Ultrafast sepsis test pinpoints treatment in just 13 hours

The immune system is generally a pro at fighting off microbial infections. But sometimes it goes rogue, turning itself on the body’s own organs and tissues, resulting in a life-threating condition called sepsis. According to the World Health Organization, in 2017 sepsis was responsible for 20% of deaths worldwide. Pinpointing the bacterial culprit and the right antibiotic treatment for sepsis can take a harrowing 2–3 days.

“You need to find the optimal therapeutic as soon as possible,” says Tae Hyun Kim, a research scientist at the California Institute of Technology. “The survival rate decreases every hour.”

Kim and his colleagues developed a testing method for sepsis that reduces the turnaround time from collecting a blood sample to finding the right treatment to about 13 h (Nature 2024, DOI: 10.1038/s41586-024-07725-1).

“Our objective was to make this into a commercial device that could be used to help patients,” says Kim, who did the work as a postdoctoral scholar in the lab of Sunghoon Kwon, a biomedical engineer at Seoul National University.

The most time-consuming step in diagnosing sepsis is the first one: culturing a patient’s blood sample. Because pathogens in the blood are so few in number, it usually takes about a day for them to grow to detectable levels in a culture. Kim and his colleagues set out to bypass this step with the help of a naturally occurring immune system peptide called β-2-glycoprotein I (β2GPI), which recognizes a broad range of pathogens, including the virus that causes COVID-19. When they add a magnetic nanoparticle decorated with the synthetic version of this peptide, the researchers could quickly pull different bacterial species from a blood sample.

Using magnetic separation, the researchers pulled out the microbial particles and then used 50 µm disksfunctionalized with probes that identify and fluorescently label individual bacterial species based on snippets of their DNA. Finally, they designed an assay consisting of a chip with a 96-well plate on which they could test the bacteria’s response to different antibiotics at different concentrations. They used an image-processing algorithm to examine the wells and determine the most effective treatment.

In a clinical study of 190 patients thought to have an infection, the system correctly identified the bacterial species involved. When they retrospectively obtained bacterial strains from six patients with confirmed sepsis, their platform shaved about 48 h off the process of identifying the right antimicrobial therapy.

“The interesting thing about this paper is it’s actually not about one particular discovery but rather about creating a pathway to do rapid sepsis testing,” says Lara Mahal, a chemist and glycomics expert at the University of Alberta who was not involved in the work. “It’s a lot of really cool bits of technology put together in a smart way.”

The system still needs some engineering improvements, including improving the yield of microbes captured by the peptide-covered nanoparticles. But on a practical level, Kim says, what would make the platform most effective in a hospital setting is creating a fully automated version of the system—one in which a staff member pops a blood sample into a machine and then simply waits for the result. Few hospitals have a 24 h lab, and for people who arrive during off hours, “that’s another huge delay,” he says.