One of the key foundations of epidemiology is causation (Hill’s causal criteria were first detailed in 1965) – and that conclusions about causation are based on a number of factors including a plausible biologic mechanism. Thus, research papers outlining associations between risk factors and disease typically outline the likely or possible mechanisms that would link the risk factor with the disease endpoint. Despite convincing evidence from observational studies that link numerous risk factors with cancer endpoints, in most cases we can’t point to a mechanistic pathway and say “this is how.” In part, this is likely because many risk factors operate through multiple pathways. But understanding those pathways remains an important priority in cancer prevention research.
While a sufficient body of evidence exists that numerous independent bodies have concluded that physical activity is causally related to a reduced risk of several cancers, including colon and breast, the mechanisms remain relatively poorly understood. A few candidates are typically mentioned and are the focus of most current research in this area – physical activity has anti-inflammatory effects, reduces insulin resistance and may alter sex hormone levels.
Mechanism studies tend to be conducted in animals or cell lines and if conducted in humans, tend to be relatively small or from non-randomized trials. This week, results came from one of the largest supervised exercise training randomized controlled trial studies conducted in women. 316 women between 18 and 30 were enrolled in a 16 week aerobic exercise intervention. The women randomized to the intervention arm participated in physical activity at levels recommended by public health guidelines: at least 30 minutes of activity at least 5 times a week. Adherence and retention rates were both high indicating a methodologically rigorous study.
Overall, the study found little that could be a link in the causal pathway between activity level and cancer. The insulin-like growth factor (IGF) family of hormones and binding proteins, which are involved in growth and development, have long been suspected as possible mediators between exercise and cancer. Yet, apart from a statistically significant lower level of IGFBP-3 in controls, none of the other IGF levels were different between exercisers and non-exercisers. The same held for levels of glucose, insulin, and a measure of insulin resistence (HOMA).
What does this mean? Previous studies have suggested that the IGF family is associated with premenopausal breast cancer risk – specifically that IGF-1 levels are associated with increased risk of ER+ breast cancer. It may be that the physical activity levels recommended by public health guidelines aren’t intense enough to materially alter levels in the IGF family. Or it may be that exercise needs to be consistent over a longer period of time to see significant changes.
Does this mean we should dismiss the data showing that exercise reduces cancer risk? Not at all. What it does highlight is that randomized trials have great utility for examining very precise questions. This trial studied one specific dose of exercise over a specific number of weeks. The trial can’t tell us what more vigorous activity would have done, or what 60 minutes/day of activity does to the IGF family or what consistent activity over several years does, if anything, to IGF.
Sometimes, randomized trials raise more questions than they answer. Thus, to move science forward, we need BOTH observational and randomized trial evidence.