Isoniazid-resistant tuberculosis treatment with first-line drugs

We read with interest the systematic review and meta-analysis by Medea Gegia and colleagues1 of use of first-line drugs for isoniazid-monoresistant tuberculosis. Among the main findings, this study revealed that 11% of new patients who received the first-line WHO standard regimen had treatment failure. On the basis of these observations, the authors suggest that standardised empirical treatment of new cases could be contributing substantially to the multidrug-resistant tuberculosis epidemic. However, the authors do not question the reliability of the initial bacteriological diagnosis that was made for these patients, which supported the clinical decision to initiate standardised first-line treatment. Most studies referenced in the systematic review and meta-analysis used phenotypic drug susceptibility testing methods, which fail to detect a substantial proportion of rifampicin-resistant strains.2 Of the 52 papers cited, only two used molecular methods for establishment of rifampicin resistance. None of the studies included sequencing of the rpoB gene and all were published between 1968 and 2014. Only since 2014 have reports originating from different regions of the world highlighted the fact that several specific mutations were regularly undetected and thus responsible for missed opportunities to diagnose multidrug-resistant tuberculosis.3, 4, 5 To date, the most alarming figures originated from Swaziland, where a third of patients with multidrug-resistant tuberculosis had the rpoB Ile491Phe mutation,6 which is hardly detected by phenotypic drug susceptibility testing methods and missed by commercial molecular assays. In this context, assessment of the efficacy of first-line regimens among patients for whom whether they have isoniazid-monoresistant or undetected multidrug-resistant tuberculosis is of questionable value. In most studies used for this systematic review and meta-analysis, the frequency of negative clinical outcomes is similar to the expected level of detection of multidrug-resistant tuberculosis when suboptimal techniques are used. We would have appreciated this important question being addressed.