Resources
Pretomanid Resistance Is a Question of When, Not If, and Diagnostics Have to Be Ready Before It Arrives
A new systematic review and meta-analysis pools seventeen South African studies published between 2013 and 2025 to answer a question every drug-resistant TB programme is now asking: how durable are linezolid (LZD) and pretomanid (Pa), the drugs that make BPaL-based regimens work?
The headline finding is reassuring. Pooled baseline linezolid resistance across routine cohorts was 0.53% (95% CI 0.01–1.83), and no confirmed phenotypic pretomanid resistance was identified in any included South African cohort, across 738 isolates with MICs consistently inside the susceptible range. Treatment success reached 90% for pretomanid-containing regimens. On current evidence, these regimens work, and they should continue to be used.
Read more closely, though, the paper is a warning rather than an all-clear.
The genotype arrives before the phenotype
Although no phenotypic pretomanid resistance was detected, genomic surveillance studies in the same setting already describe mutations across the prodrug activation pathway — ddn, fbiA, fbiB, fbiC, fbiD and fgd1. As the authors put it, the genetic pathways to resistance already exist, even in the absence of phenotypic expression. For linezolid the picture is further along: resistance clusters in rplC (recurrently Cys154Arg) and rrl, and while baseline resistance sat near zero, a treatment-failure cohort showed 33% resistance. That gap between 0.5% and 33% is the whole story.
Resistance to these agents is being acquired under drug pressure during treatment, not transmitted at baseline. The more successfully a programme scales a regimen, the faster it manufactures the selective pressure that erodes it.
Every TB drug that has ever worked has followed this curve. Pretomanid will not be the exception. The only variables are how quickly, and whether anyone is measuring when it happens.
Detection sensitivity is itself a diagnostic problem
The review’s meta-regression is quietly the most important result in it. Genomic and WGS-based studies predicted a resistance prevalence of 1.71%, against 0.16% in non-genomic studies. Routine phenotypic methods are under-recognising early resistance signals. The authors conclude that integrated phenotypic and molecular surveillance is essential to detect resistance amplification before clinical failure occurs.
The reassuring prevalence figures are partly a statement about our instruments, not about the bacillus. Genomic studies predicted resistance at 1.71%, against 0.16% in non-genomic studies — routine phenotypic methods are under-detecting the earliest signals.
Sequencing is the right answer at a price programmes cannot pay
Whole-genome sequencing is unambiguously the most complete tool for this job, and nobody serious argues otherwise. But cost per isolate, instrument capital, sample logistics, bioinformatics capacity and turnaround time keep it a reference-laboratory activity. South Africa has among the strongest sequencing capacity on the African continent, and even there this review had to assemble a national picture from fragmented research datasets rather than routine programmatic data. In the countries carrying the heaviest burden, across India, Indonesia, Nigeria and the Philippines, sequencing every isolate is not a deployable strategy this decade.
What deployable actually means
The practical response is targeted molecular drug-susceptibility testing that runs on qPCR instruments laboratories already own, at a cost per test a national programme can absorb, with minimal hands-on time and throughput that suits district-level volumes rather than reference-centre ones. Crucially, it has to be modular: panels must be extendable as regimens change, so that testing tracks the drugs actually being prescribed. The second-line injectables are the cautionary example, assays still report resistance to agents that no longer define pre-XDR & XDR, while the drugs that decide today’s treatment go untested.
Sequencing is the most complete way to watch resistance emerge, and the least deployable where the burden is heaviest. Closing that gap, targeted, modular molecular DST on instruments laboratories already own, is the diagnostic problem worth solving before pretomanid resistance arrives.
The window for building that capacity is now, while resistance is still rare and while the surveillance systems that will detect it can be designed rather than improvised. By the time pretomanid resistance is visible with routine phenotypic methods, it will already be established.
Reference
Kaapu, K.G.; Makondo, V.T.; Conceição, E.C.; Rukasha, I. Resistance to Linezolid and Pretomanid in the Era of Modern Drug-Resistant Tuberculosis Treatment in South Africa: A Systematic Review and Meta-Analysis. Antibiotics 2026, 15, 543. https://doi.org/10.3390/antibiotics15060543