|Home | About | Journals | Submit | Contact Us | Français|
A European multicenter study was performed to evaluate the performance of a new method, based on the transcription-reverse transcription concerted reaction (TRC-2), which enabled one-step amplification and real-time detection of the Mycobacterium tuberculosis 16S rRNA target directly in clinical specimens. A total of 633 respiratory and nonrespiratory specimens were tested, and the results were compared with those from smears and cultures. A total of 129 patients (Paris center) were followed up in order to evaluate the clinical performance of TRC-2. By using M. tuberculosis complex strains to inoculate sterile sputa, the detection limit of TRC-2 was found to be 30 to 50 CFU/ml. A total of 548 respiratory specimens and 59 extrapulmonary specimens were assessable. For pulmonary specimens, the sensitivities of TRC-2 and acid-fast smear were 86.8% and 50.4%, respectively (P = 0.002). The specificities were 97.5% and 100%, respectively. For extrapulmonary specimens, the sensitivities of TRC-2 and acid-fast smear were 83.3% and 8.3% (P < 0.0001), and the specificities were 95.8% and 100%, respectively. Fifteen of 129 patients were diagnosed with pulmonary tuberculosis (TB). The sensitivities of culture and TRC-2 were 80% (12/15) and 86.7% (13/15) (P = 0.16), and the specificities were 100% and 93.9%, respectively. Based on an 11.6% incidence of TB in our population, the positive predictive values of TRC-2 and culture were 81.3% and 100%, respectively, and the negative predictive values were 98.2% and 97.4%, respectively. These results demonstrated that detection of M. tuberculosis complex in clinical specimens by TRC-2 with ready-to-use reagents was an efficient and rapid method for the diagnosis of pulmonary and extrapulmonary TB.
Accurate and early diagnosis of tuberculosis (TB) remains a critical step in the management and control of TB (22). To this day, the primary rapid tool in case detection remains the microscopic examination of clinical specimens in order to detect acid-fast bacilli. The advantages and disadvantages of the acid-fast smear (AFS) are well defined in the literature. Although culture is considered the gold standard, it has the major disadvantage of requiring 2 to 6 weeks to obtain a result, since mycobacteria are slow-growing organisms (5). The use of an approach combining smear microscopy and specific identification at the same time will enhance predictive values for the rapid diagnosis of TB.
Nucleic acid amplification technology (NAAT) tests represent an important technical advance for microbiology laboratories. However, NAAT developed for the rapid diagnosis of TB has never achieved the same sensitivity as culture (2, 5, 11, 18, 19). In addition, NAAT tests have been time-consuming, despite the fact that some approaches are now partially automated (Cobas Amplicor) (1, 9). These systems are more rapid than culture-based methods, but they can still require up to 12 h and a series of complicated procedures to get the results. They need a high level of expertise from experienced and dedicated technicians (18). Recently, the development of real-time PCR (RT-PCR) has considerably improved the molecular diagnosis of many infectious diseases. Agarose gels and hybridization assays have been replaced by the continuous detection of amplification products as they develop in a closed system using either fluorescence resonance emission transfer probes, molecular beacons, or TaqMan probes (reference 10 and references therein). In addition, RT-PCR has demonstrated shortened turnaround times, providing the potential for parallel use of these assays with smear microscopy.
A new concept, the transcription-reverse transcription concerted reaction (TRC) (14), was developed and tested for the diagnosis of TB using respiratory specimens (6, 20). However, neither of the studies was a multicenter study, and they demonstrated sensitivities ranging from 57.1% to 72.2%. In addition, in the first version of the assay, a small number of smear-positive specimens were still TRC negative. Finally, these studies did not correlate TRC results with the final diagnosis of TB and the treatment outcome for the patients (6, 20).
A second and improved version, referred to below as TRC-2, for the detection of M. tuberculosis complex and use with a semiautomated machine (20) was developed (Extragen M.TB kit, TRCRapid M.TB kit, and TRCRapid-160 analyzer; Tosoh Corporation, Tokyo, Japan). A multicenter study involving one Parisian and two Italian centers evaluated the performance of the TRC-2 method in comparison with AFS and culture for respiratory and extrapulmonary specimens. In addition, the efficiencies of TRC-2 and culture were compared to the final diagnosis of TB (Parisian center). The results confirmed the potential of TRC-2; the assay could be performed on the same working day as AFS, efficiently fulfilling the updated recommendations of the Centers for Disease Control and Prevention for the use of commercial direct-amplification tests.
(Part of this work has been presented at the 107th General Meeting of the American Society for Microbiology, Toronto, Ontario, Canada, 21 to 25 May 2007 .)
Three European centers were involved in a multicenter study testing prospectively both pulmonary and extrapulmonary specimens (633 specimens altogether: 357 in Paris, 100 in Parma, and 176 in Rome) from patients with suspected TB. Of 633 specimens, 607 were assessable, leaving 26 specimens with exclusion criteria for further analysis. One specimen was culture contaminated, and 12 were TRC-2 inhibited, as demonstrated by the absence of a signal with the internal control. Thirteen were positive for mycobacteria other than TB, all giving a negative signal with TRC-2. The numbers of patients per center were 84, 50, and 129 in Rome, Parma, and Paris, respectively.
A total of 548 pulmonary specimens were categorized: 417 sputa, 46 gastric fluids, 68 bronchial aspirates, and 17 bronchial washes (Table (Table1).1). The 59 extrapulmonary specimens comprised 4 peritoneal fluid, 1 pericardial fluid, 3 cerebrospinal fluid, 12 lymph node puncture, 11 pleural fluid, 11 urine, 5 pus, 4 tissue biopsy, 2 semen, and 6 stool specimens.
Patient analysis and follow-up were performed in the Parisian center, and 129 patients' files were analyzed. The study population consisted of untreated, at-risk patients who, on the basis of their physicians' initial assessments, were suspected of having active TB. Patients currently receiving anti-TB therapy for more than 6 days were excluded from the protocol, as were patients who had completed treatment less than 12 months before the date of enrollment.
When necessary, all specimens were decontaminated by the NaCl-NaOH procedure, as previously described (16). Smear microscopy was performed using both auramine and Ziehl-Neelsen staining (3). Culture was performed in liquid (MGIT, MGIT 960 [Becton Dickinson], or BacT/Alert [bioMérieux]) or solid (Lowenstein-Jensen and/or Coletsos; Bio-Rad) medium, with incubation for as long as 63 days for liquid media and 3 months for solid media at 37°C. Mycobacteria isolated from culture were characterized by molecular assays (INNO-LiPA Mycobacteria, version 2 [Innogenetics] and the Genotype Mycobacterium assay [Hain]) (17, 21).
The technology of fluorescence real-time monitoring of isothermal RNA sequence amplification was applied to all specimens. Also called TRC, this technology used the intercalation activating fluorescence DNA probe to emit enhanced fluorescence by binding to a complementary sequence (14). The key point of the new version tested, TRC-2, was the sample preparation, since a previous pilot study demonstrated a lack of sensitivity for smear-positive pulmonary specimens (6). Therefore, the extraction of the nucleic acids used the new Extragen M.TB kit in addition to simultaneous amplification and detection using the TRCRapid M.TB kit (Tosoh Co., Tokyo, Japan). Specimens were frozen at a temperature below −20°C with an equal volume of RNASafer stabilizer reagent (Omega Bio-Tek, Doraville, GA) after decontamination when necessary. Sample preparation, storage, and amplification were performed according to the manufacturer's instructions (Tosoh Co., Tokyo, Japan) (15, 20). At the end of the process, reaction tubes were set up in a dedicated instrument (TRCRapid-160 analyzer; Tosoh Co., Tokyo, Japan) to measure the fluorescence intensity of the reaction product incubated at 41°C (15, 20). Altogether, the manual handling and PCR processing for 14 specimens lasted 2 h.
Evaluation of analytical sensitivity was based on Mycobacterium tuberculosis complex strains (M. tuberculosis H37Rv [ATCC 27296], Mycobacterium africanum ATCC 25420, and Mycobacterium bovis subsp. caprae) added to negative sputum samples at known concentrations (105 CFU/ml, 104 CFU/ml, 103 CFU/ml, 102 CFU/ml, 101 CFU/ml) and assayed by the TRC-2 method (in triplicate). Culture was performed on inoculated sputum to check for the CFU count.
The Fisher exact test procedure for a two-by-two cross table was used for statistical analysis. Statistical significance was accepted for a P value of <0.05.
The development of new molecular assays in the field of rapid TB diagnosis is still a challenge. A pilot study previously performed in our laboratory demonstrated the efficiency of the TRC technology in quickly identifying acid-fast bacilli detected by a positive AFS. However, several AFS-positive sputa were missed, mainly due to the sample preparation and RNA stabilization techniques (6). In addition, RNA extraction is often less efficient than DNA extraction (4). This led to the development of a newer version of the assay with the new Extragen M.TB kit and the use of the RNASafer solution. This new version (TRC-2) was evaluated by performing a European multicenter study and was compared to culture and to the final diagnosis of TB for the patient.
The detection limit of TRC-2 for mycobacterial 16S rRNA by exact CFU counts (on 7H11 agar plates) was 30 to 50 CFU per ml of sputum. TRC-2 never detected mycobacterial 16S rRNA for inocula with fewer than 30 mycobacteria per ml of sputum or in the absence of CFU as detected on agar plates (data not shown).
The distribution of pulmonary specimens and the percentages of positive specimens per center are given in Table Table1.1. Of 548 pulmonary specimens, 119 (21.7%) and 116 (21.2%) were positive by culture and TRC-2, respectively (P = 0.06). Sixty (10.9%) of the 548 pulmonary specimens were AFS positive (P = 0.002 for comparison to TRC-2) (Table (Table1).1). The overall sensitivity and specificity of TRC-2 for culture-positive pulmonary specimens were 88.2% and 97.4%, respectively (Table (Table2).2). By comparison, the sensitivity and specificity of AFS were 50.4% and 100%, respectively (Table (Table2).2). Of 60 AFS-positive pulmonary specimens, 58 were TRC-2 positive, giving a sensitivity of 96.7% for AFS-positive sputa. Of 430 AFS-negative samples, 58 were TRC-2 positive, and 47 of these were also culture positive, resulting in a sensitivity of 79.7% and a specificity of 97% for smear-negative samples. By center, the sensitivities of TRC-2 were 90.5%, 80%, and 89.5% for Rome, Parma, and Paris, respectively. These sensitivities were in the range of recently published values for traditional commercialized PCR assays: 83 to 96.7% for the Cobas Amplicor assay (Roche Diagnostics) and 85.7 to 97.8% for the AMTD2 assay (bioMérieux-Gen-Probe) on respiratory specimens (2, 11, 18, 19).
The overall sensitivity and specificity of TRC-2 for culture-positive extrapulmonary samples were 83.3% and 95.8%, respectively (Table (Table3).3). By comparison, 1 out of 12 culture-positive extrapulmonary samples was AFS positive (sensitivity, 8.3%; P < 0.0001 for comparison to TRC-2) (Table (Table3).3). None of the culture-negative extrapulmonary specimens were AFS positive (specificity, 100%) (Table (Table3).3). The only AFS-positive extrapulmonary specimen was also TRC-2 positive. The sensitivity of TRC-2 for extrapulmonary specimens was higher than the published sensitivities of in-house PCR assays for such specimens (2, 11, 18, 19), despite the low number of specimens tested. Comparison with the BD-Probe Tec assay demonstrated that more specimens were BD-Probe Tec positive than TRC-2 positive (data not shown). However, numerous specimens were culture negative, giving a specificity of 87.2%, which is lower than the specificities of TRC-2 (95.7%) and other commercialized assays (2, 11, 18, 19).
A total of 129 patients with suspected pulmonary TB were included in the study. The male-to-female ratio was 2.5. Patients originated mostly from Northern Africa (24%), Central Africa (13%), and France (35%). The median age was 44.2 years, and 13.5% of the patients were infected with human immunodeficiency virus. Fifteen patients presented and were treated for pulmonary TB, giving a TB incidence of 11.6%. Other diagnoses were mainly lung cancer, chronic obstructive pulmonary disease, asthma, sarcoïdosis, bacterial (nonmycobacterial) pneumonia, and aspergillosis. The diagnosis was established by clinicians on the basis of clinical symptoms and a positive response to a trial of anti-TB treatment, with or without positive histopathology and microbiology results.
The overall sensitivities and specificities of TRC-2 and culture compared to a final diagnosis of TB are shown in Table Table4.4. The sensitivity of TRC-2 was higher than that of culture (86.7% versus 80%). However, this difference was not significant (P = 0.16). Three of 114 untreated patients were TRC-2 positive (specificity, 97.4%) (Table (Table4).4). None of these patients was smear or culture positive. One had septicemia with Nocardia spp., and the other two presented with exacerbations of chronic obstructive pulmonary disease. Finally, one patient received anti-TB treatment, but the diagnosis of TB was not confirmed due to the absence of a positive outcome. TRC-2, AFS, and culture all gave negative results for all specimens tested from this patient.
The TB incidence rate in our population was 11.6%, allowing the calculation of positive and negative predictive values (PPV and NPV, respectively). The PPV were 100% and 81.3% for culture and TRC-2, respectively. The NPV were 97.4% and 98.2% for culture and TRC-2, respectively.
Previously published reports demonstrated that the sensitivity of NAAT could be increased either by changing the target (9, 11, 12) or by modifying the DNA extraction protocol (13). Our multicenter study demonstrated similarly increased sensitivity by improving the sample preparation and RNA stabilization. The overall sensitivity of the first version of the TRC for pulmonary specimens was 57.1%, compared to 88.1% sensitivity for TRC-2. In addition, the sensitivity for AFS-positive sputa increased from 73% to 96.7% (6; also this study).
The profile of a new molecular assay for the rapid diagnosis of TB might pose several risks: primarily, its cost relative to that of AFS, although TRC-2 is not yet commercialized in Europe; its implementation in a laboratory with dedicated technicians; and its difficulty for use on a daily basis, rendering the notion of rapidity useless. RT-PCR has considerably changed mycobacterial identification by decreasing the turnaround time from weeks to hours. Automation of the preparation and extraction step might represent a solution for facilitating the implementation of TRC-2 in a clinical mycobacteriology laboratory, as described recently (7). Decreased sensitivity to inhibitors, as observed in RT-PCR assays, allowed direct testing on nondecontaminated sputa (7), meaning that AFS and RT-PCR results might be obtained on the same working day.
TRC-2 had an NPV close to 98%, allowing potential exclusion of a diagnosis of TB when the assay is negative. In addition, the sensitivity of TRC-2 for AFS-positive pulmonary specimens is close to 100%, meaning that an identification result might be sent back to the clinician to help identify the degree of infectiousness of the tested patient. Conversely, a negative result might allow a strategic wait for further confirmation of an infection due to mycobacteria other than M. tuberculosis. This might considerably improve the clinical impact of molecular assays on TB patient management, and such an approach might be feasible in a mycobacteriology laboratory.
In conclusion, the data presented from this multicenter study demonstrate the high sensitivity and high specificity of the improved TRC-2 method relative to AFS, culture, and a final “clinical” diagnosis of TB. Although we still had negative TRC-2 specimens that were culture positive, we were able to obtain TRC-2 results in 45 min, indicating that a combined smear result and TRC-2 result can be sent together to the clinician.
We gratefully acknowledge Ben Marshall (Southampton University Hospitals Trust, United Kingdom) for careful review of the manuscript. We warmly thank Tosoh Co. (Tokyo, Japan) for its constant support and, more specifically, Kirsten Van Garsse and Shigeo Tsuchiya for careful analysis of the data and for fruitful discussions.
Published ahead of print on 9 September 2009.