The Paediatric Primaquine Pharmacokinetic (PK) Study Group aims to assess the PK properties and optimal dosing of primaquine in paediatric patients, with a focus on children under 5 years of age. The work uses nonlinear mixed-effects modelling to characterise primaquine exposure with recommended dosing and to explore how age and body weight influenced PK.

The study will generate evidence to support the development of an age- and weight-based optimised dosing of primaquine in children to improve safety, tolerability, efficacy and adherence.

8-aminoquinolines, notably primaquine and tafenoquine, are the only available antimalarials that kill dormant liver stages (hypnozoites) of Plasmodium vivax and P. ovale. G6PD testing prior prescription is required. Tafenoquine is contraindicated in G6PD-deficient individuals, resulting in increased haemolysis risk in those with intermediate G6PD activity. Primaquine is widely used and also increase the risk of haemolysis in G6PD-deficient patients but can be given as an extended weekly regimen in this group.

Because of the dormant liver stage and resulting  relapses, P. vivax and P. ovale pose a greater barrier to malaria elimination than P. falciparum. The main burden of malaria lay in young children, yet no suitable paediatric primaquine formulation is available, and paediatric dosing remained unclear. Existing age-based and weight-based schemes led to variable exposure, and pharmacokinetic data: especially in children and in P. ovale malaria were limited. An individual patient data (IPD) meta-analysis has been conducted to inform optimal dosing of primaquine in children. 

The IPD meta-analysis aimed to describe the population pharmacokinetic properties of primaquine in children and to compare them with older children and adults by:

• Characterising population PK properties of primaquine and carboxyprimaquine (absorption, distribution, elimination).
   
• Evaluating the impact of age and body weight on primaquine pharmacokinetics.
   
• Optimising dosing of primaquine in different age and weight groups, particularly in children under 5 years, compared to typical adults (body weight 60 kg and 20 years old)

Primary endpoints covered model development for primaquine and its absorption, estimation of inter-individual variability, covariate effects (including age and body weight), characterisation of carboxyprimaquine PK, and dose optimisation in children under 5 years through population simulations.

Secondary endpoints included PK in key sub-populations (for example, by disease state, CYP2D6 genotype, malnutrition or drug–drug interactions, where data allowed), optimised dosing in groups at risk of mis-dosing, and joint drug–metabolite modelling of primaquine and carboxyprimaquine.

The analysis included individual patient data from studies that:

• Collected primaquine pharmacokinetic data (sampling times and concentrations of primaquine, carboxyprimaquine and any other available metabolites), with sample matrix identified.
   
• Recorded individual dosing information (dose, start day, duration and dosing times).
   
• Captured core covariates, including age and body weight, with additional admission covariates and co-administered drugs where available.
   
• Had ethical approval in the country of origin and provided fully anonymised data that could not be traced back to individuals.
   

Individuals without dosing time information were excluded. Major outliers were first included, then omitted in sensitivity analyses, and the impact of their exclusion was reported.

Individual patient data from all eligible studies were pooled into a common database containing:

1. Pharmacokinetic data (concentrations of primaquine, carboxyprimaquine and other metabolites, plus sample times and matrix).
    
2. Dosing data (dose, schedule, and dosing times).
    
3. Covariate data (age, body weight, disease status, and other admission covariates, including co-administered drugs).
    

Concentrations from different matrices (plasma, whole blood, capillary dry blood spots) were harmonised by labelling matrix type and estimating conversion factors. All data above the limit of quantification were included. Below-quantification data were flagged and analysed with appropriate methods.

Co-administered drugs were incorporated as covariates to assess possible drug–drug interactions where data permitted. All admission covariates were tested on biologically plausible PK parameters, and missing covariates were imputed using typical population estimates. Population pharmacokinetic analysis followed standard protocols and EMA/FDA guidelines, using nonlinear mixed-effects modelling and population-based simulations to explore and refine dosing, particularly in children under 5 years.

Participating investigators agreed to the IDDO terms of submission, which ensure that all data uploaded are de-identified and obtained with informed consent, and in accordance with any laws and ethics committee approvals applicable in the country of origin. 

Data sets uploaded to the IDDO repository were standardised using the IDDO Clinical Data Management and Pharmacometric Analysis Plan and IDDO-CDISC (infectious diseases data observatory – clinical data interchange standards consortium) nomenclature into quality-assured IPD sets. 

Meta-data including study design, study sites, the methodology (e.g. pharmacokinetic assay) were also recorded.

*Correspondence: Prof. Joel Tarning, e-mail: joel.tarning@ndm.ox.ac.uk

                              Dr. Kanoktip Puttaraksa, e-mail: kanoktip@tropmedres.ac