The power of thinking big

Scientists have learned a great deal about the risk factors for cardiovascular disease (CVD), the group of conditions that includes heart disease and stroke and which kills one in three people in the UK, through studying the health of large numbers of people. But, as new risk factors emerge, how can clinicians know what to include in an accurate assessment of an individual’s risk of CVD? Moreover, which risk factors are causal, and therefore directions for potential intervention, as opposed to simply being associated with the onset of disease?

The answers are in the numbers. By conducting an analysis on a scale never attempted before, Professor John Danesh and his team at the University of Cambridge’s Cardiovascular Epidemiology Unit (CEU) in the Department of Public Health and Primary Care are gaining new insights into the causes of, and potential interventions into, CVD.

Bigger is better

Since the late 1940s, when the Framingham Heart Study in the USA first undertook the ambitious task of following the development of CVD in a large group of individuals over a long period of time (originally around 5,000 healthy men and women), many longitudinal population studies into CVD have been conducted across the world. Each strives to understand, predict and prevent a global epidemic that currently causes 17 million deaths per year.

Many of these early studies identified the major CVD risk factors that are now so familiar to us – high blood pressure, high blood cholesterol, smoking, obesity, diabetes and physical inactivity. But a significant number of people who succumb to cardiovascular events do not display any of these warning signs.  In recent years, many new or emerging risk factors have been proposed, among them mediators and products of the inflammatory system.

Over a decade ago, Danesh realised that by pulling together the large number of existing population studies into a dataset of unprecedented size and potential, he and his team could use the power of scale to understand the risk factors for CVD as never before. In 2003 he established the Emerging Risk Factors Collaboration (ERFC), an international collaboration that collates the work of 130 epidemiological studies from more than 20 countries, creating a primary dataset that comprises information collected from around 2.5 million healthy volunteers.

“With this central database we can combine exceptional power and exceptional detail,” said Danesh, Director of the Unit and Head of the University’s Department of Public Health and Primary Care. “The ERFC has information on many risk factors, both known and suspected, for CVD, and also information on the resurveys of these risk factors at different time intervals in large subsets of the people. Up to 100,000 of the 2.5 million individuals involved developed a new onset of heart attack or stroke after their entry into the study, making the collaboration extremely valuable for evaluating potential cardiovascular risk factors.”

The enterprise has not been without its challenges. Emanuele Di Angelantonio, a University Lecturer in Medical Screening and part of the team using the dataset to reanalyse and assess novel risk factors for heart disease, explained: “There are many technical issues involved in working with such a massive dataset from so many sources. Different studies define risk factors in slightly different ways, and we had to come up with clever algorithms to standardise these so we can address the questions in the correct way. Another challenge was the statistical analysis. When science reaches a certain scale, it merges with all sorts of other considerations. We had to come up with whole new statistical methodologies to support the research programme.”

Improving clinical practice

By conducting detailed re-analyses of data collected worldwide, the ERFC is starting to provide reliable answers to long-standing clinical controversies surrounding the assessment of people’s risk of heart disease, helping to define exactly what it is useful to measure. Important lipid and inflammatory biomarker findings are already emerging that have implications for clinical practice.

“If you’ve ever had your blood cholesterol measured, the convention is that you fast overnight first because that was thought to provide the best measurement for an optimal prediction of heart attack risk,” said Danesh. “In fact, a report from this collaboration suggests that non-fasted, or random samples are at least as good, potentially even better, than fasted samples. It’s also been the convention to measure levels of a type of blood fat called triglyceride, but our work suggests that once you have the cholesterol information, the triglyceride information is not at all useful.”

The work has also identified a new causative risk factor, a novel blood fat called lipoprotein-a, the relevance of which to the risk of heart attack had previously been uncertain for a long time. “Because the contribution of lipoprotein-a to the risk of heart attack is only about a quarter of the strength of the contribution of bad cholesterol, we required a much larger study to identify and confirm its effect than we would for really strong risk factors,” explained Danesh. “This positive confirmation has only been possible because of the power provided by the unprecedentedly large dataset of the ERFC.”

Such evidence has been cited by the 2011 guidelines of the European Atherosclerosis Society consensus panel for therapeutic targeting of patients with elevated triglycerides and/or low levels of good cholesterol, and by the 2010 guidelines of the European Atherosclerosis Society recommending lipoprotein-a for use in cardiovascular disease risk assessment. It has also informed phase III clinical trials of a drug called darapladib, by GlaxoSmithKline, for the treatment of atherosclerosis, and encouraged the American College of Cardiology Foundation/American Heart Association Task Force to recommend assessment of mass concentration of lipoprotein-associated phospholipase for targeted cardiovascular disease risk assessment in their 2010 guidelines.

Adding genetics to inform medical interventions

The ERFC provides useful data on the correlations between levels of biomarkers in the blood and their association with the risk of heart disease. But there is a difference between biomarker molecules that are associated with heart disease and those that actually contribute to it.

Dr Adam Butterworth, a University Lecturer in the CEU, has been trying to identify the causal biomarkers by finding related genetic variants that are associated with heart disease. “For example, the idea of a link between heart disease and inflammation has been around for a long time, but it has proven difficult to identify specific biomarkers and genes to explain that link,” he said.

A new collaboration, taking the same large-scale approach as the ERFC, has used human genetics to show that a functional allele (a variant of a gene) in an inflammation-related gene is convincingly associated with heart disease. “The IL6R Genetics Consortium is a collaboration of 46 genetic studies that was created to investigate the effects of a specific genetic variant in the interleukin-6 receptor (IL6R) gene on heart disease, inflammatory biomarkers and other vascular risk factors,” he explained.

“We integrated the data from these studies of the genetic variant and associations with heart disease, with the data on blood levels of the IL-6 and IL6R proteins and associations with heart disease from the ERFC, and tried to understand a bit more of the story.” The study, published in March 2012 in The Lancet, suggests that IL6R-mediated pathways are causally related to coronary heart disease.

“The results suggest that if you want to intervene to lower CVD risk, one option might be to reduce the signalling of the IL6 pathway,” he said. “What is interesting is that there is already a drug on the market targeting this particular pathway, currently for the treatment of rheumatoid arthritis.” This finding heralds a new, more targeted and thus potentially more cost-effective approach to drug development for therapeutic intervention.

A complementary approach being used by the CEU is to measure hundreds of thousands of genetic variants, suggested as being of potential relevance to heart disease, at the same time using microarray technology. A new consortium will deliver results of a ‘mega-analysis’ of more than 200,000 genetic variants in coming months, helping to define those variants most strongly associated with CVD. And, by 2013, further microarray studies carried out by Butterworth and colleagues are expected to yield information on more than 200,000 additional genetic variants, many that are much rarer than those previously studied.

The work of the CEU has had a considerable impact on public health, clinical guidelines and priorities for medicines development in cardiovascular disease, as Danesh explained: “Through studies such as the ERFC, we are not only building on previous findings but also producing results that are robust enough to be applied to public health interventions.”

This research is funded by the British Heart Foundation, Medical Research Council, and National Institute of Health Research Cambridge Biomedical Research Centre.