BODY SIZE IN DIFFERENT PERIODS OF LIFE AND BREAST CANCER RISK IN POST-MENOPAUSAL WOMEN Cecilia MAGNUSSON 1 *, John BARON 1,2 , Ingemar PERSSON 1 , Alicja WOLK 1 , Reinhold BERGSTRO ¨ M 1,3 , Dimitrios TRICHOPOULOS 4 and Hans-Olov ADAMI 1,4 1 Department of Medical Epidemiology, Karolinska Institutet, Stockholm, Sweden 2 Departments of Medicine and Community and Family Medicine, Dartmouth Medical School, Hanover, NH, USA 3 Department of Statistics, Uppsala University, Uppsala, Sweden 4 Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA Adult obesity has been associated with an increased risk of post-menopausal breast cancer, but it is unclear whether this relationship reflects a causal role of obesity during childhood and adolescence, of weight gain during adult life or of adult obesity per se. In a population-based case-control study in all of Sweden, we included 3,345 (84%of all eligible) women aged 50–74 yearswith invasive breast cancer, and 3,454 (82%of all selected) controls of similar age. Mailed questionnaires and telephone interviews were used to collect detailed informa- tion on anthropometric measures. Odds ratios were esti- mated through multiple logistic regression. W omen with the leanest somatotype at age 7 had about a 3-fold higher risk of breast cancer than the most obese ( P for trend 0.0009). A suggested protective effect of a high body mass at age 18 and a detrimental influence of body mass 1 year prior to data collection largely reflected the effect of weight gain after age 18, a strong predictor of breast cancer risk. Among women at least 20 yearspost menopause, those who had gained 30 kg or more since age 18 had an odds ratio of 2.04 (95% confidence interval 1.20–3.48) of breast cancer compared with those who had maintained their weight unchanged. The effect of weight gain was unequivocal among non-users but not among users of hormone replacement therapy. Our findings have impor- tant implications, suggesting weight preservation as a means for prevention of post-menopausal breast cancer as well as a causal role of childhood body build in breast cancer etiology. Int. J. Cancer 76:29–34, 1998.  1998 Wiley-Liss, Inc. Since body size is one of the few suggested breast cancer risk factors that can reasonably be modified, knowledge of the influence of body size on breast cancer risk may offer prospects for preventive actions. In most, but not all studies, adult obesity has been associated with an increased risk of post-menopausal breast cancer (Hunter and Willett, 1993). Although there are suggestions that obesity in childhood and early adulthood reduces risk of post-menopausal breast cancer (Brinton and Swanson, 1992; Barnes-Josiah et al.1995; Chu et al., 1991), and that weight gain may be an independent risk factor for the disease (Ballard-Barbash et al., 1990b; Brinton and Swanson, 1992; Chu et al., 1991; Le Marchand et al., 1988), such associations remain less certain. We used data from a large population-based study with detailed anthropometric information to disentangle the possible effects of body mass at different ages and weight change on breast cancer risk. MATERIAL AND METHODS Subjects This was a population based case-control study of women aged 50–74 years, without previously diagnosed breast cancer, born in Sweden and resident there between October 1, 1993 and March 31, 1995. We attempted to contact all incident cases of invasive breast cancer in this population. Cases were identified through the 6 Swedish regional cancer registries [to which 98% of all diagnosed cancer cases in Sweden are reported (National Board of Health and Welfare, 1993)] and were asked to participate through their physicians. A total of 3,979 eligible cases were detected of whom 3,345 (84%) participated in the study. Non-participation was due to death or physician refusal (primarily because of poor health of the patients) in 4%, and patient refusal or inability to contact the patient in 11%. The mean interval from diagnosis to data collection was 4.3 months (standard deviation 1.5 months). Control women, frequency matched to the expected age distribu- tion of the cases, were randomly selected from a continuously updated Swedish registry that provides National Registration Number, name, address and place of birth of all people residing in Sweden. Of 4,188 controls selected, 3,454 (82%) agreed to participate in the study. Women with a previous diagnosis of cancer (other than non- melanoma skin cancer or cancer in situ of the cervix), 112 cases and 91 controls, were excluded from all analyses. Furthermore, because of our limited statistical power for pre-menopausal women (198 cases vs. 152 controls) and those with unknown menopausal status (217 cases vs. 100 controls), only post-menopausal women were included in analyses reported here. Thus, the final study group included 2,818 cases and 3,111 controls. Data collection Data were obtained by means of a mailed questionnaire request- ing information on personal and medical history, including current height, weight at age 18 and weight 1 year prior to data collection (‘‘recent weight’’). Subjects were also asked to specify which of 9 pictograms most resembled their own body build (‘‘somatotype’’) at ages 7, 18 and 1 year prior to data collection (‘‘recent somatotype’’) (Fig. 1). These pictograms have been validated against body mass index (BMI) (Must et al., 1993) within a cohort of 100 white females from middle class communities with an average age of 73.1 years. In this study, the Spearman correlation coefficients between recalled somatotype and measured BMI were 0.57, 0.70, 0.75, 0.64 and 0.79 at ages 5, 10, 15, 20 and currently, respectively. Furthermore, we found in a population-based valida- tion study of 111 Swedish women (representing 72% of a random sample) aged 51–66 years that the correlation coefficients between BMI from school records and adult report of somatotype at ages 7 and 18 were 0.6 and 0.7, respectively (A. Kuskowsha-Wolk, personal communication). In the present study, the Spearman correlations between BMI at age 18 and somatotype at ages 7 and 18 were 0.4 and 0.6, respectively. Validity of self-reported height, current weight and BMI in Swedish women was found to be high, (r = 1.0, 0.9 and 0.9, respectively) (Kuskowska-Wolk et al., 1989). The questionnaire also covered reproductive and menstrual history, Contract grant sponsor: National Institutes of Health; Contract grant number: RO1 CA58427; Contract grant sponsor: Swedish Cancer Society. *Correspondence to: Department of Medical Epidemiology, Karolinska Institutet, Box 281, S-171 77 Stockholm, Sweden. Fax: (46)8-314957. E-mail: Cecilia.Magnusson@mep.ki.se Received 26 September 1997; Revised 6 Decemer 1997 Int. J. Cancer: 76, 29–34 (1998)  1998 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer