Naturally-acquired immune responses to Plasmodium
spp. infection target a variety of pre-erythrocytic and blood stage antigens of the parasite[1
]. When an endemic population demonstrates a degree of clinical or parasitological resistance, identifying an association between immunological recognition of a given antigen and resistance to malaria may indicate the antigen's potential value as a malaria vaccine candidate. Defining background responses is also useful for planning vaccine trials in endemic areas, due to the need to distinguish vaccine-induced responses from the baseline of naturally acquired responses once the vaccine is administered. Additionally, knowing this baseline and comparing immune responses post immunization with the responses obtained in malaria-naïves helps to assess whether naturally-acquired responses can be boosted by the candidate malaria vaccine.
Many studies have used immunofluorescence antibody assays (IFA) and enzyme linked immunosorbent assays (ELISA) to measure naturally acquired anti-malaria antibodies. In some cases, immunological studies have demonstrated an association between positive anti-merozoite[2
] or anti-pre-erythrocytic[3
] antibodies and incidence of malaria infection. IFA positivity has generally been defined by titres equal to or higher than the dilution of control sera not giving a positive immunofluorescence with the test antigen[4
], while ELISA positivity has generally been defined as titres greater than the mean + 3 SD of the negative control samples ("classical approach"). It has been pointed out that there are problems associated with the classical approach when negative and positive samples are not well separated and the background levels of controls are variable; in this case, a latent class model may better estimate the proportion of positive samples [6
]. However, it seems best suited to estimating the proportion of positive samples rather than identifying each sample as positive or negative, which is the objective of this study, and so was not used. Because the "the classical method" yielded many positive samples due to the small SD of negative controls, a second, more stringent method was also applied, in which a sample was deemed positive if it met criteria for the first method and simultaneously showed an optical density (OD) ≥ 0.5 at a dilution of >1/100.
Defining baseline, naturally acquired T cell immunity is more challenging, since activities of T cells measured using different assays in malaria endemic areas are low, vary over time[7
] and are HLA-restricted[9
]. Earlier studies, using proliferative or cytotoxic T cell responses to measure T-cell immunity, identified various epitopes within CSP [10
], MSP1 [14
], LSA1 SSP2/TRAP CSP [20
] and AMA1[23
] that induced recall responses in residents of endemic areas. In general, these T cell responses were short lived and did not clearly correlate with natural-transmission induced immunity defined as resistance to clinical malaria. Moreover, patent infections with P. falciparum
appeared to suppress T cell responses[25
]. Subsequently, ex vivo
ELISpot assays were shown to be more sensitive than CTL assays [26
]. However, ELISpot responses were also of low magnitude [9
] and relatively unstable over time [7
]. In studies in The Gambia and Kenya [28
], positive ELISpot activities ranged from 28-34 sfc/m to SSP2/TRAP, 32-60% of volunteers responded, and responses differed when measured one year apart [9
]. With MSP-1, responses differed when measured three weeks apart[8
]. Such studies relied on a single assay per sample, albeit often using replicate wells; so far there have been no studies reporting whether such single assays reproducibly represent T cell immunity and thus whether differences measured at different time points might reflect intrinsic variation in the assay as much as changing T cell function.
ELISpot assays are generally defined as positive when a statistically significant difference is found between test samples and medium controls using the Student t
] or chi-square comparison[32
], but often mean sfc/m induced by an antigen have been low and within the range of medium-only controls[29
To assess baseline immune responses as well as their reproducibility, in preparation for vaccine trials in endemic areas of West Africa, healthy adult subjects were recruited from two sites in southern Ghana, the rural community of Mampong about 35 kilometres northeast of Accra, and the urban community of Legon, a northern suburb of Accra and site of the University of Ghana. In Part A of this study, IFA was used to measure antibodies to sporozoites and blood stage parasites, and ELISA was used to measure antibodies to the candidate recombinant protein vaccine antigens CSP, SSP2/TRAP, EXP1, LSA1, MSP1 and MSP3[10
]. Antibody assays were defined as positive using the two approaches described above. Ex vivo
interferon-g (IFN-γ) ELISpot assays were also performed, using fresh PBMC stimulated with HLA-matched peptides from the candidate antigens CSP, SSP2/TRAP, EXP1, LSA1 and LSA3. In the ELISpot assays, positive activity was determined by requiring a statistically significant difference between test sample (done in quadruplicates) and medium control using Student's t
test, plus a 2-fold greater value than medium controls and at least a 10 sfc/m difference between test sample and medium controls.
Part B of the study was conducted two and a half years later using a different set of 12 volunteers, to assess the intrinsic reproducibility of the ELISpot assay. Cell samples were drawn from each study volunteer approximately two weeks apart and each sample was assayed three times on three different days to measure assay variability. Reproducibility of the assay was defined as the proportion of assays yielding either all positive or all negative results in the three replicate assays from a single sample time point while reproducibility over time was defined by comparing results from two time points.
Peptide pools spanning the full length of the test antigens CSP and AMA1 were used for stimulation in Part B rather than HLA-matched peptides, as a more straightforward approach that avoided having to perform HLA typing for each volunteer. Three statistical methods using different cut-off criteria of varying stringency were used to define positive responses to see which provided optimal results; (1) Least stringent: a greater than 20 sfc/m difference between peptide stimulated samples and medium controls; (2) Medium stringent (same as Part A): a significant difference on Student's t
test comparing test samples and medium controls, at least two-fold difference comparing test samples and medium controls, and at least a 10 sfc/m difference between test samples and medium controls; and (3) Most stringent: based on criteria established for chronically HIV-infected subjects where a positive response was defined as a minimum of 55 sfc/m and a 4-fold difference over the medium controls [35
]. Volunteers who developed positive ELISpot activities to one or more peptide pools were designated as positive responders.