Our Gut Microbiota as a Health Compass

The microorganisms living with us in our bodies – collectively called the microbiota – outnumber our human cells. Recognition that the human body hosts a rich microbiota began as early as the 17th century, but it was not until the last few decades that we became aware of its effects on health. Especially over the last 10 years, microbiota has become a central focus of medicine and basic scientific research, particularly the gut microbiota. In fact, the microorganisms in our bodies are not found only in the gut; the entire skin is also covered with bacterial flora. 

Note: Microbiome refers to all the genetic information (genomes) of microorganisms in a particular environment. Microbiota is the community of microorganisms. Humans have 30 trillion human cells and 20 thousand genes in our DNA. In contrast, approximately 35-100 trillion microbiota cells live with us in our bodies, while the microbiota is estimated to have around 2 million genes.

The ability of the human microbiota to provide information about the risk of “non-alcoholic fatty liver disease”, which we know as fatty liver, was discovered by an international team led by the Hans Knöll Institute in Germany. The researchers developed a model that can predict the likely course of the disease based on the microbial composition in the gut. The study was published in Science Translational Medicine.

Approximately 25% of the world’s population is affected by non-alcoholic fatty liver disease, which causes an increased formation of fat cells in the liver (referred to below as fatty liver). This is the most common chronic liver disease in industrialized countries and, unlike alcoholic fatty liver disease, is not caused by high alcohol consumption. In some people, undetected fatty liver can lead to liver cirrhosis, liver cancer, or liver failure.

In a long-term study, an international research team led by Gianni Panagiotou, leader of the systems biology and bioinformatics research group at the Hans Knöll Institute, analyzed stool and blood samples from 1200 people who did not initially have fatty liver. Panagiotou explains the goal of the study as follows: "The microorganisms in the human gut had already been shown to play a role in fatty liver. We wanted to find out whether the microbiomes of healthy people could predict whether they would develop fatty liver in the future."

When the participants were examined again after 4 years, 90 were found to have developed fatty liver in the meantime. Samples from the affected participants were compared with those from a control group of 90 people who had fatty liver neither at baseline nor during follow-up.

"Using these data, we were able to develop a model to predict, based on the microbiome, who would develop fatty liver in the future with 80% accuracy." There are currently clinical models that predict with 60% accuracy using biochemical parameters in the blood. "The model we developed combines data obtained from the microbiome with measurable information obtained from blood and can thus greatly increase reliability."

Predicting Disease with Machine Learning

The research team developed a computer model trained to remember a particular pattern in a dataset, called a machine learning model. The model then uses these patterns to analyze new datasets, that is, to predict the probability of non-alcoholic fatty liver disease. “The entire process of developing our model took 3 years because of the complexity of the data. But we finally succeeded and were able to create a useful tool for predicting fatty liver.”

End-stage non-alcoholic fatty liver disease is irreversible and may even lead to an aggressive type of liver cancer. People already at particular risk of progression to liver cancer should be identified early to counter the disease. Fatty liver is a silent disease. This means that most cases are asymptomatic (showing no symptoms) and are often discovered only by chance. Approximately 12 million Germans are estimated to have this form of fatty liver. People with preexisting conditions such as type 2 diabetes, obesity, high blood pressure, or dyslipidemia (an imbalance or elevation of blood fats) are particularly affected by fatty liver disease.

Potential Applications and Next Steps

Using machine learning models, the researchers have already been able to compare their results with patient data obtained from the US and Europe, thereby validating them. In the next step, Panagiotou plans to conduct the study worldwide and use artificial intelligence to integrate even larger datasets.

Panagiotou said, “I see microbiome-based diagnostics as something that will reach clinical practice and have great potential in the next decade.” Early treatment of risk factors for non-alcoholic fatty liver disease, such as type 2 diabetes, hypertension, and obesity, can stop the disease from developing. Therefore, early prediction is the only way to prevent the disease.