Skip to main content
Premium Trial:

Request an Annual Quote

Consortium Builds Tuberculosis Genome Database in Move Toward NGS-Based Diagnostics


SAN FRANCISCO (GenomeWeb) – Researchers are turning to next-generation sequencing to develop better methods for cataloguing mutations that confer tuberculosis drug resistance and, ultimately, to design better diagnostics.

According to the World Health Organization, 80 percent of people infected with multi-drug resistant tuberculosis in 2015 did not receive the appropriate treatment, and only half of those who started treatment were cured. This is compared to a more than 90 percent cure rate for tuberculosis that is not resistant.

One reason for these poor statistics is that the methods for diagnosing and determining whether an individual has a drug resistant strain of tuberculosis are outdated and slow, according to Marco Schito, scientific director at the Critical Path Institute in Tuscon, Arizona, who is heading the Critical Path to TB Drug Regimens project there. Schito's group at the Critical Path Institute is one partner in a University of Oxford-led consortium called Comprehensive Resistance Prediction for Tuberculosis: an International Consortium (CRyPTIC). CRyPTIC, which launched last March with $2.2 million in funding from the Bill & Melinda Gates Foundation and £4 million ($5 million) from the Wellcome Trust and Newton Fund, aims to sequence the genomes of 60,000 Mycobacterium tuberculosis isolates and to test them for susceptibility to 14 drugs over the next five years.

"There are already some well-known mutations that confer resistance," Philip Fowler, a senior researcher in the Modernising Medical Microbiology group at the University of Oxford, said. "But, the idea is to identify additional mutations. When you have a more comprehensive catalog, it makes genetics-based microbiology much more accurate."

Fowler added that tuberculosis is a good test case for the broader adoption of NGS-based infectious disease diagnostics because there is so much room for improvement on current methods for diagnosing it and screening for drug resistance.

Currently, tuberculosis is diagnosed, tested for drug-resistant mutations, and genotyped to determine how the infection clusters with other strains in order to spot potential outbreaks by a variety of methods. Typically, microscopy is performed first to identify whether the organism is part of the Mycobacterium family. Then it is grown in culture to determine whether it is tuberculosis or non-tuberculosis Mycobacterium. After that, it is grown in the presence of drugs to see whether it has any resistance, and finally, it is genotyped.

But because tuberculosis is a slow-growing organism, the whole process can take six to 12 weeks. Meanwhile, the patient is sick and has been started on first-line therapy that may or may not be effective. "We need to move away from the culture paradigm," Schito said.

There is already a number of well-known mutations that confer drug resistance in TB and researchers at the Modernising Medical Microbiology lab have developed an NGS-based test that incorporates that current knowledge, which was adopted by Public Health England earlier this year. The Oxford team initially described its test in 2015 in the journal The Lancet Respiratory Medicine.

In the study, the researchers found that the sequencing-based test took an average of nine days, whereas the standard method — which involved culturing, a genotyping test for species identification, and phenotypic drug testing to identify mutations conferring drug resistance — took an average of 31 days. The tests had comparable performance, but the sequencing test identified cases earlier, including several that were part of an outbreak as well as one case with multi-drug resistant TB.

Public Health England has now begun implementing this whole-genome sequencing protocol. An initial culture step is still needed to ensure that there are enough bacteria to be sequenced, but drug resistance and genotyping information can be determined from the genomic information.

What CRyPTIC aims to do is build a database of tuberculosis samples from all over the world. They are also adding to the current knowledge of drug resistance mutations, including screening samples against newer drugs, with the ultimate goal of building a comprehensive resource to enable other labs to also adopt sequencing-based diagnostic tests.

The CRyPTIC team is testing for drug resistance using a 96-well microtiter plate designed by Thermo Fisher Scientific. 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, or the minimum amount of a drug that's needed to prevent growth. The Critical Path Institute worked with the CRyPTIC team to develop this standardized approach.

Importantly, Schito said, the plates include new and repurposed drugs. "We don't really know what the level of resistance is for those," he said. "So there is concern about using the newer drugs in a blind manner." This method of testing strains against varying concentrations of drugs should help decide what drug should be administered and in what concentration, he added.

Fowler said that the participating labs are just now completing the validation stage of scoring the plates, making sure that results on whether a strain is resistant or susceptible to a drug are consistent.

This phenotypic drug resistance data will be added to the genomic data to give a more complete picture and to identify new mutations that may be responsible for drug resistance or sensitivity.

Separately, Fowler has also engaged the citizen science community to help score the strains' sensitivities to drugs. Known as Bash the Bug, participants who sign up receive photos of the plates. They then score each well as having growth, no growth, or partial growth. Each image is shown to 15 different individuals to build a consensus, Fowler said. Since April, more than 5,000 participants have signed up and classified more than 250,000 plates for the validation phase of the project.

Fowler said this crowdsourced approach helps complement the algorithmic-based approaches for determining drug sensitivity, which can sometimes "get confused by artifacts in the image like air bubbles." Each participating lab is also doing its own classifications, and ultimately, Fowler anticipates that pooling data from the three would yield the best results.

Jennifer Gardy, the Canada Research Chair in Public Health Genomics at the University of British Columbia, said her lab has contributed 1,500 TB genomes to the CRyPTIC project thus far.

She is also leading a separate collaborative project with Derrick Crook's laboratory at the University of Oxford and PHE, known as the Sharing Mycobacterial Analytic Capacity (SMAC) project. Part of that project involved validating the informatics pipeline that PHE now uses in its TB sequencing test. The pipeline uses the TB sequence data for speciation analyses, resistance testing, and clustering.

The groups are also working on a project to improve how the genomic information is delivered back to the end user — the scientist in the reference laboratory who reports relevant bits to physicians, epidemiologists, and other public health officials.  Gardy's team is working on tapping into computer scientists' expertise to better design the report to incorporate visualization components that make it easier to read and use. She considers this absolutely critical for ensuring the success of sequencing-based diagnostics for tuberculosis and other infectious diseases.

"We have an opportunity to get genomics into public health labs in a way that's not been done before," Gardy said. "But if the end report is indigestible, if we don't think about how to make the output clear and useful, we run the risk of people not getting on what's the biggest revolution to hit public health microbiology in decades."

The laboratories sequencing TB strains for CRyPTIC are doing so using Illumina's technology, but Fowler said there has been growing interest in evaluating Oxford Nanopore's MinIon device. Because it is small and portable, there is the potential of using it "in the field," away from large hospital labs.

Gardy added that the MinIon also has potential for designing a more rapid, point-of-care test. Although whole-genome sequencing-based diagnostics for TB require a culture, she said, a targeted sequencing approach would not. For instance, if CRyPTIC's work ultimately identifies that drug resistance mutations are confined to, say, 20 to 30 genes, then it would be feasible to design a targeted assay that could be performed directly from sputum. If that's coupled with sequencing on the MinIon, she said, results could be obtained in less than one day. The downside would be that a targeted approach would not give information that epidemiologists use for investigating outbreaks, Gardy said, but "targeted sequencing directly from sputum is great for diagnosing and rapidly initiating the appropriate therapy."