In this study, we evaluated the associations between current BMI, BMI at age 20 years, BMI change and breast cancer risk. We showed the possibility that heterogeneity may be present in the association between adult weight gain and breast cancer risk by tumor subtype. Postmenopausal women with luminal and triple-negative breast cancers showed the same impact of BMI on breast cancer risk, despite the molecular and clinical differences between these two subtypes. These findings are consistent with those of several (Phipps et al. 2011; Phipps et al. 2008) but not all previous studies (Yang et al. 2007; Yang et al. 2011; Tamimi et al. 2011).
The biological mechanisms underlying these associations are unclear, but it is assumed that obesity influences breast cancer risk through multiple mechanisms. One is a classic estrogen-dependent mechanism, in which obesity contributes to lower serum levels of sex hormone-binding globulin and higher circulating levels of endogenous estrogen (Potischman et al. 1996). Almost all reproductive factors that influence estrogen levels are primarily associated with luminal-type tumors, but there is also some evidence related to risk for non-luminal-type tumors (Yang et al. 2007; Yang et al. 2011; Tamimi et al. 2011; Islam et al. 2012). For example, our recent study in Japanese suggested that age at menarche was inversely associated with the risk of both luminal and triple-negative breast cancer (Islam et al. 2012). Similarly, the Polish Breast Cancer Study showed an association of earlier age at menarche with basal-like tumors only (Yang et al. 2007). Other mechanisms might relate to non-hormonal factors, such as an insulin-related mechanism involving insulin, insulin resistance and insulin-like growth factors (IGFs), or an inflammation-related mechanism (Stoll 2002; Kern et al. 2001). IGFs, which stimulate mitogen-activated protein kinase (MAPK) or Akt and increase cell survival, proliferation, and migration through signaling cascades, are also considered to be risk factors for breast cancer (Hankinson et al. 1998; Schernhammer et al. 2005). Davison et al. (Davison et al. 2011) reported that IGFs promote cell survival in triple-negative breast cancer cells. Further, Maiti et al. (Maiti et al. 2010) suggested that metabolic syndrome, characterized by high BMI and insulin resistance, is significantly more prevalent in triple-negative than non-triple-negative patients. On the basis of these findings, we hypothesize that the combination of these complex mechanisms contributes to an increased risk of breast cancer, particularly of triple-negative subtypes. An understanding of the etiology of less common subtypes such as triple-negative tumors, which are the most clinically aggressive, may be useful in the development of prevention strategies.
Our observation of an inverse association between BMI gain and the risk of premenopausal luminal breast cancer is to our knowledge the first description in an Asian population. A number of studies have evaluate the effect of weight change on the risk of premenopausal breast cancer, but most evidence to date is from studies of ER and PR rather than HER2 disease, and the results are inconsistent (Huang et al. 1997; Michels et al. 2012; John et al. 2010; Weiderpass et al. 2004; Lahmann et al. 2005). In the Nurses’ Health Study, weight loss since age 18 years was inversely related to both ER+/PR+ and ER-/PR- tumors (Michels et al. 2012). In contrast, a case–control study in the San Francisco Bay Area reported that inverse associations between obesity and premenopausal breast cancer risk were limited to ER+/PR+ tumors (John et al. 2010), which is consistent with our present findings. The mechanisms underlying this inverse association remain to be elucidated. Several studies have suggested that women who are obese prior to menopause are likely to have ovulatory insufficiency or anovulatory cycles, resulting in decreased estrogen and progesterone levels (Suzuki et al. 2011; Baer et al. 2010; Kawai et al. 2010; Key & Pike 1988), and we accept this theory in interpreting our present result. However, given our limited number of subgroup subjects and the inconsistent results across studies, additional studies with sufficient numbers of subjects of various ethnicities are needed.
With regard to BMI at age 20 years, we found no statistically significant association, but rather a slightly inverse association when underweight women (BMI < 18.5) were excluded (Table ). Several studies have suggested an inverse association between body weight in early adulthood and the risk of breast cancer (Suzuki et al. 2011; Baer et al. 2010; Kawai et al. 2010), but others have found no significant association (Sanderson et al. 2002; Okasha et al. 2003). A conclusive answer to this issue awaits further study.
Our study had several methodological strengths. First, it was conducted within the framework of the HERPACC study, which has a substantial number of subjects and a high response rate to completion of the questionnaire (95%). Second, the potential confounding factors of age and menopausal status were adjusted for by matching. Finally, the single-institution design of the study obviated the possibility of hormonal receptor or HER2 status misclassification resulting from different assay methods.
Since the present study was based on a hospital-based case–control study, several methodological limitations exist. First, the values for self-reported life-style factors considered as potential confounding factors might have been inaccurate. In particular, information about body weight at age 20 years was collected retrospectively. Second, the classification of tumor subtypes as defined in our study is not identical to other published data due to the lack of information on other tumor markers, such as cytokeratin (CK) 5/6 expression, epidermal growth factor receptor (EGFR), and Ki67. We were therefore unable to precisely distinguish between basal-like and triple-negative tumors, or between luminal A and luminal B tumors. Finally, subject numbers in our subgroup analyses were limited.
In conclusion, we found that current BMI and adult weight gain are associated with the risk of luminal and triple-negative breast cancer among postmenopausal women only. Our results provide additional evidence for these associations and suggest further etiological heterogeneity among tumor subtypes.