Risk and Protective Factors for Breast Cancer in Midwest of Brazil

Patterns of physical activity, body composition, and breastfeeding are closely related to health and are influenced by environmental, economic, and social factors. With the increase of sedentary lifestyle and overweight, many chronic diseases have also increased, including cancer. Breast cancer is the most common cancer in women worldwide, and the knowledge of its risk and protective factors is important to the adoption of primary prevention strategies. We aimed to investigate some risk and protective factors for breast cancer among women from Midwest Brazil. It is a case-control study of outpatient basis, carried out with 93 breast cancer cases and 186 controls. Socioeconomic, gynecological, anthropometric, and lifestyle variables were collected, and odds ratios (ORs) values were estimated (significance level, 5%; confidence interval (CI), 95%). Per capita income equal to or lower than 1/2 Brazilian minimum wage (OR = 1.88; CI = 1.06–3.29), residence in rural area (OR = 4.93; CI = 1.65–14.73), and presence of family history of breast cancer (OR = 5.38; CI = 1.46–19.93) are risk factors for breast cancer. In turn, physical activity (past 6 months) (OR = 0.23; CI = 0.10–0.55) and leisure physical activity at 20 years old (OR = 0.13; CI = 0.03–0.54) are protective factors for the disease in women who live in Midwest of Brazil.


Introduction
Around the world, populations have changed rural areas for cities, becoming more sedentary and consuming great amounts of industrialized foods and beverages, which have contributed for the increase of obesity and chronic diseases, such as cancer. These changes began in Europe, North America, and other countries that became industrialized in XIX century. In recent decades, there was acceleration in urbanization process around the world, especially in lowmiddle-income countries, like Brazil. Patterns of physical activity, body composition, and breastfeeding are determined by environmental (urbanization, urban planning, which includes transport systems), economic (social status), and social (education level) factors, influencing, thus, the occurrence of several types of cancer [1,2].
Breast cancer is the most common cancer in women worldwide and has been associated with higher socioeconomic status and urbanized populations [2,3]. Family history of breast cancer and gynecological variables, such as early menarche, late menopause, not bearing children, and late first pregnancy are well established risk factors for breast cancer [2,4]. Besides these variables, according to the World Cancer Research Fund (WCRF) and the American Institute for Cancer Research (AICR) [2], drinking alcoholic beverages is a risk factor, whereas breastfeeding is a protective factor for breast cancer. However, regarding the body fatness, adult attained height, abdominal fatness, adult weight gain, and physical activity, the evidences are not sufficiently clear [2]. Body fatness is a convincing risk factor only for postmenopausal breast cancer, and the protective influence of physical activity is not convincing yet [2]. Most studies on the determinant factors for breast cancer, especially related to physical activity, have been conducted in high-income countries [5][6][7][8]. Thus, some risk factors pointed by the WCRF and the AICR [2] may be different from that found in low-middle-income countries, such as Brazil, which has a different epidemiological profile. In most of Latin American countries, the transition experience is unlike that of the developed countries and is distinguished by the coexistence of infectious diseases and chronic diseases as important causes of death. This epidemiological profile is called "transition polarized model" [9].
Furthermore, in Brazil and other Latin American countries, there are few recent analytical studies on the risk and protective factors for breast cancer. According to a 2009 report by Brazilian National Cancer Institute [4], in Goiânia, capital of Goiás state (Midwest of Brazil), the estimated risk for breast cancer was 51.87 new cases per 100 000 women, which was close to the national mean estimate (49.27 new cases per 100 000 women). Knowledge of risk and protective factors among populations is important to identify groups that are more susceptible and to develop strategies for primary prevention [10]. Thus, the objective of this study was to investigate some risk and protective factors for breast cancer among women from Midwest of Brazil and with recent diagnosis of breast cancer. To calculate the sample size, the following criteria were considered: significance level of 5%, study power (1-β) of 80%, 2 controls for each case, an odds ratio (OR) of 2.5 [12], and a prevalence of risk for consumption of alcohol among controls of 14.4% [13]. According to these parameters, the total sample must include 186 controls and 93 cases of women with breast cancer. The inclusion and exclusion criteria (cases and controls) are shown in Table 1.

Material and Methods
This study was approved by the ethics committees of UFG (protocol no. 026/08) and Araújo Jorge Hospital (protocol no. 019/08). Written informed consent was obtained from all participants.

Data Collection.
Data were collected between May 2008 and June 2010, through a questionnaire that had been previously tested in a pilot study. This questionnaire included socioeconomic variables (per capita income-in American dollar-education level, and area of residence), family history of cancer and breast cancer, gynecological variables (number of children, age at first pregnancy, age of menarche, and age of menopause), breastfeeding information, anthropometric measures, alcoholic drink consumption, physical activity, and smoking habit. Questions about alcoholic drink consumption, physical activity, and smoking habit were asked regarding the 6 months prior to the interview date. Data on body weight and physical activity were also investigated considering the young age (20 years old) [14].
Interviewees provided data on occurrence and duration of breastfeeding (in months) for each of their children, which were added up to provide lifetime duration of breastfeeding. Anthropometric variables included height, current and previous (at 20 years-old) body weight, waist circumference, current and previous body mass index (BMI), and adult weight gain. The first 3 variables were obtained according to the protocol reported by Lohman et al. [15]. Before data collection, these measures were standardized among interviewers, according to Habicht [16]. The cut-off points used to classify nutritional status and waist circumference were established based on the World Health Organization [11]. Weight gain in adulthood was estimated by the difference between weight at the time of research and weight at 18-20 years old [14].
The daily intake of ethanol was calculated as follows: the amount of each alcoholic beverage consumed per week was multiplied by its alcohol content and divided by 7 [17]. We considered only beer, because it was the only alcoholic beverage reported by consumers. The following parameters were adopted: a can of beer equivalent to 350 mL; a bottle, 600 mL; a glass, 250 mL. We considered the alcohol content of beer to be 5 per cent (5 g/100 mL) [18] and consumption of more than 15 g of ethanol/day as excessive consumption [2].
The subjects were questioned about physical activity (past 6 months) and were classified into 4 categories: active (moderated activities ≥30 minutes/session and ≥5 days/ week, or vigorous activities ≥20 minutes/session and ≥3 days/week, or any activity adding up ≥150 minutes/week and ≥5 days/week); irregularly active A (any physical activity ≥150 minutes/week or ≥5 days/week); irregularly active B (any physical activity, but did not reach any criteria recommendation related to frequency and duration); sedentary (did not practice any physical activity at least 10 continuous minutes during the week). This classification was defined according to the eighth version of the International Physical Activity Questionnaire (IPAQ), validated to Brazilian adult women [19]. We also evaluated previous physical activity (at 20 years old), considering the practice or not of occupational, housework, and leisure physical activities.
We considered smokers as those who smoked regularly for at least 1 year [20] and ex-smokers as those who had stopped smoking at least 2 years before the interview date [17].

Statistical Analysis.
Data were analyzed using Statistical Software for Professional 8.0 (Stata Corp., College Station, TX, USA). The normality test for quantitative variables was performed using normal curves in the Stata software. Variables were analyzed in the whole sample, without considering the menopausal status, like found in some studies [21][22][23], and because some controls did not know about their menopausal status, due to the surgical extraction of their womb.
In bivariate analyses, to compare the 2 groups (case and control) in relation to quantitative variables, we applied Student's t-test (parametric distribution) and the Mann Whitney test (nonparametric distribution). Regarding the qualitative variables, we used the chi-square test or Fisher's exact test to compare case and control groups, according to frequency distribution. For all variables related to outcome, we calculated the Wald statistic and the odds ratio (OR), with a significance level of 5% and a confidence interval (CI) of 95%. All variables with a P value lower than 0.20 for bivariate analyses were included in multivariate analyses. Logistic regression was performed using a hierarchical model structured in 2 levels. Socioeconomic variables, such as education level, per capita income, and area of residence were included in the first level (distal). Height, physical activity, family history of cancer and of breast cancer, bearing children, and number of children were grouped in the second level (proximal). Bearing children and number of children, although considered as demographic variables, were included in the second level because they are influenced by socioeconomic status. Variables whose P value was lower than 0.05 remained in the logistic regression model [24].

Results
We invited 98 cases and 186 controls to participate in research, but 5 cases (5.1%) refused to be interviewed, adding up 279 women (93 cases and 186 controls). The average age among cases was 51.93 years (range 30-83 years) and that among controls was 51.73 years (range 28-81 years). There were no significant differences between the case and control groups in relation to socioeconomic, gynecological, and anthropometric variables. The median family income per capita was lower than the Brazilian minimum wage (R$ 510 per month, approximately US$ 300), being US$ 147 for cases (range US$ 0-1,470) and US$ 200 for controls (range US$ 0-1,029) ( Table 2).
Education level, per capita income, area of residence, family history of any kind of cancer, and family history of breast cancer in the first degree were significantly associated with breast cancer in the bivariate analysis (Table 3).
Anthropometric and lifestyle variables, such as alcohol consumption, smoking habit, and physical activity, are shown in Table 3. Physical activity (past 6 months) and previous leisure physical activity (at 20 years old) were associated significantly with breast cancer in the bivariate analysis (Table 4).
We considered the following variables in the multivariate analysis: education level, per capita income, area of residence, bearing children, number of children, family history of cancer (any type), history of breast cancer, height, physical activity (past 6 months), and previous occupational and leisure physical activities. After adjusting the OR values, per capita income, area of residence, family history of breast cancer, physical activity (past 6 months), and previous leisure physical activity remained associated with breast cancer (Table 5).
Per capita income equal to or lower than 1/2 Brazilian minimum wage (OR = 1.88; 95% CI = 1.06-3.29), residence in rural area (OR = 4.93; 95% CI = 1.65-14.73), and presence of family history of breast cancer (OR = 5.38; 95% CI = 1.46-19.93) were considered risk factors for breast cancer. Active women were at lower risk of developing breast cancer, compared to sedentary women (OR = 0.23; CI = 0.10-0.55), and that ones who had practiced leisure physical activity at young age (20 years-old) were less likely to develop the disease, compared to who had ever practiced (OR = 0.13; CI = 0.03-0.54) ( Table 5).

Discussion
The sample was very homogeneous concerning socioeconomic, anthropometric, and gynecological variables ( Table  2). This can be explained by the fact that all interviews were conducted with users of public hospitals, which serve a quite similar population. Women in this sample are of low socioeconomic status, which makes up 29.3% of Brazilian Midwest families with a per capita income of 0.5-1.0 minimum Brazilian wage [25]. In Table 6, we have summarized the findings of the present study, comparing to WCRF and the AICR evidences [2]. Although nutritional status was not significantly associated with breast cancer (Tables 3 and 5), it is important to emphasize that the prevalence of current overweight (BMI ≥ 25 kg/m 2 ) in both groups (cases and controls) was greater than 60%, and the prevalence of obesity (BMI ≥30 kg/m 2 ) was approximately 28% among cases and 24% among controls. These values are higher than those found in Brazil, since the proportion of obesity in adult Brazilian women, between 2008 and 2009, was 16.9% [26]. Obesity statistic found in this sample is troubling, because excessive body fat can promote chronic diseases such as type 2 diabetes and increase the level of proinflammatory substances that may enhance carcinogenesis [2]. The prevalence of excessive alcoholic drinking intake was low, under 3%, for both cases and controls (Table 4), and may be related with the low socioeconomic status of the sample.
In the present study, we found that having a low per capita income (under 1/2 Brazilian minimum wage) and living in a rural area were risk factors for breast cancer (Table 5). One possible explanation for these findings may be the lesser access to health care and health information of poor women who live in rural areas. Nevertheless, previous studies report that breast cancer is positively associated with urbanization and high socioeconomic status [2][3][4]. These associations are mainly due to variation in known risk factors, such as late pregnancy and not bearing children, found in women from urban and high-income areas [3].
It should be added that approximately 2/3 of Brazilian population is exposed to pesticides, and the most vulnerable group to this exposure is rural workers [27]. According to Peres, Peres et al. [27], rural workers have low health risk perception related to the use of pesticides. The indiscriminate use of these substances by the rural population could explain the occurrence of some cancers. Moreover, in most Brazilian inner cities, including from Goiás State, a large proportion of women is from rural area, where there are no reference breast services. So that, suspected breast cancer cases move up to capital cities to receive health attendance. Furthermore, it should be emphasized that Goiânia's urbanization process was initiated in 1950 and occurred quickly, which was called "demographic explosion." However, this urbanization was not followed by industrialization [28]. Thus, this population does not have the same characteristics as urban areas in developed countries. Besides, the Brazilian epidemiological transition model, called "transition polarized model," is different from the high-income countries models, named "classic or occidental transition model." In the "transition polarized model," both chronic and infectious diseases coexist as important cause of death, and the decrease in mortality rate is slower. In the "classic or occidental transition model," there is the prevalence of chronic diseases and the mortality rate decreases progressively. Therefore, risk factors may differ according to the epidemiological profile of each population [9,29].
We also found that family history of breast cancer in first degree is a risk factor for breast cancer (Table 5), and this result is reinforced by other ones [12,30]. According to  Journal of Environmental and Public Health the "theory of two stages" in the genetic determinism of carcinogenesis, it is necessary that mutations occur at the same locus of homologous chromosomes for the development of a malignant tumor. Thus, in nonhereditary cases (also called sporadic cases), 2 mutagenic processes are required, whereas, in hereditary cases, 1 genetic mutation already exists and only 1 more mutation in the other chromosome is necessary [31]. According to Brandt et al. [12], 15-20% of breast cancer cases are associated with family history of breast cancer, but only 15% can be explained by mutations in the BRCA1 and BRCA2 genes, known tumor suppressors. A woman who carries a mutation in BRCA-1 has a risk of 60-80% for developing breast cancer [12,31].
Regarding physical activity, we found that women who practiced at least 150 minutes/week and/or ≥5 days/week of physical activity were at lower risk of developing breast cancer, compared to sedentary women. Moreover, that ones who practiced leisure physical activity at young age (20 years old) were also at lower risk, compared to that women who did not practice leisure activity. These results show that physical activity, in general, may protect against breast cancer. This protective effect is supported by WCRF and AICR's report and by other case-control studies [2,[5][6][7][8]. This result is important because WCRF and AICR [2] concluded that the protective effect of physical activity is not convincing yet (probable in postmenopausal women, but limited in premenopausal women), and our study supports this evidence, which may increase the consistency of this association (Table 6).
Regular physical activity probably has a protective effect on breast cancer by delaying menarche, promoting irregular and anovulatory cycles, reducing serum estrogen, increasing globulins that bind to sex hormones, reducing inflammation, improving immune function, and helping to control weight and to improve sensitivity to insulin [2,6]. Considering the importance of physical activity practice in prevention of breast cancer and other chronic diseases, and that health is a determinant key of development and a precursor of economic growth, it is necessary to perform programs aimed at promoting health lifestyle. To implement these policies, it is necessary the collaboration between the health sector and other key sectors, such as education, urban planning, transportation, and communication [32].
The lack of an association of some variables with breast cancer can be explained partially by the recall bias, which is very common in case-control studies. Another difficulty of case-control studies is the time required to consider exposure factors. Some studies [22,33] considered exposure factors over 1 year. In the present study, we adopted 6 months preceding the interview to minimize memory bias. Although the exposure time considered in this study was shorter than most previous case-control studies, we interviewed cases immediately after the positive diagnosis of breast cancer to avoid misconceptions about the exposure time. In fact, it represents a great strength of the present study, since in the other studies previously cited [22,33], cases were interviewed 3-4 months after diagnosis. We recommend that other breast cancer case-control studies are conducted on populations of higher socioeconomic status from the same region of Brazil. The risk and protective factors published by WCRF and AICR [2] may apply better to such segment of the population, because their lifestyles are more similar to those from developed countries, where most case-control and cohort studies are performed. Furthermore, it is necessary to better explore the question of area of residence to clarify the association of this variable with breast cancer in the present population.
This investigation is the first case-control study on the risk and protective factors for breast cancer in a population from Midwest of Brazil. Findings from our research reinforce the associations between physical activity and family history of breast cancer, in first degree, and breast cancer. Moreover, our results reveal new data that can contribute to elucidate determinant factors for breast cancer, and to prevent this disease in epidemiological transition countries, such as Brazil (Table 6).

Conclusions
Low per capita income, living in a rural area, and family history of breast cancer are risk factors for breast cancer, whereas practicing physical activity, including at 20 years old, is a protective factor for the disease in women from Midwest of Brazil.