Malaria infection during pregnancy substantially increases the risk of morbidity and mortality to the mother, her foetus and the neonate, and thus constitutes an important public health problem 1 . In areas of high endemicity, the nefarious effects of infection by Plasmodium falciparum are most pronounced in primigravidae, and though less severe, infections in subsequent pregnancies remain associated with anaemia and low birth weight 2 . In areas of low endemicity, the relatively low levels of acquired immunity against malaria parasites increase susceptibility to more severe clinical falciparum episodes during pregnancies, which could result in foetal or maternal death 3 . The impact of infection by Plasmodium vivax, the most prevalent parasite in the Asia-Pacific region, is less pronounced than that associated with P. falciparum 4 . In a large-scale study conducted on the western border of Thailand, an area of low endemicity, P. vivax infections were more common in primigravidae than in multigravidae and were associated with mild maternal anaemia and increased risk of low birth weight 5 . Histopathological examination of the placentas from P. vivax-infected and treated women did not reveal any evidence of parasite sequestration or pathological changes 6 , thus the pathogenic mechanisms remain unclear.

One salient biological characteristic of P. vivax is the formation of dormant liver stages, hypnozoites, by a proportion of the sporozoites inoculated by the infected mosquito. Hypnozoites remain uninucleate and metabolically quiescent for varying durations, that can extend to years, before resuming their development to form mature schizonts that then initiate a new erythrocytic episode upon merozoite release. In tropical areas, P. vivax infections have a short latent period (2 to 6 weeks), and the primary episode is often followed by a succession of relapses (every 3 to 4 weeks) that wane in frequency with time 7 . In the mid 1990’s, the recurrence rate for P. vivax infections treated with chloroquine and followed up for 63 days was found to be 63% (95% CI 57-69%) in non pregnant patients on the Thai-Myanmar border 8 . In a similar study conducted towards the end of the 2000’s in adults and children in the same region the recurrence rate following chloroquine treatment increased to 79.1% (95% CI, 73.5%–84.8%), probably because of increased prevalence of chloroquine resistant P. vivax 9 . Indeed, the first case of P. vivax high-grade resistance to chloroquine in pregnancy was reported recently from the same area 4 . The only drug available to eliminate hypnozoites, thus preventing relapses, is primaquine. Thus, in 1995–1996 when the first P. vivax recurrence in non pregnant patients on the Thai-Myanmar border was treated with primaquine and chloroquine the risk of having a further vivax episode within 2 months was reduced by 96% (95% CI 83-99%) 8 . However, primaquine is contraindicated in pregnancy and in 1986 to 1997 23% (149/634) of pregnant women had two or more parasitaemia episodes 5 .

In recent years, reliable methods to genotype P. vivax populations using microsatellites have been developed 10 11 . Using this methodology it was revealed that P. vivax infections in patients from Thailand, Myanmar and India are often polyclonal, and that the genotype of the parasites in the first relapse following chloroquine treatment 12 differs from that of the initial admission P. vivax population in more than half of the patients. This suggests heterologous activation of hypnozoites, most probably acquired from earlier inoculations. This was supported by data from a study of relapses in infants and in their mothers pre- and post-partum 13 that showed that whereas admission and relapse infections are often genetically heterologous in the mothers, those observed in the children are generally homogeneous. In a recent study P. vivax genetic diversity was compared in pregnant and non pregnant patients in Colombia, and found to be similar in both groups 14 . The purpose of the study presented here is to expand knowledge on the genetic diversity of P. vivax in pregnancy beyond the few studies quoted above. Given the altered immunological and physiological status in pregnancy, it was considered important to ascertain whether the pattern of relapse and the genetic diversity of the P. vivax populations observed differed between pregnant and non-pregnant patients. To this end, a genetic analysis was conducted on archived samples that had been collected in the course of drug treatment studies conducted in villages along the Thai-Myanmar border.

A total of 116 P. vivax isolates were selected for inclusion in this study: 54 were derived from 18 non-pregnant patients who all had two recurrent episodes, and 62 were obtained from 18 pregnant patients who all had two recurrences, four of whom then had a third and two more a fourth recurrence. The characteristics of the patients are provided in Table 1. Non-pregnant patients tended to be of lower age and to have a more frequent history of P. vivax infections in the year preceding the date of recruitment to the study than the pregnant patients. They also tended to have higher temperature and parasitaemia on admission. As expected, pregnant patients had a lower haematocrit than non-pregnant patients. The mean interval time to the first reappearance did not significantly differ between the two groups, 45 days for the pregnant vs 43 days for the non-pregnant patients, nor did it differ for the second reappearance, 51 days vs 48 days. The second recurrence in non-pregnant women was treated with chloroquine and primaquine, and no further recurrences occurred. In the pregnant women the mean times to the third and fourth reappearances were 51 days and 32 days, respectively.

^a mean ± SD (min-max).

^b Number of patients (%).

^c Median (min-max).

^d Geometric mean (min-max).

^e Range.

N = number of patient.

The microsatellite genotyping data were used to calculate the mean number of distinct alleles (A), the heterozygosity (H _e ) and the mean number of distinct allelic variants for each locus in each sample (Table 2). The high number of distinguishable allelic forms observed for each locus and the high value of heterozygosity indicated that overall the P. vivax isolates circulating in the patients had a high degree of genetic diversity. The values did not significantly differ between parasites from pregnant and non-pregnant patients. There was a tendency that did not reach significance for samples from pregnant women to have a lower mean number of distinguishable alleles (A) than those from non-pregnant patients. On the other hand the number of distinct alleles per locus was significantly higher in pregnant vs non-pregnant patients. This was reflected in the higher MOI observed in the combined admission and recurrence samples from the pregnant vs non-pregnant women (Table 3), a difference that nearly reached significance. Nonetheless, the proportion of polyclonal infections was not statistically different between the samples obtained from either group (Table 3). When all the P. vivax samples from the admission vs the recurrent samples were compared (Table 2), there was a significant difference in the mean number of distinct alleles (A) observed for the loci (7.25 vs 9.5, p-value 0.0001), and the isolates from the recurrent episodes showed a higher proportion of loci for which more than one allelic variant was noted in each sample (Figure 1). Linkage disequilibrium was assessed for clonal infections, infections with unique haplotype and all infections, in samples from pregnant patients and non-pregnant patients that were subdivided as admission and recurrence samples (Table 4). No evident for linkage disequilibrium was found for the parasites in the samples obtained on admission, or in the subgroup of recurrence samples with a monoclonal infection. However, significant linkage disequilibrium was found for the recurrent samples from the two groups when considered in their entirety or for the subgroup that has a unique haplotype (Table 4).

P = pregnant women and NP = non-pregnant women.

The No. of alleles and H _e values were calculated from a data set in which only the predominant alleles at each locus was considered.

The No. of alleles per locus values were calculated from all detected alleles at each loci, divided by the total number of samples.

^a Paired t-test was used to compare the means between groups.

*MOI = multiplicity of infection.

The genetic relatedness of the parasites obtained from each patient was assessed for all paired combinations (Figure 2, Table 5). Parasites from the first recurrent episode were different from those of the admission episode in 15 of the 18 pregnant patients and in 11 of the 18 non-pregnant patients, and a similar pattern was observed when the second recurrent episode parasites were compared to those of the first recurrence, in 11/18 and in 12/18 of the pregnant and non-pregnant patients, respectively. In only 1/18 and 3/18 patients from the two groups were all the episodes of the same genotype.

Criteria for interpretation: A = 1 or 2 markers (loci) different with ≤1 possible stepwise mutation; B = 3 markers (loci) different but ≤1 repeat unit, different with ≤1 possible stepwise mutation.

* Chi-square test used to compare in two patients groups.

It is clear from the analyses above that the P. vivax populations in this hypoendemic region on the Thai-Myanmar border exhibited high genetic diversity, with about half of the isolates harbouring polyclonal infections. The differences between the parasites present in admission and recurrent samples, or those in pregnant and non-pregnant patients were relatively subtle. The complexity of the populations (as assessed by the mean number of distinct alleles for the 8 microsatellite markers) was higher in the recurrent as compared to the admission isolates in both groups of patients. The episodes in pregnant women tended to contain parasites with a slightly higher mean number of distinct alleles per locus and consequently a higher MOI. However, there are three caveats connected with sampling that need to be considered before drawing firm conclusions. First, given the relatively small differences, a higher number of patients might be needed. Second, the isolates were collected a few years apart. The fact that the parameters of genetic diversity did not significantly differ between the two groups of samples, suggests that there was little change in the extent of the parasites’ genetic diversity over the periods of sampling (1995–1999), though it is possible that the genetic diversity in the pregnant women does not reflect that in the general population. Third, the amount of blood equivalent to the DNA template aliquots used for genotyping was smaller for the non-pregnant (0.7 μl) than for the pregnant (4 μl) patients. This was inherent to the nature of the archived samples (dried blood spots and whole frozen blood, respectively). This six-fold difference in the volume of blood analyzed might account for the higher MOI in the pregnant women’s episodes, though it should be mitigated by the fact that mean parasitaemia in pregnant patients was half that in non-pregnant women, and that a 33% cut-off threshold was used to exclude the minor alleles. The main reason why this is unlikely is that it is inconsistent with the increase in allelic diversity in recurrent as compared to admission samples.

In conclusion, the parasites causing episodes in pregnant women had a higher genetic diversity than those in non-pregnant patients. This could be due to increased susceptibility to mosquito bites, reduced levels of immunity to infection, physiological changes in pregnancy especially if they affect reticulocyte dynamics, or a combination of these factors. In both patient groups the parasites at recurrence following treatment of the admission episode displayed a higher genetic complexity. In both pregnant and non-pregnant patients, the parasites at recurrence were genetically distinct from those on admission in about half the cases, as were the parasites from the first and the second recurrence. This pattern is similar to that observed earlier for relapsing P. vivax episodes in Thailand 12 13 . Differential accumulation of P. vivax in the placenta has not been noted in this area 6 , though the phenomenon was observed in deliveries at Papua New Guinea 23 , and in vitro cytoadhesion of P. vivax-infected red blood cells to ligands found in the placenta has been reported 24 25 . However, at the time when the samples analysed here were obtained, resistance of P. vivax to chloroquine, the treatment that was administered to both groups of patients, had not been recorded in Thailand 8 26 . Therefore, it is highly likely that the recurrent episodes analysed in this study were relapses originating from hypnozoites. The frequent heterologous nature of these relapse infections and their increased genetic diversity as compared to the admission infections are consistent with the activation of latent hypnozoites hypothesis 27 .

The authors declare that they have no competing interests.

RM, FN, MI, GS, ND and NW were involved in the conception and design of the study. ST performed the laboratory experiments and the analysis. RM, FN and CL were involved in the samples collection. ST and MI wrote the first draft. GS and MI wrote the final draft of the manuscript. All authors read and approved the final manuscript.