Researchers at Karolinska Institutet in Sweden and Copenhagen University Hospital in Denmark have, with the help of AI, identified a new and very promising biomarker known as thrombomodulin, for bacterial soft tissue infections. The study was published in the Journal of Clinical Investigation.
Necrotising soft tissue infections (NSTI) are bacterial infections characterised by rapid tissue degradation. These infections, often caused by streptococci, are relatively uncommon but extremely serious, in most cases requiring intensive care and may quickly become life-threatening.
The new biomarker is specific to different patient groups with soft tissue infections, say the researchers. Using machine learning, a type of artificial intelligence, the team analysed 36 soluble factors in blood plasma from the 311 NSTI patients included in the international INFECT study. Control groups included patients with suspected NSTI and sepsis, respectively. The analyses showed a new biomarker that identifies patients with tissue necrosis to a considerable degree of accuracy.
Rapid diagnosis is important in bacterial soft tissue infections to reduce the risk of severe injury or amputation. Vague symptoms such as vomiting, fever, severe pain and a heterogeneous patient group increase the risk of misdiagnosis. Extensive surgery, intravenous antibiotics and sometimes even amputation are required to prevent the infection from spreading. Many patients also develop sepsis, which further complicates the course of the condition.
Currently, various laboratory tests, including the number of white blood cells, are used as diagnostic tools, but these are low-sensitivity techniques.
“The new biomarker, thrombomodulin, proved to be superior to the laboratory parameters used clinically today. The analyses also identified biomarkers for patients with soft tissue infection caused by different types of bacteria, as well as patients who developed septic shock,” said Laura Palma Medina, a researcher at the Department of Medicine, Karolinska Institutet (Huddinge), and the study’s first author.