Turning to gene-based data for sepsis diagnosis
Inflammatix shares findings in Nature Communications supporting a bioinformatics approach that measures immune system response to better predict and diagnose sepsis
Register for free to listen to this article
Listen with Speechify
0:00
5:00
BURLINGAME, Calif.—Infection is a leading and ever-present complication for wounds if not properly seen to, and infections that evolve into sepsis are particularly worrisome. The Mayo Clinic defines sepsis as “a potentially life-threatening complication of an infection. Sepsis occurs when chemicals released into the bloodstream to fight the infection trigger inflammatory responses throughout the body. This inflammation can trigger a cascade of changes that can damage multiple organ systems, causing them to fail. If sepsis progresses to septic shock, blood pressure drops dramatically, which may lead to death.” More than 250,000 people die of sepsis in the United States every year, and healthcare costs attributed to sepsis are estimated at more than $27 billion every year.
In an effort to address this issue, Inflammatix is advancing its HostDx Sepsis molecular test as a new diagnostic option, and recently published a study in Nature Communications detailing how a gene expression model could help track immune system response in the presence of infection.
As noted in the study, “Two consensus papers suggest that continued failure of proposed sepsis therapies is due to substantial patient heterogeneity in the sepsis syndrome and a lack of tools to accurately categorize sepsis at the molecular level. Current tools for risk stratification include clinical severity scores such as APACHE or SOFA as well as blood lactate levels. While these measures assess overall illness severity, they do not adequately quantify the patient’s dysregulated response to the infection and therefore fail to achieve the personalization necessary to improve sepsis care. Some peptide markers of sepsis severity have been validated (e.g. proadrenomedullin among others), but these are not yet cleared for clinical use.”
“Current tools for sepsis identification and triage are imprecise, which often results in patients being overtreated or undertreated and wastes significant healthcare resources. The findings in this new study suggest that measuring immune dysregulation could indicate infection severity and significantly improve sepsis diagnosis,” said Dr. Tim Sweeney, lead author of the new paper and cofounder and CEO of Inflammatix. “The technology used in this study forms the basis of our HostDx Sepsis test, which we plan to bring into hospitals and urgent care settings as a rapid test to help reduce the global burden of sepsis.”
The study used gene-based models designed to predict 30-day mortality in patients with sepsis. These models were developed and evaluated on more than 20 cohorts from clinical studies with a variety of populations, totaling more than 1,000 patients with community- or hospital-acquired sepsis. The researchers used a “community approach” in this work, in that three teams—Duke University, Sage Bionetworks and Stanford University—“performed separate analyses using the same input data.” (The exclusive license to a pending patent from Stanford University for the gene expression model used in this study is held by Inflammatix.)
The combination of the Stanford gene set with clinical severity scores, the current standard of care, resulted in a substantial increase in prognostic power for 30-day mortality—from 77 percent to 87 percent. This could enable clinicians to rule out roughly 20 percent more sepsis cases, compared to clinical severity scores alone. The study saw “summary AUROCs around 0.85 for predicting 30-day mortality” from its four models, and the authors noted that “The derived discriminatory power of the gene models (AUCs near 0.85) is at least similar to the AUC of proadrenomedullin (0.83) in a recent large prospective trial (TRIAGE study).” Proadrenomedullin has been widely explored as a biomarker for infections/sepsis.
“Prediction of outcomes up to 30 days after the time of sampling represents a difficult task, given that the models must account for all interventions that occur as part of the disease course,” the authors noted in the Nature Communications paper. “An accuracy of 100 percent is likely not only unachievable but also undesirable, as it would suggest that mortality is pre-determined and independent of clinical care. Given this background, and since similar prognostic power was observed across all individual models and the ensemble model, our prognostic accuracy may represent an upper bound on transcriptomic-based prediction of sepsis outcomes. In addition, since prognostic accuracy was retained across broad clinical phenotypes (children and adults, with bacterial and viral sepsis, with community-acquired and HAIs, from multiple institutions around the world) the models appear to have successfully incorporated the broad clinical heterogeneity of sepsis.”
“This new research, combined with previously published data, demonstrate the HostDx Sepsis test’s ability to identify the presence of bacterial and/or viral infection and determine the likelihood of a patient having or developing sepsis. We believe this powerful combination will strongly drive the economic value of HostDx Sepsis, a point we look forward to validating in interventional trials,” said Sweeney.