Half of the world’s population are children, yet many of the most important developments in drug therapy ignore the needs of this most vulnerable population. With Mycobacterium tuberculosis as the cause of one of the most important chronic infectious diseases worldwide,¹ it is gratifying to see research that describes the integration of modern molecular techniques to improve the delivery of the most important component of tuberculosis treatment for the world’s children. The study by Schaaf and colleagues,² reported in this issue of the journal, focuses our attention on isoniazid therapy which is still the mainstay of tuberculosis therapy for both adults and children.

For most drugs, children need specific doses, usually defined in terms of body weight. Extrapolation of the dose from adult studies by body weight would result in the under-dosing of many drugs, as children are proportionally more efficient at clearing the drugs.³ Isoniazid is a good example. The study by Schaaf et al highlights the fact that not only are higher doses required in childhood to achieve the same levels as adults, but also that children are as metabolically heterogeneous as adults. Correct dosage is important to maximise the chances of effective treatment, and to minimise the chances of adverse effects, both serious and minor. Adherence in tuberculosis treatment of children is poor⁴ and may be improved by minimising adverse effects of therapy.

Where a drug is being used for the treatment of infections, the appropriate efficacious dose is defined by the characteristics of the organism as well as host factors such as absorption. Drug exposure must be adequate to control or eliminate the infection.^5,6 Adverse effects are host dependent and may be dose related or idiosyncratic. Many of the serious side effects associated with isoniazid, such as hepatotoxicity, although thought to be idiosyncratic, may be dependent to some extent on the dose chosen and the serum drug levels achieved.^7,8 Fortunately, serious side effects, such as dose related peripheral neuropathy are less common in children than adults.⁹ However, children who are severely malnourished may be at increased risk of side effects; especially those involving the nervous system.¹⁰ This is particularly important in the treatment of tuberculosis, as children with tuberculosis are often severely malnourished at the time they start treatment, as were almost half of the children in the Schaff et al study. In these cases, supplementation with pyridoxine rather than dose adjustment may be more important as this has been shown to reduce the risk of toxicity.¹⁰ Tolerability is dictated by dose related minor side effects such as nausea, vomiting, and anorexia. These may not require withdrawal of the drug but can influence adherence to a regimen, and could therefore influence efficacy.

The current approach to dosing children is procrustean (see box) with a one size fits all approach. WHO and the International Union Against Tuberculosis and Lung Disease (IUATLD) recommend, for all patients regardless of age, a dose of 5 mg/kg/day where the drug is given daily and 10 mg/kg/day where the drug is given three times a week. Higher doses have been recommended in severe disease.¹¹ The study of Schaaf et al presents a strong argument that dosage should be tailored based on age and possibly genotype. Pharmacogenetics has identified differences in drug clearance based on the genetic profile of the metabolising enzymes; this is most relevant for drugs that are primarily cleared by hepatic elimination.¹² The CPY450 pathways have been the most extensively investigated and reported upon. However, for INH, it is phase II acetylation by N-acetyltransferase 2 (NAT2) which is the rate limiting step in drug elimination.^13,14 The enzyme is subject to genetic polymorphism, which results in fast (FF), slow (SS), and intermediate (FS) metabolisers. Schaaf et al investigated a group of 64 children less than 13 years of age with a standard 10 mg/kg dose of isoniazid. They showed that, following a single dose of the drug, the slow metabolisers had concentrations at two hours approximately double those of the fast group and total drug exposures over three times those of the fast group. The intermediate group had a profile that was between the two, although the total exposure to the drug (as measured by the area under the drug concentration by time curve) was closer to the fast group. These differences were primarily related to differences in the drug elimination. While there were statistically significant differences between groups, there is still considerable overlap between the fast, intermediate, and slow populations as shown in fig 1. This indicates that there are other important determinants of elimination, such as age, nutritional status, disease burden, and the interactions with other drugs. There may also be random individual variation not explained by such factors. Thus, the genotype may not necessarily always be a good predictor of the phenotype. As fig 1 indicates, the combined curve showing the AUC measures for children in all three genetic groups is biphasic rather than triphasic. The FF and FS genotype have a similar phenotype, while those with the SS genotype have greater AUC values. The combined data reveals a bimodal distribution, with considerable overlap between the two groups. Through the use of modern technology, particularly microarray technology, it should be possible to develop a rapid test to assist physicians in the management of a child (or adult) with tuberculosis.¹⁵ The question that remains is, would such a step be a useful addition to the management of tuberculosis?

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 Distribution of AUC values based on genotype. This graph is based on the isoniazid AUC distribution data from Schaaf et al.

In children, metabolic rate is determined mainly by size. Traditionally this has been interpreted as a function of liver size such that younger children have more liver per body weight. Often allometric scaling is used to define the relation between size and metabolic processes.¹⁶ This relation states that the metabolic rate of an organism is proportional to (weight)^3/4:

Drug elimination ∝ Wt^¾

Several pharmacokinetic studies have shown that this relation holds for drug metabolism in children such that dose scales to (body weight)^3/4 better than body weight alone.^17,18 This may also explain why body surface area also scales well for dose as the commonly used equation approximates the allometric model.

Tuberculosis, like most infectious diseases, is predominantly a disease affecting developing countries. The World Health Organisation is the authority on which most developing countries rely for guidance on the use of drugs and vaccines. For good reasons, WHO always endeavours to produce recommendations that are practical and simple to apply in the field, thus the Procrustean approach. For many drugs this can be achieved and despite the competing influences that affect the dose-effect equation, most drugs can be adequately administered using a one size fits all approach for adults and a simple age or weight related schedule for children. Only a few drugs require dose adjustment based on therapeutic drug monitoring as, for most drugs, the therapeutic window is wide enough to ensure that the minimum effective dose is well below the toxic dose.¹⁹ The need for monitoring drug levels would make the use of a drug virtually impossible to implement in the context of the developing world, due to the costs involved and the complexity of the test and its interpretation. INH appears to be in an intermediate group between these two extremes. The currently recommended dose in children of 10 mg/kg treats the vast majority of children adequately without an unacceptably high rate of side effects. However, there are some dose dependent adverse effects, and identifying a group of children likely to have higher drug levels, and therefore at higher risk of side effects may improve treatment adherence and safety, as this group would be adequately treated with a lower dose. Identification of this group may also enable the use of a larger standard dose in children. Thus, knowledge of the patient’s metaboliser status could further optimise INH therapy.²⁰ While the technology to perform this is presently available, it is currently too expensive and inconvenient.^21,22 It is conceivable that in the not too distant future polymorphisms at a single locus could be defined by a simple bedside test on blood, urine, or cheek swab. To know whether this is a concept worth pursuing we need to know much more about the dose related toxicity of isoniazid among malnourished children living in rural areas of the developing world, at the current dose, and possibly at a higher dose. If there is significant dose related toxicity, there would be a strong argument to develop this as a bedside test for the developing world to improve the safety and acceptability of tuberculosis treatment. Isoniazid may be the ideal candidate drug to investigate the impact of the widespread application of a simple pharmacogenetic test on clinical outcomes in a whole world environment.

WHO treatment guidelines reach into all corners of the world in the form of an army of health workers and countless guidelines and training materials developed at country level. Any change in the guidelines is difficult and expensive and creates the risk of confusion in the field. Although Schaff et al do not specifically recommend a change in the guidelines, the fact that almost half the children in their fast acetylator group do not achieve the recommended levels at two and three hours raises the possibility that the dose for children under 5 years of age should be raised. WHO’s Tuberculosis Control Programme has paid little attention to the problem of childhood tuberculosis, despite growing evidence that the disease is under-diagnosed in children. The paper by Schaff et al should prompt WHO to undertake a careful review of isoniazid dosage in children. Perhaps it is time for a randomised controlled trial comparing the current dosage with a higher dose in children, examining toxicity, tolerability, and perhaps effectiveness. Based on the results of such a trial, consideration should then be given to the need for a rapid test to identify the group of slow acetylators who may be at higher risk of toxicity. Such developments could lead to fewer side effects, better adherence to therapy, and possibly more effective therapy.