Association of white blood cell count with breast cancer burden varies according to menopausal status, body mass index, and hormone receptor status: a case-control study

Breast cancer is a heterogeneous disease that among Korean women has a peak incidence in the perimenopausal period. The full epidemiological characteristics of breast cancer in Korean women are not yet properly understood. We investigated whether white blood cell (WBC) is related to breast cancer burden according to estrogen receptor (ER) and progesterone receptor (PR) status in the context of body mass index and menopausal status. We conducted a large case-control study and compared WBC counts between patients with breast cancer (N = 4,402) and propensity score-matched controls (N = 4,402) selected from the Korean National Health and Nutrition Examination Survey (KNHANES). We stratified the study sample by ER/PR status, menopausal status, and body mass index and assessed the association between WBC count and breast cancer burden using multinomial logistic regression. Compared with controls, non-obese patients with ER+/PR+ breast cancer had significantly higher WBC counts regardless of menopausal status (OR 1.293 95% CI 1.139–1.363, p < 0.001 in premenopausal and OR 1.049 95% CI 1.019–1.295, p = 0.023 in postmenopausal). There was no relationship between WBC count and ER+/PR+ breast cancer among premenopausal obese women. Furthermore, premenopausal non-obese women and postmenopausal obese women with ER+/PR+ breast cancer had higher WBC counts than those with ER−/PR− breast cancer. Further larger-scale prospective cohort studies are warranted to determine these associations in the future.

triglyceride, HDL-cholesterol, aspartate aminotransferase (AST), and alanine aminotransferase (ALT) were measured using an automated chemistry analyzer (Hitachi 7600, Tokyo, Japan). Also, an overnight-fasting venous blood specimen was collected from all patients within one week after breast cancer diagnosis. WBC counts were quantified by an automated blood cell counter (ADVIA 120, NY, USA). Fasting plasma glucose, total cholesterol, triglyceride, HDL-cholesterol, AST, and ALT were measured using Hitachi 7600-110 Chemistry System Autoanalyzer (Hitachi, Tokyo, Japan). Previous study comparing automated blood cell counters has shown that there is no difference in reference intervals for WBC counts between two analyzers used in controls and cases, respectively 28 . After definitive surgery for breast cancer, tumor specimens were stained for ER and PR. Specimens with more than 1% nuclear-stained cells were defined as positive for ER and/or PR according to the guidelines of the American Society of Clinical Oncology-College of American Pathologists 29 . Menopause status, menarchial age, menopausal age, and history of breastfeeding were documented among the patients and the controls by a self-administered questionnaire. Menopause was defined as the cessation of menstruation for more than 12 months or surgical menopausal status in cases and controls. statistical analysis. We compared the characteristics of the patients and propensity score-matched controls using paired t-tests for continuous variables and McNemar's test for categorical variables. Propensity score was created using age, the most well-known confounding variable. We examined potential effects on WBC counts of obesity (BMI ≥ 25 kg/m²) and ER/PR status (ER − /PR − , ER + /PR − or ER − /PR + , or ER + /PR + ) according to menopausal status. This stratification was based on previous meta-analysis showing the different association between body weight and breast cancer risk among ER − /PR − , ER + /PR − or ER − /PR + , or ER + /PR + by menopausal status 27 . We performed an analysis of covariance (ANCOVA) adjusted for WBC counts of controls to compare differences in WBC counts between matched patients and controls among groups in which the patients were ER − /PR − , ER + / PR − or ER − /PR + , or ER + /PR + . Using multinomial logistic regression, we measured the strength of correlation between WBC count (×10 3 cells/μL) and breast cancer burden according to ER/PR status stratified by BMI and menopause after adjusting for continuous variables (age, systolic blood pressure, fasting plasma glucose, total cholesterol, triglyceride, HDL-cholesterol, ALT, and age at menarche or age at menopause) and categorical variables (breastfeeding, hypertension medication, and diabetes medication), including significant variables (p < 0.05) in univariate analysis with clinically important variables, while further considering multicollinearity. To test the combined effect of menopause status, BMI, and WBC, we tested their interactions with the interaction term for menopause status*BMI*WBC by multinomial logistic regression models for outcome. The interaction among menopause status, BMI, and WBC was tested at a significance level of 0.2. We conducted all analyses using the SAS statistical software, version 9.2 (SAS Institute Inc., Cary, NC, USA). All statistical tests were two-sided, with statistical significance determined by p < 0.05. Table 1 shows the characteristics of the patients and matched controls according to menopausal status. The mean age of the premenopausal and postmenopausal women was 42.5 ± 6.1 years and 58.5 ± 7.7 years, respectively. The percentage of premenopausal women was 51.8%. Among the premenopausal women, the mean BMI of the controls (22.7 ± 3.1 kg/m 2 ) was higher than that of the patients (22.5 ± 3.1 kg/m 2 ; p < 0.001). There was no significant difference in mean BMI between the postmenopausal controls (24.2 ± 3.3 kg/m 2 ) and patients (24.2 ± 3.2 kg/m 2 ). Regardless of menopausal status, the patients in each group had higher WBC counts, systolic blood pressure, diastolic blood pressure, and fasting plasma glucose than the controls (p < 0.001 for each comparison). HDL cholesterol was lower in the premenopausal patients than in the matched controls, while there was no difference in HDL cholesterol between the postmenopausal patients and controls. Regardless of menopausal status, history of breastfeeding was more prevalent among the controls, whereas history of medication for hypertension was more prevalent among the patients. History of medication for diabetes was more prevalent among the postmenopausal patients than among the matched controls, however there was no difference in history of diabetes medication between the premenopausal patients and controls.

Results
Premenopausal non-obese women with ER − /PR − or ER + /PR + breast cancer had higher WBC counts than their matched controls (p = 0.010 and p < 0.001, respectively). Postmenopausal women with ER + /PR − or ER − / PR + breast cancer exhibited similar trends, but the difference between the patients and controls was not significant. In contrast to the patterns in premenopausal non-obese women, premenopausal obese women with ER − /PR − breast cancer or ER + /PR − or ER − /PR + breast cancer did not have higher WBC counts than their age-matched controls, and the trend for higher WBC counts relative to those in the controls was weakened in those with ER + /PR + breast cancer ( Table 2, Fig. 2a). WBC counts were significantly higher in postmenopausal women with ER + /PR + breast cancer compared with those in matched controls, irrespective of obesity status (p < 0.001 and p = 0.014, respectively; Table 3, Fig. 2a). Figure 2b shows the differences in WBC counts between patients and matched controls by ER/PR status, menopause status, and obesity status. ANCOVA showed that the difference in WBC counts between patients and matched controls was significantly greater for premenopausal non-obese women with ER + /PR + breast cancer than for those with ER − /PR − breast cancer or ER + /PR − or ER − /PR + breast cancer (p = 0.020 and p = 0.038, respectively). Among postmenopausal women, the difference in WBC counts between patients and controls was greater for patients with ER + /PR + breast cancer than for those with ER − /PR − breast cancer or ER + /PR − or ER − /PR + breast cancer, regardless of whether the women were obese (p = 0.002 and p < 0.001, respectively) or non-obese (p = 0.002 and p = 0.009, respectively).

Discussion
In this case-control study, we found that WBC count was associated differently with breast cancer burden depending on menopausal status, BMI, and ER/PR status. Premenopausal non-obese women with ER + /PR + breast cancer had elevated WBC counts compared with both controls and premenopausal non-obese women with ER − /PR − breast cancer. Those trends were weaker in premenopausal obese women. Postmenopausal non-obese women with ER + /PR + breast cancer had elevated WBC counts compared with controls, while postmenopausal obese women with ER + /PR + breast cancer had elevated WBC counts compared with postmenopausal obese women with ER − /PR − breast cancer.
Emerging evidence suggests that chronic low-grade inflammation plays an important role in cancer development. Both menopause and obesity can also play a crucial role in the development of breast cancer 30 , but their interaction may differ according to menopausal status 22,31,32 . Positive associations between obesity and breast cancer risk have been consistently observed in postmenopausal women [33][34][35][36] . However, there is substantial evidence that there is an inverse association between obesity and breast cancer risk in premenopausal women [37][38][39] . Our results showed that WBC count was not associated with breast cancer burden in premenopausal obese women.
WBCs, including neutrophils, monocytes, and eosinophils, produce reactive oxygen species (ROS) and nitric oxide species (NOS), which are chemically reactive molecules 40 . Unless ROS and NOS are properly neutralized by the antioxidant defense system, they can cause damage to cellular proteins, lipids, and DNA that may lead to the accumulation of genetic instability, affecting single nucleotide polymorphisms (SNPs) or upregulating the PI3K-Akt pathway for carcinogenesis 41 . Large-sample studies that attempted to evaluate the association between WBC counts and breast cancer risk without stratification by menopausal status and obesity have produced inconsistent results 24,25 . A prospective study demonstrated that leukocyte counts may be a predictor of breast cancer, but the study included only postmenopausal women 20 . Akinbami et al. 42 reported that WBC counts were higher in patients with breast cancer than in controls, but their study did not include information about menopausal status. Okuturlar et al. showed that neutrophil levels were associated with the risk of breast cancer, including Stage IV breast cancer 43 . None of the previous studies included ER/PR status in their analysis. A recent meta-analysis assessed the association between the neutrophil-to-lymphocyte ratio as a biomarker using WBC subtypes and breast cancer prognosis 44 . In that study, which was performed in patients with breast cancer without control, patients with a higher neutrophil-to-lymphocyte ratio had a higher relapse and a shorter overall survival. Subgroup analysis showed that studies performed in Eastern countries had perfectly homogeneous results, www.nature.com/scientificreports www.nature.com/scientificreports/ whereas Western countries did not. A distinguishing feature of our study is that it attempts to elucidate a role of the interaction between WBC count and ER/PR status in the context of menopause and BMI.
The prevalence of obesity among Korean women has gradually decreased since 2001 45 , but the incidence of breast cancer has increased over the last decade 4 . The age-frequency distribution of breast cancer among Korean women is unimodal, with peak incidence at 45-49 years of age 5 . Dense breast on mammography, a potent risk factor for breast cancer, is more prevalent among Korean women, especially before menopause, than among women from Western countries 46 . Those distinctive epidemiological features warrant more investigation of the interplay between well-known risk factors such as obesity and menopause and the emerging role of inflammation in cancer development.  www.nature.com/scientificreports www.nature.com/scientificreports/ While metabolic syndrome and insulin resistance, as inflammatory conditions, have been noticed to be associated with breast cancer development and subsequent progression [7][8][9] , there are controversies, regarding discrepancies according to menopause [21][22][23] , as well as some limitations in Korean women. One Korean cohort study in which 23,830 Korean women 50-64 years of age were examined reported that metabolic syndrome was related to the risk of breast cancer after adjustment for age and BMI 47 . That study did not account for menopausal status, although most of the participants were likely postmenopausal, nor did it consider different breast cancer subtypes. A recent epidemiological study of postmenopausal Korean women showed that insulin resistance was independently associated only with luminal B subtype breast cancer 10 , which is included in the ER + /PR + phenotype 48 .
ER and PR are found in about two-thirds of breast cancers, representing favorable therapeutic and prognostic factors. In terms of breast cancer pathogenesis, the risk of breast cancer development associated with weight gain, a surrogate for increasing subclinical inflammation, has been shown to be higher for ER + /PR + breast cancer than for ER − /PR − breast cancer 27 . Approximately 40% of ER + breast cancers fail to respond to hormone therapy 49 . ER and PR status can be a precipitating factor in breast cancer development through its interplay with inflammation, which may also influence endocrine resistance. Pro-inflammatory cytokines, such as tumor necrosis factor-α (TNF-α), increase the transcriptional activity of the NF-κB and JNK pathways and may subsequently induce tumorigenesis or resistance to hormone therapy 50 . In a case-case analysis of epidemiological risk factors for breast Figure 2. Comparison of WBC counts between patients and matched controls (ER − /PR − , ER + /PR − or ER − / PR + , ER + /PR + ) according to menopause status and obesity status. * P < 0.05, calculated by paired t-test (a). Differences in WBC counts between patients and matched controls according to menopause status and obesity status. * Difference between ER − /PR − vs. ER + /PR − or ER − /PR + and ER + /PR + , P < 0.05, calculated by ANCOVA (adjusted WBC counts of controls). † Difference between ER + /PR − or ER − /PR + vs. ER − /PR − and ER + /PR + , P < 0.05, calculated by ANCOVA (adjusted WBC counts of controls) (b).
www.nature.com/scientificreports www.nature.com/scientificreports/ cancer, women less than 50 years of age with ER − /PR − tumors were more likely to be obese than those with ER + / PR + tumors 51 . Obesity per se has been mainly associated with postmenopausal ER + /PR + breast cancer 27 . Our study reported that menopausal status, BMI, and WBC showed a significant interaction by multinomial logistic regression models. Considering how previous studies have shown that obesity can be associated differently with breast cancer status, our findings suggest that among non-obese women, WBC may be related to breast cancer burden. Among Korean women, breast cancer has peak incidence in the perimenopausal period, with the number of obese women being lower than that of non-obese women 10,52 .
Our study has a few limitations. First, because our study was a case-control study, the exact cause-effect relationship between WBC count and ER/PR status according to obesity status and menopausal status remains   www.nature.com/scientificreports www.nature.com/scientificreports/ unclear. Although it is plausible that WBC may reflect underlying inflammation and, in turn, affect breast cancer risk, higher WBC counts may result from the stress that comes after receiving a cancer diagnosis. Also, WBC counts in controls and cases were measured separately in different laboratories, with very high concordance of general blood cell counts among various automated hematology analyzers 53,54 , which may have led to differential misclassification of laboratory errors. Prospective longitudinal studies are needed to verify the effects of those interactions on the development of breast cancer, especially in Asian women. Second, some inflammatory Premenopause, BMI < 25 kg/m² Premenopause, BMI ≥ 25 kg/m² OR (95% CI) p value a OR (95% CI) p value a  Forest plot. Models of premenopausal women were adjusted for age, systolic blood pressure, fasting plasma glucose, total cholesterol, triglyceride, HDL cholesterol, alanine aminotransferase, age at menarche, breastfeeding, hypertension medication, and diabetes medication. Models of postmenopausal women were adjusted for age, systolic blood pressure, fasting plasma glucose, total cholesterol, triglyceride, HDL cholesterol, alanine aminotransferase, age at menopause, breastfeeding, hypertension medication, and diabetes medication. * Difference between control and ER − /PR − , ER + /PR − or ER − /PR + , or ER + /PR + ; P < 0.05, calculated by multinomial logistic regression analysis. † Difference between ER − /PR − and ER + /PR − or ER − /PR + , or ER + /PR + ; P < 0.05, calculated by multinomial logistic regression.
www.nature.com/scientificreports www.nature.com/scientificreports/ markers such as C-reactive protein, interleukin-6, serum amyloid-A, and prostaglandin E2 were not measured at the beginning of the study. Those markers can be indicators of chronic low-grade inflammation but have not been taken into consideration when relating breast cancer to the presence of obesity. Third, cases were selected from hospital registry between 2005 and 2012, but controls were selected between 2010 and 2012 from KNHANES, albeit for proper stratification in cases and for excluding duplication of participants and minimizing missing data in controls.
In conclusion, compared with those in controls, WBC counts were significantly elevated in non-obese patients with ER + /PR + breast cancer, irrespective of menopause. Further larger-scale prospective cohort studies are warranted to determine these associations in the future.