The Battle for a Better Sepsis Test

Biomarkers and predictive analytics are combining forces for better diagnostics
Laboratory medicine professionals know all too well that sepsis not only is a major problem but also that there’s no foolproof way to test for it. Current sepsis testing is “less than ideal,” said T. Scott Isbell, PhD, DABCC, associate professor of pathology at Saint Louis University School of Medicine in Saint Louis. “We don’t have a single biomarker that we can rely on to be able to say ‘if we measure this, it’s sepsis. If we detect it and it’s above this amount, it’s sepsis,’” he said. “That’s related to the fact that sepsis is a very complex, heterogeneous syndromic type of problem. It’s been very difficult for us to find one thing to latch onto.”
Some patients clearly are either at risk for sepsis or already septic, so there’s little guesswork in how to treat them. “For someone coming in the [emergency department] who’s obviously very sick, that’s easy: broad spectrum antibiotics, get the patient in the [intensive care unit], and ask questions later,” said Tim Sweeney, MD, PhD, co-founder and CEO of Inflammatix, a molecular diagnostics company developing tests for sepsis. “But for the vast majority of patients, it’s not obvious. There’s a big gap in what clinicians need and the tools that they’re offered, especially in diagnostics.”
Moreover, there is no one type of septic patient, Isbell added. “It can range from a young child with a small abrasion on a leg that turns into sepsis to a 90-year-old person in a nursing home with a urinary tract infection whom we fail to recognize is septic because she also has dementia,” he said.
Even as health systems and clinical laboratories grapple with how best to deploy existing assays like lactate and procalcitonin, academic researchers and industry entrepreneurs remain engaged in a fierce competition to find the biomarker, combination of biomarkers, or perfectly tuned statistical algorithm that can reveal the actionable insights clinicians crave.
One such company exploring the use of a novel biomarker is Swiss medtech company Abionic, which developed abioSCOPE, a nanofluidic immunoassay technology that measures for pancreatic stone protein (PSP). “PSP is the only marker to identify sepsis from noninfection inflammation 24 hours before the current methods, giving physicians a clear signal when they start the antibiotic therapy,” said Abionic CEO Nicolas Durand, PhD.
Abionic first demonstrated the kinetics of this marker by studying patients with severe burns. The company started with these patients, Durand said, because they come into hospitals without an infection, and their chances for developing one and therefore sepsis are high. Abionic found that PSP “was rising fast, much earlier than any other existing sepsis marker,” he said.
Abionic has tested abioSCOPE in 14 hospitals in Europe on several hundred patients and confirmed that in more than 85% of cases, practitioners could have diagnosed sepsis more than 24 hours earlier than the standard of care. The company is currently in its first U.S. clinical trial to validate those results for the Food and Drug Administration (FDA), working in seven hospitals. Abionic plans to market the first PSP tests in Europe in the second quarter of 2020.
Another road to a better test might be a precise combination of markers. Inflammatix—winner of AACC’s Disruptive Technology Award at the 2019 AACC Annual Scientific Meeting—is using machine learning algorithms that look at the expression of multiple messenger RNAs in the blood that reflect the immune system’s response to infection. Sweeney said that this process identifies the presence of a bacterial or viral infection and determines if a patient has or is likely to develop sepsis. The test is designed to give point-of-care results in 30 minutes or less.
“We go ask the immune system,” he said. “If we can figure out what the immune system is reacting to, then we know how to treat the patient. We think this will be the first product to say: First, let’s figure out if you have an infection, and, second, let’s figure out if you have sepsis.”
In January Inflammatix secured $32 million in funding to power its commercial launches in Europe and regulatory submissions for FDA. The new financing follows a November 2019 contract worth up to $72 million with the U.S. Biomedical Advanced Research and Development Authority to develop its tests for acute infections and sepsis. Sweeney hopes to submit Inflammatix’s HostDx sepsis test to FDA in 2021 and launch in Europe the same year.
In a third example aimed at the point of care, U.K.-based QuantuMDx and California-based Ontera are working together on a solution that they believe can determine, in one system, whether a patient has a bacterial or viral infection and then whether or not the organism is antibiotic resistant.
QuantuMDx has developed a cassette-based device that separates and concentrates pathogen cells from a sample using electronic fields, while Ontera is known for its nanopore biosensor measurement system. “The combined system is a panel that allows [the operator] to do both in that first hour” rather than waiting 24 hours for a culture, said Ontera CEO Murielle Thinard McLane. She expects the test system to be on the market at the end of 2022.
Even as promising research emerges on new biomarkers and instruments, health systems are also moving forward with big data approaches that aim to interpret existing data using machine learning. These data points include laboratory values as well as traditional vital signs and comorbidities in patient records.
In May, HCA Healthcare, which has 185 hospitals and 2,000 sites of care in 21 states and the U.K., announced that it had developed an algorithm driven, real-time system called Sepsis Prediction and Optimization of Therapy (SPOT) Technology. According to HCA, in conjunction with the use of evidence-based clinical interventions SPOT has helped save about 8,000 lives in the last 5 years.
Other health systems have been evaluating a predictive analytics approach as well, taking advantage of new algorithms offered by electronic medical record (EMR) companies. At Saint Louis University Hospital, Isbell and the hospital’s sepsis committee implemented one such system available from Epic. The algorithm runs continuously in the background to check patient records every 15 minutes and sends clinicians an alert via the EMR if the algorithm’s score predicts a patient is at risk of developing sepsis. In addition to demographics, vital signs, and comorbidities, the algorithm also calculates the scores using hematologic parameters from the clinical laboratory, as well as creatinine, HbA1c, procalcitonin, and other results.
Ultimately, experts expect the war on sepsis to continue on multiple fronts: new biomarkers, predictive analytics, and insight into immune system response. “The new tests focused on the immune response along with machine learning approaches are promising and move us toward the ultimate goal—prediction and prevention of sepsis,” Isbell commented.