Data sources

The data displayed in WWARN Explorer preview was generously contributed by colleagues during the early stages of WWARN, so that our Informatics Module could design a data repository system and interactive tools to visualise the data. The data in the preview version, drawn from more than a hundred studies, have undergone careful statistical analyses to allow for comparisons between studies. More detail on our methodologies is provided below by module: Clinical, Pharmacology, In vitro and Molecular.

The data currently featured in WWARN Explorer are only a small portion of the large body of research that actually exists on antimalarial resistance. If you're interested in contributing data, please visit our Contributing Data section to learn more about the benefits of contribution and our commitment to data contributors or contact us info@wwarn.org.

Clinical methodology

In any individual study, the analysis of clinical drug efficacy is chosen based on the rationale of the study (for instance standalone efficacy or comparative analysis in randomised clinical trials)¹. In order to compare these diverse studies, WWARN Explorer uses a uniform approach to the statistical analysis of clinical drug efficacy data². As such, the derived parameters presented in WWARN Explorer will often differ slightly from the data presented in the original publications and reports of these studies. 

Outcome classification
All patient observations are classified according to World Health Organization (WHO) definitions³. WWARN Explorer displays the Day 28 cure rate because it is highly informative of drug efficacy and the vast majority of studies follow patients for at least 28 days.

Statistical methods
WWARN estimates of clinical drug efficacy are presented both before and after adjusting for reinfections. The latter are determined according to local study protocols by PCR analysis of the parasite isolates on admission and at the time of recurrence.

Estimates of efficacy are calculated using a modified Intention-To-Treat analysis (mITT) as described previously². The evaluable population includes all patients enrolled in the study, except that late clinical failure (LCF) and late parasitological failure (LPF) outcomes are excluded from the adjusted calculations if genotyping was unsuccessful. The unadjusted and adjusted cumulative risks of recurrence are calculated by survival analysis, using the Kaplan-Meier product limit formula⁴. Patients lost to follow-up before the end of the study period, are censored on the day on the last day of observation. For the unadjusted calculations, patients with interrupted follow-up and non-falciparum new infections are censored on the last day of observation. For the adjusted calculations, patients with new P. falciparum infections are included as censored observations.

Clinical pharmacology methodology

Pharmacokinetic results describe measured drug concentrations over time and are currently reported as a median and range, stratified by treatment, age category, study site and treatment response. Future versions of WWARN Explorer will include the effect of other patient factors — such as pregnancy, concomitant medication or comorbid disease — on antimalarial pharmacokinetics.

Clinical patients lost to follow-up or with indeterminate or missing PCR results are excluded from the current clinical pharmacology analysis. Future versions will include patients with indeterminate or missing PCR results. (See the 'Clinical methodology' section above for an explanation of the WHO treatment outcomes used in WWARN Explorer.)

In vitro methodology

Data sources
Data on in vitro susceptibility of P. falciparum parasites to antimalarials displayed in the WWARN Explorer preview was contributed by the French Centre National de Référence du Paludisme (CNRP) which is composed of three laboratories:

• CNR-palu Hôpital Bichat -Claude Bernard (Paris),
• CNR-palu IMTSSA (Marseille),
• CNR-palu Hôpital La Pitié Salpetrière (Paris).

Malaria patients presenting at French hospitals, immigrants to France and returning travellers from malaria-endemic regions were referred to CNRP, where samples were collected for in vitro susceptibility testing.

P. falciparum isolates were incubated with 5% CO₂ and 10-20% O₂. Susceptibility was measured as the concentration of antimalarial which inhibited parasite growth by 50% (IC₅₀). IC₅₀ values were obtained by either isotopic methods⁵ or ELISA pLDH⁶.

Statistical methods
From 1984 to 2006, we calculated the IC₅₀ with d HN-NonLin polynomial⁷, a regression model using the log transformation of drug concentration. This model requires the user to fix the degree of smoothness.

Since 2006, IC₅₀ values and their 95% Confidence Intervals (CIs) were calculated with an Emax model based on the linear least squared function of R software⁸, also available at http://www.antimalarial-icestimator.net.

In vitro data are grouped by country of origin, as determined by patient interview. Data are presented in two different ways for each country. The very large data sets in the WWARN Explorer preview version were gathered by three laboratories using common protocols and reagents. This allowed the group to define threshold resistance values for a set of drugs analysed by each laboratory. In addition, an extensive review of the literature demonstrates that resistance to many antimalarials is not observed as a discrete change from susceptible to resistant. However, chloroquine resistance can be defined as a threshold, at an accepted concentration of 80-110 nM in the isotopic assay^{9,10,11,12,13}. Based on this rationale, we have used this metric to present most of the data in the WWARN Explorer preview.

However, when considering in vitro data gathered from many laboratories worldwide, the absolute IC₅₀ values measured for a drug can vary dramatically. For that reason, in future versions of WWARN Explorer, we will move towards displaying in vitro susceptibility as the ratio of the IC₅₀ value in the isolates compared with the susceptibility of the 3D7 laboratory adapted strain assessed under the same conditions. In this way, we expect that results from studies of isolates analysed over time, variable locations and a wide range of methodologies can be compared.

For these reasons, in vitro data are presented in the WWARN Explorer preview in two ways:

• Median IC₅₀ value for isolates represented by a box plot.

• Chloroquine IC₅₀ values are also presented as the ratio of the isolate mean IC₅₀ values divided by the geometric mean of a control 3D7 strain from the same time period.

Molecular marker methodology

The WWARN Explorer preview presents a uniform analysis of marker prevalence. Therefore, the results presented by WWARN may differ slightly from those in the original publications and reports of these studies.

Statistical methods
Prevalence of each marker is calculated by dividing the number of samples bearing the mutation or allele correlated with resistance (pure or mixed with wild-type) by the total number of samples genotyped at that locus. Samples used for prevalence estimates include those taken on Day 0 of clinical trials (pre-treatment) and those taken during cross-sectional or population surveys.

For multi-locus genotypes such as the dhfr triple mutant (108N + 51I + 59R), pure mutants are defined as the absence of wild-type at all codons. Mixed mutants include all mutant genotype samples with up to one mutant/wild-type mixed codon. Samples with more than one mutant/wild-type mixed codon are not included in the numerator, as the presence of a true multi-locus mutant genotype cannot be verified.

Confidence intervals for prevalence estimates are calculated assuming a binomial distribution:
95% CI=p±1.96 √((p(1-p))/n)

References

¹ Verret et al. The effect of varying analytical methods on estimates of anti-malarial clinical efficacy. Mal J 2009; 8:77
² Price R et al. World Antimalarial Resistance Network I: clinical efficacy of antimalarial drugs. Mal J 2007; 6:119.
³ World Health Organization. Methods for surveillance of antimalarial drug efficacy. 2009.
⁴ STATA, 2009. www.stata.com.
⁵ Desjardins et al. Quantitative assessment of antimalarial activity in vitro by a semiautomated microdilution technique. Antimicrob Agents Chemother 1979 Dec; 16(6):710-718.
⁶ Makler et al. Parasite lactate dehydrogenase as an assay for Plasmodium falciparum drug sensitivity. Am J Trop Med Hyg 1993;48(6):739-741.
⁷ Copyright H. Noedl, Armed Forces Research Institute for Medical Sciences (USAMC-AFRIMS, Bangkok, Thailand). www.meduniwien.ac.at/user/harald.noedl/malaria/.
⁸ The R Project for Statistical Computing www.r-project.org.
⁹ Le Bras et al. Plasmodium falciparum: drug sensitivity in vitro of isolates before and after adaptation to continuous culture. Exp Parasitol 1983 Aug; 56(1):9-14.
¹⁰ Le Bras ^{et al}. Dichlorquinazine (a 4-aminoquinoline) effective in vitro against chloroquine-resistant P. falciparum. Lancet, 1983, i, 73-74 (letter).
¹¹ Basco and Le Bras. In vitro activity of chloroquine and quinine in combination with desferrioxamine against Plasmodium falciparum. Am J Hematol 1993 Apr; 42(4):389-391.
¹² Pradines et al. [In vitro sensitivity of Plasmodium falciparum isolates from Gabon to chloroquine and cycloguanil] Bull Soc Pathol Exot 1999 May; 92(2):91-94.
13 Ringwald and Basco. Comparison of in vivo and in vitro tests of resistance in patients treated with chloroquine in Yaoundé, Cameroon. Bull World Health Organ 1999;77(1):34-33.