Different patterns of pfcrt and pfmdr1 polymorphism in Plasmodium falciparum isolates from Tehama region, Yemen

Introduction. Despite the efforts of the malaria control programme, malaria morbidity is still a common health problem in Yemen, with 60% of the population at risk. Plasmodium falciparum is responsible for 99% of malaria cases. The emergence in Yemen of parasite resistance to chloroquine (CQ) prompted the adoption of artemisinin combination therapy (ACT) in 2009, which involves the use of artesunate plus sulphadoxine-pyrimethamine (AS + SP). However, CQ was retained as the drug of choice for vivax malaria. To assess the impact of the change in the malaria treatment policy five years after its introduction, the present study investigated the mutations in the CQ resistance transporter (pfcrt) and multidrug resistance 1 (pfmdr1) genes. Method. A molecular investigation of 10 codons of pfcrt (72–76, 220, 271, 326, 356, and 371) and five codons of pfmdr1 (86, 184, 1034, 1042, and 1246) was conducted on P. falciparum isolates from districts with the highest malaria endemicity in the Hodeidah and Al-Mahwit governorates in Tehama region, Yemen. A total of 86 positive cases of falciparum monoinfection were investigated for the presence of mutations related to CQ and other antimalarials using a PCR-RFLP assay. Results. There was a wide prevalence of pfcrt gene mutations with the pfcrt 76T CQ resistance marker being predominant (97.7%). The prevalence of other pfcrt mutations varied from high (75E: 88%) to moderate (74I: 79.1%, 220S: 69.8%, 271E and 371I: 53.5%) or low (326S: 36%, 72S: 10.5%). Mutated pfcrt 72–76 amino acids haplotypes were highly prevalent (98.8%). Among these, the CVIET classic, old-world African/Southeast Asian haplotype was the most predominant, and was mostly found in the isolates from the Khamis Bani Saad district of Al-Mahwit (93.1%) and the AdDahi district of Hodeidah (88.9%). However, it was only found in 26.3% of the isolates from the Bajil district of Hodeidah. Surprisingly, the SVMNT new-world South American haplotype was exclusively detected in 9.3% of the isolates from the Bajil district of Hodeidah. Mutations at Y184F of pfmdr1 were found in all isolates (100%) from all districts. The mutation for codons 1034C and 86Y were found only in the isolates from the AdDahi and Khamis Bani Saad districts. Overall, the AdDahi and Khamis Bani Saad districts were similar in terms of carrying most of the mutations in the pfcrt and pfmdr1 genes, while there was a lower prevalence of mutation in the isolates from the Bajil district. Conclusion. The high prevalence of mutations in pfcrt 5 years after the official cessation of CQ use against P. falciparum suggests that there is sustained CQ pressure on P. falciparum isolates in the study area. Moreover, the low prevalence of mutations in the pfmdr1 gene could be a good indicator of the high susceptibility of P. falciparum isolates to antimalarials other than CQ. A new strategy to ensure the complete nationwide withdrawal of CQ from the private drug market is recommended.

Hodeidah. Mutations at Y184F of pfmdr1 were found in all isolates (100%) from all districts. The mutation for codons 1034C and 86Y were found only in the isolates from the AdDahi and Khamis Bani Saad districts. Overall, the AdDahi and Khamis Bani Saad districts were similar in terms of carrying most of the mutations in the pfcrt and pfmdr1 genes, while there was a lower prevalence of mutation in the isolates from the Bajil district. Conclusion. The high prevalence of mutations in pfcrt 5 years after the official cessation of CQ use against P. falciparum suggests that there is sustained CQ pressure on P. falciparum isolates in the study area. Moreover, the low prevalence of mutations in the pfmdr1 gene could be a good indicator of the high susceptibility of P. falciparum isolates to antimalarials other than CQ. A new strategy to ensure the complete nationwide withdrawal of CQ from the private drug market is recommended.

INTRODUCTION
Chloroquine (CQ), a safe and cheap antimalarial, has been the drug of choice for treating uncomplicated falciparum malaria over the past few decades. However, its effectiveness has been greatly hampered by the emergence of resistant Plasmodium falciparum strains that emerged along the Thai-Cambodian border early in 1957 and then spread to other foci in Asia, and also to South America, Papua New Guinea, and Africa (Moore & Lanier, 1961;Harinasuta, Suntharasamai & Viravan, 1965;Grimmond, Donovan & Riley, 1976;Campbell et al., 1979;Fogh, Jepsen & Effersoe, 1979). The emergence and spread of P. falciparum strains that are resistant to CQ and other antimalarials has necessitated a switch to new drugs to ensure maximum effectiveness and prevent the development of ongoing parasite resistance. It has been suggested that a combination therapy rather than monotherapies is highly efficient for falciparum malaria case management and malaria control.
Currently, artemisinin combination therapy (ACT) is the cornerstone of malaria treatment policies and control worldwide (Nosten & White, 2007;White, 2008;World Health Organization, 2015). Interestingly, the replacement of CQ with an ACT in areas where falciparum isolates were CQ-resistant resulted in the re-emergence of CQ-sensitive strains a few years after the cessation of CQ use (Liu et al., 1995;Kublin et al., 2002;Kublin et al., 2003;Mita et al., 2004).
In Yemen, malaria is still a major public health problem that threatens more than 60% of the population, with P. falciparum causing 99% of the malaria cases (World Health Organization, 2015). The emergence in Yemen of parasite resistance to CQ as well as to other antimalarials prompted the adoption of ACT in 2009, which includes artesunate plus sulphadoxine-pyrimethamine (AS + SP) and artemether plus lumefantrine (AL) as the first line and second line treatment, respectively, for uncomplicated falciparum malaria infections. However, CQ was retained as the drug of choice for vivax malaria.
Some in vivo and in vitro studies have been conducted in different malaria-endemic districts of Yemen in order to evaluate the efficacy of CQ against P. falciparum isolates. These studies found that CQ resistance varied from 16.1% in the Taiz governorate (Alkadi, Al-Maktari & Nooman, 2006) to more than 40% in the Lahj (Mubjer et al., 2011) and Hodeidah governorates (Al-Maktari et al., 2003;Al-Shamahy et al., 2007). However, despite these reports and the introduction of a new malaria treatment policy, CQ continues to be available in drug markets and private drug stores in Yemen, is still prescribed for falciparum malaria in some private health facilities, and is also used by individuals for self-administration, especially in remote malarious areas (Bashrahil, Bingouth & Baruzaig, 2010;Ghouth, 2013).
Some molecular studies have reported a high prevalence of pfcrt 76T mutations (ranging from 88% to 100%) in different districts of Yemen (Al-Mekhlafi et al., 2011;Mubjer et al., 2011;Abdul-Ghani, Farag & Allam, 2013;Al-Hamidhi et al., 2013). Nevertheless, there is still a paucity of information about the full genetic profile of pfcrt and pfmdr1 in the country. Yet, such genetic markers are useful tools for monitoring parasite susceptibility to various antimalarials. In addition, they help researchers to monitor changes in parasite genotypes after a change in treatment policy. Therefore, the present study aimed to investigate the susceptibility of P. falciparum to CQ (and some other antimalarials) five years after the adoption of an ACT by conducting a molecular analysis of the related mutations in the pfcrt and pfmdr1 genes in isolates from the Tehama region in northwestern Yemen. Furthermore, because there is a diverse market for antimalarials in this region, improving our knowledge about drug pressure would be useful in targeting interventions aimed at reducing inappropriate drug use in the region.

Study area and subjects
An active case survey that targeted individuals with fever (as a group suspected to have malaria) was conducted from March to May 2014 in two malaria-endemic districts of the Hodeidah governorate and one in the Al-Mahwit governorate in the Tehama region in northwestern Yemen (Fig. 1)  located between Hodeidah and Sana'a (about 111 km west of Sana'a), covers a total area of 2,858 km 2 , and has a total population of 597,000 (National Information Centre, Yemen, 2014). The climate in the Tehama region is a combination of tropical monsoons with occasional rain in summer and dry weather in winter with a mean rainfall of 200 mm/year. The mean temperature is 37.5 • C in summer and 24 • C in winter with humidity ranging between 70% and 90%. Malaria transmission is perennial, but there is a high malaria transmission peak from January to March. Transmission is known to be more prevalent in Hodeidah.
Villages in the districts with the highest malaria endemicity in the region were chosen based on the national malaria records for the period 2010-2013 that were provided by the National Malaria Control Programme (NMCP), Yemen. Based on these records, samples were collected from 10 villages in the AdDahi and Bajil districts in Hodeidah and the Khamis Bani Saad district in Al-Mahwit. The villages in question were Halalah, Al-Huamarah, Al-Meshaahra, Al-Rakib, Al-Rufae, Al-Sharjah, Deer Shareef, Kidf Zumailah, Shat Hajal, and Siraj. Samples were also collected from the city of Bajil.
The present study employed an in vivo efficacy trial that was designed to investigate the effectiveness of the AS + SP therapy in treating uncomplicated malaria and that used the World Health Organization (WHO) 28-day follow-up protocol. Initially, 622 individuals with fever in the 11 mentioned sites were examined. From this number, 86 individuals with falciparum malaria monoinfection, who voluntarily agreed to take part in the in vivo study, were recruited based on the inclusion criteria in the WHO guidelines (WHO, 2009). All 86 were considered for the investigation of different molecular markers related to resistance to antimalarial drugs, including the pfcrt and pfmdr1 genes.

Plasmodium falciparum isolates
A finger-prick blood sample was collected from each participant for use in the RDT test (CareStart TM Malaria HRP2-RDT) and for preparing thick and thin blood film and a filter paper blood spot. Blood films were stained with 5% of buffer-diluted Giemsa stain for 30 min and examined microscopically for malaria parasites. Filter papers were left to dry in air away from dust and then kept in an aluminium pouch together with desiccated silicon bags until used.
Parasite species and stages were recorded and parasitaemia (parasite density) was determined by counting only the asexual stages against 300 white blood cells (WBCs) and then multiplying by 25, assuming the average of the total WBC count of the individuals was equal to 7,500 cells per µl of blood (Billo et al., 2013). The level of parasitaemia was graded as low (<1,000 parasites/µl of blood), moderate (1,000-9,999 parasites/µl of blood), or severe (≥10,000 parasites/µl of blood).

DNA extraction
Two to three discs (6 mm diameter) of 3MM Whatman's filter paper blood spots (cut by a flamed-sterile puncher) were used for DNA extraction, which was performed using a Qiagen blood and tissue kit (Cat. no. 69506; QIAGEN, DNeasy R Blood & Tissue Kit, Germany) according to the manufacturer's instructions. The DNA was eluted using 100 µl of AE (10 mM Tris-Cl; 0.5 mM EDTA; pH 9.0) elution buffer (included in the kit) and kept at -20 • C until used.

Detection of gene mutations in pfcrt and pfmdr1
All extracted DNA samples of P. falciparum were subjected to mutation analysis using polymerase chain reaction (PCR) amplification followed by restricted fragment length polymorphism (RFLP) to investigate the mutations in CQ resistance transporter (pfcrt ) and multidrug resistance1 (pfmdr1) et al. (2000). However, the PCR amplification and restriction enzyme (RE) digestion in the protocol was modified for codon 184 by designing a forward primer MDR184-F (5 -ATAATAATCCTGGATCTAAATTAAGA-3 ) to replace A4 to amplify an amplicon of 155 bp instead of 560 bp and by using Swa1 endonuclease which cuts once into 123 bp and 32 bp for the mutated allele but not the wild.
Restriction enzyme digestion was done in 20 µl of reaction mixture that consisted of 6-8 µl of PCR product, 1X of the specific buffer (included in the RE kit), and 1 unit of the specific RE (New England Biolabs Inc., Hitchin, UK), then was incubated for 15-60 min (based on the RE) according to the manufacturer's instructions, and then visualised in 2.5%-4.0% of TAE buffered agarose gel (based on the size of the cleaves) stained with Sybr R safe DNA gel stain (Invitrogen, USA) using a Bio-Rad Molecular Imager Gel Doc XR System (Bio-Rad, Hercules, CA, USA).
Genomic DNA of P. falciparum strains 3D7 (MRA-102G) and HB3 (MRA-155G) were used as positive controls for the wild types of pfcrt and pfmdr1, while the Dd2 strain (MRA-150G) was used as positive control for the mutated types. All the strains were provided by the Malaria Research and Reference Reagents Resources Center (MR4, ATCCW, Manassas VA, USA).
The study protocol was approved by the Medical Ethics Committee of the University of Malaya Medical Centre, Kuala Lumpur (Ref. 974.19), and by the Ministry of Health and Population, in conjunction with the National Malaria Control Programme in Yemen.

RESULTS
A total of 86 malaria-positive individuals participated in the present study; 40 (46.5%) were male and 46 (53.5%) were female. The ages of participants ranged from 8 months to 65 years, with a mean age of 12.4 years. The age group of 5-15 years represented 58.1% of the study participants. Asexual parasite density (parasitaemia) varied from 561 to 55,555 parasites/µl, with a mean parasitaemia of 8,199 parasites/µl.
The pfcrt gene mutations were screened as molecular markers of CQ, and possibly of other antimalarials such as amodiaquine and lumefantrine. All isolates were successfully amplified (100%). Mutation at pfcrt 76T was found at a virtual fixation level in almost all districts (97.7%), followed by 88.4% for position 75E. The prevalence of mutations at other codons of pfcrt was found to vary from moderate as in codons 74I (79.1%), 220S (69.8%), and 271E and 371I (53.5%) to low as in 326S (36%) and 72S (10.5%). None of the isolates considered in the present study carried the mutated type at codon 356 of pfcrt (see Table 1).
Eight haplotypes of pfcrt 72-76 amino acids were found to be present in the study area. The CVIET classical, old-world African/Southeast Asian haplotype was the most prevalent among all pfcrt haplotypes (77.9%), followed by SVMNT (9.3%) and CVMET (7%). Only one isolate (1.2%) was found to carry each of the other mutated haplotypes, namely CVMEK, CVINT, SVMET and CVMNT, as well as the CVMNK wild haplotype.
As regards the results of the mutation analysis of the pfmdr1 gene, P. falciparum field isolates from the Tehama region showed at least a single point mutation in the pfmdr1 gene (see Table 2). Mutation at position 184 of pfmdr1 was found at the fixation level (100%), together with a low prevalence of mutation for codons 1034 and 86 of pfmdr1 of 20.9% and 16.3%, respectively. No mutation was detected for codons 1042 and 1246.
The NFSND single-mutated haplotype of phenyl alanine amino acid at position 184 was predominant (64%), followed by NFCND (19.8%), and YFSND (15.1%), with an  (184), and cysteine (1034). Interestingly, falciparum malaria in the study area, which consisted of districts known to have the highest malaria endemicity in the country, showed unexpected variation in pfcrt and pfmdr1 gene mutations. Isolates from the AdDahi district in Hodeidah and the Khamis Bani Saad district in Al-Mahwit were almost similar in terms of carrying the majority of mutated alleles for most of the codons, and consequently carrying the mutated haplotypes. In contrast, the malaria isolates from the Bajil district in Hodeidah were found to be mostly of the wild type and consequently carried less mutated haplotypes.
Isolates from the study area showed a high prevalence of CQ resistant haplotypes, with a predominance of CVIET classical old-world African/Southeast Asian haplotypes in the Khamis Bani Saad district of Al-Mahwit (93.1%) and the AdDahi district in Hodeidah (88.9%) compared to only 26.3% for isolates from the Bajil district of Hodeidah. In the same context, mutation at the pfcrt 72 codon was only reported in isolates from the Bajil district (47.4%), whereas it was totally absent from the other two districts. Accordingly, the SVMNT haplotype was found exclusively in 42.1% of isolates from Bajil. (Refer to Table 1. ) Similarly, mutations at pfmdr1 showed a geographic variation in terms of pfmdr1 allele and haplotype frequencies. That is to say, in addition to the fixation level of mutation at pfmdr1 184, mutations at 86 and 1034 of pfmdr1 were found only in the AdDahi and Khamis Bani Saad districts, whereas they were not present in isolates from the Bajil district. The triple-mutated haplotype of 86, 1084 and 1034 (YFCND) was found in one isolate from Khamis Bani Saad. (See Table 2.)

DISCUSSION
The present study provides information on the pfcrt and pfmdr1 genetic profile of P. falciparum isolates from the districts with the highest malaria endemicity in the Tehama region of Yemen. The isolates were found to mostly carry mutated alleles, especially for pfcrt codons 74I, 75E, 76T, 220S, 271E, and 371I, while no mutations were detected at codon 356 of pfcrt. This is consistent with a previous study conducted in Yemen that reported a high prevalence of mutated alleles for codons 74I and 75E (89%) and 76T (93%) (Al-Hamidhi et al., 2013). The present study found that the overall prevalence of pfcrt 76T was 74%-100%, which is similar to the figure previously reported for isolates from other parts of Yemen (Al-Mekhlafi et al., 2011;Mubjer et al., 2011;Bamaga, Mahdy & Lim, 2015).
The high level of mutation at position 76 of pfcrt (96.6%-100%) appears to indicate that, 5 years on from changing the malaria treatment policy to one that is ACT based and the official cessation of CQ use for P. falciparum, the re-emergence of a CQ-sensitive strain of P. falciparum has not yet occurred in the study area. One possible explanation for the virtual fixation of the pfcrt 76 mutation in the study area could be the continued availability of CQ; it is still being prescribed to falciparum malaria patients in Yemen particularly in private health facilities and is also being used as a self-administered treatment (Bashrahil, Bingouth & Baruzaig, 2010;Ghouth, 2013;Bamaga et al., 2014). Moreover, CQ has not yet been totally withdrawn from governmental drug stores as it is still the drug of choice for treating vivax malaria infections.
The findings of the present study support those of previous research conducted elsewhere in Yemen that have reported 79%-100% mutated alleles for pfcrt 76T (Al-Mekhlafi et al., 2011;Mubjer et al., 2011;Abdul-Ghani, Farag & Allam, 2013;Al-Hamidhi et al., 2013). The finding in previous studies such as these that there was a decrease in the prevalence of the pfcrt 76T mutated allele and a re-emergence of CQ-sensitive strains some years after abandoning CQ was controversial, particularly as many studies in Africa documented an increase in the re-emergence of the wild type of pfcrt 76 after complete CQ withdrawal (Laufer et al., 2006;Mohammed et al., 2013;Mbogo et al., 2014;Mekonnen et al., 2014). In contrast, South American isolates of P. falciparum have been reported to retain their mutated types despite the cessation of CQ use (Vieira et al., 2004;Griffing et al., 2010;Adhin, Labadie-Bracho & Bretas, 2013). Similarly, studies conducted in Ethiopia have found a fixation of pfcrt 76T mutations after cessation of CQ use for more than a decade (Mula et al., 2011;Golassa et al., 2014).
The results of the present study also revealed that the CVIET classical old-world, African/ Southeast Asian mutated haplotype was predominant (77.9%) among all pfcrt haplotypes in the study area and that seven other haplotypes were also present, including 9.3% (eight cases) of the SVMNT new-world South American mutated haplotype (only in the Bajil district) and 7.0% (six cases) of CVMET. This finding is in agreement with the only two other available studies based in Yemen that attempted to examine the mutations along the pfcrt gene (Al-Hamidhi et al., 2013;Mubjer et al., 2011). For instance, CVIET was reported for the majority of the isolates from different parts of Yemen (Dhamar 100%, Taiz 88%, and Hodeidah 70.6%), with a presence of only 4% of the SVMNT haplotype in isolates from the Hodeidah governorate (Al-Hamidhi et al., 2013). Similarly, CVIET was the only haplotype reported for isolates from the Al-Musaimeer malaria-endemic district in Lahj Province (Mubjer et al., 2011).
Interestingly, the present study found the SVMNT haplotype exclusively in 42.1% of isolates from the Bajil district in Hodeidah. This finding might be due to a 'parasite response' to a particular drug pressure other than CQ. However, data on the use of a unique antimalarial in this particular area were not available, so it is not possible to relate the occurrence of the SVMNT haplotype to the increased use of any of the antimalarials. A previous study conducted in Tanzania reported the unusual existence of the SVMNT haplotype in Africa and attributed the finding, without supporting evidence, to the increased drug pressure of amodiaquine (AQ) use in the area under study (Alifrangis et al., 2006).
Moreover, malaria transmission in the Bajil district, an area with the SVMNT haplotype, was found to be lower than that in the AdDahi and Khamis Bani Saad districts. This is in line with the study on Tanzania, which reported a higher existence of the SVMNT mutated haplotype in areas with low malaria transmission than in areas with high transmission (Alifrangis et al., 2006). As mentioned above, it has been suggested that the SVMNT haplotype occurs as a response to the intense pressure of AQ, and P. falciparum isolates with SVMNT are supposedly less susceptible to AQ monotherapy and combination therapy. Interestingly, the NMCP in Yemen conducted six in vivo clinical trials for CQ and AQ drug efficacy from 2002 to 2004; four studies looked at CQ efficacy and found that there was 30%-57% treatment failure, while two studies monitored the efficacy of AQ monotherapy and artesunate-amodiaquine (AST-AQ) combination therapy and found 44.3% and 18.5% treatment failure rate, respectively (National Malaria Control Programme (NMCP), 2013, unpublished data).
The existence of SVMNT in Bajil rather than in the other two districts raises an interesting question about the source of the evolution of this CQ resistant haplotype in the country. Typically, SVMNT is associated with new-world South American or Papua New Guinea isolates, and importation of this haplotype from South America or Papua New Guinea, or even from the Philippines or India in Asia, was logically unaccepted based on the population characteristics of the study area. Hence, the apparent independent evolution of SVMNT in Bajil isolates will need further study to confirm the finding presented herein. However, the results of a study conducted in Papua New Guinea suggest that there is a recombination of de novo point mutation that transforms the old-world classical African CVIET haplotype into a new-world South American SVMNT one (Mehlotra et al., 2001).
As regards the pfmdr1 gene, the present study found that mutations varied among the isolates from the three districts, from fixation (100%) for 184F to 20.9% for 1034C and 16.3% for 86Y, while no mutation was detected for the 1042 and 1246 codons. These results are consistent with those of a previous study in Hodeidah that reported a high prevalence (99%) of 184F and a relatively low existence (20%) of the 86Y mutated allele (Al-Hamidhi et al., 2013). Likewise, a recent study conducted in Hadramout in the southeastern part of Yemen, which has low malaria transmission, reported a low frequency (16.7%) of pfmdr1 86 mutated allele (Bamaga, Mahdy & Lim, 2015). However, a high prevalence (70%) for mutations of 1034C and 1042D has been reported for the Taiz, Dhamar, and Hodeidah governorates (Al-Hamidhi et al., 2013).
It has been found that polymorphism in the pfmdr1 gene is related to an increase in parasite tolerance/resistance to some antimalarials (Gamo, 2014). Nevertheless, the findings regarding the association of mutations in the pfmdr1 codons with antimalarial drug resistance are mixed. For instance, 86Y has been linked to CQ resistance, while mutations at positions 184, 1034, 1042, and 1246 were found to be mostly related to resistance to AQ, mefloquine, halofantrine, and quinine (Reed et al., 2000;Danquah et al., 2010;Gamo, 2014). Moreover, the occurrence of pfmdr1 N86 and 184F has been found to increase with the use of lumefantrine (Sisowath et al., 2007;Nwakanma et al., 2008;Malmberg et al., 2013). In the same vein, another study reported an increase of pfmdr1 N86 and 184F, together with the wild alleles of D1246, and concluded that this was linked to the extensive and prolonged use of AQ in combination with artesunate (Fröberg et al., 2012). In other studies, P. falciparum isolates carrying pfmdr1 haplotypes of mutated 86Y and Y184 wild alleles were found to be associated with the usage of CQ or AQ, which continued to change over time with changes in the antimalarials used. For instance, mutated 86Y and Y184 wild alleles changed to a haplotype carrying wild N86 and 184F mutated alleles years after the implementation of ACT, and the selection of parasites to one of artemisinin's derivatives (Humphreys et al., 2007;Dlamini, Beshir & Sutherl, 2010;Mungthin et al., 2010;Thomsen et al., 2011). Overall, the findings of the present study provide a broad view on the polymorphism of pfcrt and pfmdr1 genes in areas with the highest level of malarial transmission in Yemen, 5 years after official cessation of CQ.

CONCLUSION
The present study aimed to provide a genetic profile of pfcrt and pfmdr1 as molecular markers of the P. falciparum parasite's resistance to antimalarials, especially CQ. The high prevalence of pfcrt mutations and mutated haplotypes suggests a high CQ resistance in P. falciparum and a continuation of CQ pressure in the study area. It is therefore recommended that the availability of CQ in private hospitals, clinics, and drugstores should be controlled to ensure its complete withdrawal. Moreover, treatment of vivax malaria using CQ should only be available in governmental health facilities to prevent the random, uncontrolled use of CQ.
• Hesham M. Al-Mekhlafi conceived and designed the experiments, analyzed the data, wrote the paper, prepared figures and/or tables.
• Adel Al-Jasari provided logistic support for data collection and supervises fieldwork.
• Init Ithoi and Yee Ling Lau reviewed drafts of the paper.
• Johari Surin conceived and designed the experiments.

Human Ethics
The following information was supplied relating to ethical approvals (i.e., approving body and any reference numbers): The study protocol was approved by the Medical Ethics Committee of the University of Malaya Medical Centre, Kuala Lumpur (Ref. 974.19), and by the Ministry of Health and Population, in conjunction with the National Malaria Control Programme in Yemen.