SAN FRANCISCO (GenomeWeb) – Whole-genome sequencing has the potential to replace conventional phenotyping methods for drug susceptibility testing for Mycobacterium tuberculosis, according to a study published in the New England Journal of Medicine this week.
In the study, conducted by members of an international consortium studying tuberculosis drug susceptibility and the UK's 100,000 Genomes Project, researchers analyzed whole-genome sequence data from more than 10,000 TB isolates to see whether they could predict the strains' susceptibility to four first-line TB drugs from the sequence data alone.
The results were so convincing that already, the Netherlands, Public Health England, and the New York State Department of Health have opted to forgo phenotypic drug testing when whole-genome sequencing predicts that a strain is susceptible to a number of first-line drugs.
Molecular tests to predict drug susceptibility are "potentially a paradigm shift," said Tim Walker, a clinical lecturer at the University of Oxford's Nuffield Department of Medicine and an author of the NEJM study. "The data has already had an impact and I imagine other [public health agencies] will follow suit as well."
Tuberculosis is the most deadly infectious disease worldwide, causing 1.6 million deaths in 2017, according to the World Health Organization. Drug resistance is a particularly challenging and growing problem: In 2017, the WHO estimated that 558,000 new cases of TB were resistant to rifampicin, the most effective first-line drug, and globally, just over half of patients with multi-drug resistant TB are successfully treated.
The Comprehensive Resistance Prediction for Tuberculosis: an International Consortium (CRyPTIC) was launched in 2016 with $2.2 million in funding from the Bill & Melinda Gates Foundation and £4 million ($5 million) from the Wellcome Trust and Newton Fund. Ultimately, the consortium aims to sequence a total of 100,000 TB genomes.
Walker said that the sequence data included in the NEJM study was generated by members of the consortium and constituted isolates from 16 different countries across six continents. The isolates had all previously been sequenced, but the genomic data was then aggregated and brought under the CRyPTIC umbrella to enable deeper analyses. Genomics England generated about one-third of the sequence data and "provided the computational structure in which such a large dataset could be analyzed," Walker said.
In the study, the researchers compared the ability of genome sequencing to predict drug susceptibility and resistance with standard phenotypic-based approaches, such as culture-based methods, for four first-line TB drugs: isoniazid, rifampin, ethambutol, and pyrazinamide.
Full phenotypic drug susceptibility profiles were available for just over 7,500 isolates, and partial profiles for the remainder. Around 48 percent of isolates were found to be susceptible to all first-line drugs.
To make genotype-based predictions, sequence data was analyzed for nine known genes and their promoter regions.
Overall, the sequence data was able to predict resistance to isoniazid, rifampin, ethambutol, and pyrazinamide with sensitivities of 97.1, 97.5, 94.6, and 91.3 percent, respectively, and was able to predict susceptibility to each of the drugs with specificities of 99.0, 98.8, 93.6, and 96.8 percent, respectively.
The overall negative predictive value of the genotypic susceptibility predictions was 98.5 percent.
An advantage of using sequencing is that both susceptibility and resistance testing can be done in one assay. Currently, drug susceptibility testing is performed using culture-based methods, which can be time consuming. And although rapid PCR-based drug resistance tests have been developed, Walker said, resistance and susceptibility "aren't quite two sides of the same coin." For instance, he said, PCR-based tests analyze the most common mutations that confer resistance to rifampin, and if the test detects one of those mutations, it's also possible to infer the strain is resistant to other drugs, since those mutations are rarely found in isolation. On the other hand, if a resistance mutation is not identified, drug susceptibility cannot be inferred.
"The PCR-based tests only tell you which drugs not to give, not which drugs to give," Walker said.
With sequencing, however, it is possible to both predict which drugs a patient may be resistant to and which drugs he or she will most likely respond to.
In an accompanying editorial, also published in NEJM this week, Helen Cox and Valerie Mizrahi, of the Wellcome Center for Infectious Diseases Research in Africa, wrote that a diagnostic test that could determine the complete drug susceptibility and resistance profile "would make it possible to give patients the correct treatment and thereby decrease the amplification and onward transmission of drug resistance." The study "brings us a step closer to this goal," the authors added.
What's needed next is a list of all resistance mutations associated with existing, repurposed, and new drugs. "A more complete knowledge base enabling the prediction of resistance to all the drugs currently used for the treatment of multi-drug resistant or rifampin-resistant tuberculosis is required in order to implement an approach whereby treatment can be rapidly individualized," Walker said.
That, he said, is the main goal of CRyPTIC. The team is continuing to sequence TB isolates and to test them for drug resistance and susceptibility. Recently, it validated a custom-designed 96-well microtiter plate, developed by Thermo Fisher Scientific, to test tuberculosis isolates against 14 drugs — the four first-line therapies, second-line drugs, two new drugs, and two repurposed ones.
Each of the 14 drugs occupies six to seven wells at varying concentrations, so that researchers can identify what's known as the minimum inhibitory concentration (MIC), or the minimum amount of a drug that's needed to prevent pathogen growth.
Walker said that within the next 18 months or so, he expects CRyPTIC to have generated sequence and MIC data for nearly 30,000 TB isolates. The MIC metric is important, he said, because in some instances "mutations seem to be inconsistently related to a given phenotype," sometimes seeming to indicate resistance to a given drug and other times being associated with sensitivity to a drug. Walker said that one reason for this is that certain variants can result in either sensitivity or resistance, depending on the concentration of the drug that is used.
"There's a tight margin of error" for drug dosing, he said, and one big push of CRyPTIC is to identify the dosage for various drugs that is needed for a given isolate to stop growing.
This type of analysis is "shedding new light on the behavior of mutations and the predictive value of those mutations," Walker said. The group's work in the upcoming year will be particularly important for generating data about the new and repurposed drugs, he added.
In general, sequencing is increasingly being used for infectious disease testing. New York state has already begun to replace traditional phenotypic testing with whole-genome sequencing for drug susceptibility testing for tuberculosis, and Public Health England has implemented an NGS-based diagnostic test for suspected TB cases. Researchers from the University of Oxford, meanwhile, have been developing a method to sequence patient samples directly with Oxford Nanopore Technologies' MinIon, which would further decrease turnaround time and enable testing in more remote locations.
With a portable platform, the goal of delivering "individualized therapy to the bedside anywhere in the world" becomes more realistic, Walker said. "It's no longer a pipe dream."