The prediction of septic shock outcomes and the selection of early optimal treatment may become more precise, thanks to a new biomarker-based tool.
The investigative Pediatric Sepsis Biomarker Risk Model (PERSEVERE) assay is a rapid blood test that measures five biomarkers involved in sepsis and accurately stratifies patients as being at low, medium, or high risk for septic death. The test was developed by Hector R. Wong, MD, director of critical care medicine at Cincinnati Children’s Hospital Medical Center in Ohio, and colleagues over more than 10 years.
In a prospective cohort of 461 children with widely differing risk levels who were receiving intensive care for sepsis, the blood panel reliably predicted who would develop severe sepsis and accurately distinguished between pediatric survivors and nonsurvivors of sepsis at 28 days. The cohort was enrolled from a number of pediatric care centers across the country. The overall mortality rate was 12.6%.
Although the original version of PERSEVERE performed only modestly in predicting 28-day mortality, with an area under the receiver operating curve (AUROC) of .67 (95% confidence interval [CI], .58 – .75), an updated version, PERSEVERE II, had a much greater AUROC of .83 (95% CI, .77 – .88) in differentiating between survivors and nonsurvivors.
PERSEVERE II even outperformed the Pediatric Risk of Mortality (PRISM) III score of .74 (95% CI, .66 – .81; P = .03).
“This approach is not about diagnosis, who does or doesn’t have sepsis. It’s about asking among those with sepsis who’s at risk for poor outcome, and we were impressed how well the model performed,” Wong told Medscape Medical News. He and his colleagues hold patents on the approach and work with several related companies. Their findings were published online November 13 in Science Translational Medicine.
The investigators also found that blood bacterial loads were higher in children who were at greater risk of dying. That finding echoes the group’s previous results in mice, which showed that a higher-dose antibiotic rather than a high-dose anti-inflammatory was able to control the infections. Together, the observations indicate that a greater bacterial burden rather than excessive inflammation is the main pathologic impetus of sepsis, the authors note.
“The bedside-to-bench approach used in the current investigation is interesting and supports previous clinically derived results,” Jerry J. Zimmerman, MD, PhD, told Medscape Medical News. “PERSEVERE appears to add sensitivity and specificity over other more clinically based methodologies to predict outcomes from pediatric sepsis,” said Zimmerman, a critical care pediatrician at Seattle Children’s Hospital in Washington, who was not involved in the study.
“This is very important area of research,” added Roberto Posada, MD, an associate professor of pediatrics in infectious disease at the Icahn School of Medicine at Mount Sinai in New York City, who was also not involved in the study. “The study data are very interesting, but they need to be replicated in larger cohorts.”
Biological Underpinnings Needed for New Treatments
Septic shock is a major cause of morbidity and mortality worldwide and is a leading cause of largely unpreventable in-hospital death in the United States. Years of research and development have yielded no major breakthroughs in therapy or precision-medicine approaches.
“Treatment for septic shock has remained essentially unchanged for decades, relying primarily on care bundles, protocols, antibiotics, and intensive care unit–based organ support,” Wong and associates write. “An important barrier for developing new therapies for septic shock is lack of a unifying, targetable biological mechanism applicable to all patients with septic shock.”
They anticipate that the PERSEVERE system may ultimately provide such a target by shedding light on processes and proteins involved in sepsis initiation and progression. “It’s known that sepsis initiates via infection. The data suggest that a combination of high bacterial burden and excessive inflammation, perhaps orchestrated by some of the PERSEVERE biomarkers, contribute to the pathway(s) to being fulminant,” Wong said.
The blood panel’s five key protein markers derive from “agnostic, discovery-oriented transcriptomic” studies, aided by machine learning, which identified gene expression patterns associated with death from pediatric septic shock, Wong explained.
Starting with about 80 candidate genes, the researchers reduced the list to 12 on the basis of two a priori criteria ― the likelihood of linking the operative gene to sepsis biology, and the ability to measure the gene’s protein product in the serum compartment. These 12 were further reduced to five using classification and regression tree modeling to formulate a model that estimates the risk for 28-day mortality in children with septic shock.
The markers ultimately used in the panel are all involved in sepsis biology and include C-C chemokine ligand 3, interleukin 8, heat shock protein 70 kDa 1B, granzyme B, and matrix metallopeptidase 8.
For the study, the PERSEVERE-based risk assignment occurred within the first 24 hours of diagnosis of septic shock. Patients were assigned to “nodes” that indicated mortality risk derived from PERSEVERE II and that were based on the presence of each of the five biomarkers measured in picograms per milliliter and on platelets measured at number per microliter. Probability of death was dichotomized into those patients who were predicted to survive and those who were predicted not to survive by 28 days.
The five marker proteins also accurately identified low and high risk in mice with sepsis.
Wong anticipates the rapid multiplex assay panel will be in clinical use in 2 or 3 years, thanks to biomedical engineering, and he is working with other groups toward that end. “The technology exists, and it’s not fancy technology,” he said. Furthermore, data from his previous research indicate that biomarker-based risk stratification could be equally effective in adults.
Clinically, identifying those with a higher bacterial burden could mean selecting appropriate patients in a tailored way to receive higher-dose antibiotics.
“It would allow more targeted and earlier intervention for those who need it and avoid risking the side effects of unnecessary aggressive treatment in those who don’t need it,” Posada said. He noted that some treatments can cause kidney and liver injury.
Wong goes further. “This could potentially change how we think about antibiotic dosing for sepsis, which is currently done according to population data estimates,” he said. The problem with the current approach is that antibiotic pharmacokinetics can be substantially altered in critically ill sepsis patients, so standard dosing regimens, even if weight based, may be inadequate for patients with higher pathogen burden.
But the really exciting thing, Wong added, is that murine research suggests these five proteins could be not just prognostic markers but also therapeutic targets. “There might be antibodies that could neutralize them or small molecules that could block receptors for them,” he said. “In this way, PERSEVERE can move beyond prognostic enrichment toward predictive enrichment.”
PESEVERE and pediatric sepsis endotyping will be validated in a large cohort of children with septic shock as the third specific aim of the Stress Hydrocortisone In Pediatric Septic Shock interventional trial, noted Zimmerman, sponsor of the trial. “Identification of which patients may benefit from and which patients may be harmed from adjunctive hydrocortisone sepsis therapy facilitates precision medicine and addresses a key question in clinical medicine that has been debated for decades.”
The study was supported by the National Institutes of Health and Cincinnati Children’s Hospital Medicine. The Children’s Research Foundation, Wong, and coauthor Caldwell are co-holders of US patents for the PERSEVERE biomarkers and other risk-stratifying models for septic shock in children and adults. Wong serves on the advisory boards for Inflammatix, Endpoint Health, and Eccrine Systems Inc. The other authors have disclosed no relevant financial relationships.
Sci Transl Med. Published online November 13, 2019. Abstract