Abstract P6-06-14: Weight gain and educational level are strongly associated with breast cancer prognosis. Findings from the prospective SU.VI.MAX study

Abstract Purpose: To identify correlations between lifestyle risk factors and breast cancer prognosis. Methods: We screened the database of the SU.VI.MAX trial (13017 subjects randomized between low-dose beta-carotene, C and E vitamins, zinc and selenium, versus placebo, Arch Int Med 2004) to retrieve women with confirmed breast cancer occurring after inclusion. The following data were collected at baseline (before diagnosis) and updated during followup: anthropometric measurements, weight gain, physical activity, tobacco and alcohol consumption, educational level, socio-economic status, and medication use. Data have been collected prospectively before breast cancer diagnosis. All women >50y underwent screening mammograms. Medical reports were screened to collect data on tumor size, histological grade, mitotic count, ER and PR status, nodes and type of adjuvant therapy (chemo- and hormone therapy). Since the study was conducted during the 90s, HER2 status was unavailable for a majority of patients. Analyses were performed using SAS software 9.2. Results: We identified 231 pts with breast cancer included since 1994 in the SU.VI.MAX 8-year Cohort Study. Tumor size had a grossly bimodal distribution and was significantly associated with PR status (p = 0.02), grade (p = 0.0002), mitotic count (p = 0.0004) and nodal status (p<0.0001). In univariate analysis, the following variables were associated with tumor size (25 versus 8.6 mm): educational level (>12 versus <12y of schooling, p = 0.05), and weight gain >10% between baseline and the end of followup. In multivariate analysis, these associations remained significant with odds ratios of 0.38 (95%CI: 0.16-0.9) for educational level and 1.94 (95%CI: 0.93-4.05) for weight gain. Level of physical activity (walking or equivalent <1h/day versus >1h/day) was not significantly associated with tumor size. Weight gain >10% after breast cancer (data available for 176 patients) was also significantly associated with tumor size. Discussion: We confirm that weight gain is significantly associated with larger tumor size and adverse prognostic factors. Whereas education level is widely regarded as a risk factor for postmenopausal breast cancer, a favorable association between education level and breast cancer prognosis has already been described (Hussain SK et al. Int J Cancer 2008). Since all women >50 included in the study underwent mammograms, this finding might be explained by lifestyle-related factors rather than by a better compliance to screening. Although the harms of antioxidant supplementation are still a matter of debate, we did not find any association between antioxidant supplementation (at nutritional doses) and breast cancer prognostic factors. Finally, a surprising finding is that post-cancer weight gain >10% of baseline is associated with larger tumors. We already presented preliminary findings suggesting that none of the commonly admitted factors were involved into severe weight gain after breast cancer (SABCS 2011, P4-12-06). Severe weight gain raises risk for recurrence among breast cancer survivors (Caan et al. AACR 2011), but we hypothesize that this relation is not causal. Rather, it could be due to yet undetermined factors linked to adverse prognosis. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P6-06-14.