Role of Obesity in the Risk of Breast Cancer: Lessons from Anthropometry

An estimated 1.38 million new cases of breast cancer (BC) are diagnosed each year in women worldwide. Of these, the majority are categorized as invasive ductal cell carcinoma. Subgroups of BC are frequently distinguished into five “intrinsic” subtypes, namely, luminal A, luminal B, normal-like, HER2-positive, and basal-like subtypes. Epidemiological evidence has shown that anthropometric factors are implicated in BC development. Overall consistent positive associations have been observed between high body mass index (BMI) and waist-to-hip ratio (WHR) and the risk of BC among postmenopausal women, while conflicting results persist for premenopausal BC, both for BMI and for other anthropometric parameters as well as across ethnic groups. Furthermore, some evidence suggests that body size, body shape, and weight gain during childhood or adolescence may play a role in the risk of BC. In this paper, we describe the evidence linking anthropometric indices at different ages and BC risk, in order to improve our understanding of the role of body fat distribution in the risk of BC, investigate differences in these associations according to menopausal status and ethnic groups, and discuss the potential biological mechanisms linking body size and BC risk.


Introduction
Breast cancer (BC) is the most frequently diagnosed cancer in women worldwide, accounting for 23% (1.38 million) of the total new cancer cases in 2008 [1,2]. e majority of invasive breast neoplasms are categorized as "invasive ductal cell carcinoma, not otherwise speci�ed" (ICD-O 8500/3) [3]. is entity has long been recognized to include tumors with heterogeneous molecular characteristics, characterized by distinct patterns of gene expression [4] and of genomic/genetic alterations [5]. Subgroups of BC are frequently distinguished into luminal A (estrogen/progesterone-positive), luminal B, HER2+, and so-called "triple negative" subtypes [6]. Among these subtypes, HER2+ and basal-like subtypes tend to be more common among premenopausal women, as well as in women of African ancestry. Luminal subtypes are more common in postmenopausal women and among Caucasians [7]. e incidence rates of BC show a heterogeneous distribution, while Western countries present the highest incidence rates, the lowest incidences are observed in low resource countries. BC ranks as the ��h cause of death from cancer overall (458,000 deaths), but is still the most frequent cause of cancer death in women in both developing (269,000 deaths) and developed regions (189,500 deaths) [1]. Incidence and mortality rates have increased in low-incidence countries, particularly in Latin America (LA) and Africa and especially among younger women. Data from LA show that the highest BC incident rates are among women aged 30-49. In this area, BC is the most common cancer among women, and the leading cause of cancer mortality compounded by problems of access to screening, diagnosis, and treatment [8,9]. In Mexico for example, BC death has increased 3.6 fold between 1995 and 2005. In Africa, a recent study among Gambian women showed an increase of 6.5% in the incidence rate of BC from 1990 to 2006 [10] compared to 1.5% annually increase worldwide [11]. Asian countries are also recording signi�cant annual increases, for example 2% in Japan and 3-5% in China [11].
e risk factors related to reproductive factors such as delayed childbearing, lower parity, and reduced breastfeeding are becoming more prevalent in countries in economic transition but do not fully explain the increase of BC incidence rates. Other risk factors such as changes in lifestyle, physical activity, and anthropometry as well as ethnicity and genetic susceptibility play a role in the development of BC [12][13][14][15][16]. In particular, overweight and obesity have been clearly associated with an increased overall risk of BC.
Overweight and obesity have become major public health challenges throughout the world in both high and low income countries, with over 1 billion overweight and 315 million obese adults currently estimated worldwide [17,18]. In the United States (US) for example, obesity is now considered as the second leading cause of preventable death aer tobacco consumption. Over two-thirds of the adult population is overweight, with approximately one-third of adults and almost 17% of children and adolescents obese [17,18]. Most studies on the association between obesity and BC risk have been conducted on Caucasian populations from Europe, Canada, and the US. However, a limited number of studies in other ethnic groups suggest some differences in the prevalence of obesity and fat distribution among women. For example studies in the US show that American women of African or Hispanic origin are more likely to be obese than Caucasian women [19]. Given the disparities in prevalence rates of obesity, it is expected that among women, 69% of Caucasians, 87% of African Americans, and 80% of Hispanics will be overweight in 2015 [19].
Anthropometric indices, height, weight, body mass index (BMI), waist circumference (WC), hip circumference (HC), or waist-to-hip ratio (WHR), are commonly used as tools for assessing overweight/obesity and recently methods for evaluating body shape and body size at different ages have also been used. Several studies and meta-analyses have examined the associations between anthropometric indices and BC among both pre-and postmenopausal women. Overall consistent positive associations have been observed between BMI and WHR and the risk of BC among postmenopausal women [20][21][22][23][24][25][26][27][28][29], while con�icting results persist for premenopausal BC, both for BMI and for other anthropometric parameters [29][30][31][32][33][34][35][36]. Possible differences in these associations could be related to differential roles of fat and fat distribution on metabolism and to their contribution to BC development among pre-and postmenopausal women [31,32]. Waist circumference (WC) and WHR, which correlate with central (abdominal) obesity have been associated with increased BC risk in postmenopausal women [32,[37][38][39], but not in premenopausal women for whom both null [23,38,39] and increased risk have been reported [31,32,37,40]. Furthermore, the relation between premenopausal BC and anthropometry has shown inconsistencies across ethnic groups [41][42][43]. ese inconsistencies may be due to important ethnic variations in body fat distribution. We conducted a systematic review of published studies to evaluate the strength of associations between anthropometric indices at different ages and BC risk in order to improve our understanding of the role of body fat distribution in the risk of BC, to investigate differences in these associations according to menopausal status and ethnic groups, and to highlight the potential biological mechanisms linking body size and BC risk.

Methods
To review the epidemiologic literature on the association of overweight, obesity, fat distribution, and BC risk, we conducted a MEDLINE and PUBMED search including all publications using height, weight, BMI, WC, HC, WHR, anthropometric factors, body shape, early life body size, BC, premenopausal, premenopausal BC, case-control, and cohort studies as key words. Our paper follows the preferred reporting items for systematic reviews and meta-analysis (PRISMA) statement guidelines described by Liberati et al. [52]. We then examined the references from the identi�ed articles, previous review and meta-analysis between anthropometric factors and premenopausal and postmenopausal BC. e literature search included all publications from 1997 to 2011 and recent publications up to February 2012.

�. �e���t�o�� Me�s��e�e�t� ��d ���e o� ��es�t�
Overweight and obesity are de�ned as abnormal or excessive fat accumulation that may impair health [61]. ere are different ways to measure obesity; BMI is the most commonly used and is calculated as weight (kg) divided by height (m 2 ). e World Health Organization (WHO) classi�es the degree of adiposity in terms of the BMI as follows: underweight, less than 18.5; normal, 18.5-24.9; overweight, 25.0-29.9; obese, more than 30.0 kg/m 2 (Table 1(a)). While BMI is usually associated with general obesity, WC and WHR have been used as measures of central or intra-abdominal obesity, de�ned as waist-hip ratio above 0.90 for males and above 0.85 for females (Table 1(b)). However, there is evidence of important ethnic variations in body fat distribution, especially in Asians as compared to Caucasians. In Asian adults, differences in body build and body composition result in a different relationship between BMI and body fat distribution relative to Caucasians [62][63][64]. In a study of young Japanese and Australian men, Kagawa and colleagues reported that Japanese men were estimated to have an equivalent amount of body fat as the Australian men at BMI values about 1.5 units lower than those of the Australians (23.5 kg/m 2 and 28.2 kg/m 2 , resp.) [65]. Similarly, studies among Chinese reported similar relationship between BMI and body fat distribution [43]. ese observations suggest that Asian people are more likely to have higher levels of body fat overall and more abdominal fat and lower lean mass than the other ethnic groups for a given BMI [41][42][43]. Consequently, the level of BMI that the WHO recommends for overweight and obesity in Caucasians may not be applicable for Asian adults. e WHO Western Paci�c Regional O�ce (WPRO) has proposed a de�nition of obesity in Asian populations with overweight de�ned as BMI 23.0-24.9 kg/m 2 and obesity Journal of Oncology 3 T 1: (a) Combined recommendations of body mass index and waist circumference cut-off points made for overweight or obesity, and association with disease risk. (b) World Health Organisation cut-off points and risk of metabolic complications.  [29]. In most case-control and prospective studies, a positive and strong relationship has been observed between BMI and postmenopausal Asian women [20,22,67]. In a cohort study of 10,106 women, conducted in Japan, the RR for developing postmenopausal BC was 2.54; 94% CI (1. 16-5.55) in women with BMI of 25 kg/m 2 or above compared to those with less 20.5 kg/m 2 [22]. In contrast, there is a lack of data among women of African origin, but the notion that increase in BMI is associated with the risk of postmenopausal BC in this population is consistent with a recent case-control study reporting a positive association between high BMI and an increased risk of estrogen receptor and progesterone receptor positive BC (OR = 1.83; 95% CI: 1.08-3.09; P trend = 0.03) comparing the highest (>35 kg/m 2 ) to the lowest (<25 kg/m 2 ) quintile [30]. Other studies on women of African descent do not support a positive association between BMI and postmenopausal BC [34,35,40,49].  [51]. e heterogeneity among these �ndings may be due to the lack of power in African studies due to the small number of cases, in particular in postmenopausal women, re�ecting the demographic structure of African populations. Another possible explanation for these differences may be the selection bias mainly in casecontrol studies, where anthropometric measurements are oen obtained aer diagnosis and may be affected by the effects of cancer development on the general condition of the patient. e lack of association between BMI and BC risk observed in Hispanic women may re�ect different effects of fat accumulation and distribution in this Hispanic population compared to populations in high resource countries.

WHR and Breast Cancer
Risk. WHR is commonly used as a measure of central obesity [31,32,70], de�ned as waist-hip ratio above 0.90 for males and above 0.85 for females has oen been associated with the risk of developing postmenopausal BC (Table 2). However, while most studies have reported a signi�cant increased risk [33,[37][38][39]48], some studies are inconclusive [23,35,40] (Figure 2). A metaanalysis with six case-control and �ve cohort studies observed a summary risk estimate of 1.50 (95% CI: 1.10-2.04) for postmenopausal women [37]. e associations tend to be stronger in Asian women than other ethnic groups [33].  (Figure 2).  [25]. Several studies supported the hypothesis that higher level of BMI may be associated with a decrease in the risk of premenopausal BC. is hypothesis is supported by results from several case-control studies [34,39,40,60] and cohort studies [35,53]. However, others studies did not observe a statistically signi�cant association when comparing highest versus lowest levels of BMI [23,49,56]. Ethnicity appears to modify the association of overweight, obesity, and BC. While the inverse association between BMI and risk of premenopausal BC is well documented in Caucasians, the association among Asian women is inconsistent. Several studies among Asian women suggest that higher BMI may be associated with an increased risk for premenopausal BC [22,24,46,55]. A prospective study including 11,889 women from Taiwan reported that higher BMI was moderately associated with an increased risk of premenopausal BC [55], with an OR of 1  [36]. A more recent study has shown that high BMI was inversely associated with risk of BC in Hispanic population (P trend < 0.01) [58].

�. Si�ni�cance of �aist�to��ei�ht Ratio
ere is evidence suggesting that the waist-to-height ratio (WHtR) may be a more useful global clinical screening tool than WC, with a weighted mean boundary value of 0.5 [73]. e WHtR is de�ned as waist (cm) circumference divided by height (cm) and correlates well with abdominal obesity; higher values of WHtR indicate higher risk of obesity-related cardiovascular diseases [74]. However, to date the association between WHtR and BC risk has not been systematically studied. Nevertheless, a previous study of multiethnic groups reported an inverse association between WHtR and ER+PR+ BC in all premenopausal women [58].  [31]. ese �ndings suggest that body fat distribution may be associated with an increased risk for ER-negative BC among premenopausal women.

Variations with Molecular Subtypes of Breast Cancer
In a pooled analysis of tumor marker and epidemiological risk factor data from 35,568 invasive BC case patients from 34 studies participating in the BC Association Consortium, obesity (BMI ≥ 30 kg/m 2 ) in women ≤50 years was found to be more frequent in ER−/PR− than in ER+/PR+ tumors ( −7 ). In contrast, obesity in women >50 years was less frequent in PR− than in PR+ tumors ( −4 ) [75]. is study demonstrated that elevated BMI in younger women (≤50 years) was associated with the risk of ER+ or PR+ BC but not with ER−/PR− BC and suggested that triple-negative or core basal phenotype (CBP); de�ned by triple-negative and cytokeratins [CKs] 5/6+ (CBP) BC may have distinct etiology [75]. e differences between the conclusions of the two studies [31] may be linked to the use of different obesity/fat distribution measures. Further studies are needed to better understand the associations between overweight and obesity and speci�c subtypes of BC.

Effect of Early Life Anthropometry on Breast Cancer Risk
Most of the epidemiological studies have focused mainly on adult BMI and not on weight change or on the in�uence of early life body weight and body shape (silhouette). ese factors, such as birth size, body shape, and weight during childhood or adolescence may play a role in the risk of BC. Epidemiologic studies have demonstrated that independently of BMI, greater body fatness during childhood or adolescence are associated with lower BC risk in both premenopausal women [30,36,45,53,54,57] and postmenopausal women [30,35,44,45]. In a recent large prospective study conducted among 188,860 women, Baer and colleagues reported 9. Metabolic Pathways for Breast Cancer Development e complex associations between anthropometric measures of body fatness/obesity and the risk of BC suggest that metabolic conditions associated with high body fatness may in�uence this risk in several ways, with distinct effects on preand postmenopausal BC, as well as on different molecular subtypes of BC. It is now commonly accepted that the occurrence and development of BC is driven by the abnormal, clonal expansion of pools of initiated stem/progenitor cells [76,77]. e existence of different types of such progenitors, as well as the effect of molecular alterations allowing the progeny of these cells to switch from one differentiation phenotype to another, may account for a large part of the molecular and epidemiological heterogeneity of BC. According to this view, the risk of BC may depend upon whether speci�c pools of progenitor cells are either enhanced or suppressed at particular stages of breast maturation and development. Metabolic parameters associated with body fatness may affect these progenitor cells through �ve main biological mechanisms ( Figure 5). First, they may affect the cell's bioenergetic balance and favor the expansion of cells with high anaerobic glycolytic capacity, a bioenergetics adaptation which characterizes cancer cells. is effect, known as �Warburg effect, � is de�ned by intense lipogenesis and glycolysis and low mitochondrial oxidative phosphorylation capability even in the presence of sufficient oxygen [78,79]. Hyperinsulinemia and high blood glucose levels, which are frequent in obese subjects, are expected to provide a selective advantage for the growth of such cells. Second, increased adiposity may have an impact on sterol synthesis and on the metabolism of estrogens. Obesity (high BMI) has been associated with increasing sex hormone (estrogen) due to increased peripheral aromatization of adrenal androgens in adipose tissue among postmenopausal women, which can promote cell proliferation, anti-apoptotic and proangiogenic effects [80,81]. ird, high blood levels of insulin and insulinlike growth factor (IGF-I) have been found to stimulate the growth and survival of cancer cells in both pre-and postmenopausal women and their production can be increased by estrogen [82][83][84]. Fourth, obesity induces chronic low-grade in�ammation resulting in an increase of local and systemic levels of cytokines (such as TNF-, interleukin-6 (IL-6), C-reactive protein (CRP), and monocyte chemoattractant protein-1 (MCP-1)). Obesity can also increase adipokines (leptin and adiponectin). ese factors may, in turn, affect mitosis, apoptosis, angiogenesis, and cell migration and escape from immune recognition [85]. Fih, increase in free fatty acids, such as triglycerides, has been reported to increase the level of free estradiol by displacing estradiol from sex hormone binding globulin (SHBG) [86]. erefore, both decreases in SHBG and increases in triglycerides may result in increased free estradiol. e above mechanisms are not mutually exclusive. ey may operate in a complementary manner to promote speci�c forms of BC. One of the main paradox is the apparently opposite association of body fatness with pre-and postmenopausal BC. In Caucasians, most postmenopausal BC are ER+/PR+ (luminal A) subtypes and the effect of body fatness may involve increased hormone biosynthesis in adipose aer the menopause, leading to the long-term maintenance of breast progenitor cells aer the menopause. In contrast, in pre-menopausal women, the apparently protective effect of obesity may be due to hormone-independent forms of BC, which are more common among premenopausal women. Also, in premenopausal women estradiol levels are reduced in anovulatory cycles that are more frequent in obese than lean women. In addition, obese premenopausal women have been found to have reduced progesterone levels [87], accounting for the negative association between BMI and risk of premenopausal BC. e lack of the association between estradiol levels and risk of BC development in premenopausal women suggest that obesity may affect breast progenitor cells through mechanisms other than estradiol levels.

Conclusion
Obesity has become a crucial public health problem worldwide, especially for BC development and survival. Most studies have shown that BMI which re�ects general obesity is associated with a decrease of the risk of developing BC before menopause and increase aer menopause in most of the studies, while �HR which re�ects central obesity is associated with an increased risk of both pre-and postmenopausal BC. Results are consistent with differences in metabolic risk and de�nitions of obesity according to ethnicity. �ata regarding the relationship between obesity and young age and BC have demonstrated a strong inverse association between body fatness during childhood and adolescent and risk of BC throughout life in Caucasian population. e mechanisms for this inverse association are not fully understood and need further research. It will also be important to develop stringent recommendations and to maintain a healthy weight both at individual and community levels.