Naturally acquired Duffy-binding protein-specific binding inhibitory antibodies confer protection from blood-stage Plasmodium vivax infection

Abstract

Individuals residing in malaria-endemic regions acquire protective immunity after repeated infection with malaria parasites; however, mechanisms of protective immunity and their immune correlates are poorly understood. Blood-stage infection with Plasmodium vivax depends completely on interaction of P. vivax Duffy-binding protein (PvDBP) with the Duffy antigen on host erythrocytes. Here, we performed a prospective cohort treatment/reinfection study of children (5–14 years) residing in a P. vivax-endemic region of Papua New Guinea (PNG) in which children were cleared of blood-stage infection and then examined biweekly for reinfection for 25 weeks. To test the hypothesis that naturally acquired binding inhibitory antibodies (BIAbs) targeting PvDBP region II (PvDBPII) provide protection against P. vivax infection, we used a quantitative receptor-binding assay to distinguish between antibodies that merely recognize PvDBP and those that inhibit binding to Duffy. The presence of high-level BIAbs (>90% inhibition of PvDBPII-Duffy binding, n = 18) before treatment was associated with delayed time to P. vivax reinfection diagnosed by light microscopy (P = 0.02), 55% reduced risk of P. vivax reinfection (Hazard's ratio = 0.45, P = 0.04), and 48% reduction in geometric mean P. vivax parasitemia (P < 0.001) when compared with children with low-level BIAbs (n = 148). Further, we found that stable, high-level BIAbs displayed strain-transcending inhibition by reducing reinfection with similar efficiency of PNG P. vivax strains characterized by six diverse PvDBPII haplotypes. These observations demonstrate a functional correlate of protective immunity in vivo and provide support for developing a vaccine against P. vivax malaria based on PvDBPII.

Protective immunity against malaria is acquired by residents of endemic regions over a period of years after repeated exposure to malaria parasites (1). Naturally acquired immunity to Plasmodium falciparum and Plasmodium vivax malaria does not prevent infection by these parasites completely, but limits parasite densities, reduces the frequency of clinical malaria episodes, and prevents severe disease (2). Humoral immune responses against blood-stage antigens are an important component of naturally acquired immunity to malaria (2, 3). Therefore, parasite proteins engaged in erythrocyte invasion are potential targets of protective antibody responses.

Interaction between P. vivax Duffy-binding protein (PvDBP) and the Duffy antigen receptor (DA) on host erythrocytes is central to blood-stage infection by P. vivax (4, 5). Adhesion to DA is mediated by the 140-kDa PvDBP (6–8). The receptor-binding domain of PvDBP maps to the conserved, N-terminal cysteine-rich region II (PvDBPII) (9, 10). Given the complete dependence of P. vivax on the PvDBPII–Duffy interaction for blood-stage infection and recent observations that PvDBPII-specific antibodies inhibit P. vivax invasion of human erythrocytes in vitro (11), we examined the hypothesis that naturally acquired PvDBPII-specific binding inhibitory antibodies (BIAbs) are associated with protection against P. vivax infection in vivo.

To examine this hypothesis, we conducted a prospective cohort treatment reinfection study in 5- to 14-year-old children residing in Mugil, Papua New Guinea (PNG), which is hyperendemic for P. vivax and P. falciparum. Plasma samples were tested for recognition of PvDBPII by ELISA. In addition, plasma samples were tested for the presence of BIAbs by evaluating inhibition of binding between PvDBPII and its receptor, the N-terminal extracellular region of DA (12). We also examined whether BIAbs are strain-transcending, and whether their presence is associated with a reduced risk of infection with P. vivax.

Results

Malaria Infection in the Population.

At enrollment and before treatment, all four human malaria species were detected in the study children (n = 206), with 33.9% and 9.8% infected with P. vivax by PCR and light microscopy (LM), respectively, at baseline, as described in detail (13). All children in the study were treated with a 7-day course of artesunate at enrollment to clear blood-stage malaria parasite infections. Blood samples were collected every 2 weeks after enrollment and tested for the presence of P. vivax and P. falciparum infections by LM, a post-PCR ligase detection reaction-fluorescent microsphere assay (LDR-FMA) for all four human Plasmodium species, and real-time quantitative PCR (RTQ-PCR) for P. vivax. Although 81–100% of all children at risk were seen at each followup time point, 28 (13.6%) children were lost during followup because of withdrawal, or because they were missing two or more consecutive bleeds. A final blood sample was obtained from 159 children 1 year after enrollment.

The median time to P. vivax reinfection after the initial artesunate treatment as detected was 54 days by LDR-FMA and 119 days by LM; for P. falciparum, median time to reinfection occurred at 55 days by LDR-FMA and 99 days by LM. The incidence rate for P. vivax during the followup period was 2.0 infections per person per year by LM (1.7–2.5; 95% C.I.) and 5.3 infections per person per year by LDR-FMA (4.5–6.1; 95% C.I.). The incidence rate for P. falciparum was 3.2 infections per person per year by LM (2.7–3.7; 95% C.I.) and 5.0 infections per person per year by LDR-FMA (4.3–5.8, 95% C.I.). Thus, the reinfection and incidence data suggest that transmission rates for P. falciparum and P. vivax were similar.

P. vivax and P. falciparum infections detected in the first 6 weeks posttreatment by LDR-FMA were genotyped for the highly polymorphic PvDBPII alleles and P. falciparum merozoite surface protein two alleles, respectively, to identify treatment failures (13). Only one P. vivax infection detected in the first 6 weeks posttreatment by LDR-FMA had the same P. vivax PvDBPII genotype as observed at baseline, suggesting a possible treatment failure. In the case of P. falciparum infections detected in the first 6 weeks, 12 individuals had evidence of the same PfMSP2 genotype at 6 weeks after treatment. Thus, efficacy of artesunate for eradication of P. vivax and P. falciparum was 98.6% and 91.4%, respectively. All children with treatment failures were excluded from further analysis of homologous parasite species. The haplotypes of the PvDBPII alleles were identified in children who became infected with P. vivax during the first 6 months (14) [supporting information (SI) Table S1]. The AH variant was the most common (26% prevalence) and SalI (0.7% prevalence).

PvDBPII-Specific BIAbs.

Plasma from the 206 children at enrollment were tested by ELISA for recognition of the PvDBPII SalI and AH variants. The PvDBPII SalI variant is being developed as a vaccine candidate, and AH is the most common variant identified in the study population (14). Plasma were also tested for their ability to inhibit binding of PvDBPII variants to DA using an ELISA-based assay where plate wells were coated with the N-terminal 66-aa residues of DA expressed as a fusion with Fc region of human Ig (nDA-Ig, Table S2). Eighteen children (8.7%) had plasma that showed high-level binding inhibitory activity of >90% inhibition at a dilution of 1:5 (Table 1 and Table S2). Interestingly, the plasma samples with high-level BIAbs blocked binding of both SalI and AH PvDBPII variants to nDA-Ig at similar levels (Table S2). Plasma samples from 40 children (19.4%) showed intermediate-level (50–89%) of inhibition. The remaining children (148/206, 71.8%) had <50% binding inhibition; 97 of these children (97/206, 47.1%) did not have any detectable binding inhibitory activity (<20%). We then compared the median reactivity of plasma samples with PvDBPII as detected by ELISA with binding inhibitory activity against PvDBPII as measured by the nDA-Ig based binding assay (Fig. S1). Children with high-level BIAbs were all positive for recognition of PvDBPII by ELISA. Children with intermediate- and low-level BIAbs had significantly lower overall recognition of PvDBPII. There was a weak but significant correlation between BIAbs and overall reactivity with PvDBPII by ELISA (r² = 0.22, P < 0.001, linear regression analysis). The median age of children with high-level BIAbs was marginally higher than the other groups but not significant statistically (P > 0.1, Table 1). The average number of followup observations was equivalent among the three groups (Table 1).

Table 1.

Study population

Binding inhibitory antibodies*	n	Median age, yrs (range)	Females(%)	Total no. of observations(average no. of followups per person)	Average no. of observations per person(range)
High >90%	18	9.7 (7–14)	11 (61)	217 (12.1)	12 (11–13)
Intermediate 50–90%	40	9.5 (6–13)	21 (53)	458 (11.5)	12 (9–13)
Low <50%	148	9.2 (5–14)	75 (51)	1,728 (11.7)	11 (8–13)

*Normalized values indicating the percent inhibition by plasma samples (at 1:5 dilution) of PvDBPII binding to its receptor, the Duffy antigen, compared with PvDBPII binding in the absence of plasma. Details of the assay described are in Materials and Methods.

Presence of PvDBPII-BIAbs Is Associated with Protection Against P. vivax Infection.

We performed a survival analysis to determine whether the presence of high BIAbs was associated with protection against P. vivax infection. As shown in Fig. 1A, there was a significant delay in time to first P. vivax infection as determined by LM (median 96 days) among individuals with high BIAbs compared with those with low BIAbs (median 61 days, P = 0.02, log-rank test). The presence of high BIAbs had no effect on the time to first P. falciparum infection (Fig. 1B) as determined by LM compared with individuals with low or no BIAbs. When reinfection was diagnosed by the more sensitive LDR-FMA strategy, we observed no significant delay in time to first P. vivax or P. falciparum infection (Fig. 1 C and D). Children with high BIAbs had a 55% reduction in risk of P. vivax infection (Cox's proportional hazards ratio of 0.45, P = 0.04), and those with intermediate BIAb had a 25% reduced risk as assessed by LM, although the effects of the intermediate BIAbs were not statistically significant (Table 2).

Fig. 1.

Association of anti-PvDBPII BIAbs with reinfection by P. vivax and P. falciparum. Children with high PvDBPII BIAbs (n = 18, dashed line) vs. those with low BIAbs (n = 148, solid line) show a delay in time to first blood-stage P. vivax infection detected by LM of Giemsa-stained blood smears (A, P = 0.02, Log-rank test) after parasite clearance with artesunate. The presence of anti-PvDBPII high BIAbs did not affect the time to reinfection with P. falciparum as measured by LM of Giemsa-stained blood smears (B). Binding BIAbs did not affect the time to reacquisition of blood-stage infection for either P. vivax or P. falciparum as measured by PCR-based LDR-FM assay (C and D).

Table 2.

Association of anti-PvDBPII-blocking antibodies with protection against P. vivax infection

Binding inhibitory antibodies	Hazard ratio ± SE(95% C.I.)	Cumulative frequency ofP. vivax infection†		Cumulative frequency ofP. falciparum infection†		Number (proportion) of P. vivax infections with more than one strain‡
		LM	LDR-FM	LM	LDR-FM
High (>90%)	0.45 ± 0.18* (0.20, 0.98)	11 (5.1%)***	49 (23.1%)***	79 (36.4%)	110 (51.9%)	20 (40.8%)***
Intermediate (50–90%)	0.76 ± 0.19 (0.45, 1.26)	60 (13.1%)	161 (36.0%)	189 (41.3%)	235 (52.6%)	98 (60.9%)
Low (<50%)	NA	236 (13.7%)	563 (33.5%)	755 (34.9%)	811 (48.2%)	336 (59.8%)

NA, not applicable.

*, P < 0.05 Cox's proportional hazards analysis.

***, P < 0.001 χ² analysis compared to intermediate or low binding inhibitory Ab group.

^†Denominator is total number of observations for each group shown in Table 1.

^‡Denominator is number of individuals LDR-FM-positive for P. vivax (column 4).

We also examined whether high reactivity with PvDBPII as determined by ELISA was associated with protection against P. vivax infection. Children were divided into terciles based on reactivity (high, medium, and low) with PvDBPII. There was no difference in median time to reinfection between individuals with high (54 days), intermediate (49 days), or low (median 57 days, P = 0.57) reactivity. The presence of P. vivax infection at baseline was included as a covariate in the analysis and had no effect on the risk for subsequent P. vivax infection or acquisition of BIAbs. Children that acquired clinical disease due to non-vivax malaria and were treated or were lost in followup before becoming infected with P. vivax were censored from the analysis. Because acquisition of Pv infection by PCR occurred earlier after treatment, 32 (16%) children were censored in the analysis. For detection of first Pv infection by LM, n = 70 (44%) were censored.

In a second model that adjusts for repeated measures, the P. vivax and P. falciparum parasite densities were compared among individuals with high, intermediate, and low BIAbs. Here, individuals were removed from analysis for 6 weeks after their initial therapy and for 4 weeks after treatment for clinical malaria during the followup period. Children with high BIAbs had a 2.7- and 1.5-fold reduction in the frequency of P. vivax infection compared with children with low BIAbs as determined by LM and LDA-FMA, respectively (Table 2, P < 0.001). Moreover, children with high BIAbs had a 1.7-fold reduction in parasitemia compared with children with low BIAbs as determined by LM (Fig. 2A; P < 0.0001). The frequency and parasite densities for P. vivax infection were equivalent for children with intermediate and low BIAb levels. There was also a reduction in parasite density for children with high BIAbs compared with intermediate or low BIAbs by quantitative RT-PCR (Fig. 2 A; P < 0.05). There was no difference in frequency and parasite density for P. falciparum malaria between children with high BIAbs and those with intermediate or low BIAbs against PvDBPII (Table 2 and Fig. 2), demonstrating the specific protective effect of PvDBPII-specific BIAbs against P. vivax infection. Children with high BIAbs were more likely to be infected with a single P. vivax clone (based on PvDBPII genotyping) compared with individuals with intermediate or low BIAb levels who were more likely to have multiple P. vivax clones (Table 2; P < 0.001).

Fig. 2.

The association of anti-PvDBPII BIAbs antibodies with lower P. vivax density. A shows that the presence of high anti-PvDBPII BIAbs before treatment (n = 18) correlates with lower P. vivax density compared with individuals with intermediate (n = 40) or low (n = 148) BIAb levels as estimated by LM (black bars) or RTQ-PCR (stippled bars). There was no association of high anti-PvDBPII BIAbs levels with lower P. falciparum parasite densities. Parasite densities were determined from an average of 12 observations per individuals as described in Table 1 over the 6-month followup period after treatment. B shows that the presence of high overall PvDBPII reactivity measured by ELISA was also associated with a reduction in parasitemia for P. vivax and not P. falciparum. Measurement of parasitemia and the statistical methods for testing their differences between groups are described in Materials and Methods.

The same model was also used to test the relationship between reactivity against PvDBPII measured by ELISA and susceptibility to Plasmodium infection. There was a trend for a reduction in parasite density for P. vivax among children in the highest tercile for ELISA reactivity with PvDBPII compared with children in the intermediate or lowest terciles as estimated by LM (Fig. 2B). There was a statistically significant decrease in P. vivax parasite densities as estimated by quantitative RT-PCR in children in the highest tercile for reactivity against PvDBPII (Fig. 2 B; P = 0.04). There was no difference in the geometric mean LM parasitemia among the terciles for P. falciparum infection (Fig. 2B). There was also no association of high overall PvDBPII-specific ELISA reactivity and the proportion of individuals infected with a single P. vivax clone (highest tercile 31.2% with a single clone, middle tercile 22.5% with a single clone, and lowest tercile 25.6% with a single clone, P = 0.19).

High Levels of PvDBPII-Specific BIAbs Are Strain-Transcending.

Because PvDBPII is highly polymorphic (15), it is important to determine whether PvDBPII-specific BIAbs recognize PvDBPII variants and whether inhibition of receptor-binding is strain-specific. To examine this possibility, plasma with PvDBPII-specific BIAbs were tested for inhibition of binding of six different PvDBPII variants. These variants included the three most common haplotypes identified in the population (designated as AH, O, and P) and three other less-common variants in the population, including SalI (Table S1) in DA-binding assays. High-level BIAbs were strain-transcending in contrast to intermediate and low BIAbs (P = 0.002). Twelve of 18 children with high BIAbs inhibited binding of all six PvDBPII variants by >90%, and all 18 with high BIAbs inhibited binding of all six variants by >70% at a plasma dilution of 1:5 (Fig. 3A). Child no. 5 (Fig. 3B) typifies the strain-transcending binding inhibitory activity when plasma was serially diluted from 1:5 to 1:20. Three children with high BIAbs showed evidence for strain-specific inhibition after serial dilution of their plasma as shown for child no. 3 (Fig. 3C). For those children with intermediate BIAbs (n = 40), 14 showed strain-transcending binding inhibitory activity. More typical, however, was the strain-specific inhibitory activity among the majority of intermediate blocking individuals (n = 26), as illustrated by results for child no. 418 (Fig. 3D). The highest-level BIAbs were usually observed against the AH strain.

Fig. 3.

Anti-PvDBPII-specific binding BIAb levels to different variants of PvDBPII. A shows inhibition of receptor binding by six variants of PvDBPII at a 1:5 plasma dilution for the 18 individuals with high BIAbs. The other images show variant-specific binding inhibitory activities at different plasma dilutions for an individual with high (B) and intermediate BIAb levels (C and D).

Because the AH strain was the most-abundant strain, we examined whether AH-specific binding BIAbs (>50%, e.g., both high and intermediate) offered protection against reinfection with the homologous AH strain compared with the five other strains. There was 32% reduction in infection with homologous compared with heterologous strains [hazard ratio of 0.68, C.I. (0.31, 1.09), P = 0.07, n = 21], although this difference did not reach statistical significance because of the small sample size.

Levels of PvDBPII-Specific BIAbs Remain Stable During the Followup Period.

Repeated peripheral venous blood draws were obtained from 25 children, because they presented with one or more cases of clinical malaria. We examined plasma from children with high or intermediate PvDBPII-specific BIAbs at the beginning of the study and from children with three or more peripheral blood samples collected during the study along with repeated samples from some children with low BIAbs (<50%). The serial levels of BIAbs to the PvDBPII AH variant are shown in Fig. 4A for 15 children. Data for the remaining 10 children are not shown, because they had low BIAbs (<50%), and their pattern of responses was essentially the same as those shown in Fig. 4. BIAbs were remarkably stable over the followup observation time, irrespective of whether they were high, intermediate, or low. This pattern did not change appreciably if plasma samples were further diluted as represented by subject no. 5 (Fig. 4B). There was no association between the frequency and level of P. vivax (or P. falciparum) infections with subsequent BIAbs during the course of the study.

Fig. 4.

Repeated measurements on inhibition of PvDBPII-receptor binding by endemic sera during the study. (A) Serial measurements of PvDBPII-specific BIAbs for 15 children for whom three or more plasma samples were collected over 1 year. Time 0 represents a sample collected before artesunate treatment and time 11–12 months are collected on asymptomatic individuals 1 year after treatment. All intervening samples were collected at the time of an acute malaria episode (>90% due to P. falciparum). (B) BIAb levels from one subject (no. 5) at different plasma dilutions collected at four time intervals over the course of the study. The solid, shaded, and hatched bars represent binding inhibition at plasma dilutions of 1:5, 1:10, and 1:20, respectively.

Discussion

This study examined the role of antibodies directed against the receptor-binding domain of PvDBP in protection against P. vivax blood-stage infection. The study shows that ≈9% of PNG children aged 5–14 years residing in a P. vivax hyperendemic area had acquired high-level BIAbs directed against PvDBPII as a result of natural exposure to this parasite. The presence of high-level PvDBPII-specific BIAbs is associated with protection against P. vivax infection. In contrast, total antibody reactivity against PvDBPII as measured by ELISA demonstrated a weaker association with protection against P. vivax infections. This conclusion was supported by several independent analyses that demonstrate statistically significant association between presence of high-level BIAbs with a delay in time to reinfection after clearance of blood-stage infection, a reduction in the prevalence and intensity of P. vivax parasitemia, and a decrease in infection complexity as measured by the frequency of infection with multiple P. vivax strains. Importantly, the protective effect was species-specific, because children with high levels of BIAbs were protected only against P. vivax and not P. falciparum infection, which had similar transmission intensity in the population (15). Whereas amino acid sequence variation may compromise the utility of polymorphic antigens like PvDBPII to serve as reliable vaccine candidates (16, 17). PvDBPII-specific binding inhibitory activity of plasma samples from children with high-level BIAbs was strain-transcending. This suggests that a vaccine based on PvDBPII may be effective against diverse P. vivax strains.

We recently reported that naturally acquired PvDBPII-specific antibodies inhibited P. vivax invasion in vitro (11). These results suggest that naturally acquired PvDBPII-specific antibodies inhibit ligand-receptor engagement and block erythrocyte invasion by P. vivax. Together with the results presented here characterizing the role of PvDBPII BIAbs in protection against P. vivax infection in vivo, our overall findings have identified an important immune correlate for protective immunity against P. vivax in endemic populations. The functional nDA-Ig-binding assay provides an important tool for correlating PvDBPII-based vaccine immunogenicity with efficacy in vaccine trials, because it is quantitative, rapidly performed, and avoids complex adherence or parasite invasion assays (11, 12). These conclusions highlight the potential of PvDBPII as an important vaccine candidate against P. vivax malaria.

PvDBP is localized in the micronemes (internal apical merozoite organelles) and released to the merozoite apical surface only at the time of erythrocyte invasion (18, 19). The observation that naturally acquired high-level PvDBPII-specific BIAbs are uncommon may suggest that PvDBPII is not under immune surveillance and may not serve as an effective vaccine candidate. Alternatively, the limited exposure of PvDBPII during erythrocyte invasion may be a mechanism by which the parasite avoids host immune responses to the critical DA-binding motif within PvDBPII. PvDBPII-specific BIAbs may be produced only when sufficiently high parasitemia develops to stimulate an immune response, usually in young children with little immunity. Theoretically, immunization with a recombinant PvDBPII-based vaccine should elicit BIAbs against the conserved DA recognition site within PvDBPII that provide strain-transcending protection against P. vivax infection.

We observed that 9% of children in the 5- to 14-year age group had acquired high-level BIAbs through natural exposure to P. vivax that protect against infection and possibly disease; none of the five clinical cases of P. vivax malaria had high binding BIAbs. Overall, however, children in this age group from this study area were protected against high levels of P. vivax parasitemia and clinical vivax malaria (13). This suggests that by the time children have reached school age, most have acquired protection against P. vivax disease by immune mechanisms directed against an array of blood-stage antigens, including PvDBPII. Studies have yet to be performed to test the association of high-level PvDBPII-specific BIAbs with reduced risk of clinical P. vivax malaria in younger children.

Intermediate PvDBPII-specific BIAbs ranging from 50% to 90%, which were found in ≈25% of children in the current study, showed little or no protection against P. vivax infection. Our results showed that most of these children had not acquired strain-transcending BIAbs. Instead, these children had evidence for BIAbs that were strain-specific (Fig. 4) and conferred protection against parasites with homologous AH haplotype, the most common haplotype in the parasite population, compared with parasites with other haplotypes. This observation did not reach statistical significance (P = 0.07), because the study was underpowered to perform this subanalysis.

Once children acquire BIAbs, their levels remain remarkably stable, at least over the 1-year period of observation in the current study and up to 4 years among adolescents and adults (unpublished observations). We have demonstrated that affinity-purified PvDBPII-specific antibodies derived from pooled plasma from individuals with high BIAbs blocked binding to nDA-Ig and invasion of reticulocytes by P. vivax in vitro (11), ruling out the possibility that a stable plasma factor other than antibodies may be responsible for the blocking activity observed. This suggests that, once lymphocyte priming directed against PvDBPII occurs, even low levels of parasitemia can sustain BIAb responses to this protein.

There was an association between presence of a high level of PvDBPII-specific BIAbs and delay in time to reinfection by P. vivax, as measured by LM but not by the more sensitive PCR-based assay. This is not surprising, because PCR will detect more infections with low-level parasitemia that are less likely to develop clinical disease. It also suggests that PvDBPII-specific BIAbs reduce parasitemia by preventing P. vivax invasion into erythrocytes but does not completely prevent infections that remain detectable by the much more sensitive PCR methods.

In conclusion, the present study demonstrates that the essential parasite invasion ligand, PvDBPII, plays a critical role in naturally acquired immunity against P. vivax. These observations point to PvDBPII as an excellent vaccine candidate. Challenges for further studies include determining when high-level BIAbs develop in young children and what levels are sufficient to prevent clinical P. vivax malaria. Further, it is essential to learn how to formulate a PvDBPII-based vaccine that will drive antibody production to levels sufficient to be of functional relevance in blocking P. vivax invasion of erythrocytes, thereby preventing disease.

Methods and Methods

Detailed methods and procedures are provided in SI Text.

Prospective Cohort Study.

The study population of 206 children (ages 5–14 years) for this treatment reinfection study has been described (13). This study was reviewed and approved by the institutional review boards of the Papua New Guinea Medical Research Advisory Council, the Veteran's Affairs Medical Center (Cleveland), and the Walter and Elisa Hall Institute (Melbourne, Australia); written informed consent was obtained from the parents/guardians of each child, as well as the assent of the children. All children had an initial venous blood draw and were subsequently treated with a 7-day course of artesunate monotherapy (4 mg/kg on day 1 and 2 mg/kg on days 2–7). After treatment, children were monitored for malaria through biweekly active followup visits at school or at home for a total of 25 weeks beginning in June 2004 for a total of 13 followup visits for the presence of malaria infection. If a child developed clinical malaria symptoms, a peripheral venous blood sample was taken, and they were treated according to the PNG national treatment guidelines. Clinical vivax malaria was defined as temperature >37.5°C or a history of fever within the past 48 h, >1,000 parasites/μl by blood smear, and no other species of malaria parasite as determined by blood smear or PCR.

Detection of Plasmodium Species Infection.

Parasitemia levels were determined by LM and by semiquantitative post-PCR LDR-FMA that detects all four human malaria species (13, 20). The parasite density of P. vivax was determined by real-time quantitative PCR (14, 21), and P. vivax-positive samples were genotyped for the highly polymorphic PvDBPII alleles (22).

Measurement of PvDBPII-BIAbs and Total Anti-PvDBPII by ELISA.

To test plasma for their ability to inhibit binding of PvDBPII variants to DA, a construct encoding the N-terminal 60 codons of the human DA was ligated to the sequence encoding the Fc region of human IgG (nDA-Ig). The construct was cloned into the mammalian expression vector pCDM8 (12, 23). Sulfonation of tyrosine at amino acid 41 of the human Duffy antigen is essential for interaction with PvDBPII (12). After cotransfection of mammalian cells with plasmids encoding nDA-Ig and human tyrosyl protein sulfotransferase-2, nDA-Ig was expressed and purified as described (12). Recombinant nDA-Ig (1 μg/ml) was added to Immulon 4 plates followed by addition of recombinant PvDBPII protein corresponding to the different PvDBPII haplotypes (Table S1 at 0.1 μg/ml) to allow binding to nDA-Fc. PvDBPII (0.1 μg/ml) was preincubated with test plasma samples at dilutions of 1:5 and 1:10 for 1 h before addition to nDA-Ig coated plates. If binding inhibition was >80% at 1:10, the plasma was further diluted to 1:20, 1:40, and 1:80. PvDBPII (0.1 μg/ml) without plasma was used as a positive control for binding to nDA-Ig. PvDBPII with or without sera was allowed to bind to nDA-Ig for 1 h at 37°C. Bound PvDBPII was detected with anti-DBPII polyclonal rabbit sera (1:8,000), followed by mouse anti-rabbit IgG conjugated to alkaline phosphatase (1:5,000, Jackson ImmunoResearch). P-nitrophenyl phosphate was used a substrate for alkaline phosphatase and OD (A₄₂₂ nm) measured. Mean OD values of binding in the presence of plasma were normalized to %OD as compared with OD values of PvDBPII in the absence of sera. A standard curve of PvDBPII (0.003 μg/ml to 0.1 μg/ml) binding to nDA-Ig was used to estimate bound PvDBPII in the presence of plasma using %OD values. Plasma from P. vivax-unexposed individuals consistently resulted in <15% inhibition at 1:5 dilution (n = 11, data not shown).

Statistical Analysis.

Time to first infection for P. vivax was calculated as the time between the date of first treatment at enrollment and detection of P. vivax infection by LM or LDR-FMA either through active or passive case detection during 25 weeks of active surveillance. Time-to-infection data were analyzed by using standard survival analysis techniques. A log-rank test was used to evaluate the difference in nonparametric survival curves, and Cox regression was used to test for univariate and multivariate risk factors using STATA 8 statistical analysis software (Stata Corporation). For parasitemia outcomes that incorporated repeated measures on the same individual, a linear mixed model was used in SAS 9.1 (PROC MIXED). The analysis was based on a model that applied maximum-likelihood estimation techniques to the likelihood function associated with a set of error contrasts rather than to the original observations as described (24).