Fax: (718) 904-2892
Article first published online: 27 AUG 2012
Copyright © 2012 American Cancer Society
Volume 118, Issue 23, pages 5937–5946, 1 December 2012
How to Cite
Sparano, J. A., Wang, M., Zhao, F., Stearns, V., Martino, S., Ligibel, J. A., Perez, E. A., Saphner, T., Wolff, A. C., Sledge, G. W., Wood, W. C., Fetting, J. and Davidson, N. E. (2012), Obesity at diagnosis is associated with inferior outcomes in hormone receptor-positive operable breast cancer. Cancer, 118: 5937–5946. doi: 10.1002/cncr.27527
The first author, as the principal investigator, participated in all phases of this analysis, interpretation, and article preparation. All coauthors participated in data interpretation. The study biostatisticians (second and third authors) conducted all analyses. Coauthors reviewed the article contents and approved the submission version.
Presented in part at the San Antonio Breast Cancer Symposium; December 10, 2010; San Antonio, TX.
- Issue published online: 19 NOV 2012
- Article first published online: 27 AUG 2012
- Manuscript Accepted: 24 JAN 2012
- Manuscript Revised: 15 JAN 2012
- Manuscript Received: 4 DEC 2011
- breast cancer;
- hormone receptor positive;
Obesity has been associated with inferior outcomes in operable breast cancer, but the relation between body mass index (BMI) and outcomes by breast cancer subtype has not been previously evaluated.
The authors evaluated the relation between BMI and outcomes in 3 adjuvant trials coordinated by the Eastern Cooperative Oncology Group that included chemotherapy regimens with doxorubicin and cyclophosphamide, including E1199, E5188, and E3189. Results are expressed as hazard ratios (HRs) from Cox proportional hazards models (HR >1 indicates a worse outcome). All P values are 2-sided.
When evaluated as a continuous variable in trial E1199, increasing BMI within the obese (BMI, ≥30 kg/m2) and overweight (BMI, 25-29.9 kg/m2) ranges was associated with inferior outcomes in hormone receptor-positive, human epidermal growth receptor 2 (HER-2)/neu-negative disease for disease-free survival (DFS; P = .0006) and overall survival (OS; P = .0007), but not in HER-2/neu–overexpressing or triple-negative disease. When evaluated as a categorical variable, obesity was associated with inferior DFS (HR, 1.24; 95% confidence interval [CI], 1.06-1.46; P = .0008) and OS (HR, 1.37; 95% CI, 1.13-1.67; P = .002) in hormone receptor-positive disease, but not other subtypes. In a model including obesity, disease subtype, and their interaction, the interaction term was significant for OS (P = .02) and showed a strong trend for DFS (P = .07). Similar results were found in 2 other trials (E5188, E3189).
In a clinical trial population that excluded patients with significant comorbidities, obesity was associated with inferior outcomes specifically in patients with hormone receptor-positive operable breast cancer treated with standard chemohormonal therapy. Cancer 2012. © 2012 American Cancer Society.
Obesity is a major public health problem in North America, Europe, and other developed regions of the world.1 A body mass index (BMI) of 30 kg/m2 or higher has been used to define obesity in most reports, including those showing an association between obesity and increased risk of cancers, cancer mortality, and mortality from all causes.2, 3 Obesity rates have increased 2-fold in adults and 3-fold in children in the past 30 years in the United States, and are also rising in other developed regions worldwide.1 Obesity has been associated with an increased risk of breast cancer, which is the most common cancer in women and the second leading cause of cancer death in the United States and other developed nations worldwide.4, 5 Obesity at breast cancer diagnosis has also been associated with inferior survival, including breast cancer-specific survival.6 Other analyses have indicated that for a clinical trial population of patients with lymph node-negative, estrogen-receptor positive breast cancer treated with tamoxifen who have a relatively low risk of cancer recurrence, obesity is associated with an increased rate of contralateral breast cancer, second primary cancers, and other noncancer-related deaths.7, 8
Gene expression profiling studies have shown that breast cancer is a heterogeneous disease characterized by distinct subtypes that have differing clinical behaviors and response to therapy.9-11 Management decisions are based upon predictive factors for response to targeted therapies, such as estrogen receptor (ER) and/or progesterone receptor (PR) protein expression for selecting hormonal therapy,12 and human epidermal growth receptor 2 (HER-2)/neu protein overexpression and/or gene amplification for selecting anti–HER-2-directed therapy.13 Patterns of ER, PR, and HER-2 expression have been shown to correspond to genotypically defined subsets.14 Understanding the relation between host-related factors, such as obesity, and tumor-related factors, such as breast cancer subtype, may have important implications for identifying factors contributing to recurrence, and thus may impact the management of localized and advanced disease.
To disentangle the influence of obesity from other factors influencing recurrence and survival, we compared the outcomes of obese patients with others in a large cohort of women with stage I to III breast cancer who participated in 3 National Cancer Institute (NCI)-sponsored trials evaluating the role of chemotherapy or endocrine therapy regimens.15-17 These trials were completed before trastuzumab was approved as adjuvant therapy for HER-2–positive disease.18, 19 All of the trials required normal cardiac, renal, hepatic, and bone marrow function and good performance status, thereby excluding patients with significant comorbidities, potentially minimizing this as a confounding factor. In addition, patients received standardized care as stipulated by the trial. We found that obesity was independently associated with inferior outcomes specifically in patients with hormone receptor-positive, HER-2/neu–negative disease, but not other breast cancer subtypes.
MATERIALS AND METHODS
Patient Selection and Treatment
The analysis included patients enrolled on 3 NCI-sponsored clinical trials coordinated by the Eastern Cooperative Oncology Group (ECOG) for whom BMI data were available. Briefly, patients with operable adenocarcinoma of the female breast with axillary lymph node metastases (T1-3; N1 or 2) or high-risk lymph node-negative disease (T2-3, N0 for E1199 only) without distant metastases were eligible. Other details regarding eligibility, treatment administered, and results have been previously reported and are summarized in Table 1.15-17 Chemotherapy was prescribed by actual body weight and was not adjusted for obese patients in all trials. The primary analysis was initially performed for E1199, which was the most recent trial and included the most contemporary chemotherapy regimens (doxorubicin, cyclophosphamide, and paclitaxel or docetaxel) and endocrine therapy (tamoxifen or aromatase inhibitors). After a relation between obesity and breast cancer was defined in the E1199 study population, similar analyses were performed in the 2 older studies (E5188, E3189) that provided confirmatory findings.
|Characteristic||Trial E1199||Trial E5188||Trial E3189|
|Analysis||Primary analysis||Confirmatory analysis||Confirmatory analysis|
|No. with BMI data included in analysisa||4770||1502||613|
|Time period study performed||October 1999-January 2002||July 1989-February 1994||September 1989-April 1993|
|Population||Node positive and high risk node negative||ER positive, node positive, premenopausal||ER negative, node positive|
|Median follow-up for living patients||7.9 years||14.0 years||14.4 years|
|Chemotherapy||AC taxane||CAF||CAF vs 16-week regimen|
|Endocrine therapy||Tamoxifen or tamoxifen followed by an AI||No therapy vs goserelin vs goserelin plus tamoxifen||None|
|Age at diagnosis, median y (range)||51 (22-84)||43 (24-57)||47 (25-78)|
|Tumor size, cm|
|Axillary nodal status|
The primary trial endpoints were disease-free survival (DFS) and overall survival (OS). DFS was defined to be time from randomization to a first event, including disease recurrence, contralateral breast cancer, or death from any cause. OS was defined as time from randomization to death from any cause. To determine whether inferior outcomes were attributable to breast cancer recurrence rather than comorbidities, we also evaluated breast cancer-specific survival (BCSS), which was defined as death attributed to breast cancer or preceded by a breast cancer recurrence, with follow-up censored at the time of death from other causes. Event-time distributions for DFS and OS were estimated using Kaplan-Meier analysis. Cox proportional hazards methods were used to estimate hazard ratios (HRs) and test for significance for event times. The relation between BMI as a continuous variable and DFS/OS was evaluated using a flexible family of curves (a natural spline with 3 degrees of freedom) to model the effect of BMI in a proportional hazards model.20 Patient characteristics were compared using Fisher exact test if they were categorical and Wilcoxon 2-sample test if they were continuous. The adverse events were compared using Fisher exact test. All P values are 2-sided; confidence intervals (CIs) are at the 95% level. The analysis was based on a dataset downloaded in April 2011 for trial E1199, in which patients are still being followed for recurrence and survival; the median follow-up for surviving patients was 95 months (7.9 years; range, 0-119 months), at which time there were 1234 DFS events and 891 deaths (including 695 BCSS events). The BCSS events included 568 patients who were coded by the treating institutions as dying from breast cancer (64% of all deaths and 82% of all BCSS events) and 127 patients who had a breast recurrence before death whose death was coded by the treating site as from an unknown cause (14% of all deaths and 18% of all BCSS events). Other deaths included 118 patients who were coded as dying from other causes (13% of all deaths), and 78 patients who were coded as dying from an unknown cause and who did not have breast cancer recurrence before death (9% of all deaths). The median time from recurrence to death for those coded as dying from breast cancer was 15.2 months; for those coded as dying from an unknown cause who had a breast cancer recurrence before death it was 12.4 months.
Data Management and Regulatory Issues
The studies were sponsored by the NCI, reviewed and approved by the Cancer Therapy Evaluation Program at NCI, and developed and coordinated by the ECOG (ClinicalTrials.gov identifier, NCT00004125). The protocol was reviewed and approved by the institutional review board at each participating institution, and all patients provided written informed consent.
The characteristics of obese and nonobese patients for trial E1199 are shown in Table 2. Of the 4770 patients with BMI data, 1745 (36.6%) were obese, 1540 (32.3%) were overweight, 1447 (30.3%) had a normal BMI, and 38 (0.8%) were underweight at the time they were enrolled on the trial after surgery and before initiation of chemotherapy. Obese and overweight patients were older and more likely to be postmenopausal and black. Obese patients and overweight patients had somewhat larger primary tumors and were more likely to have breast-conserving surgery, and exhibited somewhat different distribution of nodal metastases, although the median number of nodes involved was similar. There were no significant differences in the biologic characteristics of the tumor, as reflected by ER, PR, or HER-2 expression, nor in the type of endocrine therapy or chemotherapy given.
|Characteristic||Normal or Underweight, BMI <25 kg/m2a||Overweight, BMI 25-29.9 kg/m2||Obese, BMI ≥30 kg/m2||P, Obese vs Other||P, Obese vs Overweight vs Normal|
|No.||1485 (31%)||1540 (32%)||1745 (37%)|
|Median y [range]||49 [22-79]||51 [24-81]||52 [25-84]|
|≤45 years||568 (38.3%)||435 (28.3%)||434 (24.9%)|
|46-65 years||801 (53.9%)||931 (60.5%)||1130 (64.8%)|
|≥66 years||116 (7.8%)||174 (11.3%)||181 (10.4%)|
|Premenopausal, No.b||811 (54.6%)||684 (44.4%)||724 (41.5%)||<.0001||.0005|
|White||1314 (88.5%)||1310 (85.1%)||1407 (80.6%)|
|Hispanic||41 (2.8%)||67 (4.4%)||69 (4.0%)|
|Black||57 (3.8%)||109 (7.1%)||234 (13.4%)|
|Other||65 (4.4%)||49 (3.2%)||31 (1.8%)|
|Unknown||8 (0.5%)||5 (0.3%)||4 (0.2%)|
|Median cm [range]||2.5 [0.1-17]||2.5 [0.1-14]||2.5 [0.1-15]|
|≤2 cm||593 (39.9%)||532 (34.6%)||601 (34.4%)|
|>2 and ≤5 cm||768 (51.7%)||838 (54.4%)||926 (53.1%)|
|>5 cm||111 (7.5%)||156 (10.1%)||201 (11.5%)|
|Unknown||13 (0.9%)||14 (0.9%)||17 (1.0%)|
|Median [range]||2 [0-40]||2 [0-30]||2 [0-41]|
|0||151 (10.2%)||178 (11.6%)||224 (12.8%)|
|1-3||869 (58.5%)||859 (55.8%)||917 (52.6%)|
|4-9||332 (22.4%)||352 (22.9%)||400 (22.9%)|
|≥10||127 (8.6%)||143 (9.3%)||193 (11.1%)|
|Unknown||6 (0.4%)||8 (0.5%)||11 (0.6%)|
|Hormone receptor positivec||.40||.30|
|Positive||1084 (73.0%)||1091 (70.8%)||1236 (70.8%)|
|Negative||381 (25.7%)||426 (27.7%)||486 (27.9%)|
|Unknown||20 (1.4%)||23 (1.5%)||23 (1.3%)|
|Positive||302 (20.3%)||295 (19.2%)||343 (19.7%)|
|Negative||1043 (70.2%)||1086 (70.5%)||1215 (69.6%)|
|Unknown||140 (9.4%)||159 (10.3%)||187 (10.7%)|
|Triple-negative diseased||262 (17.6%)||289 (18.8%)||327 (18.7%)||.61||.56|
|Most extensive surgery||.015||.011|
|Breast-sparing surgery||541 (36.4%)||610 (39.6%)||728 (41.7%)|
|Mastectomy||935 (63.0%)||922 (59.9%)||1010 (57.9%)|
|Unknown||9 (0.6%)||8 (0.5%)||7 (0.4%)|
|Given||814 (54.8%)||885 (57.5%)||995 (57.0%)|
|Not given||671 (45.2%)||655 (42.5%)||750 (43.0%)|
|Paclitaxel every 3 weeks||369 (24.9%)||396 (25.7%)||449 (25.7%)|
|Paclitaxel weekly||371 (25.0%)||394 (25.6%)||417 (23.9%)|
|Docetaxel every 3 weeks||389 (26.2%)||363 (23.6%)||438 (25.1%)|
|Docetaxel weekly||356 (24.0%)||387 (25.1%)||441 (25.3%)|
|Endocrine therapy givene||.49||.50|
|No. with hormone receptor-positive disease||1084||1091||1236|
|Tamoxifen alone||412 (38.0%)||388 (35.6%)||446 (36.1%)|
|Tamoxifen, then AI||598 (55.2%)||631 (57.8%)||693 (56.1%)|
|AI alone||43 (4.0%)||33 (3.0%)||54 (4.4%)|
|None||22 (2.0%)||23 (2.1%)||31 (2.5%)|
|Unknown||9 (0.8%)||16 (1.5%)||12 (1.0%)|
Delivery of Adjuvant Therapy and Adverse Events
The administration of adjuvant chemotherapy and grade 3 to 4 adverse events in obese compared with nonobese patients for the E1199 trial are shown in Table 3. There were generally no significant differences in the proportion who received at least 90% of cycle 1 dosing for cyclophosphamide, doxorubicin, paclitaxel, or docetaxel, indicating that clinicians adhered to the protocol and prescribed chemotherapy dosing by actual body weight. When evaluating all cycles, drug administration was similar for doxorubicin, cyclophosphamide, and docetaxel, although obese patients were less likely to receive at least 90% relative dose intensity for every 3-week paclitaxel (80% vs 87%; P = .007) and weekly paclitaxel administration (78% vs 85%; P = .008). During doxorubicin and cyclophosphamide chemotherapy, obese patients exhibited less grade 4 neutropenia (32% vs 40%; P < .0001) and more grade 3 to 4 cardiac toxicity (0.2% vs 0%; P = .05), but had comparable rates of grade 3-4 febrile neutropenia, infection, and overall cardiac toxicity. During taxane therapy, obese patients exhibited less grade 4 neutropenia (12% vs 15%; P = .027) and febrile neutropenia (3% vs 5%; P = .022), but more grade 2 to 4 neuropathy (21% vs 19%; P = .042) and grade 3 to 4 neuropathy (7% vs 5%; P = .0041).
|Drug Administration/AEs||BMI <30 kg/m2||BMI ≥30 kg/m2||Pa|
|Drug administration, at least 90% of intended dose given|
|Paclitaxel every 3 weeks|
|Docetaxel every 3 weeks|
|AC chemotherapy AEs|
|Neutropenia, grade 4 only||40%||32%||<0.0001|
|Taxane therapy AEs|
|Neutropenia, grade 4 only||15%||12%||.027|
Relation Between Obesity and Clinical Outcomes
Univariate analyses were performed evaluating the relation between obesity and clinical outcomes in the E1199 trial. When the entire study population was included, compared with nonobese women, obesity was associated with significantly inferior DFS (HR, 1.17; 95% CI, 1.04-1.31; P = .0077) and OS (HR, 1.23; 95% CI, 1.08-1.40; P = .0025). For patients with hormone receptor-positive/HER-2–negative/unknown disease, obesity was likewise associated with inferior outcomes, including DFS (HR, 1.31; 95% CI, 1.12-1.53; P = .0009) and OS (HR, 1.46; 95% CI, 1.21-1.77; P = .0001; Fig. 1A, B), an effect that was more pronounced than in the overall population. This relation was not observed, however, in patients with triple-negative disease (Fig. 1C, D) or HER-2–positive disease (Fig. 1E, F). Only 0.8% (n = 38) of patients were underweight (BMI, <18.5 kg/m2), and the results were not altered if these patients were excluded (DFS: HR, 1.31; 95% CI, 1.11-1.53; P = .001; OS: HR, 1.45; 95% CI, 1.20-1.76; P = .0001 for hormone receptor-positive, HER-2–negative disease in obese compared with nonobese women). The results were also similar if outcomes for obese women (BMI, ≥30 kg/m2) were compared with those for women with a normal BMI (18.5-24.9 kg/m2), including DFS (HR, 1.40; 95% CI, 1.15-1.70; P = .0007) and OS (HR, 1.69; 95% CI, 1.32-2.15; P < .0001) for women with hormone receptor-positive, HER-2–negative disease, but not other subtypes (data not shown).
Analysis for Interaction Between Obesity and Breast Cancer Subtype
To further evaluate the interaction between obesity and breast cancer subtype (hormone receptor-positive/HER-2–negative or unknown vs other) in the E1199 trial, we constructed models including obesity, the breast cancer subtype, and their interaction. For OS, the interaction term of P = .021 was statistically significant (HR for the obese over nonobese for hormone receptor-positive/HER-2–negative or unknown was 1.45, P = .00014 vs HR 1.05 for other subtypes, P = .62). For DFS, the interaction term of P = .070 indicated a strong trend (HR for hormone receptor-positive/HER-2–negative or unknown was 1.30, P = .0012 vs HR 1.05, P = .60). For BCSS, the interaction term of P = .0087 was statistically significant (HR for hormone receptor-positive/HER-2–negative or unknown was 1.47, P = .0008 vs HR 0.96, P = .74).
Relation Between BMI as a Continuous Variable and Clinical Outcomes
The relation between BMI as a continuous variable after adjustment for other covariates was also performed for the E1199 trial. There was a significant relation between increasing BMI and inferior DFS (overall P = .00063, nonlinearity P = .83) and OS (P = .00067, nonlinearity P = .31) in patients with hormone receptor-positive/HER-2–negative or unknown disease (Fig. 2). A similar relation was not observed for patients with triple-negative disease (P = .89 for DFS [nonlinearity, P = .13] and P = .45 for OS [nonlinearity, P = .22]), nor in HER-2–positive disease (P = .51 for DFS [nonlinearity, P = .34] and P = .61 for OS [nonlinearity, P = .55]).
Evaluation of Obesity in Multivariate Models and Validation in Other Trials
After identifying a relation between obesity and inferior outcomes in trial E1199 specifically in patients with hormone receptor-positive, HER-2–negative/unknown disease, we next constructed multivariate models (Table 4) adjusted for other covariates (including age, race, menopausal status, tumor size, number of positive axillary lymph nodes, and type of surgery), and also performed similar analyses in 2 other clinical trial populations that included only patients with hormone receptor-positive disease (E5188) or hormone receptor-negative disease (E3189). When adjusted for other covariates, obesity remained strongly associated in the E1199 trial population with inferior DFS (HR, 1.24; 95% CI, 1.06-1.46; P = .0079) and OS (HR, 1.37; 95% CI, 1.13-1.67; P = .0015) only for patients with hormone receptor-positive, HER-2–negative/unknown disease in trial E1199. For the E5188 trial, which included only premenopausal women with ER-positive, axillary lymph node-positive breast cancer, obesity was likewise associated with inferior DFS (HR, 1.41; 95% CI, 1.19-1.67; P < .0001) and OS (HR, 1.51; 95% CI, 1.24-1.83; P < .0001). For the E3189 trial, which included only patients with ER/PR-negative disease, there was no relation between obesity and outcomes. Although trials E5188 and E3189 were performed before routine HER-2/neu testing in clinical practice and the information was not available for this analysis, the findings are nonetheless consistent with the relation observed in the E1199 trial.
|Trial||Population||HR for Disease-Free Survival (95% CI)||HR for Overall Survival (95% CI)||HR for BCSS (95% CI)|
|E1199||Hormone receptor-positive, HER-2–negative/unknown disease||1.24 (1.06-1.46), P = .0079||1.37 (1.13-1.67), P = .0015||1.40 (1.11-1.76), P = .0042|
|Triple-negative disease||1.02 (0.80-1.30), P = .87||1.11 (0.85-1.46), P = .45||1.00 (0.74-1.36), P = .996|
|HER-2–positive disease||1.06 (0.82-1.38), P = .65||0.99 (0.73-1.34), P = .93||1.00 (0.71-1.40), P = .98|
|E5188||ER-positive disease||1.41 (1.19-1.67), P < .0001||1.51 (1.24-1.83), P < .0001||1.54 (1.26-1.88), P < .0001|
|E3189||ER/PR-negative disease||0.90 (0.70-1.16), P = .41||0.83 (0.63-1.09), P = .18||0.85 (0.63-1.15), P = .29|
To provide additional confirmation that breast cancer recurrence was contributing to the inferior DFS and OS observed only in specific breast cancer subtypes, we also evaluated BCCS. For trial E1199, BCSS was likewise inferior for those with hormone receptor-positive, HER-2–negative disease (HR, 1.40; 95% CI, 1.11-1.76; P = .0042), but not for other subtypes (Table 4). Inferior BCSS was also observed in trial E5188, which included only patients with ER-positive disease (HR, 1.54; 95% CI, 1.26-1.88; P < .0001), but not in trial E3189, which included only patients with ER-negative disease (HR, 0.85; 95% CI, 0.63-1.15; P = .29). In addition to the results using cause-specific hazard analysis based on the Cox proportional hazard methods described above, we also conducted the subdistribution hazards analysis for competing risks,21 which produced consistent results (data not shown).
We evaluated the relation between BMI and clinical outcomes in a cohort of 6885 women with stage I to III breast cancer enrolled in 3 clinical trials that included adjuvant doxorubicin-containing chemotherapy. Similar to previous individual reports and a meta-analysis of these other reports,6 we found that obese patients, defined as having a BMI of 30 kg/m2 or higher, exhibited a significantly higher risk of recurrence and death. This is likewise consistent with a population-based study that demonstrated inferior outcomes for obese patients.22 However, analysis of the E1199 dataset is the first report to demonstrate that worse outcomes are observed specifically in women with hormone receptor-positive/HER-2–negative disease who were treated with contemporary chemoendocrine therapy (including anthracyclines, taxanes, and aromatase inhibitors) but not other breast cancer subtypes. Furthermore, there was a continuous relation between increasing BMI and inferior outcomes in this specific subtype, indicating that excessive weight may also be a surrogate for other factors contributing to recurrence not only in obese but also in overweight patients.
Although numerous previous studies have shown an association between obesity and inferior breast cancer outcomes, this report addresses critical limitations of previous analyses, including uniform adjuvant therapy administration, adjustment for other potentially confounding covariates, evaluation of outcomes in specific breast cancer subtypes, and sufficient maturity of follow-up data to detect late relapses typically associated with hormone receptor-positive disease.23 Few previous studies have evaluated the effect of obesity stratified by hormone receptor expression, and none also included information regarding HER-2/neu overexpression. Majed et al reported that in a cohort of 14,709 patients treated at the Curie Institute over a 19-year period, there was a statistically significant interaction between obesity and distant recurrence for ER-positive compared with ER-negative disease (P = .05), although only about 30% of the population received adjuvant chemotherapy, and not all patients with ER-positive disease received endocrine therapy.24 In contrast, de Azambuja et al reported worse outcomes for ER-positive compared with ER-negative disease, but the interaction P values were not significant.25 Sestak et al reported that in postmenopausal women with ER-positive disease enrolled in the ATAC (Arimidex, Tamoxifen Alone or in Combination) trial, not only was high baseline BMI associated with more distant recurrence, but also better outcomes were observed for adjuvant anastrozole compared with tamoxifen primarily in women who were not obese.26
Obesity is known to be associated with more comorbidities.27 However, the studies included in this analysis required an excellent performance status and normal cardiac, renal, hepatic, and bone marrow function, thereby excluding those with cardiovascular disease or other comorbidities compromising normal organ function or functional status. Moreover, by evaluating BCSS, we confirmed that the inferior DFS and OS was attributable to higher risk of breast cancer recurrence and not solely to comorbidities.
An important consideration in interpreting this work is that the relation between increasing BMI and increased recurrence is biologically plausible for several reasons. First, ER-positive breast cancer exhibits significantly higher gene expression of the insulin growth factor pathway (eg, IRS1, IGFR1, IGFB2) and downstream elements, such as Ras pathway (RhoB, RhoC, RAB27B, GGPS1) and mitogen-activated protein (MAP) kinase (MAPK3) pathway,28 and there is significant cross-talk between the estrogen and insulin signaling pathways.29 Second, higher insulin levels, even within a normal physiologic range, have been associated with increased recurrence in operable breast cancer30, 31; in addition, elevated C-peptide levels are associated with increased risk of cancer-related death specifically in ER-positive breast cancer.31 Binding of insulin to the insulin receptor activates the MAP kinase and PI3 kinase pathways, and also actives ER-alpha–mediated transcription. Finally, obesity is known to be associated with hyperinsulinemia, and metabolic syndrome (characterized by central obesity and hyperinsulinemia) is known to be associated with increased breast cancer recurrence.32 Although increased production of endogenous estrogens in obese women has been postulated as a contributing factor, this has not been confirmed.33, 34
The results of this analysis clearly establish a relation between higher BMI at the time of breast cancer diagnosis and higher risk of recurrence and death, specifically in hormone receptor-positive, HER-2–negative disease, which accounts for about ⅔ of all breast cancers. What remains uncertain, however, is whether dietary and lifestyle interventions resulting in weight loss after a breast cancer diagnosis could substantially reduce the risk of recurrence, and also perhaps provide secondary benefits in reducing cardiovascular morbidity and mortality. The effect of dietary fat reduction in early stage breast cancer has already been evaluated in 2 clinical trials, including the WINS (Women's Intervention Nutrition Study)35 and WHEL (Women's Healthy Eating and Living) trials.36 Dietary fat reduction associated with modest average weight loss of 6 pounds was associated with reduced recurrence in the WINS trial (HR, 0.76; P = .034 for adjusted Cox model analysis), but was not observed in the WHEL trial, in which the dietary intervention did not produce weight loss. The 25% reduction in the risk of recurrence observed in the WINS trial is comparable to the treatment effect associated with adjuvant chemotherapy.37 Other studies have shown that significant weight loss may be achieved with other dietary strategies, and was also associated with comparable reduction in insulin levels,38 suggesting that patients may have several dietary options that may be effective in reducing recurrence risk. Further evaluation of dietary and lifestyle modification is warranted.
Supported in part by grants from the Department of Health and Human Services and the National Institutes of Health (NIH): CA14958 to the Albert Einstein College of Medicine, CA23318 to the ECOG statistical center, CA66636 to the ECOG data management center, CA21115 to the ECOG coordinating center and chairman's office, CA32012 to the Southwest Oncology Group, CA11789 to the Cancer and Leukemia Group B, CA25224 to the North Central Cancer Treatment Group, CA49883 to the Indiana University School of Medicine, and CA16116 to Johns Hopkins Oncology Center. The contents of the article are solely the responsibility of the authors and do not necessary represent the official view of the National Center for Research Resources, NIH, or American Society of Clinical Oncology.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 21A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999; 94: 496-509., .
- 25The effect of body mass index on overall and disease-free survival in node-positive breast cancer patients treated with docetaxel and doxorubicin-containing adjuvant chemotherapy: the experience of the BIG 02-98 trial. Breast Cancer Res Treat. 2010; 119: 145-153., , , et al.
- 28Gene expression profiling of phenotypically-defined hormone-receptor positive breast cancer: evidence for increased transcriptional activity of the insulin growth factor receptor pathway and other pathways [abstract]. Breast Cancer Res Treat. 2009. Abstract 5165., , , et al.
- 33Gruppo Italiano Mammella (GIM). Effects of body mass index (BMI) on plasma levels of estrone sulfate (ES) in postmenopausal women with breast cancer (BC) during letrozole (L) treatment [abstract]. ASCO Meeting Abstracts. 2011; Vol. 29, No 15_suppl (May 20 Supplement): Abstract 515., , , et al and
- 34Do obese women benefit less from adjuvant endocrine therapy with aromatase inhibitors? Preliminary analysis on the association of BMI and aromatase inhibitor plasma levels [abstract]. ASCO Meeting Abstracts. 2011; Vol 29, No 15_suppl (May 20 Supplement), 2011: e11118., , , et al.