The protocol for this trial and supporting CONSORT checklist are available as supporting information; see Checklist S1
and Protocol S1
The study was conducted in Donéguébougou, Mali, at a clinic operated by the Malaria Research and Training Center of the University of Bamako. Donéguébougou is a rural village of approximately 1300 inhabitants in which malaria transmission occurs mainly during the rainy season extending from June to November. Entomologic inoculation rates (EIR) as determined by the human landing catch method vary between zero in the dry season and 50 to 60 infective bites/person/month at the height of the rainy season in September and October 
The study was approved by the institutional review boards of the University of Bamako and the US National Institute of Allergy and Infectious Diseases, and was conducted under an investigational new drug application (BB-10944) to the US Food and Drug Administration.
After obtaining permission from the village elders to conduct the study, 54 participants were enrolled into 1 of 3 cohorts. Written informed consent was obtained from healthy volunteers between the ages of 18 and 45 years. Volunteers were excluded if they had evidence of clinically significant systemic disease; were pregnant or breast feeding; had serological evidence of chronic hepatitis B or C infection; were receiving corticosteroids or immunosuppressive drugs; or had been immunized with a live vaccine within the previous month.
Recombinant AMA1-FVO and AMA1-3D7 were manufactured, mixed, and adsorbed to Alhydrogel (HCl Biosector) as described previously 
. Each 0.5 mL dose contained either 5, 20, or 80 µg AMA1-C1 and 800 µg Alhydrogel. Potency studies in mice conducted every 6 months confirmed that all lots were stable and potent throughout the trial. Recombivax HB (Merck&Co.) was supplied in single-dose vials containing 10 µg of recombinant hepatitis B surface antigen adsorbed to amorphous aluminum hydroxyphosphate sulfate at a final volume of 1.0 mL. Vaccines were transported to the study site using temperature monitoring devices to ensure maintenance of the cold chain.
Within each cohort, participants were randomized to receive either AMA1-C1/Alhydrogel (n
12) or Recombivax (n
6), with the first, second, and third cohorts enrolled successively at three-week intervals. Participants randomized to AMA1-C1 received 5, 20 and 80 µg in the first, second, and third cohorts, respectively. Vaccinations were administered by intramuscular injection in the deltoid muscle on study days 0, 28, and 360.
The safety of study participants was monitored throughout the trial by an independent Malian physician. Interim safety reports were reviewed by a data and safety monitoring board prior to vaccination of the second and third cohorts, and before administration of the third vaccinations on study day 360.
The primary objective was to estimate the frequency of vaccine-related adverse events, graded by severity, for each dose of AMA1-C1 being tested. Secondary objectives included evaluation of the allele-specific antibody response to vaccination, assessment and comparison of the duration of antibody response to AMA1-FVO and AMA1-3D7, measurement of the inhibition of parasite growth as measured by the in vitro GIA to the FVO and 3D7 clones of P. falciparum, and determination of the relationship between anti-AMA1 antibody concentration, as judged by ELISA, and degree of in vitro growth inhibition of P. falciparum by GIA.
Following vaccinations, volunteers were directly observed for 30 minutes and then evaluated 1, 2, 3, 7, and 14 days post-vaccination for evidence of local and systemic reactogenicity, and then monthly throughout the malaria transmission seasons following the second and third vaccinations until the end of the study approximately 6 months after the third vaccination. Injection sites were examined for erythema, swelling, and tenderness at the site of injection. Solicited systemic adverse events included fever or chills, headache, nausea, myalgia, and arthralgia. Adverse events were graded as either mild (easily tolerated), moderate (interfered with activities of daily living), or severe (prevented activities of daily living), and assigned causality relative to the study vaccine. Injection site erythema and swelling were graded as mild (>0 to ≤20 mm in diameter), moderate (>20 to ≤50 mm), or severe (>50 mm). Oral temperature was graded as mild (>37.5°C to≤38.0°C), moderate (>38.0°C to≤39.0°C), or severe (>39.0°C). A complete blood count, serum creatinine and alanine aminotransferase concentration were performed immediately prior to each vaccination as well on the third and fourteenth days following vaccination. Complete blood counts were also performed at the monthly visits during the malaria transmission seasons of each study year.
Anti-AMA1 antibodies were measured using a standardized ELISA 
. ELISA plates were coated with AMA1-FVO, AMA1-3D7, or AMA1-L32, a P. pastoris
-expressed recombinant protein based on the sequence of the L32 strain of P. falciparum
(Genbank accession number EF221749) and that differs from the FVO and 3D7 antigens by 26 and 24 amino acids, respectively. Serial dilutions of a standard serum pool were included on each test plate to generate a standard curve, which was used to convert the absorbance of individual sera into antibody units. Participant samples from days 0 to 180 and from days 270 to 540 were tested at separate times.
Antigen-specific IgG subclasses were measured by a flow cytometric suspension array assay. Serum samples were mixed with microspheres coupled to AMA1 (Luminex Corporation). Mouse anti-human IgG subclass antibodies (anti-human IgG1, IgG2, IgG3, and IgG4) and a secondary donkey anti-mouse IgG phycoerythrin-labeled antibody (Jackson ImmunoResearch) were added to develop the reactions. Mean fluorescence intensities were detected by Luminex X-MAP using Bioplex software (BioRad).
IgG from vaccinated individuals were tested for their ability to inhibit in vitro
growth of P. falciparum
3D7 and FVO parasites using a standardized GIA procedure 
. Values obtained with test samples were compared to those obtained from parasites incubated with a pool of malaria-naïve human serum and with uninfected red cells to obtain the percent inhibition in growth.
Although not powered to detect differences in the incidence of specific adverse events or immune responses between AMA1-C1 and Recombivax or between the different dose concentrations of AMA1-C1, a group size of 12 per dose concentration of AMA1-C1 was chosen to give a reasonable probability of detecting one or more serious or severe vaccine-related adverse events. A group size of 10 individuals would provide a power of 80% to detect an adverse event that occurs with a probability of 0.15; an extra 2 participants were recruited into each group in case of withdrawal or loss to follow-up.
In addition, each dose concentration of AMA1-C1 was compared to Recombivax, which served as a comparison for both immunologic and safety assessments. A total of 18 controls, 6 per each dose concentration of AMA1-C1, were enrolled to allow a 2
1 ratio of AMA1-C1 to Recombivax within each cohort; thus, each of the 3 cohorts contained 12 volunteers receiving AMA1-C1 plus 6 receiving Recombivax for a total sample size of 54.
Within each cohort, participants were randomized to receive either AMA1-C1 or Recombivax by use of sealed envelopes labeled with a unique participant study number and containing the vaccine assignment. A set of 18 envelopes were prepared for each cohort such that 12 contained assignments to AMA1-C1 and 6 to Recombivax. Study numbers were assigned in the order that participants arrived at the clinic on the day of first immunization. Randomization envelopes were opened by a study pharmacist on the day of first immunization and immediately re-sealed after vaccine assignment.
Study participants and investigators who assessed outcomes were blinded to vaccine assignment. Access to randomization codes was limited to the study pharmacists. Syringe barrels were masked with opaque tape to disguise the contents, since the volumes of the two study vaccines were different. To reduce investigator bias, injections were administered by physicians who were not involved in post-vaccination safety assessments or study analysis.
Differences in the proportion of individuals experiencing each adverse event (of any severity) between vaccine allocations (AMA1-C1 vs. Recombivax) and between vaccinations (first versus second, etc.) within each dose group were analyzed using Fisher's exact and McNemar's tests, respectively. The exact two-sided Cochrane-Armitage test for trend using rank scores was used to test for a dose effect for each solicited adverse event; for these analyses, the Recombivax group was assigned a dose of 0 µg of AMA1-C1.
Differences in the change in antibody level from baseline to post-vaccination time points were compared between AMA1-C1 and Recombivax using exact Wilcoxon rank sum tests within each cohort. Differences in change of antibody (days 0 and 42) between dose groups were analyzed using the two-sided Wilcoxon rank sum test, assuming the absence of an effect due to differing vaccination times for the cohorts in relation to the malaria transmission season (this assumption was not violated by a Kruskal-Wallis test for cohort effect on day 0 values [p
>0.4]). Agreements between anti-AMA1-FVO, anti-AMA1-3D7, and anti-AMA1-L32 antibody responses were calculated using the random marginal agreement coefficient (RMAC) with the squared difference cost 
. Correlations between percent growth inhibition and antibody responses were assessed by the Spearman rank test. The SAS (version 9.1; SAS), R (version 2.4.0; R Foundation for Statistical Computing) and STATA (version 8.0; StataCorp) software packages were used, and p
<.05 was considered significant.