‘Big data’ - harnessing the power of pooled analyses to prolong the efficacy of treatment

“Big data” is a global initiative that countries such as the USA and the UK are investing millions of dollars in to enable secure storage, standardisation, analysis and sharing of large clinical, drug discovery, engineering and broader scientific data. A similar approach has been championed by WWARN since 2009 to support the malaria community to share, store, standardise and analyse malaria clinical trial results.

Research groups and pharmaceutical drug developers have contributed to the WWARN Data Centre – a repository that stores over 350 clinical, molecular, in vitro, and pharmacokinetic malaria studies. The repository is comprised of approximately 90,000 patient records, an estimated 70% of all published clinical patients enrolled in ACT studies. It includes data on some of the current and most widely used artemisinin combination therapies (ACTs) such as artemether-lumefantrine (AL), artesunate-amodiaquine (ASAQ), and dihydroartemisinin-piperaquine (DP).

Why is this Data Centre useful?

Over the last two years this secure centre has enabled Study Groups, comprised of 130 or more malaria researchers, to examine clinical trial results for each of the key antimalarial drugs. Working together they have developed key questions around malaria treatment efficacy to assess whether all patient groups were receiving the optimal dose level of particular drugs, and if not whether patients who are underdosed  are at increased risk of treatment failure. To achieve this, thousands of data sets were analysed to identify key trends and anomalies.

The overall analyses confirm the excellent news that the clinical efficacy of all three ACTs is still very high – in fact they are more than 95% effective. Despite this confirmation, it was also possible to examine large numbers of patients who did fail treatment and consider: 

“Do the patients that fail treatment share certain factors in common that might increase their risk of treatment failure?”

What we learnt from a pooled analysis

Firstly, the groups realized that it was necessary to develop a standardized approach to storing and curating the data that would enable them to then pool and analyse the patient responses that were all presented with the same standard. This increased the statistical power, and enhanced the depth and quality of each analysis by giving greater visibility of treatment patterns across a large number of patients that were treated in multiple countries and contexts.

Carol Sibley, Scientific Director at WWARN explains “We were able to determine that patients who failed treatment with any of these ACTs did indeed have some risk factors in common. First, they were more likely to have a high numbers of parasites already in their blood when they began treatment.  Second, they were also likely to have received a lower dose of the drug than is recommended. These factors stood out particularly because very young children (under 5 years old) often shared both of these factors.”

The researchers confirmed that there are probably several reasons why this occurs. In the first instance, getting the dose right for children is difficult; the correct dose may require that tablets are split. Second, the dose is determined by weight bands, and the lightest children are likely to be overdosed while heavier children are often underdosed.  In addition, the  biochemical and physiological effects of drugs on the body in children, and the relationship between drug concentration and effect,  often differ between children and adults [1]. This can exacerbate the problem.  It is striking that in areas of high and moderate malaria transmission, young children are a major segment of the patient population, so the number of people who are underdosed and then threatened with treatment failure can be high.

Dr Martin De Smet, Leader of the Malaria Working Group at Médecins Sans Frontières (Doctors Without Borders) supports the pooled analysis, “Working with broad age and weight ranges is the best choice where large scale routine malaria treatment is required and conditions are tough. However, these guidelines may be less appropriate when we focus on specific vulnerable groups such as malnourished, hospitalised children and children below one year of age. For these groups a more accurate dosage should be aimed for, as such a greater insight into the relationship between antimalarial dosing and drug efficacy is valuable.”

Why does the drug dosing level matter?

When resistant parasites first evolve they are rare.  Driven by drug exposure the resistant parasites do increase in the parasite population, but slowly. Treatment failure is hardly noticed in the patients until the resistant parasites are at high levels and threaten vulnerable communities in countries such as Cambodia, Thailand and Myanmar.  Exposing parasite populations with small numbers of resistant parasites to low concentrations of the drug - enough to kill the sensitive parasites, but not the resistant ones - is an excellent way to rapidly increase resistance.  That is exactly what happens when patients receive a dose of an effective drug that is too low to treat their malaria symptoms effectively.

A major advantage of pooled analyses of this kind is that they allow drivers of resistance like underdosing to be identified even when the clinical efficacy of the existing drugs used is still very high. This information can then be used by researchers and governments to respond to signs of resistance very early on, and slow or prevent further resistance spreading by modifying treatment regimens.  Overall, this will prolong the effectiveness of current antimalarials and strengthen the potential impact of new drugs as they are developed and administered in the coming years.

See the first of three pooled analyses papers on dihydroartemisinin-piperaquine (DP) that the WWARN study groups hope to publish this year on each major ACT antimalarial drug treatment.

The WorldWide Antimalarial Resistance Network DP Study Group (2013). The Effect of Dosing Regimens on the Antimalarial Efficacy of Dihydroartemisinin-Piperaquine: A Pooled Analysis of Individual Patient Data. PLoS Med 10(12): e1001564. doi:10.1371/journal.pmed.1001564

1. Lees P et al (2004). Principles of pharmacodynamics and their applications in veterinary pharmacology. J. Vet. Pharmacol. Ther. 27 (6): 397–414; doi:10.1111/j.1365-2885.2004.00620.