Selection of Parasites with Diminished Drug Sensitivity by Amodiaquine-Containing Antimalarial Regimens in Uganda

Abstract

Background

Amodiaquine (AQ) is paired with artesunate (AS) or sulfadoxine-pyrimethamine (SP) in recommended antimalarial regimens. It is unclear how readily AQ resistance will be selected with combination chemotherapy.

Methods

We collected 61 Plasmodium falciparum samples from a cohort of Ugandan children randomized to treatment with AQ/SP, AS/AQ, or artemether-lumefantrine (AL) for uncomplicated malaria. In vitro sensitivity to monodesethylamodiaquine (MDAQ) was measured with a histidine rich protein-2-based ELISA, and potential resistance-mediating polymorphisms pfmdr-1were evaluated.

Results

Parasites from subjects previously treated with AQ/SP or AS/AQ within 12 weeks were less sensitive to MDAQ (n=18; mean IC₅₀ 62.9 nM; range 12.7–158.3 nM) than parasites from those not treated within 12 weeks (n=43; mean IC₅₀ 37.5 nM; range 6.3–184.7 nM; p=0.0085) or only those in the treatment arm that did not contain AQ (n=20; mean IC₅₀ 28.8 nM; range 6.3–121.8 nM; p=0.0042). The proportion of strains with polymorphisms expected to mediate diminished response to AQ (pfmdr-1 86Y and 1246Y) increased after prior AQ therapy, although differences were not significant.

Conclusions

Prior therapy selected for diminished response to MDAQ, suggesting that AQ-containing regimens may rapidly lose efficacy in Africa. The mechanism of diminished MDAQ response is not fully explained by known mutations in pfmdr-1.

INTRODUCTION

The control of Plasmodium falciparum malaria is seriously challenged by resistance to many available drugs [1]. Among drugs with diminished antimalarial activity are the 4-aminoquinolines, notably chloroquine (CQ) and amodiaquine (AQ). However, despite very high levels of resistance to CQ in Africa, AQ often offers effective antimalarial treatment, and it remains an important part of our antimalarial drug portfolio [2]. With increasing resistance, there is now a clear consensus that uncomplicated falciparum malaria should be treated with drug combinations, ideally artemisinin-based combination therapy (ACT) [3]. ACT regimens all rely on an artemisinin component and a longer-acting partner drug [4]. AQ plays a key role in current treatment strategies. When partnered with artesunate (AS), AQ has shown excellent antimalarial efficacy in multiple studies in Africa, and this combination is now available as a co-formulated drug [5–9]. In addition, the older combination of AQ plus sulfadoxine-pyrimethamine (SP) has shown good efficacy in many areas [5, 10, 11], and is recommended by the World Health Organization to treat uncomplicated malaria when the efficacy of the individual components is acceptable and ACTs are not available [2].

With all ACTs there is a risk of selection of resistance to artemisinin partner drugs, which circulate long after elimination of the short-acting artemisinin component [12]. This is a particular concern for AQ, as resistance to this drug is already well established in parts of Africa [8, 13, 14]. AQ appears to act principally after conversion to an active metabolite, monodesethylamodiaquine (MDAQ) [15]; parasites are ~3-fold less sensitive in vitro to MDAQ than to the parent compound [16]. Parasites with a range of in vitro sensitivity to MDAQ have been described in Africa [17–21]. In general >90% of isolates from different sites in Africa were classified as sensitive to MDAQ (using the IC₅₀ cut-off of 60 nM MDAQ), although a recent study from Kenya reported that 26% of baseline isolates had an IC₅₀ for MDAQ >60 nM [21]. In South America and Asia, resistance to MDAQ or AQ has been more common [16, 22–25]. Relatively little information on P. falciparum sensitivity to MDAQ is available from East Africa, the location of our study, and an area where resistance to AQ is quite common.

The mechanism of P. falciparum resistance to AQ or MDAQ is incompletely understood; mutations in genes encoding two putative drug transporters, pfcrt and pfmdr1, appear to play important roles. The pfcrt 76T mutation is the principal mediator of resistance to CQ [26]. In areas where parasites with wild type pfcrt still circulate, the 76T mutation predicted poor response to therapy with AQ [27–29], and this mutation was selected in parasites that caused recurrent infections after treatment with AQ [11, 28–31]. The pfcrt 76T mutation was also associated with decreased in vitro sensitivity to MDAQ in field isolates [32] and in genetically altered laboratory strains [33]. Polymorphisms in a second putative transporter, pfmdr1, impact upon the sensitivity of malaria parasites to a number of drugs [34]. The pfmdr1 86Y polymorphism predicted poor AQ treatment outcome in most [28, 29], but not all [27] studies from Africa and was selected by prior therapy with AQ in a number of studies [11, 28–31, 35]. However, prevalence of the pfmdr1 86Y mutation was not associated with in vitro drug sensitivity in a study from Africa [32], and in samples from Colombia the pfmdr1 86Y mutation was seen in only ~10% of strains with in vitro resistance to MDAQ [22]. Taken together, recent studies suggest that the molecular basis of resistance to AQ is complex, with apparent contributions of polymorphisms in pfcrt and pfmdr1, and likely additional factors contributing to the resistant phenotype.

As we move to routine treatment of malaria with combination regimens, it is unclear how readily resistance to AQ will be selected by treatment with AQ-containing combinations. To gain insight in this area, we evaluated the in vitro activity of MDAQ against 61 clinical samples from Ugandan children treated with one of three combination regimens, two of which included AQ. We then searched for associations between in vitro drug sensitivity, clinical outcomes, prior drug use, and known molecular mediators of resistance. Most notably, we found that prior use of AQ-containing combinations selected for subsequent infections with diminished responsiveness to MDAQ.

METHODS

Clinical trial

All samples were from a cohort of 601 children randomly selected from a community in Kampala, Uganda, aged 1–10 years at enrollment in 2004–05, as previously described [9, 36]. Upon presentation with the first episode of uncomplicated malaria, participants were randomly assigned to receive AQ/SP, AS/AQ, or AL, which they received thereafter for each episode of uncomplicated malaria. With each treatment, efficacy was assessed following WHO guidelines [37], with genotyping to distinguish recrudescence and new infection after therapy, all as previously described [9]. For this study we analyzed a subset of samples, collected between August, 2006 and March, 2007 for which parasites were successfully grown in short-term culture to allow determination of in vitro drug sensitivity. The clinical trial was approved by the Uganda National Council of Science and Technology and the Institutional Review Boards of Makerere University and the University of California, San Francisco.

In vitro sensitivity to MDAQ

Upon diagnosis of uncomplicated malaria, and before the initiation of therapy, blood was collected into heparinized tubes and delivered within 30 minutes to our laboratory. Specimens were centrifuged, supernatant and buffy coat were removed, erythrocytes were washed twice in RPMI 1640 medium pre-warmed to 37°C, samples were diluted to 0.05% parasitemia and 2% hematocrit, and parasites were assessed for drug sensitivity under sterile conditions in RPMI 1640 supplemented with 0.5% Albumax and 100 µg/ml gentamicin in 96-well cell culture plates that had been coated with 7 serial dilutions of MDAQ (6.25–400 nM, each in duplicate) and dried overnight. Wells without drug served as controls. For each sample, 200 µl of culture was added to each well, with duplicate wells for each concentration of MDAQ. Plates were incubated for 72 hours at 37°C in a candle jar, and samples were then frozen (−20°C overnight) and thawed before analysis. Parasite growth was assayed by comparing levels of histidine rich protein-2 (HRP-2) in treated and control cultures [38]. HRP-2 was quantified using a commercial ELISA test kit (Malaria Ag Celisa, Cellabs). Samples were diluted (1:4–1:10; the same dilution for each sample in an experiment) in water, and 100 µl of each hemolyzed sample preparation was added to an ELISA plate pre-coated with anti-HRP-2 and incubated at room temperature for 1 h. Plates were then washed four times with the kit washing solution, 100 µl of secondary antibody was added to each well for1 h at room temperature, plates were again washed 4 times, 100 µl of chromogen substrate was added to each well, plates were incubated for 15 min in the dark, and 50 µl of stopping solution was added. Absorbance (450 nm) was then measured for each well with a SpectraMAX 340 spectrophotometer (Molecular Devices). Optical density values were fitted to normal curves based on serial dilutions of HRP-2 standards and a four-parameter curve model (Softmax Pro 2.1.1, Molecular Devices), and IC₅₀s were calculated based on a nonlinear regression model, with attention to test validity based on adequate readings above background, as previously described [39].

Genetic polymorphisms in pfmdr-1

Upon diagnosis of uncomplicated malaria blood spots were also dried on filter paper, which was stored with dessicant in sealed plastic bags at room temperature. DNA was subsequently extracted with chelex and previously characterized polymorphisms in pfcrt and pfmdr-1 were analyzed by nested PCR followed by mutation-specific restriction endonuclease (ApoI, AflIII, DraI, DdeI, AseI and EcoRV) cleavage, as previously described for the pfcrt K76T and pfmdr-1 N86Y, Y184F, S1034C, N1042D, and D1246Y polymorphisms, respectively [40]. DNA fragments were separated by agarose gel electrophoresis, and electrophoretic band patterns were categorized as wild type, mixed, or mutant genotypes by visual inspection of gels and comparison with DNA from control strains 7G8 and FCR3, obtained from the Malaria Research and Reference Reagent Resource Center (MR4).

Statistical analyses

Clinical data were entered and verified with Access (Microsoft Corporation). Statistical analyses were performed with Stata, version 10 (StataCorp). Clinical outcomes were assessed as previously described [9]. Values for in vitro drug sensitivity of parasite samples were not normally distributed, and were assessed using the Kruskal-Wallis rank test. Prevalences of mutations were compared with Fisher’s exact test. For all assessments, p values <0.05 were considered statistically significant.

RESULTS

In vitro drug sensitivity of fresh P. falciparum samples

Blood was collected from children presenting with uncomplicated falciparum malaria, and the in vitro sensitivity of infecting parasites to MDAQ, the principal active metabolite of AQ, was determined. Of 72 samples cultured during the time frame of the study, 61 had successful in vitro analyses, with a wide range of in vitro drug sensitivity (figure 1, supplemental table). Samples that could not be evaluated included 1 that was contaminated with bacteria, 4 that did not grow in culture, and 6 that failed to provide an adequate dose response curve for analysis. Using a previously assigned cut-off for resistance of IC₅₀ >60 nM [17], 13 (21.3%) of the parasite samples were categorized as resistant to MDAQ, and 8 (13.1%) had IC50 >100 nM. Control parasite lines (reference numbers from MR4 are provided) yielded IC₅₀ results of 30.9 nM for D6 (MRA-285), 29.8 nM for HB3 (MRA-155), 267.7 nM for W2 (MRA-157), and 262.1 nM for K1 (MRA-159); the first two strains are considered sensitive, and the last two resistant to CQ.

Figure 1. Sensitivity of clinical samples to MDAQ.

Parasites were placed into culture and sensitivities were assessed by comparing levels of HRP-2 with those of untreated controls with an ELISA assay.

Association of in vitro drug sensitivity and clinical treatment failure

The three combination regimens included in our clinical trial were all quite efficacious, and so few recrudescences occurred after therapy [9]. Only 7 of the 61 samples (11.5%) were from episodes of malaria classified as recrudescent. For these samples, and for both the AQ-containing treatment arms and the AL arm, recrudescence was equally likely in those with parasites sensitive and resistant to MDAQ (table 1). Thus, although sample size was small for this analysis, we saw no clear association between MDAQ sensitivity and treatment failure.

Table 1.

Clinical outcomes stratified by in vitro MDAQ sensitivity.

Study regimen	MDAQ sensitivity (IC50, nM)	Number with indicated sensitivity	Number with recrudescent outcomes (%)
AQ/SP	< 60	10	2 (20)
	> 60	5	1 (20)
AQ/AS	< 60	18	1 (5.6)
	> 60	6	0
AL	< 60	20	2 (10)
	> 60	2	1 (50)

Parasites with diminished sensitivity to MDAQ were selected by prior therapy with AQ

To assess the impact on drug sensitivity of recent prior therapy, we compared the MDAQ sensitivity of parasites from subjects that were previously treated with AQ/SP or AS/AQ within 12 weeks with those from patients not treated with these drugs within this interval and with only those from patients in the AL treatment arm, and so not treated with AQ during the course of the study. Parasites from subjects who were previously treated with AQ/SP or AS/AQ within 12 weeks were less sensitive to AQ (n=18; mean IC₅₀ 62.9 nM; range 12.7–158.3 nM) compared to parasites from those not treated with an AQ-containing regimen within 12 weeks (n=43; mean IC₅₀ 37.5 nM; p=0.0085; range 6.3–184.7 nM) or compared to parasites from those in the treatment arm that did not contain AQ (AL subjects; n=20; mean IC₅₀ 28.8 nM; p=0.0042; range 6.3–121.8 nM) (figure 2). Similar associations were seen when drug sensitivities were compared only for parasites causing new infections (not shown) and for infections occurring within 6, 8, or 10 weeks of prior therapy, although with intervals ≤6 weeks differences in drug sensitivities were not statistically significant (table 2).

Figure 2. Selection of parasites with diminished response to AQ.

Sensitivities were determined as described in Methods, and results are plotted for parasites from all patients that did not receive an AQ-containing regimen within 12 weeks (No Prior AQ; A) or only patients in the AL treatment arm, who did not receive an AQ-containing regimen for the course of the study (No Prior AQ; B), in both cases compared to results for subjects who did receive an AQ-containing regimen within 12 weeks prior to the time of this analysis (Prior AQ). Means are indicated by horizontal lines. Differences for both comparisons were significant, as described in the text.

Table 2.

MDAQ sensitivity of parasites causing infections soon after prior infections

Time since prior treatment with AQ	Average MDAQ IC50, nM	n	p value1
6 weeks	46.4	11	0.076
8 weeks	54.5	14	0.019
10 weeks	57.3	17	0.0073
12 weeks	62.9	18	0.0042

¹Mean IC₅₀ values were compared with those for samples from patients in the AL treatment arm (n=20; mean IC₅₀ 28.8 nM) by the Kruskal-Wallis test.

Association of parasite genetic polymorphisms with prior therapy with AQ

We also characterized genetic polymorphisms that have previously been identified in the pfcrt and pfmdr1 genes. Consistent with other results from Kampala [41], all 61 samples that provided in vitro sensitivity results had only the mutant pfcrt allele (76T) that has previously been associated with CQ resistance [26]. As seen previously [11, 28–31, 35], prior treatment with AQ (within 12 weeks) was associated with two pfmdr1 polymorphisms, 86Y and 1246Y, that have been associated with decreased responsiveness to this drug, although baseline prevalences of these polymorphisms were high, and differences were not statistically significant (figure 3). Another polymorphism, 184F, was not selected by prior therapy with AQ, and two other mutations, 1034C and 1042D, were not seen in any samples. Of note, the lower prevalence of pfmdr1 86Y in subjects from the AL treatment group was probably due, in part, to selection of the wild type sequence by AL, as has been shown previously.

Figure 3. Selection of pfmdr1 alleles.

The proportions of samples with mutations at the alleles indicated are shown for the two comparisons described in figure 2.

Association of parasite genetic polymorphisms and MDAQ sensitivity

We also attempted to identify associations between pfmdr1 polymorphisms and in vitro sensitivity to MDAQ. Straightforward correlations were not seen (figure 4). The prevalence of the mutant allele was slightly higher in parasites categorized as resistant for N86Y (prevalence 86Y 38/48 (79.2%) in sensitive vs. 11/13 (84.6%) in resistant), prevalences were the same for 1246Y (prevalence 1246Y 37/48 (77.1%) in sensitive vs. 10/13 (76.9%) in resistant), and the prevalence of the mutant allele was lower in resistant parasites for Y184F (prevalence 184F 9/48 (18.8%) in sensitive vs. 1/13 (7.7%) in resistant). No differences in allele prevalence were statistically significant.

Figure 4. Correlation of in vitro sensitivity to MDAQ with pfmdr1 genotypes.

For each sample studied, the sequence at the three alleles of interest and the in vitro sensitivity to MDAQ are shown.

DISCUSSION

ACT is now the international standard for the treatment of P. falciparum malaria, as multiple new combination regimens offer excellent antimalarial efficacy [4]. However, there is concern that, since all ACTs include a short-acting artemisinin and long-acting partner drug, the regimens will select for resistance to partner drugs, especially in areas where reinfection after treatment is common. This concern is arguably greatest for AQ-containing combinations, as resistance to this drug is already fairly common. AQ is rapidly metabolized to MDAQ [15], an active metabolite which has a half life of about 2 weeks [12]. Thus, after therapy with a regimen containing AQ, MDAQ circulates at decreasing levels for many weeks. We hypothesized that children treated with AQ-containing regimens would be at increased risk of AQ-resistance in subsequent infections. To test this hypothesis, we studied sensitivity to MDAQ in P. falciparum parasites causing uncomplicated malaria in a cohort of children in Kampala who received AQ/SP, AS/AQ, or AL for each episode of uncomplicated malaria. Kampala is known to have a fairly high prevalence of AQ-resistant malaria [42], and parasites that caused malaria in our cohort demonstrated a wide range of sensitivities to MDAQ. Importantly, parasites that caused infections within 12 weeks of a prior treatment with AQ had decreased sensitivity to MDAQ compared to those that caused infections in individuals not recently treated with AQ. Thus, as we hypothesized, prior treatment with AQ selected for parasites with diminished drug sensitivity. This result suggests that resistance to AQ may quite rapidly be selected by treatment of malaria with combinations including AQ.

We identified a broad range of sensitivities to MDAQ in parasites causing uncomplicated malaria in our cohort of Ugandan children. Resistant parasites (based on a cut-off of MDAQ IC₅₀ >60 nM) were seen in 21% of infections, consistent with prior clinical trials in Kampala showing frequent treatment failures with AQ monotherapy [42]. These results and recent similar findings from Kenya (26% of isolates had MDAQ IC₅₀ >60 nM) [21] suggest a more extensive problem with AQ resistance in East Africa than in some other regions of Africa, where >90% of parasites were sensitive to MDAQ in vitro in recent studies in Madagascar [43], Ghana [44], Cameroon [17], Congo [20], Senegal [18], Rwanda [45] and in a collection of isolates from different countries [19]. Older studies that considered sensitivity to AQ, rather than MDAQ, also generally reported high levels of sensitivity, although 16% of samples were reported to demonstrate in vitro resistance in a study from the Central African Republic [46]. In our study, MDAQ-resistant parasites caused infections in those with and without recent prior therapy with AQ; selective pressure from circulating AQ was not required for infection with a resistant strain. However, prior therapy with AQ led to an increased predilection for infection with MDAQ-resistant parasites.

It was also of interest to determine if the in vitro sensitivity of parasites to MDAQ was associated with clinical outcomes. However, of the 61 parasite samples studied, only 7 represented recrudescent infections as determined by multi-locus genotyping, and only 4 of these were in subjects in an AQ-containing treatment arm. Therefore we lacked power to assess associations between in vitro sensitivity to MDAQ and treatment outcomes. Nonetheless, recrudescences occurred after infection with parasites sensitive and resistant to MDAQ, arguing against any simple association between in vitro drug sensitivity and clinical response. Similarly, in studies in Gabon [47] and Kenya [21] in vitro sensitivity to MDAQ did not correlate with clinical response to therapy with AQ. Indeed, it is likely that clinical responses to AQ-containing combination regimens are complex, with effects of varied pharmacokinetics [15], pharmacogenomics [48], host immunity, and other factors in addition to the drug responsiveness of parasites. Nonetheless, parasite drug sensitivity is clearly an important component of a successful treatment response. A better appreciation of mediators of parasite resistance to AQ will be of value in optimizing utilization of available drug regimens.

We also searched for associations between in vitro drug sensitivity and polymorphisms that have previously been linked to altered sensitivity to AQ. The 76T polymorphism in the putative drug transporter pfcrt, which is the primary mediator of resistance to chloroquine, also predicts poor response to AQ [27–29] and is selected in new infections following therapy with AQ [11, 28–31]. The pfcrt 76T mutation was also associated with decreased in vitro sensitivity to MDAQ in field isolates from the Central African Republic [32] and in genetically altered laboratory strains [33]. Considering our results, it is noteworthy that, even with 100% prevalence of 76T in our set of parasites, 79% of samples demonstrated sensitive in vitro responses to MDAQ. Thus, the common pfcrt polymorphism, although predictive of decreased response to MDAQ, does not by itself dictate MDAQ resistance. Polymorphisms in a second putative drug transporter, pfmdr1, have not been as clearly linked to AQ treatment outcome, but in some studies from Africa the pfmdr1 86Y polymorphism predicted poor AQ treatment outcomes [28, 29]. Also, as seen in this study, pfmdr1 86Y was selected by prior therapy with AQ in a number of studies [11, 28–31, 35]. However, prevalence of the pfmdr1 86Y mutation was not associated with in vitro drug sensitivity in samples from the Central African Republic [32], from Colombia [22], or in our study from Uganda. Taken together, our results are consistent with those from other areas, suggesting contributions of polymorphisms in both pfcrt and pfmdr1 to AQ resistance, but likely involvement of additional host (e.g. genetics, pharmacokinetics, immunity) and parasite (e.g. additional polymorphisms) factors in high level AQ resistance.

Our study had some limitations. Our evaluations of in vitro sensitivity of clinical samples were necessarily limited to parasites capable of growing in short term culture to allow measurement of in vitro drug sensitivity. Exclusion of parasites that did not grow in culture may have introduced bias. Infections in Kampala are commonly polyclonal, and both complexity of infection [49] and key polymorphisms in pfmdr1 [50] have been seen to change after in vitro culture. Most of our in vitro sensitivity measurements were thus based on assessment of a mixed population of parasites in which competition between sensitive and resistant strains might obscure results for parasites capable of mediating resistant outcomes. Comparisons with results from other studies must take into account the fact that a number of different assays have been used to measure in vitro drug sensitivity; comparisons of values between studies may be misleading. In vitro measurements leave potential for error, as, since they are measured only during the first life cycle after parasite collection, they cannot be repeated. Thus, for any single measurement, there is the possibility that human error or other factors led to misrepresentation of the true results. Despite these limitations, considering our large set of evaluated samples, our results strongly suggest that prior therapy with AQ selects for decreased sensitivity to the drug.

Our results add concern regarding the long-term prospects for AQ-containing combination therapies for the treatment of P. falciparum malaria. Selection of resistant parasites was readily apparent after prior therapy with AQ. These results suggest diminishing efficacy of AQ-containing combination regimens as they are increasingly used in Africa. The results further highlight the need for continued scrutiny of the efficacies of current antimalarial regimens, as efficacies will likely diminish over time, and for continued efforts to develop new antimalarial treatments that circumvent drug resistance.