.101 |
Among postmenopausal women | 157.1 (7.4) | 159.6 (7.5) | 1.87 | (1.13–3.10) | |
Weight (per 1-kg increase)f—overallh | 64.6 (15.4) | 71.9 (18.2) | 1.00 | (0.99–1.02) | NA |
Among premenopausal women | 64.4 (15.5) | 67.9 (17.8) | 0.98 | (0.96–0.99) | <0.001 |
Among postmenopausal women | 65.0 (15.0) | 78.5 (17.2) | 1.05 | (1.02–1.08) | |
BMI (per 1-kg/m2 increase)f—overallg | 25.8 (5.8) | 28.2 (6.8) | 1.00 | (0.97–1.03) | NA |
Among premenopausal women | 25.7 (5.9) | 26.6 (6.4) | 0.95 | (0.91–0.99) | <0.001 |
Among postmenopausal women | 26.3 (5.6) | 30.0 (6.7) | 1.11 | (1.04–1.18) | |
Obesity (BMI ≥ 30 kg/m2)—overallg | 137 (21.5) | 51 (37.0) | 1.00 | (0.63–1.61) | NA |
Among premenopausal women | 89 (18.6) | 25 (29.1) | 0.69 | (0.37–1.28) | 0.055 |
Among postmenopausal women | 48 (30.0) | 26 (50.0) | 1.73 | (0.84–3.60) |
Abbreviations: BMI, body mass index; CI, confidence interval; NA, not applicable; OR, odds ratio; SD, standard deviation.
aN (%), except if otherwise specified.
bP for interaction.
cOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala).
dOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala), age (continuous), and number of live births (0–1, 2–3, 4–5, and ≥6 years).
eOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala), age (continuous), education (0, 1–5, 6–7, 8–10, 11–12, and ≥13 years), BMI (continuous), menopausal status (no/yes), and interaction term BMI*menopausal status.
hMean (SD).
gOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala), age (continuous), number of live births (0–1, 2–3, 4–5, and ≥6 years), and education (0, 1–5, 6–7, 8–10, 11–12, and ≥13 years).
hOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala), age (continuous), number of live births (0–1, 2–3, 4–5, and ≥6), education (0, 1–5, 6–7, 8–10, 11–12, and ≥13 years), and height (continuous).
Higher weight and BMI were associated with an increased risk of breast cancer among postmenopausal women, whereas they were protective among premenopausal women (P for interaction < 0.001). When analyzing height separately from BMI, there was an 87% overall increase in the risk of breast cancer for each 10-cm increase in height among postmenopausal women (aOR 1.87; 95% CI, 1.13–3.10).
Risk factors for breast cancer, by subtype
Higher education (≥8 years) was a risk factor for breast cancer across all tumor subtypes, but this association was stronger for TNBC (Table 3). Higher parity was a protective factor, but only for HR-positive tumors overall (≥6 vs. 0–1 live births: aOR 0.22; 95% CI, 0.08–0.64; P for trend = 0.003) and, more specifically, for HR-positive/HER2-negative tumors (aOR 0.20; 95% CI, 0.06–0.68; P for trend = 0.007); whereas no association with the development of HR-negative tumors overall (aOR 0.91; 95% CI, 0.24–3.51; P for trend = 0.543) or TNBC (aOR 0.58; 95% CI, 0.12–2.79; P for trend = 0.287) was observed. Among postmenopausal women, the influence of height was stronger for HR-negative tumors (aOR 2.81 per 10-cm increase; 95% CI, 1.41–6.03). The risk of breast cancer increased with higher weight and BMI among postmenopausal women in all subtypes. Their protective effect was apparent in premenopausal women in all subtypes, with a significant association being observed for HR-positive tumors and the HR-positive/HER2-negative subtype.
. | HR status . | Classic tumor subtypes . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|
. | HR-negative (46 cases) . | . | HR-positive (72 cases) . | . | HR+/HER2− (55 cases) . | . | HER2+ (32 cases) . | . | TNBC (31 cases) . | . |
. | Adjusted OR (95% CI) . | Pa . | Adjusted OR (95% CI) . | Pa . | Adjusted OR (95% CI) . | Pa . | Adjusted OR (95% CI) . | Pa . | Adjusted OR (95% CI) . | Pa . |
Age (years)b | ||||||||||
21–39 | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA |
40–49 | 3.14 (1.36–7.26) | 3.80 (2.03–7.08) | 4.09 (2.02–8.28) | 3.39 (1.27–9.08) | 3.05 (1.18–7.88) | |||||
50–64 | 3.72 (1.78–7.77) | 2.58 (1.37–4.84) | 2.59 (1.26–5.30) | 4.10 (1.69–9.96) | 2.78 (1.16–6.70) | |||||
Education (complete school years)c | ||||||||||
0 | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA |
1–7 | 1.83 (0.66–5.15) | 0.81 (0.38–1.71) | 0.75 (0.32–1.72) | 0.93 (0.31–2.83) | 2.96 (0.74–11.85) | |||||
≥8 | 4.16 (1.34–12.91) | 1.82 (0.79–4.18) | 1.48 (0.58–3.78) | 3.19 (0.98–10.39) | 5.16 (1.17–22.70) | |||||
P for trend | 0.007 | 0.032 | 0.154 | 0.009 | 0.037 | |||||
Parityd | ||||||||||
Nulliparous (0 live births) | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA |
Parous (≥1 live births) | 1.03 (0.22–4.87) | 0.65 (0.23–1.88) | 0.49 (0.16–1.47) | 1.33 (0.16–11.18) | 1.70 (0.21–13.93) | |||||
Number of live birthsd | ||||||||||
0–1 | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.] | NA |
2–3 | 1.44 (0.47–4.37) | 0.75 (0.35–1.63) | 0.67 (0.28–1.63) | 1.38 (0.40–0.74) | 1.26 (0.36–4.42) | |||||
4–5 | 0.76 (0.21–2.80) | 0.46 (0.19–1.13) | 0.46 (0.17–1.27) | 0.69 (0.16–2.97) | 0.64 (0.14–2.84) | |||||
≥6 | 0.91 (0.24–3.51) | 0.22 (0.08–0.64) | 0.20 (0.06–0.68) | 0.81 (0.17–3.79) | 0.58 (0.12–2.79) | |||||
P for trend | 0.543 | 0.003 | 0.007 | 0.441 | 0.287 | |||||
Height (per 10-cm increase)—overalle | 1.43 (0.90–2.26) | NA | 1.15 (0.79–1.68) | NA | 1.03 (0.67–1.58) | NA | 1.86 (1.07–3.23) | NA | 1.30 (0.74–2.28) | NA |
Among premenopausal women | 0.88 (0.48–1.63) | 0.021 | 1.00 (0.62–1.60) | 0.287 | 0.90 (0.53–1.52) | 0.328 | 1.35 (0.67–2.72) | 0.158 | 0.83 (0.41–1.70) | 0.047 |
Among postmenopausal women | 2.81 (1.41–6.03) | 1.54 (0.80–2.95) | 1.39 (0.68–2.83) | 3.18 (1.22–8.25) | 2.83 (1.09–7.36) | |||||
Weight (per 1-kg increase)—overallf | 1.01 (0.99–1.03) | NA | 1.00 (0.98–1.02) | NA | 0.99 (0.97–1.01) | NA | 1.02 (0.99–1.04) | NA | 1.01 (0.98–1.03) | NA |
Among premenopausal women | 0.99 (0.96–1.01) | 0.004 | 0.97 (0.95–0.99) | <0.001 | 0.96 (0.94–0.99) | <0.001 | 1.00 (0.97–1.03) | 0.013 | 0.98 (0.95–1.02) | 0.003 |
Among postmenopausal women | 1.06 (1.02–1.10) | 1.05 (1.02–1.09) | 1.05 (1.01–1.08) | 1.06 (1.02–1.11) | 1.07 (1.02–1.13) | |||||
BMI (per 1-kg/m2 increase)—overalle | 1.02 (0.96–1.07) | NA | 1.00 (0.95–1.04) | NA | 0.98 (0.93–1.02) | NA | 1.03 (0.97–1.09) | NA | 1.02 (0.96–1.08) | NA |
Among premenopausal women | 0.98 (0.91–1.04) | 0.045 | 0.93 (0.88–0.99) | <0.001 | 0.90 (0.85–0.97) | <0.001 | 0.99 (0.92–1.07) | 0.061 | 0.97 (0.90–1.05) | 0.019 |
Among postmenopausal women | 1.10 (0.99–1.21) | 1.14 (1.05–1.23) | 1.12 (1.03–1.22) | 1.12 (1.01–1.25) | 1.14 (1.02–1.29) | |||||
Obesity (BMI ≥ 30 kg/m2)—overalle | 1.10 (0.54–2.24) | NA | 0.86 (0.47–1.59) | NA | 0.82 (0.41–1.61) | NA | 0.69 (0.28–1.68) | NA | 1.57 (0.67–3.65) | NA |
Among premenopausal women | 0.78 (0.28–2.11) | 0.315 | 0.44 (0.19–1.03) | 0.012 | 0.34 (0.12–0.90) | 0.005 | 0.67 (0.20–2.21) | 0.914 | 0.90 (0.28–2.83) | 0.142 |
Among postmenopausal women | 1.61 (0.57–4.57) | 2.27 (0.87–5.97) | 2.74 (0.92–8.18) | 0.74 (0.20–2.74) | 3.36 (0.87–13.01) |
. | HR status . | Classic tumor subtypes . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|
. | HR-negative (46 cases) . | . | HR-positive (72 cases) . | . | HR+/HER2− (55 cases) . | . | HER2+ (32 cases) . | . | TNBC (31 cases) . | . |
. | Adjusted OR (95% CI) . | Pa . | Adjusted OR (95% CI) . | Pa . | Adjusted OR (95% CI) . | Pa . | Adjusted OR (95% CI) . | Pa . | Adjusted OR (95% CI) . | Pa . |
Age (years)b | ||||||||||
21–39 | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA |
40–49 | 3.14 (1.36–7.26) | 3.80 (2.03–7.08) | 4.09 (2.02–8.28) | 3.39 (1.27–9.08) | 3.05 (1.18–7.88) | |||||
50–64 | 3.72 (1.78–7.77) | 2.58 (1.37–4.84) | 2.59 (1.26–5.30) | 4.10 (1.69–9.96) | 2.78 (1.16–6.70) | |||||
Education (complete school years)c | ||||||||||
0 | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA |
1–7 | 1.83 (0.66–5.15) | 0.81 (0.38–1.71) | 0.75 (0.32–1.72) | 0.93 (0.31–2.83) | 2.96 (0.74–11.85) | |||||
≥8 | 4.16 (1.34–12.91) | 1.82 (0.79–4.18) | 1.48 (0.58–3.78) | 3.19 (0.98–10.39) | 5.16 (1.17–22.70) | |||||
P for trend | 0.007 | 0.032 | 0.154 | 0.009 | 0.037 | |||||
Parityd | ||||||||||
Nulliparous (0 live births) | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA |
Parous (≥1 live births) | 1.03 (0.22–4.87) | 0.65 (0.23–1.88) | 0.49 (0.16–1.47) | 1.33 (0.16–11.18) | 1.70 (0.21–13.93) | |||||
Number of live birthsd | ||||||||||
0–1 | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.) | NA | 1 (ref.] | NA |
2–3 | 1.44 (0.47–4.37) | 0.75 (0.35–1.63) | 0.67 (0.28–1.63) | 1.38 (0.40–0.74) | 1.26 (0.36–4.42) | |||||
4–5 | 0.76 (0.21–2.80) | 0.46 (0.19–1.13) | 0.46 (0.17–1.27) | 0.69 (0.16–2.97) | 0.64 (0.14–2.84) | |||||
≥6 | 0.91 (0.24–3.51) | 0.22 (0.08–0.64) | 0.20 (0.06–0.68) | 0.81 (0.17–3.79) | 0.58 (0.12–2.79) | |||||
P for trend | 0.543 | 0.003 | 0.007 | 0.441 | 0.287 | |||||
Height (per 10-cm increase)—overalle | 1.43 (0.90–2.26) | NA | 1.15 (0.79–1.68) | NA | 1.03 (0.67–1.58) | NA | 1.86 (1.07–3.23) | NA | 1.30 (0.74–2.28) | NA |
Among premenopausal women | 0.88 (0.48–1.63) | 0.021 | 1.00 (0.62–1.60) | 0.287 | 0.90 (0.53–1.52) | 0.328 | 1.35 (0.67–2.72) | 0.158 | 0.83 (0.41–1.70) | 0.047 |
Among postmenopausal women | 2.81 (1.41–6.03) | 1.54 (0.80–2.95) | 1.39 (0.68–2.83) | 3.18 (1.22–8.25) | 2.83 (1.09–7.36) | |||||
Weight (per 1-kg increase)—overallf | 1.01 (0.99–1.03) | NA | 1.00 (0.98–1.02) | NA | 0.99 (0.97–1.01) | NA | 1.02 (0.99–1.04) | NA | 1.01 (0.98–1.03) | NA |
Among premenopausal women | 0.99 (0.96–1.01) | 0.004 | 0.97 (0.95–0.99) | <0.001 | 0.96 (0.94–0.99) | <0.001 | 1.00 (0.97–1.03) | 0.013 | 0.98 (0.95–1.02) | 0.003 |
Among postmenopausal women | 1.06 (1.02–1.10) | 1.05 (1.02–1.09) | 1.05 (1.01–1.08) | 1.06 (1.02–1.11) | 1.07 (1.02–1.13) | |||||
BMI (per 1-kg/m2 increase)—overalle | 1.02 (0.96–1.07) | NA | 1.00 (0.95–1.04) | NA | 0.98 (0.93–1.02) | NA | 1.03 (0.97–1.09) | NA | 1.02 (0.96–1.08) | NA |
Among premenopausal women | 0.98 (0.91–1.04) | 0.045 | 0.93 (0.88–0.99) | <0.001 | 0.90 (0.85–0.97) | <0.001 | 0.99 (0.92–1.07) | 0.061 | 0.97 (0.90–1.05) | 0.019 |
Among postmenopausal women | 1.10 (0.99–1.21) | 1.14 (1.05–1.23) | 1.12 (1.03–1.22) | 1.12 (1.01–1.25) | 1.14 (1.02–1.29) | |||||
Obesity (BMI ≥ 30 kg/m2)—overalle | 1.10 (0.54–2.24) | NA | 0.86 (0.47–1.59) | NA | 0.82 (0.41–1.61) | NA | 0.69 (0.28–1.68) | NA | 1.57 (0.67–3.65) | NA |
Among premenopausal women | 0.78 (0.28–2.11) | 0.315 | 0.44 (0.19–1.03) | 0.012 | 0.34 (0.12–0.90) | 0.005 | 0.67 (0.20–2.21) | 0.914 | 0.90 (0.28–2.83) | 0.142 |
Among postmenopausal women | 1.61 (0.57–4.57) | 2.27 (0.87–5.97) | 2.74 (0.92–8.18) | 0.74 (0.20–2.74) | 3.36 (0.87–13.01) |
Abbreviations: BMI, body mass index; CI, confidence interval; HR, hormone receptor; NA, not applicable; OR, odds ratio; TNBC, triple-negative breast cancer.
aP for interaction.
bOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala).
cOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala), age (continuous), and number of live births (0–1, 2–3, 4–5, and ≥6).
dOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala), age (continuous), education (0, 1–5, 6–7, 8–10, 11–12, and ≥13 years), BMI (continuous), menopausal status (no/yes), and interaction term BMI*menopausal status.
eOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala), age (continuous), number of live births (0–1, 2–3, 4–5, and ≥6), and education (0, 1–5, 6–7, 8–10, 11–12, and ≥13 years).
fOR adjusted for province (Maputo City/Province, Gaza, Inhambane, Sofala), age (continuous), number of live births (0–1, 2–3, 4–5, and ≥6), education (0, 1–5, 6–7, 8–10, 11–12, and ≥13 years), and height (continuous).
Discussion
With this case–control study, we showed for the first time that multiparity is a protective factor for the development of HR-positive and HR-positive/HER2-negative tumors among native Sub-Saharan African women. A higher educational level was associated with an increased risk of breast cancer across all tumor subtypes, whereas being taller was associated with HR-negative tumors among postmenopausal women. In addition, higher weight and BMI were risk factors among postmenopausal women regardless of the subtype, whereas they were protective for premenopausal women.
Education is a proxy for socioeconomic status and is usually related to reproductive behaviors (e.g., delayed age at first childbirth, reduced parity; ref. 32) and access to healthcare (e.g., cancer screening; ref. 33). Multiple studies, including from Africa, have shown a positive association between a higher educational level and breast cancer risk (11, 13, 32, 34), even after adjustment for parity or age at first birth. However, none of them evaluated this association by subtype. In Western countries, there are some case–case studies assessing the risk of each tumor subtype according to region/neighborhood socioeconomic status. One large study observed higher incidence rates of HR-positive breast cancer among white and black women residing in areas with higher versus lower socioeconomic status, but not of HR-negative tumors (35). Yet, as individual-participant data were not available, these estimates were not adjusted for parity or breastfeeding. Our case–control study, with individual-subject data of native African women from all socioeconomic strata, shows that a higher educational level was a risk factor for all breast cancer subtypes, even following adjustment, and especially for HR-negative tumors and TNBC, which was intriguing. We would expect this effect to be more evident for HR-positive tumors, as they are more biologically indolent and frequently diagnosed by mammographic opportunistic screening, which is more accessible to highly educated women in this low-resource setting. However, it can also be the case that there is an underdiagnosis of breast cancer among women with a lower educational level, who may be treated by traditional healers and/or cannot reach the health system and thus were not included in this study. This may be especially true for women with HR-negative tumors and particularly for those with TNBC, who frequently have faster and more aggressive clinical courses.
Previous African studies have found that multiparity protects from breast cancer (10–13). Nonetheless, we have observed, for the first time in African women, that the protection given by multiparity is restricted to HR-positive tumors. A large meta-analysis also demonstrated that the protective effect of parity (parous vs. nulliparous) was limited to HR-positive tumors (3) and had no effect on the development of HER2-positive/HR-negative or TNBC. Yet, some studies have suggested that parous women who have never breastfed have a higher risk of developing HR-negative or TNBC compared with nulliparous women (36–38). The mammary gland reaches its full development during pregnancy and lactation, and after its cessation, the breast undergoes involution (39). A lower degree of terminal duct lobular units involution has been associated with higher breast cancer risk (40), especially of basal-like breast cancers (41). Thus, it has been proposed that long-term breastfeeding could reduce the risk of TNBC/basal-like breast cancers by promoting involution. However, this has yet to be proven. Although we did not collect data on breastfeeding, this is a common practice in Mozambique: most children are still breastfed at 1 year of age (94% in 2003, 86% in 2013; ref. 42), and the mean duration of breastfeeding ranges between 11 and 17 months (43). Moreover, given the high fertility rate, the lifetime duration of lactation among our participants should be high. This generalized breastfeeding adherence may explain why, in our study, there was no association between multiparity and HR-negative tumors nor with TNBC.
The prevalence of obesity is increasing in most Sub-Saharan African countries (44), currently being around 10% among adult Mozambican women (24). However, most studies assessing the impact of high BMI on breast cancer risk have been performed among white women in the United States or Europe, and show that obesity is a risk factor among postmenopausal women, whereas it is protective among premenopausal women (5, 45, 46). This applies mostly to HR-positive tumors, as its effect in HR-negative or TNBC is less well-defined (47). However, data in Africa are scarce and without information on tumor subtypes. One large study from Ghana indicated that a “heavier” body size (classified by pictograms) conferred a higher risk of breast cancer (aOR 1.50), but without stratification for menopausal status (11). In a Nigerian study, higher BMI had a protective effect among premenopausal women (per 5-kg/m2 increase: aOR 0.89), whereas it appeared to have no effect in postmenopausal women (16). However, they showed that a higher waist–hip ratio was a risk factor both in pre- and postmenopausal women (highest vs. lowest quartile category: aOR 2.15; ref. 48). Interestingly, this effect persisted in normal-weight women (BMI < 25 mg/kg2), indicating that increased central adiposity was also a risk factor per se (48). These findings somehow differ from the large AMBER Consortium report, which included over 3,000 Afro-American women (49). Among postmenopausal women, obesity was a risk factor for HR-positive tumors, but it was protective for TNBC (aOR 0.60). However, women with central obesity have been found to have a higher risk of TNBC, which may be related to components of metabolic syndrome (e.g., insulin resistance) that are related with carcinogenesis (50). Curiously, BMI appeared to have no effect on the risk of breast cancer among premenopausal women, regardless of the subtype.
Nevertheless, these somehow contradictory findings are not so surprising. Although BMI is a standardized body measurement, it does not capture other aspects of body composition and metabolism, such as the proportion of central/subcutaneous adiposity, muscle mass, and insulin sensitivity (51, 52). Afro-American women tend to have more lean mass and lower fat mass than non-Hispanic white women with a similar BMI do (53), but tend to be more insulin-resistant (54). In addition, Africa is a vast region, with large genetic, cultural, and socioeconomic diversity (55); therefore, we should be cautious when generalizing findings from Afro-American women to the African context or even across African regions. Similarly, the African women lifestyle may be very different from Afro-American women, namely in terms of diet, physical activity, and reproductive behaviors. Our finding that higher weight and BMI were risk factors for both HR-positive and HR-negative tumors (including TNBC) among postmenopausal women, whereas these were protective in premenopausal women, should thus be further analyzed. It would be interesting to integrate data on waist–hip ratio, diet, alcohol consumption (56), and physical activity (57) to explore this relationship in future studies.
Most studies assessing the association between height and breast cancer risk were conducted in Western countries, where attained height is not conditioned by nutritional status during childhood. Currently, the prevalence of stunting in Mozambique is still around 43% among children under 5 years of age (58). Therefore, it is likely that many of our participants had an insufficient energy intake during childhood/adolescence, which may explain their shorter stature compared with those in developed countries (59, 60). Still, we have seen that higher height is a risk factor for HR-negative tumors in postmenopausal women, even after adjusting for education (a proxy for socioeconomic status), age, and number of live births. Data from Nigeria mirrored this finding, but without information by subtype (16). However, in the US-based Black Women's Health Study, being tall was a risk factor for breast cancer, but for HR-positive tumors among Afro-American women, most of whom were premenopausal (59). The hypothesis that a higher attained height is associated with increased sex hormones and insulin-like growth factors during the pubertal growth spurt, a period in which the breast is also developing (60, 61), would be more in line with a higher risk for HR-positive tumors. Yet our findings go in the opposite direction, deserving further clarification in future studies.
This study has several strengths. Both the WHO-STEPS survey and the Moza-BC cohort collected data on weight/height that was measured by the investigators, instead of self-reported (11). The classification of tumor subtypes was done prospectively and according to the ASCO/CAP guidelines (28, 29), allowing a reliable assessment of the HR and HER2 status, which was not the case in some previous reports (e.g., that did not use in situ hybridization to determine HER2 status; refs. 62, 63). Moreover, the controls were selected from a population-based study, unlike other African studies that used hospital-based controls (14). In addition, we analyzed an ethnically homogeneous population, as almost all participants were black (98%, among both cases and controls), African women born in Mozambique.
Nevertheless, besides the limitations already discussed, others should be mentioned. Although we analyzed the effect of parity, we did not have data on age at first child, age at menarche, or age at menopause among controls, which impaired its analysis. Likewise, we have no information on breast cancer family history among the controls, but as the proportion of cases with a positive family history was very low (4%), this may be of little relevance within this population. Around 25% of the cases were HIV-positive (27), but this information was not available for controls. Yet, although some U.S. studies have reported a lower incidence of breast cancer among women with HIV infection compared with the general population (64), African studies show no effect of HIV on breast cancer risk (65, 66).
Three quarters of the cases had advanced breast cancer (stage III/IV) at the time of diagnosis, which might raise concerns regarding the reliability of their weight at diagnosis as a proxy to their usual weight. Nevertheless, in the Moza-BC cohort, the majority of women were overweight or obese at enrolment (23% and 36%, respectively), and only 4% of cases were underweight. Furthermore, unlike what is seen in head and neck cancer, where patients frequently present with critical weight loss at the time of cancer diagnosis, this is rarely seen among women with breast cancer, even among those with advanced disease (67). Hence, we expect our weight and BMI measurements to be a reliable proxy to the patient's usual body parameters.
Our sample size is much smaller compared with larger studies from the United States and other developed countries. However, we could detect significant effects for many of the tested factors and also when stratifying the analysis by menopausal status or subtype. To do so, we combined data from two studies, conducted during different time periods, but as these were close (2014–2015 vs. 2015–2017), we do not expect differences in the distribution of risk factors between cases and controls to be due to the timing of data collection. We did not have information regarding previous cancer history among controls, thus we could not exclude women with a previous diagnosis of breast cancer; yet, the likelihood of having controls with a previous breast cancer is very low, thus it should not influence the validity of our estimates. As the cases were followed at the Maputo Central Hospital, most of them lived in the South of the country, which partially impairs the generalization of these findings to the entire country.
As this is the first study exploring the risk effects of different factors according to breast cancer subtypes among native African women, it has implications for the management of breast cancer in Sub-Saharan Africa. We confirm that multiparity protects from HR-positive breast cancer, but unlike what was suggested by other studies, for HR-negative tumors, the point estimates suggest a much weaker association, and there was no evidence of a linear trend. Still, an absence of effect in HR-negative tumors could be explained by the high adherence to breastfeeding among Mozambican women, which reinforces the need to promote this behavior—especially since the proportion of women with HR-negative and TNBC is already high in Mozambique (33% and 25%, respectively; ref. 27).
We have also confirmed that higher weight and BMI are risk factors for breast cancer among postmenopausal women, regardless of the subtype. As the prevalences of overweight/obesity are rapidly rising throughout Africa (44), this may contribute to an increasing breast cancer incidence. Besides, obesity seems to be associated with worse prognosis after the diagnosis of breast cancer regardless of menopausal status (68–70), making its control crucial. Therefore, it is essential that awareness campaigns be conducted in this setting to promote healthier diets and lifestyles.
In conclusion, these results highlight the etiological heterogeneity of breast cancer among native African women, namely in terms of the differential effect of multiparity, education, and body parameters in breast cancer risk. As the prevalence of obesity is increasing in Africa, these findings are important to inform public health policies regarding cancer prevention, by highlighting that obesity is a modifiable risk factor for breast cancer among African women.
Authors' Disclosures
M. Brandão reports grants from FCT – Fundação para a Ciência e a Tecnologia, I.P. during the conduct of the study and grants and personal fees from Roche/GNE outside the submitted work. A. Guisseve, G. Bata, M. Gudo-Morais, J. Ferro, S. Tulsidás, and C. Carrilho report grants from U.S. NCI during the conduct of the study. C. Silva-Matos reports grants from Mozambique Ministry of Health and WHO during the conduct of the study. No disclosures were reported by the other authors.
Authors' Contributions
M. Brandão: Conceptualization, data curation, formal analysis, funding acquisition, methodology, writing–original draft, project administration, writing–review and editing. A. Guisseve: Data curation, investigation, project administration, writing–review and editing. A. Damasceno: Conceptualization, resources, supervision, funding acquisition, methodology, project administration, writing–review and editing. G. Bata: Resources, data curation, writing–review and editing. C. Silva-Matos: Conceptualization, resources, supervision, funding acquisition, methodology, project administration, writing–review and editing. M. Alberto: Investigation, writing–review and editing. J. Ferro: Resources, investigation, writing–review and editing. C. Garcia: Resources, investigation, writing–review and editing. C. Zaqueu: Resources, investigation, writing–review and editing. C. Lorenzoni: Resources, investigation, writing–review and editing. D. Leitão: Investigation, writing–review and editing. O. Soares: Resources, data curation, writing–review and editing. A. Gudo-Morais: Resources, data curation, writing–review and editing. F. Schmitt: Supervision, investigation, methodology, writing–review and editing. S. Morais: Investigation, methodology, writing–review and editing. S. Tulsidás: Resources, data curation, writing–review and editing. C. Carrilho: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing. N. Lunet: Conceptualization, formal analysis, supervision, funding acquisition, methodology, project administration, writing–review and editing.
Acknowledgments
The Moza-BC cohort study was supported by the Beginning Investigator Grant for Catalytic Research (BIG Cat) program, an AORTIC program with support from the U.S. NCI (grant number 59-210-6-004). The WHO-STEPS Survey was funded by the Mozambican Ministry of Health and by the WHO. In addition, this study was supported by national funding from FCT – Fundação para a Ciência e a Tecnologia, I.P., under the Unidade de Investigação em Epidemiologia – Instituto de Saúde Pública da Universidade do Porto (EPIUnit; UIDB/04750/2020). Samantha Morais was cofunded by FEDER through the Operational Programme Competitiveness and Internationalization and FCT (POCI01-0145-FEDER-032358; PTDC/SAU-EPI/32358/2017). The funders had no involvement in the analysis and interpretation of data, writing of the report, or decision to submit the article for publication. The authors thank the Calouste Gulbenkian Foundation (Portugal) and partners Camões, Institute of Cooperation and Language, Portugal; Millennium BCP Foundation, Portugal; and Millennium BIM; Mozambique, for funding the short-term training program of Assucena Guisseve at the Centro Hospitalar Universitário de São João, under the Project “Improving the diagnosis and treatment of oncological diseases in Mozambique.” The authors also thank Dr. Maria Alice Franzoi, from the Institut Jules Bordet, for her comments on the article.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.