Abstract
Objective
To investigate breast cancer risk factors in Chinese women residing in urban and rural areas of eastern China using a large-scale cross-sectional survey.
Methods
In this multistage, stratified cluster sampling epidemiological survey, information on demographic characteristics, diet and lifestyle variables were gathered using a carefully designed questionnaire. Multivariate logistic regression analysis and subgroup analyses of the data were performed, including separate analyses of data from women residing in urban and rural areas.
Results
A total of 122 058 women were included in the survey. Age, body mass index, number of miscarriages, family history of breast cancer and menopausal status were found to be risk factors for breast cancer, while the consumption of soya bean products was a protective factor. Among women residing in urban areas, high or moderate intake of soya bean products and red meat were protective factors. Among women residing in rural areas, obesity and a high intake of milk were identified as risk factors for breast cancer, while a moderate intake of soya bean products was a protective factor.
Conclusions
This type of data is crucial for understanding the risk factors for breast cancer and could facilitate the development and targeting of effective intervention strategies, with the ultimate aim of breast cancer prevention.
Keywords
Introduction
It is estimated that the majority of malignant cancers, as many as 70%, are the result of harmful factors related to diet, lifestyle or the surrounding environment. 1 Breast cancer, which is the most prevalent malignancy in women worldwide, is of growing concern because of its rising incidence and ongoing regional disparities in incidence. 2 In eastern China, which accounts for 38% of the Chinese population, there has been a sharp increase in breast cancer incidence in recent decades. 3 The incidence of breast cancer reached 63.90 new cases per 100 000 women in 2008. 4 The incidence of breast cancer has been shown to be lower in rural areas compared with urban areas; however, in the 15 years between 1993 and 2008, the increase in breast cancer incidence in rural areas (8.9%) was much greater than in urban areas (3.7%). 5 As a result, the gap in breast cancer incidence between rural and urban areas is narrowing, especially in some developed areas. 6
In different geographical regions, there are differences in diet, lifestyle and environmental conditions. It is possible that these differences are partly responsible for the regional disparities observed in breast cancer incidence. In China, the ratio of women living in urban areas to rural areas was ∼0.36 : 0.64 in 2000, and the prevalence of female breast cancer differed between these two areas, being 249.3/100 000 in urban areas and 267.1/100 000 in rural areas. 4 Factors such as a long menstrual life, nulliparity, later age at first birth and limited breastfeeding are traditionally associated with an increased risk of breast cancer in the Chinese population,4,7,8 but are not associated with regional distribution. According to Lee et al, 9 women with breast cancer in urban areas were more likely to be nulliparous and to have had fewer live births than those in rural areas. However, age at menopause, history of breast feeding and family history of breast cancer did not vary significantly by region. 9 In addition, over the last 10 years Chinese people have undergone marked dietary changes, moving from a low-fat, vegetable-based diet to a high-fat and animal protein-based diet. 10 As a result, the incidence of obesity and diabetes mellitus have increased greatly.11,12 A low-fat diet has been shown to reduce breast cancer incidence. 13 In addition, soya bean products are thought to be protective for breast cancer. 14
A clear understanding of the risk factors involved in breast cancer is crucial for preventive actions to be implemented. Previous studies have been carried out in China,15,16 but, to the authors’ knowledge, the present study is the first large-scale cross-sectional survey of breast cancer risk factors in Chinese women. Investigating the relationships between different lifestyles and environments and breast cancer incidence, specifically in Chinese women from rural and urban backgrounds, would give an increased understanding of the risk factors involved. This increased knowledge could then be used to educate the population with regard to everyday diet and lifestyle changes that may help to reduce the incidence of breast cancer.
Patients and methods
Patients
Participants in the present study were obtained from a multistage, stratified cluster sampling survey carried out between 15 July and 15 September 2008.4,17 The survey was performed in the eastern provinces of Shandong, Jiangsu, Hebei and Tianjin. Target regions and communities were chosen from each province using multistage stratified cluster sampling, as previously described. 4
Participants were female, aged between 25 and 70 years and of Han ethnic origin (90% of the Chinese population), and had lived in the study area for ≥ 2 years. Written informed consent was obtained from all the survey participants. Participants were recorded as living in urban or rural areas according to their registration card for permanent residence issued by the Chinese government, which states whether the area they live in is classified as urban or rural.
The survey protocol was approved by the ethics committees of the Second Hospital of Shandong University, Jinan, Shandong, China, and the collaborating institutions in the selected regions.
Questionnaire
The design of the interview questionnaire was based on numerous published articles, the Gail, Claus and International Breast Cancer Intervention Studies (IBIS) breast cancer prevention models 18 and discussions with a variety of experts in breast surgery, epidemiology, statistics, nutrition and molecular biology. 4 In order to minimize recall bias, several similar questions were asked in different sections of the questionnaire. A preliminary investigation was performed to assess the practicality and effectiveness of the survey. 19 After repeated revisions, the final version covered the following aspects: (i) demographic information, including age, residence, weight, height, marital status, and a family history of breast cancer in first- and second-degree relatives; (ii) reproductive factors, including age at first pregnancy, parity, number of miscarriages, number of children, age at menarche and menopausal status; (iii) history of breast cancer; (iv) frequency of consumption of various foods, including soya bean products (bean curd, soya bean milk), fresh soya beans, red meat, milk, corn, carrots, fried food, vegetables or fruit other than carrots, garlic, ham sausage and pickled food; (v) cigarette smoking and alcohol consumption.
Questionnaires were completed by trained investigators during face-to-face interviews with each participant. Investigators for each region were recruited before implementation of the study, to ensure objectivity and authenticity. In order to minimize bias, all information in sections (iii), (iv) and (v) related to within 1 year of any breast cancer diagnosis.
Further investigations
After being interviewed, each participant underwent a physical examination performed by an independent breast surgeon (F.W. in Shandong, M-M.G. in Jiangsu, Y-Y Li in Hebei; L.L. in Tianjin) Subjects with any palpable or suspicious symptoms on examination underwent further ultrasonography scans and mammographic screening. All results, including the questionnaire responses, were evaluated by eight experts (two ultrasonography experts, two radiologists and four breast surgeons ([co-authors Z-G.Y; Q-Y. F; D-Z. G.; Q.Z.]) to determine whether biopsy should be performed. Pathological results, including oestrogen receptor status, lymph node status and tumour size, as well as the clinical stage of all patients (whether diagnosed before or during the survey) were documented.
Statistical analyses
The database was established using Epidata 3.1 software (Epidata Association, Odense, Denmark). Descriptive analysis was used to describe the general characteristics of participants. Data concerning demographic characteristics, diet and lifestyle were expressed as n and percentage of participants and tested for significance using the χ2-test. Univariate logistic regression analysis was performed to identify significant variables, which were then used to produce odds ratios by multivariate unconditional logistic regression, with subgroup analyses for rural and urban areas. A two-sided P-value < 0.05 was considered to be of statistical significance. All statistical analyses were performed using SPSS® 16.0 software (SPSS Inc., Chicago, IL, USA).
Results
A total of 147 538 women were initially selected from the target regions; of these, 124 758 women completed the survey, giving a 15.44% rate of loss. Of the 124 758 women, 1037 were excluded for being <25 years of age, 1629 were excluded for being > 70 years of age and 34 were excluded because of missing data, leaving a survey population of 122 058 women, with a sampling concordance rate of 97.84%. Among the 122 058 women surveyed, 320 (0.262%) had breast cancer and 121 738 were healthy controls.
Demographic characteristics
Demographic characteristics showing statistically significant differences in women with breast cancer and controls living in urban and rural areas of Eastern China.
Data presented as n (%).
Statistical significance calculated using χ2-test.
Demographic characteristics were also analysed separately in women residing in urban and rural areas. The same significant differences seen in the overall study population (age, BMI, number of miscarriages, postmenopausal status and history of breast cancer) were also seen between those with breast cancer and controls in both the urban and the rural groups (Table 1). Likewise, no significant differences were found with regard to marital status, age at first birth, breastfeeding or age at menarche in either group (data not shown).
Diet and lifestyle characteristics
Relative risk estimates for the association between breast cancer and the consumption of various dietary elements and lifestyle characteristics in women with breast cancer and controls living in urban and rural areas of Eastern China.
For dietary elements: 4, eats nearly every day (5–7 days per week); 3, eats 3–4 days per week; 2, eats 1–2 days per week; 1, almost never eats.
Data presented as n (%).
NS, no statistically significant between-group differences (P ≥ 0.05) using the χ2-test.
Diet and lifestyle characteristics were also analysed separately in women residing in urban and rural areas. Among the women living in urban areas, significant differences were identified between those with breast cancer and controls, with regard to the consumption of soya bean products (P = 0.003), red meat (P = 0.001) and fried food (P = 0.003), but there were no significant differences in other variables (Table 2). Among the women living in rural areas, a significant difference was identified between those with breast cancer and controls with regard to the consumption of milk (P < 0.001), but there were no significant differences in the other variables (Table 2).
Multivariate logistic regression analysis
Multivariate logistic regression analysis of the associations between various demographic characteristics, dietary elements and lifestyle characteristics and breast cancer incidence in women living in urban and rural areas of Eastern China.
Subgroup analyses
Subgroup analyses of relative risk estimates for the associations between body mass index and the consumption of soya bean products, milk and red meat and breast cancer incidence in women living in urban and rural areas of Eastern China.
For dietary elements: 4, eats nearly every day (5–7 days per week); 3, eats 3–4 days per week; 2, eats 1–2 days per week; 1, almost never eats.
Discussion
There has been an assumption that differences in breast cancer incidence between Asian and Western populations are uniform and reflect the benefits of the Asian lifestyle. Asian populations have generally been treated as homogeneous in this respect, and variations in breast cancer incidence among the countries that make up this large and diverse geographical area have so far been overlooked. The present study was the first large-scale cross-sectional survey of breast cancer risk factors in Chinese women. A wide range of demographic characteristics and diet and lifestyle variables were analysed in women residing in urban and rural areas of eastern China. The main findings were (i) that in all of the women surveyed, age, BMI, number of miscarriages, family history of breast cancer and menopausal status were significantly associated with breast cancer, and (ii) among women residing in urban areas, moderate or high intake of soya bean products or red meat were found to be protective factors, while among women residing in rural areas, obesity and a high intake of milk were risk factors for breast cancer, whereas a moderate intake of soya bean products was a protective factor.
The mean age at diagnosis of breast cancer is 45–55 years in China, but is 65 years in the USA and other Western countries. 20 There is an interesting double-peak pattern in the age-specific incidence curve of breast cancer in China, with one peak at about 45–55 years and another at 70–74 years. 21 However, a study by Yang et al 6 showed that in the period between 2004 and 2008 in Beijing, in urban areas the peak age group for female breast cancer incidence was 60–64 years, whereas in rural areas there were two peaks at 50–54 years and 80–84 years. In the present study, in urban areas there were two peaks in the incidence of breast cancer at 35–44 years and 55–64 years, whereas in rural areas the peak was at 45–64 years. Because the present study did not include those aged >70 years, a further peak in older women in rural areas cannot be confirmed. These results indicate a shift towards Western patterns, with an older median age at diagnosis, especially in rural areas.
Body mass index is a measure of relative weight based on an individual’s weight and height. Many studies have shown that a high BMI is an important risk factor for breast cancer, 22 especially in postmenopausal women. 23 A BMI categorized as overweight or obese has been reported to be associated with increased all-cause mortality. 24 In the present study a BMI ≥ 28 kg/m2 in rural areas was associated with a significantly higher risk of breast cancer. Huang and colleagues 25 stated that avoiding adult weight gain may play an important role in preventing breast cancer after menopause, particularly among women who are not treated with postmenopausal hormones. The association between the number of miscarriages and breast cancer is well established. 26 A family history of breast cancer is also a well-established risk factor for breast cancer, 27 with women with a family history of breast cancer having a 1.2-times higher risk than women without. 28 In the present study, the number of miscarriages and a family history of breast cancer were confirmed as risk factors on multivariate logistic regression analysis.
A number of studies have indicated a relationship between dietary patterns or specific nutrients and breast cancer risk. Differences in breast cancer incidence in different ethnic and social groups could be explained by different dietary patterns. Furthermore, certain nutrients could reasonably increase the risk of cancer because they contain carcinogens (or their precursors), or cancer-promoting substances. The association between soya bean products and breast cancer is controversial. Soya bean products were thought to be protective for breast cancer, and a high soya intake was reported to reduce the risk of breast cancer in later life. 29 However, a study by Hirose and colleagues 30 reported a statistically significant association between tofu or isoflavone intake and breast cancer in premenopausal women in Japan, while no statistically significant association was seen in postmenopausal women. The chemical structure of isoflavones, the main ingredient in soya bean products, is similar to that of oestrogen. Isoflavones bind to and activate the oestrogen receptor, competing with oestrogen. Either an increase or a decrease in the risk of breast cancer associated with isoflavone intake is therefore plausible. A Japanese public health centre-based prospective study found isoflavones to be a protective factor against breast cancer, 31 while the California Teachers Study in the USA found isoflavones to be harmful. 32 Overall, the evidence is inconclusive. In the present study, a moderate or high intake of soya bean products was found to be beneficial in an analysis of all the study participants. When divided into urban and rural populations, the beneficial effects were only observed in those from urban areas with a moderate or high intake and in those from rural areas with a moderate intake; a high intake of soya bean products showed no association with breast cancer in the rural population. These results suggest that the consumption of soya bean products is not a risk factor for breast cancer in eastern China and that moderate intake has beneficial effects.
Some dairy products, such as whole milk and cheese, have a relatively high saturated fat content that may increase the risk of breast cancer. 33 Moreover, milk products may contain contaminants such as pesticides, insulin-like growth factor I and oestrogens, which have been shown to have carcinogenic potential or to promote breast cancer cell growth.34,35 In contrast, the calcium and vitamin D content of dairy products have been hypothesized to reduce breast cancer risk. 36 In a prospective cohort study of 48 844 premenopausal Norwegian women, analyses according to the type of milk consumed and milk fat consumption did not reveal any clear associations with breast cancer. 37 Milk is consumed in China, but unlike in Western countries, other dairy products such as cheese or butter are rarely consumed, especially in rural areas. Therefore consumption of milk specifically, rather than dairy products, was investigated in the present study. The results demonstrated that consumption of milk was significantly associated with breast cancer incidence in women residing in rural areas. This finding was in accordance with some previous studies.38,39 However, in urban areas, milk consumption was not associated with breast cancer risk, which is in accordance with other findings.40,41 This difference may possibly be explained by the subtype of milk consumed. In rural areas, people consume higher quantities of whole milk, whereas in urban areas people prefer processed milk, which contains higher levels of calcium and vitamin D. 42
In 2007 the UK Women’s Cohort Study 43 reported a significant association between red meat consumption and breast cancer risk in postmenopausal women, with a stronger effect observed for processed meat. This result was confirmed in a case–control study in Polish women. 44 Exogenous hormones administered to cattle to increase muscle mass could be passed on to consumers of the meat and stimulate the growth of hormone-sensitive breast tumours. Heterocyclic amines and polyaromatic hydrocarbons created during the high-temperature cooking of red meat are established mammary carcinogens. 45 However, a pooled analysis of several cohort studies measuring meat intake in relation to breast cancer risk found no significant association and revealed considerable differences between studies in the methods for assessing meat intake. 46 Taking into account cooking methods, Kabat and colleagues 47 found no significant associations between red meat consumption and breast cancer risk. Meat contains potential anticarcinogenic factors, including omega-3 polyunsaturated fatty acids and conjugated linoleic acid. Red meat is an important source of micronutrients with anticancer properties, including selenium and vitamins B6, B12 and D. 48 Each of these factors may play a role in cancer protection in different situations. 49 The present study found no significant association between red meat consumption and breast cancer risk in the rural population, while moderate or high intake of red meat were protective factors in the urban population. This may be explained by the fact that those residing in urban areas can afford to be more discerning about the type of meat they consume (and the cooking method). 50 The meat chosen will affect its fat content; urban residents select more meat from wild animals, 51 which is lower in fat overall, with a lower proportion of saturated fatty acids and a higher proportion of polyunsaturated fatty acids compared with meat from farmed animals. Furthermore, roasting meat instead of frying or barbecuing it leads to lower cooking temperatures, which in turn leads to avoidance of the formation of heterocyclic amines and polyaromatic hydrocarbons. 52
Determining the mechanisms by which lifestyle patterns influence cancer risk will not only illustrate biological plausibility for the observed associations but will also provide evidence of causality, which can then be used to inform the population of the steps necessary to reduce the risk of cancer.
Differences in cancer incidence have been observed between urban and rural communities for many decades. 53 In India, breast cancer incidence is approximately twice as high in urban women than in rural women. 54 A study of colorectal cancer in the USA observed that the risk factors for cancer differed between urban and rural areas. 55 In the present study some differences were identified between the urban and rural populations. Future studies need to focus on examining the specific characteristics of both the individuals and the environment that may affect access to and use of care in rural areas.
The present study had several limitations. First, this was a cross-sectional survey rather than a standard case–control study. There are also some uncontrolled biases, although use was made of several similar questions to minimize recall bias and to keep the data accurate and authentic. Secondly, the only lifestyle characteristics studied were cigarette smoking and alcohol consumption; many more variables should be included in future studies. Third, the dietary characteristics studied were also limited and could in future include the amount consumed, the cooking method and the regularity of meals. Lastly, the cross-sectional design of the study did not allow for changes in behaviour over time to be incorporated. A longitudinal study is required to confirm these findings and a study including intervention and follow-up would provide even stronger evidence.
In conclusion, the present study investigated the demographic characteristics and diet and lifestyle variables that might be associated with breast cancer incidence in women in eastern China, and analysed these data according to geographical location (urban or rural). Age, BMI, number of miscarriages, family history of breast cancer and menopausal status were found to be risk factors for breast cancer. Multivariate logistic regression analyses showed some differences in the risk factors and protective factors between urban and rural populations. A high or moderate intake of soya bean products and red meat in urban women, and a moderate intake of soya bean products in rural women, were associated with a decreased risk of breast cancer.
Regardless of predisposing factors, diet and lifestyle are known to influence morbidity and mortality during the course of life. 56 The ultimate goal of research such as that undertaken for the present study is to set up a model suitable for Chinese women to predict the risk of developing breast cancer similar to the Gail model 18 for Western populations. At present, the identification of factors in diet and lifestyle that are associated with breast cancer can help guide women to avoid risk factors as far as possible and lead a healthy life.
Footnotes
Acknowledgements
The authors would like to thank all the survey investigators for their involvement, as well as members of the Centers of Disease Control and Prevention and the Family Planning Departments of Shandong, Hebei, Jiangsu and Tianjin Hospitals for their collaboration and cooperation. We also thank two ultrasonography experts FeiXue Zhang and Mei Wu at the B-Ultrasound Department, The Second Hospital of Shandong University, Shandong, two radiologists Wei Hua and Hai Zhang, The Second Hospital of Shandong University, Shandong, for their involvement.
Declaration of conflicting interest
The authors declare that there are no conflicts of interest.
Funding
This study was funded through the Ministry of Health, China (Project No. 07090122). The initial stages of the study were also sponsored by Novartis China.