This study was a part of the efficacy and safety evaluation of new artesunate and amodiaquine (AS/AQ) FDC as compared to the same drugs given separately (AS+AQ). Results of the efficacy part of this evaluation have already been reported . The trial was conducted in Burkina Faso, at the Healthcare District of Pissy between September and November 2006.
The study was approved by the Burkina Faso Ethics Committee for Health Research and the WHO Secretariat Committee on Research Involving Human Subjects and was carried out in accordance with the ICH Guidelines for Good Clinical Practice. Written informed consent was obtained from parents or guardians before patients began treatment. The trial is registered with the open clinical trial registry  under the identifier number ISRCTN07576538.
Patients of both sexes, between six months and five years of age, weighing = 5 kg with Plasmodium falciparum monoinfection of more than 1,000 parasites/μL and a measured fever (axillary temperature ≥ 37.5°C) were included in the study. Patients were excluded if they: (i) had features of severe and complicated malaria ; (ii) had taken the study drugs or other anti-malarial drugs within seven days before inclusion (or within three days if an artemisinin was used); or (iii) if they were receiving treatment with antibiotics with anti-malarial activity. A randomization list was computer generated. Individual treatment allocations were kept in sealed envelopes and opened after patients were admitted into the study. All treatments were administered by a study nurse. Fixed dose combination (AS/AQ) was administered according to a newly designed, age based dosing regimen  and loose dose combination (AS+AQ) according to the manufacturer's instructions for the Arsucam® blister pack (Sanofi-Aventis, Paris, France). For both regimens, the target doses for artesunate and amodiaquine base were 4 mg/kg/day and 10 mg/kg/day, respectively, with, therapeutic ranges of 2 to 10 mg/kg/day and 7.5 to 15 mg/kg/day .
The FDC contained 25 mg of AS and 67.5 mg of AQ. The dose was one (age <12 months) or two (age of 12 to 60 months) tablets. The loose blister pack contained 50 mg tablets of AS (Arsumax®, Sanofi-Aventis) and 153 mg tablets of AQ (Flavoquine®, Sanofi-Aventis). The dose was half or one tablet of both drugs according to age, as described above. Patients remained in the health centre for an hour following each treatment administration. Vomiting during this one hour observation period resulted in re-administration of the same dose of the study drugs.
Following the first visit, patients were seen after 24, 48, and 72 hours, 7, 14, 21 and 28 days for clinical (symptoms, temperature, adverse effects) and parasitological assessments. Parasite density was determined by counting the number of asexual parasites per 200 leucocytes on a Giemsa-stained thick film and expressed as the number/μL, assuming a leukocyte count of 8,000/μL. If up to 500 parasites were counted before reaching 200 leucocytes, the counting process was stopped at the end of the last field. Gametocytes were counted and expressed as the number per 1,000 leucocytes (thick film). Routine haematology (Pentra 60®) and biochemistry (Hospitex Screen Master Tecno®) blood samples were taken and analysed on Days 0, 7 and 28.
Patients failing or not tolerating their treatment were withdrawn from the study and rescued with 25 mg/kg/day of oral or parental quinine base in three divided doses for 7 days. There was a systematic investigation of all patients lost to follow-up.
The primary efficacy endpoint was the PCR corrected parasitological cure, assessed by Kaplan Meier survival analysis. The criteria for treatment failure followed broadly those of the WHO : (i) signs of severe malaria or danger signs at any time during follow up, (ii) parasitaemia at Day-2 greater than parasitaemia at Day-0, (iii) Day-3 parasitaemia greater than or equal to 25% of parasitaemia measured on Day-0, (iv) parasitaemia on Day-7, and (v) a recurrent parasitaemia after Day-7 that was a PCR proven, recrudescent parasitaemia.
Patients with recurrent parasitaemia after Day-7 were classified as failure or new infections by analysing sequentially by PCR polymorphic segments of three parasite genes: first the merozoite surface protein (MSP) 2, then MSP 1 then glutamate rich protein (GLURP), according to a previously published method . A new infection was diagnosed if the allelic pattern for any one of the loci differed completely between the baseline and recurrent samples. All other allelic patterns were diagnosed as a resistant infection.
Safety of treatment was assessed by collecting clinical (symptoms, signs) and routine laboratory data during follow up. The standard International Conference on Harmonisation (ICH) definitions of an adverse event and a serious adverse event were used . Adverse events criteria were determined using the Common Toxicity Criteria (US NIH, CTC v2 1999) and graded as mild (1), moderate (2), severe (3) or very severe (4).
Sample size and sampling
A sample size of 140 children was deemed adequate for the population PK model given the limitation of taking blood from small children. Children were divided into two groups of 70 (35 in each treatment arm). The first 70 recruited children formed the AS pharmacokinetics group. All children were sampled three times: (i) before the first dose, (ii) after the first dose, randomly at either 0.5, 1, 2 or 4 hours, (iii) after the third dose, randomly at either 0.5, 1, 2 or 4 hours. The second group (patients 71 to 140) participated in the study of AQ pharmacokinetics. All children were sampled four times: (i) before the first dose, (ii) at 4 hours after the 3rd dose, (iii) on Day-7 or 14 of the follow-up period, (iv) on Day-21 or 28 of the follow-up period.
Each time, a sample of 1 mL of venous blood was collected in lithium heparin tubes. Plasma was separated immediately by centrifugation and stored frozen at approximately -20°C until sample analysis within three months.
The concentrations of AS and AQ and their respective pharmacologically active metabolites dihydroartemisinin and desethylamodiaquine in plasma samples collected during the study, were determined using sensitive and specific Liquid Chromatography/Mass Spectrometry/Mass Spectrometry (LC-MS/MS) methods developed and validated by PAREXEL International.
Artesunate, dihydroartemisinin and the internal standard (artesunate-D4) were separated from human plasma by solid-phase extraction (SPE) using Oasis HLB C18 extraction cartridges and analysed by reversed-phase LC-MS/MS in the Turbo Ion Spray positive mode. The assay was carried out using a 200 μL sampling volume of human plasma and the lower limit of quantification (LLOQ) of the LC-MS/MS method was 1.0 ng/mL for both artesunate and dihydroartemisinin. The coefficients of variations (CV%) during the analysis of artesunate and dihydroartemisinin were 11.3% and 6.5% at low (3 ng/mL), 3.4% and 5.0% at medium (100 ng/mL) and 3.9% and 9.0% at high (200 ng/mL) concentrations.
Amodiaquine, desethylamodiaquine and the internal standard (Amodiaquine-D10) were separated from human plasma by a different solid-phase extraction (SPE) using Oasis HLB C18 extraction cartridges and analysed by reversed-phase LC-MS/MS in the Turbo Ion Spray positive mode. The assay was carried out using a 200 μL sampling volume of human plasma and the lower limit of quantification (LLOQ) of the LC-MS/MS method was 1.0 ng/mL of amodiaquine and desethylamodiaquine. The CV% during the analysis of amodiaquine and desetylamodiaquine were 5.9% and 8.7% at low (3 ng/mL) and 6.3% and 8.0% at moderate (100 ng/mL) and 6.1% and 5.3% at high (200 ng/mL) concentrations.
Sample collection during this clinical study was performed in lithium heparin tubes and did not include stabilization with phenylmethyl sulfonylfluoride (PMSF). Long term stability of amodiaquine and desethylamodiaquine in human plasma samples collected on lithium heparin without PMSF was demonstrated for a 6-month period at approximately -20°C .
Long term stability of artesunate and dihydroartemisinin under these conditions  was demonstrated after 115 days of storage at -20°C. However, human plasma samples collected during this study were stored at -20°C for up to 147 days prior to last sample analysis. This period of storage is beyond the validated 16 weeks. Consequently, the long-term stability of AS and its pharmacologically active metabolite dihydroartemisinin in human plasma with lithium heparin as anti-coagulant and without PMSF was retested and demonstrated satisfactory stability for artesunate containing samples for 173 days and dihydroartemisinin for 142 days for samples prepared as described in the study.
The effects of haemolysis on the determinations were investigated during the respective validation studies and were found to have no effect on the assay of amodiaquine and desethylamodiaquine. However, the assay of AS and dihydroartemisinin was significantly affected in an unpredictable way by haemolysis.
Therefore, the presence of evident haemolysis observed in many of the plasma samples led to their classification as not reportable (no quantitative result) for AS and dihydroartemisinin.
Amodiaquine concentrations were usually low or not detectable. The population pharmacokinetics of desethylamodiaquine were modelled using the non-linear mixed effects approach. One and two compartment models were investigated. Due to the small number of detectable samples per subject (at most 3 per subject) and the timing of samples, the absorption rate constant was fixed (as 0.13 h-1 from previously published work ) and at most two random effects which were not correlated could be fitted at the same time. Children with detectable AQ concentrations before the first dose were excluded from the analysis. Children with detectable desethylamodiaquine levels and corresponding negative AQ levels before the first dose were included in the population modelling, and a first order elimination process was assumed for these samples.
Pharmacokinetic parameters of DHA and the total anti-malarial activity, defined as the sum of the plasma levels of DHA and artesunate calculated in nmol/L units (taking molecular weight as 384.4 for artesunate and 284.9 for DHA) were also modelled (expressed as DHA equivalents) using the population approach as described previously . This assumes equal anti-malarial activity. A one compartment model with three parameters (absorption rate constant, total apparent volume of distribution (V/F) and total oral clearance CL/F) was fitted for each. It was assumed that artesunate was completely converted to DHA for determination of the dose in the pharmacokinetic analysis of DHA and of the total anti-malarial activity. Due to the small number of detectable samples (at most 2 per subject, one after the first dose and one after the third dose) only one random effect could be fitted. Inter-subject variations in clearance and volume of distribution were examined separately and in the final model, the parameter with larger variation was fitted as a random effect. The effect of the dosing period after which sampling was done was examined by including a binary covariate (0 = sampling after dose 1/1 = sampling after dose 3) and was not accounted for by the random effects structure. Inter-subject variations in clearance and volume of distribution were examined separately and in the final model, the parameter with larger variation was fitted as a random effect.
In each model, the inter-subject variability in pharmacokinetic parameters was modelled with a log-normal error structure, for example: (CL/Fi) = (CL/F) exp(ηi
CL/F), where CL/Fi is the parameter value for the ith individual, CL/F is the population mean, ηi
CL/F is the random effect with zero mean and variance σCL/F, which represents the inter-subject variability for the parameter. The magnitude of the inter-subject variability was expressed as a coefficient of variation (CV%) approximated by the square root of the variance estimate, while the residual variability was expressed as the standard deviation of the residual error. The variability in pharmacokinetic parameters was investigated by examining the effects of the following covariates: age, weight, temperature, respiratory rate, sex, logarithm of enrolment parasitaemia, presence of gametocytaemia on enrolment, drug formulation and day of measurement for DHA and total anti-malarial activity.
The log of the likelihood function was used to determine which covariates should be included in the model, in the forward selection procedure. The goodness of fit of each model was also assessed by the examination of the scatter plots of residuals versus predicted drug levels. The actual time of the sampling was used in the analysis. All compartmental analyses were performed using the S plus programme (SPLUS 6.0 for Windows, Mathsoft, Inc).
The final population models were used to calculate posterior estimates of AUC for each individual based on the actual dose received and based on the median total dose received in this population (AUC = dose/clearance). The AUCs were then compared between treatment arms by calculating their ratio and the 95% CI.
Data are summarized using medians and ranges. Continuous variables were compared between treatment groups using the Mann-Whitney test and categorical variables were compared using the chi-square test or Fisher's exact test, as appropriate.
Cure rates were estimated using the Kaplan-Meier method. Patients who developed new infections during the follow-up or were lost to the follow-up were censored in the analysis at the last visit. Patients with recurrent parasitaemia and no PCR results were excluded from the analysis. The difference in cure rates between treatments was calculated and the 95% confidence interval was estimated using Newcombe's method  and effective sample size .