Comparison between Two Amplification Sets for Molecular Diagnosis of Toxoplasmosis by Real-Time PCR

ABSTRACT PCR is now commonly applied to the diagnosis of toxoplasmosis. Although several methods are available, comparative studies are few, making it difficult to compare the performance of each technique. We compared the sensitivities of two real-time PCR assays through a prospective study on fetuses, neonates, and immunocompromised patients and on the ocular diagnosis of toxoplasmosis. The first system targeted the widely used B1 gene (GenBank accession number AF179871) while the second (RE) targeted a more recently described sequence repeated roughly 200 to 300 times (GenBank accession number AF146527). We demonstrated that molecular diagnosis requires the duplication of PCR assays, especially with the B1 system, as only one PCR was positive in 33.3% of cases. Our study showed that the RE target was more sensitive for all biological samples (amniotic fluid, placenta, aqueous humor, whole blood, and cerebrospinal and bronchoalveolar fluids) and significantly improved the performance of the diagnosis of toxoplasmosis. Taking into consideration all clinical samples, the mean gain in the crossing point value was 4.2 ± 1.7 cycles and was even more significant for amniotic fluid (5.8 ± 1.7 cycles).

Toxoplasma gondii is a parasitic protozoan that is responsible for generally benign infections except when the disease occurs in pregnant women or immunocompromised individuals, such as human immunodeficiency virus-positive or grafted patients. PCR has clearly improved the diagnosis of toxoplasmosis and is today an inescapable technique for revealing the presence of the parasite in clinical specimens. However, no commercial kit is yet available for this application. As a consequence, each laboratory uses its own method and a great heterogeneity exists between laboratories, mainly concerning the choice of primers, molecular targets, and specific corresponding probes. Very few studies comparing the performances of the various protocols have been published (1,5,7,10,16,17,21).
Although the use of real-time PCR has grown considerably over the last few years, publications reporting comparative results between different targets are scarce (5,12,21). Moreover, the lack of a Toxoplasma gondii DNA reference prevents interlaboratory studies. However, such studies are essential to determine the most sensitive systems, as it is generally accepted that fetuses or immunocompromised patients with toxoplasmosis must be treated as early as possible. Several studies involving allogeneic stem cells or solid organ transplantation show that a sensitive PCR technique and an early diagnosis are crucial factors for outcome of the disease, especially when the immunological diagnosis remains nonconclusive (4,15,18).
In this prospective study, we compared the performances of two pairs of primers and probes for real-time PCR using fluorescence resonance energy transfer (FRET) on a Roche LightCycler (LC). The first system targeted the B1 gene (6,13), repeated approximately 30 times. This target has been used in our laboratory since 1993 for routine diagnosis. The second system targeted a more recently described repeated element (14) named the RE sequence. This sequence is more repetitive than the B1 gene, approximately 200 to 300 times, and is highly conserved (21). This region of the Toxoplasma gondii genome has been reported to be a very specific and sensitive target for toxoplasmosis diagnosis (14,21).
This study was performed during the routine molecular diagnosis of toxoplasmosis. It includes prenatal and neonatal diagnosis, ocular toxoplasmosis, and diagnosis in immunocompromised patients, such as human immunodeficiency virusseropositive or transplanted patients.

MATERIALS AND METHODS
Clinical samples. Between June and October 2003, all samples received by our laboratory for suspicion of toxoplasmosis were tested simultaneously during routine diagnosis using both the B1 and RE systems. From October 2003 to December 2004, this prospective study was continued for 1 year on amniotic fluids only, leading to a total of 136 samples. During this second period, among nonamniotic samples, only those presenting a positive result with the above-cited RE sequence were compared using both methods. We therefore included 152 fresh clinical specimens from patients suspected of T. gondii infection in the study: 52 placenta, 74 amniotic fluid, six cerebrospinal fluid, five aqueous humor, five bronchoalveolar lavage, nine whole blood, and one pulmonary biopsy samples.
Toxoplasma gondii isolates. DNA was extracted from the RH reference strain of T. gondii. The strain was maintained in female Swiss Webster mice by intraperitoneal passage of 5 ϫ 10 5 tachyzoites twice a week, as previously described (2). Parasites were harvested from peritoneal exudates. Tachyzoites were then counted in a Thoma cell and adjusted to 2 ϫ 10 7 parasites per ml in 0.9% NaCl. Tenfold dilutions of this suspension with 1 to 10 5 toxoplasmas/ml were used for quantitative assays.
DNA extraction. Extraction from amniotic fluid involved the centrifugation of two volumes of 4 and 7 ml at 900 ϫ g for 20 min. A 200-l pellet was extracted using the High Pure PCR template preparation kit (Roche Molecular Biochemicals), following the manufacturer's instructions, with treatment for elimination of PCR inhibitors included. DNA was eluted with 200 l of elution buffer.
For bronchoalveolar lavage, a volume of 1.5 ml was centrifuged under the same conditions and then DNA was extracted from the pellet following the same procedure.
Due to the very limited volume of aqueous humor available (about 20 l), DNA was extracted after the addition of 50 l of buffer containing 10 mM sodium hydroxide, 0.5% Tween 20, and 0.5% Nonidet P-40 and then heated for 10 min at 95°C (13). Samples were then centrifuged at 14,200 ϫ g for 10 min. PCR was performed on the supernatant.
DNA extraction from cerebrospinal fluid, from an initial volume of 500 l, followed the same procedure as for aqueous humor (n ϭ 3) or bronchoalveolar lavage (n ϭ 3) samples.
As previously described (20), whole blood was collected with EDTA and DNA was extracted after leukocyte separation; 1 ml of whole blood was mixed with 1 ml of a 2% dextran solution (Pharmacia). After red blood cell sedimentation at 37°C for 30 min, the supernatant (plasma, leukocytes, and tachyzoites when present) was collected and centrifuged for 10 min at 1,000 ϫ g. The cellular pellet was lysed for 2 h at 56°C with 200 l of a mix containing 10 mM Tris-HCl, pH 8.5, 50 mM KCl, 2.5 mM MgCl 2 , 0.9% Tween 20, and 0.8% proteinase K. Classical phenol-chloroform extraction and cold ethanol precipitation were then performed (22).
The placentas and the biopsy sample were lysed in the same conditions using 200 l of the proteinase K lysis buffer and submitted to phenol-chloroform extraction. Placenta DNA was extracted from 100 l of crushed sample that was initially prepared for inoculation in mice (3,11).
Genomic DNA was extracted from 200 l of each dilution of the RH strain using the High Pure PCR template preparation kit and 200 l of elution buffer.
Real-time PCR. Real-time PCR was performed on an LC in a final volume of 20 l using the DNA master hybridization probes kit (Roche Molecular Biochemicals) with an MgCl 2 concentration adjusted to 5 mM; 5 l of DNA was used for the PCR assay. Each sample was tested in duplicate for each method, except for amniotic fluids. In these cases, the assay was not a duplicate but two PCRs were performed on initial volumes of 4 and 7 ml.
The nucleotide sequences of the primers and probes for the LC-PCR assay targeting the B1 gene (4,8,9,19,23) and the RE repeated element are listed in Table 1. The two sets of hybridization probes recognized adjacent sequences within a 126-bp and a 133-bp amplified fragment on the B1 gene and the RE sequence, respectively. Concerning this last sequence, the amplified fragment was shorter and internal compared to the one described by Reischl et al. (21) or Homan et al. (14). Primers were purchased from Eurogentec and were used at a final concentration of 0.5 M. The concentrations of the fluorescein-labeled probes and LC-Red 640-labeled probes (TibMolBiol, Germany) were 0.1 M and 0.5 M, respectively.
Cycling began with 2 min of incubation at 50°C to allow the activation of the uracil-N-glycosylase (Eurogentec), followed by a 10-minute step at 95°C to inactivate uracil-N-glycosylase and denature the DNA. Then 50 cycles of amplification were performed as follows: 95°C for 5 s, 60°C for 10 s, and 72°C for 15 s (ramp rate, 20°C/s). Each PCR run included a control without DNA (containing the reaction mix alone), a control without toxoplasmic DNA (genomic DNA from a nonimmunized individual), and low (one parasite per reaction) and high (100 parasites per reaction) positive controls obtained from mouse peritoneal fluids infected with the RH strain. Results are expressed as the crossing point (Cp), i.e., the number of cycles at which a significant signal is observed in a given positive sample.
For each technique, an assay was considered positive if at least one test of the duplicate was positive.
Amplification of the human beta-globin gene was performed to assess the correct progress of DNA extraction and to underline the presence of PCR inhibitors. Qualitative analysis. The study included 136 clinical samples that were systematically and simultaneously analyzed in the B1 and RE systems. Qualitative results, expressed as the presence or absence of toxoplasmic DNA, were compared for the two systems.
Comparative sensitivity between the B1 and RE systems. Analysis of the Cp values was performed using the Student test to compare positive clinical samples obtained with the B1 and RE systems. A Wilcoxon test was used to compare Cp values obtained from amniotic fluid with 4-and 7-ml volumes and for serial dilutions of the RH strain.

Qualitative analysis of clinical samples in routine diagnosis.
Qualitative results obtained using B1 and the RE sequence for the detection of T. gondii were compared for the 136 clinical samples analyzed during the first part of the study. The details of the results are presented in Table 2. The results obtained using both methods are in agreement in 134 cases (98.5%): 119 negative and 15 positive samples. Among the latter, both duplicates were positive in the two amplification systems for 11 cases. In four samples (two placentas and two amniotic fluids), a PCR was performed in duplicate for each amplification system for all specimens received by our laboratory and suspected of toxoplasmosis. "Other" clinical samples included whole blood (n ϭ 8), bronchoalveolar lavage (n ϭ 5), aqueous humor (n ϭ 3), cerebrospinal fluid (n ϭ 3), and pulmonary biopsy (n ϭ 1).
b Two bronchoalveolar lavage fluid specimens, one aqueous humor specimen. c Whole blood.
VOL. 44, 2006 COMPARATIVE REAL-TIME PCR FOR T. GONDII 721 the two PCR assays were positive with the RE amplification system, whereas only one of the two assays was positive in PCR targeted with the B1 gene (26.7%). For two samples (one placenta, one whole blood), results were discordant with positive duplicates for the RE system and negative duplicates for the B1 system. No samples were found to be positive with the B1 gene amplification and negative in the RE LC assay. Comparative sensitivity between B1 and RE systems. In order to assess with greater precision the difference in sensitivity between the B1 and RE systems, we compared the crossing point values obtained for 10-fold dilutions of T. gondii genomic DNA of the RH strain. Amplification efficiencies were identical and close to 100% for both protocols. We obtained a 4.12 Ϯ 0.34 cycle gain using the RE LC assay, calculated with DNA concentrations equivalent to from 1 to 10 5 parasites per reaction (details in Table 3). The Cp value for DNA equivalent to 0.1 parasite was not included in the calculation of Cp delta as it constitutes the limit of detection and therefore was more variable and not consistently positive.
The greater sensitivity of the RE-LC assay was confirmed in positive clinical samples (n ϭ 33), including amniotic fluid (n ϭ 8), placenta (n ϭ 15), aqueous humor (n ϭ 3), bronchoalveolar lavage fluid (n ϭ 2), whole blood (n ϭ 2), and cerebrospinal fluid (n ϭ 3), with a gain in Cp of 4.23 Ϯ 1.69 cycles (P Ͻ 0.05) compared to the B1 gene amplification system (Table 4). Considering that PCR efficiencies were high and identical for both methods, an average gap of 4.1 cycles was equivalent to a more than 10-fold increase in sensitivity for the RE system, estimated for the RH strain as well as for clinical samples. Moreover, during the second stage of the study, we found one more placenta sample for which both duplicates were negative using the B1 gene and positive using the RE sequence, giving a total of three discordant results out of 33 positive samples (9.1%). Six more samples (four placenta, one aqueous humor, and one cerebrospinal fluid) were found to be positive in one of the duplicate assays with the B1 system when the two assays were positive with the RE-LC assay. One sample was positive for only one of the duplicates in each method.
Finally, we compared the results obtained on amniotic fluid (n ϭ 8) using 4-and 7-ml initial volumes of fluid (data not shown). With the B1 system, 6 samples were positive for both 4-ml and 7-ml volumes. Two samples were positive for only one volume (one for a 4-ml volume and one for a 7-ml volume).
The difference in Cp value between the 4-and 7-ml samples using the B1 gene amplification system was not significant (0.17 Ϯ 1.62 cycles). All amniotic fluid samples were positive for both 4-and 7-ml volumes with the RE system, providing a significant gain of 1.03 Ϯ 0.72 cycles with the latter (P Ͻ 0.05).

DISCUSSION
Over the course of the entire study, we showed that 33.3% (11 of 33) of the positive samples gave only one positive signal in the B1 amplification LC assay. Among these, only one gave a single signal in the RE LC assay. These results justify the fact that PCR assays should be carried out in duplicate for the diagnosis of toxoplasmosis. Nevertheless, the final interpretations of the test were identical for both methods as a single positive amplification (in a duplicate assay) was sufficient to give a positive diagnosis for toxoplasmosis. All these samples showed an elevated mean Cp value of 35.3 Ϯ 1.89 cycles in the a For amniotic fluids, the first assay of the duplicate was performed with DNA from an initial volume of 7 ml; the second one was performed with DNA from an initial volume of 4 ml.
RE system, indicating a poor parasitic load. The approximately ten-fold greater sensitivity of the RE LC assay compared to the B1 LC assay has been reported by Reischl et al. (21) and Hierl et al. (12).
In comparison with these studies, we used the same set of primers and probes for the B1 gene but a different set for the RE sequence. The study by Reischl et al. (21) was performed retrospectively from a collection of amniotic fluids. The study by Hierl et al. (12) was a prospective one, carried out on immunocompromised patients. Our study was performed during routine diagnosis of both congenital infection and ocular toxoplasmosis and also on immunocompromised individuals, with samples from various origins such as amniotic fluid, bronchoalveolar lavage, cerebrospinal fluid, and blood. The three samples presenting negative duplicated results with the B1 system and positive duplicates with the RE system (9.1%) were a blood sample and two placenta samples. The blood was from an immunosuppressed patient who received a bone marrow allograft. The diagnosis of toxoplasmosis was clinically compatible.
In a multicentric study, Martino et al. (18). demonstrated that many patients who develop toxoplasmosis disease presented specific circulating DNA in peripheral blood prior to toxoplasmosis infection. For placenta samples, the congenital infection was proved through mouse inoculation and by the presence of specific immunoglobulin M in the serum of one of the newborns. Moreover, antenatal diagnosis had been positive for the two infants. Infection was underlined by the presence of T. gondii in amniotic fluid using both mouse inoculation and PCR. At birth, discordant PCR results between the two targets were probably linked to the low parasitic load in the placenta due to the antitoxoplasmic treatment. These results showed that the lack of sensitivity of the B1 gene LC assay could lead in some cases to a false-negative result. Considering all clinical samples, the mean gain in Cp (4.2 Ϯ 1.7 cycles) using LC-PCR targeted to the RE repeated sequence was comparable to the result reported by Reischl on amniotic fluids (4.6 Ϯ 1.1 cycles) (21). Restricted to amniotic fluids, the gain in Cp was even more important in our study, for 4-ml (5.8 Ϯ 1.7 cycles) as well as for 7-ml (5.0 Ϯ 0.7 cycles) volumes.
With the B1 gene amplification system, comparative results obtained with 4-and 7-ml volumes of amniotic fluids demonstrated a narrow nonsignificant gap in Cp values and a large standard deviation. This can be explained by the fact that Cp values are close to the limit of detection. It is well known that in this case, Cp values are variable and less reproducible. Using the RE LC assay, we showed an earlier signal with an initial volume of 7 ml of amniotic fluid than with the 4-ml volume. The gain in sensitivity was significant (1.03 Ϯ 0.72 cycles, P Ͻ 0.05). Throughout the study, however, we never missed a positive signal in a 4-ml volume. Nevertheless, a one-cycle gain is not negligible in cases of low parasitic loads, which give late signals.
In conclusion, the RE LC assay can improve the performance of routine diagnosis of toxoplasmosis compared to the B1 gene LC assay. However, it is noteworthy that during our study, we found no discordances between the two methods in prenatal diagnosis. The greater sensitivity of the RE assay was more marked in the placenta samples for neonatal diagnosis. For other samples (aqueous humor, cerebrospinal fluid, and whole blood), our results need to be confirmed on a more extensive group of patients.