NEW YORK (GenomeWeb) – Aiming to develop a faster, more precise test for tuberculosis, researchers at Oxford University's John Radcliffe Hospital developed one based on next-generation sequencing that was able to diagnose clinical samples as well as identify antimicrobial resistance mutations.
Reporting this week in the The Lancet Respiratory Medicine, the team demonstrated that the test is comparable in terms of performance to standard diagnostic methods, but with a faster turnaround time. In addition, it has the potential to be more cost-effective.
Louise Pankhurst, a co-lead author of the study and a researcher in the Modernising Medical Microbiology group at the University of Oxford, told GenomeWeb that Public Health England is now conducting a feasibility study of around 2,000 samples to see if the test can be implemented into routine diagnostics.
She added that the Modernising Medical Microbiology team is also looking to develop other NGS-based diagnostic tests. They decided to tackle Mycobacterium tuberculosis first because it was "low-hanging fruit," due to the often very long turnaround times of current methods, and because the pathogen itself is "straightforward" in that it does not evolve as quickly as other pathogens and its method of acquiring resistance is more straightforward then gram-negative bacteria like Escherichia coli, which swap plasmids and DNA, and evolve very quickly, she said.
Currently, standard methods to diagnose Mycobacterium tuberculosis complex (MTBC) involve culturing, a genotyping test for species identification, and phenotypic drug testing to identify mutations conferring drug resistance. These tests can take upwards of two months to complete, Pankhurst said, leading to delays in treatment and poor clinical outcomes.
Instead, she said, the team developed a whole-genome sequencing approach that could identify both the species as well as resistance mutations. The group, led by Tim Walker, a co-lead author of the study, began sequencing M. tuberculosis genomes in 2008, Pankhurst said, and has built up a large database and developed algorithms that predict antibiotic resistance and can do cluster outbreak analysis.
For the study, they tested their whole-genome shotgun sequencing approach on the Illumina MiSeq on liquid samples from eight laboratories between September 2013 and April 2014.
The liquid culture samples were all from clinical samples that physicians had predicted were positive for tuberculosis, but were being sent for confirmatory testing. On all samples, both the sequencing approach and the standard approaches were used.
The routine diagnostic method included a genotype-based species identification test known as Hain, as well as culturing with isoniazid, rifampicin, ethambutol, and pyrazinamide to establish drug susceptibility. Some centers also cultured the bacteria with streptomycin.
The samples were sequenced locally at each of the eight laboratories, but were processed centrally with the same bioinformatics pipeline. The researchers wanted to compare diagnostic accuracy, turnaround time, and cost between the NGS and standard approaches. In addition, they calculated genetic distance to previously sequenced MTBC isolates to detect outbreaks.
During the time period, 356 samples were submitted for diagnostic testing. The routine method diagnosed 345 samples. Of those, whole-genome sequencing was concordant with 342 samples, or 93 percent. There were two samples where routine diagnostics failed, but that the NGS method diagnosed, and there were nine samples where both methods failed.
Pankhurst noted that most of the discrepancies were not due to the NGS method being inaccurate. Often, she said, there was not enough culture. This problem could be avoided in the future, she said, because laboratories running the test would have access to the entire liquid culture sample. Another cause for discrepant results was that some of the resistance predictions involved mechanisms that were not yet in the team's catalog, which has since been updated. But, she noted, as more samples are sequenced, it is likely that additional resistance mechanisms will be discovered, so the database will have to be continually updated.
And finally, she said that the team estimated that about 5 percent of the clinical samples had actually been mislabeled. "We had a couple of examples where I'm fairly sure there was either a switch on our side or a switch on the clinical lab side," she said.
The researchers also found that the NGS-based method was significantly faster than the standard testing method. The full whole-genome sequencing diagnosis and drug resistance prediction took an average of nine days, while the standard method took 31 days on average.
The NGS method was also able to identify 15 cases that were part of an outbreak in the UK and was able to identify one case that had multidrug resistant TB well before the routine diagnostic was complete and identified a new multidrug resistant cluster.
In terms of cost, the researchers estimated that the NGS method cost approximately £481 ($729) per culture-positive specimen, whereas routine diagnosis costs £518 ($785), equating to a WGS-based diagnosis cost that is 7 percent cheaper annually than are present diagnostic workflows. The team also analyzed the cost of running the sequencing method in conjunction with drug-resistance phenotype testing and found that cost to be £540 per sample ($819). Pankhurst said that initially, it was likely that the sequencing test would be run in conjunction with drug-resistance phenotyping. "That's the one element that clinicians have the most concern about, so we want to make sure that everyone has faith in the [NGS] answer before phenotyping is phased out," she said.
In addition, she said, the cost analysis estimate was conservative since the team did not assess the cost of having to do repeat phenotype testing, which she said occurs a lot. They also did not include the potential savings that could be realized from having a faster test, such as getting a patient on antibiotics faster or shifting patients out of the isolation unit sooner. "In reality, the cost savings could be even greater," Pankhurst said.
Pankhurst said that she is hopeful that the NGS-based TB test will be rolled out in the UK within the next year. She anticipated it would initially be run alongside the routine tests, but thought that those would eventually be phased out.
The Modernising Medical Microbiology group will now focus on developing NGS-based tests for Escherichia coli and Staphylococcus aureus that will also hopefully eventually replace current techniques.