Assessment of copy number variation in genes related to drug resistance in Plasmodium vivax and Plasmodium falciparum isolates from the Brazilian Amazon and a systematic review of the literature

Sixty-seven patients infected by P. vivax or P. falciparum from three Brazilian states (31 from Mato Grosso, 26 from Rondônia and 10 from Amapá) were included in this study (Table 1). Epidemiological data indicate that malaria transmission in these areas is generally hypo- to meso-endemic [13]. In Rondônia and Amapá, the individuals were locally infected. However, individuals from Mato Grosso may have acquired the infection in other localities of the Amazon region because there is no transmission in the municipality of blood collection. The eligibility criteria included (i) patients with symptomatic P. vivax or P. falciparum single infection with any parasitaemia by microscopic examination; (ii) the absence of severe complications of malaria; and, (iii) if the patient was female, the absence of pregnancy. The confirmation of Plasmodium spp. infection by microscopy was based on Giemsa-stained thick blood smears evaluated by well-trained microscopists in accordance with the malaria diagnosis guidelines of the Brazilian Ministry of Health. DNA samples were extracted from peripheral blood collected in EDTA-containing tubes for parasite genomic analysis using QIAamp DNA kit (QIAGEN, Chatsworth, CA, USA). Samples were collected between August 2002 and May 2012. Molecular detection and identification of Plasmodium species were confirmed later by nested PCR amplification of the 18S rRNA gene in the laboratory to exclude mixed malaria infections [38].

The copy number variation of Plasmodium genes was determined by quantitative real-time PCR (qPCR) using specific hydrolysis probes and oligonucleotide primers for each gene. Probes and primers previously described were used for amplification of pfmdr1 [27], pfgch1 [28] and pvmdr1 [31]. Amplification reactions were performed in a total volume of 10 µL in the presence of 5 µL of Taqman ^® Universal PCR Master Mix 2× (Applied Biosystems, AB, Foster City, CA, USA), 1 µL DNA (≈100 ng/µL), 900 nM (pfmdr1, pfgch1 and pvmdr1) or 300 nM (pftubulin and pvtubulin) of forward primer, 900 nM of reverse primer (for all genes), and 200 nM (pfmdr1, pfgch1 and pvmdr1) or 250 nM (pftubulin and pvtubulin) of the probe. The cycling parameters for PCR were as follows: initial denaturation at 95 °C for 10 min, 40 cycles of 15 s at 95 °C and 1 min at 60 °C. The PCR was performed in triplicate in the Applied Biosystems Viia7 real-time PCR system (AB) in 384 plates. The single-copy β-tubulin gene was used as a reference gene (normalizer) [27, 31], and a field sample with a single copy of the target gene was used as a calibrator. The ΔΔCt method was used to estimate the copy number of pfmdr1 and pvmdr1 genes relative to a standard calibrator sample. For the pfgch1 gene, a calibration curve was used with plasmids containing the pfgch1 insert. The samples were considered to have a copy number equal to 1 when the value of the relative quantification was between 0.5 and 1.5, and values with a minimum relative quantification >1.5 were defined as amplified. Only samples with a cycle threshold <32 and a Ct standard deviation <0.3 were considered in the analysis. Each experiment was performed in triplicate, and gene amplification was determined by at least two independent experiments. The accuracy of the qPCR assay was determined from at least three independent experiments, each performed in triplicate, of 19% of the total samples assayed and for the reference laboratory isolate W2mef (2 copies for pfmdr1 and 1 copy for pfgch1) (Additional file 1).

PCR followed by restriction fragment length polymorphism analysis (RFLP) was performed to identify SNPs at codons 76 of pfcrt and 86 of pfmdr1, according to Durand et al. [39] and Lopes et al. [40], respectively. The amplicons were submitted to enzyme digestion with ApoI (New England BioLabs, Ipswich, MA, USA) for 2 h at 50 °C. Then, the enzyme was inactivated by exposure to a temperature of 80 °C for 20 min at and the fragments were visualized on a silver-stained 12% polyacrylamide gel.

A literature search was conducted between May 2014 and March 2017. An advanced search was made using key words (Plasmodium, copy number, resistance, gene, copy number resistance and gene copy). These key words were identified in the title, abstract and text, according the search algorithm of each database. Manual exclusion criteria included: studies with other Plasmodium species than P. falciparum and P. vivax; studies with laboratory strains grown in vitro; other systematic reviews; and studies with genes related to anti-malarial resistance other than pfmdr1, pfgch1 and pvmdr1. All duplicate articles and review articles were excluded; the clinical studies were included in this systematic review. The title, abstract and methods section of all 309 articles (Additional file 2) were scanned to identify the studies according to the inclusion criteria. From the eligible studies the following data were extracted: author, year of sample collection (when available), country where the study was performed, sample number and the frequency of CNV in parasite population by gene assayed.

31 samples of patients infected by P. falciparum were evaluated for copy number variation in the pfmdr1 gene and 25 samples for pfgch1. Parasite isolates were sampled in three regions of Brazil (the Mato Grosso, Rondônia and Amapá states) over a period of 10 years (Table 1). For both genes analysed in this study all P. falciparum isolates carried only one copy of the gene (Fig. 1; Additional file 3A, B).

Polymorphisms in pfmdr1 are related to resistance to several anti-malarials, including CQ, MQ, QN and HF [41]. An increase in the copy number of this gene is the most important determinant of resistance to MQ and to reduced artesunate sensitivity in vitro [26, 27, 42]. In Brazil, P. falciparum resistance to anti-malarials has been reported since 1910 (QN—1910; CQ—1960s; SP—1980s; amodiaquine and MQ—1990s) [6]. Another study also assessed the presence of CNV in the pfmdr1 gene from samples collected over four decades in Brazil (from 1984 to 2011), in which patients were treated with QN plus tetracycline, MQ or ACT [43]. During that period, the rate of pfmdr1 amplification reached 42%. Particularly, the usage of drugs such as MQ, which is related to selection of resistant parasites with pfmdr1 amplification, could be an explanation for the observed high rates of amplification. A noticeable difference between the present study and that by Inoue et al. is the region and the period of sample collection (2000s herein), which could lead to the observed differences in genetic diversity. Accordingly, geographical foci of P. falciparum with distinct population structures have been described in the Brazilian Amazon region [44]. However, as data from the previous study have not been shown by region, differences in CNV distribution due to the location of the infection could not be assessed here. Alternatively, cases of a decline in resistant isolates by the withdrawal of drug pressure could have occurred as previously reported in Africa [45, 46]. The genome of P. falciparum is known to be plastic [47], which means that the parasite can experience rapid and extensive variation in response to changes in its environment; this is also known as phenotypic plasticity [48]. In a short timescale, this phenotypic plasticity can modify or produce new phenotypes. Thus, this phenomenon enables parasites to maximize fitness by mechanisms such as differential gene expression [49]. An example was shown by Preechapornkul et al. where strains containing a single copy of the gene preferentially survived in the absence of MQ compared to those who had multiple copies [25]. Parasites that had fewer copies had a greater survival rate due to this energy demand generated by P-gh1. This protein acts as an efflux pump that transports substrates against a concentration gradient with ATP hydrolysis [50]. In Brazil, MQ was officially introduced in 1987 for the treatment of CQ- and SP-resistant P. falciparum, and few years later, in 2001, MQ was substituted with the combination of QN plus doxycycline, followed by primaquine (PQ) as a first line therapy [6]. At this time, MQ plus PQ was instituted as a second line drug combination, and only after 2006 did the ACT become adopted in Brazil [6]. Thus, the lower pressure exerted by MQ may have led to selection of parasites bearing no genetic amplification after the first decade of the 2000s.

Plasmodium falciparum resistance to SP is associated with SNPs in two genes in the folate pathway, dihydrofolate reductase (dhfr) and dihydropteroate synthetase (dhps) [51, 52]. However, polymorphisms in these genes may alter their efficiencies. The pfgch1 gene encodes the first enzyme of this pathway and an increased copy number of this gene may act as a compensatory mechanism since it is associated with a greater number of point mutations in dhfr and dhps [28, 53, 54]. Few studies have evaluated the role of CNV in resistance to SP, particularly in the Americas. The present study was the first to evaluate gch1 CNV in P. falciparum isolates from Brazil. Samples analysed here date from the 2000s, more than twenty years after the withdrawal of SP in the Brazilian therapeutic scheme in the late 1980s. Thus, the absence of amplification of the pfgch1 gene in the Brazilian isolates may be the result of antifolates withdrawal, which could have favored the spread of parasites bearing a single copy of the gene. In fact, the presence of CNV in this gene seems to not be advantageous in the absence of the drug pressure, having impact on parasite growth [53]. Still, since the evolution of antifolate-resistant parasites is multifaceted and complex and the activity of the gch1 gene is linked with other enzymes (e.g., dhps and dhfr), further analysis is required to confirm the significance of gch1 CNV on the gain in SP resistance.

The samples of P. falciparum were assayed for two polymorphisms in pfcrt (K76T) and pfmdr1 genes (N86Y). Twenty-seven samples (84% [27/32]) were successfully genotyped (10 from Amapá, 16 from Mato Grosso and 1 from Rondônia) for both genes. For pfcrt gene, all samples carried the polymorphism at codon 76, while the substitution N86Y in the pfmdr1 gene was absent (Fig. 1). The fixation of the 76T allele is in agreement with other studies in Brazil [43, 55, 56], despite the withdrawal of CQ from national treatment guidelines in the mid-1980s [57]. The presence of the 76Y allele is in accordance with the description of isolates resistant to CQ in the same regions analysed here [43, 55]. The K76T mutation has been proven essential to CQ resistance, as demonstrated by transfection experiments and it was suggested that this mutation in the pfcrt gene confers resistance to CQ by reducing the amount of drug in the digestive vacuole of the parasite [33, 34, 58, 59].

Field studies have observed a significant association between the pfcrt (76T) and pfmdr1 (86T) alleles, suggesting a joint contribution of these two genes to CQ resistance [60, 61]. Reinforcing these findings, the study by Veiga et al. using zinc-finger nucleases to genetically modify the pfmdr1 gene confirmed the contribution of the N86Y substitution to CQ and amodiaquine resistance. In contrast, this substitution increased parasite susceptibility to MQ and lumefantrine [62]. Additionally, an association between the 86Y mutant and reduced susceptibility to artemether was found using in vitro tests [63], although other studies have reported contradictory results with the N86Y substitution associated with increased artemether sensitivity [45, 64, 65]. In samples analysed here, the N86Y substitution was absent, a similar result to that observed by other studies that have also reported low frequency or absence of this substitution in pfmdr1 in Brazil [43, 55, 56]. On the other hand, Inoue et al. reported the emergence of the 86Y mutant in 25 and 43% of isolates from Brazil and Guyana, respectively, over the last few decades. These findings led the authors to hypothesize that the increase in prevalence of the 86Y mutant could be related to the indiscriminate use of ACT in Guyana and the high flow of gold mine workers between Brazil and this neighboring country. Although not analysed here, other polymorphisms (e.g. Y184F, S1034C, N1042D and D1246Y) that influence parasite response to different drugs such as CQ, MQ, QN, HF and artemisinin [41, 66, 67] are found with high frequency in South America [43, 55, 68‐70]. In this way, it seems that the markers of CQ resistance may differ between South America and other endemic areas. Thus, it is necessary to define better and more reliable polymorphisms to characterize South American samples.

To characterize the pattern of copy number variation of P. vivax isolates from the Brazilian Amazon region, thirty-five samples collected in various time periods and from two different areas (the states of Mato Grosso and Rondônia) were analysed. Overall, an amplification rate of 20% (7/35) was found for the pvmdr1 gene (Fig. 1). In Mato Grosso, six of 13 (46%) isolates had multiple copies of the pvmdr1 gene (Additional file 3C). Despite being collected in the state of Mato Grosso, the probable locality of infection for these individuals carrying isolates with multiple copies of pvmdr1 included others areas of the Amazon region, including the states of Pará and Rondônia, as well as the neighboring country French Guiana. In addition, the majority of these isolates were collected in the 2010s decade. For 22 samples collected in Rondônia only one (4%) had multiple copies of the pvmdr1 gene (Additional file 3C). Overall, pvmdr1 copy number variation was not related to parasitaemia (median: 3150 [IQR 2935-12,000 parasites/μ] for individuals carrying isolates with multiple copies vs. 2483 [992-6175 parasites/μ] for samples with one copy of pvmdr1; Mann–Whitney U test, P = 0.201) or age (mean: 32 [SD 14.8 years] for samples with multiple copies vs. 37 [15.1 years] for samples with one copy of pvmdr1; Student’s t test, P = 0.469). Among samples with multiple copies of pvmdr1, the recurrence of the disease was reported in two individuals from Mato Grosso in a period of 45 days to 3 months after the initial infection.

In Brazil and in several other malaria endemic areas CQ and PQ are still the main drugs used to treat P. vivax [71]. However, resistance to CQ has been reported in the Brazilian Amazon [14‐17]. Similar to P. falciparum, the copy number variation on pvmdr1 is supposed to be related to drug resistance [31]. In Brazil, only one study described pvmdr1 copy number variation, showing an amplification rate of 0.9% in isolates from the state of Acre [72]. Furthermore, the authors have not observed any relation between SNPs described in pvmdr1 and CQ resistance. In this study, pvmdr1 amplification was observed in 20% of patients infected in different areas of Northern Brazil. One individual was declared to have been infected in French Guiana, where a high number of isolates with multiple copies of the pvmdr1 gene has previously been reported [73, 74]. The clinical features of age and parasitaemia, which are important predictors of patient’s response to the treatment [27, 75, 76] did not differ between individuals carrying parasites with one or multiple copies of pvmdr1 in the present study. Thus, for the two individuals bearing isolates with multiple copies of the gene who also had episodes of recurrence, there is the possibility of treatment failure due to parasite resistance to anti-malarials. However, as the individuals were not followed up after drug therapy with plasma drug level and parasitaemia measurements, the authors could not exclude the possibility of inadequate drug absorption. Additionally, the authors could not exclude the possibility of recurrence due to CQ or PQ-impaired metabolism by variants of cytochrome P450 isoenzymes [77], as the genes that encode the enzymes were not assayed here. On the other hand, a new infection seems to be unlikely since these individuals reside in an area without active transmission of malaria and did not travel to other endemic areas in the Brazilian Amazon after their initial infections.

As previously suggested for other geographical locations, pvmdr1 amplification could be associated with MQ pressure [31, 78] in a mixed infection context, when this drug was used alone or in combination to treat P. falciparum malaria. Accordingly, P. vivax resistance to MQ has been reported in the Brazilian Amazon region [16, 55, 79]. In P. falciparum, resistance to MQ has been associated with increased copy number of pfmdr1 [27]. The mechanisms of MQ resistance seem to be similar between P. vivax and P. falciparum. In areas of Southeast Asia with intense and sustained MQ pressure, gene amplifications of pvmdr1 were significantly more common than in those patients from other localities where there had been less parasite exposure to MQ [31]. Additionally, Suwanarusk et al. reported amplification of 21% in the pvmdr1 gene in P. vivax isolates from Thailand, which was associated with a twofold increase in MQ IC₅₀ [32]. Mefloquine and CQ seem to exert selection pressure in opposite direction on pvmdr1, with gene amplification associated with an increase of susceptibility to CQ [32]. Therefore, the observed pattern of pvmdr1 copy number variation in the Brazilian Amazon region and elsewhere would be the result of pressure exerted by the two drugs. For Brazilian samples, an increase in the prevalence of pvmdr1 amplification was observed over the 7-years period analysed. However, this finding requires further investigation as it might indicate that the P. vivax population is undergoing an increased susceptibility to CQ.

A total of 309 articles were selected from four databases (PubMed, Malaria Journal, Science Direct and CAPES), and 82 of these were included in the systematic review according to the established criteria. The data of articles included in this study are available in supplementary material (Additional file 4). Studies from four continents constituted the review, where Asia and Africa were the regions with the majority of selected articles (88%). The majority of the studies were about P. falciparum, related to copy number variation in pfmdr1. Only two studies analysed the pfgch1 gene. For the pvmdr1 gene, eight studies were eligible for inclusion.

The distribution of resistant isolates to anti-malarials is different between the endemic regions of the world, as well as the profile of CNV in genes of the parasite. In the Americas and Africa, the amplification of pfmdr1 is not common and is restricted to certain localities. Thus, the majority of the studies in the Americas reported an absence of pfmdr1 amplification in P. falciparum isolates; the amplification rates varied from 6 to 38% (Fig. 2; Additional file 4) [43, 69, 80‐83]. For some countries such as Brazil and French Guiana, there was approximately 30% pfmdr1 amplification (between 2 and 4 copies) in the samples analysed. In contrast, in Africa, the prevalence of isolates carrying one copy of pfmdr1 reached almost 100% in different countries, with the exception of Ethiopia (Fig. 2). Unlike these regions, Asian countries present a high rate and distribution of isolates with pfmdr1 amplification (Fig. 2). The studies show that on the Thailand-Myanmar border, the amplification rate is greater, reaching 100% [84, 85]. In general, Southeast Asian countries had a greater prevalence of P. falciparum isolates with pfmdr1 amplification, while isolates from Southwest Asia showed no gene amplification [86‐90]. Taken together, these differences in gene amplification rates may be the result of the adoption of different treatment regimens for malaria by different countries over the years. In fact, there is evidence that amplification of the pfmdr1 gene in P. falciparum has arisen as multiple independent events, suggesting that this region of the genome is under strong selective pressure [91]. Since gene amplification has a cost to the parasite’s fitness [25, 92], the drug pressure withdrawal could also have favored parasites without gene amplification contributing to the genetic variability observed.

Regarding pfgch1, only two studies evaluated the profile of CNV in field isolates. In Africa, Kiwuwa et al. have not detected any amplification in the 21 isolates analysed [93]. However, in Southeast Asia a significant difference in the rates of pfgch1 amplification was reported for isolates from Thailand (72%) and Laos (2%), countries that have contrasting selection histories with antifolate drugs [28]. Whereas SP was not extensively used in Laos until 2006, despite their use as official second line treatment for malaria [94], Thailand has had a longstanding history of antifolate use [95]. This geographical differentiation on pfgch1 CNV suggests local adaptations to drug pressure, which has been experimentally corroborated [53].

As described for P. falciparum, there is a difference regarding the distribution of CNV allelic frequencies of P. vivax isolates in the endemic areas. In the Americas and Africa, only a few studies have assessed pvmdr1 CNV, finding very low rates of amplification (1–3%) (Fig. 3) [72, 74, 83, 96]. Conversely, a higher rate (up to 59%) of isolates harboring multiple copies of pvmdr1 was reported in French Guiana, where MQ was widely used alone or in combination with artesunate for treatment of uncomplicated P. falciparum malaria, and therefore P. vivax were subjected to indirect selection pressure by the drug [73, 74]. In Asia, pvmdr1 amplification was reported in Cambodia (4 to 33%) [37, 74] and in Thailand (7–39%) [37, 97]. Commonly, in these areas where pvmdr1 amplification is frequent, there has been intense current or past use of mefloquine to treat uncomplicated P. falciparum malaria [37, 74].