Is the new WHO definition of extensively drug-resistant tuberculosis easy to apply in practice?

To the Editor:

The World Health Organization (WHO) recently endorsed a new definition of extensively drug-resistant tuberculosis (XDR-TB) and, for the first time, introduced the category of pre-XDR-TB [1]. Pre-XDR-TB is defined as multidrug resistance/rifampicin resistance (MDR/RR) in conjunction with resistance to any fluoroquinolone (levofloxacin or moxifloxacin), whereas the conditions for XDR-TB are now met by additional resistance to a group A drug (bedaquiline or linezolid) [1].

WHO also laid important scientific groundwork to support this transition (e.g. it revised the critical concentrations for phenotypic drug-susceptibility testing (pDST) of fluroquinolones and rifampicin) [2, 3]. We agree that the revisions to the definitions were needed but note that measuring resistance to these drugs comprehensively is not straightforward.

DST for fluroquinolones is an essential pre-condition for the initial selection of the most appropriate MDR-TB regimen, because of its predictive value for adverse outcomes and the high rates of fluroquinolone resistance (approximately 10–30%, depending on the setting) [4]. Therefore, fluoroquinolone resistance had already been a criterion for the old definition XDR-TB [1, 5]. Yet, despite the availability of pDST and rapid genotypic DST (gDST) solutions (figure 1), only 71% of the notified MDR/RR-TB cases globally were tested for fluoroquinolone resistance in 2019, with considerable variations among different regions [1, 6]. This is partly due to the fact that the only WHO-endorsed gDST assay (i.e. the GenoType MTBDRsl VER 2.0; Hain Lifescience) is relatively labour intensive and less reliable for direct testing of clinical samples [4]. The cartridge-based Xpert MTB/XDR (Cepheid), which is currently being evaluated by WHO, has the potential to narrow this diagnostic gap by enabling more decentralised testing [1]. Nevertheless, additional capacity for pDST and, potentially, targeted next-generation sequencing (tNGS) is needed given that a greater proportion of resistance to fluoroquinolones than rifampicin is due to low-frequency variants that are below the limit of detection of these assays [1, 6].

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Overview of options for genotypic and phenotypic drug-susceptibility testing (DST) of group A drugs for treating rifampicin-resistant tuberculosis (RR-TB). Commercial genotypic DST assays are only listed if they are approved for clinical use in at least one country (if a manufacturer has multiple assays on the market that are approved, only one is shown). Methods endorsed by the World Health Organization (WHO) are marked by ^# and the additional assay currently being reviewed by WHO is highlighted by ^¶. BDQ: bedaquiline; CC: critical concentration; FQs: fluoroquinolones; S: susceptible; LZD: linezolid.

DST for bedaquiline and linezolid is more challenging for a number of related reasons. First, no rapid gDST assays exist for these agents as advocated by the End TB Strategy, which means that gDST is only possible with either tNGS or whole-genome sequencing, neither of which is currently available in the vast majority of countries with higher MDR-TB incidences (figure 1) [1, 6].

Second, even where gDST is routinely used, the interpretation of the results is hampered by the incomplete understanding of the genetic basis of resistance and/or the impact of mutations on the minimum inhibitory concentrations (MICs) [1, 6]. This is a particular challenge for bedaquiline, for which a large spectrum of resistance mutation is possible, whereas data from other bacteria indicate that the number of variants for linezolid is likely small [7].

Third, there is a lack of capacity for pDST [1]. A recent survey conducted by the European TB Reference Laboratory Network (ERLTB-net) and coordinated by the European Centre for Disease Prevention and Control revealed that in 2019 only 61% and 32% of 28 participating TB laboratories tested for linezolid and bedaquiline, respectively (reassuringly, mostly reporting correct results) [8].

Fourth, pDST results for these drugs can also be difficult to interpret. On the one hand, the positive predictive value of pDST will be poor in settings where the true prevalence of resistance is low (e.g. in a setting with only susceptible isolates, approximately 1% of those isolates would be misclassified as resistant) [3]. On the other hand, it is becoming increasingly clear that pDST at the critical concentration does not detect mutations conferring only modest MIC increases reliably [9]. Whether such MIC increases are clinically relevant is not clear but if they are, MIC testing with a carefully validated and controlled method would be needed [3, 9, 10].

These challenges will adversely affect individual patient treatment. In addition, the rates of XDR-TB measured during surveillance studies will be strongly dependent on the method used, including the amount of retesting conducted (i.e. both over- and underreporting can be a problem for both gDST and pDST). This, in turn, will result in a worse understanding of countries’ epidemiological profile of the most dangerous form of TB; lack of appropriate global prevention and control activities, such as equitable access to universal DST and anti-TB drugs regimens; lower capacity of healthcare providers and public health authorities in implementing appropriate national strategies; and limited efficiency in allocating health resources based on countries’ shared experience [4, 5].

Given sufficient political will, the coronavirus disease 2019 pandemic has underlined that rapid technological advances are possible. There is a clear need for diagnostics in TB enabling universal DST access, and for rapid triage of people with XDR-TB. Ideally, a genome-based technology that is easy-to-use and rapid, and able to provide extensive coverage of genomic targets, is needed. WHO is due to publish updated target product profiles that reflect the revised needs for TB diagnostics. We call on assay developers, pharmaceutical companies and researchers, as well as funders and regulators, to renew their efforts to tackle the aforementioned questions in a coordinated fashion. We cannot afford to repeat the mistakes of the past and risk the rapid emergence and spread of resistance to more group A drugs, thereby eroding the hard-won gains in the treatment of MDR-TB [1, 2].

Shareable PDF