NEW YORK (GenomeWeb) – Researchers in Germany and Austria have demonstrated the feasibility of identifying organisms responsible for bloodstream infections by sequencing circulating cell-free nucleic acids in patient blood samples.
"Our motivation, really, is to get a handle on the identity of the infectious organisms that are causing sepsis," Kai Sohn, a researcher affiliated with the Fraunhofer Institute for Interfacial Engineering and Biotechnology in Stuttgart and the University of Stuttgart, told GenomeWeb. "The infectious organisms [leave] evidence as genetic traces in the blood."
In a paper appearing in Genome Medicine last week, Sohn and his colleagues outlined results from a pilot study of this circulating cell-free microbial nucleic acid sequencing approach for narrowing in on suspicious bacteria, viruses, fungi, or other pathogens in blood samples from seven individuals with sepsis. Indeed, they found that it was possible to tease out microbial DNA from patient blood samples, identifying sepsis-causing microbes and, in some cases, narrowing in on antibiotic resistance markers in those infectious agents.
With the growing recognition that rapid antibiotic treatment can curb morbidity in individuals with bloodstream infections that have progressed to sepsis, several teams are pursuing methods to diagnose infections and identify those involving organisms that might respond to antibiotics or other treatments.
While some groups are pursuing infection diagnoses based on gene expression patterns in host blood, Sohn and his team are focusing their attention on pinpointing organisms behind such infections, reasoning that the identity of those microbes and their potential virulence or resistance features will also impact the effectiveness of antimicrobial treatment used to tackle the infection.
Although both classical culture-based microbiological tests and molecular diagnostic tests are available to ascertain pathogens behind a given infection, Sohn argued that the "state-of-the-art technologies are somewhat limited … with respect to reliability and also with respect to detection rates, false positives, and false negatives."
In the case of culture-based diagnostic tests, for example, contamination may occur during the blood draw, leading to false positives. On the other hand, false negatives may be a problem, particularly when dealing with rare infection events.
To avoid potential biases associated with PCR-based searches for specific microbial sequences, meanwhile, the team aimed for an approach that would pick up as many different microbial culprits as possible from a single blood sample.
"We are not focused or targeted on some pre-defined sets of organisms," Sohn said. "We are open to bacteria, viruses, fungi, parasites, or whatever."
"You're using any information which is present in the blood or the plasma, irrespective of whether it's from mitochondrial genes or whatever," he explained. "Every single fragment which can be unambiguously defined and assigned to microbiological databases is of use for the detection and for the diagnosis."
The team designed the analytical arm of the sequencing pipeline to determine whether or not an infection exists, on the one hand, while using a sepsis-indicating quantifier score to quantitatively assess nucleic acid levels and weed out potential contaminants or harmless commensal organisms.
For their pilot study of the cell-free DNA-based sequencing approach to diagnosing bloodstream infections, the researchers tested blood samples from seven patients enrolled in a Heidelberg University Hospital observational clinical study who had septic infections confirmed using classical microbiological data, along with six surgical patients, and a dozen healthy volunteers.
The team used Illumina HiSeq 2500 sequencing instruments for its quantitative diagnostic sequencing experiments, owing to the technology's high throughput and sequencing depth.
Based on some 25 to 30 million sequence reads per sample, the researchers saw a slight rise in cell-free DNA concentrations in both the surgery and sepsis patients compared to the healthy volunteer controls, with human DNA reads representing roughly 96 percent of sequence reads in each case. There was also an uptick in the number of non-human reads they could classify, on average, in samples taken around the time of sepsis onset in those patients.
In patient samples collected periodically for four weeks, the team was able to trace microbial dynamics — including those relevant to patient infections — with the quantification score.
In samples taken over time from a patient who developed pneumonia and became septic after a stomach surgery, for example, the group consistently detected Staphylococcus aureus DNA, while nucleic acids stemming from the herpes simplex virus 1 rose at the one-week point.
Meanwhile, when they focused in on Enterococcus faecium sequences circulating in the blood of a patient diagnosed with sepsis after a liver transplant, the researchers detected markers for resistance to antibiotics such as vancomycin and tetracycline, consistent with the clinical and microbiological data available for that patient.
In principle, Sohn said the same general approach should be applicable to a wide range of infectious organisms, including RNA viruses, though the team has mainly focused on circulating cell-free DNA so far.
And while the method may not be able to identify new or emerging pathogens at the species level, he explained, it can uncover sequences that resemble those found for known organisms in custom or public sequence databases at a genus, family, or other classification level, spurring more detailed, de novo analyses of the microbe in question.
Stanford University's Purvesh Khatri, who was not involved in the study, noted that sequencing-based tests are inherently limited not only by a reliance on similarities to known pathogens, but also by the amount of pathogen DNA available in the bloodstream.
"The bacterial or viral load may be so small that, unless you're sequencing at a very high depth, you may not pick up those sequences, which means you're doing high-depth sequencing, which is going to be expensive," said Khatri, whose own team is pursuing a point-of-care infection diagnostics device based on host gene expression signatures.
Though he noted that the host expression and pathogen sequencing approaches may be complementary ways of rapidly uncovering an infection and then characterizing features of the pathogen involved, Khatri cautioned that sequencing may be excessive for some types of infections, particularly those not caused by bacteria.
Sohn noted that while the sequencing-based approach is a pricier way of identifying microbes than culture-based methods, he expects the cost to decline as such tests become better established, similar to the price trajectory of non-invasive prenatal tests.
He noted that the team's workflow currently takes about 26 hours, which is on par with the time needed for conventional culture-based methods for identifying pathogens but much slower than PCR-based identification methods.
With that in mind, Sohn and his colleagues are pushing to bring the time needed to run the pipeline down to eight hours as soon as this year, if possible. They are currently assessing the speed, accuracy, and throughput of third-generation sequencing platforms for detecting circulating nucleic acids from infectious organisms.
The team is also exploring efficient enrichment techniques for situations where it is advantageous to get deep coverage of the population of infectious organisms — for example, to profile drug resistance or virulence patterns.
Results published so far have relied on sequencing without enrichment for microbial DNA, Sohn explained, which is important when attempting quantitative assessments of microbial sequences.
The team has gone on to automate the nucleic acid isolation, sample preparation, and sequencing steps, coupled with a fixed and optimized bioinformatics workflow. It is now applying that automated pipeline to more than 250 patient blood samples — a set that includes samples testing positive or negative for infection with blood culturing methods.
The researchers are continuing to put the circulating cell-free nucleic acid sequencing method through its paces in that larger study, Sohn said, to get "a more precise idea of the sensitivity, the specificity, and [whether] we think a patient has an infection or not."
From there, the team anticipates undertaking an even larger, multicenter study. It has already filed for a patent related to the diagnostic algorithm used to interpret infection features from circulating microbial DNA sequence data.