Annals of Oncology 15: 875–884, 2004 DOI: 10.1093/annonc/mdh222

Original article

Body mass index as a prognostic feature in operable breast cancer: the International Breast Cancer Study Group experience G. Berclaz1*, S. Li6, K. N. Price6, A. S. Coates7, M. Castiglione-Gertsch3, C.-M. Rudenstam10, S. B. Holmberg11, J. Lindtner12, D. Erien12, J. Collins8, R. Snyder9, B. Thürlimann4, M. F. Fey2, C. Mendiola13, I. Dudley Werner14, E. Simoncini15, D. Crivellari16, R. D. Gelber6 & A. Goldhirsch5,17 On behalf of the International Breast Cancer Study Group 1

Department of Obstetrics and Gynecology; 2Institute of Medical Oncology, Inselspital; 3IBCSG Coordinating Center, Bern; 4Kantonsspital, St Gallen; Oncology Institute of Southern Switzerland, Lugano, Switzerland; 6IBCSG Statistical Center, Dana-Farber Cancer Institute and Frontier Science and Technology Research Foundation, Boston, MA, USA; 7The Cancer Council Australia and University of Sydney, Sydney; 8Department of Surgery, The Royal Melbourne Hospital, Melbourne; 9Department of Oncology, St Vincent’s Hospital, Melbourne, Australia; 10West Swedish Breast Cancer Study Group, Sahlgrenska University Hospital, Göteborg; 11Department of Surgery, SU/Moelndal’s Hospital, Moelndal, Sweden; 12The Institute of Oncology, Ljubljana, Slovenia; 13Madrid Breast Cancer Group, Madrid, Spain; 14Groote Schuur Hospital and University of Cape Town, Cape Town, South Africa; 15Oncologia Medica-Spedali Civili, Brescia; 16Centro di Riferimento Oncologico, Aviano; 17European Institute of Oncology, Milan, Italy 5

Received 8 October 2003; revised 3 February 2004; accepted 4 February 2004

Background: Current information on the prognostic importance of body mass index (BMI) for patients with early breast cancer is based on a variety of equivocal reports. Few have data on BMI in relationship to systemic treatment. Patients and methods: Patients (6792) were randomized to International Breast Cancer Study Group trials from 1978 to 1993, studying chemotherapy and endocrine therapy. BMI was evaluated with eight other factors: menopausal status, nodal status, estrogen receptor status, progesterone receptor status, tumor size, vessel invasion, tumor grade and treatment. BMI was categorized as normal (≤24.9), intermediate (25.0–29.9) or obese (≥30.0). Results: Patients with normal BMI had significantly longer overall survival (OS) and disease-free survival (DFS) than patients with intermediate or obese BMI in pairwise comparisons adjusted for other factors. Subset analyses showed the same effect in pre- and perimenopausal patients and in those receiving chemotherapy alone. When assessed globally and adjusted for other factors, BMI significantly influenced OS (P = 0.03) but not DFS (P = 0.12). Conclusions: BMI is an independent prognostic factor for OS in patients with breast cancer, especially among pre-/perimenopausal patients treated with chemotherapy without endocrine therapy. Key words: body mass index, breast cancer, obesity, prognostic factor, survival

Introduction Previous studies have noted a modest negative association between obesity and survival of breast cancer [1, 2]. However, the methodologies used differed widely [3]. In most of the reports, the effect on prognosis did not persist after adjustment for other factors [4]. Studies reporting the effect of body mass index (BMI) allowing for other prognostic factors such as nodal involvement [5] or menopausal status [6] are also inconsistent, and few studies included patients receiving systemic adjuvant therapy [3]. Given these equivocal results, the importance of BMI as a prognostic factor for breast cancer remained uncertain. To examine the impact of BMI on the prognosis of patients with breast cancer and on outcome with different adjuvant treatments, *Correspondence to: Dr. G. Berclaz, Department of Obstetrics and Gynecology, Inselspital, Effingerstrasse 102, CH-I3010 Bern, Switzerland. Fax: +41-31-632-1652; E-mail: [email protected] © 2004 European Society for Medical Oncology

we analyzed data from 6792 eligible breast cancer patients randomized in trials I–VII of the International (formerly Ludwig) Breast Cancer Study Group (IBCSG).

Patients and methods Patient selection From 1978 to 1993, the IBCSG conducted three generations of trials (I–IV, V and VI–VII) that studied chemotherapy and endocrine therapy. The seven trials accrued 6792 eligible patients who were randomized to receive one of the treatments listed in Table 1. All trials have been reported previously [7–15]. This report includes a cohort of 6370 patients (94%; Table 1) for whom height and weight were recorded in the database. The median follow-up was ∼14 years. BMI [weight (kg)/height (m)2] categories were selected according to the World Health Organization definition: underweight, 40.0 kg/m2 [16]. Since the proportions of patients in the underweight group and in the grade 3 overweight group were small (2% in each), we combined the grade 2 overweight group with the grade 3 overweight group and combined the underweight group with the normal weight group, resulting in three BMI categories: normal (≤24.9), intermediate (25.0–29.9) and obese (≥30.0). We examined the relative risk of relapse and death with regard to the BMI categories adjusting for eight factors known to be predictors of disease-free survival (DFS) and overall survival (OS): menopausal status, nodal status, tumor size, vessel invasion, estrogen receptor (ER) status, progesterone receptor status, tumor grade and treatment regimens (Table 2). In this analysis, we did not adjust for age because there was a significant correlation between age and BMI (r = 0.35, P