Obesity, diabetes, and survival outcomes in a large cohort of early-stage breast cancer patients

Abstract

Background

To determine the relationship between obesity, diabetes, and survival in a large cohort of breast cancer patients receiving modern chemotherapy and endocrine therapy.

Patients and methods

We identified 6342 patients with stage I–III breast cancer treated between 1996 and 2005. Patients were evaluated according to body mass index (BMI) category and diabetes status.

Results

In a multivariate model adjusted for body mass index, diabetes, medical comorbidities, patient- and tumor-related variables, and adjuvant therapies, relative to the normal weight, hazard ratios (HRs) for recurrence-free survival (RFS), overall survival (OS), and breast cancer-specific survival (BCSS) for the overweight were 1.18 [95% confidence interval (CI) 1.02–1.36], 1.20 (95% CI 1.00–1.42), and 1.21 (95% CI 0.98–1.48), respectively. HRs for RFS, OS, and BCSS for the obese were 1.13 (95% CI 0.98–1.31), 1.24 (95% CI 1.04–1.48), and 1.23 (95% CI 1.00–1.52), respectively. Subset analyses showed these differences were significant for the ER-positive, but not ER-negative or HER2-positive, groups. Relative to nondiabetics, HRs for diabetics for RFS, OS, and BCSS were 1.21 (95% CI 0.98–1.49), 1.39 (95% CI 1.10–1.77), and 1.04 (95% CI 0.75–1.45), respectively.

Conclusions

In patients receiving modern adjuvant therapies, obesity has a negative impact on RFS, OS, and BCSS; and diabetes has a negative impact on RFS and OS. Control of both may be important to improving survival in obese and diabetic breast cancer patients.

introduction

Obesity has reached epidemic proportions, with about two-thirds of the US population and one-third of the world population being either overweight or obese [1]. Obesity is an established risk factor for developing postmenopausal breast cancer [2–4]. Recent data suggest that obesity is also associated with worse survival outcomes (prognosis) in both pre- and postmenopausal breast cancer [5–8]. Diabetes has also reached epidemic proportions, with 17%–20% of the US population over age 65 being diabetic [9]. Diabetes is correlated with an increased risk of breast cancer [10], as well as with poorer breast cancer outcomes [11–13].

Obesity and diabetes share overlapping etiologies. Patients with both conditions have elevated serum insulin levels and show evidence of a chronic inflammatory state [14]. Both conditions are associated with worse breast cancer outcomes in most but not all studies. To our knowledge, no studies have examined the relationship between both obesity and diabetes, and clinical outcomes, in the context of patients receiving modern adjuvant chemotherapy and endocrine therapy. To address these gaps in knowledge, we examined the relationship between body mass index (BMI), diabetes, and survival in a large cohort of early-stage breast cancer patients.

methods

patient selection

Data for this study were obtained from a database prospectively maintained by the Department of Breast Medical Oncology at The University of Texas MD Anderson Cancer Center. We retrospectively studied stage I–III breast cancer patients treated between 1996 and 2005. After exclusion of patients who were male, had multiple episodes of breast cancer, had inflammatory breast cancer, or were treated with neoadjuvant chemo- or endocrine therapy, 6342 patients were included in the analysis.

statistical analysis

For the statistical analysis, patients were evaluated according to BMI (normal weight, overweight, obese) and diabetes status. Patients with BMI <25 were classified as normal or underweight, hereafter referred to as the normal weight group. Those with BMI between 25 and 30 were classified as overweight. And those with BMI ≥30 were classified as obese. Diabetes status at the time of diagnosis was determined from the medical record. Charlson comorbidity scores were calculated as previously described [15].

Outcome measures were defined as follows. Recurrence-free survival (RFS): time from diagnosis to local or distant recurrence or death from any cause, whichever occurred first. Overall survival (OS): time from diagnosis to death from any cause. Breast cancer-specific survival (BCSS): time from diagnosis to death due to breast cancer. Death due to breast cancer was defined as death occurring after recurrence of breast cancer [16]. Patients were censored at last follow-up if the event did not occur.

The χ² test was used to evaluate the association between categorical variables. Survival distributions were estimated using the Kaplan–Meier method and compared via the log-rank test. Multivariate Cox proportional hazards models were used to determine the effects of BMI and diabetes status on survival distributions after adjusting for other potential risk factors (age, race, Charlson comorbidity score, stage, grade, and ER/PR status) and treatments received (adjuvant chemotherapy and adjuvant endocrine therapy). All tests were two sided. P-values <0.05 were considered statistically significant. In pairwise comparisons of BMI groups, P-values <0.017 were considered statistically significant. All analyses were conducted using SAS (version 9.1, Cary, NC) and S-plus (version 8.0, TIBCO Software, Inc., Palo Alto, CA) statistical software.

results

patient characteristics

The final study cohort consisted of 6342 women with a median age of 53 years (Table 1). Most patients had stage I or II disease. Fifty-two % had grade 1 or 2 tumors, 77% were estrogen-receptor (ER) positive, and 21% were HER2 positive. Fifty-six % of patients received adjuvant chemotherapy, and 63% adjuvant endocrine therapy.

Table 1.

Patient, tumor, and treatment characteristics by BMI category and diabetes status

Variables	All patients	BMI category			P-value	Diabetes status		P-value
		<25	25–30	≥30		No	Yes
Total	6342 (100%)	2398	1816	1779		5845	497
Age at diagnosis (years)
<50	2567 (40%)	1153 (48%)	659 (36%)	604 (34%)	<0.001	2462 (42%)	105 (21%)	<0.001
≥50	3775 (60%)	1245 (52%)	1157 (64%)	1175 (66%)		3383 (58%)	392 (79%)
Race
White	4701 (74%)	1926 (80%)	1343 (74%)	1206 (68%)	<0.001	4420 (76%)	281 (57%)	<0.001
African American	622 (10%)	96 (4%)	161 (9%)	304 (17%)		522 (9%)	100 (20%)
Other	1019 (16%)	376 (16%)	312 (17%)	269 (15%)		903 (15%)	116 (23%)
BMI category
<25	2398 (40%)					2336 (42%)	62 (13%)	<0.001
[25–30)	1816 (30%)					1700 (31%)	116 (25%)
≥30	1779 (30%)					1491 (27%)	288 (62%)
Unknown	349					318	31
Diabetes status
No	5845 (92%)	2336 (97%)	1700 (94%)	1491 (84%)	<0.001
Yes	497 (8%)	62 (3%)	116 (6%)	288 (16%)
Charlson comorbidity score
0	6180 (97%)	2357 (98%)	1764 (97%)	1718 (97%)	0.009	5714 (98%)	466 (94%)	<0.001
1	106 (2%)	27 (1%)	37 (2%)	40 (2%)		90 (1%)	16 (3%)
2+	56 (1%)	14 (1%)	15 (1%)	21 (1%)		41 (1%)	15 (3%)
Stage
I/II	5840 (93%)	2239 (94%)	1665 (93%)	1621 (92%)	0.028	5397 (93%)	443 (90%)	0.008
III	430 (7%)	135 (6%)	130 (7%)	134 (8%)		382 (7%)	48 (10%)
Unknown	72	24	21	24		66	6
Nuclear grade
I/II	3220 (53%)	1276 (55%)	914 (52%)	862 (50%)	0.006	2974 (53%)	246 (50%)	0.318
III	2917 (48%)	1051 (45%)	843 (48%)	867 (50%)		2674 (47%)	243 (50%)
Unknown	205	71	59	50		197	8
ER or PR status
ER and PR negative	1334 (23%)	478 (21%)	383 (23%)	402 (24%)	0.122	1222 (23%)	112 (24%)	0.536
ER or PR positive	4572 (77%)	1772 (79%)	1302 (77%)	1274 (76%)		4212 (77%)	360 (76%)
Unknown	436	148	131	103		411	25
HER2 status
Negative	2733 (79%)	1014 (77%)	799 (80%)	824 (81%)	0.072	2489 (78%)	244 (84%)	0.015
Positive	733 (21%)	304 (23%)	206 (20%)	197 (19%)		688 (22%)	45 (16%)
Unknown	2876	1080	811	758		2668	208
Adjuvant chemotherapy
No	2789 (44%)	1050 (44%)	784 (43%)	787 (44%)	0.811	2534 (43%)	255 (51%)	0.001
Yes	3553 (56%)	1348(56%)	1032 (57%)	992 (56%)		3311 (57%)	242 (49%)
Adjuvant endocrine therapy
No	2335 (37%)	840 (35%)	642 (35%)	677 (38%)	0.102	2146 (37%)	189 (38%)	0.560
Yes	4007(63%)	1558 (65%)	1174 (65%)	1102 (62%)		3699 (63%)	308 (62%)

Regarding the study groups of primary interest, normal, overweight, and obese BMI groups represented 40%, 30%, and 30% of the total patient population, respectively (Table 1). Overweight and obese patients were significantly more likely to be older, diabetic, and to have higher stage and higher grade disease than normal weight patients (Table 1). There was no difference in the proportion of patients receiving adjuvant chemotherapy or endocrine therapy between the three BMI groups.

There were 497 diabetic patients (8% of all patients, Table 1). Diabetic patients were significantly more likely to be older, have higher BMI, and have higher stage disease than their nondiabetic counterparts (Table 1). Diabetics were also less likely to receive adjuvant chemotherapy than nondiabetics, but there was no significant difference in adjuvant endocrine therapy between the two groups.

Regarding treatments administered, of the patients who received chemotherapy, 41% received anthracycline-based regimens (predominantly FAC × 6 cycles) and an additional 50% received anthracycline- and taxane-based regimens (predominantly weekly paclitaxel × 12 weeks followed by FAC × 4 cycles). Of the patients who received endocrine therapy, 46% received tamoxifen, 29% received an aromatase inhibitor (AI), and 25% received both.

survival outcomes

Median follow-up was 5.4 years for the censored observations. There were 1175 recurrences and 951 deaths, with 673 deaths due to breast cancer. Univariate analysis by BMI groups showed that RFS and OS distributions were significantly different between the three BMI groups (overall P = 0.011 and overall P = 0.001, Figure 1A and B, respectively). Pairwise comparisons showed that the overweight and obese groups had significantly decreased RFS (P = 0.013 and 0.008, respectively), as well as OS (P = 0.052 and P < 0.001, respectively), compared with the normal weight group. Pairwise comparisons of the obese versus overweight groups did not meet significance (P = 0.837 for RFS and P = 0.087 for OS). Univariate analysis by diabetic groups showed that diabetic patients had significantly decreased RFS (P = 0.034, Figure 1D) and OS (P < 0.001, Figure 1E).

Figure 1.

Survival outcomes by BMI category (A–C) and diabetes status (D–F). RFS, recurrence-free survival; OS, overall survival; BCSS, breast cancer-specific survival.

To determine whether the increased risk of death was related to deaths from breast cancer or deaths from other causes, we examined the end point of BCSS. In this analysis, deaths resulting from breast cancer (defined as deaths occurring after recurrence of breast cancer) were counted as events, and deaths from other causes were censored. In the univariate analysis, higher BMI was associated with decreased BCSS, with a borderline significance (overall P = 0.058, Figure 1C). Pairwise comparisons showed this was driven by the difference in the obese versus normal weight groups (P = 0.016), as the overweight versus normal (P = 0.223) and obese versus overweight (P = 0.267) comparisons were not significant. In contrast to BMI, diabetes status was not associated with decreased BCSS (P = 0.431, Figure 1F).

We also carried out analyses of survival outcomes according to ER and HER2 status. This showed that overweight and obese groups had significantly worse RFS and OS, and a trend for worse BCSS, in the ER-positive subset (Figure 2). This was not true for the ER-negative and HER2-positive subsets (supplementary Figure S1, available at Annals of Oncology online). For the ER-positive subset, we examined survival outcomes by adjuvant endocrine therapy given. We found that overweight and obese groups had significantly worse RFS and OS in patients treated with tamoxifen alone (Figure 3), but not in patients treated with AI alone (Figure 3) or both therapies (supplementary Figure S2, available at Annals of Oncology online).

Figure 2.

Survival outcomes by BMI category for hormone receptor-positive patients.

Figure 3.

Survival outcomes by BMI category for hormone receptor-positive patients treated with tamoxifen alone (A–C) or AI alone (D–F).

multivariate models

We used multivariate Cox proportional hazards models to estimate the hazards of recurrence or death for each cohort after adjusting for BMI, diabetes, Charlson comorbidity score, race, age, stage, grade, hormone receptor status, adjuvant chemotherapy, and adjuvant endocrine therapy. Three models were developed, with the end points of RFS, OS, and BCSS (Table 2). The normal weight BMI or nondiabetic groups were used as the reference groups. In the first model with RFS as the end point, overweight patients were significantly more likely to recur or die compared with normal weight patients [HR, 1.18; 95% confidence interval (CI) 1.02–1.36; P = 0.022] (Table 2). A similar result was seen for the comparison of obese and normal weight patients (HR, 1.13; 95% CI 0.98–1.31; P = 0.098). Diabetic patients were P > 0.05 more likely to suffer a recurrence or die (HR, 1.21; 95% CI 0.98–1.49; P = 0.070).

Table 2.

Univariate and multivariate models

Variable	RFS		OS		BCSS
	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
Unicovariate analysis
BMI ([25–30) versus <25)	1.18 (1.04–1.34)	0.013	1.18 (1.00–1.38)	0.052	1.12 (0.93–1.36)	0.228
(≥ 30 versus <25)	1.19 (1.05–1.36)	0.008	1.36 (1.46–1.60)	<0.001	1.26 (1.04–1.52)	0.017
Diabetes status (Yes versus No)	1.22 (1.01–1.46)	0.034	1.67 (1.35–2.05)	<0.001	1.12 (0.84–1.50)	0.431
Multicovariate analysis
Age at diagnosis (years)
(≥50 versus <50)	0.75 (0.67–0.85)	<0.001	0.95 (0.82–1.11)	0.507	0.70 (0.59–0.84)	<0.001
Race (Black versus White)	0.98 (0.81–1.19)	0.873	1.12 (0.89–1.41)	0.344	1.11 (0.85–1.46)	0.438
(Other versus White)	0.91 (0.77–1.08)	0.274	0.82 (0.66–1.02)	0.078	0.93 (0.73–1.19)	0.580
BMI ([25–30) versus <25)	1.18 (1.02–1.36)	0.022	1.20 (1.00–1.42)	0.045	1.21 (0.98–1.48)	0.072
(≥ 30 versus <25)	1.13 (0.98–1.31)	0.098	1.24 (1.04–1.48)	0.016	1.23 (1.00–1.52)	0.050
Diabetes status (Yes versus No)	1.21 (0.98–1.49)	0.070	1.39 (1.10–1.77)	0.007	1.04 (0.75–1.45)	0.803
Charlson comorbidity score
(1 versus 0)	1.74 (1.21–2.51)	0.003	1.98 (1.32–2.99)	0.001	1.22 (0.63–2.36)	0.562
(2+ versus 0)	2.20 (1.34–3.62)	0.002	3.11 (1.85–5.22)	<0.001	2.70 (1.27–5.76)	0.010
Stage (III versus I/II)	2.24 (1.86–2.68)	<0.001	2.47 (1.98–3.09)	<0.001	2.54 (1.98–3.24)	<0.001
Nuclear grade (III versus I/II)	1.58 (1.38–1.81)	<0.001	1.85 (1.56–2.19)	<0.001	2.24 (1.81–2.77)	<0.001
ER or PR status
(Positive versus negative)	0.86 (0.72–1.03)	0.091	0.80 (0.64–0.99)	0.039	0.69 (0.54–0.89)	0.005
Adjuvant chemotherapy
(Yes versus No)	1.28 (1.12–1.47)	<0.001	1.01 (0.85–1.19)	0.917	1.74 (1.38–2.18)	<0.001
Adjuvant endocrine therapy
(Yes versus No)	0.6 (0.53–0.73)	<0.001	0.69 (0.57–0.84)	<0.001	0.74 (0.58–0.94)	0.014

RFS, recurrence-free survival; OS, overall survival; BCSS, breast cancer-specific survival; HR, hazard ratio; CI, confidence interval.

With regard to OS, compared with normal weight patients, overweight and obese patients also had significantly increased risks of death, with HRs of 1.20 (95% CI 1.00–1.42; P = 0.045) and 1.24 (95% CI 1.04–1.48; P = 0.016), respectively (Table 2). Diabetic patients had a significantly higher risk of death compared with nondiabetic patients (HR, 1.39; 95% CI 1.10–1.77 P = 0.007).

With regard to BCSS, higher BMI was associated with a significantly increased risk of breast-cancer specific death for obese compared with normal weight patients (HR, 1.23; 95% CI 1.00–1.52; P = 0.050) (Table 2). In contrast, diabetes status was not associated with differences in BCSS (HR, 1.04; 95% CI 0.75–1.45; P = 0.803).

discussion

The main finding of this study is that in a modern cohort of early-stage breast cancer patients receiving anthracycline- and taxane-based chemotherapy and tamoxifen- and aromatase inhibitor (AI)-based endocrine therapy as indicated, obesity has a negative impact on various survival outcomes. It has a significant negative impact on OS, and a trend or borderline significant negative impact on RFS and BCSS. This is after adjustment for multiple confounding variables. To our knowledge, this is the first report of such a finding in a large cohort being treated with essentially current standard of care therapies, and adjusted for the presence or absence of diabetes. We also found that while diabetic patients had poorer RFS and OS, the effect of diabetes on BCSS was not significant. This suggests that diabetic patients may have poorer survival due to diabetes and its complications, but that they may not have worse outcomes related to breast cancer.

Previous studies have reported similar findings [5, 17, 18]. A recent meta-analysis of the effects of obesity on survival in breast cancer, involving 43 studies with a median sample size of 1192 (range 100–424 168), found a pooled HR for OS of 1.33 and for BCSS of 1.33 [19]. After excluding studies that did not adjust for age, menopausal status, ER/PR status, and HER2 status, the pooled HR for OS was 1.28. This agrees well with our HR for OS of 1.24, which is adjusted for more factors. It is likely that the pooled HR for BCSS, which was not reported, would have been similarly close to ours.

The survival curves suggest that differences arise primarily after 5 years. Indeed, annualized HRs showed no differences between BMI groups for RFS, OS, and BCSS at years 1–3, and minimal differences at years 4–5 (data not shown). So the differences in survival outcomes between BMI groups are mainly due to events after the first 5 years (obesity is associated with late recurrences, consistent with a recent report [5]).

Several hypotheses have been proposed to explain the poorer survival outcomes observed with increasing BMI, involving factors from diagnosis to treatment of obese breast cancer patients [6]. Obese patients may undergo less mammographic screening, and their increased breast adiposity may delay tumor detection and diagnosis until tumors are larger [20–22]. We did observe a small increase in stage with increasing BMI, but this is also consistent with more aggressive tumor biology. Another possibility is that obese patients were undertreated relative to normal weight patients [23]. We cannot directly address this issue because we do not have data on chemotherapy doses or number of cycles given. However, the recommended approach in the Department of Breast Medical Oncology was to treat obese patients with chemotherapy doses based on actual body weight, consistent with recent American Society of Clinical Oncology practice guidelines [24]. Nevertheless, we cannot completely rule out an effect of suboptimal chemotherapy dosing on the clinical outcomes of obese patients. Another possibility is death due to comorbid conditions leading to poorer survival. The similar magnitudes of the hazard ratios for BCSS and OS argue against this. In addition, the multivariate model adjusted for comorbid conditions. This leaves the possibility of altered biology leading to more aggressive tumors in obese patients, which we believe may be a significant component of the poorer outcomes for obese patients in our study and recent similar studies.

The analysis by ER and HER2 marker status suggests that higher BMI is associated with inferior outcomes in the ER-positive, but not the ER-negative and HER2-positive, subsets. Our results for ER-positive patients are consistent with previous reports. Sparano et al. [25] recently reported worse DFS, OS, and BCSS in overweight and obese patients in the ER-positive, but not the HER2-positive or triple negative, subsets in the ECOG 1199 trial. The NSABP B-14 trial of tamoxifen versus placebo in ER-positive, node-negative patients reported decreased BCSS and OS in obese (but not overweight) compared with normal weight patients [16]. In the ATAC trial of endocrine therapy in postmenopausal ER-positive patients, it was found that the HR for death after breast cancer recurrence was 1.16–1.20 for patients with a BMI 25–30 (nonsignificant), 1.21 for BMI 30–35 (significant), and 1.55 for BMI >35 (significant), when compared with BMI <23 [26]. These results are biologically plausible, due to the known association of adipose tissue with higher aromatase activity and estrogen levels [27]; and due to increased activation and potential crosstalk between the insulin/IGF and ER signaling pathways in obese patients [28]. Regarding the lack of association between BMI and survival in the ER-negative and HER2-positive subsets, we note that the numbers of patients in these groups in our study is relatively small and could limit the ability to detect these associations. A recent meta-analysis by Niraula et al. [29] suggests that higher BMI is associated with inferior survival outcomes in both ER-positive and ER-negative breast cancer.

The analysis of the ER-positive subset according to adjuvant endocrine therapy suggests that higher BMI is associated with worse outcomes in patients treated with tamoxifen, but not with AI or both. Similar to our results, a recent retrospective study of postmenopausal ER-positive patients did not find poorer DFS and OS in overweight and obese, when compared with normal weight, patients [35]. In contrast, the ABCSG-12 trial (in premenopausal) and the ATAC trial (in postmenopausal) patients did find a correlation between higher BMI and worse outcomes with AI, but not tamoxifen, therapy [26, 30]. The differences in these results are likely attributable to differences in patient populations, disease characteristics, treatment, and analysis methods; and highlight that the effects of BMI are highly context dependent and require further study.

Regarding diabetes and survival outcomes, we found in univariate analyses that diabetes was significantly associated with worse RFS and OS, but not BCSS. This was confirmed in multivariate analyses. Our HR for OS of 1.39 is consistent with a recent meta-analysis that found a HR for all-cause mortality of 1.49, which decreased to 1.41 after adjusting for publication bias [13]. Our finding contrasts with the report of Srokowski et al. [31] that diabetics treated with chemotherapy had decreased BCSS when compared with nondiabetics. It should be noted, however, that that study did not adjust for obesity, and focused on older diabetic patients, who may have been undertreated due to older age and higher prevalence of diabetes and medical comorbidities.

The association of diabetes with differences in RFS and OS, but not BCSS, suggests that diabetic (when compared with nondiabetic) patients suffer more breast cancer recurrences and overall death, but not death due to breast cancer. This may be due to differences in diagnosis, comorbid conditions, and treatment. We note that our study does not exclude the possibility of differences in BCSS, as these may emerge with longer follow-up. We further note that there are reasons to believe that diabetic patients may have an underlying biology that promotes tumor recurrence [12].

With regard to treatment, diabetic patients were less likely to receive chemotherapy than non-diabetic patients (49% versus 57%, P = 0.001), though there was no difference in taxane-based chemotherapy (50% versus 52%, P = 0.664). Diabetic patients that did not receive chemotherapy had worse RFS than nondiabetic patients (P = 0.004). There were no significant differences in RFS between diabetic and nondiabetic patients that did receive chemotherapy (P = 0.2415). Thus, part of the difference in RFS and OS between diabetics and nondiabetics may be related to undertreatment with chemotherapy. However, the numbers of patients and events in these subset analyses were small, so these results are exploratory. The diabetes drug metformin may have antitumor effects that prevent recurrence and lead to improved breast cancer outcomes [33]. If so, metformin might have blunted the differences in survival outcomes between diabetics and nondiabetics. Data on metformin administration were not available in our database.

The strengths of this study include its use of a large cohort of breast cancer patients being treated with modern therapies, namely anthracycline- and taxane-based chemotherapy and tamoxifen- and aromatase inhibitor-based endocrine therapy. It is the first study to examine both obesity and diabetes and their relationships to survival outcomes in a large patient cohort. Finally, this study adjusted for multiple potential confounding variables, including BMI, diabetes, comorbidities, patient- and tumor-related variables, and treatment. The limitations of the study include its heterogeneous population of patients and treatments, such that specific effects of obesity in particular patient subsets or treatment regimens might be obscured. We did not have data regarding chemotherapy dosing in relation to BMI. The study involved a modest number of diabetic patients, due to the low prevalence of diabetes in the general (and breast cancer) population. Relatively few HER2-positive patients were treated with trastuzumab, so this study does not address how trastuzumab therapy might alter differences in breast cancer outcomes by BMI.

In conclusion, in a cohort of patients receiving modern adjuvant therapies, obesity has a negative impact on RFS, BCSS, and OS. In addition, diabetes has a negative impact on OS. This study has implications for clinical practice which need to be confirmed in prospective studies before being implemented into daily oncology practice. It adds to the body of evidence implying that control of both obesity and diabetes may be necessary for improved outcomes in obese and diabetic women with breast cancer.

Regarding therapeutic considerations for these patients, this is an area of active investigation. One promising candidate is the diabetes drug metformin, which may reverse the adverse biology associated with obesity and diabetes and which is being evaluated in both neoadjuvant and adjuvant trials [34, 35].

funding

This work was supported by the Nellie B. Connally Breast Cancer Research Fund; the American Society of Clinical Oncology [ASCO-Breast Cancer Research Foundation Young Investigator Award to SJ], the American Association for Cancer Research [AACR-Amgen Fellowship in Clinical/Translational Cancer Research to SJ], the Barbara Rattay Foundation [Barbara Rattay Advanced Scholar Award to SJ]; and by the NCI at the National Institutes of Health [K07CA109064 to SHG].

disclosure

The authors have declared no conflicts of interest.