Is Real-Time PCR Targeting Rep 529 Suitable for Diagnosis of Toxoplasmosis in Patients Infected with Non-Type II Strains in North America?

Toxoplasma gondii DNA detection is essential to antenatally diagnose a congenital infection and reactivation of a past infection in an immunocompromised patient. Initially, PCR methods targeted the 35-fold repetitive B1 gene, and more recently, coding sequence Rep 529 has been preferred, as it was reported to be repeated 200- to 300-fold and yielded far better sensitivity than amplification of the B1 sequence. To date, few data are available in regard to the efficacy of Rep 529 for non-type II genotypes.

ABSTRACT

Toxoplasma gondii DNA detection is essential to antenatally diagnose a congenital infection and reactivation of a past infection in an immunocompromised patient. Initially, PCR methods targeted the 35-fold repetitive B1 gene, and more recently, coding sequence Rep 529 has been preferred, as it was reported to be repeated 200- to 300-fold and yielded far better sensitivity than amplification of the B1 sequence. To date, few data are available in regard to the efficacy of Rep 529 for non-type II genotypes. In this study, we compared the results of B1 quantitative PCR (qPCR) with those of two different Rep 529 qPCRs performed on 111 samples in two different laboratories (Rep 529-1 and Rep 529-2). The performances of the 3 qPCRs were also compared according to the genotypes of the isolates for 13 type II and 21 non-type II samples. The performance of the Rep 529 target was superior to that of the B1 target regardless of the genotype (threshold cycle [C_T] values for the Rep 529-1 and Rep 529-2 qPCRs were lower than those for the B1 qPCR [P < 0.001 and P < 0.01, respectively]). The same results were observed when a comparison was made according to the genotype of the strain (type II and non-type II genotypes). To our knowledge, these results provide the first relative quantitative data revealing that the efficiency of Rep 529 qPCR does not depend on the genotype of T. gondii isolates and that, in fact, it is superior to B1 qPCR.

INTRODUCTION

Diagnosis of Toxoplasma gondii infection is routinely made by serology, but parasite detection is needed to diagnose a congenital infection antenatally and to confirm reactivation of a past infection, primarily in immunocompromised patients (1). Since the end of the 1980s, PCR has proven to be a useful tool to detect the presence of the parasite in amniotic fluid (AF), blood, various fluids, or biopsy specimens from patients (2, 3). Initially, PCR methods targeted the 35-fold repetitive B1 gene (2, 3). Since the early 2000s, the noncoding sequence Rep 529 has been preferred, as it was reported to be repeated 200- to 300-fold and yielded far higher sensitivity than amplification of the B1 sequence (4, 5) in the setting of diagnosis of congenital infection during the antenatal or postnatal period and of toxoplasmosis from reactivation (2, 4, 6–11). Therefore, in European countries, the Rep 529 sequence has now become the standard quantitative-PCR (qPCR) target for both in-house methods and new commercial assays (11). However, some studies recently questioned whether the genotypes of the strains could impact the sensitivity of PCR methods targeting Rep 529, as Rep 529 qPCR performed equally well or worse in several reports from South America (4, 5, 12–16). This aroused suspicion as to the true number of repetitions of Rep 529 in atypical strains or possible mutations or deletions leading to lack of amplification. Actually, the impact of the parasite genotype would be not be readily detected in European countries, where the vast majority of strains isolated from human cases and animals belong to type II (17–19). In contrast, in South America, T. gondii strains show high genetic diversity, with mainly atypical strains (20–23). Recently, an unexpectedly high prevalence of non-type II strains has been highlighted in the United States (24–26). Therefore, it is very important to determine the performance of Rep 529 PCR for the diagnosis of non-type II strains. In this study, we compared the results of B1 qPCR with those of two different Rep 529 qPCRs performed in two reference laboratories, and the results were analyzed according to the genotypes of the strains.

MATERIALS AND METHODS

Patients and sample collection.

One hundred eleven DNA specimens tested by qPCR at the time of diagnosis from 2003 to 2012 at the Palo Alto Medical Foundation Toxoplasma Serology Laboratory (PAMF-TSL) (laboratory 1) were randomly selected and included in this study. They were collected from 103 patients referred to laboratory 1 for laboratory diagnosis of congenital toxoplasmosis (37 patients), acute infection (5 patients), or reactivation of a past infection (46 patients). For 15 patients, no information was provided on clinical signs and/or serology. Of the 111 samples tested, 39 were from fetuses or newborns and 72 from patients aged 4 to 87 years. Selected samples were sent to laboratory 1 as part of routine diagnosis and treatment with no unnecessary invasive procedures. Genotyping data obtained by microsatellite analysis were available for 34 samples from a previous study (26). Thirteen isolates belonged to type II, 5 to type III, and 16 to atypical lineages. The study was approved by the Institutional Review Board at the Palo Alto Medical Foundation Research Institute (IRB no. 1191799).

Molecular diagnosis.

For the study, the 111 samples were retested simultaneously in laboratory 1 by qPCR targeting B1A (B1) and Rep 529 (Rep 529-1) and in the Parasitology-Mycology Laboratory of the University Hospital of Rennes, Rennes, France (laboratory 2), using a different set of primers targeting Rep 529 (Rep 529-2). In laboratory 1, the primer and probe sequences of the B1A target and Rep 529-1 were as follows: 5′-AAA TGT GGG AAT GAA AGA GAC GCT AAT GT-3′ (forward), 5′-GCG ACC AAT CTG CGA ATA CAC CAA AGT-3′ (reverse), and 5′-TCG CCA GCA GAG GGG AGC-3′ (probe) for the B1A target and 5′-GTT GGG AAG CGA CGA GAG TC-3′ (forward), 5′-ATT CTC TCC GCC ATC ACC AC-3′ (reverse), and 5′-AGA AGA TGT TTC CGG CTT GGC TGC TT-3′ (probe) for the Rep 529 target (10). All samples were analyzed in the same run when retested for B1A and Rep 529 in laboratory 1. The qPCR conditions for both targets were as previously described, using ViiA7 as the platform (9, 10). An internal control (TaqMan exogenous positive control; Applied BioSystems) was used, and runs were considered valid when the internal control was positive.

In laboratory 2, samples were tested for the Rep 529 target following the protocol previously described (4). In this study, the threshold cycle (C_T) values were used as a relative quantification and as a means to compare the results for B1, Rep 529-1, and Rep 529-2.

In laboratories 1 and 2, the criteria to declare a positive result were as follows: at least one of the duplicates, tested twice, amplified in each run with a C_T of amplification value below or equal to 40 and with an amplification curve similar to that of the positive control. When there was no amplification or an atypical amplification curve, the result was considered negative if amplification of the internal control was positive. If amplification of the internal control failed and there was no amplification or an atypical amplification curve for the target of interest, the sample was considered inhibited. In both laboratories, each run was validated when the positive, negative, and internal controls were acceptable.

Statistical analysis.

The VassarStats Website for Statistical Computation (http://vassarstats.net/) was used for statistical analyses. The kappa coefficient was used to compare the results of the 3 qPCRs (B1, Rep 529-1, and Rep 529-2). Perfect agreement equated to a kappa coefficient of 1, and chance agreement equated to a kappa coefficient of 0. Kappa coefficients between 0.81 and 0.99 were considered nearly perfect, those between 0.61 and 0.80 indicated substantial agreement, and those between 0.41 and 0.60 indicated moderate agreement. Statistical analyses were performed with the SAS software package (SAS Institute, Inc., Cary, NC). P values of <0.05 were considered statistically significant. The C_T values obtained with B1, Rep 529-1, and Rep 529-2 qPCRs were compared using the Wilcoxon paired test. The C_T means obtained with each qPCR were compared using the Mann-Whitney test.

RESULTS

Results of qPCRs and interlaboratory comparison of results.

The 111 samples were tested by 3 different qPCR methods: B1, Rep 529-1, and Rep 529-2 (Table 1). A good correlation of qualitative qPCR results was found between B1 and Rep 529-1, as well as between both Rep (Rep 529-1 and Rep 529-2) qPCRs (Table 1).

TABLE 1.

Qualitative and quantitative positive results and correlations among 3 qPCRs performed on 111 samples^a

qPCR	No. of positive results	Range of CT values
B1	74	17.97–38.06
Rep 529-1	76	15.12–38.48
Rep 529-2	74	15.17–40.00

^aKappa coefficients (95% confidence intervals): B1 versus Rep 529-1, 0.96 (0.91 to 1); Rep 529-1 versus Rep 529-2, 0.92 (0.85 to 0.98).

When the C_T values of all the sample types were analyzed, there was a statistical difference between the results for B1 qPCR and both Rep qPCRs, with higher C_T values obtained with B1 qPCR than with Rep 529-1 and Rep 529-2 qPCRs (P < 0.001 and P < 0.01, respectively). (Table 2). The 3 qPCRs were positive simultaneously for 71 samples, with mean C_T values of 29.79 (±0.64), 26.90 (±0.66), and 28.45 (±0.78) for B1, Rep 529-1, and Rep 529-2 qPCRs, respectively. The mean gain in C_T amplification ranged from 2.89 between B1 and Rep 529-1 qPCRs to 1.34 between B1 and Rep 529-2 qPCRs. Results were simultaneously negative by the three qPCRs in 34 out of 111 samples.

TABLE 2.

Qualitative and quantitative results for 111 samples tested by 3 different qPCRs according to sample type

Sample type	qPCR B1 (laboratory 1)		qPCR Rep 529 (laboratory 1)		qPCR Rep 529 (laboratory 2)		P value (Wilcoxon test)
	Positive results [no./totala (%)]	Mean CT ± SEMb	Positive results [no./total (%)]	Mean CT ± SEM	Positive results [no./total (%)]	Mean CT ± SEM
Amniotic fluid	2/9 (22)	28.13 ± 2.99	2/9 (22)	24.97 ± 2.35	2/9 (22)	23.76 ± 4.27	NAc
Tissue (brain biopsy and heart)	9/10 (90)	21.72 ± 1.07	9/10 (90)	19.02 ± 0.97	9/10 (90)	21.06 ± 1.69	<0.01 (B1 vs Rep 529-1) 0.426 (B1 vs Rep 529-2)
Bronchial lavage fluid	5/5 (100)	25.78 ± 2.58	5/5 (100)	21.97 ± 2.72	5/5 (100)	21.46 ± 3.01	NA
Placenta	1/3 (33)	25.75	1/3 (33)	21.77	1/3 (33)	24.1	NA
Other fluids (pleural, pericardial, urine)	0/6 (0)	NA	0/6 (0)	NA	0/6 (0)	NA	NA
Whole blood	9/18 (50)	31.1 ± 1.55	9/18 (50)	27.47 ± 1.64	9/18 (50)	28.56 ± 2.04	<0.01 (B1 vs Rep 529-1)<0.05 (B1 vs Rep 529-2)
Ocular fluids	23/29 (79)	31.63 ± 0.92	24/29 (83)	29.47 ± 1.00	24/29 (83)	30.56 ± 1.17	<0.001 (B1 vs Rep 529-1)0.080 (B1 vs Rep 529-2)
CSF	25/31 (81)	32.52 ± 0.78	26/31 (84)	30.17 ± 0.85	24/31 (77)	31.89 ± 0.93	<0.001 (B1 vs Rep 529-1)0.643 (B1 vs Rep 529-2)
All samples	74/111 (67)	30.09 ± 0.63	76/111 (68)	27.52 ± 0.67	74/111 (67)	28.71 ± 0.76	<0.001 (B1 vs Rep 529-1)<0.01 (B1 vs Rep 529-2)

^aNumber of samples that tested positive/total number of samples.

^bSEM, standard error of the mean.

^cNA, not applicable due to the small number of samples.

Some discrepancies were observed among the 3 qPCRs for ocular fluid and cerebrospinal fluid (CSF) samples (Table 2). Three samples (2 CSF and 1 ocular fluid) that tested positive by B1 and Rep 529-1 qPCRs were negative by Rep 529-2 qPCR; their C_T values by B1 qPCR ranged from 36.81 to 37.72. One ocular fluid sample tested negative by the two qPCRs in laboratory 1 but was found to be positive by Rep 529-2 qPCR, with a C_T value of 35.61. Two samples (CSF and ocular fluid) found negative by B1 qPCR tested positive by both Rep 529 qPCRs (C_T values of 38.48 and 36.3 and C_T values of 34.6 and 35.79 in laboratory 1 and laboratory 2, respectively). For ocular fluid, when the C_T values of the 22 simultaneously positive samples were compared with the three qPCR methods, there was a statistical difference between B1 and Rep 529-1 qPCR results, with Rep 529-1 qPCR performing better than B1 qPCR (P < 0.001) (Table 2). Similarly, for CSF, whole-blood, and tissue samples, Rep 529-1 qPCR performed better than B1 qPCR (P < 0.001, P < 0.01, and P < 0.01, respectively). Significant differences were less often found between B1 and Rep 529-2 qPCRs when sample types were analyzed separately, but overall, both Rep 529 qPCR methods yielded significantly lower C_T results than B1 qPCR (P < 0.001 and P < 0.01 for B1 versus Rep 529-1 and B1 versus Rep 529-2 qPCRs, respectively) (Table 2).

Evaluation of Rep 529 qPCRs according to the genotype of the strain.

For 34 samples, the genotypes of the isolates were available. The means of the C_T values obtained by the three different qPCRs were 25.98 (±0.71), 22.99 (±0.71), and 23.79 (±0.78) for B1, Rep 529-1, and Rep 529-2 qPCRs, respectively. For the 34 samples, the C_T values obtained with B1 qPCR differed significantly from that of either of the Rep 529 qPCRs (P < 0.001), with better performance of both Rep qPCRs than of B1 qPCR. No significant differences were observed between the two Rep 529 qPCRs (P = 0.20). When statistical analyses were done separately on the 13 type II genotype samples, both Rep 529 qPCRs still performed better than B1 qPCR (P < 0.001 and P < 0.01, respectively) (Table 3). Analysis of the results of the 21 non-type II samples also showed that Rep 529 qPCRs were better than B1 qPCR (P < 0.001 and P < 0.05, respectively). The mean gain in ΔC_T between Rep 529-1 or Rep 529-2 and B1 qPCRs were 1.67 and 2.68, respectively, for non-type II strains and 3.34 and 3.47, respectively, for type II strains. A deeper analysis according to the sample type was not possible due to the small number of samples per category.

TABLE 3.

Means of qPCR C_T values for each target, B1, Rep 529-1, and Rep 529-2, according to sample type and genotype (type II or non-type II) of the isolate

Sample type (no. of positive samples)a	Type II					Non-type II
	No. of positive samples	Mean CT ± SEMb			P value (Wilcoxon test)	No. of positive samples	Mean CT ± SEM			P value (Wilcoxon test)
		B1A (laboratory 1)	Rep 529				B1A (laboratory 1)	Rep 529
			Laboratory 1	Laboratory 2				Laboratory 1	Laboratory 2
AF (2)	2	28.13 ± 2.99	24.97 ± 2.35	23.76 ± 4.27		0
OF (8)	3	29.39 ± 1.39	26.15 ± 2.16	23.71 ± 0.95		5	25.97 ± 1.16	23.66 ± 1.46	24.57 ± 1.47
BB (9)	2	24.46 ± 2.33	20.47 ± 2.57	27.44 ± 0.14		7	20.94 ± 1.11	18.60 ± 1.08	19.24 ± 1.55
BAL (4)	2	26.93 ± 4.24	23.05 ± 4.28	20.97 ± 4.61		2	26.08 ± 6.55	22.94 ± 6.72	24.37 ± 6.98
CSF (6)	2	29.66 ± 0.20	27.40 ± 0.80	25.02 ± 1.84		4	27.50 ± 1.10	24.58 ± 1.21	27.97 ± 1.74
Placenta (1)	1	25.75	21.77	24.10		0
WB (4)	1	24.98	20.88	22.36		3	29.16 ± 1.84	25.36 ± 1.84	25.66 ± 2.46
Total (34)	13	27.48 ± 0.90	24.06 ± 1.04	24.00 ± 0.94	<0.001 (B1 vs Rep 529-1)<0.01 (B1 vs Rep 529-2)	21	25.05 ± 0.97	22.32 ± 0.94	23.58 ± 1.14	<0.001 (B1 vs Rep 529-1)<0.05 (B1 vs Rep 529-2)

^aOF, ocular fluid; BB, brain biopsy specimen; BAL, bronchoalveolar lavage fluid; WB, whole blood.

^bSEM, standard error of the mean.

Overall, the genotypes of the isolates do not appear to have an impact on the performance of Rep 529 qPCRs, both of which yielded lower C_T values than B1 qPCR.

DISCUSSION

Molecular techniques rank first in diagnosing congenital toxoplasmosis, as well as reactivation of a past infection in immunocompromised patients (27). The performance of qPCR methods has been largely evaluated in the setting of congenital toxoplasmosis, and the most recently described target, Rep 529, has proven to allow a significant gain in sensitivity in European studies, with a reduction of about 3 cycles of amplification, compared to B1 PCR, whatever the sample type (28). Our study results are consistent with this finding, with a gain of about 3 cycles between Rep 529 qPCRs and B1 qPCR for type II isolates. In a recent retrospective study on 76 clinical samples, the relative sensitivity of B1 qPCR was shown to be only 54% of that of Rep 529 qPCR (4). However, whereas there is high genetic homogeneity of strains circulating in Europe, particularly in France, much greater heterogeneity has been recently highlighted in the United States, but less than in South America (26). The sensitivity of the Rep 529 target was questioned for non-type II T. gondii isolates. Indeed, a Brazilian study found that the sensitivity of the Rep 529 target by conventional PCR was lower than that of B1 (12.7% versus 63.5%) (14). Some hypotheses were raised to explain this lack of sensitivity: (i) great variability of the Rep 529 sequence or possible deletions in atypical-genotype strains, which predominate in Brazil, and (ii) variation in the copy number of Rep 529. In our study, we retrospectively evaluated the Rep 529 target sequences on clinical samples routinely analyzed in the reference laboratory PAMF-TSL, and we compared our results to those obtained in a French reference laboratory using a Rep 529 qPCR technique previously evaluated in several clinical studies (4, 11, 27, 29). Overall, there was good correlation among the three qPCR methods. Three negative results were observed in laboratory 2, whereas results were positive by the other 2 qPCRs. Several hypotheses could explain the lack of amplification of these DNAs: (i) degradation during transport, (ii) variability of Rep 529 sequence at the site of primer annealing (unfortunately, the genotypes of these three samples were not available to confirm a possible link with an atypical genotype), (iii) PCR inhibition, or (iv) a very low parasite load, as the lowest C_T value of these samples was 36.81. Extraction methods also have an impact on PCR results (30). However, in our study, the DNA was already extracted when tested in both laboratories.

When the results of all the samples were pooled, there were few differences regarding the qualitative results obtained with B1 qPCR compared to both Rep 529 qPCRs. No differences were found in AF for the antenatal diagnosis of congenital toxoplasmosis, but some were highlighted for diagnosis based on CSF (used for congenital toxoplasmosis and immunocompromised patients) and on ocular fluid (used for the diagnosis of ocular toxoplasmosis). Ocular fluid and CSF represented almost half of our sample set. These matrices are less studied than amniotic fluid in Europe, and data comparing the performance of B1 and Rep 529 targets in these matrices are scarce (4, 31, 32). In our study, optimal results were obtained with Rep 529 compared to B1. Thus, we confirmed the better performance of the Rep 529 target than the B1 target in our sample set.

Interestingly, when only genotyped samples were considered, the mean C_T of Rep 529 qPCRs was still lower than that obtained with B1 qPCR. When we analyzed genotype II and non-genotype II samples separately, we observed a similar trend. However, the mean gain in C_T between B1 qPCR and Rep 529 qPCRs was lower for non-type II strains than for type II strains, especially for the Rep 529 qPCRs performed in laboratory 2. These results could be explained by differences in the performance of qPCR depending on the Rep 529 target used. Not all the Rep 529 qPCRs seem to perform equally. Indeed, this observation was also made by Sterkers et al., who found in a multicentric evaluation that for the same Rep 529 sequence, some primers were less efficient than others (33). However, it must be underlined that the comparison of the mean C_T of amplification of Rep 529 qPCRs in laboratory 1 and laboratory 2 must be viewed with caution, as the qPCRs were performed on different qPCR devices and with different amplification master mixes, both of which influence the C_T result. Therefore, only Wilcoxon rank tests of results should be taken into account, although we also provide mean C_T values for information. A limitation of our study is the fact that the parasite loads were quite high in the vast majority of samples, as shown by the low mean C_T values, which could explain the close qualitative performances of both PCR targets. The difference might have been more perceptible for low parasite loads. A high load is likely the result of a lack of prenatal screening and treatment of pregnant women in the United States, a situation rarely observed in France (34). Fortunately for this study, it allowed genotyping to be performed on about half of the DNA samples, which is not possible when parasite loads are low. In addition, numerous samples from ocular toxoplasmosis and from immunocompromised patients with uncontrolled parasite multiplication were included in this study. Data from these samples are less frequent in European studies.

Taken together, this work provides for the first time relative quantitative data, although on a small number of samples, showing that the efficiency of Rep 529 is not altered by the genotype of the strain, despite a lower gain in efficiency of amplification of non-type II strains.