A surrogate marker of piperaquine-resistant Plasmodium falciparum malaria: a phenotype-genotype association study

Abstract

Background: Western Cambodia is the epicentre of Plasmodium falciparum multidrug resistance and is facing high rates of dihydroartemisinin-piperaquine treatment failures. Genetic tools to detect the multidrug-resistant parasites are needed. Artemisinin resistance can be tracked using the K13 molecular marker, but no marker exists for piperaquine resistance. We aimed to identify genetic markers of piperaquine resistance and study their association with dihydroartemisinin-piperaquine treatment failures.

Methods: We obtained blood samples from Cambodian patients infected with P falciparum and treated with dihydroartemisinin-piperaquine. Patients were followed up for 42 days during the years 2009-15. We established in-vitro and ex-vivo susceptibility profiles for a subset using piperaquine survival assays. We determined whole-genome sequences by Illumina paired-reads sequencing, copy number variations by qPCR, RNA concentrations by qRT-PCR, and protein concentrations by immunoblotting. Fisher's exact and non-parametric Wilcoxon rank-sum tests were used to identify significant differences in single-nucleotide polymorphisms or copy number variants, respectively, for differential distribution between piperaquine-resistant and piperaquine-sensitive parasite lines.

Findings: Whole-genome exon sequence analysis of 31 culture-adapted parasite lines associated amplification of the plasmepsin 2-plasmepsin 3 gene cluster with in-vitro piperaquine resistance. Ex-vivo piperaquine survival assay profiles of 134 isolates correlated with plasmepsin 2 gene copy number. In 725 patients treated with dihydroartemisinin-piperaquine, multicopy plasmepsin 2 in the sample collected before treatment was associated with an adjusted hazard ratio (aHR) for treatment failure of 20·4 (95% CI 9·1-45·5, p<0·0001). Multicopy plasmepsin 2 predicted dihydroartemisinin-piperaquine failures with 0·94 (95% CI 0·88-0·98) sensitivity and 0·77 (0·74-0·81) specificity. Analysis of samples collected across the country from 2002 to 2015 showed that the geographical and temporal increase of the proportion of multicopy plasmepsin 2 parasites was highly correlated with increasing dihydroartemisinin-piperaquine treatment failure rates (r=0·89 [95% CI 0·77-0·95], p<0·0001, Spearman's coefficient of rank correlation). Dihydroartemisinin-piperaquine efficacy at day 42 fell below 90% when the proportion of multicopy plasmepsin 2 parasites exceeded 22%.

Interpretation: Piperaquine resistance in Cambodia is strongly associated with amplification of plasmepsin 2-3, encoding haemoglobin-digesting proteases, regardless of the location. Multicopy plasmepsin 2 constitutes a surrogate molecular marker to track piperaquine resistance. A molecular toolkit combining plasmepsin 2 with K13 and mdr1 monitoring should provide timely information for antimalarial treatment and containment policies.

Funding: Institut Pasteur in Cambodia, Institut Pasteur Paris, National Institutes of Health, WHO, Agence Nationale de la Recherche, Investissement d'Avenir programme, Laboratoire d'Excellence Integrative "Biology of Emerging Infectious Diseases".

Figure 1

Location of study sites (provinces) where dihydroartemisinin–piperaquine clinical efficacy studies were done in 2009–15 (42-day follow-up)

Figure 2

Manhattan plot showing the significance of copy number variations between whole-genome exome sequences of 23 piperaquine-resistant and eight piperaquine-sensitive culture-adapted lines collected in Cambodia in 2012 and phenotyped using the in-vitro piperaquine survival assay Each dot represents a gene in the set of 31 culture-adapted parasites, according to chromosome. The x axis represents genomic location, and the y axis represents the log₁₀ transformed Wilcoxon test p values. *Wilcoxon test p=0·139; after Benjamini-Hochberg correction, only two genes, PF3D7_1408000 (plasmepsin 2) and PF3D7_1408100 (plasmepsin 3) achieved genome-wide significance (p=0·03795).

Figure 3

Ex-vivo piperaquine survival assay (PSA) survival rates and single (n=67) and multicopy plasmepsin 2 (n=67) as estimated by qPCR in isolates collected before dihydroartemisinin–piperaquine (DHA–PPQ) treatment stratified by K13 genotype Patients were enrolled in clinical studies done in 2014–15 in Mondulkiri, Rattanakiri, Siem Reap, and Stungtreng provinces (see table 1). K13 polymorphisms were detected in 65 of 69 piperaquine-resistant isolates (64 C580Y, one Y493H) and 17 of 65 piperaquine-susceptible isolates (15 C580Y, one C469F, and one A626E). Three parasite lines with discordant data were recorded: two resistant lines with non-amplified plasmepsin 2 and plasmepsin 3 loci (6246 and 6395) and one sensitive line with two plasmepsin 2 copies (6369; table 2). The ex-vivo PSA survival rate (%) corresponds to the ratio of number of viable parasites in the PPQ-exposed cultures versus the number of viable parasites in the non-exposed culture.

Figure 4

Patients enrolled in clinical studies done in 2009–15 in 12 provinces across Cambodia to assess the efficacy of the 3-day dihydroartemisinin–piperaquine (DHA–PPQ) regimen, and isolates used to detect molecular signatures associated with in-vitro piperaquine survival assay (PSA) resistance and DHA–PPQ clinical failure Supervised DHA–PPQ was given once daily for 3 days (day 0, 24 h, 48 h). Dosing was based on bodyweight: less than 19 kg, 40 mg DHA–320 mg PPQ per day; 19–29 kg, 60 mg DHA–480 mg PPQ per day; 30–39 kg, 80 mg DHA–640 mg PPQ per day; greater than 40 kg, 20 mg DHA–960 mg PPQ per day. For children unable to swallow tablets, DHA–PPQ was dissolved in 5 mL of water. Patients were observed for 1 h post-dosing and were re-dosed with a full or half dose if vomiting occurred within 30 min or between 31 and 60 min, respectively. Those who vomited after the second dose were withdrawn from the study and were given parenteral rescue treatment (intramuscular artemether). Patients with axillary temperatures of 37·5°C were treated with paracetamol. Patients were seen daily to day 3 and then weekly for 6 weeks (day 42) for clinical examinations (axillary temperature, symptom check) and malaria blood films. Home visits were done if patients failed to come back for their follow-up appointments. Withdrawn patients, patients lost to follow-up, and patients classified as reinfected (based on msp1, msp2, and glurp genotypes) were excluded from the analysis.

Figure 5

Cumulative proportion of non-recrudescent patients treated with a 3-day course of dihydroartemisinin–piperaquine (A) Plasmepsin 2 (PM2) gene copy number. Log-rank test: p<0·0001 overall; p<0·0001 (hazard ratio [HR] 32·2 [95% CI 17·9–58·0]) for single copy vs two copies; p<0·0001 (HR 49·0 [23·0–104·2]) for single copy vs three of more copies; p=0·017 (HR 1·53 [1·04–2·25]) for two copies vs three or more copies. (B) PM2 gene copy number and K13 genotype detected in isolates collected at the time of enrolment, before treatment. Log-rank test: p<0·0001 overall; p<0·0001 for K13 wild-type–PM2 single copy vs K13 wild-type–PM2 multicopy; p=0·002 for K13 wild-type–PM2 single copy vs K13 mutant–PM2 single copy; p<0·0001 for K13 wild-type–PM2 single copy vs K13 mutant–PM2 multicopy; p=0·001 (HR 6·9 [0·5–96·6]) for K13 wild-type–PM2 multicopy vs K13 mutant–PM2 single copy; p=0·07 (HR 2·6 [1·3–5·5]) for K13 wild-type–PM2 multicopy vs K13 mutant–PM2 multicopy; p<0·0001 (HR 17·5 [12·2–25·2]) for K13 mutant–PM2 single copy vs K13 mutant–PM2 multicopy.