Abstract 2627: A mathematical model to predict prognosis in breast cancer survivors following an exercise intervention

Abstract

Abstract BACKGROUND Metabolic syndrome (MetS) and obesity are associated with increased risk of cardiovascular disease, type 2 diabetes, and cancer recurrence, and are higher in breast cancer survivors than age-matched postmenopausal women. We previously reported that exercise improves MetS and reduces pro-inflammatory biomarkers in obese breast cancer survivors. Whether these exercise-induced changes impact cancer outcomes is currently unknown. In this pilot study, we apply a robust mathematical analysis to identify biomarkers of MetS and obesity associated with clinical outcomes in obese breast cancer survivors following participation in a 16-week exercise intervention. EXPERIMENTAL DESIGN Eleven obese postmenopausal breast cancer survivors were randomized to either the exercise or control group. The exercise group participated in 16 weeks of supervised aerobic and resistance exercise sessions 3 times/week. Fasting blood and adipose tissue samples were analyzed for cytokine secretion, macrophage phenotype, and MetS. Prognostic outcomes included disease-free survival (DFS), overall survival (OS), distant DFS (DDFS), and recurrence-free interval (RFI). Partial least squares regression (PLSR) was used to quantify the importance of specific patient measurements in predicting the clinical response. PLSR is a multivariate regression analysis that quantifies the relationships between the participant characteristics and tissue and plasma measurements (inputs) and the clinical response (outputs) and can be used to identify which patient measurements most significantly associate with specific clinical outcomes. RESULTS MetS, macrophage phenotype, and inflammatory biomarkers were significantly improved in the exercise group compared to the control group (p<0.01) and these biomarkers were used to build the PLSR model. An accurate and predictive PLSR model could be constructed for two clinical responses: distant DFS and RFI. Overall, the percentage of type 1 macrophages (M1) is predicted to significantly contribute to predicting the clinical responses of distant DFS and RFI, where patients with a lower percentage of M1 have a better clinical response. Other patient measurements were identified as significantly contributing to distant DFS (prolactin, serum amyloid A, type 2 macrophages, brain-derived neurotrophic factor, interleukin-6, and insulin-like growth factor binding protein-1) which impact the clinical response to different extents. CONCLUSIONS Exercise improves biomarkers related to MetS, obesity, and inflammation. The PLSR model quantifies patient characteristics and measurements associated with clinical outcomes. Our analysis provides quantitative insight into potential biomarkers to predict response to exercise and prompts the need for an in depth longitudinal study to determine the effects of exercise on long-term survival in obese breast cancer patients. Citation Format: Christina M. Dieli-Conwright, Jean-Hughes Parmentier, Steven D. Mittelman, Nathalie Sami, Kyuwan Lee, Stacey Finley. A mathematical model to predict prognosis in breast cancer survivors following an exercise intervention [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2627.