Fig 1.
Outline of TcI/TcVI/TcII Chagas-Flow ATE-IgG2a for serodiagnosis of Trypanosoma cruzi infection.
(A) The experimental procedure display a schematic representation of genotype-specific T. cruzi ATE-parasite Mix Platforms using TcI (Colombiana strain) = black bar, TcVI (CL strain) = light gray bar and TcII (Y strain) = dark gray bar antigens in separate batches. (B) Representative gating strategies used to select the target antigens (amastigote-AMA, trypomastigote-TRYPO and epimastigote-EPI) on each Chagas-Flow ATE-IgG2a platform and the histograms employed to quantify the genotype-specific anti-T. cruzi IgG2a reactivity, expressed by the percentage of positive fluorescent parasites (PPFP), based on the positivity limit (PPFP<2%), set based on the internal control.
Fig 2.
Overall reactivity profile of TcI/TcVI/TcII Chagas-Flow ATE-IgG2a for universal and genotypic-specific diagnosis of T. cruzi infection.
The Chagas-Flow ATE-IgG2a reactivity was determined for sera samples from T. cruzi-infected mice, including TcI/TcVI/TcII genotype-representative strains, including: TcI/Colombiana strain (black dotted frame, n = 29), TcVI/CL strain (light gray dotted frame, n = 29) and TcII/Y strain (dark gray dotted frame, n = 35) as well as non-infected mice (white dotted frame, n = 10). Genotype-specific IgG2a reactivity to each target-antigen (amastigote-AMA, trypomastigote-TRYPO and epimastigote-EPI) from T. cruzi genotype I (left panels), genotype VI (middle panels) and genotype II (right panels) was assessed at eight serum dilutions (1:500 to 1:64,000). The results are expressed as the percentage of positive fluorescent parasites (PPFP), using the box plot format, stretching from min to max values with outliers represented by gray-shaded dots and the box defining the 25th and 75th percentile and the median value (line across the box). Comparative analyses were performed by the Kruskal-Wallis followed by Dunn’s post test for multi-group comparisons. Significant differences were considered at p<0.05. The light gray continuous rectangle selects the pair of attributes (“target antigen/serum dilution”) with the most consistent ability to discriminate non-infected mice from all T. cruzi-infected hosts (Colombiana, CL and Y strains). Therefore, these features (anti-TcII TRYPO reactivity at 1:500) were selected for universal diagnosis of T. cruzi infection. The dark gray dotted frame select the pair of attributes “target antigen/serum dilution” with the most promising perspective to distinguish the reactivity of sera samples amongst host infected with Colombiana, CL or Y T. cruzi strains. Therefore, these features (anti-TcII AMA reactivity at 1:1,000; anti-TcI TRYPO reactivity at 1:4,000 and anti-TcVI EPI reactivity at 1:1,000) were selected for genotype-specific diagnosis of T. cruzi infection.
Fig 3.
Performance of TcI/TcVI/TcII Chagas-Flow ATE-IgG2a for universal diagnosis of T. cruzi infection.
(A) The anti-TcII TRYPO reactivity at 1:500, pre-selected as attributes pairs for universal T. cruzi infection diagnosis, were compared by Kruskal-Wallis followed by Dunn’s post test for multi-group comparisons and significant differences at *p <0.05, **p<0,001 and ***p<0,0001, highlighted by connecting lines. Data are expressed as median PPFP values for non-infected mice (white bar) and T. cruzi-infected hosts (TcI/Colombiana strain = black bar, TcVI/CL strain = light gray bar and TcII/Y strain = dark gray bar). The similarity amongst the anti-TcII TRYPO IgG2a reactivity at 1:500 observed for the three T. cruzi infected groups (Colombiana + CL + Y strains) allows the establishment of a single group referred to as T. cruzi infected hosts (n = 93) and the performance of the TcI/TcVI/TcII Chagas-Flow ATE-IgG2a in the universal diagnosis of T. cruzi infection carried out as compared to a group of non-infected mice (NI, n = 10). (B) ROC-curve analysis was applied to define the appropriated cut-off to discriminate the PPFP values from NI and T. cruzi-infected host (Colombiana + CL + Y strains). Additional performance indices were also calculated and provided in the figure, including the area under the curve (AUC), defined as global accuracy, the sensitivity (Se) and the specificity (Sp). (C) Representative scatter plot illustrates the ability of the selected set of attributes (“target-antigen/serum dilution/cut-off”) to discriminate the reactivity of the sera from non-infected (NI) and T. cruzi-infected hosts (Colombiana+CL+Y). The dotted line represented the cut-off of PPFP = 20% defined by the ROC-curve analysis. (D) TG-ROC analysis was also performed to confirm the cut-off selection at higher “Se” and “Sp”, highlighted by dark gray dotted frame.
Fig 4.
Overall reactivity of TcI/TcVI/TcII Chagas-Flow ATE-IgG2a applied for genotype-specific diagnosis of T. cruzi-infection.
(A) The anti-TcII AMA reactivity at 1:1,000, (B) the anti-TcI TRYPO reactivity at 1:4,000 and (C) the anti-TcVI EPI reactivity at 1:1,000, pre-selected as attributes for genotypic-specific diagnosis of T. cruzi-infection, were further evaluated by histogram plot distributions and trendlines built by nonlinear regression. The results were expressed as the proportion of samples displaying a given PPFP values amongst T. cruzi-infected hosts, including: TcI/Colombiana strain (red bar, n = 29), TcVI/CL strain (green bar, n = 29) and TcII/Y strain (blue bar, n = 35). The reactivity amongst T. cruzi-infected groups was further compared for each pair of attributes, including: (D) anti-TcII AMA reactivity at 1:1,000, (E) the anti-TcI TRYPO reactivity at 1:4,000 and (F) the anti-TcVI EPI reactivity at 1:1,000. The results are expressed as median PPFP values in box plot format with the outliers underscored by gray-shaded dots. Data analyses were carried out by Kruskal-Wallis followed by Dunn’s post test for multi-group comparisons The significant differences were indicated by asterisk at *p <0.05, **p<0,001 and ***p<0,0001 and highlighted by connecting lines.
Fig 5.
Performance of combined TcI/TcVI/TcII Chagas-Flow ATE-IgG2a for genotype-specific diagnosis of T. cruzi infection in two population/prototype scenarios.
3D plots were employed to identify clusters of TcI/TcVI/TcII Chagas-Flow ATE-IgG2a reactivity amongst sera samples from T. cruzi-infected mice in two population/prototypes including (A) TcI/Colombiana (red circle, n = 29) vs TcVI/CL (green circle, n = 29) vs TcII/Y (blue circle, n = 35) or (D) TcI/Colombiana (red circle, n = 29) vs TcII/Y(blue circle, n = 35), using the three selected pair of attributes (“target-antigen/serum dilution”). Data are expressed as Log of PPFP values for anti-TcI TRYPO at 1:4,000 (left lateral axis), anti-TcII AMA at 1:1,000 (vertical axis) and anti-TcVI EPI at 1:1,000 (right lateral axis). Decision trees were constructed using the set of attributes (“target-antigen/serum dilution/cut-off”) to create algorithms (root and branch attributes) to classify T. cruzi infected mice in a population/prototype including (B) TcI/Colombiana (red rectangle) vs TcVI/CL (green rectangle) vs TcII/Y (blue rectangle) or including (E) TcI/Colombiana (red rectangle) vs TcII/Y(blue rectangle). Global accuracy and leave-one-out-cross-validation-LOOCV are provided in the Figure. Bar charts representing the performance of the decision trees demonstrate the number of animals that ranked within each branch amongst the T. cruzi-infected hosts for (C) (TcI/Colombiana = red bar vs TcVI/CL = green bar vs TcII/Y = blue bar) and (F) (TcI/Colombiana = red bar vs TcII/Y = blue bar).
Fig 6.
Criteria to define universal and genotype-specific diagnosis of T. cruzi infection by TcI/TcVI/TcII Chagas-Flow ATE-IgG2a in two population/prototype scenarios.
(A) Reactivity boards were built to provide a panoramic snapshot of TcI/TcVI/TcII Chagas-Flow ATE-IgG2a applied to the universal and genotype-specific diagnosis of T. cruzi infection. Data mining approaches were used to pre-select the target-antigens and specific cut-off edges to define positive results for universal diagnosis purpose (TcII TRYPO, PPFP>20%) and genotype-specific diagnosis criteria (TcI TRYPO, PPFP>50%; TcII AMA, PPFP>40% and TcVI EPI, PPFP>45%) based on the differential positive reactivity of sera samples (TcI/Colombiana strain = black rectangle, TcVI/CL strain = light gray rectangle and TcII/Y strain = dark gray rectangle) from negative reactivity (white rectangle) observed for T. cruzi infected hosts and non-infected mice. The pre-selected sera dilutions defined by decision tree analysis are underscored by dotted rectangles and include TcII TRYPO PPFP>20% at 1:500 for universal diagnosis and the set of attributes (TcI-TRYPO/PPFP>50%/4,000 followed by TcII-AMA/PPFP>40%/1,000 and TcVI-EPI/PPFP>45%/1,000) for genotype-specific diagnosis criteria. Reactivity panels were constructed to define the diagnosis conclusion when applying TcI/TcVI/TcII Chagas-Flow ATE-IgG2a for (B) universal diagnosis of T. cruzi infection and genotype-specific diagnosis in a population/prototype including TcI/Colombiana strain (COL), TcVI/CL strain (CL) or TcII/Y (Y) strain or (C) including TcI/Colombiana strain (COL) or TcII/Y strain (Y).