NEW YORK (GenomeWeb) – A team led by researchers at the University of Tübingen in Germany has used whole-genome sequencing and proteomics to detect prognostic biomarkers derived from a pathogen that causes bloodstream infection (BSI). The team hopes that the biomarkers can help clinicians stratify patients at risk of death from the infection in a hospital setting.
In the study, described in a BioRxiv preprint, Matthias Willmann, lead author and assistant professor at the Institute of Medical Microbiology and Hygiene at the University of Tübingen, and his team discovered four pathogen-derived prognostic biomarker candidates by studying genetic characteristics and proteins of Pseudomonas aeruginosa strains from patients with BSI. The team announced the study's results last month at the European Congress of Clinical Microbiology and Infectious Diseases in Madrid.
In terms of biomarker research, Willmann explained, researchers typically focus on patient-derived biomarkers instead of biomarkers coming from pathogens. Rather than examining the patients' genomes, Willmann's team chose to use a multi-omic approach to screen P. aeruginosa strains from patients in order to identify markers that are potentially linked to patient morbidity.
The team collected 166 blood samples from critically ill patients for genetic and proteomic analysis across multiple hospitals over several years. They used three sets of blood samples for each part of the multi-omics approach.
After sample collection, the team first sequenced the genomes of all P. aeruginosa isolates, containing a pangenome of over 23,000 genes. They found that recombination events only slightly contributed to shaping the diversity of the strains collected from patients. Closely related isolates were commonly found in only one hospital in a narrow time frame, indicating a spatial and temporal clustering.
In terms of genomics features, the team found a wide range of antibiotic susceptibility in the P. aeruginosa strains. Statistical analysis also revealed that the isolates could be divided into four accessory genome clusters. The team saw that one cluster, accessory cluster 2 (acc-cluster 2), was independently linked to 30-day mortality, indicating the presence of genetic factors that lowered the patient survival rate.
Willmann's group then used quantitative proteomic analysis through mass spectrometry to define the isolates' cellular proteome. While identifying 7,757 unique proteins, the analysis did not demonstrate clustering linked to the survival status of patients. In addition, the team did not observe any distinct patterns linked to an acc-cluster or a strong relationship in a cross-comparison of protein and acc-clusters.
The team then applied multilevel Cox regression analysis of single genomic and protein level factors to investigate the relationship between patient characteristics and pathogen features — 2,298 accessory genes, 1,078 core protein levels, and 107 parsimony-informative variations in reported virulence factors — and a 30-day mortality rate.
They identified four pathogen-derived predictors — one genomic and three proteomic — linked to patient mortality. Three of the four pathogen-derived predictors — helP, Prot7, Prot214 — increased the risk of death, while Prot330 had an anti-virulence effect.
Importantly, the researchers found that helP, which encodes a helicase enzyme that unwinds RNA structures to influence translation, doubled the mortality rate in sepsis patients. While the team saw that helP appeared in 22 of the study's strains, it occurred much more frequently in the high-risk acc-cluster 2 strains. Because helP appeared in multiple distinct strains, Willmann hypothesized, it might have been spread by horizontal gene transfer across multiple strains.
Elevated levels of Prot7 (the bacterial flagellum protein FliL) and Prot214 (a bacterioferritin-like protein) also increased the risk of death in patients. While FLiL may serve as a future target, the researchers noted that Prot214's role as a risk factor for a patient's fatal outcome "remains elusive."
According to Willmann, the entire multi-omics process — from blood draw to statistical analysis — took around two months. He noted that his team found clinical sample collection from critically ill patients the most challenging element of the study.
Although the researchers immediately preserved P. aeruginosa strains after detection and established protein levels in the first subculture after thawing, Willmann acknowledged that his team is unsure as to how accurately the protein level profiles mimic pathogen protein levels in a patient's bloodstream.
The researchers therefore focused on helP, a genomic biomarker, due to its stability under different pre-analytical conditions. In addition, due to helP's relevant predictive power and simple detection through PCR, the group believes that the helP genotype is the most attractive biomarker to stratify patient risk in clinical situations.
Robert Hancock, a professor of microbiology and immunology at the University of British Columbia, who was not affiliated with the study, highlighted that "the modest [mortality] ratio means that successful identification of the gene would only tell [a doctor that] a patient was twice as likely to die and not how to treat them." Because of how rapidly Pseudomonas infections can spread, as well as the disease's ability to resist treatment "due to intrinsic, acquired, and adaptive antibiotic resistance, there may be little chance to demonstrate how diagnosis can improve treatment," he said.
Hancock also noted that one of the study's major issues is that the team did not discover any clear mechanistic link between the presence of helP and the increase in fatalities. While he agreed that helP is a potential biomarker, he said that the team will need to uncover the relationship over time in future studies.
However, Hancock believes the researchers' combined multi-omics approach is the "way of the future," as the technique "enables the use of multivariate statistics, which generate new knowledge about biomarkers that is not always evident in any individual dataset."
While Willmann and his team have not applied for a patent regarding using helP as a biomarker, he explained that they will push for commercial development in the near future.
"If helP holds as a stable biomarker, we'd like to talk to commercial companies and try to incorporate it in their panels that [are] searching for P. aeruginosa in sepsis," he said.
In future studies, the authors noted, they will "need to validate the biomarkers found in the study and push the pathogen-derived prognostic biomarkers into the clinics."
Willmann said he envisions helP used as a tool for infection control, since researchers will have established the pathogenic potential for the strain.