Study Design and Samples

The study was conducted during September 2008–October 2010 in Korogwe District in the Tanga Region of northeastern Tanzania, where P. falciparum is the predominant malaria-causing species. A prospective cohort of 924 pregnant women was monitored from first attendance at the antenatal clinic through delivery. The study has been described in detail (22–24). In brief, pregnant women with a gestational age of <24 weeks, as estimated by using ultrasound, were enrolled if they had lived in Korogwe District for >6 months and were willing to give birth at Korogwe District Hospital. Study participants attended 3 additional prescheduled antenatal visits at weeks 26, 30, and 36 of pregnancy, and they were attended to by a study nurse/clinician. Obstetric history and maternal anthropometric measurements were recorded for all women (23). A venous blood sample was collected at each antenatal clinic visit, and venous and placental blood samples were collected at the time of delivery.

Diagnosis, Treatment, and Prevention of Malaria

All blood samples were tested for malaria parasites by using a rapid diagnostic test (RDT) (Parascreen, Zephyr Biomedicals, Goa, India; Paracheck Pf, Orchid Biomedical Systems, Mumbai, India; or ParaHIT-f, Span Diagnostics Ltd, Surat, India) and by microscopy. Blood smears for women with negative RDT results were examined retrospectively, whereas those for women with positive RDT results were examined immediately if deemed necessary by the study physician for a treatment decision (25). Parasite density was determined as the number of asexual stage parasites/200 leukocytes (500 leukocytes if <10 parasites) and converted to the number per microliter, as described (22,25); >100 fields were double-examined before a blood smear was declared negative.

Women with positive RDT results were treated with the antimalarial drug artemether-lumefantrine (Coartem Dispersible, Norvatis Corporation, Suffern, New York, USA) or with quinine. For infections occurring during the first trimester of pregnancy, quinine sulfate–coated tablets (ELYS Chemical Industries Ltd, Nairobi, Kenya) were used, and for severe cases, quinine dihydrochloride injection (Vital Healthcare PVT Ltd, Mumbai, India) was used.

Two doses of IPTp-SP (Sulphadar, Shelys Pharmaceutical Ltd, Dar es Salaam, Tanzania) were given >1 month apart as directly observed treatment; each dose contained 1,500 mg of sulfadoxine and 75 mg of pyrimethamine. Women with a gestational age of >20 weeks at enrollment were given the first IPTp-SP dose at the study-inclusion visit and the second dose during the third trimester. Women with a gestational age of <20 weeks at enrollment were given the first dose at 20 weeks of gestation. Women who had received IPTp-SP before study inclusion but earlier than recommended by the World Health Organization (i.e., after quickening in the second trimester) received a second dose after 20 weeks of gestation, and a third dose was given in the third trimester. Use of SP from private pharmacies/drug shops for malaria treatment before and after study inclusion was also documented. All study participants were provided with a voucher for procuring insecticide-treated nets.

Laboratory Methods and Birthweight Measurements

EDTA-preserved venous blood was used to estimate hemoglobin levels (KX-21N Automated Hematology Analyzer, Sysmex, Kobe, Japan). Live newborns whose birthweights were measured by using a spring scale (Fazzini, Vimodrone, Italy) with an accuracy of <50 g or a digital strain gauge scale (ADE, Hamburg, Germany) with an accuracy of <10 g within 24 h of delivery were included in the birthweight analysis. Newborns with severe malformations, twins, and those born to women with preeclampsia were excluded from analyses because these conditions can severely affect birthweight (23).

EDTA-preserved blood (50 μL) was spotted on Whatman number 3 filter paper (VWR– Bie & Berntsen, Herlev, Denmark), dried at room temperature, and stored in separate zip-lock bags. DNA was extracted by using the Chelex 100 method, as described (26). The DNA supernatant was transferred to a 96-well PCR plate and stored at −20°C until use. The parasite DNA was amplified by outer and nested P. falciparum–specific PCRs, as described (27); the products were analyzed by electrophoresis in 1.5% ethidium bromide–stained gel, as described (22).

To determine the multiplicity of infections, block 2 of the merozoite surface protein 2 domain was amplified by using fluorescent PCR (28). The results were analyzed by using GeneScan software, version 3.7 (Applied Biosystems, Naerum, Denmark).

Parasite DNA was amplified by outer and nested PCR with specific primers targeting the Pfdhfr and Pfdhps genes, as described (29). Single-nucleotide polymorphisms (SNPs) in the Pfdhfr and Pfdhps genes were identified by using a sequence-specific oligonucleotide probe ELISA technique, as described (29) with minor modifications. In brief, we used sequence-specific oligonucleotide probes targeting Pfdhfr codons c50/51 CI/CN, c59 (C/R), c108 (S/N/T), and c164 (I/L) and Pfdhps codons c436/437 (AA/AG/SA/SG/FG), c540 (K/E), c581 (A/G), and c613 (A/S). Individual SNPs were combined to deduce the different infecting mutant haplotypes.

Data Management and Statistical Analyses

We double-entered and validated data in Microsoft Access version 2007 (Redmond, WA, USA). Statistical analyses were conducted by using Stata version 10 (StataCorp, College Station, TX, USA) deploying parametric and nonparametric methods, as appropriate. The effect of infecting allelic haplotypes on birthweights was investigated by using multiple linear regression and dichotomized (as 6 and <6 SNPs) to infecting haplotypes; variables with a p<0.20 in univariate analysis were included in the multivariate models. By using a stepwise backward elimination approach, we obtained final models including only variables with a p<0.10. A 2-sided p-value of <0.05 was considered significant. Final models included only women without missing values.

Demographic and Parasitologic Characteristics of the Study Cohort

Of 1,171 screened pregnant women, 995 met the study inclusion criteria; 924/995 women completed follow-up. For the entire study cohort, 5,555 venous and placental blood samples were collected during antenatal care and at delivery. Among the women completing follow-up, 76 had a total of 96 episodes of malaria. Some women had >1 infection, as determined by RDT and/or blood smears. The median asexual parasite density for study participants was 2,570 asexual stages/μL (range 40–390,749).

A total of 91 samples from women with RDT parasite-positive results were examined by P. falciparum–specific PCR, and 65 (71.4%) were positive. Of these 65 samples, 54 (83.1%) from 49 women were successfully typed in subsequent molecular analyses; 21 (38.9%) of these 54 samples were obtained at study enrollment. The median age of women with samples included in the genotyping analysis was 22 years (range 17−35), and at study inclusion, they had a median gestational age of 17.7 weeks (range 6.9–23.9). Overall, 91.8% of the women in the study cohort received 2 doses of IPTp-SP.

Multiplicity of Infections and P. falciparum dhfr/dhps Genotypes

Sixty percent of the genotyped parasites were polyclonal, with a mean of 2.5 clones/sample (range 1.0–8.0 clones/sample). Of the few women with repeat P. falciparum infections and sequence-specific oligonucleotide probe ELISA data (n = 5), 1 had a recrudescent and 1 had a new infection; genotype results of the 3 remaining infections were inconclusive.

Figure 1

Figure 1. . . . Pfdhfr/Pfdhps haplotyping results for Plasmodium falciparum parasite isolates from 54 parasite isolates in 49 pregnant women, Korogwe District, Tanga Region, Tanzania, September 2008–October 2010. A) Proportion of single...

Of the 54 parasite isolates, 49 (90.7%) were triple Pfdhfr mutants (CIRNI [C50, 51I, 59R, 108N, and I164]); 4 (7.4%) isolates were double Pfdhfr mutants CICNI/CNRNI; and the remaining 1 (1.9%) isolates was wild-type CNCSI (Figure 1A). Two major Pfdhps haplotypes were identified in the 54 isolates: the double mutant AGEAA/SGEAA (A436/S436, A437G, and K540E) haplotype in 28 (51.9%) isolates and the triple mutant Pfdhps SGEGA (S436A, A437G, K540E, and A581G) haplotype in 25 (46.3%) isolates (Figure 1B). An AAKAA wild-type haplotype was identified in only 1 (1.9%) of the 54 isolates (Figure 1B).

The different SNPs were combined to generate the following mutant Pfdhfr/Pfdhps haplotypes: quadruple haplotype (double Pfdhfr [CICNI/CNRNI] with double Pfdhps [AGEAA/SGEAA]); quintuple haplotypes (triple Pfdhfr with double Pfdhps [CNRNI–AGEAA/SGEAA] or double Pfdhfr with triple Pfdhps [CICNI–SGEGA]); and sextuple haplotype (triple Pfdhfr [CIRNI] and triple Pfdhps [SGEGA]). The only Pfdhfr CNCSI wild-type isolate was a double Pfdhps (SGEAA) mutant, and the only Pfdhps AAKAA wild-type haplotype was a triple Pfdhfr (CIRNI) mutant. The isolates with these 2 haplotypes were grouped as quadruple or less when generating the combined Pfdhfr/Pfdhps haplotypes. The quadruple or less, quintuple, and sextuple haplotypes were observed for 9.3%, 46.3%, and 44.4% of the malaria infections, respectively (Figure 1, panel C). For the woman with repeated infections, the highest level of mutation was used in the analyses on pregnancy outcome.

Trends of Pfdhfr and Pfdhps Allelic Haplotypes by Gestational Age

Figure 2

Figure 2. . . . Proportion of mutant Plasmodium falciparum dihydrofolate reductase and dihydropteroate synthetase haplotypes among pregnant women, Korogwe District, Tanga Region, Tanzania, September 2008–October 2010. Proportions are shown by gestational age;...

Quadruple and less mutated haplotypes were observed mainly during early pregnancy, but the quintuple and sextuple haplotypes were observed throughout pregnancy (Figure 2); there was no clear trend indicating a selection by IPTp-SP use for the most mutated haplotypes (Table 1). Among the women who did not receive IPTp-SP before malaria infection, 3 with quintuple and 1 with sextuple haplotype infections reported using SP for treatment of suspected malaria before study inclusion. Even when both exposures to SP (i.e., as IPTp-SP and as SP treatment) were taken into account, no trend toward accumulation of resistant haplotypes as a result of SP use was observed (Table 1).

Effect of Sextuple Haplotypes on Pregnancy Outcome

There were no statistically significant differences in the characteristics of women infected with highly mutated parasites and those with less mutated parasites with respect to age, gravidity, and anthropometry (Table 2). Of the 26 women with quintuple or less haplotype infections and 23 with sextuple haplotype infections, 24 and 20, respectively, received 2 doses of IPTp-SP. Two women with quintuple or less haplotype infections and 2 with sextuple haplotype infections received no or 1 dose of IPTp-SP, and 1 woman with sextuple haplotype infection received 3 doses. The levels of parasitemia were also similar between the 2 groups, and there were no differences in the incidence of fever (axillary temperature >37.5°C) or level of hemoglobin (g/dL) at the time of infection or at delivery (Table 2).

Among the 49 newborns born to women for whom P. falciparum strains were genotyped, there was 1 case of preeclampsia, 2 twin deliveries, and 1 newborn with congenital malformation. Among the remaining 45 singleton newborns, 36 had their birthweight measured within 24 hours of delivery and, thus, were eligible for analysis. There was no difference in the distribution of haplotypes between mothers of the included and excluded newborns (data not shown).

The median birthweight for newborns of women with sextuple haplotype infections was significantly lower than that for newborns of women with less mutated haplotype infections (Table 2). When using z-scores (adjusting for gestational age at delivery and sex of newborn) based on a Tanzanian reference chart from the same cohort (23), borderline significance toward lower z-scores in women with sextuple mutant haplotype infections was observed (p = 0.06). There was no difference in the incidence of low birthweight newborns or of newborns that were small for gestational age at birth among women with sextuple haplotype and less mutated haplotype infections (Table 2). When the effect on birthweight was further analyzed by using a stepwise multiple linear regression model adjusting for covariates, newborns of women with sextuple mutant haplotype infections had a significantly reduced birthweight (mean reduction 359 g; 95% CI, −601 to −117) compared with newborns of women with quintuple or less haplotype infections (p = 0.005) (Table 3) Covariates for the linear regression model were mid upper arm circumference at study inclusion, parasite density, hemoglobin level, number of IPTp-SP doses, gestational age at inclusion and at delivery, number of infections, maternal age, weight/weight gain during pregnancy, HIV status, place of delivery, sex of the newborn, and gravidity.

Women in whom sextuple haplotype parasites were identified more often had repeated infections than those in whom less mutated haplotypes were identified (Table 2). In the univariate analysis, there was a borderline significance toward decreased birthweight after repeated infections (p = 0.08). However, this effect was not observed in the multivariate model (Table 3).

There was a trend toward decreased birthweight with increasing numbers of IPTp-SP doses received (Table 3). This trend was explained by the fact that newborns of the 3 women who received 0 or 1 IPTp-SP dose had much higher birthweights than newborns of the 33 women who received 2 doses (3,690 g vs. 2,920 g, respectively; p = 0.04). However, in the entire cohort (924 women) there was no association between the number of IPTp-SP doses and birthweight (data not shown). Furthermore, the number of IPTp-SP doses received before time of infection and the time between the most recent IPTp-SP dose and when infection occurred was not associated with birthweight. Adding these variables to the multivariate model did not alter the result (data not shown).