GIA of Kenyan plasma samples for various P. falciparum lines. Panel A, box plot of GIA result distribution for N = 197 samples for each parasite line tested (D10, 3D7, and FVO). Horizontal bars indicate the median percent growth inhibition. D10 GIA was conducted with different methodology than that of 3D7 and FVO GIA. All assays utilized one cycle of parasite growth and comparable methods of measuring parasitemia. Plasma exhibited higher growth inhibition against FVO than 3D7 (p = 0.0005, Mann-Whitney test). Panel B, box plot of GIA result distribution for N = 54 samples for D10 and 3D7 GIA utilizing the same methodology. Horizontal bars indicate the median percent growth inhibition. Plasma samples demonstrated more inhibitory activity against 3D7 than D10 (p<0.0001, Mann-Whitney).

Box plot of GIA distribution for each parasite line tested (D10, 3D7, and FVO) divided by child (n = 98) and adult (n = 99). Horizontal bars indicate the median % growth inhibition. The child and adult growth inhibition medians differ significantly from each other for each parasite line tested (Mann-Whitney test). Child growth inhibition is consistently higher than adult growth inhibition for each parasite line tested.

Box plot of GIA distribution of 197 plasma samples stratified by BS status at baseline and age for each parasite line examined (D10, 3D7, FVO). BS- children (n = 17), BS+ children (n = 81), BS− adults (n = 66) and BS+ adults (n = 33). No statistical difference was detected between any comparison groups (ie. BS− children vs. BS+ children with D10 GIA; Mann-Whitney test). A trend that BS− groups had higher growth inhibition than BS+ groups was observed. P values of visually dissimilar pairs are included.

Comparison between different GIA methods. Panel A depicts box plots for D10 GIA using the same 54 dialyzed Kenyan plasma samples to compare one growth cycle (harvest at ring stage) to two growth cycles (harvest at trophozoite stage). Parasitemia was assessed by flow cytometry (10× SYBR Green 1 stain for one cycle, GFP for two cycles). Two growth cycles of D10 GIA had a statistically significant higher median inhibition (18.5% inhibition) compared to one cycle D10 GIA (8.3% inhibition; p = 0.0003; Wilcoxon signed-rank test). Panel B shows the Bland-Altman plot to assess the degree of agreement between paired results after one vs. two growth cycles. The points display the difference in growth inhibition between the two assays (y-axis) against their corresponding average values (x-axis). The horizontal lines correspond to the mean difference (solid line) ±2SD (dashed lines). The mean shows a bias of -11.05, indicating the two assays are producing different results with a trend of increasing differences with increased averages. Panel C depicts box plots for 3D7 GIA using the same 54 Kenyan samples. All samples were dialyzed prior to use. 3D7 GIA using one cycle of parasite growth, harvest at the trophozoite stage with parasitemia measured by pLDH was compared to 3D7 GIA using two growth cycles, harvest at the trophozoite stage with parasitemia measured by flow cytometry (EtBr). No statistically significant difference was noted (Wilcoxon signed-rank test). Panel D shows the Bland-Altman plot to assess the degree of agreement between paired assay results. The points display the difference in growth inhibition between the two assays (y-axis) against their corresponding average values (x-axis). The horizontal lines correspond to the mean difference (solid line) ±2SD (dashed lines). Panel E depicts box plots for D10 GIA using the same 54 Kenyan samples either dialyzed or not dialyzed. GIA was performed using one cycle of parasite growth (harvest at ring stage) followed by flow cytometry to measure parasitemia (SYBR Green 1 stain for dialyzed samples, Hoechst stain for non dialyzed samples). No statistically significant difference was noted (Wilcoxon signed-rank test). Panel F shows the Bland-Altman plot to assess the degree of agreement between paired assay results. The points display the difference in growth inhibition between the two assays (y-axis) against their corresponding average values (x-axis). The horizontal lines correspond to the mean difference (solid line) ±2SD (dashed lines). The mean has minimal bias (−2.8) with a small trend of increasing differences with increased averages.

GIA of 5 representative Kenyan samples tested in parallel against 3 parasite lines (D10, 3D7 and FVO). Individual A has significant inhibitory activity against the 3 lines tested whereas individual D has essentially no inhibitory activity against all lines tested. Individual B has a growth inhibition pattern that reflects that of the entire population (Figure 1). Other individual samples demonstrate variable growth inhibitory activity against the different lines tested.

Box plot of GIA distribution for each parasite line tested (D10, 3D7, and FVO) with only children's plasma samples (n = 98) divided by age group: 1–4 years (n = 28), 5–9 years (n = 48), and 10–14 years (n = 22). The youngest age group consistently had higher growth inhibition compared to children in older age groups (Mann-Whitney test).

Relationship of growth inhibition with time to infection. Time to infection in individuals with upper quartile GIA results (>60% inhibition) tested against 3D7 compared to lower quartiles GIA results (<60% inhibition) controlling for age (Cox regression, p = 0.0438). Kaplan-Meier curves are divided by child and adult to illustrate age effects on growth inhibition. Similar analyses controlling for age were performed with D10 and FVO GIA.