Malaria remains one of the most serious vector diseases in the world, and there are between 300 to 500 million cases of malaria annually. Although considered less dangerous than Plasmodium falciparum, Plasmodium vivax infections occasionally cause fatal cases 1234 and have a large economic impact in endemic countries. There are estimated to be 70–80 million cases of P. vivax infection annually 5.

French Guiana, an overseas French territory in equatorial South America, suffers endemic malaria with areas of moderate and high transmission. The epidemiological profiles in the two main endemic foci differ. Plasmodium falciparum is largely predominant along the Maroni River, western border with Surinam, particularly in "Bush Negro territory". Plasmodium vivax is more common in the Oyapock focus, eastern border with Brazil and in the eastern slightly inland areas with a substantial increase since mild 2001 6. Plasmodium malariae is much rarer. The inland regions of French Guiana, located between the two Border Rivers, are uninhabited with the exception of two villages. In coastal regions where three quarters of the population (205,000 inhabitants according the 2008 official censuses) live, occasional bouts of malaria, mainly imported malaria, are recorded.

Plasmodium vivax is characterized by relapses at various time intervals associated with the presence of hypnozoites. Investigations of relapses of P. vivax infection report activation of heterologous hypnozoites in East Timor 7, Thailand and India 8. There are genetic differences between the P. vivax populations originating from these two continents 910. So far, there are few available data on hypnozoite activation during P. vivax relapse in patients in South America. Therefore, a better genotypic characterization of P. vivax at a regional scale is necessary 11. To develop reliable epidemiological tools, appropriate genetic markers and practical molecular techniques are needed. Althought microsatellites are the most highly discriminating markers, their development is laborious and their analysis costs expensive. Merozoite Surface Proteins (MSP) have been extensively studied. These proteins of the erythrocytic stage of the Plasmodium life cycle are potential targets for vaccine and can also be used for genetic typing 12. Pvmsp3a and Pvmsp1 are single copy genes that are highly polymorphic and currently used for P. vivax genotyping 13. The MSP3a protein has a molecular weight of between 148 and 150 KD, and the highly polymorphic region of the gene encodes an alanine-rich central domain. Nucleotide polymorphism of msp3a has been studied by PCR-RFLP in different geographical regions (Iran and Papua New Guinea 14151617) and has also been analysed by gene sequencing in Venezuela and Thailand 1819. Pvmsp1, another well-studied gene 91220212223242526, encodes a protein of approximately 1,720 amino acids, and sequence comparisons indicate that the gene is found as two major sequence types, known as Belem and Salvador 27. Interallelic recombination of these two sequences has been also described 22. The Pvmsp1 gene is long, so many genetic studies have focused on block 5, a polymorphic region suitable for molecular typing.

The objectives of the present study were firstly to analyse the genetic polymorphism of Pvmsp3a and Pvmsp1_b5 markers in French Guiana; and secondly to propose a protocol for distinguishing between new infection or relapse from heterologus hypnozoites and treatment failure or relapse from homologus hypnozoites. The protocol was tested with blood samples from patients with multiple malaria episodes.

Plasmodium vivax diversity has been less extensively studied than that of P. falciparum. A better characterization of P. vivax populations in South America was particularly lacking. A large number of samples must be analysed to describe the genetic structure of a population, so genetic markers that are quickly and easily detectable at a reasonable cost are required. The aims of this work were to study the polymorphism of msp3a and msp1_b5 markers and to develop a method for distinguishing between new P. vivax infections or relapses from heterologous hypnozoites and treatment failure or relapse from homologous hypnozoites. It is of crucial importance clinically when selecting drug treatment, and in particular chosing between drugs with or without effect on the exo-erythrocytic stages. msp3a and msp1 block 5 sequence analysis indicated a high degree of polymorphism, according to different continents and countries, making these genes informative for molecular epidemiological studies. Nevertheless, they also contain conserved regions allowing amplification of the nucleotide sequences of interest. The polymorphism of msp3a can be evaluated rapidly at low cost by PCR-RFLP 14151617. Msp1_b5 sequence reveals substantial diversity 9122021222324252627. The RFLP technique allows a better discrimination of polyclonal infection samples than sequencing 29, so this protocol associating RFLP and sequencing could be used in regions with high rates of polyclonal infections.

Work on relapses of P. vivax infections demonstrates activation of heterologous hypnozoites. The genotype of the acute case is not the same as that of the recurrent infection in about 71% of relapses observed in Australian soldiers returning from East Timor 7, 61% of Thai and Burmese patients and 55% of Indian patients 8.

In French Guiana, 11 different msp3a genotypes were identified among 109 samples and 13.8% had mixed-species infections. Polyclonal infections were thus less frequent than described by Bruce et al and Cui et al 1314, who reported 23% in Papua New Guinea and 19.3% of in Thailand, respectively. Three previously described types were found: in 85.1 (type A), 9.6 (type B) and 5.3% (type C) of the monoclonal samples. The frequencies of these types reported by Bruce et al were 70.5%, 6.7% and 22.8% and by Cui et al were 74.8%, 6.5% and 18.7% for A, B and C, respectively. A Venezuelan study found frequencies of 59.3 for A, 21.9 for B and 18.8% for C 18, and type C has been reported to be much more frequent in southern Iran 15 than in the studied population. Over half (54.3%) of the isolates carried one of only three alleles, A1, A4 and A7, with the other eight genotypes accounting for the remaining 45.7% of the monoclonal samples.

Polymorphism was also studied of the msp1_b5-6 sequence. For this protocol, genotyping msp1 is required in cases for which the msp3a genotype is identical in consecutive samples. Twenty-two different sequences in the 57 samples tested were observed. Four genotypes were strongly represented: H1A, H2B, SAL-1 E and SAL-1 H made up 51.5% of the alleles and 25% overall were H1A. The frequencies of the other 18 alleles were between 4.7 and 1.7%. A study in Iran of msp1_b5 sequences in 145 isolates reported 30 genotypes 26. Seven (12.3%) mixed infections by msp1 sequencing, and like msp3a polymorphism analysis, this suggests that polyclonal infections are less frequent here than in Papua New Guinea with 38% 20. The possible causes for this difference include malaria endemicity, population and vector density.

From these data a maximal msp1 allelic frequency of 5%, at which two samples from a same patient found with the same genotype can be considered to be a treatment failure or a relapse from homologous hypnozoites. In other words, relapse should be concluded with a reasonable risk of error up to 5%. If the genotype found is more represented, error risk in the conclusion of the analysis will be proportional to the genotype occurrence found.

The two methods combined revealed 45 genotypes among the 57 samples and 12 (21%) cases of polyclonal infection. Only four of these cases of polyclonal infection were identified as such by both genotyping methods and eight were identified only by one technique. Five patients were found by msp3a PCR-RFLP to have polyclonal infection, but msp1 sequencing suggested mono-infection. In cases of mixed infection with one genotype being very poorly represented in the biological sample, sequencing may not easily detect the minority genotype. In contrast, msp3a PCR involving nested amplification is more sensitive, and thus an additional allele should be more easily detected as it will give a 'parasite' band in addition to the known PCR band or known RFLP pattern. Thus, PCR-RFLP seems to be more sensitive than direct sequencing for detecting polyclonal infections 29. Inversely, three patients were detected with a polyclonal infection with msp1 sequencing although msp3α PCR-RFLP detected only one genotype (two with genotypes A1 and one with A7). As the frequencies of msp3a genotypes A1 and A7 were high (27.7 and 13.8%, respectively), it is plausible that the two msp1 genotypes in these samples have the same msp3a genotype. The polymorphism of Pvmsp1_b5 and Pvmsp3a as described in many parts of the world, has been observed in French Guiana.

Samples from ten patients who had suffered two or three bouts of malaria (R001 to R010) were analysed. Five episodes were considered to be new infections or relapse from heterologous hypnozoites (patients R001, R002, R004, R005 and third episode for R006). Six of the second bouts were identified as being treatment failures or relapses from homologous hypnozoites. This was concluded for three patients with less than 5% risk of error and with more than 5% for also three others. The time interval between episodes for these six patients was 27 to 50 days. Molecular typing methods cannot distinguish between the both outcomes and treatment failure can occurs one month after the first episode. Therefore, based on these data, the question remains open and it is of crucial importance in clinical trials in endemic areas to know the cause of the latter episode. Moreover, to be sure that in sequential patient sample genotypes are identical, it should be verified using more efficient markers, such as microsatellites.

A high diversity of msp3a and msp1_b5 genetic markers was observed among P. vivax strains in French Guiana and revealed a low polyclonal infection rate. The described protocol associates two molecular techniques for P. vivax genotyping: msp3a PCR-RFLP, which is a low cost method based on a highly polymorphic gene, and msp1_b5 sequencing, a highly discriminatory genetic tool, which provides complementary genotyping data, if required. The methodology can be easily used as a moderate- or high-throughput analysis protocol for clinical trials or epidemiological studies.

The authors declare that they have no competing interests.

VV was involved in all stages of this work and was responsible for the Pvmsp3α PCR-RFLP and for writing the paper. EL was responsible for the supervision of the msp1 B5 sequencing and helped with editing the manuscript. JY, BV and SS participated in the laboratory work. BC helped with editing the manuscript. All authors read and approved the manuscript.