Development of a two-step nucleic acid amplification test for accurate diagnosis of the Mycobacterium tuberculosis complex

The Mycobacterium tuberculosis complex (MTBC) remains one of the top 10 leading causes of death globally. The early diagnosis of MTBC can reduce mortality and mitigate disease transmission. However, current nucleic acid amplification diagnostic test methods are generally time-consuming and show suboptimal diagnostic performance, especially in extrapulmonary MTBC samples or acid-fast stain (AFS)-negative cases. Thus, development of an accurate assay for the diagnosis of MTBC is necessary, particularly under the above mentioned conditions. In this study, a single-tube nested real-time PCR assay (N-RTP) was developed and compared with a newly in-house-developed high-sensitivity real-time PCR assay (HS-RTP) using 134 clinical specimens (including 73 pulmonary and 61 extrapulmonary specimens). The amplification efficiency of HS-RTP and N-RTP was 99.8% and 100.7%, respectively. The sensitivity and specificity of HS-RTP and N-RTP for the diagnosis of MTBC in these specimens were 97.5% (77/79) versus 94.9% (75/79) and 80.0% (44/55) versus 89.1% (49/55), respectively. The sensitivity and specificity of HS-RTP and N-RTP for the diagnosis of MTBC in pulmonary specimens were 96.3% (52/54) versus 96.3% (52/54) and 73.7.0% (14/19) versus 89.5% (17/19), respectively; in extrapulmonary specimens, the sensitivity and specificity of HS-RTP and N-RTP were 100% (25/25) versus 92% (23/25) and 83.3% (30/36) versus 88.9% (32/36), respectively. Among the AFS-negative cases, the sensitivity and specificity of HS-RTP and N-RTP were 97.0% (32/33) versus 90.9% (30/33) and 88.0% (44/50) versus 92.0% (46/50), respectively. Overall, the sensitivity of HS-RTP was higher than that of N-RTP, and the performance was not compromised in extrapulmonary specimens and under AFS-negative conditions. In contrast, the specificity of the N-RTP assay was higher than that of the HS-RTP assay in all types of specimens. In conclusion, the HS-RTP assay would be useful for screening patients suspected of exhibiting an MTBC infection due to its higher sensitivity, while the N-RTP assay could be used for confirmation because of its higher specificity. Our results provide a two-step method (screen to confirm) that simultaneously achieves high sensitivity and specificity in the diagnosis of MTBC.

www.nature.com/scientificreports/ (screening followed by confirmation) achieved a higher sensitivity and specificity in the detection of MTBC and that its performance was not compromised in extrapulmonary specimens or under AFS-negative conditions.

Methods.
Multiple nucleotide sequence alignment of IS6110 and IS1081 fragments in 8 Mycobacterium tuberculosis complexes (MTBCs). The alignment results for IS6110 were previously described 35  Design of primers and probes. The primers and probes were selected to target the IS6110 and IS1081 sequences of the conserved region of MTBC using the online PrimerQuest Tool from the IDT website (https ://sg.idtdn a.com/Prime rQues t/Home/Index ), which were compared to all the available sequences with BLAST (http://www.ncbi.nlm.nih.gov/BLAST /). The IS6110-specific primer and probe set (IS4F and IS4R) was designed as previously described. As an internal control, a primer and probe for bacteriophage lambda (cI857ind 1 Sam 7) were designed to amplify a 111 bp amplicon within the 35,000-36,000 region, for which there is no conserved sequence in the BL21(DE3) E. coli strain, and were designated Ld2F and Ld2R, respectively. The fluorogenic probes IS4P, 1081P, and Ld2P were labeled with a unique fluorescent reporter dye ], [4,4,7,2′,4′,5′,7′-hexachloro-6-carboxyfluorescein (HEX)], and [indocarbocyanine (Cy5)] at the 5′-end, respectively, and an internal ZEN Quencher and a 3′-Iowa Black Fluorescent Quencher (IBFQ) at the 3′-end.
Real-time assay and cycling conditions. The HS-RTP and N-RTP PCR assays were performed using

Results
Primers and probes designed for HS-RTP and N-RTP. Previously, we successfully developed IS4 primer/probe sets for the detection of MTBC. Here, we designed an outer primer pair, Nes6110, that covers the IS4 amplicon and amplifies a larger fragment in the first-round amplification reaction. The amplicons produced by Nes6110 in the first round of PCR are used as the DNA template for the secondary amplification step for IS4. In addition to designing Nes6110, we designed a new primer/probe set for Lambda plasmid DNA, Ld2, which was used as an internal control in this study. The Nes6110 primer was designed with a higher melting temperature (Tm) = 65 °C than those for IS4 and Ld2 (Tm = 60 °C for both). The amplicons of IS4 and Ld2 were designed to exhibit different fluorescence excitation and emission wavelengths, in which the 6-FAM-labeled IS4P probe emits blue fluorescence, and the Cy5-labeled Lambda DNA probe (RPLd) emits red fluorescence. In the preliminary analyses, we tested over 6 primer/probe sets to eliminate cross-reactivity between IS4 and the newly designed primer/probe pairs (data not shown). While Ld2 and Nes6110 elicited interference from IS4, the effect was minor. In practice, of HS-RTP, IS4 and Ld2 were used for the amplification of IS6110 and Lambda DNA. The amplification stage involved initial denaturation at 95 °C for 2 min, followed by 45 cycles of denaturation at 95 °C for 5 s and annealing/extension at 60 °C for 10 s. In practice, the first amplification stage of N-RTP (for Nes6110 amplification) involved initial denaturation at 95 °C for 2 min, followed by 10 cycles of 95 °C for 5 s www.nature.com/scientificreports/ and 65 °C for 10 s, and the second amplification stage involved 45 cycles of denaturation at 95 °C for 5 s, after which the temperature was decreased to 60 °C for 10 s for annealing/extension to allow IS4 and Ld2 binding and amplification ( Fig. 1a,b). In summary, we successfully established primer/probe pairs for N-RTP (with Nes6110) and HS-RTP (without Nes6110) assays.

Validation of the HS-RTP and N-RTP assays with internal standards revealed efficient amplification and increased sensitivity.
To verify the performance of HS-RTP and N-RTP amplification, primer and probe set sensitivity was investigated. Synthetic IS6110-containing plasmid DNA targeted by the IS4 primer/probe set was obtained, which was serially diluted tenfold from 10 5 to 10 1 genomic DNA copies, mixed with bacteriophage lambda DNA (equal to 10 3 genomic DNA copies), and used to produce a standard curve. A comparison of HS-RTP and N-RTP showed that the fluorescence resulting from N-RTP was higher than that from HS-RTP ( Fig. 2a, 2), meaning that effective amplification in the N-RTP assay may increase sensitivity (Fig. 2d). In the study, we used Lambda DNA as the internal control. Specifically, we added synthetic vectors containing Lambda DNA as the internal control. Given that the Lambda DNA primer/probe sets may interrupt primer efficiency of HS-RTP and N-RTP assays, we spiked lower copies of Lambda DNA and decreased the primer/probe concentration. Our tests suggested that 30 cycles are required to detect 1,000 copies of Lambda DNA (data not shown).
The N-RTP assay LOD was estimated to be a single copy. IS6110-containing plasmid DNA with five gene copies or a single gene copy was used to determine the LOD of the assay, and all tests were performed in twenty replicates 36 . The N-RTP assay revealed a 100% positive rate, while the HS-RTP assay showed lower positive rates (100% for five copies and 95% for a single copy) ( Table 2). The detection limit of N-RTP is estimated to be a single copy and one to five copies for HS-RTP.
Analytical specificity. To check for cross-reactivity, eleven NTM species, six bacterial species, and nine fungal species were evaluated in the HS-RTP and N-RTP assays. Finally, no amplification was observed with any of the primers when genomic DNA was extracted from individual species, and the input was 10 6 copies. All the strains were obtained from the microbiology laboratory of Chang Gung Memorial Hospital (CGMH) in Linkou (Supplemental Table 1).
Comparison of the HS-RTP and N-RTP with a standard NAAT and mycobacterium culture. There were 134 clinical specimens subjected to testing by the HS-RTP and N-RTP assays. Cobas TaqMan MTB (CTM) qPCR assay (Roche) was used as the standard NAAT. Mycobaterium culture was used as the referral standard. Among 79 culture-positive specimens, the HS-RTP and N-RTP assays yielded positive results in 77 (97.5%) and 74 (93.7%) samples, respectively (P > 0.05). Additionally, among 55 culture-negative specimens, the HS-RTP and N-RTP assays revealed negative results in 44 (80%) and 49 (89.1%) specimens (P > 0.05), respectively, meaning that more false-positive specimens (five specimens) were obtained by N-RTP (Table 3, panel: All specimens). In both pulmonary and AFS-positive specimens, the results were similar between HS-RTP and N-RTP for culture-positive specimens, while the N-RTP assay elicited three additional false-positive results from nineteen and five of these specimens, respectively (Table 3,   www.nature.com/scientificreports/ AFS-negative specimens that were not detected as positive by the N-RTP assay in culture-positive specimens. Among 36 and 50 culture-negative extrapulmonary and AFS-negative specimens, respectively, there were an additional 2 specimens identified as positive by the N-RTP assay (    Table 4.

Speed and costs.
This study evaluated two molecular methods, N-RTP and HS-RTP, for the diagnosis of MTB. In terms of expense the N-RTP assay costs more than HS-RTP because it requires more materials, including an external primer, a high concentration of DNA polymerase, and abundant dNTPs. Taking into consideration the cost of the real-time PCR machine, the running cost of the N-RTP assay was approximately NT$25, representing an increase of 20% over the HS-RTP assay. In addition, N-RTP involves two PCR runs, and the first PCR step represents an additional step and, thus, additional time compared with HS-RTP. In conclusion, the HS-RTP assay is rapid and inexpensive, and the N-RTP assay is more expensive, time-consuming and precise.

Discussion
In our previous study, an in-house IS6110-based qPCR assay (IS4) for the diagnosis of MTBC was developed. Compared to the commercially available Roche Cobas TaqMan MTB (CTM) qPCR assay, the results show high sensitivity in both pulmonary and extrapulmonary specimens, and the performance is not compromised under AFS-negative conditions; however, the specificity of the IS4 assay is much lower than that of CTM 35 . To improve the specificity of the IS4 assay, a single-tube nested-PCR assay (N-RTP) and a high-sensitivity real-time PCR assay (HS-RTP) were developed in this study by pairing IS4 and Ld2 and obtaining a new primer/probe set for the amplification of Lambda DNA as an internal control for MTBC detection. In addition, we designed a new external primer set (Nes6110) for the N-RTP assay. The performance of HS-RTP and N-RTP was evaluated in 134 clinical specimens and in comparison with mycobacterium culture results. The results indicated that the specificity in N-RTP was obviously increased in the different groups of specimens tested compared with HS-RTP, which is consistent with our expectations. However, we also noted that the sensitivity of N-RTP was slightly decreased in extrapulmonary and AFS-negative specimens, although this difference was not significant (P > 0.05). One possibility is that the Nes6110 primer was affected by inhibitors that were included in the extrapulmonary specimens or AFS-negative specimens. Currently, there are two main testing platforms used in clinical examinations, for screening and confirmation, which are usually used for detecting drug abuse and conducting toxicology testing. Screening, which is also referred to as presumptive testing, is usually cost effective, rapid, and yields fast results; however, it is not very precise, and there are likely to be more false test results due to lower sensitivity and specificity. In contrast, confirmation testing requires greater time and expense than screening but usually provides definitive results. A similar concept was employed in this study. In the HS-RTP assay, which is cheaper, faster and shows high sensitivity (97.5%), the overall specificity is only 80%. In contrast, the N-RTP assay provides a more precise result and shows an obvious increase in specificity to 89.1%, but it is time-consuming and costly. On the basis of comprehensive consideration, we propose screening for suspected MTBC cases by first conducting the HS-RTP assay and then carrying out confirmation with the N-RTP assay. Two-stage nucleic acid amplification testing for screening and confirmation by HS-RTP and N-RTP in the diagnosis of MTBC achieved higher sensitivity and specificity.
Many studies have shown that the detection of MTBC by applying PCR to different genes or a combination of genes achieves higher sensitivity and specificity. IS1081 is a highly conserved gene in MTBC. It consists of four to six copies in the MTBC genome, even in M. bovis (which only contains a single copy of IS6110). As the number of copies between different MTBC members remains stable, several studies have reported the use of IS1081 as the NAAT target. It has been reported that IS1081-PCR sensitivity is increased compared with the use of rpoB, which encodes the β-subunit of bacterial RNA polymerase in MTBC, as a target in pleural fluid specimens 37 . IS1081 has also been combined with 23S rDNA and IS6110 to develop a single-tube triplex PCR assay, and the results indicated that this approach increased the differentiation between MTC and NTM 38 . Khosravi et al. demonstrated that among five target genes (IS1081, IS6110, hsk65kd, mbp64 and mtp40), only IS1081-based PCR showed an identical positivity rate (30.8%) to the results of AFS staining in pleural fluid specimens, and this strategy showed a significantly improved positivity rate in bone and wound specimens (33.3%) 39 . In addition, Fatolahzadeh et al. successfully used IS1081-PCR for the detection of pulmonary tuberculosis, and 78.2% of their isolates yielded positive results 40 . The sensitivity of IS1081-PCR has been reported to be higher than that of IS6110-PCR 41 ; thus, it is a realistic screening method for the rapid identification of positive MTBC cases; however, it lacks the sensitivity of single copy-based strategies 42 . Ultra is a commercialized kit that targets IS6110, Figure 2. Single-tube nested real-time PCR assay using five samples of which Lambda DNA (1000 copies) was mixed with tenfold serial dilutions of IS6110-containing synthetic vectors from 10 5 -10 1 DNA copies. Amplification curve of conventional real-time PCR (a) and single-tube nested real-time PCR (b). Detection of IS6110 target sequence with a FAM-labeled double quenched probe (495-520 nm) showing the increasing number of cycles required to detect reducing DNA copies. The relative fluorescence units (y axis) of the reaction is plotted against the Ct values (x axis). The amplification curves are representative of three independent experiments. (c) Standard curve of single-tube nested real-time PCR (red circles) and conventional PCR (blue circles). The Ct values (y axis) are plotted against the log of the starting quantity of IS6110-containing synthetic vectors DNA copies (x axis) for each dilution. There are representatives of three independent experiments in each dilution. The slope, Ct values, Y-intercept, and R 2 are shown. (d) The Ct values and primer efficiency with and without nested PCR are shown. www.nature.com/scientificreports/ and IS1081 for detection and has been recommended by the WHO as a replacement for the Xpert cartridge (targeting only IS6110). Many reports have shown that the Ultra assay might exhibit higher sensitivity than Xpert; however, the WHO reported that the specificity of the Ultra assay is lower than that of the Xpert assay, and the cost is higher, which means that it will yield a higher proportion of false-positive results 43 . In summary, IS1081 may be a good target for the detection of MTBC in both pulmonary and extrapulmonary specimens, but the variable sensitivity of this approach in specimens of different origins has raised concern. In this study, a single-tube multiplex real-time PCR assay using IS6110/IS1081 as the molecular target was developed to identify MTBC in suspected cases. A comparison of partial IS1081 sequences among the MTBC members is shown in supplemental Fig. 1. Two and one single-nucleotide polymorphisms occurred in M. africanum and M. canettii, respectively. The IS1081 DNA sequence shares 99.9% identity with MTBC, which is similar to the result for the IS6110 sequence and meets our expectations. In addition, we designed a new outer primer (Nes1081) and inner primer/probe set (IS1081F, IS1081R and IS1081P) for IS1081 nested PCR in this study. To verify the performance of IS1081 nested PCR amplification, the sensitivity of the primer and probe set was investigated using the method described previously. The IS1081 amplicon was used to design a 6-carboxyl-X-rhodamine (ROX)labeled IS1081P probe emitting red fluorescence (610 nm). The comparison of HS1081 and N1081 shows that   www.nature.com/scientificreports/ the fluorescence of N1081 is slightly higher than that of HS1081 (Supplemental Fig. 2a, 2b). The primer efficiency of qPCR was 102.8% and 105.8% in the HS1081 and N1081 assays, respectively, and the overall results show a similar correlation coefficient between HS1081 and N1081 (Supplemental Fig. 2c). Compared with HS1081 (Ct values from 23.9 to 36.4), the N1081 results produced amplification curves with relatively lower Ct values (15.3 to 27.9), indicating that effective amplification occurred in the N1081 assay (Supplemental Fig. 2d), which may increase its sensitivity. The LOD of N1081 was estimated to be lower than 10 copies (Supplemental Table 2). On the basis of other reports, we assumed that the use of IS1081 may increase the sensitivity and specificity of the assay when combined with IS6110 for the detection of MTBC. However, an unexpected result was that among the 79 culture-positive specimens, only one specimen (from pleural fluid) yielded a positive result in N1081 and not in N-RTP. However, six specimens were N-RTP positive, while 2 pulmonary specimens and 3 extrapulmonary specimens were N1081 negative. One possibility is that the number of copies of IS6110 is higher than that of IS1081 in most patient specimens, and very few specimens show exactly opposite results. To our knowledge, M. bovis is the only species in this group in which the copy number of IS1081 is higher than that of IS6110 in the genome. However, among 3,321 patient isolates from Taiwan, 3,306 (99.5%) were found to be M. tuberculosis, and only 15 (0.5%) were M. bovis, and this pattern was especially prevalent in aborigines 44 . This result suggests that the prevalence of M. bovis is very low and may explain why we were unable to obtain significant positive results in terms of increased sensitivity and specificity in this study.
There are several limitations of this study that affect future applications. First, the sample size was not sufficient to reach statistical significance; thus, more specimens are required to investigate the diagnostic performance of the developed assays. In addition, we focused on developing IS61110 and IS1081 primer/probe sets, and many other genes that are suitable for NAATs should be compared with this assay in the future. Moreover, other NAAT methods have been used to diagnose TB, such as loop-mediated isothermal amplification (LAMP) 45 and rolling circle amplification (RCA) 46 ; these methods both exhibit high sensitivity and specificity and can be used for comparisons, including comparisons of speed and cost. Moreover, NAAT cannot discriminate the viability of MTBC pathogens. Thus, disease activity of MTBC cannot be determined in which other information, including AFS, mycobacterium culture, and clinical conditions are still necessary for accurate diagnosis and management of MTBC infection.

Conclusion
The two-step nucleic acid amplification test provides high sensitivity and high specificity and thus can be used for screening and confirmation of MTBC. The performance of the test do not compromised in extra-pulmonary or AFS-negative specimens.