MIT’s New Nanoparticle Sensor Can Distinguish Between Viral and Bacterial Pneumonia

Doctors might be able to utilize this test to avoid providing antibiotics in circumstances when they won't work.

Pneumonia may be caused by a variety of bacteria and viruses, but it's difficult to pinpoint which organism is to blame for a patient's disease. Because the drugs frequently used to treat bacterial pneumonia won't assist individuals with viral pneumonia, this ambiguity makes it more difficult for clinicians to identify appropriate therapies. Additionally, reducing antibiotic usage is a critical step in combating antibiotic resistance.

Researchers at MIT have developed a sensor that can distinguish between viral and bacterial pneumonia infections, which they believe may aid clinicians in selecting the best treatment option.

“The challenge is that there are a lot of different pathogens that can lead to different kinds of pneumonia, and even with the most extensive and advanced testing, the specific pathogen causing someone’s disease can’t be identified in about half of patients. And if you treat a viral pneumonia with antibiotics, then you could be contributing to antibiotic resistance, which is a big problem, and the patient won’t get better,” says Sangeeta Bhatia, the John and Dorothy Wilson Professor of Health Sciences and Technology and of Electrical Engineering and Computer Science at MIT and a member of MIT’s Koch Institute for Integrative Cancer Research and Institute for Medical Engineering and Science.

The researchers demonstrated that their sensors could reliably differentiate bacterial and viral pneumonia in mice within two hours using a simple urine test.

Signatures of infection

Because there are so many germs that may cause pneumonia, including bacteria like Streptococcus pneumoniae and Haemophilus influenzae, as well as viruses like influenza and respiratory syncytial virus, it's been difficult to distinguish between viral and bacterial pneumonia (RSV).

The researchers opted to focus on assessing the host's response to infection rather than trying to identify the virus itself when building their sensor. Infections caused by viruses and bacteria elicit several immunological responses, including the activation of proteases, which break down proteins. The MIT researchers discovered that the pattern of activity of those enzymes may be used as a bacterial or viral infection hallmark.

More than 500 proteases are encoded in the human genome, with many of them being employed by cells that respond to infection, such as T cells, neutrophils, and natural killer (NK) cells. Purvesh Khatri, an associate professor of medicine and biomedical data science at Stanford University and one of the paper's authors, headed a group that gathered 33 publicly accessible datasets of genes expressed during respiratory infections. Khatri was able to discover 39 proteases that appear to respond differently to different forms of infection after evaluating the data.

The data was then used by Bhatia and her students to construct 20 distinct sensors that can interact with the proteases. Nanoparticles coated with peptides that can be cleaved by certain proteases make up the sensors. A reporter molecule is attached to each peptide, which is released when the peptides are broken by proteases that are increased during infection. Eventually, those reporters are eliminated in the urine. After that, mass spectrometry may be used to determine which proteases are most active in the lungs.

The researchers put their sensors to the test in five distinct mouse pneumonia models generated by Streptococcus pneumoniae, Klebsiella pneumoniae, Haemophilus influenzae, influenza virus, and pneumonia virus infections.

The researchers analyzed the data using machine learning after reviewing the pee test results. Based on just 20 sensors, scientists were able to develop algorithms that could discriminate between pneumonia and healthy controls, as well as determine if an infection was viral or bacterial.

The researchers also discovered that their sensors could discriminate between the five pathogens they examined, but with lesser accuracy than the test for viruses and bacteria. One avenue the researchers may follow is building algorithms that can not only discriminate between bacterial and viral illnesses, but also identify the class of germs that are causing a bacterial infection, allowing clinicians to select the appropriate medication to battle that type of bacteria.

The urine-based readout might potentially be detected using a paper strip in the future, similar to a pregnancy test, allowing for point-of-care diagnosis. In order to accomplish this, the researchers chose a subset of five sensors that may bring at-home testing closer to reality. Greater research is needed to see if the reduced panel will work in people, who have more genetic and clinical heterogeneity than mice.

Patterns of response

The researchers discovered certain patterns in the host's reaction to various forms of infection as part of their investigation. Because neutrophils respond more to bacterial infections than viral infections, proteases released by neutrophils were more prevalent in mice with bacterial infections.

T cells and NK cells, on the other hand, were activated by viral infections and produced protease activity. A protease called granzyme B, which causes programmed cell death, was related to one of the sensors that produced the greatest signal. The researchers discovered that this sensor was strongly active in the lungs of mice with viral infections, and that the response involved both NK and T cells.

The sensors were injected directly into the trachea in mice, but the researchers are currently working on human equivalents that may be delivered via a nebulizer or an inhaler similar to an asthma inhaler. They're also developing a method to detect the findings using a breathalyzer rather than a urine test, which might provide results even faster.