Occult rifampicin-resistant tuberculosis: better assays are needed

Drug-resistant tuberculosis remains an important public health concern. A 2017 meta-analysis of published studies from India suggested that more than 40% of all Mycobacterium tuberculosis isolates included in the analysis were multidrug-resistant (MDR). Drug resistance is associated with high mortality despite treatment. Furthermore, prolonged therapy with multiple anti-tuberculosis drugs is onerous both for patients and the health system. Current WHO-endorsed diagnostic tests for tuberculosis include Xpert MTB/RIF (Cepheid, Sunnyvale, CA, USA) and GenoType MTBDRplus (Hain Lifescience, Nehren, Germany); however, these assays do not detect mutations outside of the 81-bp rifampicin-resistance-determining region (RRDR) of the rpoB gene of M tuberculosis, which contains 95% of known rifampicin-resistance-associated codons.

Furthermore, the current standard resistance assay used to validate novel molecular assays is the automated mycobacteria growth indicator tube 960 liquid culture system (BD, Baltimore, MD, USA), which has been reported to miss mutations associated with rifampicin resistance outside of the RRDR. An important study by Ndivhuho A Makhado and colleagues in The Lancet Infectious Diseases reports the prevalence of previously undetected rifampicin-resistant tuberculosis. The group used M tuberculosis strains with isoniazid monoresistance, diagnosed by WHO-endorsed assays, and tested these isolates for the Ile491Phe mutation, which confers rifampicin resistance but is not detected by the assays currently endorsed by WHO. The authors randomly selected 277 (15%) of 1823 isoniazid-monoresistant isolates and compared these with strains from a survey in eSwatini (formerly Swaziland) in which 30% of MDR tuberculosis isolates had the Ile491Phe mutation.

Using deep sequencing (Deeplex-MycTB) and whole-genome sequencing, the authors were able to genotype, predict drug resistance, and phylogenetically analyse the isolates. The Ile491Phe rifampicin-resistance-associated mutation was detected in 37 (14%) of the South African isolates, thereby reclassifying them as MDR. Additionally, isolates from both eSwatini and South Africa were resistant to all first-line drugs according to Deeplex-MycTB. Ile491Phe was associated with worse prognosis, with five-times increased odds of the mutation being present in cases of unfavourable treatment outcomes compared with favourable outcomes.

In South Africa (and probably elsewhere), WHO-endorsed molecular assays are used for diagnosis of M tuberculosis and the detection of rifampicin resistance. In routine practice and in accordance with dogma that accepts that rifampicin resistance indicates multidrug resistance, additional tests assessing isoniazid sensitivity are only requested if rifampicin resistance is detected; if a sample is reported as rifampicin sensitive, isoniazid sensitivity is not routinely requested.

The increased rollout of isoniazid preventive therapy in countries with a high HIV burden and the increased prevalence of isoniazid monoresistance reported in surveys from South Africa and Botswana make the risk of undetectable rifampicin resistance alarming. Isoniazid resistance appears to be a precursor for multidrug resistance.

More disturbing was the detection of Rv0678 mutations, implying bedaquiline and clofazimine resistance, among a proportion of isolates with the Ile491Phe mutation, which could result in total drug resistance, although the clinical significance of Rv0678mutations is uncertain. Notably, Rv0678 has been found in areas without bedaquiline exposure.

Although the authors suggest a link to South Africa’s tuberculosis control programme, which established a compassionate bedaquiline access programme replacing injectables (kanamycin and amikacin) in selected individuals, only 200 patients received bedaquiline in 2013. The programme’s regimen includes clofazimine, which makes testing for Rv0678 crucial. The authors report two distinct strains harbouring Ile491Phe, the eSwatini strain (ST34) found both in South Africa and eSwatini, and a newly described South African strain (ST94) found among specimens only from South Africa. ST94 is of concern because it was previously unrecognised and suggests independent evolution of this strain.

Clearer guidance and algorithms for isoniazid testing and subsequent screening for Ile491Phe, especially when isoniazid resistance is present, is needed. Testing for Rv0678 mutations and Ile491Phe should be included in routine drug-resistance surveillance programmes and, if high prevalence is found, as reported by Makhado and colleagues, detection for these mutations should become part of clinical care. Although whole-genome sequencing is expensive and currently unaffordable for routine surveillance in low-resource settings, current or novel PCR techniques that can identify existing and new drug-resistance-conferring mutations will be important tools for understanding the prevalence of such mutations. Ideally, these diagnostics should be available for all countries, and efforts should be made to aggressively reduce cost and simplify laboratory processes so that those most in need can benefit.