Validation of a duplex qPCR for detection of Chlamydia trachomatis DNA in ocular samples and comparison with a direct immunofluorescence method using samples from endemic and non-endemic areas in Brazil CURRENT STATUS: POSTED

Trachoma, caused by the bacteria Chlamydia trachomatis, is the world leading infectious cause of blindness. The disease is associated with poor living conditions, especially in developing countries. In these countries, diagnosis is usually based on clinical evaluation, although laboratory confirmation is necessary. Serological-based tests for trachoma laboratory confirmation are cheaper than molecular-based tests, but the later are more sensitive and specific. Among the available molecular tests, qPCR has the best cost-benefit. With this in mind, the present study developed a new duplex qPCR reaction, which concomitantly detects C. trachomatis cryptic plasmid and the human 18S rRNA gene, as an internal reaction control. The new qPCR was validated using 50 previously qPCR-characterized samples for trachoma infection, and showed 95% specificity and 100% sensitivity, with an estimated LOD95 of 600 ag/µL. Next, 50 samples from an endemic area (Marajó, Pará) and 12 from a non-endemic area (Curitiba, Paraná) were investigated using direct immunofluorescence assay (DFA) or the new duplex qPCR. Among the 50 endemic samples, three were positive by clinical evaluation (6%), 18 by DFA (36%) and 48 by qPCR (96%). All samples positive by the clinic evaluation were also positive by qPCR. From the 18 DFA-positives, qPCR identified 16 as positives as well. On the other hand, 32 samples that were DFA-negative due to the low number of elementary bodies (<5 per slide) were positive by qPCR. The results show that the new duplex qPCR has sensitivity and specificity in similar levels to commercial qPCR tests available, and that qPCR indeed is more sensitive than clinical evaluation or DFA, thus allowing earlier treatment start. The ubiquitous presence of C. trachomatis DNA in samples from the endemic region confirms that constant monitoring is necessary. Additionally, effective measures for the elimination of trachoma and the detection of bacterial DNA in the active infection are of fundamental importance, and this newly developed duplex qPCR is an important tool towards this goal.

that qPCR is a good tool for monitoring the incidence of trachoma in low-infrastructure laboratories [20].
The present study aimed to develop and validate a new research duplex qPCR reaction for concomitant detection of C. trachomatis cryptic plasmid and the human 18S rRNA gene in ocular swab samples. We also compared the performance of the new qPCR to DFA or clinical evaluation for the diagnosis of C. trachomatis infection and found that, as expected, qPCR is more sensitive and specific.

Methods
Samples. Different sets of samples were used. Commercial DNA: Commercial DNA for Chlamydia trachomatis serovar D or serovar J were purchased from ATTC (VR-885D and VR-886D , respectively).
Lyophilized DNA was diluted in water to 10 ng/µL, aliquoted and stored at -80 °C until use. Trachomanegative DNA was extracted from whole blood of non-endemic area volunteers using the High Pure  Direct Immunofluorescence detection (DFA). Samples were methanol-fixed in glass slides, transported at 4-8 °C and stored at -20 °C until analyzed. DFA was performed according to the manufacturer's instructions of the kit "Chlamydia T" (Biocientífica S.A., Buenos Aires, Argentina). Briefly, fixed samples were exposed to fluorescent-labelled monoclonal antibodies against anti-MOMP from C.
trachomatis. Slides were analyzed in a fluorescence microscope, and were considered valid when containing a minimum of 100 cells and 5 clearly visible Chlamydia antigenic structures. Positive and negative controls were performed in all DFA assays. Statistical analysis. All qPCR assays in the ABI7500 system were performed in technical triplicates, except for limit of detection (LOD) determination (8-12 replicates). Results are expressed as mean ± standard deviation of the quantification cycle (C q ) values. Paired Student's t-test (95% confidence level) were calculated using GraphPad Prism v5.0 software (Graph Pad Prism Inc, USA). The 95% limit of detection (LOD 95% ) for detection of the Chlamydia genomic target was calculated by Probit

Results
The performance of the new qPCR was evaluated using commercial genomic extracted DNA from two distinct C. trachomatis serovars, D and J. Figure 1 shows linear dynamic range for detection of the C. trachomatis genomic target in serovar J. Insert in Figure 1 shows representative traces of the reactions used to calculate the linear regression (red lines a-f) obtained with concomitant detection of the human 18S rRNA gene (green lines). Reaction parameters for C. trachomatis DNA detection were efficiency of 92% (slope -3.52), R 2 of 98.6%, and Y-intercept of -32.79. Table I shows average ± SD of the quantification cycle (C q ) for each DNA concentration used in Figure 1. Data show that the reaction is more sensitive towards the detection of serovar J, since the lowest concentration (100 ag/µL) was not detected for serovar D DNA.
Using data from Figure 1, a Probit analysis determined an LOD 95% of 600 ag/µL when assessed with commercial genomic DNA serovar J. When a synthetic double strand DNA containing the same genomic sequence was used as template, LOD 95% reached 1.61 copies/µL, which translates to 8 copies per reaction (Supplemental Figure S1).
Next step was to validate the new duplex qPCR with previously characterized samples. DNA from fifty samples previously characterized by the Aptima Combo2 assay was analyzed by the duplex qPCR described in Figure 1. Table II compares the results obtained with both protocols. Samples 1-30 were previously characterized as "Positive" and samples 31-50 were characterized as "Negative" for C.
trachomatis DNA. Both assays yielded the same result for all samples except for one. "Negative" sample (#34) was classified as "Inconclusive" by the qPCR because the reaction failed to detect the human DNA marker, thus invalidating the results. This sample would have to be re-processed in a clinical setting because failure to detect the human DNA might be attributed to degraded DNA template. These data yield a true-positive rate (sensitivity) of 100% and a true-negative rate (specificity) of 85%, with no false-negatives or false-positive detections, and one inconclusive (Table   III). Cohen's kappa coefficient was calculated to be 0.657, which means that both assays have substantial agreement.
Next, we decided to compare the new duplex qPCR protocol to a DFA protocol that uses a monoclonal anti-MOMP antibody. Figure  However, only three of these samples were clinically positive for C. trachomatis (Table IV), confirming an already known strong disparity between the two diagnostic techniques. Figure 2C shows the dispersion of the C q for each sample detected in Figure 2A plotted over the linear regression of the detection of serially diluted C. trachomatis serovar J DNA. It can be observed that the samples have high C q , highlighting the importance of the linear detection of 100 ag/µL (Table I) and non-linear detection of up to 10 ag/µL ( Figure 2C). Table IV shows that samples 1-50 (endemic region) were primarily negative for the clinical evaluation as well as for immunodetection of MOMP, whereas 48 of those were classified as "Positive" by qPCR.
Samples 51-62 (non-endemic region) were classified as "Negative" for all three diagnostic techniques (clinical evaluation, DFA, or qPCR). There was no correlation between the clinical evaluation, the number of elementary bodies (EBs) as observed by DFA, or the results of the direct immunodetection protocol with the qPCR results. Indeed, Kappa coefficient analysis showed poor correlation between clinical evaluation and qPCR or between DFA and qPCR (less than 0.2 for both cases).

Discussion
The work presented here shows the development and validation of a new duplex qPCR test for detection of C. trachomatis DNA in ocular samples. To reach the necessary sensitivity, we chose a target in the cryptic plasmid, which can have up to ten copies per bacterial genome [24, 28].
The development of the new reaction was performed using commercial DNA from two serovars, D and J, which exhibited different LOD (Table I) [34]. Molecular tests such as qPCR detects and amplifies sequences of the bacterial genome in a very specific pattern, thus being more reliable than DFA. Therefore, despite recent advances, DFA still is less sensitive than qPCR [8-10, 31, [35][36][37][38]. Indeed, since the clinical signs of trachoma infection are not so unique [39], clinical evaluations correlate poorly with PCR detection levels in low prevalence areas or after mass antibiotic treatments [40].
Our results corroborate the differences between the all three techniques. When 62 uncharacterized samples were probed for the presence of C. trachomatis antigens (DFA) or for C. trachomatis DNA (qPCR), almost opposite results emerged (Table IV). Clinical evaluation or detection of EB by DFA showed very few positives amongst the endemic samples, while 96% were positive by qPCR (48 out of 50 samples). This means that, although the patients did not show clinical signs of active infections or scars from repetitive infections, they have been recently infected because the bacterial DNA was still present, possibly in tissue that was not yet removed by the body. Indeed, latent class analyses of clinical examination versus qPCR detection suggest that qPCR positivity is a better predictor for determining chlamydial infection than clinical inspections [41].
However, limitations of the qPCR technique must not be understated. Molecular assays for trachoma diagnosis might present a false-negative rate as high as 20% [41], possibly because of the infection timing or poor DNA extraction efficiencies [42][43][44][45]. Second, the assay presented herein relies on the presence of the cryptic plasmid inside the bacteria, which could be missing [46]. Some authors have also suggested that other species of Chlamydia can also cause trachomatous inflammation [47].
Other limitations that were not evaluated in our study include differences between conjunctival and epithelial specimens, human conjunctival cell yield, DNA extraction efficiency, and thorough removal of molecular inhibitors that may also affect test performance [42][43][44][45].

Conclusions
This work shows the development and validation of a new duplex qPCR that concomitantly detects C. independent experiments for each serovar. ND, not determined.

Concentration (per µL)
C q (mean ± SD) for Serovar J C q (mean ± SD) for Serovar  Table II. Validation of qPCR shown in Figure 1 with pre-characterized patient samples. Samples were pre-characterized by the Aptima Combo2 CT/NG assay (Hologic, USA). Extracted DNA was then analyzed by the NAT Trachoma kit described in the present paper.   Figure 1 28 Linear dynamic range for detection of the C. trachomatis genomic target in serovar J DNA.
C. trachomatis DNA was serially diluted in human DNA and evaluated using the new duplex qPCR. The reaction shows efficiency of 92% and R2 of 98.6%. Insert in Figure 1 shows representative traces of the reactions used to calculate the linear regression (red lines a-f), obtained with concomitant detection of the human 18S rRNA gene (green lines). Traces and linear regression are representative of more than 10 independent experiments with 3-12 replicates per run for each concentration.

Figure 2
Detection of C. trachomatis DNA in human samples. Samples from an endemic and a non-