Background

Malaria in pregnancy (MiP) is a significant public health problem, with substantial adverse effects on both mother and fetus, including maternal anaemia, fetal loss, premature delivery, intrauterine growth retardation, and delivery of low birth-weight infants, which is a risk factor for death [1,2,3,4]. MiP current control measures used in most endemic countries, according to WHO recommendations, include the use of insecticide treated nets (ITNs), intermittent preventive treatment with sulfadoxine-pyrimethamine (IPTp-SP) and effective case management of malaria, which, since 2010, includes the use of artemisinine-based combination therapy (ACT) [5, 6].

Current efforts of malaria control during pregnancy rely mostly on the effectiveness of anti-malarial drugs used for both IPTp and case management. Indeed, although there are other limiting factors, including low attendance rate of antenatal services [7,8,9,10], low coverage and compliance to the preventive treatment by pregnant women [11,12,13] and inadequate protection of fewer than three SP doses where malaria transmission is intense [12, 14], the major hindrance of the effectiveness of IPTp-SP policy is the spread of Plasmodium falciparum resistance to SP [15,16,17,18,19,20]. Given the number of studies reporting the increased resistance of P. falciparum to SP, there has been several responses: (i) the increase of IPTp-SP doses by WHO in 2012 [21]; (ii) the evaluation of alternative drugs to SP [22, 23]; and, (iii) the assessment of alternative or improved strategies to IPTp-SP [24]. However, none of the alternative drugs or strategies tested has prompted the replacement of the IPTp-SP policy [25, 26]. Reducing the burden of malaria during pregnancy in high transmission setting remains challenging.

Since 2015, WHO recommends the use of ACT for the treatment of P. falciparum uncomplicated malaria during the second and third trimester of pregnancy [27, 28], and such recommendation has already been adopted and implemented by all sub-Saharan African countries [29]. However, there is limited knowledge on the effect of ACT, treatments such as artemether-lumefantrine (AL), on the selection of P. falciparum resistance markers during pregnancy that can affect the treatment outcome. Indeed, available information in non-pregnant women have shown that the P. falciparum multidrug resistance 1 (pfmdr1) N86 and D1246 alleles might be associated with AL resistance [30, 31]. In addition, the combination of N86, 184F and D1246, forming the ‘NFD’ haplotype, led to a decreased susceptibility to AL and treatment with AL selects for such a haplotype [32,33,34]. There is therefore a need for pharmacovigilance studies to monitor any delayed parasite clearance by AL and to assess risk factors associated with the carriage of P. falciparum resistance markers.

Between 2013 and 2016, a multi-centre, cluster-randomized, controlled trial (COSMIC) was conducted in three West African countries with high (Burkina Faso, Benin) and low (The Gambia) malaria transmission, to assess the protective efficacy of adding community-scheduled screening and treatment of malaria during pregnancy (CSST) to standard IPTp-SP (CSST/IPTp-SP) [35, 36]. The CSST/IPTp-SP strategy was based on monthly active follow-up by community health workers using rapid diagnostic tests (RDTs). The aim of the combined CSST/IPTp-SP strategy was to provide an opportunity to detect and treat malaria infections during pregnancy with AL and reduce the prevalence of placental malaria [35]. As part of the COSMIC trial, it has shown a high prevalence of the triple dhfr mutation with presence of quintuple mutants (triple dhfr and double dhps) in Burkina Faso, confirming concerns about the efficacy of IPTp-SP in the near future [37]. This study aimed to determine the prevalence and factors associated with the carriage of pfmdr1 polymorphisms (pfmdr1 N86Y7, Y184F, D1246Y) among pregnant women within the COSMIC trial in Burkina Faso.

Methods

Study site

This study was conducted in Nanoro Health District (NHD) in Burkina Faso between 2013 and 2016. NHD is a rural setting located in the centre-west region of the country, 85 km from the capital city, Ouagadougou. Malaria transmission in the region is endemic and highly seasonal. There is year-round malaria transmission with high transmission season occurring from July to December, corresponding to the wet period in the country. Most malaria cases are due to P. falciparum.

Study participants and sample size

The samples analysed in this study were collected from pregnant women enrolled in a clinical trial assessing the effectiveness of MiP preventive treatments, known as COSMIC trial (Clinical Trials.gov Identifier: NCT1941264). In the COSMIC trial, the protective efficacy of adding CSST during pregnancy to the standard IPTp-SP (CSST/IPTp-SP, intervention arm) was compared to IPTp-SP alone (control arm) [35]. CSST intervention was implemented by community health workers through monthly screening using RDTs and treatment of malaria infection with AL. AL treatment was given for both malaria episodes and infections detected by RDTs in the intervention arm, whereas in the control arm (standard IPTp-SP alone), AL was given for clinical episode treatment only, according to the national guidelines.

For the present analysis, a total of 324 dried blood spots (DBS) collected from finger prick at enrolment (first antenatal care visit (ANC-1)) and at delivery were selected from participants in Burkina Faso based on light microscopy (LM) results as shown in Fig. 1: (i) all available DBS collected at delivery from pregnant women with a P. falciparum infection (N = 162); (ii) all available DBS collected at ANC-1 from pregnant women who experienced P. falciparum infection both at recruitment and delivery (N = 42); and, (iii) a random selection of DBS collected at ANC-1 from pregnant women with a P. falciparum infection (N = 120).

Fig. 1
figure 1

Flow diagram of samples selection

Full size image

Laboratory methods

DNA extraction

Plasmodium falciparum genomic DNA was extracted using the QIAamp 96 DNA Blood Kit (Qiagen, Germany) following the manufacturer’s instructions.

PCR-RLFP

pfmdr1 86Y, 184F and 1246Y mutations were determined using nested PCR followed by restriction fragment length polymorphism (RFLP) analysis as previously described [38]. PCR products were digested with AflIII (NEB), DraI (NEB), and EcoRV (NEB) for the determination of pfmdr1 N86Y, Y184F, and D1246Y alleles, respectively (Table 1). The digested products were visualized by electrophoresis using 2.5% agarose gel for 2 h at 80 V. The nested PCR and digestion reactions were run using 3D7 as wild-type control for the three SPNs and V1 (for N86Y) and 7G8 (for Y184F and D1246Y) as mutant controls.

Table 1 Primer sequences used for the nested PCR and restriction enzymes used for the RLFP
Full size table

Sulfadoxine-pyrimethamine resistance markers

The assessment of P. falciparum mutations in the dhfr gene (codons N51, C59, S108) and the dhps gene (codons A437, K540) was performed as part of the COSMIC trial [35, 37] and, re-analysed in this manuscript. The dhfr and dhps genes were amplified by nested PCR and products were sequenced to identify the targeted mutations as previously reported [37].

Statistical analysis

Data were analysed using STATA version 14.0 (StataCorp, USA). The pfmdr1 genotype profile was determined according to the presence or absence of wild/mutant alleles. Samples in which both wild and mutant alleles were detected were considered as mutant allele carriers. Differences between samples collected at ANC-1 and at delivery were estimated using the Chi-square test for proportions. Factors associated with pfmdr1 86Y mutant allele carriage were assessed by univariate and multivariable logistic regression analyses. Variables with P values ≤ 0.10 in univariate analyses were included in the multivariable models. The investigated variables included: MiP preventive strategy, parasitaemia, age, gravidity, number of malaria episodes, number of IPTp-SP doses received during pregnancy, AL treatment during pregnancy, haemoglobin levels at delivery, and bed-net usage. P values less than 0.05 were considered statistically significant.

Results

Characteristics of study population

The characteristics of the study participants are shown in Table 2. The mean age of pregnant women infected both at ANC-1 and at delivery (21.5 ± 5.6 years) was significantly lower than that of women infected only at ANC-1 (23.9 ± 5.7 years) and of women infected only at delivery (26.5 ± 6.1 years) (P < 0.001). The median parasite density was 980.5 (IQR: 412.5–2,949.2) parasites/µl and 2,237.2 (IQR: 613.5–11,425.7) parasites/µl at ANC-1 and at delivery, respectively (P = 0.02). Among women infected at ANC-1, the median parasite density was significantly higher among those infected both at ANC-1 and at delivery (1,752.5 (IQR: 558.5–5,373.5) parasites/µl) compared to those infected only at ANC-1 (903 (IQR: 324.5–2,308.5) parasites/µl; P = 0.01). Among women infected at delivery, there was no significant difference of the median parasite density between those infected only at delivery and those infected both at ANC-1 and at delivery (P = 0.84). There was no significant difference of the proportion of women who benefited from CSST intervention between the three groups (P = 0.40). The proportion of primigravid and secondigravid women infected both at ANC-1 and at delivery was significantly higher than that of women infected only at ANC-1 and those infected only at delivery (P < 0.001).

Table 2 Characteristics of study population
Full size table

Among pregnant women infected at ANC-1, there was a significant difference of bed-net use between those infected both at ANC-1 and at delivery (72.7% (24/42)) and those only infected at ANC-1 (92.8% (103/120); P = 0.004). Such a difference was not observed among women infected at delivery (P = 0.30). Among pregnant women who experienced malaria infection at delivery, 69% (29/42) of those infected both at ANC-1 and at delivery received at least one AL treatment during pregnancy against 44.2% (53/120) of those infected only at delivery (P = 0.02). There was a significant difference of the time of AL treatment to delivery between women infected only at delivery and that of women infected both at ANC-1 and at delivery (P = 0.05). The proportion of women who received at least three IPTp-SP doses was not significantly different between those infected only at delivery and that of those infected both at ANC-1 and at delivery (P = 0.21). The carriage of triple dhfr 51/59/108 mutation was similar among women infected only at delivery and those infected both at ANC-1 and at delivery (P = 0.60). Only two women were found to carry a double dhps 437/540 mutant allele at delivery. Of note, data on dhfr-dhps mutations among the study popualtion were obtained from the main COSMIC trial [36, 37] and re-analysed in this manuscript.

Plasmodium falciparum resistance markers genotyping success rate

In total, the success rate for pfmdr1 D1246Y genotyping was 92.0% (298/324), with rates of 90.7% (147/162) and 93.2% (151/162) for samples collected at ANC-1 and at delivery, respectively. The pfmdr1 Y184F and N86Y genotyping showed similar success rates as the samples underwent the same nested PCR procedure. The success rate for the samples collected at ANC-1 was 93.2% (151/162), while the success rate for the samples collected at delivery was 71.6% (116/162), giving a total success rate of 82.4% (267/324) for both Y184F and N86Y codons.

Prevalence of pfmdr1 alleles among the study population

The mutant pfmdr1 86Y allele was detected among the study participants with a prevalence of 13.2% (20/131) and 12.1% (14/116) at ANC-1 and at delivery, respectively (P = 0.77). Among women who experienced malaria infection both at ANC-1 and at delivery, the prevalence of the mutant pfmdr1 86Y allele was 7.9% (3/38) and 16.1% (5/31) at ANC-1 and at delivery, respectively (P = 0.25). There was no significant difference of the proportion of pfmdr1 86Y allele carriage between women infected at ANC-1 and those infected at delivery for both women who received the standard IPTp-SP treatment (P = 0.63) and those who received the CSST/IPTp-SP treatment (P = 0.38). By contrast, no mutant alleles corresponding to pfmdr1 Y184F and D1246Y codons were observed in the samples collected from either ANC-1 or delivery (Table 3).

Table 3 Prevalence of pfmdr1 N86Y alleles among the study population
Full size table

Factors associated with pfmdr1 86Y mutant allele carriage at ANC-1 and at delivery

Univariate and multivariable logistic regression analyses were performed to assess the factors associated with the pfmdr1 86Y mutant allele at ANC-1 and at delivery (Tables 4 and 5). None of the variables investigated in the univariate analysis was significantly associated with the carriage of pfmdr1 86Y mutant allele at ANC-1 (P > 0.1). Consequently, no multivariable analysis was undertaken for further assessment (Table 4). At delivery, univariate analyses showed a tendency of an increase risk of pfmdr1 86Y mutant allele carriage with parasite density above the median (P = 0.08). By contrast, there was a tendency towards reduced risk of pfmdr1 86Y mutant allele carriage among women who received at least three IPTp-SP doses during pregnancy (P = 0.06) and those who received at least one AL treatment during pregnancy (P = 0.10). In multivariable analyses, these associations were confirmed as infections at delivery with a parasite density more than the median (2,237.2 (IQR: 613.5–11,425.7) parasites/µl) was associated with an increase risk of pfmdr1 86Y mutant allele carriage (AOR = 5.5 (95% CI 1.07–28.0); P = 0.04). In addition, both three or more IPTp-SP doses (AOR = 0.25 (95% CI 0.07–0.92); P = 0.04) and one or more AL treatment (AOR = 0.25 (95% CI 0.07–0.89); P = 0.03) during pregnancy were associated with a significant reduce risk of pfmdr1 86Y mutant allele carriage at delivery. None of the other variables of interest, including MiP preventive strategy, triple dhrf 51/59/108 mutation, timing of AL treatment to delivery, gravidity, and malaria infection both at ANC-1 and at delivery, showed a significant association with pfmdr1 86Y mutant allele carriage (Table 5).

Table 4 Univariate analyses assessing factors associated with pfmdr1 86Y mutant allele at ANC-1
Full size table
Table 5 Univariate and multivariable analyses assessing factors associated with pfmdr1 N86Y mutant allele at delivery
Full size table

Discussion

Despite the widespread implementation of IPTp-SP to prevent MiP, pregnant women in endemic countries often experience peripheral and/or placental malaria infection at delivery [15, 39, 40]. Although WHO revised IPTp-SP guidelines and increased the SP dose, which has been shown to improve birth outcomes [41, 42], IPTp-SP strategy is still threatened by increasing Plasmodium falciparum resistance to SP. Consequently, there is a need to develop new alternative or improved strategies as part of IPTp-SP policy. In line with the latter, CSST of MiP, in addition to standard IPTp-SP (CSST/IPTp-SP), was tested in Burkina Faso, Benin and The Gambia as an intervention to improve maternal health and birth outcomes in areas of different malaria transmission intensity (COSMIC trial, NCT1941264) [35]. In such a context, the current study was conducted to determine prevalence and factors associated with the carriage of pfmdr1 polymorphisms among pregnant women participating in the COSMIC trial in Burkina Faso.

Mutations in the gene-encoding pfmdr1 are known to be associated with aminoquinoline resistance [43], and therefore represent key P. falciparum markers for monitoring resistance in both susceptible groups (children under 5 years and pregnant women) and the general population. In this study, the analysis was focused on: (i) mutations in pfmdr1 N86Y and D1246Y codons, which have been associated with resistance to chloroquine and amodiaquine [44,45,46], whereas wild-type sequences in these alleles were associated with reduced sensitivity to lumefantrine [44, 47, 48]; and, (ii) mutation in the pfmdr1 Y184F codon, which was associated with altered sensitivity to artemisinins and mefloquine [49]. A prevalence of 13.2 and 12.1% of the pfmdr1 86Y mutant allele was found at ANC-1 and at delivery, respectively. No mutant allele was observed for pfmdr1 Y184F and pfmdr1 D1246Y codons at both ANC-1 and at delivery. The observed prevalence of these mutations at positions 86, 184 and 1246 in this study lack comparable data in the country as previous studies differ with regard to study population (children vs adults vs pregnant women), Plasmodium falciparum isolates (clinical vs asymptomatic infections), and study periods [50,51,52,53]. However, looking at available reports in the study area, these results showed a higher prevalence of the pfmdr1 86Y mutant allele in pregnant women at ANC-1 (13.2%) compared to that in patients with uncomplicated malaria two years before (pre-treatment prevalence of 8.3% in 2010–2012) [50]. An early study conducted by the time of adoption of ACT as a first-line treatment for uncomplicated falciparum malaria in the country (2005), reported a higher prevalence of pfmdr1 86Y mutant allele (35–40%) in children aged 6–59 months with uncomplicated malaria [51]. Surprisingly, no pfmdr1 184F mutant allele was detected among the study population while a prevalence of about 50% was reported among P. falciparum uncomplicated malaria patients in 2010–2012 [50]. In the same study, only three pfmdr1 1246Y mutant alleles were detected in 660 isolates, corresponding to a prevalence of 0.4%.

It has been shown that wild-type sequences of pfmdr1 N86Y, Y184F and D1246Y codons are selected by prior use of AL treatment in malaria patients [47, 48, 51,52,53]. By contrast, selection of the pfmdr1 184F mutant allele has been observed in prior therapy with AL in malaria patients in Uganda [47]. To explore the potential selection of wild-type/mutant sequences of pfmdr1 polymorphisms following AL treatments during pregnancy, the prevalence of pfmdr1 86Y mutant alleles at ANC-1 was compared to that at delivery. No significant difference of the pfmdr1 86Y mutant allele was found neither in the general study population (P = 0.77) nor in sub-groups represented by women infected both at ANC-1 and at delivery (P = 0.25), women who received the standard IPTp-SP (P = 0.63) and women who benefitted from additional screening and treatment of P. falciparum infections (asymptomatic infections) using AL (P = 0.38).

To further assess the potential selection of pfmdr1 N86Y mutant/wild-type alleles by specific factors, factors associated with pfmdr1 86Y mutant allele carriage at ANC-1 and at delivery were investigated. Among the variables evaluated, none was significantly associated with pfmdr1 86Y mutant allele carriage at ANC-1. However, Plasmodium falciparum infection at delivery with high parasitaemia was significantly associated with nearly 5.5 times increase risk of pfmdr1 86Y mutant allele carriage at delivery after adjusting by confounding factors (P = 0.04). By contrast, uptake of at least three IPTp-SP doses and at least one AL treatment were found to be significant protective factors against pfmdr1 86Y mutant allele carriage at delivery in multivariate analyses (75% reduction for both with P = 0.04 and P = 0.03, respectively). These results suggest a benefit in reducing the risk of aminoquinoline resistance marker carriage by high coverage of IPTp-SP doses, which has been shown to reduce parasite load in P. falciparum infection during pregnancy [54, 55]. Moreover, these findings suggest a positive selection of pfmdr1 N86 wild-type allele at delivery following AL treatment during pregnancy in women receiving IPTp-SP, similar to that reported in non-pregnant women in Southeast Asia (Thailand) and East Africa (Kenya) [30, 31] and in patients of all age groups in West Africa (Burkina Faso) [50].

Studies have demonstrated that recent AL use has more of an impact on pfmdr1 N86 wild-type allele prevalence than less recent AL use, as lumefantrine levels decline over time, resulting in less selection [56, 57]. On the other hand, it has been shown that AL could select for pfmdr1 N86 wild-type allele a few months post-treatment in children [58]. This could be explained by genetic variations or other factors including the administration of AL with fatty foods, leading some individuals to exhibit longer artemether or lumefantrine half-lives than other, allowing longer periods of selection [58, 59]. Although, no evidence of an impact of the timing of AL treatment to delivery on the selection of pfmdr1 N86 wild-type allele was observed in this study, the existence of a specific selective window in pregnant women should not be ruled out given the limited number of pfmdr1 86Y mutant allele carriers. In addition, the sub-group of women who experienced malaria infection both at ANC-1 and at delivery did not show a significant association with pfmdr1 86Y mutant allele carriage at delivery and no significant difference was found for triple dhfr 51/59/108 mutations carriage in this sub-group compared to those infected only at delivery (P = 0.6), suggesting new infection parasite population after IPTp-SP and eventually AL treatments. In this regard, the lack of Plasmodium falciparum genotyping to distinguish recrudescent parasites to new infection parasites could be seen as another limit of this study. Future investigations on factors associated with pfmdr1 gene polymorphisms selection in pregnant women living in endemic countries should include large sample size and parasites population genotyping.

Conclusion

This study showed that the uptake of three or more IPTp-SP doses and one or more AL treatments are significantly associated with a reduced risk of pfmdr1 86Y mutant allele carriage in pregnant women at delivery. These findings suggest that both high coverage of IPTp-SP and the use of AL for the treatment of malaria infection/disease during pregnancy select for pfmdr1 N86 wild-type allele at delivery.