Background
Lifestyle factors such as nutrition and physical activity represent actionable targets to improve survival in breast cancer survivors. While high BMI at diagnosis is consistently associated with poorer prognosis, BMI does not distinguish between adipose and lean mass or capture fat distribution (visceral vs. subcutaneous), although these body composition components may differentially influence survival. Evidence on detailed body composition phenotypes and breast cancer survival remains limited, but recent advances in artificial intelligence tools now enable large?scale, automated extraction of detailed body composition features from routine CT scans, offering new opportunities for research at scale.
Besides, metabolic alterations linked to excess adiposity or low muscle mass could potentially explain their association with survival. Treatment?related metabolic changes may also differ according to baseline body composition. Physical activity during treatment may mitigate such metabolic deterioration, but whether benefits vary by body composition phenotype is unknown.
Objectives
This project aims to characterize CT-scan derived body composition phenotypes and related metabolic alterations in localized breast cancer patients, and to evaluate their associations with survival and response to physical activity during treatment.
Methods
This project leverages both observational and interventional study designs. Associations between body composition, circulating biomarkers, and survival are investigated using a retrospective cohort of 2,732 women diagnosed with non-metastatic breast cancer between 2010 and 2020, who had surgery at the Léon Bérard comprehensive cancer center (Lyon, France) (n=343 deaths, average follow-up=7.1 years). For participants with available CT-scan within 3 months of diagnosis (n=2,400), body composition was analyzed using a validated, automated algorithm. For participants with available blood samples collected before any systemic treatment (n=557), inflammatory, metabolic and muscle mass regulation biomarkers were quantified (n=45 inflammatory proteins; CRP, IGF-1, C-peptide, leptin, adiponectin, triglycerides, total, HDL- and LDL-cholesterol, myostatin, activin, follistatin). Cross-sectional associations between body composition components and metabolic alterations will be evaluated linear regression models. Associations of body composition parameters and biomarkers concentrations with survival outcomes (disease-free, overall and breast cancer-specific) will be evaluated using Cox proportional hazards models adjusted for patient and tumor characteristics, treatments, and year of diagnosis.
Further, the impact of physical activity on body composition and circulating biomarkers changes during treatment will be examined in the DISCO trial (NCT03529383, France), a completed randomized trial evaluating the efficacy of two 6-month physical activity interventions concomitant to (neo)adjuvant treatments, using data from participants with available repeated blood samples (n=244).
Results
Analyses are underway, and preliminary results focusing on survival outcomes will be presented at the conference.
Conclusions/Implications
This project will deepen understanding of how body composition and related metabolic alterations influence breast cancer survival, and how physical activity during treatment could help mitigate those associations, by integrating observational and interventional approaches. Large?scale automated CT assessment, which can be embedded in routine care, may help identify patients at higher risk and inform targeted tertiary prevention strategies.