Whole genome sequencing for the rapid prediction of drug susceptibility in patients with suspected multi-drug resistant tuberculosis in Vietnam

Lead Research Organisation: University of Oxford
Department Name: Tropical Medicine


Mycobacterium tuberculosis is a bacteria that causes tuberculosis, an infectious disease that kills nearly 2 million people each year worldwide. Despite the availability of effective drug treatment for more than 50 years, and coordinated global efforts to control the disease, tuberculosis kills more people than any other infectious disease. One of the major reasons for the world's failure to control tuberculosis, and why it poses such a threat for the future, is that the bacteria are becoming increasingly resistant to the drug treatments. This problem is compounded by the difficulty in detecting drug resistant bacteria, which grow very slowly in the laboratory and require special facilities to handle and test safely. The current standard methods for detecting tuberculosis drug resistance take 8-12 weeks to give results and are only available in specialist laboratories. In Vietnam, for example, which has >100,000 people with tuberculosis each year, there are only two laboratories in the country that can perform full tuberculosis resistance testing. The inadequacy and inaccessibility of currently available resistance detection methods means people with drug resistant tuberculosis are either diagnosed late, or not at all, which leads to more severe disease, on-going transmission of the infection to others, and more tuberculosis deaths.
This problem is especially serious if a patient has so-called 'multi-drug resistant' (MDR) tuberculosis. MDR tuberculosis is defined by resistance to at least two of the most active 'first line' anti-tuberculosis drugs. It is much harder to treat, requiring four or five 'second line' drugs that are less effective than the 'first line' treatments and must be given for 12 months or more. Ideally, resistance testing would help doctors select the best drugs to treat people with MDR tuberculosis quickly and easily and thereby ensure the patient gets better faster. However, the current resistance detection methods are too slow (and are often unavailable) to help doctors and their patients.
Our proposal will investigate a new solution to the problem of tuberculosis drug resistance testing. The team from the University of Oxford in the UK have been at the forefront of using the bacteria's genetic code (or DNA) to detect resistance against all anti-tuberculosis drugs. They have developed relatively simple methods that are able to look at the whole genetic code of a bacteria (or 'whole genome sequence') and rapidly predict whether or the not the bacteria are resistant to any or all of the anti-tuberculosis drugs by detecting the genetic mutations associated with resistance. These methods take a couple of days, not the 8-12 weeks of the current standard methods, and are so accurate that Public Health England now uses them in their routine service. Other developed countries are set to follow suit, but these genetic methods have yet to be tested in less well-resourced countries like Vietnam.
Our research team will develop 'whole genome sequencing' tuberculosis resistance testing in Vietnam's national tuberculosis reference laboratory in Hanoi. We will investigate whether the method can rapidly and accurately predict susceptibility or resistance to all anti-tuberculosis drugs from a culture of the bacteria, or directly from patients spit. We will do so in patients who are strongly suspected to have MDR tuberculosis. We will also investigate whether the method can be used to predict new drug resistance in patients who are failing treatment for suspected drug susceptible tuberculosis. In all these cases, we aim to provide proof-of-principle that the new method will give rapid and accurate results that will ultimately improve the way patients are treated. We will also assess the economic benefits of the new approach. We hope that the information provided by the project will help support the future adoption of genetic resistance detection into Vietnam's tuberculosis treatment and control programme.

Technical Summary

Multi-drug resistance (MDR) is a growing threat to global tuberculosis (TB) control. Central to the problem is the difficulty determining drug susceptibility sufficiently quickly to guide effective therapy. Current standard laboratory methods include phenotypic (culture-based) drug susceptibility testing (DST), which requires specialist laboratories and takes 8-12 weeks. Only two laboratories in Vietnam perform full DST. Inadequate diagnosis leads to missed cases, onward transmission, morbidity and mortality, and increased costs. There is a clear need for fast, accurate DST that guides appropriate treatment.
Our proposal addresses one overarching question: can Whole Genome Sequencing (WGS) replace current methods for the DST of MDR-TB in Vietnam's tuberculosis National Reference Laboratory (NRL)?
We will address four hypotheses:
1. WGS of M. tuberculosis cultured from patients with suspected MDR-TB can predict susceptibility to second-line drugs faster and with equivalent accuracy to current phenotypic methods
2. Direct WGS of sputum from patients with suspected MDR-TB can predict susceptibility of MTB to second-line drugs with equivalent accuracy to WGS of cultures
3. WGS of sputum taken from patients with suspected drug-susceptible TB who are failing treatment (bacteria still seen in sputum by microscopy) can detect mutations associated with MDR-TB development
4. WGS generates economic benefits resulting from the earlier diagnosis and treatment of MDR-TB.
We will test these hypotheses by developing a sequencing and bioinformatics platform within the NRL. We will sequence sputum and the bacteria subsequently cultured from patients with suspected MDR-TB to address hypotheses one and two, and sputum from patients failing therapy for suspected drug susceptible disease to address hypothesis three. To address hypothesis four we will perform detailed microcostings of the different diagnostic methods for detecting MDR-TB and the associated treatment strategies.

Planned Impact

Our goal is that the proposed work contributes to building sufficient capacity in Vietnam to launch a sustainable whole-genome sequencing service in Vietnam's National Reference Laboratory for MDR-TB in the short-term, and for all TB cases in the medium-term. The expected impact of a switch to WGS-based diagnostics will be:
Direct Impacts
a) Precision diagnostics will increase the likelihood of a drug regimen containing sufficient efficacious drugs. In Vietnam, this could result in >8000 MDR-TB patients receiving correct individualized treatment per year
b) Efficacious MDR-TB treatment is a necessary, though not sufficient, condition to improve treatment success. A treatment success rate of 92%, reflecting that of non-MDR-TB, is an ambitious target and would correspond to an additional 1360 cases successfully treated (up from the 75% current treatment success rate for MDR-TB, and based on 8,000 cases per year).
c) Accurate therapy leads to reduced infectivity and less onward transmission that would break the cycle of primary MDR-TB transmission.
d) WGS will provide added value with epidemiologic data to guide public health interventions aimed at breaking the chain of transmission.
Indirect impacts:
a) By building up capacity for TB, there is the potential to share expertise with other pathogen diagnostic services such as HIV or other antimicrobial resistant bacteria.
b) Having an established WGS diagnostic system in place will increase epidemic preparedness and provide in-country expertise and technology for rapid identification of emerging diseases.
c) Vietnam would become a key site for future clinical trials of individualized vs. regimental therapies.
Economic impacts of scale-up to sustainable WGS-based services:
a) At $4,289 per episode of MDR-TB, an additional 1360 successfully treated cases each year will correspond to $5.8 million saved (see 2017 WHO TB report, p.128).
b) Economies of scale could save health systems money by linking TB diagnostics to diagnostics for other pathogens.
c) Precision diagnostics are expected to reduce treatment costs by avoiding the prescription of drugs that are unlikely to be efficacious. The amount saved is not possible to calculate due an absence of data on current drug susceptibility patterns.
d) Appropriate use of drugs will prolong their use-life and reduce dependence on newer drugs that will be more expensive whilst under patent.
e) Instituting WGS in Vietnam could contribute to its adoption in other countries in the region. That in turn is likely to further stimulate the WGS market and reduce prices.
Outcome-level impacts:
a) Expertise will have been built in Hanoi to sustain future WGS-based diagnostic services.
b) It will have been demonstrated that WGS can accurately predict 2nd and 3rd line drug susceptibility, and direct therapies. This will help inform decisions by the Vietnamese Department of Health on investing further in WGS in TB.
c) Demonstrating that WGS can be performed accurately directly from clinical samples will inform decisions around whether a WGS service can be implemented on a culture-free basis or not.
d) Identifying reasons for treatment failure will be key to informing the (re-)design of both diagnostic and treatment algorithms in Vietnam.
e) Demonstrating the economic benefits of WGS for TB will be key to any decision around routine uptake by the Vietnamese Department of Health.


10 25 50
Description Collaboration with the National Lung Hospital of Vietnam 
Organisation Vietnam National Lung Hospital
Country Viet Nam 
Sector Hospitals 
PI Contribution We work with the NLH and NTP to conduct the study - they are the primary partners for the project. We provide the technical advice and training on whole genome sequencing of Mycobacterium tuberculosis, and helped design and implement the clinical study associated with the project
Collaborator Contribution They provide the laboratory space and facilitate the study within the national TB programme.
Impact No outputs yet, in the first year of the project
Start Year 2019