Doctors face a major challenge when determining who with obesity is most likely to develop obesity complications and who to prioritise for treatment. Precision medicine could help address these challenges, according to researchers.
In a paper from the IMI SOPHIA consortium, published in the journal Nature Medicine, a new precision prediction algorithm is described that reveals previously unknown subtypes of obesity that raise risk of developing type 2 diabetes and heart disease.
“At a population level, being heavier is generally worse for health. But, when you look more closely, at an individual level, more complex patterns exist that can be harnessed to improve disease prediction,” commented Dr. Ewan Pearson, Professor of Diabetes Medicine from Dundee University.
“The amount of fat or sugar in a person’s blood, for example, can be much higher or lower than you’d expect when considering their body weight alone, which in turn affects the person’s risk of obesity complications,” said the paper’s lead author, Dr. Daniel Coral, from Lund University Diabetes Centre in Sweden. “This is missed by standard clinical prediction tools, meaning that about 1 in 5 people who might need early interventions are overlooked. The algorithm we’ve developed may help clinicians and patients in the future”, explains Coral.
The research focused on 170,000 adults from the UK, the Netherlands and Germany in whom detailed clinical information had been gathered. Using an artificial intelligence method called “machine learning”, the researchers developed powerful algorithms that split obesity up into 5 separate diagnostic profiles with contrasting risk of developing obesity complications.
"Obesity is both common and heterogeneous, meaning that the health risks one person with obesity confronts may differ substantially from those faced by someone else with obesity. Figuring out who has the highest health risks is important because this may lead to more precise, accurate and timely prevention and treatment," added Dr. Paul Franks, Professor of Genetic Epidemiology at Lund University Diabetes Centre, the paper’s senior author.
The research was led by scientists at Lund University Diabetes Centre in Sweden, and Maastricht Centre for Systems Biology and Erasmus MC University Medical Centre in The Netherlands, in collaboration with other researchers from the IMI SOPHIA consortium.
Key findings included:
- Most people (~80%) had health markers that matched what’s expected for their body weight.
- About 8% of women had higher blood pressure than expected for their weight, coupled with higher "good" cholesterol (HDL) and a lower waist-to-hip ratio (WHR), which means they had more fat in their hips and less around their waist than expected for their weight. This wasn’t seen in men.
- Around 5% of women and 7% of men had a profile with high "bad" cholesterol (LDL), high triglycerides (fat in the blood), a high WHR (more fat around the waist), and higher blood pressure than expected for their weight.
- About 5% had high liver enzymes (ALT) and a high WHR for their weight.
- About 4% showed higher inflammation (measured by CRP) than expected for their weight.
- Around 2.5% had high blood sugar and lower LDL for their weight.
The full paper can be viewed at:
"Subclassification of obesity for precision prediction of cardiometabolic diseases"