Introduction

MetS is a constellation of metabolic dysfunctions that include central obesity, dyslipidemia, hypertension, and insulin resistance, and significantly increases the risk of chronic diseases such as coronary heart disease, stroke, type-2 diabetes, and cancer [1,2,3,4]. Individual components of MetS are associated with increased risk of cancer [5,6,7,8,9,10]; however, studies on the association of MetS with breast cancer mortality are limited. Of the eight published studies evaluating breast cancer outcomes till date [11,12,13,14,15,16,17,18], three observed increased risk of breast cancer mortality among women with MetS [11, 15, 18], and four studies reported significant associations between MetS and other breast cancer outcomes—including distant metastasis [12] and aggressive phenotypes [11, 13, 14]. Of the three studies that directly examined the association between MetS and breast cancer mortality, two were limited by small sample sizes [15, 18], and one was conducted among European populations, limiting generalizability [15]. Moreover, while breast cancer hormonal subtypes are established prognostic indicators for breast cancer [19], results from the only two studies that evaluated the association between MetS and breast cancer subtypes were inconclusive [17, 20], and none of the previous studies simultaneously investigated whether the associations between MetS and breast cancer mortality varied by menopausal status and BMI.

It remains unclear whether MetS is a risk factor for breast cancer mortality overall and by subtype, and whether these associations vary by BMI, especially given the limited number of cohort studies and strong potential for residual confounding by BMI in cross-sectional studies. While obesity, a major component of MetS, is recognized as an established risk factor for breast cancer among post-menopausal women [21], obesity is also a risk factor for breast cancer mortality among women of all ages [22, 23]. Additionally, most of the other risk factors for breast cancer outcomes also vary by menopausal status, including the distribution of hormone-receptor subtype [19], suggesting that the association between MetS and breast cancer mortality may vary by menopausal status and subtype [20]. It is also unclear whether the individual MetS components—including hypertension, high cholesterol, and high fasting glucose—show independent or synergistic associations with breast cancer mortality across these categories.

The aim of this study was to examine the association between MetS, individual components, and breast cancer mortality in a large prospective cohort, and to evaluate these associations across menopausal status, BMI, and subtype. We also examined the association between different combinations of MetS components and risk of breast cancer mortality by menopausal status to identify which combination was most strongly associated with mortality risk.

Materials and methods

Data from 94,555 women in the National Institute of Health-American Association of Retired Persons (NIH-AARP) Diet and Health Study were analyzed to investigate the prospective association between MetS and risk of breast cancer mortality. Over half a million NIH-AARP members were recruited to the NIH-AARP Diet and Health Study in 1995-1996, and baseline data on demographics, behavioral, and other participant characteristics were collected. The NIH-AARP Diet and Health Study has been described in detail elsewhere [24]. Participants were followed up prospectively to ascertain cancer incidence and mortality outcomes [25]. After excluding male participants (n = 325,171), those with cancer diagnosis (n = 23,971) or mortality (n = 1375) at baseline, poor general health (n = 3407), proxy respondents (n = 15,760), those who did not return Risk Factor Questionnaire (n = 58,374), and those with missing data for the main study variables (n = 43,501), there were 94,555 women included in the present analysis. There were significant differences in the distribution of missing values by menopausal status, race, age, and education (p values < 0.0001), but no difference by BMI (p value = 0.117).

Metabolic syndrome

MetS was defined following the harmonized criteria [26] based on the presence of three of the following components: (1) High waist circumference (WC): >88 cm; (2) Dyslipidemia or self-reported history of high cholesterol level; (3) High blood pressure or self-reported history of hypertension; (4) High glucose or self-reported history of diabetes. The NIH-AARP dataset did not include data on high-density lipoprotein cholesterol (HDL) and triglycerides; therefore, the current analysis is based on 4 out of the 5 MetS components. Women with races other than Black or Non-Hispanic White had substantial missing values for the main covariates, thus the present study was limited to women who self-reported Non-Hispanic White or Black race.

Breast cancer mortality

Breast cancer incidence was ascertained from cancer registries in 6 US states including California, Florida, Louisiana, New Jersey, North Carolina, and Pennsylvania, and two metropolitan areas (Atlanta, Georgia and Detroit, Michigan) with complete case ascertainment, and from registries in three additional states (Arizona, Nevada, and Texas) where participants relocated to during the follow-up period [27]. Breast cancer mortality was determined based on data from the National Death Index (NDI), Social Security Administration Death Master File (SSADMF), Cancer Registry Returns, and Follow-up Questionnaire Responses.

Covariates

Baseline demographic covariates in this study included age (categorical), BMI derived from weight and height (normal BMI: 18.5 to < 25 kg/m2; overweight/obese: ≥ 25 kg/m2), region (Mid-West, North East, South, West), race (Black or White), and marital status (married, widowed, divorced, separated, or never married). Baseline behavioral covariates included physical activity (never, rarely, 1–3 times/month, 1–2 times/week, 3–4 times/week or ≥ 5 times/week), and smoking status (yes/no). Breast cancer hormonal receptor subtypes (estrogen receptor (ER), progesterone receptor (PR)) were identified from cancer registry data according to established protocols [19]. However, human epidermal growth factor receptor 2 (HER2) status was not assessed in the majority of women and was not included in the analysis.

Statistical analysis

Differences in baseline characteristics by breast cancer mortality status were assessed using Chi-squared (χ2) tests for categorical variables. To estimate the HR of breast cancer mortality, a series of Cox proportional hazards regression models were fitted with MetS as the exposure—comparing women with MetS with those without MetS. The primary outcome of interest was risk of breast cancer specific mortality assessed from time of entry into the cohort. The proportional hazard assumption was tested using cumulative sum of martingale residuals and Kolmogorov-type supremum test [28]. In separate models, each individual MetS component and number of MetS components present were evaluated in relation to breast cancer mortality. Each model was adjusted for age, race, education, region, smoking, physical activity, marital status, and BMI (in models not stratified by BMI), and in sensitivity analysis, BMI-stratified models were further adjusted for continuous measures of BMI to address residual confounding by BMI. These associations were also evaluated by BMI (normal weight and overweight/obese), menopausal status (pre- and post-menopause), and hormone-receptor subtypes (ER+/PR+, ER+/PR−, and ER−/PR−) analysis. Effect modification in the stratified models was assessed using the Breslow-Day–Tarone test [29] and likelihood ratio test for interaction terms. Individuals were censored at the time of death due to non-breast cancer causes, loss to follow-up, or December 31, 2011, whichever happened first. Results are presented as adjusted HRs and corresponding 95% Confidence Intervals (CIs); p values ≤ 0.05 were considered statistically significant, and for interaction terms, p values ≤ 0.1 were considered statistically significant. All statistical analyses were conducted using SAS version 9.4, SAS Institute Inc., Cary, NC, USA.

Results

During a mean follow-up of 14 years (SD 3.6), 6631 women in the NIH-AARP cohort developed breast cancer, and 607 (9%) died due to breast cancer. Compared with women without MetS, those with MetS were more likely to be overweight/obese (91% vs. 52%), have high waist circumference (96% vs. 30%), and were less likely to be current hormone therapy users (40% vs. 47%), all p values < 0.05 (Table 1). Among women with normal BMI, those with MetS had relatively lower 10-year survival probability than those without MetS (Fig. 1a), but there was no significant difference in 10-year survival probability by MetS among women with overweight/obesity (Fig. 1b).

Table 1 Baseline characteristics of women in the NIH-AARP cohort by baseline MetS (N = 94,555)
Full size table
Fig. 1
figure 1

Ten-year survival probability by MetS status. a Normal BMI, b overweight/obese

Full size image

In Cox regression models (Table 2), MetS was associated with a twofold increased risk of breast cancer mortality in crude age-adjusted models (HR 2.12, 95% CI 1.35, 3.31) and a 73% increased risk of breast cancer mortality in the fully adjusted model (HR 1.73, 95% CI 1.09, 2.75). High WC (HR 1.32, 95% CI 1.03, 1.70), high cholesterol (HR 1.24, 95% CI 1.05, 1.46), and high blood pressure (HR 1.24, 95% CI 1.05, 1.46) were each associated with increased risk of breast cancer mortality overall in multivariable adjusted models. Furthermore, compared to individuals without any component of MetS present, the risk of breast cancer mortality increased steeply as the number of MetS components increased. For one component, the risk was 44% higher (HR 1.44, 95% CI 1.02, 2.04), for two by 87% (HR 1.87, 95% CI 1.29, 2.71), and for three or more the risk doubled (HR 2.02, 95% CI 1.29, 3.17). Risk of breast cancer mortality increased by 24% for every additional component of MetS present (HR 1.24, 95% CI 1.09, 1.41, p-trend = 0.001). When stratified by stage, MetS was associated with 29% increased risk of cancer mortality for early-stage (stage I–II) diagnoses (HR 1.29, 95% CI 0.60, 2.77), and over a twofold increased risk for late-stage diagnoses (HR 2.59, 95% CI 1.28, 5.24).

Table 2 Hazard ratios (HRs) and 95% confidence intervals for the association between MetS and breast cancer mortality by BMI
Full size table

MetS was associated with a 28% (HR 1.28, 95% CI 0.30, 5.45) and 48% (HR 1.48, 95% CI 0.81, 2.72) increased risk of breast cancer mortality among women with normal BMI and overweight/obesity, respectively, as shown in Table 2. However, these associations were not statistically significant, and the formal test of effect modification by BMI was not statistically significant (Breslow-Day Test P = 0.716; P interaction = 0.147). High cholesterol was associated with a 39% increased risk of breast cancer mortality among women with normal BMI (HR 1.39, 95% CI 1.05, 1.83), while high waist circumference (HR 1.40, 95% CI 1.04, 1.89) and high blood pressure (HR 1.26, 95% CI 1.03, 1.58) were associated with increased risk of breast cancer mortality among women with overweight/obesity. In sensitivity analysis considering non-breast cancer deaths as competing risk, the association between MetS (HR 2.40, 95% CI 1.74, 3.30), high WC (HR 1.25, 95% CI 1.05, 1.48), high cholesterol (HR 1.12, 95% CI 1.00, 1.25), high blood pressure (HR 1.20, 95% CI 1.00, 1.43), and high fasting glucose (HR 1.61, 95% CI 1.33, 1.94) with breast cancer mortality became stronger (data not shown).

In analyses stratified by menopausal status (Table 3), MetS was associated with a statistically significant twofold higher risk of breast cancer mortality among post-menopausal women in the fully adjusted model (HR 2.07, 95% CI 1.32, 3.25). Among this specific group, MetS was not significantly associated with breast cancer mortality in women with normal weight (HR 0.91, 95% CI 0.22, 3.66), but associated with non-significantly increased risk of mortality among women with overweight/obesity (HR 1.15, 95% CI 0.81, 1.64). However, high cholesterol (HR 1.39, 95% CI 1.05, 1.84) and having one (HR 1.72, 95% CI 1.11, 2.67) or two (HR 1.96, 95% CI 1.14, 3.38) components of MetS in normal BMI post-menopausal women increased the risk of breast cancer mortality. The combination of high blood pressure, high fasting blood glucose, and high WC was associated with 72% higher risk of breast cancer mortality (HR 1.72, 95% CI 1.05, 2.83), while other combinations did not reach statistical significance (Table 4). In models evaluating differences by hormone-receptor status (Table 5), MetS was associated with higher risk of breast cancer mortality for ER+/PR + subtypes (HR 1.28, 95% CI 0.52, 3.16) and ER+/PR- (HR 1.89, 95% CI 0.36, 10.05), but lower risk of mortality for ER-/PR- subtypes (HR 0.44, 95% CI 0.11, 1.75) breast cancer. However, none of these associations by subtype reached statistical significance.

Table 3 Hazard ratios (HRs) and 95% confidence intervals for the association between MetS and breast cancer mortality by menopausal status
Full size table
Table 4 Hazard ratios (HRs) and 95% confidence intervals for the association of combinations of MetS components with breast cancer mortality
Full size table
Table 5 Hazard ratios (HRs) and 95% confidence intervals for the association between MetS and breast cancer mortality by breast cancer subtype
Full size table

Discussion

In the large prospective NIH-AARP cohort study, MetS was significantly associated with increased risk of breast cancer mortality, especially among post-menopausal women in a dose–response manner. That is, the risk of breast cancer mortality increased significantly as the number of MetS components increased, with over a threefold increase in risk of breast cancer mortality observed among women with four components of MetS compared with women with none. MetS was also associated with increased risk of breast cancer mortality in women who had normal BMI and overweight/obesity; however, these increases were not statistically significant. Further, the combination of high WC, high cholesterol, and high blood pressure was most strongly associated with increased risk of breast cancer mortality.

These findings are consistent with two previous reports in the US showing that women with MetS had 26% [11] to twofold [18] higher risk of breast cancer mortality, and with a European study [15] documenting that women with MetS had 23% higher risk of breast cancer mortality. Other studies also support an association between MetS and poor breast cancer prognosis with results indicating an association of MetS with advanced cancer stage, metastasis, recurrence, and mortality [12, 14]. Our findings on the associations between individual MetS components and risk of breast cancer mortality are also consistent with the result of a Norwegian study that found women in the highest tertiles of cholesterol and blood pressure had a 29% and 41% higher risk of breast cancer mortality, respectively [16]. A previous study observed that high blood glucose was associated with poor breast cancer prognosis [30]. Studies evaluating the differences in the association between MetS and breast cancer mortality by menopausal status have been limited and results are inconsistent [31]. Bjørge et al. [15] evaluated the associations by age group (< 50, 50–59, and ≥ 60 years) and found a positive association between MetS and breast cancer mortality in those ≥ 60 years old but found no association in younger age groups. Some studies show that high WC is associated with poor breast cancer prognosis among post-menopausal women [32, 33], while the association in pre-menopausal women is less clear [31, 32]. To our knowledge, no study has simultaneously evaluated whether the association between MetS and breast cancer mortality varies by both BMI and menopausal status. Among post-menopausal women, there was consistently higher risk of breast cancer mortality associated with MetS, and a 45% increased risk of breast cancer mortality associated with high waist circumference. However, the association between MetS and breast cancer mortality among post-menopausal women were attenuated in BMI-stratified analysis. Among post-menopausal women with normal BMI, high cholesterol was associated with 39% higher risk of breast cancer mortality and the risk almost doubled in those with at least 2 components of MetS. Recent studies that evaluated the association of MetS with breast cancer mortality by subtype [17, 20] found no significant association between MetS and breast cancer survival in models of subtype analysis; however, low HDL cholesterol was associated with poor survival among women with triple-negative breast cancer [20]. Although not statistically significant, we observed a trend towards increased breast cancer mortality among women with ER + breast cancer, and an inverse association among women with ER- breast cancer, highlighting the need for more research in this area.

Several concerted pathways may predispose patients with MetS or metabolic dysregulation in general to poor breast cancer prognosis. Hyperinsulinemia, a common feature of MetS, promotes tumor growth, angiogenesis, and metastasis, and has anti-apoptotic properties [34, 35]. Central obesity and the related increased adiposity may intensify aromatase activity, which converts androgens to estrogen, a hormone that promotes breast tumor growth and tumor cell survival [36,37,38]. The hormonal changes associated with menopause may also lead to central adiposity [39], and is seen as a driving factor for MetS after menopause [40]. Central adiposity is also related to a risk of insulin resistance, glucose intolerance, high blood pressure, and dyslipidemia [41]. Carriers of a single-nucleotide polymorphism (SNP) APOA1 rs670 A/A associated with dyslipidemia, a component in MetS, were found to show about a threefold higher risk of breast cancer recurrence and nearly a fourfold higher risk of breast cancer mortality [42], suggesting a potential role for gene–environment interaction driving MetS-associated breast cancer mortality risk. Abdominal obesity is also associated with chronic inflammation [43], which in turn is related to poor breast cancer prognosis [44, 45], and studies suggest that cholesterol, another component of MetS, may be involved in cell signaling pathways in breast cancer [46] and may promote cancer metastasis [47, 48]. The mechanisms through which MetS results in poor breast cancer outcomes are likely complex, multifaceted, and not yet fully elucidated.

There are several strengths and limitations relevant to the interpretation of these results. First, we were able to investigate the prospective association of MetS and components with breast cancer mortality in a large US cohort, and add important information to this literature regarding the association among pre- and post-menopausal women, those with normal weight and overweight/obesity, and by hormone-receptor subtype. Second, exposure and covariate data were obtained at baseline in a cancer-free cohort, reducing the risk of recall bias and/or reverse causality since measurement of these variables preceded diagnosis of breast cancer. Third, data on breast cancer mortality in the NIH-AARP cohort were obtained from cancer registries with high levels of ascertainment accuracy (> 95% ascertainment rate) [49], thereby minimizing the risk of outcome misclassification [24]. The major limitations of this study were the limited sample size in sub-group analysis that may have limited our statistical power in detecting the existence of statistically significant associations, especially in analyses stratified by BMI and menopausal status. In addition, lack of detailed treatment data in the NIH-AARP cohort precluded our ability to adjust for these important variables, so future studies with treatment information will be needed to fill this critical gap. While we were able to evaluate associations by hormone-receptor subtype focusing on estrogen and progesterone receptors (ER and PR), we were limited by lack of data on HER2 status in the cohort to comprehensively evaluate the major breast cancer subtypes. Another limitation of this study is regarding the data on MetS exposure; since there were no clinical assessments of women in NIH-AARP cohort at baseline, we relied on self-reports of MetS components and could only assess 4 out of the 5 components. This may have resulted in underestimation of MetS prevalence. However, given the prospective cohort design, we expect that any misclassification would likely be non-differential in relation to the outcome and bias the estimates towards the null. Future studies with larger sample sizes and objective baseline measures of MetS are needed to fully validate our findings.

In conclusion, MetS was observed to be a strong risk factor for breast cancer mortality in a large US prospective cohort, especially among post-menopausal women. Future studies are needed to better elucidate the mechanisms linking metabolic dysregulation with tumor subtypes and mortality outcomes, especially among younger patients. Nevertheless, given the link between poor metabolic health and poor health outcomes in general, clinical strategies and lifestyle modification addressing MetS may be beneficial if incorporated into routine cancer care.