Evaluation of the Performance of the Loopamp Trypanosoma cruzi Detection Kit for the Diagnosis of Chagas Disease in an Area Where It Is Not Endemic, Spain

In Spain, PCR is the tool of choice for the diagnosis of congenital Chagas disease (CD) and serology for diagnosing chronic CD. A loop-mediated isothermal amplification test for Trypanosoma cruzi DNA detection showed good analytical performance and ease of use.

as CD is not included in the category of notifiable diseases except in the region of Catalonia (3). Records from the National Centre for Microbiology (CNM, Spanish acronym) from 1997 to the present show more than 6,000 infected individuals, of which 51 were congenital cases, 5 were transfusion-associated cases, and 2 were due to transmission through organ transplant (CNM data, unpublished).
The serological screening of at-risk blood donors has been mandatory since 2005 (4). Also, half of the Spanish regional health authorities recommend serological screening in pregnant women and their newborns, with the regions of Catalonia, Galicia, and Valencia pioneering the inclusion of a diagnostic algorithm for congenital CD in their official guidelines (3,(5)(6)(7). In the rest of Spain, serological screening in pregnant women and at-risk migrants is also conducted in the main public hospitals at the discretion of their health professionals.
Most of the CD cases in settings of nonendemicity are in the chronic phase, when the parasite burden is usually low and intermittent (8)(9)(10). Therefore, suspected chronic CD is usually confirmed by serological tests (11). Since seroreversion to negative following successful treatment of a chronic CD case may take years, serology is still recommended for treatment follow-up (12). Although PCR is the best option for early detection of circulating parasites, in chronic patients, it is mainly used as a marker of therapeutic failure (8,9).
In Spain, the acute phase is observed mainly in newborns and is characterized by high degree of parasitemia, making diagnosis by parasitological tests feasible (11). Unfortunately, the sensitivity of parasitological tests is suboptimal, and infants who are diagnosed by serology 9 to 10 months after birth, when maternal antibodies disappear (13)(14)(15), are often missed. In Europe, in-house or commercial PCR tests are routinely used to diagnose congenital CD cases (16)(17)(18). Real-time PCRs targeting satellite DNA (SatDNA) or kinetoplast (kDNA) minicircles are commonly used (9,19,20).
Years ago, Eiken Chemical Co. Ltd., Japan, developed a simple approach for T. cruzi DNA detection based on the loop-mediated isothermal amplification test (Tcruzi-LAMP). This test amplifies the T. cruzi SatDNA in less than 1 h, under isothermal conditions, without specific instrumentation, and interpretation of results is quite straightforward (21). Besuschio et al. reported good analytical sensitivity and specificity for this test and showed its potential usefulness in a small number of clinical samples (21). Based on this, we aimed to evaluate the sensitivity, specificity, and accuracy of Tcruzi-LAMP using well-characterized samples from confirmed congenital and chronic CD cases born and living in Spain. We also estimated the agreement between Tcruzi-LAMP and PCR tests targeting kDNA and SatDNA.

MATERIALS AND METHODS
Study design. The evaluation was a retrospective study using convenience sampling from the clinical specimen collections of the CNM, Instituto de Salud Carlos III (ISCIII), Madrid, Pharmacy and Food Sciences Faculty, Universitat de Barcelona (PFSF-UB), Barcelona, and Hospital de la Santa Creu i Sant Pau (HSCSP), Barcelona. The time frame of sample and data collection was between 2003 and 2016 (Table 1), and the convenience sampling was performed considering eligibility criteria (see below).
The study was designed as a case-control model for evaluation of Tcruzi-LAMP in congenital infection; for each positive congenital case, we included one uninfected baby. The chronic samples were Samples. (i) Eligibility criteria. DNA and guanidine-EDTA-blood (GEB) samples were selected from the above-mentioned collections based on (i) previous results of laboratory routine diagnostic tests epidemiological and clinical background (see Data set S1 in the supplemental material) and (ii) availability of enough volume to run index (Tcruzi-LAMP) and PCR tests.
The epidemiological data considered related to CD were the country of birth, age, stays or trips to areas of endemicity, mother from an area of endemicity, and risk exposure to T. cruzi (blood transfusion, transplantation, laboratory accident). We also considered the following clinical data: presence of cardiac or digestive alterations or any other symptom related to CD. The affected population residing in Spain is generally asymptomatic; thus, epidemiological data had greater relevance for the inclusion of cases. Raw epidemiological and clinical data are included in Data set S1.
(ii) Congenital CD cases. A total of 39 samples from infected children born in Spain from women with CD were included. The diagnosis of congenital CD had been confirmed following the guidelines in Madrid and Catalonia, i.e., (i) before 9 months after birth, a positive result by parasitological tests (direct microscopy, microhematocrit, or culture) and/or PCR or (ii) after 9 months of age by positive serology, plus no travel to areas of endemicity from birth to time of diagnosis (3,7,22,23).
(iii) Uninfected children. This group included 48 samples from babies born to T. cruzi infected mothers for whom congenital CD had been ruled out by negative parasitological and/or molecular tests before 9 months of age and negative serology from this point on (Table S1 and Data set 1 in supplemental material).
(iv) Chronic CD cases. We selected 174 samples from patients with chronic CD who had a positive result in at least two different serological tests (Table S2 and Data set S1).
(v) Nonchagasic individuals. A total of 34 samples from individuals at risk for T. cruzi infection for whom CD had been ruled out by negative serology (Table S2 and Data set S1) were included.
(vi) Positive controls for molecular tests. DNA was obtained from T. cruzi cultures of discrete typing units TcV and TcI using the High Pure PCR template preparation kit (Roche Diagnostics, Germany) following the tissue protocol. The DNA concentration was measured with a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, USA). For the positive controls, the DNA concentration was adjusted to 10 and 1 parasites/ml. We assumed 100 fg DNA as the equivalent of one parasite (24). Standard curves for quantitation were made of 10-fold serial dilutions of T. cruzi TcV genomic DNA, spiking genomic DNA from uninfected individuals.
Test methods. (i) Loopamp Trypanosoma cruzi detection kit (Tcruzi-LAMP, index test). Tcruzi-LAMP was performed in a single test following the manufacturer's instructions using a portable realtime fluorimeter (Genie III, OptiGene, UK).
DNA from GEB samples was purified using a High Pure PCR template preparation kit (Roche Diagnostics, Germany) following the recommendations of Ramírez et al. (25). The starting volume of GEB was 300 ml, and the volume of elution was 100 ml; 5 ml of DNA per reaction was tested. Amplification conditions were the first step of 5 min at 95°C, followed by 40 min at 65°C. Reading of the results was by visual examination immediately after the end of the amplification process; no color change was recorded as a negative result, and a color change, as a positive result. Results provided by the fluorimeter were displayed at the end of the reaction and recorded as the time to positivity in minutes and seconds (for more details, see Fig. S1).
Two operators were trained in the management of biological samples, DNA extraction, and molecular tests 2 weeks before starting the study; they did not participate in the selection of the samples, but performed the DNA extraction and Tcruzi-LAMP blinded to previous results.
(ii) Reference tests. (a) Parasitological tests. Parasitological confirmation was based on direct observation of trypomastigotes in a blood sample, either after a fresh examination or with a microhematocrit technique (MHT) (26). Alternatively, 100 ml of the blood sample was inoculated in Novy-MacNeal-Nicolle (NNN) culture medium supplemented with liquid medium liver infusion tryptose, 10% fetal bovine serum, and antibiotics; the supernatant of the culture was examined at 15 and 30 days postinoculation. These tests were performed at the time of sample collection.
(c) Conventional kDNA PCR (kDNA-PCR). This PCR targets a variable sequence of kDNA minicircles. DNA purification was performed according to the standard CNM procedure (10). The starting volume of GEB samples was 300 ml, and DNA was dissolved in 75 ml of molecular-grade water. Samples and positive and negative controls were processed in duplicate. Amplification was routinely performed in a total volume of 75 ml, using 10 ml of DNA in each duplicate. Positive and negative controls were included in duplicate for each amplification run. PCR products of 330 bp were analyzed by electrophoresis on a 2% agarose gel and using GelRed nucleic acid gel stain (Biotum, USA) as the dye. For this study, the kDNA-PCR was rerun on all samples at the same time as Tcruzi-LAMP.

Evaluation of LAMP for Chagas Disease Diagnosis
Journal of Clinical Microbiology Data analysis. The composite gold standard according to age of diagnosis was used for sample classification as case or control: parasite detection by any parasitological or molecular test confirmed the T. cruzi infection before 9 months of age, after which the diagnosis was based on serology ( Fig. 1). In this context, sensitivity, specificity, accuracy, and their 95% confidence interval (CI) of Tcruzi-LAMP were estimated by binomial distribution (http://statpages.info/confint.html). Differences regarding the composite gold standard were analyzed with McNemar's test. kDNA-PCR or Sat-qPCR after 9 months of age are mainly used to determine the presence of parasites; thus, the agreement between molecular tests was measured with Cohen's kappa statistic (κ) and the Spearman rank correlation (r s ). The Mann-Whitney U test was used to analyze differences among time of positivity and degree of parasitemia (continuous responses) regarding the congenital and chronic CD groups (categorical factors). To explore the influence of degree of parasitemia (continuous response) in the measurement of agreement, the overall data were stratified according to the different combinations of test results for all the molecular tests. This meant that the positive cases were classified in two groups: (i) samples with positive results in all molecular tests, concordant group, and (ii) samples with a positive result in any test, discordant group (categorical factors). Differences were determined with the Mann-Whitney U test. In Sat-qPCR, not quantifiable values were considered equivalent to the limit of detection. Analyses were performed using Minitab 18.1.

RESULTS
A total of 295 samples from an population at risk of being infected with T. cruzi were tested by Tcruzi-LAMP (Fig. 2).
Congenital Chagas disease. At CNM, parasites were seen in 21 cases out of 22 for whom a blood sample was examined by direct microscopy; for the remaining case, parasite was detected by culture ( Table 2). Tcruzi-LAMP and PCR tests yielded positive results in all congenital CD cases. In samples from children previously testing negative for T. cruzi infection (n = 25), Tcruzi-LAMP returned a positive result in four infants when the reaction was read by visual examination, three of which were also positive by fluorimeter. All of them were negative by PCR tests.
At PFSF-UB/HSCSP, only one case, 5 months old, was positive by MHT (1/6). All congenital CD cases were positive by PCR tests (10/10), and Tcruzi-LAMP yielded a positive result in 9 of them (9/10). The undetected case was a child older than 9 months of age (case 38; Table 2). For uninfected children, Tcruzi-LAMP and kDNA-PCR yielded negative results in all of them (23/23), whereas Sat-qPCR returned a positive result in 1 infant (Table S1 summarizes the results that rule out T. cruzi infection in uninfected children).
When comparing the global positivity of parasitological and molecular tests (Fig.  3A) among 12 infected children diagnosed from 0 to 1 month after birth, 83.3% (10/12) had circulating parasites in the bloodstream. From 1 to 9 months after birth, 92.3% (12/ 13) of cases were positive by parasitological tests. For infants older than 9 months of age, only 1 (1/3) was positive for parasite detection. Both Tcruzi-LAMP and PCR tests detected 100% (29/29) of infected children younger than 9 months of age. Among infants older than 9 months found positive by serological and PCR tests (10/10), the 1 case not detected by Tcruzi-LAMP (1/10) had a low degree of parasitemia, while most of the infected children (25/39) showed levels of parasitemias higher than 40 parasites/ ml (Tables 3 and 5).
Unlike what is observed in congenital infection, Tcruzi-LAMP returned a positive result before 20 min, just in 59.8% of chronic CD cases, and in 78.1% of cases, the level of parasitemia was lower than 40 parasites/ml (Table 3; P , 0.001).
Sensitivity, specificity, and accuracy. Tcruzi-LAMP for congenital CD diagnosis displayed a sensitivity of 97.4% for both types of readings and 100% for infected children younger than 9 months of age (Table 4), while, depending on the type of reading, specificity was 91.7% by visual examination and 93.8% by fluorimeter. Subsequently, the accuracy for both types of reading and situations was similar, between 94.2% and 96.1%. Differences in performance of Tcruzi-LAMP were not statistically significant (P . 0.05).
In chronic CD diagnosis, sensitivity for Tcruzi-LAMP was low (47.1%), but specificity was 100%. As a consequence, accuracy was 55.8%. Performance of Tcruzi-LAMP was similar to that of Sat-qPCR (P = 0.332) but different from that of kDNA-PCR (P , 0.001).
Agreement analysis. The frequency of concordant results between molecular tests was high for congenital CD diagnosis, 97.4% in positive cases and 89.6% in uninfected children (Table 5); thus, the agreement was almost perfect (κ of 0.89 and 95% confidence interval [CI] of 0.79 to 0.98, to κ of 0.91 and 95% CI of 0.82 to 1.00; Table S3).
For chronic CD diagnosis, the frequency of concordant results in nonchagasic individuals was 100%, whereas in chagasic cases it was 77% (κ of 0.67 and 95% CI of 0.58 to 0.77, to κ of 0.83 and 95% CI of 0.75 to 0.91, substantial agreement; Table S3).
Comparing the time to positivity of Tcruzi-LAMP with the cycle threshold (C T ) values of Sat-qPCR, the correlation was strong in congenital infection (r s , 0.653; P , 0.001) and moderate in chronic CD (r s , 0.458; P , 0.001).

DISCUSSION
Our data provide substantial evidence on the usefulness of Tcruzi-LAMP and its role as a molecular test in the diagnosis of congenital CD in settings of nonendemicity. As the prevalence of congenital T. cruzi infection in countries of nonendemicity is quite low, our results are relevant since we include a cohort with a high number of cases (n = 39).
According to the Pan American Health Organization and the World Health Organization, direct parasitological techniques (MHT and direct observation) remain the gold standard for parasite detection in the diagnosis of congenital CD. Should these tests be negative, or if no   tests were done before, detection of T. cruzi-specific antibodies in infants from 8 to 10 months old becomes the gold standard (31,32). Although MHT is the fastest method to detect T. cruzi (less than 20 min), its sensitivity depends on the operator's skills, the degree of parasitemia, and whether the sample is examined within 12 h after collection (13,26,33). Reports from some South American countries showed that MHT sensitivity was 34.2% in the first month of age (34), 76.1% up to 6 months of age (15), and 94% in 1 year of follow-up (14), whereas qPCR could detect up to 84.2% of infected infants before 1 month of age (34). Others, using conventional PCR, achieved 100% sensitivity in infected children who were diagnosed before 6 months of age (35). Due to losses to follow-up, there is not enough PCR data for the first year of life in children born to T. cruzi-infected mothers.
In Spain, both parasitological tests and PCR are used without distinction for congenital CD diagnosis. Recently, Basile et al. reported that among 22 congenital CD cases born in Catalonia, in 1 year of follow-up, 5 were diagnosed by MHT and 10 by qPCR (36), while in Murcia, among 12 infected infants, 1 was positive by MHT, 8 by culture, and 100% by conventional PCR, also in 1 year of follow-up (18).
There is less information on the use of LAMP for CD diagnosis. So far, three studies were published in populations in areas of endemicity. The first targeted the 18S rRNA gene (37), and the other two, a prototype, targeted SatDNA (21,38). The first two showed the potential of LAMP in a limited number of samples (seven and five cases, respectively), while the third study showed good sensitivity detecting 10 cases of congenital CD.
In our study, Tcruzi-LAMP and PCR tests detected 100% (29/29) of cases younger than 9 months, while parasitological tests detected just 88% (22/25) of them (Table 1). In infants older than 9 months, Tcruzi-LAMP was negative in one case (1/10) on a stored DNA sample that could have degraded even under the best storage conditions, considering that this sample was collected in February 2008 and stored at 240°C. This sample was positive by kDNA-PCR and Sat-qPCR tested at the same time. This result may reflect a slightly lower analytical sensitivity for Tcruzi-LAMP. Certainly, it is well known that the number of copies of SatDNA in TcI, TcIII, and TcIV is lower than that in TcII, TcV, and TcVI (24,25,29,39). Thus, a low degree of parasitemia could be the cause of a lower sensitivity of techniques targeting SatDNA, such as Tcruzi-LAMP or Sat-qPCR (40). Regarding specificity, four uninfected infants were found positive by Tcruzi-LAMP and one by Sat-qPCR, yet all had negative kDNA-PCR and serology after 9 months of age (Table S1). This may be explained by contamination at any step of the procedure. False positives due to contamination can happen in any laboratory, even when using automated systems (34), but this is easy to trace and control with good laboratory practices (41). For this study, new DNA extractions and Tcruzi-LAMP were performed by two new laboratory technicians, and the controls (tested in parallel) yielded the expected results. In our experience, the best way to detect false positives due to contamination is to examine the sample in duplicate after DNA extraction and to retest when a positive result is observed in the later steps of amplification. Due to the scarcity of available biological material, we were not able to repeat the DNA extraction to confirm whether these false-positive results were due to contamination during that process. In experiments not shown, it was observed that guanidine produced a color change similar to that observed in a positive sample, but this change was not detected by the fluorimeter. That is, if the guanidine had not been correctly eliminated during DNA extraction, we will have false positives by visual examination but not by fluorimeter. In our study, one of the false positives could have been caused by the presence of guanidine traces. In the other three samples, we cannot discard nonspecific amplification, though.
In terms of cost, the implementation of Tcruzi-LAMP could be less expensive than PCR (31). Comparing ready-to-use techniques, Tcruzi-LAMP is the only low-cost test (Tcruzi-LAMP e7.5 versus commercial qPCR e20 per reaction) giving results faster than the current qPCRs and that can be performed without needing sophisticated instrumentation. The additional advantage of using a real-time isothermal fluorimeter is that the reading eliminates any subjectivity, and time to positivity may allow for estimating the degree of parasitemia. The correlation between the time to positivity in Tcruzi-LAMP and the C T values of qPCRs was significant in samples from congenital CD, but further research on this observation is required.
The main limitation of our study is the relatively low number of samples of uninfected individuals to estimate specificity. As Tcruzi-LAMP is based on SatDNA, we expected low sensitivity; therefore, in the study design, we tended to include a high number of positive cases. In experiments not shown, we tested samples from malaria and leishmaniasis patients with all molecular T. cruzi tests (n = 20), and all samples tested negative. Considering these data and results of all uninfected individuals (48 samples from uninfected children and 34 samples from nonchagasic individuals), the overall specificity for Tcruzi-LAMP by visual examination was 96.1% (95% CI, 90.3 to 98.9%) and by fluorimeter was 97.1% (95% CI, 91.6 to 99.4%). New studies will improve these estimates.
Although this study was retrospective and based on convenience and available sampling, the results clearly support the usefulness of the Loopamp Trypanosoma cruzi detection kit in the early diagnosis of congenital T. cruzi infection. Tcruzi-LAMP showed a suitable performance, and running in duplicate, it could be used as a screening test before 9 months of age and afterward as an alternative or complement to current diagnostic tools. In the chronic phase, its role would be similar to that of the Sat-qPCR.