Assessment of ex vivo antimalarial drug efficacy in African Plasmodium falciparum parasite isolates, 2016-2023: a genotype-phenotype association study

Abstract

Background: Given the altered responses to both artemisinins and lumefantrine in Eastern Africa, monitoring antimalarial drug resistance in all African countries is paramount.

Methods: We measured the susceptibility to six antimalarials using ex vivo growth inhibition assays (IC₅₀) for a total of 805 Plasmodium falciparum isolates obtained from travellers returning to France (2016-2023), mainly from West and Central Africa. Isolates were sequenced using molecular inversion probes (MIPs) targeting forty-three genes across the parasite genome, of which nineteen are drug resistance genes.

Findings: Ex vivo susceptibility of all assessed antimalarial compounds was consistent with their potent activity. The median IC₅₀ values for the six drugs were 1.1 nM [IQR: 0.8-1.7] for DHA, 16.7 nM [9.9-27.4] for LMF, 29.5 nM [19.1-45.5] for MFQ, 23.4 nM [17.1-39.0] for MDAQ, 26.7 nM [18.0-41.2] for CQ, and 18.5 nM [15.1-24.3] for PPQ. Only four isolates carried a validated pfkelch13 mutation. Multiple mutations in pfcrt and one in pfmdr1 (Asn86Tyr) were significantly associated with altered susceptibility to multiple drugs, and their frequencies decreased over time. Pfcrt and pfmdr1 mutations altered susceptibility to lumefantrine and mefloquine in an additive manner, with the wild-type haplotype (pfcrt K76-pfmdr1 N86) exhibiting the lowest susceptibility.

Interpretation: Our study on P. falciparum isolates from West and Central Africa indicates a low frequency of molecular markers associated with artemisinin resistance and a modest but significant decrease (1.6-2.3X) in the frequency of multidrug resistance markers. These genotypic changes likely mark parasite adaptation to sustained drug pressure and call for intensifying the monitoring of antimalarial drug resistance in Africa.

Funding: This work was supported by the French Ministry of Health (grant to the French National Malaria Reference Centre) and by the Agence Nationale de la Recherche (ANR-17-CE15-0013-03 to JC). JAB was supported by NIH R01AI139520. JR postdoctoral fellowship was funded by Institut de Recherche pour le Développement.

Keywords: Antimalarial drug resistance; Ex-vivo susceptibility; Genotype; Growth inhibition assay.

Fig. 1

Geographical origin and collection year of isolates. a) Map of Africa showing the origin of malaria cases imported into France. Colours indicate sample size, and countries with fewer than three isolates are not shown. b) Bar plot showing the temporal distribution of imported malaria cases included in this study (2016, n = 165; 2017, n = 166; 2018, n = 167; 2019, n = 100; 2020, n = 46; 2021, n = 39; 2022, n = 112; 2023, n = 10). The top ten countries (n > 21 samples) are displayed. Countries with fewer than 20 samples are represented in the group “n < 20”.

Fig. 2

Half-maximal inhibitory concentration (IC₅₀) for six antimalarial drugs. Distribution of IC₅₀ for the six antimalarial drugs from 2016 to 2023 (2016, n = 165; 2017, n = 166; 2018, n = 167; 2019, n = 100; 2020–21, n = 85; 2022–23, n = 122). Box plots show the median IC₅₀ (in nM) and interquartile range. Dashed red lines indicated the IC₅₀s cut-offs for drug resistance as defined in some studies (see Table S4).

Fig. 3

Frequency of key mutations associated with resistance to different drugs. a) Frequency of key mutations in pfcrt, pfdhfr, pfdhps, and pfmdr1 genes detected in isolates from this study (n = 427–785). b) Frequency of mutations in the pfkelch13 gene (n = 611–761). Green bar plots indicate the frequency of mutations in the N-terminal-coding domain, while orange bar plots indicate the frequency of mutations in the propeller domain. SNPs marked with an asterisk (∗) are validated or candidate SNPs by WHO. c) Frequency of the main key mutations in pfcrt, pfdhfr, pfdhps, and pfmdr1 by year (n = 796). d) Frequency of pfcrt Lys76Thr-pfmdr1 Asn86Tyr haplotypes over time (n = 664). Colour code for the 4 haplotypes: purple, wild-type Lys76-Asn86; green, single mutant Lys76-86Tyr; blue, single mutant 76Thr-Asn86; red, double mutant 76Thr-86Tyr. a–c: mixed genotypes, i.e., a sample with a reference allele and a mutant allele in a given sample, were considered mutant regardless of the within-sample mutant allele frequency. c & d: 95% confidence intervals are represented by error bars. In panel d, haplotypes were reconstructed for all samples using the major allele in a given codon (i.e., the allele with a within-sample allele frequency larger than 75%); alleles with a frequency between 25% and 75% were considered unresolved and discarded from the analysis.

Fig. 4

Manhattan plot showing the significance of SNPs associated with six antimalarial drugs. Each dot represents 1 of 362 SNPs with MAF > 0.01 coloured by chromosome. The x-axis represents the chromosomal location of the SNPs, and the y-axis represents the −log10 of the p-value obtained from the linear regression model analysis. The red line represents the p-value threshold after Bonferroni correction (p ≤ 10⁻⁴). The full list of SNPs associated with IC₅₀ is shown in Table S10.

Fig. 5

Effect of pfcrt Lys76Thr-pfmdr1 Asn86Tyr haplotypes on IC₅₀ for six drugs. a) IC₅₀ for mefloquine (MFQ), lumefantrine (LMF), and dihydroartemisinin (DHA) disaggregated by pfcrt-pfmdr1 haplotypes. b) IC₅₀ for chloroquine (CQ), monodesethylamodiaquine (MDAQ), and piperaquine (PPQ) disaggregated by pfcrt-pfmdr1 haplotypes. Haplotypes were built with pfcrt Lys76Thr and pfmdr1 Asn86Tyr using the major allele per position. Dashed lines indicated the IC₅₀s cut-offs. n: number of isolates carrying the haplotype. Differences in the IC₅₀ of wild-type (WT|WT, crt-K76|mdr1-N86, n = 497) versus single mutant (Mut|WT, crt-76T|mdr1-N86, n = 126), wild-type vs single mutant (WT|Mut, crt-K76|mdr1-86Y, n = 28), and wild-type vs double mutant (Mut|Mut, crt-76T|mdr1-86Y, n = 13) were calculated with Pairwise Wilcoxon tests with Benjamini–Hochberg correction. ∗: p < 0.05; ∗∗: p < 0.001.