In-house ELISA method to analyze anti-Trypanosoma cruzi IgG reactivity for differential diagnosis and evaluation of Chagas disease morbidity

Introduction: The goal was to develop an in-house serological method with high specificity and sensitivity for diagnosis and monitoring of Chagas disease morbidity. Methods: With this purpose, the reactivities of anti-T. cruzi IgG and subclasses were tested in successive serum dilutions of patients from Berilo municipality, Jequitinhonha Valley, Minas Gerais, Brazil. The performance of the in-house ELISA was also evaluated in samples from other relevant infectious diseases, including HIV, hepatitis C (HCV), syphilis (SYP), visceral leishmaniasis (VL), and American tegumentary leishmaniasis (ATL), and noninfected controls (NI). Further analysis was performed to evaluate the applicability of this in-house methodology for monitoring Chagas disease morbidity into three groups of patients: indeterminate (IND), cardiac (CARD), and digestive/mixed (DIG/Mix), based on their clinical status. Results: The analysis of total IgG reactivity at serum dilution 1:40 was an excellent approach to Chagas disease diagnosis (100% sensitivity and specificity). The analysis of IgG subclasses showed cross-reactivity, mainly with NI, VL, and ATL, at all selected serum dilutions. Based on the data analysis, the IND group displayed higher IgG3 levels and the DIG/Mix group presented higher levels of total IgG as compared with the IND and CARD groups. Conclusions: These findings demonstrated that methodology presents promising applicability in the analysis of anti-T. cruzi IgG reactivity for the differential diagnosis and evaluation of Chagas disease morbidity.

Lilian da Silva Santos 1 , Rosália Morais Torres 2 , Girley Francisco Machado-de-Assis 1 , Maria Terezinha Bahia 1,3 , Helen Rodrigues Martins 1,4 , Andréa Teixeira-Carvalho 5 , Jordana Grazziela Alves Coelho-dos-Reis 6 , Pedro Albajar-Viñas 7 , Olindo Assis Martins-Filho 5 and Marta de Lana 1,8 Chagas disease is endemic to a wide region extending from Mexico to Argentina and is one of the parasitic diseases with a major social impact in Latin America, mainly due to its high morbidity and mortality, including the occurrence of sudden death 1 .In the past decades, Trypanosoma cruzi infection has been increasingly found in non-endemic countries, such as the United States, Canada, Japan, Australia, and many European countries, mainly due to increasing population mobility 2 .
Chagas disease has two successive phases: the acute and the chronic phase, with different clinical forms referred to as asymptomatic (indeterminate form (IND)) for many years throughout the lifetime of the majority of infected people.In approximately 30% of patients, the disease evolves to cardiac (CARD), digestive (DIG), and/or mixed clinical forms 3 .
During the acute phase or during reactivation due to immunosupression, the diagnosis is carried out by the direct detection of circulating parasites.However, during the chronic phase, because of low parasitemia, the diagnosis is performed mainly by serological methods.The enzyme-linked immunosorbent assay (ELISA) is the most used serological technique due its high sensitivity and specificity, which is dependent on the antigen employed, the cutoff point used, and the ability to process large quantities of samples rapidly 4 .
However, while the serological methods used in routine diagnosis of T. cruzi infection are very effective in detecting infected individuals with high and medium antibody titers, they fail to detect patients with low immunoglobulin levels 5 .This and other limitations have been the main reason of the continuing research on the ELISA technique and its targeted antigens for the purpose of eliminating the occurrence of false-positive and false-negative results.

METHODS
Santos LS et al -ELISA in the differential diagnosis of Chagas disease and morbidity evaluation and American tegumentary leishmaniasis (ATL) was assessed using a whole lysate and recombinant (CRA + FRA) antigen 6 .The recombinant ELISA showed 100% sensitivity and specificity, with no cross-reactivity with other infections.However, the conventional ELISA showed serological cross-reactivity with VL.Although the use of purified antigen preparations increased the specificity of the ELISA tests, its availability in health systems is still limited due to its higher cost.
Several authors have already studied the relationship between the intensity of serological reactivity of different IgG subclasses and Chagas disease morbidity [7][8][9][10][11][12] .Morgan et al detected increased levels of IgG2 in CARD patients using an ELISA test based on epimastigotes of the Brazil strain.Verçosa et al. also verified increased levels of IgG2 in CARD patients using an ELISA test based on the flagellar portion of the T. cruzi.Increased levels of IgG1 and IgG2 in all Chagas disease patients, independent of their clinical form, were found.At the same time, the highest levels of IgG2 were found in the CARD group 9 .Different results using other antigenic preparations were observed and have been verified to be predominantly IgG1 and IgG3 in all studied Chagas disease patients, while IgG2 reactivity was observed in only 50% of cases 13 .However, equal concentrations of IgG1, IgG2, IgG3, and IgG4 in all studied Chagas disease patients, independent of their clinical form, were verified 7 .Associations between other classes of immunoglobulins and clinical forms of Chagas disease have also been found.Higher levels of IgM were predominant in patients presenting the cardiac form, while higher levels of IgA were predominant in those with the digestive form 10 .The predominance of IgA was also verified in patients in the acute phase and in those with the digestive form in the chronic phase 14 , corroborating previous data from this team 15 .
Taking into account the limitations of the available serological tests for Chagas disease diagnosis and the existent controversy regarding the correlation between serological reactivity and clinical manifestations of Chagas disease, the objectives of this investigation were to develop an in-house ELISA, to propose criteria for the analysis of the reactivity of total IgG and IgG subclasses to T. cruzi antigens, and to evaluate the applicability of the in-house method in the differential diagnosis of Chagas disease and the analysis of correlations with the clinical forms of the disease.

Study area
The municipality of Berilo is located in the Jequitinhonha Valley, northeast of Minas Gerais State, Brazil (10 o 57'06"S; 42o 27'56"W), at an altitude of 401m.The territorial area of Berilo is 586km 2 , and its total population is 17,549 inhabitants, 77% of whom live in the rural area.The economic activity of the population is mainly based on agriculture and livestock production.

Patients and sera samples
Sixty Chagas disease patients (CH) were selected in Berilo municipality, Minas Gerais State, Brazil, from March to November 2006.All patients presented positive serological results in the diagnosis of T. cruzi infection (ELISA and indirect haemaglutination assay (IHA)).The clinical evaluation was performed through anamnesis, physical examination, conventional electrocardiogram (ECG), thorax X-ray, and esophagus and colon barium-contrast X-rays 16 .Additionally, echocardiographic evaluation of all patients was performed using a Cypress portable device with unidimensional, bidimensional, continuous, pulsate and color Doppler adapted for tissue Doppler imaging.A multifrequency transductor, at 3.0 to 3.5MHz, with simultaneous electrocardiography registration was used.A standard echocardiography protocol was performed assessing the following variables: cardiac dimensions, left ventricular systolic and diastolic functions, and the global and segmental contractibility of the left ventricle.The patients were clinically classified according to the Brazilian Consensus on Chagas Disease 17 .Those with neither Chagas disease symptomatology nor alterations in the physical examination, ECG, echocardiography, thorax X-ray, and esophagus and colon contrast X-rays were classified under indeterminate form (IND) (n=15, including 9 males and 6 females, with ages ranging from 13 to 57 years).The CARD group (n=30, including 5 males and 25 females, with ages ranging from 25 to 60 years) included patients with any type of cardiac alteration detected through either referred symptoms or detected signs in the physical examination, ECG, and echocardiography; and without any alteration in the thorax X-ray, and esophagus and colon contrast X-rays.Another group of Chagas disease patients simultaneously presenting digestive alterations detected by esophagus and colon contrast X-rays, with or without cardiac alterations, were classified as digestive/mixed (DIG/Mix) group (n=15, including 6 males and 9 females, with ages ranging from 37 to 59 years).
All patients in this study were integrated into the clinical assistance program of the Chagas disease project of Jequitinhonha Valley, Minas Gerais, and received continuous assistance when necessary.
A control group of noninfected individuals (NI) (n=15, including 6 males and 9 females, with ages ranging from 17 to 54 years) was selected from the local population of Berilo municipality, Minas Gerais, Brazil; all of them presented negative results of Chagas disease serological diagnosis.These subjects were also clinically evaluated through anamnesis, physical examination, conventional ECG, and chest X-ray.

In-house ELISA for evaluation of anti-Trypanosoma cruzi total IgG and IgG subclasses
The in-house ELISA methodology modified 18 proceeded as follows.Briefly, each well of the ELISA plate was sensitized overnight at 4 o C with 100µL of T. cruzi Y-strain antigen (4.5μg/mL) prepared by alkaline extraction.Following washing steps and blocking procedures with phosphate-buffered saline (PBS) supplemented with 10% fetal bovine serum, 100µL of serial dilution from each sera sample (1:10, 1:20, 1:40, 1:80, 1:160, 1:320, 1:640, and 1:1,280) was added to selected wells and the plate incubated for 45min at 37 o C.After washing with PBS 0.05% Tween 20, 100µL of mouse anti-human anti-IgG and anti-IgG subclass biotinilated antibodies (Sigma, St. Louis, USA) were added to specific wells and the plate incubated for 45min at 37 o C. The standardization of each conjugate dilution was previously performed using a range of serial dilutions (anti-IgG -1:7,500, 1:10,000, 1:15,000, and 1:25,000; anti-IgG1 -1:500, 1:1,000, 1:2,500, and 1:5,000; anti-IgG2 -1:2,500, 1:5,000, and 1:7,500; and anti-IgG3 -1:1,000 and 1:2,000) to assay positive and negative sera samples at 1:80 dilution.The anti-IgG4 conjugate was also titrated and did not show any reactivity; therefore, it was excluded from further analysis (data not shown).The selected dilution for each conjugate was established considering the highest difference between the median OD for positive and negative samples (delta median -Δ) (Figure 1).By this approach, 1:10,000, 1:500, 1:2,500, and 1:1,000 dilutions were selected for anti-IgG, anti-IgG1, anti-IgG2, and anti-IgG3 conjugates, respectively (Figure 1).Following incubation and washing procedures with PBS 0.05% Tween 20, 100µL of streptavidin/horseradish peroxidase (HRP) (Sigma, St. Louis, USA) was added to all wells and the plates incubated for 45min at 37 o C. The standardization of the streptavidin/HRP dilution was previously performed using a range of serial dilutions (1:500, 1:1,000, and 1:2,500), and the dilution 1:1,000 was selected as the best condition to segregate positive and negative samples.After incubation and washing procedures with PBS 0.05% Tween 20, 100µL of substrate solution (1mg of orthophenylenediamine (OPD), 3L of H 2 O 2 (30vol), and 15mL of citrate buffer) was added to each well and the plates incubated for 5min at 37 o C. Following incubation, the enzymatic reaction was stopped by adding 32µL of 2.5M H 2 SO 4 solution to each well.The optical density (OD) was determined using a 490nm filter (SOFTMAX® PRO 4.0-Life Sciences Edition).

Data interpretation and statistical analysis
The best sera dilution to segregate the IgG reactivity of sera samples from T. cruzi-infected patients and NI was initially established along the titration curve, considering the highest difference between the median OD for positive and negative samples (delta median -Δ) (Figure 2).Thereafter, the sera dilution that was shown to have higher performance by the construction of the receiver operating characteristic curve (ROC) using the Medcalc statistical program (Broekstraat, Mariakerke, Belgium) was selected and identified as the most useful condition to discriminate chagasic patients from NI individuals.The cutoff point to discriminate serological reactivity was selected for each immunoglobulin (total IgG, IgG1, IgG2, and IgG3), taking into account the ROC curve parameters, including the specificity, sensitivity index, and area under the curve (AUC) (Figure 2).The selected conditions were used to evaluate the IgG reactivity of sera samples from patients with other relevant infectious-parasitic diseases (Figure 3).The data were further analyzed with the two-graph receiver operating characteristic (TG-ROC) using the CMDT software package 19 .This method reads valid pairs of sensitivity and specificity considering pre-assigned cutoff values and allows the selection of the best condition for data interpretations.The cutoff of the sensitivity and specificity intersection point (θ 0 ) provides equal test indices.According to Greiner et al., whenever the value of θ 0 is lower than a preselected accuracy level (90% or 95%), two cutoff values are selected to represent the limits of an intermediate range (IR).The IR can be considered as a borderline range for the clinical interpretation of the test results.The selection of cutoff values was corroborated by graphical displays of logarithmically transformed likelihood ratios (LR) (LR+=Se/ (1-Sp) and LR-=(1-Se)/Sp).The Efficiency and Youden (J) indices were also presented to support the cutoff selection (Figure 4).

Differences in the OD median values of IgG reactivity between different clinical groups of Chagas disease patients (IND, CARD,
and DIG/Mix) were evaluated by the Kruskal-Wallis test, followed by Dunn's test, using the GraphPad Prism software (San Diego, CA, USA).Statistical significance in all cases was considered at p<0.05 (Figure 5).

Ethical considerations
All volunteers were included in the study after they had read and signed an informed consent form approved by the Ethics Committee of the Centro de Pesquisas René Rachou (CPqRR), FIOCRUZ, Belo Horizonte, Minas Gerais, Brazil (Process 0007/02).

Discrimination between Trypanosoma cruzi-infected and noninfected individuals by the differential reactivity of total IgG and IgG subclasses
Upon its standardization, the in-house ELISA (Figure 1) method was used to evaluate the reactivity of total IgG and IgG subclasses, with the aim of discriminating T. cruzi-infected (CH) from noninfected individuals (NI).During the standardization procedure, the best dilution of second-step reagents to be applied for diagnosis purposes was chosen by considering the delta median (Δ median) between positive (CH) and negative samples (NI), as well as the ROC curve indexes for each immunoglobulin.Using a similar approach to select the best sera dilution to be applied for diagnosis purposes, our findings demonstrated that only the first three serum dilutions presented an outstanding ability to discriminate the reactivity of total IgG, and IgG1, IgG2, and IgG3 subclasses along the titration curve, to segregate patients with T. cruzi infection (CH) from noninfected controls (NI).The data analysis demonstrated distinct Δ median values for IgG (0.39, 0.35, and 0.34), IgG1 (0.29, 0.15, and 0.09), IgG2 (0.27, 0.23, and 0.16), and IgG3 (0.45, 0.42, and 0.41) at serum dilutions of 1:10, 1:20, and 1:40, respectively.Although the Δ median value was higher at 1:10 dilution for all IgG reactivity, the selected sera dilution also took into account the best performance in the ROC curve analysis including sensitivity, specificity, and AUC.Using these criteria, the sera dilution 1:40 elected for IgG, 1:10 for IgG1 and IgG2, and 1:20 for IgG3 (Figure 2).The use of the cutoff edge indicated by the ROC curve analysis revealed outstanding performance indexes for IgG, IgG1, and IgG2, with 100% sensitivity and specificity, and AUC=1.0.Moreover, the LR ratio analysis by the ROC curve further confirmed these indicators (Figure 2).The analysis of IgG3 demonstrated indexes with lower power to discriminate the clinical groups CH and NCH, with 98.3% sensitivity and AUC=0.998(Figure 2).

Discrimination between Trypanosoma cruzi infection and other infectious diseases by the differential reactivity of total IgG and IgG subclasses
Considering the excellent performance of the in-house ELISA in discriminating chagasic patients from noninfected controls, an additional analysis was subsequently performed to evaluate the reactivity of sera samples from chagasic patients in parallel with samples from noninfected controls and also from patients with other relevant infectious-parasitic diseases, including HIV, HCV, SYP, VL, and ATL.The analysis of total IgG at serum dilution 1:40 was able to discriminate patients with T. cruzi infection from patients with the other studied infectious diseases (HIV, HCV, SYP, VL, and ATL) and NI individuals, with maximum performance indexes, including 100% sensitivity and specificity, AUC of 1.0, and LR+ and LR-values indicating infinite probability of finding a positive result for CH and a negative result for NCH (Figure 3).
Although the analysis of IgG1 at serum dilution 1:10 showed 100% sensitivity, indicating that all CH presented a positive result, the low specificity (90.7%) indicated the presence of cross-reactivity in 5 out of 54 tested samples, including 7.4% false-positive results due to four VL samples and 1.9% due to one ATL sample.The LR+=10.38, and LR-=0 indicated lower performance as compared with total IgG (Figure 3).
The evaluation of IgG2 reactivity at serum dilution 1:10 also showed 100% sensitivity but also revealed lower specificity (88.9%), with cross-reactivity in 11.1% of the tested samples, due to the existence of false-positive results in three VL and three ATL samples.The AUC=0.976, LR+=9, and LR-=0 indicated lower performance as compared with total IgG (Figure 3).
The analysis of IgG3 reactivity at serum dilution 1:20 demonstrated low sensitivity (95%), with three out of 60 patients with T. cruzi infection showing OD values below the cutoff point.The specificity of 98.1% was due to cross-reactivity observed in one VL sample.The AUC=0.992 showed high accuracy.Although the LR+=51.3indicated that a positive result has higher probability of belonging to the T. cruziinfected population as compared with NCH, the LR-=0.05showed borderline performance in identifying negative results (Figure 3).
Considering the good performance of the in-house ELISA when applied to the diagnosis of Chagas disease and aiming to validate our findings with supplementary performance analysis, the data were further evaluated using novel parameters, including the TG-ROC, the LR ratio in log base graphs, and the simultaneous analysis of the and Youden indexes (Figure 4).In the first approach, valid pairs of sensitivity and specificity were read for pre-assigned cutoff values directly from the TG-ROC plots 19 .The TG-ROC analysis showed that the in-house ELISA used to detect total IgG reactivity did not present any intermediate range (IR) that can be considered a borderline range for the clinical interpretation of test results, therefore indicating that the method displayed 100% valid data with no incorrect results.These findings were also confirmed by the maximum LR+ and LR-, and the Efficiency and Youden index analysis (Figure 4).Together, these analyses showed that the anti-T.cruzi IgG detected by the in-house ELISA at sera dilution 1:40 using the cutoff point=0.272 was adequate to confirm the diagnosis of Chagas disease and exclude non-chagasic samples (Figure 4).
Although the analysis of IgG2 performance also evidenced the absence of an intermediate range in the TG-ROC graph, as reported in the previous analysis (Figure 3), it was not possible for the IgG2 reactivity to conclusively discriminate T. cruzi-infected individuals from non-chagasic patients, considering the cross-reactivity with sera from VL and ATL patients.
The most unsatisfactory results were found in the analysis of IgG1 and IgG3 reactivity.The analyses of IgG1 and IgG3 anti-T.cruzi reactivity revealed the presence of an intermediate range, corresponding to the existence of gray zones that do not guarantee a clear discrimination between T. cruzi-infected and noninfected individuals.Moreover, the LR analysis of IgG1 reactivity did not show outstanding performance.Therefore, together with the previous analysis (Figure 3), these findings did not recommend the use of the IgG1 and IgG3 detected by the in-house ELISA proposed in this study as a valuable method for diagnosis of Chagas disease (Figure 4).

Applicability of the in-house ELISA in analyzing total IgG and IgG subclasses reactivity and monitoring Chagas disease morbidity
The analysis of IgG reactivity at serum dilution 1:40 revealed that the DIG/Mix group presented significantly higher OD reactivity as compared with the IND and CARD groups (Figure 5).On the other hand, the IgG1 reactivity at serum dilution 1:10 and the IgG3 reactivity at serum dilution 1:20 demonstrated that the median OD for the IND group was significantly lower as compared with that for the CARD group (Figure 5).No significant differences were observed in IgG2 reactivity at sera dilution 1:10 among the distinct clinical groups evaluated (Figure 5).

DISCUSSION
In this study, an in-house ELISA method was standardized to better determine the reactivity of total IgG, and IgG1, IgG2, and IgG3 subclasses in T. cruzi-infected and noninfected patients.Additionally, the in-house ELISA technique was applied to verify the reactivity of these immunoglobulin classes and subclasses in groups of individuals with T. cruzi infection and other relevant infectious-parasitic diseases (HIV, HCV, SYP, VL, and ATL).Initially, after the standardization procedures to define the best dilutions of the second step reagents (Figure 1), the goal of the study was to verify the best serum dilutions able to discriminate Chagas disease patients from the other selected groups.Our findings demonstrated that only total IgG reactivity showed no cross-reactivity with sera from patients without T. cruzi infection.However, the IgG1 and IgG2 analysis showed crossreactivity with sera samples from VL and ATL patients.IgG2 showed higher cross-reactivity with sera samples from ATL as compared with IgG1 and IgG3.This finding is in agreement with previous reports showing that T. cruzi-infected patients presented high levels of crossreactivity of IgG2 with Leishmania braziliensis antigens 20 .
The cross-reactivity between anti-Leishmania and anti-T.cruzi IgG2 may explain these data.Although IgG3 showed lower crossreactivity with other diseases, with only one false-positive result, the performance of this subclass was not validated due to the low sensitivity index.These findings agreed with the data in the literature demonstrating the existence of shared antigens between T. cruzi and Leishmania sp., such as carbohydrate epitopes [20][21][22][23] .On the other hand, the use of crude antigens in this in-house ELISA may explain this cross-reactivity since the same phenomenon was not observed in the sera of patients with VL, ATL, SYP, HIV, and HCV when CRA+FRA recombinant antigens were used 6 .Thus, the discovery and availability of purified antigens or antigenic fractions from T. cruzi with higher specificity and similar sensitivity to avoid cross-reactions among these and other infections may represent a next step in further investigations to validate in-house ELISA methods.Nevertheless, the present study demonstrated that the detection of total IgG using an in-house ELISA technique with T. cruzi crude alkaline antigenic extracts, as proposed by our group at the Laboratório de Doença de Chagas of the Universidade Federal de Ouro Preto, represent a valuable tool for the diagnosis of T. cruzi infection, with outstanding performance indexes.The major changes performed in the present investigation as compared with the original methodology proposed 18 included the use of biotinilated second-step reagents, the use of serial dilutions for all tested serum samples, and a standardized incubation temperature of 37 o C. The use of biotinilated reagents may lead to increased performance of this in-house ELISA due to the reagents' ability to expand the detection capacity of the system from that in the original methodology.Moreover, these reagents require a shorter development time (5min) as compared with that in the original method (15min), which may reduce the occurrence of cross-reactivity and false-positive results.

FIGURE 1 -
FIGURE 1 -Standardization of in-house ELISA method for the detection of total IgG and IgG subclasses (IgG1, IgG2, and IgG3) using Trypanosoma cruzi Y-strain alkaline antigen.

FIGURE 2 -
FIGURE 2 -Analysis of in-house ELISA reactivity of total IgG and IgG subclasses in sera samples from Chagas disease patients (CH) and noninfected controls (NI) using Trypanosoma cruzi Y-strain alkaline antigen.
Titration curves were used to select the best sera dilution to discriminate Trypanosoma cruzi-infected (CH) from noninfected (NI) patients, with the Δ median values and the ROC curve indexes as basic parameters.The gray rectangles indicate the selected serum dilution with higher Δ median total IgG, IgG1, IgG2, and IgG3.The ROC curve and the scatterplot are referent to the selected sera dilutions (IgG=1:40, IgG1=1:10, IgG2=1:10, and IgG3=1:20).The performance indexes at the selected sera dilutions are provided in the figure.The horizontal lines in the scatterplot indicate the cutoff point for segregating positive and negative results.

Santos
LS et al -ELISA in the differential diagnosis of Chagas disease and morbidity evaluation p were further segregated into noninfected controls (NI=○) and patients with other infectious diseases, such as human immunodeficiency virus (HIV=), hepatitis C virus (HCV=), syphilis (SYP= ), visceral leishmaniasis (VL=), and American tegumentary leishmaniasis (ATL=).The performance indexes, including sensitivity, specificity, area under curve (AUC), LR+, and likelihood ratios (LR), are provided in the inserted table.NCH: non-chagasic patients.