The goal is to develop a test that can identify the pathogen involved and predict its antibiotic susceptibility within six hours of taking a urine sample and at a cost that is "attractive to hospital laboratories," according to Justin O'Grady, a medical microbiologist at the University of East Anglia's Norwich Medical School and one of the project's leaders.
Members of his team presented early results from a pilot study last weekend at the American Society for Microbiology's Interscience Conference of Antimicrobial Agents and Chemotherapy in San Diego, which was held jointly with the International Congress of Chemotherapy and Infection. They plan to submit their results for publication in a journal in the near future.
More than 30,000 patients in the UK each year are diagnosed with Escherichia coli bacteremia, or blood infections, the majority of which start as urinary tract infections. At the moment, severe urinary tract infections that can lead to urosepsis, where the infection spreads into the bloodstream, are diagnosed by cell culture, O'Grady said, which takes about two days and tells doctors about the pathogen and its antibiotic susceptibility.
A sequencing-based test could potentially deliver this information faster, which could be essential for patients in intensive care units with suspected urosepsis. O'Grady's team has been interested in using DNA sequencing for clinical microbiology for a while, and has been participating in Oxford Nanopore's MinIon Access Program since last summer. Last fall, he and his colleagues published a paper on using the MinIon to characterize antibiotic resistance islands in Salmonella.
"We knew there would be sufficient bacteria present in urine samples, and we knew that this technology would work quite well for this application," he told GenomeWeb, "so we decided to test urinary tract infection samples from the local hospital and see if we could detect these organisms and their resistance genes using this technology."
For their pilot study, the researchers analyzed 10 urine samples from patients treated at the Norfolk & Norwich University Hospital, as well as two control samples that were spiked with E. coli bacteria, one of which had been previously analyzed with Illumina sequencing technology.
To prepare the samples for sequencing, they developed a protocol to enrich for pathogen DNA that includes differential centrifugation to remove human cells and a lysis step that selectively breaks up human cells and destroys their DNA. This is followed by automated bacterial DNA extraction using a Roche MagNA Pure system.
After library preparation, sequencing, and bioinformatic analysis, the scientists were able to identify the pathogen species correctly in six of the clinical urine samples. For the first four samples, the sequence data yield was too low because they were run on early versions of the MinIon flow cell or because human DNA was not adequately removed, but all later samples worked well.
Also, the MinIon identified the same acquired resistance genes in the control sample as the Illumina platform. However, the bioinformatic pipeline sometimes called the same gene multiple times because it did not generate a consensus sequence from the noisy MinIon reads first. In the future, this can be fixed by increasing sequencing coverage and developing consensus calling, O'Grady said.
In the six successful clinical samples, resistance gene profiles derived from the MinIon data agreed well with culture-based drug resistance profiles. But because the culture-based tests did not cover all the resistance genes the MinIon found, they need to confirm the MinIon results by comparing them with Illumina sequencing data, which they are currently generating.
The entire assay currently takes about 12 hours, including a couple of hours to process the sample and extract the DNA, two hours of library preparation, and the remainder for sequencing and data analysis.
The first part — identifying the pathogen — can now be done within the first five minutes of sequencing, using an application called "What's in my pot?" or WIMP that Oxford Nanopore recently introduced for its Metrichor analysis framework. This enables the analysis of the sequence reads in real time, O'Grady said.
After that, the run needs to continue for a few more hours to ensure sufficient coverage of the pathogen genome, so no resistance genes are missed in the analysis that follows. "What we're not good at doing so far is detecting single nucleotide variants that lead to drug resistance in chromosomal resistance genes," O'Grady said. "That will improve with improving technology and a greater yield of sequence data."
He and his colleagues are now working on bringing the assay time down to six hours, which they hope to achieve within the next month or so. In part, this will rely on Oxford Nanopore adding an antibiotic resistance gene database to its Metrichor framework, which would allow the resistance analysis to become automated and proceed in real time. They are also planning to test new flow cells from Oxford Nanopore that double the speed of data acquisition, as well as new library prep kits that reduce the time required for that process.
The assay currently costs on the order of $1,000, he said, including about $900 for the flow cell, $100 for library prep reagents, and $50 for "everything else." However, at its user meeting this spring, Oxford Nanopore talked about plans to introduce 'pay-as-you-go' sequencing for as little as $20 per 2 gigabases, he said, sufficient "to tell us everything we would need" about a sample. "If you could get down to that range, $20 per sample, that would really make a huge difference" for clinical applications, he said.
Getting the test into routine clinical use would require validation studies, though, as well as changes to the workflow, who would analyze the data, and how the results would be presented to doctors. "That would take time," O'Grady said. "But the technology is there."
In parallel to the urine test, he and his colleagues are developing a similar, blood-based test for sepsis, which is more challenging than urine because there is much more human DNA in blood than pathogen DNA. They have already developed strategies to enrich the pathogen DNA and have achieved "some good results" from sequencing so far, he said.