A team led by researchers at Cincinnati Children's Hospital Medical Center has developed a proteomic test for stratifying adult sepsis patients according to their mortality risk.
Detailed in a study published this week in Critical Care Medicine, the test, which combines a 12-protein panel with conventional clinical variables, could be used to select subjects for drug trials, benchmark hospital performance, and aid in patient management, Hector Wong, a CCHMC researcher and one of the leaders of the effort, told ProteoMonitor.
He and his colleagues have also developed a sepsis risk stratification test for pediatric patients using the same protein panel and are exploring the protein interleukin-27 as a possible biomarker for diagnosing the disease. With seed investor CincyTech, CCHMC is working to launch a start-up firm, Persepsys Biomedical, based on the sepsis biomarker work.
Sepsis is the tenth leading cause of death in the US, with sepsis infections costing the country's healthcare system an estimated $16.7 billion per year.
The adult panel published this week stems from the group's pediatric risk panel, which it first presented in a 2012 study in Critical Care. That panel was developed using transcriptomic data the group had collected on CCHMC sepsis patients for more than a decade, identifying 117 proteins that appeared potentially useful in predicting sepsis mortality risk. According to Wong, the researchers used transcriptomics as opposed to proteomics for their discovery work due to the greater robustness of transcriptomic tools compared to proteomic approaches when they began the work ten years ago.
Taking the list of 117 candidates, the researchers selected a subset of analytes based on "what was biologically plausible and what we could measure as a protein in the serum compartment" of a patient sample, Wong said. "And based on that we identified about 12 or 15 candidates and developed a protein assay to measure them."
In their development of an adult panel, the researchers used the same markers but applied different cutoffs and models. While there are differences between pediatric and adult sepsis, Wong said that he and his colleagues decided to "take a chance" applying the child data to adult patients, "and to our satisfaction, it worked very well."
In a validation cohort of 209 sepsis patients, the panel predicted which patients would die within 28 days with a sensitivity of 85 percent, specificity of 60 percent, positive predictive value or 61 percent, and negative predictive value of 85 percent.
The authors noted in the paper that, given the test's high negative predictive value, it could prove most useful as a rule-out tool for determining patients who are not at high risk of mortality. The researchers also developed a classification tree-based approach to using the markers in which the protein and clinical measures are used to place patients in one of 12 classification nodes – six high-risk and six low-risk, but each with its own specific mortality probability.
"When you look at positive predictive value, negative predictive value, sensitivity, you're looking at [the model] in a dichotomous manner," he said. "But the way I like to look at it is that each of the nodes assigns a different mortality probability, so it's a model for assigning a range of mortality probabilities."
Wong noted as well that if applied in a dichotomous way, the test likely has better positive predictive value than is apparent based on the study due to false positives arising from clinical intervention.
"You have a substantial number of false positives, patients who are predicted to have a high rate of mortality but actually survive, and I would argue that that's to be expected," he said. "Because if you believe what we do in the ICU actually makes a difference, then you should have" patients initially predicted to die who, in fact, survive.
He added that the researchers tried to tease out this effect in their pediatric patients and found that compared to true negatives, false positive patients typically had much more severe sepsis, with longer ICU stays and higher organ failure burdens, suggesting that they were, in fact, initially at higher risk of dying.
Wong said he thought that one of the primary applications for the test would be in selecting patients for clinical trials, where "the very heterogeneous cohorts with very wide mortality probabilities" currently used have presented significant challenges in development of therapies for sepsis.
Because trials often include both patients at low risk who are unlikely to benefit from treatment beyond standard of care and extremely high-risk patients who are unlikely to be helped by any treatment, benefits gained by patients with "significant, but modifiable, mortality risk" are often missed, leading, consequently, "to a negative clinical trial," the authors wrote.
Beyond this, Wong said the test could also be useful for benchmarking hospital performance. While data exists on average survival rates for sepsis survivors, it is difficult to use effectively due to the different case mixes in different hospitals, he said.
"Say you have a mortality rate of 40 percent in your ICU," he said. "That's kind of high relative to national averages, but if you are admitting very high-risk patients, then that may be the mortality that you are going to see. But right now there's no way to know that."
Wong said he also expected the panel to help with patient management. "For example, if you are in a relatively small community ICU and you have a very high-risk patient based on this modeling, you may want to choose to transfer that patient to a hospital that has more resources," he said. "Or if you're in a hospital that has more than one ICU at different levels of care you may decide to triage those patients to different ICUs."
In addition to Wong and his CCHMC colleagues, several other outfits are working on protein marker-based sepsis tests, including risk stratification assays. For instance, a team led by researchers at University of Missouri-Kansas City School of Medicine has filed patents on a metabolomics-based assay that distinguished between sepsis survivors and non-survivors with area under the curve of 0.74 in a study of 86 sepsis patients published last year in Science Translational Medicine.
On the diagnostic front, Thermo Fisher Scientific offers its procalcitonin assay for the diagnosis of the condition. Protein biomarker firms Pronota and Astute Medical also have proteomic tests for sepsis in their development pipelines.
Nucleic acid-based approaches are also under development, with firms like Cepheid and BioMérieux, for instance, working on PCR-based tests for identifying the infectious agents responsible for the disease.