Selection and spread of artemisinin-resistant alleles in Thailand prior to the global artemisinin resistance containment campaign

Abstract

The recent emergence of artemisinin resistance in the Greater Mekong Subregion poses a major threat to the global effort to control malaria. Tracking the spread and evolution of artemisinin-resistant parasites is critical in aiding efforts to contain the spread of resistance. A total of 417 patient samples from the year 2007, collected during malaria surveillance studies across ten provinces in Thailand, were genotyped for the candidate Plasmodium falciparum molecular marker of artemisinin resistance K13. Parasite genotypes were examined for K13 propeller mutations associated with artemisinin resistance, signatures of positive selection, and for evidence of whether artemisinin-resistant alleles arose independently across Thailand. A total of seven K13 mutant alleles were found (N458Y, R539T, E556D, P574L, R575K, C580Y, S621F). Notably, the R575K and S621F mutations have previously not been reported in Thailand. The most prevalent artemisinin resistance-associated K13 mutation, C580Y, carried two distinct haplotype profiles that were separated based on geography, along the Thai-Cambodia and Thai-Myanmar borders. It appears these two haplotypes may have independent evolutionary origins. In summary, parasites with K13 propeller mutations associated with artemisinin resistance were widely present along the Thai-Cambodia and Thai-Myanmar borders prior to the implementation of the artemisinin resistance containment project in the region.

Fig 1. Geographic distribution of the K13 propeller alleles in Thailand in 2007.

Pie charts show K13 propeller allele frequencies among 417 parasite isolates in 10 Thailand provinces. The different alleles are color coded. The results are shown on top of the clinical burden map of P. falciparum in Thailand in 2007 (Malaria Atlas Project) [34]. Light grey areas are P. falciparum malaria free and dark grey areas have an unstable risk of malaria transmission (i.e. annual case incidence, or API, is reported at less than 1 per 10,000). Map shows mean estimate for the clinical burden in the range from 0 (light green) to 10,000 (dark green/blue) clinical cases per year. The clinical burden predictions are based on a Bayesian geostatistical model.

Fig 2. K13 flanking microsatellites of parasites from Thailand.

The figure comprises data from 16 western and 10 eastern Thai parasites with the C580Y mutation. Allele lengths are shown for 12 microsatellites positioned at: 72.3kb, 31.5kb, 31.0kb, 15.1kb, 8.6kb, 3.4kb, -0.15kb, -3.74kb, -6.36kb, -31.9kb, -50.0kb, and -56.0kb. Teal green shading and lines indicate identical allele sizes. DNW (in grey) = indicates no successful amplification after a third attempt or not enough DNA was available to repeat the analysis.

Fig 3. Heterozygosity valley around K13 propeller alleles in Thailand.

The expected heterozygosity of parasite isolates with: (A) the C580Y mutation (N = 26) and wild type alleles (N = 82); (B) the C580Y from the east (N = 10) and west (N = 16). Diversity was reduced at all 9 K13 propeller microsatellite loci for C580Y compared to wild type alleles. The mean He in (A) for C580Y (0.3526 ± 0.08), wild type (0.6246 ± 0.06), neutral (0.7650 ± 0.05); (B) C580Y east (0.2755 ± 0.05) and C580Y west (0.4360 ± 0.03 ± 0.03). The error bars indicate ± standard deviation (SD).

Fig 4. Population structure by geography.

Scatter plots from principal component analysis (PCA) based on flanking K13 microsatellite analysis of parasite isolates with the C580Y allele (A) and wild type allele (B). Results of PCA based on neutral microsatellites for parasite isolates carrying the C580Y allele, are shown in (C).

Fig 5. Association between genetic dissimilarity and geographic distance.

(A) Genetic dissimilarity and geographic distance between pairs of C580Y mutants (A) and wild type parasites (B). Genetic dissimilarity of flanking microsatellites for C580Y is highly associated with geographic distance (Pearson's correlation coefficient: 0.44, 95% CI:0.34–0.52). Average genetic dissimilarity is lower between C580Y and wild type parasites (t-test p-value < 0.001).