The five-year project, called Inhale, started in early 2016 and initially aims to assess the diagnostic performance of three respiratory disease tests — running on the Curetis Unyvero and BioMérieux's BioFire Diagnostics FilmArray, both PCR-based; and Oxford Nanopore Technologies' MinIon sequencer — to select the one that performs best for a prospectcive randomized controlled clinical trial comparing the platform against conventional cell culture.
Inhale is a collaboration between University College London (UCL), the University of East Anglia (UEA), and four hospitals — Norwich & Norfolk University Hospitals, University College London Hospitals, Great Ormond Street Hospital Children's Charity, and BUPA Cromwell Hospital — and is funded by the UK's National Institute for Health Research.
The original plan was to assess three PCR-based platforms, but after one of the manufacturers, Mobidiag, pulled out earlier this year, the project replaced that system with the Oxford Nanopore MinIon and started analyzing the first samples with nanopore shotgun metagenomic sequencing earlier this month.
"The ultimate goal here is to try to improve the treatment of patients, to look at a more personalized approach to antibiotic treatment," said Justin O'Grady, senior lecturer in medical microbiology at the University of East Anglia's Norwich Medical School, and one of the organizers of the project. "It's a move away from the guesswork that we currently use to diagnose and treat people."
Because pneumonia can be a life-threatening disease, especially in children, the elderly, and critically ill patients, doctors usually start treatment with antibiotics immediately. Two or three days later, when a microbiology lab has cultured the pathogen from the sample and determined its antibiotic resistances, treatment is often adapted to target the causative agent more specifically.
The idea is to shorten that timeframe to a few hours by using a molecular diagnostic test, O'Grady said, so patients can receive the most appropriate treatment more quickly.
For the clinical trial, "we will look at what drugs people got, when they were changed to a different drug, if they had the same outcomes, and if the antibiotic stewardship was better — if you were reserving the most potent antibiotics for those who really need it," he explained.
The trial will also have a health economics component to study the costs involved with moving to a molecular diagnostic platform, and a behavioral science part to look at whether doctors will actually act upon the results of a rapid molecular test.
For the first part of the project, which assesses the performance of the systems and whether they find the same pathogens and resistances as culture-based testing, the researchers plan to analyze about 600 samples on all three platforms. Samples come from the intensive care units of the participating hospitals and are from patients with hospital-acquired pneumonia, including patients on ventilators. Three sample types are being studied: bronchoalveolar lavage (BAL) samples, sputum, and endotracheal tube aspirates.
So far, the researchers have tested about 200 samples with the Curetis and BioFire platforms, some of which were frozen and will also be tested with the MinIon sequencer. The plan is to collect about 600 additional samples over the coming months for analysis with all three platforms, which are installed at the two testing laboratories at UCL and UEA.
About a year from now, one of the platforms will be chosen for the randomized controlled trial. O'Grady said the selection criteria will not only be the tests' specificity and sensitivity but also their cost and ease of use.
While the results of the initial evaluation will not be in for a while, in general terms, the three platforms have a number of pros and cons. For example, the Curetis test, which is already on the market, and the BioFire test, which is still in development, both run on "sample-in-answer-out" platforms that are simple to use and require little hands-on time, O'Grady said, whereas the MinIon metagenomic sequencing test, which his lab developed in house, is not as mature yet.
On the other hand, "the depth of information provided by metagenomics is vastly superior over what you get out of a PCR," he said. It includes the full genome of the pathogen and all resistance genes, for example, whereas the PCR tests only target selected organisms and resistances.
A shotgun metagenomics assay will also likely not miss any pathogen, he said, even unusual bacteria, whereas the PCR tests run panels that may cover 90 or 95 percent of the culprits, but not all of them.
In addition, metagenomic sequencing covers all organisms in a sample, pathogens and commensals, which O'Grady said is very useful because it provides an idea of whether a pathogen is a major or a minor component of a sample. Like a culture plate, a metagenomics test lets labs "see" quickly what is there, "whereas with PCR tests, you just pick out the pathogens and you don't know what else is possibly there," he said.
Being able to detect all resistances, even less common ones, is also an advantage, he said, but right now, bioinformatics pipelines are not ready yet to fully interpret genome sequence data for antibiotic susceptibility.
The platforms also differ in turnaround time. The Curetis test and the nanopore sequencing test currently take about six hours each to perform, whereas the BioFire test only takes about an hour.
What's important is to keep the turnaround time under eight hours, O'Grady said, because doctors usually put patients on a broad-spectrum antibiotic immediately and only need the test result before the patient takes the second dose about eight hours later.
It is not clear yet how the three tests will compare in price. The nanopore sequencing test currently costs the lab on the order of $120 per sample in consumables, including sample prep reagents, if six samples are combined on one flow cell, O'Grady said. This is less expensive than the Curetis test, which costs on the order of $230 per sample, but the price could be different if Oxford Nanopore decided to develop the test into a commercial offering, for example. The price of the BioFire test is not known yet since it has not been fully commercialized, he added.
Whichever test is chosen for the randomized clinical trial, where it would be used to guide patient treatment, will need to be approved by the study funders and an ethics board first. It is unclear yet whether this will require CE-IVD marking, US Food and Drug Administration approval, or whether it could be validated as a lab-developed test, O'Grady said.
Curetis submitted its Unyvero system and lower respiratory tract (LRT) panel to the FDA earlier this year and the BioFire FilmArray system is already FDA-cleared and CE-marked.
Prior to entering the nanopore respiratory test into the Inhale project, O'Grady's lab evaluated it in a proof-of-concept study, using about 50 sputum or BAL bacterial pneumonia samples. The analysis is still ongoing, but initial results look like the shotgun metagenomic approach has "excellent concordance with culture for pathogen identification," O'Grady said. The researchers plan to publish the results of the pilot in the near future, he added.
Sample prep for the test starts with a yet-unpublished depletion method the lab developed that removes human DNA. This is followed by DNA extraction and library preparation, using Oxford Nanopore's Rapid Low Input by PCR Barcoding Kit, which requires 1 to 5 nanograms of input DNA. Depending on how much DNA is available from each sample — BAL samples, which are more dilute, generally yield less DNA than sputum or endotracheal tube aspirate — the lab usually performs 15, 20, or 25 cycles of PCR. Every five PCR cycles add about 30 minutes to the procedure, O'Grady explained, and the lab is working on ways to shorten the elongation step of the PCR.
At the moment, the lab multiplexes six samples per MinIon flow cell to keep costs down, but once Oxford Nanopore comes out with its Flow Cell Dongle, or Flongle, a low-cost MinIon flow cell adapter, the researchers plan to run a single sample per flow cell. Oxford Nanopore said previously that it intends to pursue regulatory approval for the Flongle, in a version to be called MinIon Dx, for use in diagnostics.
Right now, the team obtains 2 million reads on average in a 48-hour run, with average read lengths of 2 to 4 kilobases. The error rate of the nanopore reads is not a problem for pathogen identification, O'Grady said, but rapid identification of mutational resistance will require real-time consensus calling, which he believes Oxford Nanopore's team is working on.
To analyze the data, the lab currently uses real-time analysis workflows developed by Oxford Nanopore — 'What's In My Pot,' or WIMP, for taxonomic classification of microbial organisms and ARMA for antimicrobial resistance profiling.
Others have also been exploring shotgun metagenomics for infectious disease diagnostics.
Charles Chiu, an associate professor at the University of California San Francisco School of Medicine and director of the UCSF-Abbott Viral Diagnostics and Discovery Center, for example, said that his lab just completed a prospective clinical trial for meningitis and encephalitis diagnosis in which the team compared its clinical metagenomic next-generation sequencing (mNGS) assay to conventional testing, including culture and PCR testing, and evaluated clinical utility, cost-effectiveness, and patient outcomes. That test uses an Illumina sequencing platform and the patented SURPI algorithm for analysis.
The researchers are currently writing up the results for publication, Chiu said, and he plans to present them at a conference in September. Earlier this month, UCSF's CLIA-certified clinical microbiology laboratory launched the test, which analyzes cerebrospinal fluid, for patients with neurological diseases and said it was working on making testing available for patients with sepsis and pneumonia. The lab said it currently conducts the test in about 72 hours, and results are generally available within one or two weeks.
Also, PathoQuest, a spinout from the Institut Pasteur in France, has developed a metagenomic sequencing test for blood pathogens called iDTECT Blood, for which it obtained CE-marking last year. That test also uses an Illumina platform.
In addition, researchers at the University of Utah, Arup Laboratories, and bioinformatics firm IDbyDNA have also developed a metagenomic sequencing-based test for diagnosing respiratory and other diseases, presenting results earlier this year at the Biology of Genomes meeting.