Although higher body mass index (BMI) convincingly increases the incidence of many cancers, evidence is contradictory whether higher pre-diagnosis BMI is associated with poorer or improved survival among cancer patients, often referred to as “obesity paradox”. This uncertainty hampers efforts of making clear recommendations on weight management in cancer survivors. One plausible explanation for this discrepancy in results of the relationship with survival is the limitation of BMI to distinguish between adipose and muscle tissue and to capture body fat distribution. Yet, body composition and body fat distribution have been shown to substantially impact on cancer survival. Higher pre-diagnosis BMI has also been associated with increased risks of second primary cancers (SPC), both in men and women. However, the same limitations of BMI might limit our understanding of the genuine relationship between obesity phenotypes and risk of SPC. Recently, a principal component (PC) meta-analysis on multiple anthropometric traits (weight, height, BMI, waist circumference, hip circumference, and waist-to-hip ratio) identified four PCs for body shape phenotypes and six novel corresponding genetic loci. These composite body shape phenotypes are a novel and promising concept and are potentially more specific in terms of body composition and body fat distribution and may improve risk discrimination for cancer outcomes. We hypothesize that principal components of multiple anthropometric traits capturing different body shape phenotypes are associated with survival and risk of SPC in cancer patients. We will address this hypothesis by combining two complementary state-of-art epidemiological approaches.1. Prospective analysis of large cohorts including the EPIC study, the German National Cohort (GNC), and UK Biobank. These cohorts complement each other in their uniqueness and together provide a large number of validated cancer outcomes ascertained post-recruitment, repeated measures of anthropometry, advanced assessment of body composition and body fat distribution based on magnetic resonance imaging and precise data on relevant confounding variables.2. Mendelian randomization (MR) approaches, which address key biases of observational studies such as confounding and reverse causation.The findings of this project will inform management of cancer patients based on anthropometric traits that are commonly measured in clinical practice, and will also inform public health decisions and cancer prevention strategies.
