NEW YORK (GenomeWeb) – For bloodstream infections, combining rapid molecular identification of pathogens with antibiotic susceptibility testing (AST) directly from whole blood on a single platform is the "holy grail" to guide appropriate antibiotic use.
A proof-of-concept study has now shown that using a sped-up PCR method, high-resolution melt analysis, and a technique to detect bacteria in their demise after a brief antibiotic treatment, can yield full ID-AST results in around eight hours, with the potential to go even faster.
The study, which was published in Clinical Chemistry earlier this month, was conducted by researchers at Stanford University, Johns Hopkins University, and the University of Utah School of Medicine. It describes the initial results of work under a five-year, $4 million joint NIH grant which they recently received to develop a digital microfluidic ID-AST assay for sepsis.
The method involves removing blood cells from whole blood then performing a brief bacterial enrichment. The sample prep and pre-enrichment step takes about six hours, with the ID and AST steps each taking approximately one hour, according to the study.
The bacterial detection and identification are performed using broad-spectrum qPCR amplification of an internal spacer region between the 16S and 23S regions, followed by high-resolution melt analysis, or HRMA.
The HRMA method was first described in 2003 by the University of Utah's Carl Wittwer, a collaborator on the recent study. Last year, Wittwer and colleagues also published a paper describing the use of microfluidics to dramatically decrease the speed of HRMA.
Stanford's Samuel Yang, corresponding author of the recent study, noted that his team's method is essentially a "sequence fingerprinting approach" that monitors the melting of double-stranded DNA as temperature increases. It uses a saturating amount of a fluorescent dye that binds DNA and this can lead to changes in fluorescence as the DNA melts.
"Because the shape of the melt curve can reflect sequence differences, it can be used as a powerful post-PCR amplicon genotyping technology," Yang said.
The method has advantages in simplicity, assay chemistry, and time to result, Yang said, and can also easily incorporated into fluorescence-based qPCR platforms without the need for additional instrumentation, as compared to, say, sequencing, microarrays, or mass spectrometry for amplicon analysis.
For the PCR step, the researchers also described incorporating Wittwer's rapid-cycle real-time PCR methods, as well as a few of the aspects of his Extreme PCR technique. Namely, they used modestly increased primer and polymerase concentrations, although in an interview Wittwer noted that this was not to produce "extreme" PCR conditions, as the reaction itself takes about 30 minutes, due in part to the fact that the product length is about 625 base pairs. This amount of time, however, an 83 percent reduction from a three-hour run time for a standard qPCR-based method, according to the study.
In the study, the researchers use the LS-32 thermal cycler, a limited production instrument from BioFire Defense that "combines the speed of the carousel technology underlying the Roche LightCycler with the high-resolution DNA melting found in the [BioFire Defense] HR-1," Wittwer said.
The group also ran some amplifications in an "extreme" instrument prototype in about 10 min, he added, but noted that the details were not reported in the current publication. "The perfect machine for extreme PCR does not yet exist, but we're about to submit a manuscript ... that describes a system that gets closer," he noted.
For suceptibilty testing, genetic AST methods typically detect alterations to the bacterial genome that enable a bug to resist being killed by an antibiotic, while phenotypic methods usually involve enriching the bacteria, distributing it in small aliquots, and applying varying concentrations of different antibiotics to find the minimum inhibitory concentration, or MIC, that halts growth of the colony.
Both methods have limitations. Genotypic methods suffer because "the current set of known genetic resistance markers is very limited," Yang said. Bacteria are constantly adapting and evolving new resistance tricks, as well, including harboring resistance markers that don't even convey a resistance phenotype.
Phenotypic methods, on the other hand, are the diagnostic standard but usually require the bacteria to grow in culture, which is constrained by the doubling time of the particular strain. Furthermore, to detect the lack of cell division when the MIC is achieved requires waiting however long it takes the bacteria to reproduce.
The AST method described in the Clinical Chemistry study is what Yang calls "phenotypic molecular AST," using molecular methods to measure the response to antibiotics, rather than looking for genetic resistance markers. The method can also provide MIC, which standard genetic AST cannot do.
Initially, the group used the ribosomal RNA gene for identification, because it confers a lot of genetic information that allows for broad pathogen ID, Yang said. "We are also now using it as a molecular AST target, measuring the change in its copy number to reflect susceptibility," he added.
And, to further accelerate AST without relying on bacterial growth, the group is exploring the use of ribosomal RNA "as a biosynthetic marker to assess cell viability," with the theory being that a cell that is sensitive to antibiotics would decrease biosynthesis, and thus their rRNA levels.
Preliminary data in the Clinical Chemistry study suggests this approach can yield AST results in about five minutes, although Yang said further work is still needed to see if the rRNA method can be a universal AST marker for a broad range of species.
As noted in an accompanying editorial in the journal, the only assay currently approved for rapid ID-AST, the Pheno system from AccelerateDx, takes about 1.5 hours for the ID step and around seven hours for the AST step, and these steps are performed from blood cultures, not directly from blood.
Other researchers and firms are developing systems for rapid, direct ID-AST. For example, researchers led by Rustem Ismagilov at the California Institute of Technology have demonstrated a digital method that uses isothermal nucleic acid amplification and takes less than 30 minutes directly from patient urine samples. Ismagilov is also a co-founder of Talis Biomedical, a firm that recently garnered $10 million in funding to develop the digital assay into an ID-AST test for bloodstream infections, among others.
Yang said he has been collaborating with JHU's Tza-Huei "Jeff" Wang, a co-author on the recent study, for around 15 years. Yang works on assay development, as well as the use of broad-range PCR combined with HRM and machine learning for analytics of melt curves. Meanwhile, Wang's lab focuses on microfluidics instrumentation.
Yang had also been a long-time admirer of Wittwer's work, he said, and happened to meet him at a conference. Wittwer, in turn, said he decided to participate in the project because faster results in terms of ID and antibiotic susceptibility reduce cost, morbidity, and mortality in infectious disease. "Anything that gets results to the physician and patient faster makes a difference," Wittwer said.
Although there are many players now in the sepsis diagnostics market, Yang said that is probably for the best, to continue to advance the science. The goal is to replace the blood culture system with molecular approaches in order to do broad pathogen detection, identification, and susceptibility testing rapidly on the same platform, to enable the earliest possible guidance of targeted antibiotic selection. "I don't think we are there yet, with all the technologies that are out there," he said, suggesting that their proof-of-concept study showing direct-from-blood broad range ID-AST in a single assay may be the first to do so.
"There are technologies that can do direct-from-blood detection and ID, but they have no AST, or they have ID and AST but those are all post-blood culture," he said. "This is an ongoing effort, and recently with advances in digital PCR, and using HRMA as a way to do broad pathogen ID, and the ability to do molecular AST, we're finally able to put everything together into a single platform," Yang said.
Near-term, the group plans to integrate the assay onto the digital PCR platform it is developing to do single-cell analysis. Yang's lab is also working to create a more comprehensive database of melt curves for all clinically relevant pathogens. The group typically uses machine learning for automated classification of unknown curves against a database for large-scale pathogen ID, but he said they are also currently exploring the use of machine learning to better predict melt curves directly from raw sequence input. The team is also exploring the platform's use for other sample types important in infectious diseases.