Over the last 30 years, the traditional methods used to understand the links between nutrition and cancer have been observational, based on analysing data collected from people who have been followed up for many years, such as the European Prospective Investigation into Cancer and Nutrition (EPIC) study.
These studies have established very large and important datasets, and have continued to be the go-to resource for performing analyses, providing us with interesting and insightful results on the links between nutrition and disease.
What do we know so far?
We now know that around 40% of cancer cases and cancer deaths in high-income countries are thought to be explained by known modifiable lifestyle and environmental risk factors.
As shown in the WCRF/AICR Third Expert Report, these could be aspects of diet (eg alcohol consumption; eating large amounts of red and processed meat; or diets low in fruits, vegetables, wholegrains and dietary fibre) and other related lifestyle factors (eg overweight and obesity, physical inactivity, smoking, and metabolic factors) that work collectively to increase our cancer risk.
The patterns of lifestyle behaviours that are more likely to protect us against cancer led World Cancer Research Fund to develop our Cancer Prevention Recommendations.
So why the need for more tools?
Studies have shown that adherence to our Recommendations may substantially reduce cancer risk1,2, but when we get to specific nutrients or foods, there are still inconsistent associations, using the traditional methods mentioned above.
Why is this? Simply because these traditional methods are vulnerable to potential biases, and subsequent clinical studies don’t always confirm initial leads. They fail to tease out the specific benefit or harm of each nutrient independently, rather than being part of an overall diet.
These biases that we still cannot fully overcome could be:
- confounding between factors, because a lot of the factors do not operate in isolation.
- exposure misclassifications – such as incomplete medical records or incorrect information from questionnaires that measure dietary intake – because people cannot always recall the exact information when it comes to their dietary habits.
- reverse causality, which is the instance of a reverse or not-expected exposure-disease direction.
What is the new tool?
Mendelian randomisation (MR) is the tool that will hopefully help with some of the above challenges. As experts explain, it’s like a magnifying glass that looks at the specific nutrients independently and their interactions with other components of the diet or general lifestyle, then helps disentangle the direction of causality.
MR is a statistical method, a way to analyse data, that uses people’s genetic variation – in other words the different genes that we carry and which form our unique footprint.
This variation helps to approximate a dietary exposure independently of other dietary exposures and – ultimately – pinpoint a causal effect to a specific nutrient. This effect will be distinct from other factors that might be correlated with that nutrient, as part of people’s diet or lifestyle.
This is possible because, if there is a lot of data available, at a population level that unique footprint is less likely to be associated with environmental factors, which are precarious and may confound the associations found from observational studies.
World Cancer Research Fund supports many MR projects and the research community that delves into this field to shed light on the causal relationship between nutrition and multiple cancer sites.
We have already funded 9 projects in the UK, France, Japan, Germany and Sweden. More recently, we have supported the MR in Nutrition and Cancer workshop, a working group from experts in the field.
A position paper from that workshop provides two key outputs:
- an overview of MR, discussing applications and providing examples of MR studies in nutrition and cancer.
- the limitations of the tool, and how to overcome those to translate findings from MR studies into nutritional recommendations and policies for cancer prevention, that will be used from the scientific community around the world.
Challenges of Mendelian randomisation
Of course, as with every statistical method, MR comes with its own challenges. You need studies with large sample sizes to avoid biases, which is often difficult to achieve unless different teams work together and pool data from smaller studies.
It’s crucial to have enough data to showcase a robust association between nutritional traits and genetic variation in the population.
This association, in turn, increases our level of certainty in the findings and their interpretation – which is called statistical power – and the chances of detecting relationships of modest strength that are not obvious.
Another challenge originates from the limited understanding of the biological mechanisms that underpin the detected associations and this requires caution when interpreting the findings.
Scientists have achieved great progress in developing methods to mitigate those challenges, such as performing more sensitive analyses, and forming large consortia and international collaborations. In an attempt to expand the focus of MR studies beyond populations of European ancestry to tailor findings to other populations as well, they have also increased the number of trans-ethnic databases.
Nutrition is a puzzle of complex and interconnected relationships, but the scientific community is devoted to unravelling the mystery. The aim is to give us clear explanations and modifiable nutritional risk factors for cancer development and progression, and the belief is that MR has the potential to contribute significantly to those efforts.
This is of particular importance to us as we embark on the next phase of the Global Cancer Update Programme. MR will complement existing methodologies for evidence synthesis and interpretation, and will equip us with robust evidence and up-to-date recommendations for specific cancer sites.
Bethany van Guelpen
> The role of inflammation in the link between obesity and colorectal cancer
> The impact of obesity and insulin on cancer risk