In a recent study published in Nature Microbiology, researchers developed integrated host-microbe plasma metagenomics to facilitate sepsis diagnosis.
Sepsis accounts for 20% of all fatalities worldwide and 20% to 50% of hospital deaths in the United States. For timely and effective antibiotic therapy crucial for sepsis survival, initial detection and identification of microbial infections are required. However, no etiologic pathogens are identified in more than 30% of cases. Distinguishing sepsis from non-infectious systemic disorders is essential since they frequently appear clinically similar during hospitalization.
About the study
In the present study, researchers created a sepsis diagnostic tool that combined host transcriptional profiling along with broad-range pathogen identification.
At two tertiary care hospitals, the team conducted a prospective observational examination of critically ill adult patients admitted to the intensive care unit (ICU) from the emergency department (ED). Patients were divided into five subgroups based on the presence or absence of sepsis. These patients included those who had: (1) clinically adjudicated sepsis as well as confirmed bacterial bloodstream infection (SepsisBSI); (2) clinically adjudicated sepsis as well as a confirmed non-bloodstream infection (Sepsisnon-BSI); (3) suspected sepsis characterized with negative clinical microbiological testing (Sepsissuspected); (4) patients having no evidence of sepsis and an explanation for their critical disease (No-sepsis); or (5) patients with an indeterminate status (Indeterm).
By conducting ribonucleic acid (RNA) sequencing on whole blood samples, the team first examined transcriptional variations between patients having clinically and microbiologically proven sepsis and those without symptoms of infection. A technique called gene set enrichment analysis (GSEA) detects clusters of genes within a dataset with related biological functions.
A differential gene expression (DE) study across the SepsisBSI and Sepsisnon-BSI groups was conducted to identify further variations between sepsis patients with infections in the bloodstream versus peripheral sites. The team developed a universal sepsis diagnostic classifier based on whole-blood gene expression patterns in response to the practical requirement to diagnose sepsis in SepsisBSI as well as Sepsisnon-BSI patients. The team utilized a bagged support vector machine (bSVM) learning strategy to choose the genes that most successfully differentiated patients with sepsis (SepsisBSI and Sepsisnon-BSI) and those without sepsis (No-sepsis).
A median of 2.3 × 107 reads was acquired after sequencing the RNA from obtained patients whose plasma specimens were available. Furthermore, DE analysis was performed to determine if a biologically plausible signal could be used to differentiate patients who did and did not have sepsis.
Heart failure exacerbation, overdose/poisoning, cardiac arrest, and pulmonary embolism were the most frequently diagnosed conditions in the No-sepsis group. Irrespective of the subgroup, most patients required vasopressor support and mechanical ventilation. Patients in the SepsisBSI and Sepsisnon-BSI who had proven sepsis did not show any difference from No-sepsis patients with respect to age, sex, race, ethnicity, APACHE III score, immunocompromise, intubation status, maximal white blood cell count, or 28-day mortality. In the group of patients without sepsis, all but one patient demonstrated two or more systemic inflammatory response syndrome (SIRS) criteria.
The study also revealed the downregulation of pathways linked to ribosomal RNA processing and translation along with the upregulation of genes involved in innate immune signaling and neutrophil degranulation in sepsis patients. Using DE analysis, the team found 5,227 genes. The Sepsisnon-BSI cohort displayed enrichment in genes associated with defensins, antimicrobial peptides, and G alpha signaling as well as other pathways. On the other hand, the SepsisBSI cohort showed enrichment in genes associated with immunoregulatory interactions between non-lymphoid and lymphoid cells and CD28 signaling, among other functions.
The bSVM model displayed a mean cross-validation area under the receiver operating characteristic (ROC) curve (AUC) of 0.81. Samples with transcript counts lower than the quality control (QC) threshold had a lower mean input mass than samples with sufficient counts.
Interestingly, a number of differentially expressed genes have been identified as sepsis biomarkers, including increased CD177, repressed human leukocyte antigen – DR isotype (HLA-DRA), indicating a biologically significant transcriptome signature from plasma RNA. In the Sepsisnon-BSI group, plasma deoxyribonucleic acid (DNA) metagenomic next-generation sequencing (mNGS) revealed three out of eight bacterial urinary tract infection (UTI) pathogens and two out of 25 bacterial lower respiratory tract infection (LRTI) pathogens. None of the three patients with severe colitis caused by C. difficile had this pathogen. In eight out of 73 patients with proven sepsis, additional potential bacterial pathogens not identified by culture were found.
Overall, the study findings showed that reliable sepsis diagnosis is facilitated by the combination of host gene expression profiling with metagenomic pathogen identification from plasma nucleic acid. Future research is required to verify and gauge the therapeutic utility of this culture-independent diagnostic strategy.
- Kalantar, K., Neyton, L., Abdelghany, M., Mick, E., Jauregui, A., & Caldera, S. et al. (2022). Integrated host-microbe plasma metagenomics for sepsis diagnosis in a prospective cohort of critically ill adults. Nature Microbiology. doi: 10.1038/s41564-022-01237-2 https://www.nature.com/articles/s41564-022-01237-2