NEW YORK – DNA sequencing-based approaches for diagnosing drug-resistant TB have been gaining acceptance and adoption worldwide to improve treatment for a disease that killed approximately 1.4 million people worldwide last year, making it the leading cause of death attributable to a single infectious disease pathogen.
In its Global Tuberculosis Report for 2020, released in October, the World Health Organization noted that sequencing can provide detailed information on resistance across multiple genes, and specifically called out examples of targeted next-generation sequencing tests in development at TGen and GenoScreen.
The Foundation for Innovative New Diagnostics (FIND) is currently evaluating such technologies as part of a three-year project called Seq&Treat. The global nonprofit organization is validating a targeted NGS solutions that can be used in low-resource settings, guided in part by a number of new research studies and existing TB sequencing workflows, including one that is now used for routine TB testing at New York State Department of Health's Wadsworth Center.
Quickly determining the resistance profile of a TB infection is critical to get patients on a treatment regimen most likely to cure them. There were roughly half a million drug-resistant TB infections worldwide last year, according to the WHO report. These showed resistance to a first-line drug called rifampin, but nearly 80 percent were also resistant to other classes of drugs and thus deemed multi-drug resistant.
Since TB is caused by a slow-growing organism, gold-standard culture-based drug-resistance testing for the disease can take up to six weeks to generate complete information.
In recent years, PCR tests have enabled the rapid detection of some kinds of resistance, with firms like Cepheid crafting increasingly more complex assays like the recently-launched 10-color Xpert MTB/XDR test to detect extensively drug-resistant TB.
Unlike PCR-based tests that look for a handful of genetic indicators of resistance, whole-genome sequencing can theoretically detect all resistance-conferring signatures. Targeted next-generation sequencing, meanwhile, can be used to focus on signatures that are more frequent, with potential time and cost benefits.
A five-year evaluation of more than 7,000 patients in seven countries — Azerbaijan, Bangladesh, Belarus, Pakistan, the Philippines, South Africa, and Ukraine — published in The Lancet in 2018 suggested that WGS resistance profiling could be a valuable tool for TB surveillance, even in resource-limited settings, provided the methodologies could be standardized. Meanwhile, a 2018 study in the New England Journal of Medicine of more than 10,000 TB isolates concluded that sequencing-based testing could replace culture-based susceptibility testing, and a study in Madagascar that same year demonstrated the Oxford Nanopore MinIon sequencer could be used for TB susceptibility testing in rural areas.
In parallel, in 2018 the WHO initiated development of a TB sequencing database adapted from one called ReSeqTB for drug-resistant TB.
More recently, an article in Nature Reviews Microbiology last yearprovided workflow guidance and best practices for WGS of TB organisms, while a study in The Lancet Infectious Diseases suggested that WGS is the most promising approach to drug susceptibility testing for TB in order to prevent inaccurate treatment.
This year, the field advanced with a study published in The Lancet in August that demonstrated WGS could be performed directly from patient sputum samples at a cost per sample of approximately $260, generating high-resolution genomic epidemiology information.
Sequencing directly from sputum, together with an improved interpretation catalog of drug resistance-defining mutations, could potentially allow genotypic drug susceptibility testing to replace phenotypic testing for a large fraction of cases, according to a review published in October in Frontiers in Immunology. However, the majority of patients affected by tuberculosis live in resource-limited settings, and "finding funding and performing operational research on the implementation of precision medicine for tuberculosis in these settings will be one of the great challenges for the future," the authors concluded.
FIND's Seq&Treat program
FIND's Seq&Treat program is a rising to this great challenge. The program, funded by a $14.5 million grant from Unitaid, got underway a little more than a year ago.
Anita Suresh, head of the sequencing program at FIND, said in an email that the three-year project aims to generate evidence and catalyze the market for "end-to-end targeted NGS solutions for diagnosis of drug-resistant TB in low- and middle-income countries." It will initially be implemented in Brazil, China, Georgia, India, and South Africa.
The first year has focused on the evaluation of multiple targeted sequencing end-to-end solutions, Suresh said, including laboratory validation using a blinded reference panel.
"Data from this study is being reviewed by technical experts, including at WHO, and will be used to determine selection of technologies that will advance to the next phase, involving a multi-center clinical evaluation study," Suresh explained. The clinical study will evaluate the diagnostic accuracy and performance of up to three targeted NGS solutions in the settings of their intended use, she said.
In addition, FIND and partners have developed the first global catalogue of mutations associated with drug resistance in TB, which is being reviewed by a WHO technical experts group and will be subsequently published by the WHO, she said.
Suresh noted that sequencing should be thought of as complementary to rather than competing with PCR-based approaches.
For example, a technical guide from FIND and WHO regarding sequencing for TB notes that the technique can be used as a reflex test after PCR testing, or for all samples as part of routine surveillance for resistance.
Although the cost and expertise required for sequencing would seem to make it an unlikely fit for low-resource settings, Suresh argued that compared to culture, sequencing has many advantages. Culture takes up to eight weeks and requires a biosafety level 3 lab that is expensive to maintain and operate on a large scale, she said, while sequencing costs actually decrease as more samples are put on a sequencer.
"Therefore, sequencing-based approaches are already more amenable for use in low-resource settings relative to current practice," she said.
Other global efforts seem to back this up. For example, the Africa Pathogen Genomics Initiative, which launched earlier this year with $100 million in support from the Bill and Melinda Gates Foundation, Microsoft, Illumina, Oxford Nanopore Technologies, and the US Centers for Disease Control and Prevention, expects to connect public health labs across Africa to establish and demonstrate the value of pathogen NGS, including for tuberculosis.
To make sequencing further suitable for low- and middle-income countries, the Seq&Treat evaluations will include clinical performance and ease of use assessments. The goal is to generate evidence-based policy and implementation guidance for routine use in high-burden TB programs, Suresh said. As part of this work, the FIND team will also assess and guide national programs on optimal placement and capacity utilization of NGS solutions.
The project will also facilitate the inclusion of endorsed NGS workflow components, including instruments and related consumables, in global procurement lists such as the Global Fund and the Global Drug Facility, "to enable public sectors in [low- and middle-income countries] to procure them with the applicable donor funding," Suresh said.
Suresh co-authored a review, published in September in the Journal of Clinical Microbiology, describing molecular diagnostic approaches to TB that emphasized the fact that nucleic acid amplification tests and next-generation sequencing can provide results substantially faster than traditional phenotypic culture for drug susceptibility testing.
The review also offered updates on efforts in Brazil, India, and South Africa to adopt sequencing-based approaches to TB, as well as on two specific commercial technologies in development.
One was the Deeplex Myc-TB from Lille, France-based Genoscreen, which uses ultradeep sequencing of 24-plex amplicon mixes for mycobacterial species identification, genotyping, and drug susceptibility testing. The test was CE-marked in March of this year.
The second was a targeted sequencing assay called DeepChek-TB developed by the Translational Genomics Research Institute in Flagstaff, Arizona. Last year, TGen licensed the assay to Advanced Biological Laboratories in Luxembourg for marketing and distribution.
An evaluation of the GenoScreen Deeplex-MycTB published in August showed it could be used to predict drug resistance in clinical specimens, while another, published in September, showed it could be used directly from sputum.
Furthermore, different sequencers are being evaluated for TB testing. A comparison of Illumina MiSeq and Nanopore MinIon targeted NGS workflows published in Clinical Chemistry in June led a global team of researchers to conclude that the two approaches are interchangeable. Meanwhile, a comparison published in the Journal of Clinical Microbiology in September that Suresh co-authored showed that the workflow and sequencing data obtained from Deeplex on MinIon are comparable to those for the Illumina MiniSeq.
Although the application of sequencing technologies to TB is progressing rapidly, the COVID-19 pandemic is expected to be detrimental to testing and treatment. The WHO TB report presented data modeling as many as 400,000 excess deaths from TB due to COVID-19-lockdown-related delays in diagnosis and treatment.
The Seq&Treat project has also been impacted by the COVID-19 pandemic, Suresh said, but has pivoted online as much as possible and has used the time to assess the sequencing technology landscape for COVID-19 surveillance and management, and to provide technical guidance to countries on building sequencing capacity for pandemic preparedness.
Wadsworth's workflow
While national TB programs in the Netherlands and the UK have already taken up sequencing-based testing, the US Centers for Disease Control and Prevention has been sequencing isolates from all culture-confirmed TB cases in the United States.
New York State has also transitioned to evaluating resistance in all of its 800 or so TB cases each year using sequencing at the Department of Public Health's Wadsworth Center.
Kimberlee Musser, chief of bacterial diseases at Wadsworth, described the WGS approach the state has developed last month at a presentation at the annual Association for Molecular Pathology, held virtually this year.
Wadsworth began development of the test in 2013, and has gradually validated and scaled it.
"We were doing many different molecular assays ... and we thought, could we have one test, a one-stop shop that could provide us everything we needed to know about an organism?" Musser said.
Currently, Wadsworth performs its WGS test on 1 milliliter of liquid culture from a Mycobacterium growth indicator tube, or MGIT, with MiSeq instrumentation and Nextera XT library preparation. The cost is approximately $200 per sample using batched testing, which leads to a turnaround time of about seven days.
The bioinformatics analysis pipeline was a tricky part, because the Wadsworth team had to consider results reporting, LIMS, and validation of the test, Musser said.
"We wanted this pipeline to be able to give us a high level of resistance prediction, comprehensive assessment of the high-confidence mutations ... and some additional phylogentic analysis," she said.
The pipeline it now uses was developed by an in-house bioinformatician, Pascal Lapierre, and provides species identification and resistance information. It also performs spoligotyping — a method for simultaneous detection and typing of M. tuberculosis — which it uses to help provide relatedness information about strains to local epidemiologists.
The team looks at different high-confidence mutations that impact susceptibility to nine different drug classes. It has established a method to update the pipeline when new mutations crop up, but also locked down the pipelines it uses at different versions, "so that we know, until we make our own update, nothing can change about that pipeline in the meantime," Musser said.
"There are hundreds of these mutations, insertions, and deletions that we detect, and we have many more under evaluation," she said.
To validate the test, "we pulled out every strain in our freezer that had a discordant susceptibility result and molecular result, and any that were just unusual that we really wanted to see what WGS would additionally provide," Musser said. "It answered every question we had," she said, for example detecting mutations in the primer binding site that led to molecular testing failure.
The team needed to standardize the testing and reporting, and bring this workflow through the Clinical Laboratory Evaluation Program. "In the end, it was the biggest validation document we've ever put together, almost 450 pages," Musser said.
The report that the testing generates identifies the bacteria, gives a susceptibility profile, and lists the mutations identified.
Along the journey, the team has been constantly evaluating the susceptibility predictive value. For the four most common drugs, the pipeline has an approximately 98 percent positive predictive value for susceptibility, Musser said, noting that this data is from 8,253 resistance calls.
More than 80 percent of strains seen since 2016 have been pan-resistant, while 8 percent were isoniazid resistant, and nearly 3 percent are MDR and XDR strains. The remainder are other resistances, including to a single first-line drug, or rare resistances that may not have been picked up with traditional testing, Musser said.
Interestingly, the team also compares new sequences to its database to look for closely-related strains that differ in fewer than 20 SNPs across the 4 megabase genome. This triggers a report to the state's epidemiologists, who have also noted that fewer than five SNPs indicates a high likelihood of recent transmission. There were 30 such cases in the state this year that could then undergo transmission investigations.
The most recent update also includes an assessment of pan-susceptible strains, and when the pipeline flags one of those, no further drug susceptibility testing is done. This enables the team to focus on strains likely to have resistance, so that unknown mutations can be investigated further.
The team also recently evaluated Oxford Nanopore MinIon-based approaches and saw high concordance with its clinically validated Illumina MiSeq sequencing assay at "a very competitive cost per sample."
In 2021, the Wadsworth group will be looking into further improving costs by increasing throughput with a NextSeq format as it continues to look at how it might utilize the MinIon, Musser said. The lab is also focusing on direct specimen NGS, she added.
"We are also working hard to provide our pipeline externally to some US public health labs that are interested in being able to do similar work," Musser said.
The Wadsworth team's work is also enmeshed in the larger global community. Musser said the center has worked closely with the UK team of Timothy Walker, Derrick Crook, and others at Oxford as they helped bring on Public Health England's sequencing-based TB testing program.
Over the last few years, FIND has also interacted with Wadsworth scientists via visits, webinars, and discussions, Suresh said, by "seeking and sharing feedback and experience on sequencing protocols, data analyses, and implementation." The FIND team now plans to actively engage with Wadsworth as it moves further along in the Seq&Treat project, "especially on assessment of various implementation models for sequencing and planning for scale, given their invaluable experience in this space," Suresh said.