A bug in the system – the difficulties of linking the microbiome to cancer aetiology

Cancer

Variations in the human gut microbiome have been linked to cancer an exciting prospect for cancer prevention, but teasing apart causation from correlation is no easy task says Dr Kaitlin Wade…

Cancer remains one of the leading causes of death worldwide as well as one of the greatest economical burdens on health care systems. And yet, evidence indicates that over 40% of all cancers are likely explained by preventable causes.

One of the main challenges is identifying so-called ‘modifiable risk factors’ for cancer – aspects of our environment that we can change to reduce the incidence of disease. One very promising avenue of research has been the gut microbiome. There is growing evidence – from human and predominantly mouse models – supporting the relationship between the human gut microbiome and cancer aetiology.

Human studies have largely been observational, and investigations have so far been unable to offer convincing causal evidence

We know the gut microbiome can have a substantial impact on host metabolism, inflammation, and immune response to external infections, so there are many plausible biological mechanisms by which it could influence cancer development and progression. However, findings have been inconsistent, or even contradictory, and very few hypotheses have been reliably supported with data from multiple model organisms.

Human studies have largely been observational, and investigations have so far been unable to offer convincing causal evidence. This isn’t helped by several important limitations in the design of studies and analyses of data linking the gut microbiome and cancer aetiology. Common causes of gut microbiome variation and cancer (confounding), the ability for cancer to influence the gut microbiome (reverse causation) and various biases can distort results. This, of course, affects our ability to find out what variation in the human gut microbiome, if any, may cause cancer. Distinguishing correlation from causation therefore requires very precise data analysis.

Human genetics, microbial life

The gut microbiome is a complex system of microorganisms aiding digestion, providing protection against pathogens and creating essential metabolites. Variation in the gut microbiome has been linked to many common cancers.

Taking colorectal cancer (CRC) as an example, there is compelling in vivo and in vitro evidence that modifying gut microbiota may reduce the incidence of the disease. Alongside this, epidemiological studies suggest a lower microbiota diversity in people with CRC. There is also research showing lower levels of some bacteria, such as Bifidobacterium and Roseburia, as well as higher levels of others, such as Fusobacterium and Porphyromonas, in those with CRC.

Despite a lack of causal evidence, there is still a growing market for commercial products targeting the microbiome

Despite a lack of causal evidence, there is still a growing market for commercial products targeting the microbiome – several companies now offer sequencing of faecal samples and prescribe “personalised” nutritional information. There is controversy around this given the uncertainty of the likely impact, which is not helped by a lack of consistency between observational studies and large-scale randomised controlled trials. Something which is clearly a barrier when trying to harness the gut microbiome to tackle disease. What this does highlight though, is the public demand for such information which could suggest an untapped opportunity to make important population-based health interventions. Therefore, we need alternative approaches to interrogate causality and tease apart the links between the gut microbiome and cancer aetiology.

Interrogating causality

One way of improving causal inference has been the integration of human genetic variation within population health sciences. With the growth in genome-wide association studies (GWASs), we now know thousands of genetic variants across the genome that influence almost every aspect of human physiology and even elements of behaviour.

We need alternative approaches to interrogate causality and tease apart the links between the gut microbiome and cancer aetiology

Within the last few years, GWAS has been used to understand the relationships between human genetic variation and the gut microbiome. These studies have provided evidence for the contributions of human genetics on features of the gut microbiome such as diversity, abundance and enterotype. While this is not in itself causal evidence, knowledge of the relationships between human genetic variation and various characteristics has provided an opportunity to tease out causality from observational epidemiological associations.

Established in the early 2000s and applied mainly to understand the links between modifiable risk factors and cardiometabolic diseases, Mendelian randomisation (MR) is a method that enables the interrogation of causality.

MR utilises human germline genetic variation – usually single nucleotide polymorphisms – to help investigate whether the gut microbiome changes the risk of cancer or whether cancer changes the gut microbiome. Genetic variation can’t be influenced by the gut microbiome or disease. Therefore, if people who are genetically predisposed to having a higher abundance of certain bacteria within their gut also have a lower risk of cancer, this would strongly suggest a causal and protective role of those bacteria in cancer aetiology.

With the recent growth in GWASs focusing on the gut microbiome, there have been a handful of studies applying MR to assess the impacts of gut microbiome variation on several cardiometabolic, inflammatory and auto-immune diseases. As yet however, there are no studies that focus on the appropriate application of these methodologies to cancer – something I aim to change as part of my Cancer Research UK fellowship. I plan to use human genetic information to shed light on the relationship between the gut microbiome and cancer aetiology.

Knowledge of the relationships between human genetic variation and various characteristics has provided an opportunity to tease out causality from observational epidemiological associations

Using MR should largely avoid the limitations of observational epidemiological studies however, the specifics of the way the method is applied is very important and is something that can be hard to get right. Clearly, I want to ensure my results give a reliable indication of causality. So, first order of business for this study will be to apply a more robust way to tease apart correlation from causation. To be able to do this, there are a number of caveats to the current use of MR in the field that need to be considered carefully.

These caveats centre around the core assumptions of the MR framework. First, that human genetic variation must be associated with the gut microbiome. Second, that there must be no confounding – that is to say common causes of the gut microbiome and cancer and, third, there must be no relationship between microbiome-related genetic variation and cancer independent of the gut microbiome. However, the current applications of MR to try and understand the role of the gut microbiome on health outcomes rarely consider these caveats carefully enough.

Inter-disciplinary triangulation

The appropriate application of MR to interrogate causality of the gut microbiome in cancer has begun to show promise. However, early work has also highlighted the importance of inter-disciplinary collaboration between population health, genetic and basic sciences. We really do need a triangulation of evidence to unpick causation from correlation. Any research conducted within one discipline cannot provide concrete evidence to support or challenge the role of the gut microbiome in cancer aetiology.

I am hopeful with this Fellowship, and the support from my team of experts in microbiology, basic sciences and population health sciences, we can take a new and important step towards refining the current applications of complex integrative methodologies in cancer research. And it is this which will in turn allow more accurate evaluation of potential treatments or protective factors for cancer prevention.

About the author

Dr Kaitlin Wade is a lecturer in epidemiology and Co-Director of the MSc in Epidemiology in the MRC Integrative Epidemiology Unit based at the University of Bristol. She was awarded a Cancer Research UK Population Research Postdoctoral Fellowship in 2020.

Acknowledgements
The research conducted as part of my CRUK Population Research Postdoctoral Fellowship will be supported by the following collaborators: Nicholas Timpson, Caroline Relton, Jeroen Raes, Trevor Lawley, Lindsay Hall and Marc Gunter. Additional thanks to Chloe Russell, who supplied the image for this piece.

More on this topic

Products You May Like

Articles You May Like

High junk food consumption in childhood increases risk of unhealthy body fat by early adulthood
A Mix-And-Match Approach To COVID-19 Vaccines Could Provide Logistical And Immunological Benefits
Hormone Pellet Safety Data ‘Not Very Reassuring at All’ for Women
How Effective Was the Latest COVID Vaccine Candidate?
Researchers report a new genetic cause for infantile fractures

Leave a Reply

Your email address will not be published. Required fields are marked *