856. Physical activity and cancer: an umbrella review of the literature including 22 major anatomical sites and 770,000 cancer cases
Commentary by Ms Yi Yang, PhD Candidate
Associate Professor Brigid Lynch, Principal Fellow Cancer Epidemiology and Intelligence Division at Cancer Council Victoria
Cancer is expected to become more prevalent in coming years, as global trends in population ageing continue, and low and middle-income countries increasingly adopt high income country lifestyle habits. It is estimated that by 2020 the number of incident cancer cases and cancer deaths will rise to approximately 17 million and 10 million, respectively. Thus, it is critical to clearly understand how modifiable lifestyle factors, such as physical activity, contribute to cancer risk.
With the increasing number of systematic reviews and meta-analyses on physical activity and cancer, an umbrella review helps to synthesise the existing evidence, and highlights whether independent groups of researchers reached consistent (or contradictory) conclusions. An umbrella review provides a broader picture of the current state of evidence than a single systematic review or meta-analysis, which is helpful for formulating future research foci.
De Rezende et al. recently published an umbrella review that quantified the methodological quality of systematic reviews and meta-analyses related to physical activity and cancer risk, and generated fixed effect models to meta-analyse findings. Overall, their results were very similar to the findings from the World Cancer Research Fund/American Institute of Cancer Research Second Expert Report and subsequent Continuous Update Project findings. A key limitation identified by this umbrella review was that, for cancer sites other than breast and colon, the evidence was limited by small study effect bias.
While we commend de Rezende et al. for synthesizing the available evidence in a rigorous and quantitative fashion, it is important to understand that umbrella reviews cannot overcome some of the methodological challenges related to physical activity research. For example, there is considerable heterogeneity in how physical activity was quantified across the studies incorporated in this umbrella review. Further, the meta-analyses performed within this review pooled risk estimates derived from comparing the highest to lowest physical activity categories within each study. The differences in how physical activity is categorized between studies (e.g. some studies may have an upper category of ≥ 10 MET h/wk, whereas for other studies the upper category may be ≥ 30 MET h/wk) may lead to an underestimation of the effect on cancer risk.
These particular limitations can be overcome by harmonising and pooling data from relevant studies. The National Cancer Institute’s Cohort Consortium physical activity pooling project harmonized leisure-time physical activity data from 12 cohort studies involving 1.44 million adults from across the U.S. and Europe. Analysis by Moore et al. found that leisure-time physical activity decreased the risk of 13 different cancer types: oesophageal adenocarcinoma, liver, lung, kidney, gastric cardia, endometrial, myeloid leukaemia, myeloma, colon, head and neck, rectal, bladder, and breast. Estimated risk reductions ranged from 42% (95% CI: 0.37, 0.89) for oesophageal adenocarcinoma to 10% (95% CI: 0.87, 0.93) for breast cancer, with seven sites (oesophageal adenocarcinoma, liver, lung, kidney, gastric cardia, endometrial, and myeloid leukaemia) associated with a risk reduction of at least 20%.
A limitation of the harmonized pooling project, umbrella reviews and meta-analyses of studies on physical activity and cancer is that evidence has been derived from a single exposure time-point (typically reflecting behaviour at the time of study entry). Whether and how changing lifestyle, especially at mid-life or later, will affect cancer risk is not well understood. A recent review from Lynch and Leitzmann suggests that exploiting longitudinal data may improve the understanding of how interventions to decrease physical inactivity could affect cancer risk.
It has not yet been established whether the associations between physical activity and some cancers are causal. It is not feasible to conduct randomised controlled trials of physical activity to assess effect on cancer risk. To establish causality, triangulation of evidence from different approaches to this research question, including epidemiological, experimental and mechanistic studies, is required.