Plasmodium falciparum variant erythrocyte surface antigens: a pilot study of antibody acquisition in recurrent natural infections

In this study, samples from children presenting with mild malaria who were included in a Phase III, randomized, non-inferiority trial was used to assess the efficacy and safety of dihydroartemisinin-piperaquine (Artekin) in comparison with artemether–lumefantrine (Coartem©) in Pingilikani, Kenya, with a follow up period of 84 days. This study has been described in detail elsewhere [17]. Children were enrolled if presenting with uncomplicated malaria and fulfilling all criteria as described [17]. They were monitored for 3 days on-site during the administration of therapy and the mother/guardian was asked to return with the child for scheduled visits on days 7, 14, 21, 28, 35, 42 and 84 post-enrollment, or if any symptoms occurred. During this follow up period, samples (serum, blood samples for gDNA extraction and live P. falciparum isolates) were collected at any time the children were parasitaemic according to microscopic examination as described [17]. Serum was additionally collected on days 3, 7, 14, 28 and 42. All samples were anonymized. Serum was stored at − 80 °C. Full blood containing parasitized red blood cells was collected and mixed with cryosolution (28% glycerol, 3% (w/v) sorbitol, 0.64% saline) at a volumetric ratio of about 1:4 (packed cell volume:cryosolution) and stored in liquid nitrogen.

During the follow up period recurrent parasitaemias were screened for recrudescences using MSP2 genotyping. Initial screening was done using the RFLP method described by Felger et al. [18] and isolates from 9 patients were selected for the study described here if the patients harboured a recrudescence within the first 42 days of follow up and if the parasitaemia was > 0.1% (Tables 1 and 2). Flow cytometry data was obtained from 7 of these patients, and out of these, 6 patients could later be confirmed as harbouring recrudescence by a superior resolution capillary electrophoresis MSP2 genotyping method [19] using oligo N5 with JOE instead of VIC, on the ABI genetic analyzer 3130 (Additional file 1: Table S1). However, it cannot be excluded that this will include false recrudescences by infections caused by parasites containing the same MSP2 alleles.

Sera from these 7 patients, taken at day of recruitment and at follow-up visits (days 0, 3, 7, 14, 28, 42 and 84, and when patient visited clinic with parasitaemia during the follow-up period), were used to study antibody acquisition towards the respective infecting parasites in each of these patients over time. Whenever a serum is tested against an isolate from the same patient, they are referred to as homologous serum and isolate.

To test heterologous recognition of the isolates, i.e., testing serum derived from different patients than the parasite isolate tested, a panel of heterologous sera (Table 3) from 14 (blood group A) and 4 (blood group AB) patients were chosen. The panel of 4 sera (blood group AB) was used to test parasites from patient 171, the only patient with blood group AB (Table 2). All sera originated from the same study described above, from children who suffered from recurrent parasitaemias within the follow up period. Four of the sera in the panel originated from patients whose isolates were further characterized in this study (119, 150, 178 and 221), but when the serum-isolate pair (derived from the same patient) were tested this was recorded as homologous testing, not heterologous testing. From all 18 patients of the heterologous serum panel, sera from day 0 (acute) and 14 (convalescent) were tested.

Improved MSP2 analysis using capillary electrophoresis [19] was performed as described above on the isolates derived from the patients whose sera were included in the heterologous serum panel to differentiate between sera derived from patients who experienced recrudescences and reinfections during the follow up period.

Live isolates were removed from liquid nitrogen, washed with ice cold 3.5% NaCl, then washed with culture medium (RPMI 1640; Gibco), and transferred into culture flasks where the isolates were allowed to stabilize for 18–24 h under standard culture conditions [20] until trophozoite stage and were then analysed using flow cytometry. The procedure of freezing and thawing of the already small blood sample volumes originating from young children reduces the volume further mainly due to red blood cell lysis. For each test, 0.5 µl of the isolate pellet were incubated 30 min in 40 µl of PBS (0.1% BSA) with 0.1 mg/ml Ethidium Bromide (Sigma) and 1 µl serum. Negative controls were no serum and non-immune pooled blood group AB serum from non-exposed European adults, and a positive control, hyperimmune blood group AB adult serum from the Kilifi area also used in previous studies [14, 21] was included. After washing (with PBS, centrifuging at 900g for 5 min) cells were incubated in 30 µl of PBS (0.1% BSA) containing antihuman-CD45-PC7 (1:30 v/v) (IM3548; Beckman Coulter), antihuman IgG-FITC (1:1 v/v) (FA004; the Binding Site) and antihuman IgM-SPRD (1:30 v/v) (735986; Beckman Coulter). After washing three times (with PBS, centrifuging at 900g for 5 min) samples were resuspended in PBS (0.1% BSA) and analysed using an FC500 Flow Cytometer (Beckman Coulter).

FlowJo and Excel were used for the initial flow cytometry analysis, where first nucleus containing cells stained by ethidium bromide were included, i.e. leukocytes and infected red blood cells (irbcs) at the trophozoite stage. At this stage ring stage parasites were excluded by low ethidium bromide staining. Next CD45-containing leukocytes were excluded and finally the IgG recognition measured as mean fluorescence intensity (MFI).

The flow cytometer was set to count 1000 irbcs. Samples where less than 200 irbcs (due to small sample volumes and low parasitaemia) had been measured were excluded from analysis. When the volume of cells was smaller than expected (pipetting errors on semi-packed cell pellets), the last cells measured would give unusually many counts with high MFI values. To exclude these erroneous measurements, samples where the whole population of measured cells (mostly noninfected red blood cells) had more than 2.5% IgG positive cells, defined by a cut-off at MFI arbitrary units of 12, were excluded from analysis. Manual analysis confirmed that no samples with high parasitaemia were excluded by this process. All flow cytometry results are presented in this study as the MFI for each isolate-serum pair, with the individual “no serum” sample MFI subtracted. The “no serum” value for each parasite isolate was subtracted rather than the value of the negative control pool to allow comparison of the data with the negative control group. The MFI values for each measurement together with parasite isolate and serum information is presented in the Additional file 1: Table S1.

If the no serum sample was missing (excluded by reasons explained above) or differing by more than 90% from the median of all no serum samples, the whole series, i.e. all data from the corresponding isolate, was excluded.

To ensure that the measured fluorescence was not simply an effect of higher parasitaemia in the sample, i.e. an experimental artefact, linear regression analysis was performed for fluorescence vs. parasitaemia but no correlation was found (P = 0.27, not shown).

All flow cytometry results used for analysis is presented in Additional file 1: Table S1.

Graphpad Prism 6.0e for MAC OS X and Genstat^® 16th Edition were used for analysis and graphical presentation. A mixed-effects linear regression model using a REstricted Maximum Likelihood (REML) [22] approach was used in each analysis described below. In all analyses below, random effects for patient and serum were used to allow for correlation in reactivity of sera and isolated parasites originating from the same patient.

To test for intrinsic differences in the ability of the parasites to be recognized, the data were initially transformed by natural logarithm, log_e (MFI + 0.145) to make the smallest observation positive and to normalize the distribution. Fixed effects for response type (originating from a patient that responded to its parasite isolate vs. a patient that did not) and serum day (day 0 or 14), together with their interaction were then tested in the model.

The same analysis was performed to test for a difference between recognition of baseline vs. recurrent isolates. In this case the fixed effects were isolate type (baseline or recurrent) and serum day (day 0 or 14).

The ability of homologous sera taken before recurrence to recognize the recurrent parasites was also compared to that of their overall recognition by heterologous sera with a linear mixed effect regression model. In this analysis, only sera taken before the recurrent infection were used. The data were transformed by log_e (MFI + 0.062) to make the smallest observation positive. Fixed effects tested in the model were serum type (homologous recognition of recurrent isolates vs. recognition of the same recurrent isolates by heterologous sera) and serum day (day 0 or 14).

To compare the anti-VSA antibody repertoire of children who eliminated the baseline parasites to that of those children who did not, a similar linear mixed effect regression model was used. The data were transformed by log_e (MFI + 0.145) to make the smallest observation positive. Fixed effects for serum type (originating from a patient that eliminated its parasite infection vs. a patient that did not) and serum day (day 0 or 14) were fitted.

Histograms or residuals, fitted-value plots and normal plots were performed to test the suitability of the transformed data in the linear mixed effect regression models.

For this study, samples from children presenting with mild malaria who were included in a dihydroartemisinin-piperaquine vs. artemether–lumefantrine drug trial were used [17]. In total 14 parasite isolates from 7 of these patients were analysed for antibody recognition (Tables 1, 2). All the children involved are within the age bracket where Plasmodium infections transit from causing disease to being asymptomatic. Table 2 also shows whether the recurrent infections were asymptomatic (tympanic temperature < 37.5) at the time of detection.

Sera derived from children in the same study were used to test antibody recognition of the isolates described above. Homologous sera (sera derived from the same patient as the parasite isolate tested) were used to test acquisition of antibodies to the infecting isolates of the respective patients. Sera from all available time points during the study follow up were used.

Heterologous sera (sera derived from other patients than the tested isolate) were used to test general recognition of the isolates in the population. For this a panel of sera from in total 18 patients in the same study were chosen to characterize heterologous recognition of the parasite isolates (Table 3). Sera from day 0 (acute) and 14 (convalescent) were used. All the patients contributing serum to this panel had recurrent infections during the follow-up period of 84 days, and Table 3 shows the day of recurrence and whether the detected recurrent infection was asymptomatic at the time of detection. Active weekly follow-up as in this study is likely to detect an increased number of asymptomatic infections, here > 50% of the recurrent infections were asymptomatic.

Antibodies to the infected red blood cell (irbc) surface were measured by immunofluorescence and flow cytometry. To explore the kinetics of antibody responses to irbc sampled at baseline (day 0) and at the time of recurrent parasite infections each parasite isolate was tested against homologous sera (sera derived from the same patient as the parasite isolate tested) sampled at different time points. Figure 1a, b shows the kinetics of antibody acquisition. 3/6 patients developed antibodies to their baseline parasites (blue lines), with antibody acquisition rising over time (patients 119, 221 and 233). The other three made either a very weak, or no response at all i.e. patients 171, 178 and 245. This variation in antibody responses appeared not to be due to intrinsic differences in the ability of the parasites to be recognized by antibodies because there was no significant correlation between the homologous and heterologous recognition of these parasite isolates (P = 0.221, mixed-effects linear regression model) (Fig. 1c and Additional file 2: Figure S1), and there was no overall difference in antibodies to baseline vs. recurrent isolates measured in heterologous sera (P = 0.215, mixed-effects linear regression model). One explanation could be the duration of the infection (or more precisely: the infection with the clone or clones that were responsible for the malaria episode) before the patients presented to us. If diagnosed quickly and combined with the fast acting ACT used in this trial parasites could have been removed before an effective antibody response was mounted. This hypothesis could not be tested here.

As shown in Fig. 1a recognition of recurrent parasites by homologous sera taken earlier than the recurrence was generally poor. Figure 2a illustrates this further, where homologous recognition of the individual recurrent isolates is shown and the day of recurrent parasitaemia is set at day 0. To explore this further, this pre-existing homologous recognition of recurrent isolates was compared to recognition by a panel of heterologous day 0 (acute) and day 14 (convalescent) sera from other patients in the same study (Table 3). Recurrent parasites tended to be recognized less well by homologous sera than heterologous sera, although with only borderline significance (Fig. 2b) P = 0.063, mixed-effects linear regression model). Overall these observations are consistent with the idea that parasites express antigens corresponding to gaps in the repertoire of antibodies carried by the host.

All sera used to test heterologous recognition of the isolates described above were derived from patients who had also experienced a recurrent, i.e. secondary infection during follow-up. Those patients who were unable to clear their parasites in spite of treatment, resulting in a recrudescent infection are of specific interest. Therefore, MSP2 genotyping analysis was performed on the parasite DNA isolated from these patients to differentiate between recrudescence (persistent blood stage infection) and re-infection (new infection emerging from the liver). A recrudescent infection is defined by the presence of at least one genotype during follow-up also detected before treatment. Thus, 6/18 patients whose sera were part of the heterologous serum panel were confirmed to have recrudescent infections within the follow-up period (Table 3).

Next, the anti-VSA antibody repertoires of the 6 patients with recrudescence was compared to the 12 patients who cleared their initial infections (Table 3). For this their ability to recognize the 14 parasite isolates from the 7 patients described above was measured (Table 2 and Additional file 2: Figure S1). The anti-VSA antibody repertoire of the sera from children who carried recrudescent infections was higher than that of children with only reinfections (Fig. 3), although only borderline significant P = 0.059 as tested using a linear mixed effect regression model. To exclude that this is not simply an effect of reduced drug sensitivity of the parasite, existing data from all children in the initial study experiencing recurrences [17] was used to re-analyse risk of recrudescence as a function of the IC₅₀ values for the long-acting drugs lumefantrine and piperaquine. A recrudescence is not determined by the sensitivity of the baseline isolate to the long-acting drugs used in the study (Additional file 3: Table S2). There was no indication of tolerance or resistance in the response to dihydroartemisinin in the study [17].

It is possible that submicroscopic infections (undetected microscopically but persisting until the recrudescence was detected), may themselves boost antibodies. However, the anti-VSA antibody repertoire in children who later had recrudescent parasites was higher than in those who later had recurrent infections at day 0 (Fig. 3). This suggests that sustained submicroscopic infection in children with recrudescences was not responsible for the difference in antibody levels between these two groups of children. One explanation for the difference is that the ability to recognize heterologous parasites promotes chronic asymptomatic infection as previously proposed by others [9, 23‐25].