The finding matters because the diagnostic backbone of tuberculosis control in high-burden countries is built around rifampicin. Rapid molecular tests such as Xpert MTB/RIF detect mutations in the rpoB gene that confer rifampicin resistance, and a rifampicin-susceptible result is generally treated as a green light for standard first-line therapy. When isoniazid resistance flies under that radar, patients may receive regimens that are partially ineffective, promoting treatment failure, relapse, and the further accumulation of resistance mutations. The study’s authors argue that the predominance of Hr-TB in western Kenya means rifampicin-based diagnostics risk systematically under-detecting a substantial share of drug-resistant disease.

To look beneath the surface of routine testing, the team selected an enriched subset of 316 isolates for whole-genome sequencing, of which 290 yielded high-quality genomes. The sequencing workflow was technically thorough: genomic DNA was extracted with a modified cetyltrimethylammonium bromide method optimized for the tough mycobacterial cell wall, libraries were prepared with the NEBNext Ultra II FS kit, and paired-end 2 by 150 base pair reads were generated on the Element AVITI platform. Bioinformatic processing used the DuffyNGS pipeline, aligning reads to the H37Rv reference genome with Bowtie2 and calling variants with Samtools and BCFtools, while TBProfiler annotated resistance determinants against the World Health Organization mutation catalogue, version 2.

The genomic data confirmed the classical resistance mechanisms. Rifampicin resistance was driven mainly by the rpoB S450L substitution, the canonical mutation in the RNA polymerase gene that most molecular assays are designed to catch. Isoniazid resistance was dominated by katG S315T, which disables the catalase-peroxidase that activates the drug, together with mutations in the inhA promoter, including the -15 C>T and -8 T>A variants, which alter expression of the enzyme targeted by both isoniazid and ethionamide. Pyrazinamide resistance mapped largely to pncA mutations such as L172P, ethambutol resistance to embB M306V, streptomycin resistance to rpsL K43R, and fluoroquinolone resistance to gyrA mutations. Notably, no resistance-conferring mutations were found for newer or repurposed agents such as bedaquiline and clofazimine, and all isolates were genotypically susceptible to the injectables amikacin and kanamycin.

Among the 290 sequenced isolates, whole-genome sequencing identified resistance-associated mutations in 43 cases, or 14.8 percent. Within this enriched subset, Hr-TB was again the most frequent resistance class at 16 isolates, followed by rifampicin-resistant tuberculosis at 11 and multidrug-resistant tuberculosis at 9, with three isolates classified as pre-extensively drug-resistant based on fluoroquinolone resistance mutations layered onto multidrug resistance. The sequenced population skewed male, at 72 percent, with a median age of 34 years, and 36 percent of patients were living with HIV, reflecting the heavy HIV-TB co-infection burden of the region.

Geographically, the study revealed striking heterogeneity. Resistant isolates clustered heavily in the Lake Victoria basin, with the highest counts in Kisii, Homa Bay, Migori, and Siaya counties, while Bomet and Bungoma contributed none. Kisii alone accounted for 12 of the 43 WGS-detected resistant isolates, or 27.9 percent. The authors suggest this concentration points to localized transmission or shared epidemiologic drivers within the basin, and they emphasize that such county-level variation argues for granular, local genomic surveillance rather than reliance on national averages that can mask hotspots.

The genetic population structure of the isolates was diverse. Susceptible isolates were spread across Lineage 4, the globally distributed Euro-American lineage, at about 40 percent, Lineage 3, common in East Africa and South Asia, at about 38 percent, and Lineage 2 at about 21 percent, with a single Mycobacterium bovis isolate also identified, a reminder of potential zoonotic transmission through cattle or unpasteurized dairy. Among resistant isolates, Lineage 4 and Lineage 3 predominated, though no statistically significant association between lineage and resistance emerged, suggesting that drug resistance is arising across multiple co-circulating strain populations rather than within one dominant clone. Resistance appeared particularly prominent within sub-lineages 4.3.3 and 4.6.1.2, although the authors caution that small sample sizes make some sub-lineage estimates unstable.

The most consequential result, however, was the discordance between routine testing and sequencing. Routine methods, either line probe assays or phenotypic drug susceptibility testing in liquid culture, flagged 56 isolates as resistant, while WGS called only 43 resistant. Concordant resistance was seen in 35 isolates, but 21 isolates were resistant by routine testing yet genomically susceptible, and 8 were resistant by WGS yet susceptible or untested by routine methods. A McNemar’s test with continuity correction confirmed the difference was statistically significant, with p equal to 0.026. Critically, re-evaluation against the WHO catalogue showed that 18 of the 21 routine-positive, WGS-negative cases carried mutations classified as being of uncertain significance, meaning the global reference database simply cannot yet interpret them confidently.

This discordance cuts both ways and exposes the limits of each method. For drugs such as pyrazinamide and ethambutol, resistance mechanisms are incompletely defined, and phenotypic testing itself suffers from technical variability in critical concentrations and pH sensitivity. Biological factors such as heteroresistance, transient efflux pump activity, and altered cell wall permeability can produce phenotypic resistance without any detectable resistance mutation, and culture-based sequencing may itself select subpopulations that do not represent the full within-host diversity. Conversely, the eight WGS-positive, routine-negative isolates included genuine discordant cases involving rifampicin, isoniazid, fluoroquinolones, and capreomycin, plus two streptomycin resistance mutations that routine testing had never evaluated, showing that sequencing can also catch resistance that conventional workflows miss.

The authors conclude that whole-genome sequencing and phenotypic testing should operate as complementary tools rather than competitors, with sequencing providing genome-wide, high-resolution mutation detection while phenotypic assays confirm resistance where genomic interpretation remains uncertain. Their recommendations are concrete: integrate WGS into Kenya’s national tuberculosis surveillance, beginning at central reference laboratories, and prioritize better detection of isoniazid resistance given the prominence of Hr-TB. They also stress that continued generation of local genomic and phenotypic data from high-burden settings is essential for refining the global mutation catalogues on which genomic resistance prediction depends. The study’s limitations, including its retrospective design, referral-laboratory sampling that enriches for resistant cases, and the absence of SNP-based transmission clustering, mean the proportions reported describe the study population rather than true regional prevalence. Even so, the message is clear: in western Kenya, a hidden epidemic of isoniazid resistance is circulating beneath a rifampicin-focused diagnostic screen, and closing that gap will require both wider sequencing capacity and a more complete global map of what tuberculosis resistance mutations actually mean.

Subject of Research: Genomic surveillance of drug-resistant Mycobacterium tuberculosis in western Kenya

Article Title: Genomic characterization of drug-resistant Mycobacterium tuberculosis complex in western Kenya reveals predominance of isoniazid resistance and discordance with routine susceptibility testing

Article References: Musau, S., Odera, S., Musyoki, V. M., Lamont, E. I., Onyango, N., Morrison, B., Murithi, W., Wandiga, S., Nduba, V., Sherman, D., & Mureithi, M. W. (2026). Genomic characterization of drug-resistant Mycobacterium tuberculosis complex in western Kenya reveals predominance of isoniazid resistance and discordance with routine susceptibility testing. Frontiers in Tuberculosis, 4, Article 1855780. https://doi.org/10.3389/ftubr.2026.1855780

Image Credits: AI Generated

DOI: 10.3389/ftubr.2026.1855780

Keywords: tuberculosis, whole-genome sequencing, drug resistance, isoniazid, rifampicin, Mycobacterium tuberculosis, Kenya, Hr-TB, diagnostics, WHO mutation catalogue, genomic surveillance, line probe assay