Reply to: Real-world safety signals of isoniazid in adults and children: a complementary FAERS analysis to inform paediatric dose optimisation

Extract

We have carefully reviewed the correspondence from Y. Lin and co-workers, which highlights the need for safety analysis related to the optimisation of isoniazid dosing, particularly dose increases in specific patient populations, such as N-acetyltransferase 2 (NAT2) fast metabolisers and malnourished children.

Shareable abstract

All children receiving anti-tuberculosis treatment require thorough monitoring to prevent adverse events. All pharmacokinetic studies should analyse efficacy and safety data. https://bit.ly/4b8c4K0

Reply to Y. Lin and co-workers:

We have carefully reviewed the correspondence from Y. Lin and co-workers, which highlights the need for safety analysis related to the optimisation of isoniazid dosing, particularly dose increases in specific patient populations, such as N-acetyltransferase 2 (NAT2) fast metabolisers and malnourished children.

To support their statements, they conducted a retrospective study using the US Food and Drug Administration Adverse Event Reporting System (FAERS), covering a 22-year period (2004–2025). As acknowledged by the authors, this study spans a long period, during which both tuberculosis treatment practices and adverse event reporting standards have evolved. Nevertheless, it provides useful insight into the safety profile of isoniazid in children, and this complementary investigation offers an overview of the past two decades.

We agree with Y. Lin and co-workers that pharmacokinetic analyses should ideally be associated with clinical outcome analysis, encompassing both efficacy and safety outcomes [1]. Efficacy analyses rely on pharmacodynamic data (e.g. clinical or bacteriological cure, relapse or death), and this component could be integrated into our individual participant data meta-analysis (IPD-MA) for a subgroup of patients [2]. Safety data, conversely, require either the support of a pharmacovigilance infrastructure or a rigorous clinician-led monitoring of adverse effects. This aspect was not included in our study because these data were unavailable. Importantly, the primary objective of our IPD-MA was to characterise pharmacokinetic variability and exposure determinants rather than to perform a formal safety evaluation. Both types of analyses should ideally be conducted prospectively and require medium- to long-term follow-up, which may limit feasibility and necessitate different study designs. Even though the FAERS analysis is complementary, it remains retrospective and cannot substitute for a rigorous pharmacovigilance evaluation.

Furthermore, FAERS analyses differs from causal dose–toxicity assessment. Some adverse events are dose-dependent, whereas others may occur regardless of the administered dose. Because the FAERS investigation does not account for dosing, it does not provide evidence that the optimised dosing regimen proposed in our IPD-MA would be associated with increased toxicity. The correspondence emphasises that patient subgroups requiring higher dosing regimens, such as NAT2 fast metabolisers and malnourished children, should undergo careful prospective monitoring. We believe that all children receiving anti-tuberculosis treatment require appropriate monitoring, particularly for neurological and hepatobiliary adverse events, and for increased dosing regimens [3, 4]. We acknowledge that isoniazid has often shown safety signals. Adverse effects associated with isoniazid are well-known and well-described, and monitoring is mandatory regardless of the dosing regimen [5]. Furthermore, other anti-tuberculosis drugs may also contribute to specific adverse events or potentiate existing toxicities [6]. In tuberculosis treatment, adverse effects analyses must always be interpreted cautiously because of confounding factors and the concomitant use of multiple medications. Of the four drugs that constitute drug-susceptible tuberculosis treatment, rifampicin is probably the one that should be optimised to achieve the right balance between efficacy and safety. Since higher rifampicin dosing did not show an increase in adverse events, it could be used to attain an adequate exposure [7, 8].

This correspondence also reinforces our message that dosing regimen should be individualised according to specific covariates to avoid subtherapeutic exposure. Our objective is not to increase the dose indiscriminately, as we support the current World Health Organization (WHO)-recommended dosing regimen for isoniazid, but rather to identify patients at risk of low isoniazid exposure, explain the underlying mechanisms, and propose feasible solutions when possible (e.g. NAT2 genotyping, nutritional status assessment, isoniazid drug monitoring). Moreover, we have remained very cautious in our dosing suggestion, which remained within currently recommended dosing ranges used in clinical practice. While supratherapeutic exposure carries a risk of toxicity, insufficient exposure is associated with unfavourable clinical outcomes and/or the emergence of bacterial drug resistance. As a reminder, tuberculosis remains a major global health problem and the leading cause of death from a single infectious agent worldwide. In 2024, children accounted for 11% (1.2 million) of the 10.7 million people affected by tuberculosis. During the same year, among the 1.23 million tuberculosis-related deaths, 175 000 (14%) occurred in children [9].

To conclude, all pharmacokinetic studies should incorporate clinical data collection to enable both efficacy and safety analyses. We reaffirm our support for the WHO-recommended dosing regimen and suggest that, despite the possibility of rare adverse effects, dosing strategies could be adapted to NAT2 genotype and nutritional status to minimise subtherapeutic exposure. As recommended, all children receiving anti-tuberculosis treatment require thorough monitoring to prevent adverse events, particularly hepatobiliary disorders.