The Malaria Immunology and Genetics in the Amazon (MIGIA) study takes place in a community called Zungarococha (there are 1907 individuals enrolled in our study from this community) with a maximum P. falciparum infection rate of 0.42 infections/person/year (details in Branch et al 12).

In 2003 and 2004 active case detection (ACD) was undertaken during the malaria transmission season (February – July) and passive case detection (PCD) was performed throughout the year. ACD was conducted in two different ways. In three community-wide surveys as many community members as possible were sampled using home-based visits: March 2003 (N = 363), February 2004 (N = 1156), and August 2004 (N = 861). Additionally, each month during malaria season (April-July 2003 and also March-July 2004), approximately 220 community-members were randomly selected from 100-meter regions surrounding homes where P. falciparum was detected during the prior month and were enrolled for one month of weekly home-based visits. There were 1,722 participants who completed at least four ACD visits in one malaria season: 671 in 2003 and 1051 in 2004, of which 423 completed at least four ACD visits both in 2003 and 2004.

Additionally, throughout 2003 and 2004 infections were detected via PCD at the local health post. All individuals entering the Zungarococha health post complaining of febrile illness or other malaria-like symptoms had a blood slide read by an expert microscopist. Malaria treatment is free and tightly controlled (observed therapy), and there is no access to private physicians.

At each ACD and PCD, a detailed clinical and epidemiologic evaluation was made by a trained physician and blood was collected (0.25 ml in ACD and 3–6 ml in PCD). Thin and thick blood-smears were made, haematocrit capillary tubes were filled, and blood was separated into packed erythrocytes and sera and stored at -70°C.

All individuals in the community were invited to participate in the ACD study protocol and clinic protocol if they were diagnosed with malaria either in ACD or PCD. Ethical clearance was received from the University of Alabama at Birmingham, the Universidad Peruana Cayetano Heredia (Lima, Peru), and the Ministry of Health (MINSA, Lima, Peru).

Thick blood smears were examined using Giemsa staining and oil-immersion microscopy by three highly qualified microscopists. If >50 ring-stage parasites were detected after reading 200 leukocytes, parasitaemia was calculated from these fields. Otherwise, at least 500 leukocytes were counted. Parasitaemia was expressed as the number of parasites/μL of blood, assuming 6,000 leukocytes/μL. The blood-smear was considered negative if no parasites were seen in 200 fields.

Whether in active or passive case detection, all treatments were given through the MINSA authorities, following the MINSA National Drug Policy Guidelines. Plasmodium vivax treatment is chloroquine (10 mg/kg for three days) with primaquine (0.5 mg/kg for seven days). Plasmodium falciparum treatment is mefloquine (12.5 mg/kg daily for two days) with artesunate (4 mg/kg daily for three days) in non-pregnant patients older than one year of age.

The recombinant P. falciparum MSP1-19kD, corresponding to Ugandan-PA P. falciparum strain (with the MSP1-19kD 4 aa variant positions of E-K-N-G)23, was expressed in Sacharomyces cerevisiae (provided by Dr. David Kaslow). The variants E-K-N-G and Q-K-N-G are the only P. falciparum MSP1-19kD detected in this population. One hundred and twenty eight randomly selected samples were tested (by ELISA). The anti-EKNG and anti-QKNG IgG responses (Negative/Low Positive/High Positive) were nearly identical (concordance = 92%).

An enzyme-linked-immunoabsorbent assay (ELISA) for total IgG was performed as described previously 7. Briefly, plates (Dynatech) were coated with 50 μL/well of recombinant MSP1-19kD (at a concentration of 0.5 ng/μL) and blocked with 1% bovine serum albumin (BSA). The sera were diluted 1:100 (1.5% non-fat milk in washing-solution [0.15 M Na₂HPO₄, 0.15 M NaH₂PO₄, NaCl, 0.05% Tween20, and 0.05% BSA]). The IgG that bound to the MSP1-19kD on the plate were detected with peroxidase-conjugated goat-antihuman-IgG (Chemicon) diluted 1:2000. The plates were washed, 50 μL/well of 3,3',5,5'-tetramethylbenzidine (KPL) was added, the reaction was stopped using 50 μL/well of 0.25 M HCL, and then read at absorbance 450 nm (A₄₅₀) with an ELISA plate-reader (BIORAD).

Ten serum samples from healthy non-exposed Peruvians were used as negative controls. The positive controls included five samples from five different P.falciparum-infected individuals and a "positive pool" made-up of three samples.

A positive pool control standard-curve was made for each experiment day. The positive pool was diluted at 1:50, 1:100, 1:400, 1:1600, 1:3200, and 1:12800. The results from each day showed similar curves. The average positive pool optical density (OD) at a 1:400 dilution for all experiment days was subtracted from the positive pool OD for each individual experiment day. The resulting correction factor was subtracted from the initial OD of each sample, thereby standardizing by experiment day.

Samples with an OD greater than the negative cutoff (the average of the negative controls plus 2 standard deviations) were considered positive. The positive group was divided into "Low" and "High," where an OD greater than 3 times the negative cutoff was called "High-Positive."

Of 1,722 individuals, 600 individuals were selected who participated in at least one ACD community survey: 300 children (1–14.5 years-old) and 300 adults (14.6–104 years-old). Of these 600 age-grouped and then randomly selected individuals, 448 participated in 2 of the 3 community surveys, and 49 participated in all three community surveys. There were a total of 868 sample-events in the three community surveys.

Seventy-nine P. falciparum infections (from 79 different individuals) were selected from the 182 infections detected in ACD because these represented individuals who had sera samples available from before, during, and after their infection. Forty-one of these were adult infection samples.

"Before", "During", and "After" infection time points were selected. From samples collected within 3 months before infection, the sample closest to 1 month before was selected as "Before": median days until infection (1^st–3^rd quartile) = 38 days (14–68). From samples collected within 3 months after infection, the sample closest to 1 month after was selected as "After": 22 days (14–53). The before, during, and after sera samples were encountered for 79 individual infections.

An infection was classified as "symptomatic" if the individual had a fever greater than or equal to 38.3°C, reported a fever in the past 2 days, or had a haematocrit of less than 30 PCV either at the time of or within one week after infection.

Statistical analysis was conducted using SAS (SAS Cary, NC, version 9). Chi square (χ²) analysis tested infection and seroprevalence differences. A non-parametric Wilcoxon Rank Sums (WRS) tested antibody level differences. A general linear model (GLM with General Estimating Equations (GEE) obtained by SAS procedure "GENMOD") determined variables associated with antibody level changes upon infection. The independent variables considered were age, logarithmic parasite density, sex and days separating the measurements of IgG. The GLM dependent variable was the IgG OD "During" minus the IgG OD "Before" time point. When testing frequency of fever with antibody response description over time (response profile), a Fisher's Exact and Mantel-Haenszel Chi-Square was used. Logistic regression analysis (also using GENMOD) determined if the above-listed variables could predict symptomatic infections, using risk estimates calculated by the GEE.

Infection prevalence (Figure 1) was calculated during the ACD community surveys (point prevalence) and also during the month of weekly ACD (monthly prevalence) for each individual. Each month, adults usually demonstrated a higher frequency of infection than children (Figure 1); however, this was not significantly different (χ², p = 0.0901). The study-period PCD data as well as the historic records and the entomologic data 1224 agree that malaria infections rarely occur within the dry season of August through December. In addition, during a recent small community survey that was conducted specifically to verify that there is negligible transmission during the dry season, 238 blood samples were collected in November through December, 2007. This survey included 123 febrile individuals as well as 115 non-febrile family members who simply accompanied the febrile individual to the community health center (sampling these individuals fell within the same ethically-reviewed protocol as described in this manuscript). Of these 238 individuals, all were microscopy negative for malaria and only one was PCR (sub-microscopy) positive for P. vivax. None were PCR positive for P. falciparum.

Returning to our 2003–2004 data, as expected due to the dry season beginning in July, the infection prevalence was near zero in the August 2004 community survey (Figure 1). Even though there was a similar infection prevalence in adults and children, significantly fewer children were IgG positive, regardless of the community survey epoch, versus adults (χ², p < 0.05 for each community survey, Figure 1). For example, in the 2003 malaria season community survey (March 2003), 34.8% of children and 76.8% of adults were IgG-positive (point seroprevalence).

Despite low transmission, the αMSP1-19kD seroprevalence in the beginning versus the end of the transmission season was similar (Figure 1). Although it is possible that some individuals had a malaria infection on or before the February 2004 survey, only a small proportion of the αMSP1-19kD positivity could be attributable to such a rare early-season malaria infection given the low transmission. Specifically, given the maximum P. falciparum transmission rate found in ACD and PCD, together, in the highest transmission zones of this community 12, the probability of infection during the 4 months between these two community surveys was only 0.06.

In both Plasmodium-negative adults and children, the frequency of having a High Positive IgG response varied by <5% between the community survey conducted at the beginning of the 2004 malaria season (February 2004) and that conducted at the end of the 2004 malaria season (August 2004) (χ², p < 0.05; Table 1).

Several Plasmodium-negative individuals in the February 2004 survey were infected with P. falciparum at some time during the following malaria season. One-hundred and thirty-five individuals were evaluated who were Plasmodium-negative in February 2004 and also who were in both the February 2004 and the August 2004 community surveys (e.g., represented twice in Table 2). Seven of the 29 individuals who were IgG negative in February 2004 were infected between March and July, 2004. In August 2004, five of these recently (but not currently) infected individuals were αMSP1-19kD IgG positive. The only two individuals who were IgG negative, despite this recent infection, were children.

Similarly, Plasmodium-negative adults, but not children, had high median αMSP1-19kD levels (1.040 versus 0.301; Table 2). The P. falciparum-infected adult median αMSP1-19kD IgG level was not significantly different than that seen in Plasmodium-negative adults, and the majority of Plasmodium-negative adults had a high αMSP1-19kD IgG level (3^rd quartile = 0.505). The median IgG level in P.falciparum-infected children (1.339) was also similar to that seen in infected adults (Table 2). Although most Plasmodium-negative children had low αMSP1-19kD IgG levels, some had a high αMSP1-19kD IgG level (3^rd quartile = 0.827).

The αMSP1-19kD IgG level in P. vivax-infected individuals (detected in community surveys) was similar to that in Plasmodium-infected children (Table 2). The low homology between the P. vivax and P. falciparum MSP1-19kD 25 supports this observation.

Sera samples taken 'During' as well as approximately one month 'Before' and one month 'After' infection were evaluated to determine the dynamics of the αMSP1-19kD IgG antibody response and to see how this response changes between age groups (Figure 2). A positive αMSP1-19kD IgG response in the Before sample was observed in 36%, 69%, 84% and 85% of the 0–6, 7–14, 15–30, and >30 years-olds, respectively. In the 0–6 year-old group, the median αMSP1-19kD IgG was only 0.216 Before, but 0.712 During and then 1.08 After infection. Only the 0–6 year-old group had an increase in median αMSP1-19kD level after infection. The Before IgG level clearly increased with age. In the older age groups, there was little difference in median IgG level across During and After time points.

The antibody level change upon detection of an infection (which was predicted to be early in the infection due to our ACD) was evaluated by subtracting the Before from the During infection IgG OD in each individual (Figure 2). A step-down multivariate model of During-minus-Before showed that each age-group of individuals >6 years-old had a significantly greater αMSP1-19kD level change upon infection versus the 0–6 year-olds (GEE, p < 0.0418). None of the other variables tested (see methods) were associated with this detection-associated change in αMSP1-19kD level (p > 0.1).

The αMSP1-19kD response profile was defined for each individual infection by the responses measured in the Before, During, and After infection samples as follows: