The studies in which the samples were collected were approved by the ethical committees of the participating institutions, according to the principles expressed in the Declaration of Helsinki. Written informed consent forms were signed by the adult study subjects and from parents/guardians on behalf of all minor subjects. All samples were pre-existent at the time of this international study and were anonymized before being processed.

Seronegative human blood samples were spiked with cultured epimastigotes of Sylvio X10 and CL-Brener stocks (TcId and TcVI, respectively) and were immediately mixed with one volume of guanidine hydrochloride 6 mol/L EDTA 0.2 mol/L buffer, pH 8.00 (GE).

Peripheral blood samples from 156 CD patients were distributed into eight groups according to their geographic origin, as follows. Group 1 (G1) included samples from four seropositive patients from Mexico, two patients with acute CD (G1a) and two patients with asymptomatic chronic CD (G1b). Group 2 (G2) included samples from two patients from French Guiana with acute CD acquired by oral transmission. One patient was positive and the other patient was negative for IgG serologic studies. Both patients experienced cardiac symptoms and were infected with TcI. Group 3 (G3) included samples from five seropositive patients from Bolivia; two patients with acute CD acquired by oral transmission (G3a) and three patients with asymptomatic chronic CD acquired from vectors or congenitally (G3b). Group 4 (G4) included samples from five seropositive patients from Venezuela with acute CD acquired by oral transmission. Two of these patients were infected with TcI. Group 5 (G5) included samples from 13 seropositive patients from Colombia with asymptomatic chronic CD acquired from vectors or congenitally. Five of these patients were infected with TcI. Group 6 (G6) included samples from 21 Bolivian seropositive patients, resident in Spain, with chronic CD acquired from vectors or congenitally. One patient experienced digestive symptoms and the others were asymptomatic. Group 7 (G7) included samples from 31 seropositive patients from Brazil with chronic CD and the following clinical manifestations: asymptomatic (n = 5), cardiac (n = 17), digestive (n = 2), and cardiodigestive (n = 7). Thirteen patients were infected with TcII. Group 8 (G8) included samples from 75 seropositive patients from Argentina with chronic CD and the following clinical manifestations: asymptomatic (n = 27), cardiac (n = 34), digestive (n = 1), and cardiodigestive (n = 13). Fifty-one patients were infected with TcV or TcVI or combinations of TcII, TcV, and TcVI.

In addition, samples from 50 persons from Argentina with negative serology for T. cruzi were included as negative controls to address the specificity of the procedures.

The blood samples were obtained and immediately mixed with an equal volume of GE (GEB). After 48 to 72 hours at room temperature GEB samples were boiled for 15 minutes (except for G3 and G5 groups and seronegative samples) and stored at 4°C for DNA extraction and PCR analysis.

GEB samples were processed with the High Pure PCR Template Preparation kit (Roche Diagnostics Corp., Indianapolis, IN) as described in Duffy et al.⁹ Those samples with cycle threshold (Ct) values lower than the Ct values for the most concentrated point of the standard curve were properly diluted in seronegative human blood treated with GE, and DNA extraction and qPCR procedures were repeated. To build the standard curves for quantification of parasitic loads, DNA from spiked blood samples were obtained in the same way as reported for the clinical samples. The DNA eluate was stored at −20°C until use in qPCR analysis.

Two duplex qPCR procedures were compared, Satellite DNA (SatDNA) qPCR and kinetoplastid DNA (kDNA) qPCR assays. The former targets the satellite sequence of the nuclear genome of the parasite and the sequence of an internal amplification control (IAC) as described in Duffy et al,⁹ and the latter is a modification of the method reported by Qvarnstrom et al¹¹ which targets the conserved region of the minicircle parasite sequences with the addition of primers and TaqMan probe for the IAC.⁹ Both reactions were performed with 5 μL of re-suspended DNA, using FastStart Universal Probe Master Mix (Roche Diagnostics GmbHCorp., Mannheim, Germany) in a final volume of 20 μL.

Optimal cycling conditions for both qPCR assays were a first step of 10 minutes at 95°C, followed by 40 cycles at 95°C for 15 seconds and 58°C for 1 minute. The amplifications were performed using Rotor-Gene 6000 (Corbett Life Science, Cambridgeshire, United Kingdom) or ABI7500 (Applied Biosystems, Foster City, CA) devices.

Standard curves were plotted with 1/10 serial dilutions of total DNA obtained from a GEB-seronegative sample spiked with 10⁵ parasite equivalents per milliliter of blood (par. eq./mL). TcId- and TcVI-DNA–based standard curves were used to quantify parasitic loads in G1a, G1b, G2, G3a, G4, and G5 and in G3b, G6, G7, and G8 samples, respectively.

A negative control and two positive controls that contained different concentrations of T. cruzi DNA (namely a high-positive control and a low-positive control near the limit of detection) were included in every run as recommended.⁶

A pZErO-2 recombinant plasmid that contains an inserted sequence of Arabidopsis thaliana aquaporin was used as a heterologous extrinsic IAC.⁸

To check for DNA integrity, clinical samples were tested with TaqMan RNase P Control Reagents Kit (Applied Biosystems) in an ABI7500 Real-Time PCR device.

SatDNA qPCR quantifiable samples collected from laboratories that did not perform T. cruzi DTU typing (G1, G3, and G6) were genotyped with PCR-based strategies targeted to nuclear genomic markers, as described in Burgos et al.²¹

The 15 samples provided by the laboratory of Disciplina de Parasitología, Universidade Federal do Triângulo Mineiro (Uberaba, Brazil), were evaluated by hemoculture. The assay was performed according to the method described in Chiari et al.²² Immediately after collection, 30 mL of blood was centrifuged at 4°C to remove plasma. The packed cells were washed by centrifugation at 4°C in liver infusion tryptose medium. The sedimented erythrocytes were resuspended in 30 mL of liver infusion tryptose medium and uniformly distributed in six test tubes. Cultures were maintained at 28°C and homogenized weekly. The culture was microscopically examined 30, 60, and 90 days after culture in 10-μL aliquots of suspension.

The Cohen κ coefficient²³ was used to compare the clinical sensitivity of SatDNA and kDNA qPCRs for the detection of T. cruzi DNA in samples from chronic CD patients. The unpaired t-test was used to compare the means of parasitic loads of quantifiable samples from acute versus chronic CD patients and asymptomatic versus symptomatic chronic CD patients for both qPCR methods. In addition, nonparametric analysis of variance was used to compare the parasitic loads of quantifiable samples grouped according to their T. cruzi genotypes for each qPCR assay. Bland-Altman bias plot⁶ was used to analyze the closeness of the agreement between the quantifiable results of both qPCR methods. Finally, the paired t-test was used to compare the means of IAC Ct values of both qPCR assays, and the Tukey criterion²⁴ was used to detect samples with outlier Ct values of IAC (Ct > 75th percentile + 1.5× interquartile distance of median Ct), which indicated PCR inhibition or material loss during sample DNA extractions.

The analytical validation results obtained for both qPCR assays are shown in Table 1. For the inclusivity, qPCR methods were assayed with DNA from strains that represented the six T. cruzi DTUs, plus TcI stocks representative of three different TcI SL-IR–based groups (TcIa, TcId, and TcIe).

kDNA qPCR gave the same analytical sensitivity of 0.0625 fg/μL DNA, the lowest concentration tested, for all T. cruzi stocks analyzed. SatDNA qPCR gave an analytical sensitivity of 0.0625 fg/μL DNA for stocks representative of TcIa, TcII, TcIII, TcV, and TcVI; 0.25 fg/μL DNA for stocks belonging to TcId and TcIV; and 1 fg/μL DNA for the TcIe stock.

Exclusivity was assayed in T. rangeli and Leishmania spp. Both qPCR methods were nondetectable when up to 1000 pg/μL DNA, the highest concentration tested, from Leishmania stocks was analyzed. In the case of T. rangeli, 10 fg/μL DNA could be amplified by kDNA qPCR, whereas SatDNA qPCR required an input of at least 10 pg/μL T. rangeli DNA to obtain a detectable PCR result.

The reportable range was determined with 10 spiked GEB samples that contained serial dilutions of TcId- and TcVI-cultured epimastigotes. Linear regression analysis yielded the equation y =1.013x − 0.058 (R² =0.992) and y =1.001x − 0.005 (R² = 0.998) for SatDNA qPCR, and y = 0.813x + 0.824 (R² =0.969) and y =1.011x − 0.048 (R² =0.984) for kDNA qPCR and for TcId and TcVI representative stocks, respectively. Accordingly, the reportable range was from 10⁵ to 1 par. eq./mL for TcId stock and from 10⁵ to 0.25 par. eq./mL for TcVI stock, for both qPCR methods.

The LOD was determined for TcVI DTU. It was 0.16 par. eq./mL (95% CI, 0.13–0.24 par. eq./mL) and 0.23 par. eq./mL (95% CI, 0.18–0.35 par. eq./mL) for kDNA qPCR in boiled and nonboiled samples, respectively (P = 0.013). For SatDNA qPCR the LOD was 0.46 par. eq./mL (95% CI, 0.36–0.64 par. eq./mL) and 0.70 par. eq./mL (95% CI, 0.54–1.01 par. eq./mL) in boiled and nonboiled samples, respectively (P = 0.044). The LOD of kDNA qPCR was lower than that of SatDNA qPCR for both boiled (P = 0.013) and nonboiled samples (P = 0.044).

Estimates of precision were calculated for nonboiled GEB samples spiked with TcVI stock. The precision was higher for kDNA qPCR (CV = 31.98%) than for SatDNA qPCR (CV = 46.60%) at concentrations closer to the LOD, but it was higher for SatDNA qPCR (CV = 6.00% and 1.72%) than for kDNA qPCR (CV = 8.79% and 2.92%) for concentrations of 10 and 1000 par. eq./mL, respectively.

The LOQ was derived from a 20% threshold value of the CVs obtained in the precision experiments. Linear least squares regression for the equation, y = y₀ + axe^−bx, resulted in the best fit (R² = 1.0) for both qPCRs. On the basis of the derived equation (y = 1.61 + 157.75xe^−1.81x and y =1.79 + 43.39xe^−0.91x), the absolute LOQ_20%CV was estimated in 1.53 par. eq./mL and 0.90 par. eq./mL for SatDNA and kDNA qPCRs, respectively.

The performance of both qPCR methods was compared with DNA obtained from 156 GEB samples that covered different epidemiologic and clinical settings. In addition, 50 GEB samples from persons with negative serology for T. cruzi were tested. No amplification was detected for any of these negative controls with the use of both SatDNA and kDNA qPCR methods (clinical specificity, 100%).

The results obtained for the different patients’ groups, including their geographic origin, routes of transmission, phase of CD, T. cruzi DTUs (only performed for samples with parasitic loads above SatDNA qPCR LOQ), qPCR positivity, and median parasitic loads, are summarized in Table 2.

The IAC amplification was used for quality control throughout the procedure, from the DNA extractions to qPCR assays. Higher Ct values of IAC were obtained for the clinical samples with the use of SatDNA qPCR (19.05; 18.53 to 19.41) than kDNA qPCR (18.90; 18.40 to 19.25) (P < 0.0001); threshold Cts for outlier values were 20.72 and 20.52, respectively. However, the same three samples with IAC Ct outliers (24.89, 21.00, and 21.49, and 24.74, 20.98, and 21.45) were identified for both SatDNA and kDNA qPCRs, respectively. These three samples were detectable for T. cruzi DNA by both qPCRs, but their parasitic loads were below the LOQ of the corresponding qPCR assay (Table 1); except for the one with the lowest Ct values of IAC (21.00 and 20.98) which presented a parasitic load of 0.94 par. eq./mL (0.97 Log₁₀ par. eq./10 mL) with the use of kDNA qPCR.

In addition, to evaluate DNA integrity of GEB samples, RNase P analysis was performed in a separate amplification reaction. All clinical samples were PCR detectable for this endogenous control with Ct values (22.32; 21.20 to 23.27) between cycle 19 and 27.

The parasitic loads of a subset of 15 GEB samples from Brazilian CD patients were compared with hemoculture results performed at the time the samples were collected (Table 4). Despite the time passed, an overall agreement was found between parasitic loads obtained by both qPCR methods and hemoculture results: the three samples with quantifiable parasitic loads corresponded to patients with five or six positive hemoculture results of a total of six replicates, whereas the three samples with nondetectable parasitic loads corresponded to patients with negative results by hemoculture assay, except one case with only one positive hemoculture result (Table 4).

This study presents the analytical validation and evaluation of two duplex qPCR methods on the basis of TaqMan probes designed for detection and quantification of T. cruzi DNA in human blood samples. The study was performed in the context of an international study organized in an attempt to establish standard operative procedures for quantification of parasitic loads in blood samples, using two methods ranked among the best ones by a previous qualitative PCR international study.⁷

This study involved the evaluation of DNA samples from parasite stocks representative of the different T. cruzi DTUs, including three distinct TcI SL-based groups. Analytical validation was performed with seronegative blood spiked with known numbers of cultured epimastigotes and treated with GE buffer. Furthermore, blood samples from diverse clinical settings of different geographic regions and harboring different parasite DTUs were assayed. The parasitic loads of these samples were compared with the same DNA extraction protocols, qPCR amplification procedures, master mixes, PCR thermocyclers, and quality controls in the same laboratory.

Analytical sensitivity was more uniform among the different T. cruzi DTUs for kDNA qPCR than for SatDNA qPCR, because this latter method was less sensitive for some TcI and TcIV strains, indicative of lower gene dosage in their genomes.^8,25 Thus, in practice it would be advisable to construct standard qPCR curves with the use of regional strains representative of the prevailing DTUs in the affected population.

As expected, parasitic loads in acute CD samples were higher than in chronic CD cases, reaching concentrations of up to 8 Log₁₀ par. eq./10 mL for both qPCR methods. The four samples with the highest parasitic loads belonged to Mexican and Bolivian acute CD patients infected with TcI and TcIV, respectively. Accurate quantification of these samples was achieved after diluting them 1:10,000 in seronegative human GEB before doing the final DNA extractions and qPCR amplifications. One of the two acute CD samples from French Guiana belonged to a seronegative patient with suspicion of orally acquired T. cruzi infection on the basis of clinical and epidemiologic findings.²⁶ The qPCR positivity of this case points to the usefulness of molecular diagnostic methods for detecting acute CD cases before seroconversion.

Parasitic loads in chronic CD patients ranged from nonquantifiable to values of 3.67 and 4.84 Log₁₀ par. eq./10 mL with the use of SatDNA and kDNA qPCRs, respectively. We observed differences in the parasitic loads between chronic CD patients infected with TcI (eight asymptomatic persons from Colombia plus one from Mexico) and TcII (11 symptomatic patients from Brazil), when they were analyzed with kDNA qPCR. This finding is in agreement with previous observations obtained by Moreira et al²⁷ with the use of a Sybr Green SatDNA qPCR assay in samples from Colombian and Brazilian symptomatic chronic CD patients recruited for the BENEFIT (Benznidazole Evaluation for Interrupting Trypanosomiasis) trial. Nevertheless, no differences were observed between the parasitic loads of asymptomatic and symptomatic chronic CD patients for the samples tested in our study. This is in agreement with previous observations that showed no correlation between T. cruzi parasitemia and clinical manifestations.^28,29 However, studies in the murine model reported that T. cruzi reinfections leading to an increase of parasitemia could be related to the variability and severity of the clinical course of CD.³⁰

Bland-Altman analysis was performed to summarize the agreement between parasitic loads obtained by both qPCR methods by calculating the bias and by estimating the mean difference and the SD of the differences. It is also common to determine the limits of agreement, which are by convention set at the 95% CI of the difference between the methods, usually specified as bias ± 2 SD. If the 95% CI for the mean difference includes zero, such as in our study, there is statistically no evidence of bias.⁶

Some samples presented discordant qPCR results by either of the methods tested; kDNA qPCR detected more samples than SatDNA qPCR, which is reasonable because the former method had higher analytical sensitivity (lower LOD). Moreover, in most of the discordant samples the parasitic load was below the LOD, and in all cases below the LOQ, of the corresponding qPCR assay that gave detectable results.

In this study, correlation between parasitic loads and frequency of positive hemocultures was also observed, strengthening the notion that detectable PCR results are indicative of live parasites.³¹

Previous studies for screening and quantification of parasitic loads of T. cruzi with the use of different real-time PCR approaches included a host DNA sequence, such as RNase P human gene, as an internal control.^8,10,27 This type of endogenous control is useful for qualitative purposes, as in the validation of the use of archival samples. In the present study, for instance, the RNase P assay was useful for checking the DNA integrity of GEB samples stored for >10 years. Nevertheless, we do not recommend the use of endogenous controls for quantitative purposes because the content of human blood cells can be highly variable, depending on the nutritional, metabolic, and immunologic status of the persons.⁹ Therefore, a normalized amount of DNA of a plasmid that contains a heterologous sequence was added before DNA extractions and was used as an internal amplification control to monitor the whole procedure. Indeed, in our study, the IAC was useful to detect three samples with outlier Ct values.

Comparison of the analytical parameters for both qPCR methods suggests that kDNA qPCR possesses higher sensitivity for detection and quantification of samples with low parasitic loads. However, in some Central and South American countries, such as Venezuela, Guatemala, Panama, Colombia, El Salvador, and some regions of Brazil, where T. rangeli might cause a false-positive diagnosis of T. cruzi infection,³² SatDNA and not kDNA qPCR should be the qPCR-based method of choice (Table 1). Furthermore, qualitative and quantitative SatDNA real-time PCR approaches were recently used in clinical trials with new antiparasitic drugs and proved to be useful to detect treatment failure.^33,34

Clinical sensitivities of both qPCR methods when tested in samples from chronic CD patients (80.69% and 84.14% for SatDNA and kDNA qPCRs, respectively) were similar to those obtained by the four best performing methods selected in a previous international PCR study for T. cruzi detection.⁷ However, these sensitivities are not good enough for the application of PCR as confirmatory testing of blood donors or clinic patients who are serologically positive. Future prospective studies must be conducted to determine the optimal real-time PCR-based algorithm for diagnosis of T. cruzi infections in other epidemiologic and/or clinical scenarios, such as in early detection of congenital or oral transmission, and reactivation of infection in immunocompromised patients due to organ transplantation or HIV coinfection. Finally, the availability of these standardized and validated qPCR methods opens up new possibilities to monitoring patients in clinical trials with trypanocidal drugs,^27,33,34 contributing to improve the quality of life of CD patients.