Ethical clearance for this study (NMIR-IRB CPN 024/14-15) was obtained from the Institutional Review Board (IRB) of Noguchi Memorial Institute for Medical Research (NMIMR). Written informed consent was obtained before enrolling participants for this study.

The study was conducted in Simiw ( Figure 1 ); a peri-urban community in the Komenda–Edina–Eguafo–Abrem (KEEA) Municipal Assembly of the Central Region of Ghana. Malaria transmission in Simiw is usually perennial and overlaps with the rainy season peaking in-between May and July. The dry season however begins in November and ends in April (Acquah et al., 2020). Study participants were pupils of the Simiw M/A Basic School-aged from 6 to 12 years.

A total volume of 1 ml of the venous blood sample was collected from each of the 84 study participants once every 2 months beginning November 2017 until May 2018. Prior to sample collection, the axillary temperature was measured using a digital thermometer. Thick and thin smears were prepared according to WHO guidelines (Organization, 2016). Hemoglobin levels were determined using a drop of whole blood (15 μl) and the Urit 12 HB meter (URIT Medical Electronic Co. Ltd., China). Three 50 μl drops of whole blood were also spotted onto Whatmann #3 filter paper (GE Healthcare, USA). The remaining blood was separated into plasma and packed cells and each stored at −20°C.

The thick and thin blood smears were processed according to WHO guidelines (Organization, 2016) and as previously described (Amoah et al., 2019b). The smears were observed under 100× oil immersion objectives by two independent trained microscopists and any inconsistency in deciding a positive or negative slide was settled by the reading of a third microscopist. Parasite density was estimated by multiplying the number of parasites counted per 200 white blood cells (WBCs) by 40 based on the assumption that 1 μl of blood contains 8,000 WBCs.

The Saponin/Chelex DNA extraction method was used as previously described (Amoah et al., 2018). Briefly, two 3 mm discs were punched out of the dried filter paper blood spot into a 1.5 ml Eppendorf tube. A measure of 1 ml solution of 1× phosphate-buffered saline and 50 µl of 10% saponin was added to each tube, vortexed, and kept overnight at 4°C. The discs were washed twice with 1 ml phosphate-buffered saline and kept in 30 µl of 20% chelex and 70 µl of Dnase/Rnase water was incubated at 95°C for 10 min with intermittent vortexing. Finally, the tubes were spun for 6 min at 13,000 rpm and the supernatant containing the extracted gDNA was transferred into a sterile 0.5 ml microfuge tube and preserved at −20°C.

Nested polymerase chain reaction, was used to amplify the 18S rRNA gene of P. falciparum as previously described (Amoah et al., 2019b). The Nest 1 reaction mixture contained 80 nM of primers (rPLU5/rPLU6), 5 μl DNA template, 1× polymerase chain reaction buffer, 167 nM dNTPs, 2.5 mM MgCl₂, and 1 U of OneTaq DNA polymerase, which totaled up to 15 μl mixture. The nest 2 reaction mixture contained 133.33 nM of primers (rFAL1/rFAL2) and 0.5 μl of the nest 1 product. The cycling conditions were set as follows; initial denaturation (95°C for 5 min), then 35 cycles of denaturation at 94°C for 30 s, annealing at 55 and 58°C (for nest 1 and nest 2 respectively) for 1 min and an extension at 68°C for 1 min with a final extension at 68°C for 5 min. The nest 2 polymerase chain reaction products were gel electrophoresed and visualized using UV illumination.

The IgG levels were determined according to previously described protocols (Amoah et al., 2019a). The Pfs230 antigen used in this study was the prodomain, amino acid (aa) residues 443 to 590 while the Pfs48/45 antigen comprised of the C-terminal 6C region (aa 291–428); both of which were expressed in Lactococcus lactis (Acquah et al., 2017). The negative control samples comprised of pooled plasma from malaria naïve individuals. The plasma used for plotting the standard curve and the positive control sample comprised of pooled plasma from individuals who were previously identified as containing high levels of IgG against these two antigens (Arredondo et al., 2012; Jones et al., 2015; MacDonald et al., 2016; Acquah et al., 2017). The Maxisorp plates (Nunc, Thermo Fisher) were coated with 100 µl of antigen (Pfs48/45 or Pfs230) diluted to 1 µg/ml in carbonated buffer (pH of 9.2) and the plasma samples were diluted 1:200 in 1% skimmed milk in 1× PBS containing 0.5% Tween 20. The rest of the procedure was followed through to the reading of the plates using the ELx808 plate reader (BioTek) set at 450 nm as previously described (Amoah et al., 2019a).

The relative avidity of IgG in each sample was determined as previously described (Amoah et al., 2019a). A procedure similar to the ELISA described above was used with an extra step incorporated after the plasma incubation step. Briefly, each plasma sample was added to four wells of the plate, after the set incubation time of 1 h, a 100 μl solution of 2.4 M sodium thiocyanate was added to two of the four wells for 15 min. After this incubation step, the wells of the plates were washed and the procedure followed through with the secondary IgG incubation step to the stop reaction exactly as with the indirect ELISA described above.

The OD values obtained from the plate reader were converted into weighted concentrations (wConcs) using ADAMSEL (Ed Remarque, BPRC). Graphs were plotted using GraphPad Prism version 6 and Microsoft Excel. The Kruskal–Wallis test and Dunn’s multiple comparison post-hoc tests were used to assess statistical significance between median values of quantitative data between age groups (overall, ≥10 yrs and <10 yrs) and time points (November 2017, January 2018, March 2018 and May 2018). Within-individual variation of the antibody and avidity levels across the study period was assessed with the Friedman test. Association between microscopic and submicroscopic P. falciparum prevalence was determined by chi-square test and correlation between IgG responses against Pfs230 and Pfs48/45 were determined using Spearman’s rank correlation (using GraphPad version 6). Linear regression models were fitted for the avidity of antibody levels in the final visit (May 2018) using the number of infections by microscopy, PCR, or both as the predictor variables. The association of P. falciparum carriage in previous months with that of May and their corresponding odds ratio was also determined using Graphpad prism. Statistical significance was determined at P <0.05. The relative avidity index (RAI) for Pfs230 IgG, Pfs48/45 IgG was calculated as the ratio of the IgG concentration of the sodium thiocyanate-treated sample to the IgG concentration of the corresponding untreated sample multiplied by 100.

Plasma samples were assigned seropositive for antibodies against an antigen if the concentration of the antibodies in the sample was greater than the average plus two times the standard deviation of that for the negative control sample (pooled malaria-naïve plasma).

The study recruited non-febrile school children aged from 6 to 12 years residing in the Simiw community in the Central Region of Ghana. A total of 84 children with a median (IQR) age of 9 (8–11) years were recruited in November 2017 and followed until May 2018. A total of 43 of the children were below 10 years old. No significant differences were observed in the median haemoglobin levels (Kruskal–Wallis statistic = 5.250, P = 0.1544) and temperature (Kruskal–Wallis statistic = 5.966, test P = 0.1133) of the children at all the four-time points ( Table 1 ).

Asexual parasite prevalence estimated by microscopy was highest in March 2018 (36.9%) and lowest (7.14%) in May 2018, with the highest and lowest median parasite densities detected in May 2018 and January 2018 respectively ( Table 2 ). Parasite prevalence estimated by polymerase chain reaction was highest (74.4%) in May and lowest (27.3%) in January. Variation in asexual parasite density was not statistically significant over the four time points (P = 0.2303, Kruskal–Wallis statistic = 4.305). None of the samples used in this study tested positive for gametocytes by microscopy ( Table 2 ) at any time point. Parasite carriage in May was significantly associated with that of March (Fisher’s exact text, p = 0.0118) with participants who carried parasites in March having odds of 5.3 of being infected in May ( Additional File 1 ).

The overall median IgG level against Pfs48/45 did not show statistically significant variation from November 2017 to March 2018. In May, however, overall anti-Pfs48/45 IgG levels increased significantly compared to all preceding months of the dry season assessed (p <0.0001). The median Pfs48/45 IgG levels in both the young children (<10 yrs) and older children (≥10 yrs) did not vary significantly from November 2017 to March 2018 except the period from January to March, where IgG levels of the young children decreased significantly (P = 0.867). In May, however, median anti-Pfs48/45 antibody levels increased significantly relative to March levels for both young children (P<0.001) and older children (P<0.01) ( Figure 2A ). Between the two age groups, older children had higher antibody levels than younger children in only March 2018, the second half of the dry season, but statistically the same for the other months assessed.

The median relative avidities of anti-Pfs48/45 IgG in all the study participants (overall and age stratified) did not vary statistically from November to January. In March 2018 however, median relative avidity increased significantly for all children (P <0.001). Transitioning from March to May produced a significant decrease in the overall median relative avidities (P = 0.0292) although it was statistically unchanged for both the young children (P = 0.106) and older children (P = 0.103) ( Figure 2B ) sub-groups. Anti-Pfs4845 antibody levels but not avidity correlated positively (r = 0.282, P = 0.0095) with age in only March, the end of the dry season ( Additional File 1 ). Within individuals, both the level and relative avidity of anti-Pfs48/45 IgG varied significantly between the various time points (P <0.0001 for both) ( Additional File 1 )

The median levels of anti-Pfs230 antibodies of the overall participants significantly increased in March (P = 0.001) and in May (P = 0.0125). In younger children, median IgG levels against Pfs230 increased significantly in only May (P = 0.014). In older children, however, median anti-Pfs230 antibody levels had a statistically significant increase in March (P = 0.002) and remained the same in May (P = 0.250). Median antibody levels were higher for young children than older children in November 2017, the beginning of the first half of the dry season (P = 0.036), while the reverse occurred in March 2018 (P = 0.0007), close to the end of the second half of the dry season ( Figure 3A ). There were no statistically significant differences in the median relative avidities for the participants as a whole across the four time points (P = 0.0673). The relative avidity of anti-Pfs230 IgG in older children (>10 yrs) did not show any statistically significant variations (P = 0.325) over the duration of the study ( Figure 3B ). In younger children (<10 yrs) however, the median relative avidities of anti-Pfs230 IgG dropped significantly in May (P <0.001), to values lower than recorded in the older children ( Figure 3B ). Similar to Pfs48/45, anti-Pfs230 antibody levels correlated positively with age (r = 0.358, P = 0.008) but the relative avidity did not. In May however, both the levels and relative avidities of IgG against Pfs230 were positively correlated (r = 0.225, P = 0.0397 and r = 0.236, P = 0.0305, respectively) with age. The anti-Pfs230 antibody levels and relative avidities also varied significantly within individual participants (P = 0.0049 and P = 0.026, respectively) ( Additional File 1 )

Percentage seroprevalence of IgG antibodies against both Pfs48/5 and Pfs230 of all the participants had a similar trend, decreasing from the beginning of the dry season (November 2018) to the middle of the dry season (January) after which it increased again by the end of the dry season (March) and the early rainy season (May) ( Figure 4 ). The observed trend of overall seroprevalence for both antigens was similar to that of median parasite densities of participants. The percentage seroprevalence of both the younger children and older children categories followed the same trend as the overall participants. In November and January, younger children had a higher seroprevalence of IgG against both antigens than older children. In March and May, however, older children had a higher seroprevalence of antibodies against Pfs230 than the younger children. For anti-Pfs48/45 however, older children had a higher Percentage seroprevalence than younger children in only March ( Figure 4 ).

In the overall dataset, anti-Pfs48/45 IgG levels were higher in children without active infections in November (P = 0.027) and January (p = 0.019). In March however, the IgG levels in children without active infections dropped to similar levels as children with active infections, whose IgG levels remained unchanged from November to March. In May, the early parts of the rainy season anti-Pfs48/45 IgG in both children with and without active infections rose significantly relative to March levels (P <0.001 for both months) ( Figure 5A ).

Children with and without active infections had similar relative anti-Pfs48/45 IgG avidity at all assessed time points. The relative avidities were unchanged between November and January but increased significantly in March (P <0.001) and remained at the same level in May ( Figure 5B ) for both children with and without active infection.

Each of the two age groups (children <10 and ≥10 yrs) were sub-grouped into those with active infections (parasites detected by polymerase chain reaction, PCR+) and those without active infections (no parasites detected by polymerase chain reaction, PCR−) and their anti-Pfs48/45 and anti-Pfs230 IgG antibody levels and avidities compared. It was observed that the antibody responses (both IgG levels and avidities) in the two groups of children (with and without active infection) were similar throughout the study period except in January when the Pfs48/45 IgG levels of young children without active infections was higher than their counterparts with active infections (P = 0.0112) ( Figure 5A ). In May, when younger children without an active infection had higher relative avidity (P = 0.019) than older children with active infection ( Figure 5B ).

Overall, children with active infections had similar anti-Pfs230 IgG levels as those without active infections at all-time points. Anti-Pfs230 IgG levels however increased in children without active infection from January, mid dry season, to March which is towards the end of the dry season. Children with active infections in May had higher anti-Pfs230 IgG levels than in March (P = 0.0041). Children with and without active infections had similar median relative avidities at all-time points in all children as a whole or in the two age categories ( Figure 6B ). In March, older children (≥10 yrs) without active infections had significantly higher anti-Pfs230 IgG levels than their age mates with active infections (P = 0.040) and young children (<10 yrs) with (P = 0.0003)) or without active infections (P = 0.00135) ( Figure 6A ).

In May, the median relative avidity of Pfs230 IgG in older children (≥10 yrs) without active infections was significantly higher (P = 0.0263) than in younger children <10 yrs with active infections ( Figure 6B ). Older children with active infection had significantly lower relative avidity of anti-Pfs230 IgG compared to their age mates without active infections (P = 0.0378) or young children with active infections (P = 0.0315) in January. In May, however, the median relative avidity of older children with active infections rose to that of older children without active infection whose median relative avidity was significantly higher than young children without active infection ( Figure 6B ). The individual antibody levels and relative avidities used in this section can be found in Additional File 1 . Regressional analyses showed that the number of times an individual tested positive for parasites by PCR or microscopy were not significant predictors of the antibody levels and avidities against the two antigens. Consequently, there was no statistical difference between the number of times a person had been infected during our visits and the anti-Pfs230 parasite and anti-Pfs48/45 antibody levels and avidity ( Additional Files 1 and 2 ).

In November, no correlation was identified between the measured IgG levels and avidity against Pfs48/45. Moderate and positive correlations were identified in January which changed to a weak inverse correlation in both March and May. All the correlations identified other than in November were significant.

Significant negative correlations existed between IgG levels and avidity against Pfs230 throughout the study period. The correlation was strong in the first half of the study period but weak towards the peak season (in the second half of the study period) ( Table 3 ).

The quality of the antibodies against an antigen is determined by higher binding affinity as well as functionality. This study did not determine the functionality of the antibodies beyond determining the relative avidity index. Functional assays such as antibody-mediated complement lysis or Opsonic phagocytosis are important measures of antibody functionality. We could however not perform these assays due to limited funding. Furthermore, submicroscopic gametocyte densities, as well as the entomological inoculation rate during the period of the study, were also not determined. These could have provided additional measures of exposure rates as well as the risk of parasite infection due to gametocytes whose surface antigens were studied in the current study. However, studies have identified submicroscopic gametocyte densities in samples that were classified as negative for gametocytes by microscopy (Shekalaghe et al., 2007; Lamptey et al., 2018) as well as have identified a very low prevalence of mosquito vectors in the off-peak malaria season (Magombedze et al., 2018; Epopa et al., 2019).