Moving on to the ICU setting, Dr. Farnsworth turned next to the SeptiCyte Rapid, a gene-expression–based method that runs on the Idylla platform.
The test detects RNA transcripts PLA2G7 and PLAC8, which are combined to produce a score that falls into four interpretation bands, with higher scores indicating higher probabilities of sepsis. “It gives you a number, which can be semiquantitative,” Dr. Farnsworth said. The assay is indicated as an aid in differentiating infection-positive (sepsis) from infection-negative systemic inflammation in patients suspected of sepsis on their first day of ICU admission.
In a clinical validation study of SeptiCyte Rapid, retrospective (n=356) and prospective (n=63) samples were tested from adult patients in the ICU who had systemic inflammatory response syndrome (SIRS) or were suspected of having or were diagnosed with sepsis (Balk R, et al. J Clin Med. 2024;13[5]:1194). Patients were clinically evaluated by a panel of three expert physicians blinded to the SeptiCyte test results.
Under the Sepsis-2 framework, SeptiCyte Rapid performance for the combined retrospective and prospective cohorts had AUCs ranging from 0.82 to 0.85. It had a sensitivity of 94 percent and a negative predictive value of 0.91 for band one. “So for ruling out sepsis, it would do a decent job,” he said. For band four, the specificity is 90 percent and the positive predictive value is 81 percent.
“Based on what band they fall into—again, these are very sick patients in the ICU—what’s the probability they had sepsis?” Dr. Farnsworth asked. “In band one, 10 percent of patients still had sepsis, versus 81 percent [without]. So it does separate them fairly well.” There was, however, a nine percent false-positive rate in band one. “To me, as a laboratorian, that would imply you wouldn’t necessarily want to stop antibiotics in a band one [patient]. The thought then becomes, ‘How do we use this?’” Another potential limitation: Indeterminate cases were excluded. “Nonetheless, it does separate these groups quite well, with good NPV and PPV,” he said.
In a follow-up study using the same data, the authors stratified a subgroup of their cohort into three clinically defined sepsis phenotypes, with 176 patients phenotyped and 16 clinical variables used (Balk R, et al. J Clin Med. 2024;13[20]:6044). The aim was to identify variables that might affect the ability of the assay to discriminate between sepsis and SIRS and to determine if any patient subgroups appeared to present a diagnostic challenge for the test. No significant differences in performance were observed between most groups and subgroups, though assay performance was enhanced in a phenotypic subgroup defined by a combination of clinical variables suggesting a septic shock response.
They found that SeptiCyte Rapid performs similarly across sepsis subtypes. Assay performance for discriminating sepsis from SIRS never fell below AUC 0.81 for any subgroup. It was 0.93 for the subgroup most likely to have septic shock (of the 15 patients in the subgroup, 11 were clinically diagnosed with septic shock and three with severe sepsis based on the Sepsis-2 definition, with no viral infections). “So it does perform best in the inflammatory group, but that’s also the group most likely to be sickest,” Dr. Farnsworth said. No significant differences were seen between the sepsis subgroups with respect to infection site or type of infecting pathogen.
SeptiCyte Rapid has high sensitivity and specificity, compares well across sepsis phenotypes, and provides sample-to-answer processing in an hour. Its limitation: If patients are sick enough to be admitted to the ICU, they may already be on antibiotics. In addition, the ICU indication raises questions, Dr. Farnsworth said. For instance, “As laboratorians, if we were to use this, would we have to consider validating it for an ED population?” And is an NPV of 91 percent high enough to withhold antibiotics? “You may need higher sensitivity in low-risk bands to stop antibiotics,” he said.
The TriVerity transcriptomics assay from Inflammatix was the last of the assays Dr. Farnsworth spoke about.
It uses isothermal amplification of 29 mRNAs and machine learning algorithms on the Myrna instrument to determine the likelihoods of bacterial infection, viral infection, and need for critical care interventions within seven days.
The assay is similar to the MeMed BV in that it differentiates bacterial from viral infection, “except you get two separate scores, one for each,” he said (because of the possibility of co-infection), as well as a score for severity. Each score ranges from zero to 50 and is further separated into five interpretation bands (very low, low, moderate, high, and very high) that reflect increasing likelihoods of the corresponding infection type or severity.