Metabolome analysis reveals risk for multiple diseases simultaneously
To prevent diseases, it is important that people who are at particularly high risk are detected as early as possible. However, current screening methods are often costly and limited to single diseases. Scientists from the Berlin Institute of Health at Charité (BIH), Charité – Universitätsmedizin Berlin and University College London have now analysed blood values of 168 metabolites and the medical histories of over 100,000 people. Using artificial intelligence, they were thus able to calculate the risk of several diseases at the same time and indicate where early intervention might be worthwhile. They published their findings in the journal Nature Medicine*.
Prevention is better than cure: This motto was in the back of the minds of researchers from BIH, Charité and University College London when they gained insight into the vast trove of data in the UK Biobank. The British study has been tracking the fate of more than 500,000 participants for more than 15 years. Because all Britons have had electronic patient records since the 1990s, the development of diseases can be monitored here – pseudonymously – over long periods of time.
Recently, the UK Biobank published an immense data package: The frozen blood samples of the participants, some of which were more than 15 years old, had been analysed for their content of 168 metabolic products using nuclear spin spectroscopy. The method is considered reliable, easy to implement and relatively inexpensive. Substances such as cholesterol or blood sugar are measured, but also molecules that are less well known and also less frequently determined in blood tests. “Recent studies have shown that individual metabolites – or metabolites – are relevant to the development of a variety of diseases. We suspected that the combination of several different metabolites might provide clues to risk for different diseases at the same time. And that’s what we wanted to investigate,” explains Jakob Steinfeldt, assistant physician at the Medical Clinic for Cardiology at Charité Campus Benjamin Franklin.
Together with colleagues from the BIH’s Digital Health Centre, the scientists then examined the participants’ data for 24 common diseases, including metabolic disorders such as diabetes, cardiovascular diseases such as heart attacks and cardiac insufficiency, as well as neurological diseases such as Parkinson’s, muscle diseases and various cancers. For each of the 24 diseases, they first determined which participants had developed the disease during the course of the study and then combined this with the composition of the metabolites in the blood serum, the metabolome, which had been taken before the onset of the disease. From this, they used artificial intelligence to calculate a model that predicted the likelihood that a given metabolite combination in the blood would predict future disease.
“We tested the metabolite profiles for their predictive power and compared them with conventional methods for calculating risk,” reports Thore Bürgel, a doctoral student at BIH’s Digital Health Centre and first author of the paper together with Jakob Steinfeldt. “This showed that the profiles improved risk prediction for the majority of the diseases studied when we combined them with information about the age and biological sex of the participants.”
For example, the combination of age, biological sex and metabolome was better able to predict the risk of diabetes or cardiac insufficiency than established risk models based on a conventional determination of blood glucose or cholesterol in the blood. At a cost of less than 20 euros, the metabolome test is also relatively inexpensive. “This is interesting because we can use the metabolome to estimate the risk of many diseases at the same time,” explains Prof. Ulf Landmesser, M.D., director of the Medical Clinic for Cardiology at Charité Campus Benjamin Franklin. “Of course, after a ‘risk warning’ due to abnormalities in the blood, we would further examine the patient before intervening. Nevertheless, this is exactly the direction we also want to take with the new ‘Friede Springer – Cardiovascular Prevention Center’: motivating people to have regular examinations after a certain age so that they can take preventive action in good time if the worst comes to the worst.”
The scientists took their model a step further and calculated which threshold values could be suitable for preventive interventions. Specifically, at which thresholds could the new method best identify people in order to prevent them from developing heart muscle weakness, for example, through the use of medication? “Again, we saw that metabolomics analysis combined with information on age and biological sex was equivalent or even better than conventional analyses at identifying people who could benefit from preventive intervention in the form of medication or lifestyle changes,” says Prof. Roland Eils, founding director of BIH’s Digital Health Centre at Charité. He adds, “We have subsequently been able to successfully validate our model in four other large population studies from the Netherlands and the UK, indicating that our models are broadly applicable.”
Also closely involved in the work was Prof. John Deanfield, MD, a cardiologist from University College London. As an Einstein BIH Visiting Fellow, funded by the Charité Foundation, he regularly visits his host Prof. Landmesser in Berlin; conversely, Prof. Landmesser and Prof. Eils have been with him in London. Both emphasize: “Science transcends borders between countries and disciplines. The connection to London, coupled with the great openness of the UK Biobank to make its data available for studies worldwide, has enabled us to do this great work.”
Medizinische Klinik für Kardiologie (Campus Benjamin Franklin)
Digitale Gesundheit am Berlin Institute of Health in der Charité (BIH)