Emerging implications of policies on malaria treatment: genetic changes in the Pfmdr-1 gene affecting susceptibility to artemether-lumefantrine and artesunate-amodiaquine in Africa

Abstract

Artemether-lumefantrine (AL) and artesunate-amodiaquine (AS-AQ) are the most commonly used artemisinin-based combination therapies (ACT) for treatment of Plasmodium falciparum in Africa. Both treatments remain efficacious, but single nucleotide polymorphisms (SNPs) in the Plasmodium falciparum multidrug resistance 1 (Pfmdr1) gene may compromise sensitivity. AL and AS-AQ exert opposing selective pressures: parasites with genotype 86Y, Y184 and 1246Y are partially resistant to AS-AQ treatment, while N86, 184 F and D1246 are favoured by AL treatment. Through a systematic review, we identified 397 surveys measuring the prevalence of Pfmdr1 polymorphisms at positions 86 184 or 1246 in 30 countries in Africa. Temporal trends in SNP frequencies after introduction of AL or AS-AQ as first-line treatment were analysed in 32 locations, and selection coefficients estimated. We examined associations between antimalarial policies, consumption, transmission intensity and rate of SNP selection. 1246Y frequency decreased on average more rapidly in locations where national policy recommended AL (median selection coefficient(s) of -0.083), compared with policies of AS-AQ or both AL and AS-AQ (median s=-0.035 and 0.021, p<0.001 respectively). 86Y frequency declined markedly after ACT policy introduction, with a borderline significant trend for a more rapid decline in countries with AL policies (p=0.055). However, these trends could also be explained by a difference in initial SNP frequencies at the time of ACT introduction. There were non-significant trends for faster selection of N86 and D1246 in areas with higher AL consumption and no trend with transmission intensity. Recorded consumption of AS-AQ was low in the locations and times Pfmdr1 data were collected. SNP trends in countries with AL policies suggest a broad increase in sensitivity of parasites to AS-AQ, by 7-10 years after AL introduction. Observed rates of selection have implications for planning strategies to cycle drugs or use multiple first-line therapies to maintain drug efficacy.

Keywords: epidemiology; malaria; parasitology; public health; systematic review.

Figure 1

Prevalence of the 86Y, 1246Y and 184F Plasmodium falciparum multidrug resistance 1 (Pfmdr1) mutants in infected individuals in Africa over different time periods. Prevalences before 2004 are shown in online supplementary figure S1.

Figure 2

The frequency or prevalence of Plasmodium falciparum multidrug resistance 1 (Pfmdr1) mutations over time. Where possible, mutant frequencies were extracted from publications or estimated (see the Methods section), otherwise mutant prevalence data are shown. Locations are coloured by geographical region and include the following countries. East Africa: Uganda, Tanzania, Zanzibar, Tanzania, mainland, Swaziland, Sudan, Rwanda, Mozambique, Malawi, Madagascar, Kenya, Ethiopia, Eritrea, Comoros; West Africa: Senegal, Nigeria, Mauritania, Mali, Liberia, Guinea-Bissau, Ghana, Gambia, Burkina Faso, Benin; Central Africa: Sao Tome and Principe, Republic of Congo, Gabon, Equatorial Guinea, Democratic Republic of Congo, Central African Republic, Cameroon, Angola. See also online supplementary figure S2-4 for the data separately by location and online supplementary data for the full data.

Figure 3

Selection of Plasmodium falciparum multidrug resistance 1 (Pfmdr1) polymorphisms in 32 locations within countries with different first-line artemisinin-based combination therapies (ACT) policies, as shown by the change in the ratio of Pfmdr1 mutant to wild type frequency/prevalence over time since introduction of ACT policy. Panels show natural log mutant:wild type ratios at Pfmdr1 positions 86 (A–C), 184 (D–F) and 1246 (G–I). Areas had (A,D,G): artemether–lumefantrine (AL) policy, (B,E,H) artesunate–amodiaquine (AS-AQ) policy or (C,F,I) both AL and AS-AQ policy. Circles indicate the data, coloured by location and lines the fitted weighted linear regression at each location (see online supplementary figure S5-7 for the data separately by location). The slope on the natural log scale indicates the selection coefficient at each location. Where available, frequencies of each allele were used to estimate the slopes (solid lines), or otherwise prevalence data were used (dashed lines – see also the Methods section).

Figure 4

Consumption of artemisinin-based combination therapies (ACT) and Plasmodium falciparum multidrug resistance 1 (Pfmdr1) single nucleotide polymorphism (SNP) selection. Panels A–C: The proportion of all antimalarials taken/given which are artemether–lumefantrine (AL) (blue), artesunate–amodiaquine (AS-AQ) (red) or ACT of unknown type (grey, may include AL, AS-AQ or other), according to national first-line policy. Data are either self-reported antimalarial use by caregivers of children aged 0–5 years recorded in national household surveys (DHS) or market share of ACT type out of all antimalarials sold/prescribed as recorded by representative national surveys of public and private dispensing outlets (ACTWatch). Data are shown for all countries with any Pfmdr1 data in our review. Legend indicates colours in panels A–C. Panels D–F: estimated AL consumption and selection of Pfmdr1 SNPs in countries with AL first-line policies. Selection coefficients for each SNP were estimated in locations with at least two measures of SNP frequency/prevalence over time periods spanning at least 3 years. The dashed line indicates zero change in frequency. AL consumption is the mean estimated % febrile RDT-positive 0–5 year olds taking ACT during the time period over which SNP trends were available for each location, from analysis by Bennett et al, assuming that all ACTs taken in these countries are the first-line policy, AL. None of the associations between AL consumption and SNP selection are statistically significant. Countries are indicated: AO: Angola, BJ: Benin, ET: Ethiopia, KE: Kenya, MZ: Mozambique, RW: Rwanda, TZ: Tanzania, UG: Uganda.

Figure 5

Relationship between prevalence of different Plasmodium falciparum multidrug resistance 1 (Pfmdr1) single nucleotide polymorphisms (86Y, Y184, 1246Y), measured in the same surveys and coloured by geographical region as in figure 2. Here, we show the wild type Y184 prevalence and its relationship with 86Y and 1246Y mutant prevalence in the interests of highlighting possible occurrence of the Pfmdr1 YYY haplotype by region (exact haplotype frequencies are not known in the majority of surveys due to mixed infections).