A.3.1. Digestion, decontamination and concentration of specimens
Upon receipt in the laboratory, respiratory samples submitted for TB diagnosis are decontaminated and concentrated. Decontamination ensures that other microorganisms present in samples, such as respiratory bacterial flora, do not outgrow the slow-growing TB in solid and liquid media, rendering the test uninterpretable and hence limiting the sensitivity of culture. Concentration separates mucus and debris from mycobacterial cells, which increases the sensitivity of smear microscopy. Decontamination and concentration processes also improve the analytical performance of molecular assays performed on primary clinical samples by increasing the available TB molecular template and eliminating potential PCR inhibitor substances. To verify the success of the decontamination process, monitoring of contamination rates is recommended.
Primary sample decontamination and concentration are processes which are not automated. They require laboratory personnel time and need to be quality controlled. Notably, sterile samples, such as cerebrospinal fluid and tissue biopsies, do not need to be subjected to decontamination nor concentration procedures and can be processed immediately for microscopy, NAAT and/or culture.
A.3.2. Acid-Fast smear and microscopy
The early and rapid diagnosis of TB still relies on the traditional acid-fast bacilli (AFB) smear. Overall, smears have a reported sensitivity of 20-80%, depending on many factors, including the type of specimen, stain used and the experience of the technologist.66–69 In addition, fluorescence microscopy can be performed by conventional mercury vapor fluorescence microscopes or newer, light-emitting diode microscopes, which have many practical advantages and have been endorsed by the World Health Organization (WHO). The sensitivity of all staining methods is inferior to that of culture. The threshold of detection of AFB in concentrated specimens using a fluorochrome stain is 5,000-10,000 bacteria/mL of sputum and is 100,000 bacteria/mL using a carbol fuchsin stain (eg, Ziehl-Neelsen). Although they are more rapid, “direct smears” without digestion, decontamination and concentration steps are discouraged because of the inherent lack of sensitivity.68 The threshold of detection in unconcentrated smears is 10-fold higher, resulting in much lower sensitivity. Culture is more sensitive than smear and can detect a bacillary load as low as 10 bacteria/mL.
The following guidelines should be observed:
Slides should be individually prepared and fixed to prevent cross-contamination.
Control slides that contain known acid-fast (positive control) and non-acid-fast (negative control, to assess for possibility of contamination) organisms should be run with each batch of smears prepared.
Primary specimen smears should be stained and reviewed using the fluorochrome method. Laboratories should confirm new AFB-positive smears by a second reader. Smears that are questionable should be repeated or can be stained using a carbol fuchsin method for review.
Fluorochrome and carbol fuchsin stain performance should be confirmed with each new lot of reagents by reviewing AFB-positive and AFB-negative control slides prior to reading patient smears.
Smears should be reported following an established grading system, such as that recommended by the WHO,48 Clinical & Laboratory Standards Institute (CLSI)70 or Association of Public Health Laboratories (APHL)71 (see Table 1).
Laboratories should participate in an approved proficiency program that includes acid-fast smears.
The American Thoracic Society, U.S. Centers for Disease Control and Prevention (CDC) and the Canadian Thoracic Society recommend that laboratories not performing a minimum of 15 AFB smears/week should refer specimens to another laboratory or reference laboratory.
A.3.3. Molecular detection of mycobacteria directly from clinical samples
Molecular detection of M. tuberculosis deoxyribonucleic acid (DNA) from clinical samples uses NAATs, of which polymerase chain reaction (PCR) is the most common method. For this reason, clinicians often use the terms NAAT and PCR interchangeably. In addition to commercial assays, there are many protocols for laboratory-developed molecular assays. Unlike standardized, commercial NAATs, in-house NAATs have been historically associated with unreliable results.50 Validation studies should, therefore, be conducted before implementation and the tests used only in accredited laboratories with quality-assurance systems in place.
Previously, performance of NAATs required sophisticated laboratory infrastructure and highly skilled technicians. This changed when partially or fully automated systems such as the Xpert MTB/RIF© test (Cepheid Inc, Sunnyvale, CA) were progressively introduced in Canadian clinical laboratories. Xpert MTB/RIF© test is a cartridge-based, automated, nested, real-time PCR test utilizing the GeneXpert© platform, which can simultaneously detect M. tuberculosis for diagnosis and mutations conferring resistance to rifampicin, a marker of multidrug-resistant TB (MDR-TB), in less than two hours with minimal hands-on technical time. Even with such automated and cartridge-based molecular assays, the risk of contaminating the test site with amplified DNA demands following stringent standardized laboratory processes and implementation of quality-control procedures. At the time of writing, the following assays are commercially available and Health Canada approved: Roche (COBAS® Taqman® MTB; real-time-PCR); Becton Dickson (BD ProbeTec®, strand displacement amplification); Gen-Probe (Amplified Mycobacterium tuberculosis Direct, [AMTD], transcription mediated amplification); Hain Lifescience (GenoType® Mycobacteria Direct, PCR); and Cepheid (Xpert MTB/RIF®, automated cartridge-based nested PCR). The COBAS® Taqman® MTB, AMTD, and Xpert MTB/RIF tests are approved for direct testing on sputum specimens. A living registry of assays can be found by searching the Medical Devices Active License Listing online query website.72
A Cochrane systematic review on the accuracy of Xpert MTB/RIF identified 18 published studies.73 The majority were performed in low- and middle-income countries. Although the test was originally presented as a point-of-care assay, in 17 of the 18 studies, Xpert was performed by trained technicians in reference laboratories. In the meta-analyses for M. tuberculosis detection (15 studies, 7,517 participants), pooled median sensitivity and specificity were 88% (83%, 92%) and 98% (97%, 99%) respectively. Xpert could distinguish between M. tuberculosis and non-tuberculosis myobacteria (NTM) in clinical samples with high accuracy. Of 139 specimens with NTM, cross-reactivity was observed in only one specimen. A more recent Cochrane report examined the newer Xpert MTB/RIF Ultra© test (Cepheid Inc, Sunnyvale, CA) and compared it to the Xpert MTB/RIF© test. The operating parameters for the classic test were effectively unchanged in this updated meta-analysis: sensitivity and specificity of 84.7% (78.6 to 89.9) and 98.4% (97.0 to 99.3). In comparison, for the newer test (Ultra), the pooled sensitivity and specificity from seven studies (2,834 participants) against culture were 90.9% (86.2 to 94.7) and 95.6% (93.0 to 97.4). Overall, the available evidence shows high accuracy for TB detection for both tests, with higher sensitivity and lower specificity for the Xpert Ultra.52 The laboratory should therefore be alert with the transition to the Ultra assay for the possibility of observing NAAT-positive, culture-negative samples. Of note, this evidence is mostly from high-burden countries and involves the use of spontaneous sputum samples. Similar data from low-incidence settings and with the use of induced sputum samples are lacking. However retrospective studies demonstrated that in low TB-burden settings where there is very limited TB transmission, the specificity of Ultra is very high (99.3%, 95% CI 96-99).53 Operational data, although limited, also suggest that Xpert MTB/RIF is able to significantly reduce the time to diagnosis and treatment.74
NAATs are currently recommended for use only on respiratory specimens, although upon special request they can be used on other specimens (eg, cerebrospinal fluid) from laboratories that have validated their assays for those sample types. As specified in Chapter 3: Diagnosis of Tuberculosis Disease and Drug-resistant Tuberculosis, NAATs should not be used for monitoring TB treatment response or for infection-control purposes after the start of treatment.
In some cases, results may be “indeterminate” because of inhibitors in the specimen or a very low bacterial load. Appropriate controls should be included when applicable to rule out inhibition by the specimen. Special care should be taken to avoid cross-contamination of NAAT samples. Laboratories should ensure that there is a clean environment and should follow proper molecular-testing practices in the preparation of solutions used in NAAT tests to effectively prevent contamination. There should be a physical separation of the laboratory areas used to prepare molecular reagents, handle the DNA template and conduct post-amplification detection. It is advisable not to conduct molecular assays in the containment level 3 laboratory, where mycobacterial cultures grow, as this increases the opportunities for contamination.
Laboratory-developed PCR methods can be less costly than commercially available methods but require advanced technical skills. Such methods can be used for detection of MTBC75 in specimens not recommended for testing with a commercial kit, such as formalin-fixed tissue blocks or extra-pulmonary samples. The analytical sensitivity of such tests should be reported with the results. Before implementing an MTBC molecular assay (laboratory-developed or commercial), laboratories should consult the CLSI guideline, Molecular Diagnostic Methods for Infectious Diseases, for guidance on the validation and implementation of a new molecular diagnostic test.76 Validation of any new or adapted test methods should be completed to evaluate the performance characteristics and technical competence of the test. All test methods should be verified as being appropriate and adequate before being undertaken. Quality-control and -assurance programs should be established to monitor the performance of NAAT assays and avoid false-positive and false-negative results.
A.3.4. Mycobacterial culture
Culture remains the gold standard for a positive laboratory diagnosis of TB. MTBC typically has a faster growth rate in liquid media than on solid agar. Also, liquid cultures are 15-30% more sensitive than solid cultures.77 Three automated liquid culture systems are approved by Health Canada: Becton-Dickinson (Bactec MGIT [mycobacterial growth indicator tube]); bioMérieux (BacT/ALERT); and Trek Diagnostic Systems Inc. (Myco-ESP culture System II). These are fully automated systems that use either fluorometric or colorimetric detection of mycobacterial growth and permit a higher throughput of specimens for testing. For pulmonary TB, the sensitivity of three sputum cultures exceeds 90%, although six specimens are required to achieve 100% sensitivity. Three sputum cultures are recommended for the diagnosis of a new case, as this represents the best balance between high sensitivity and efficiency.78
At least one liquid medium should be inoculated from each clinical specimen for culturing of AFB and, depending on the sample, labs may also do a solid medium culture. Cultures should be kept for a minimum of six and up to eight weeks for observation of growth. Positive cultures of MTBC should be retained for at least one year, in case additional testing is required.28,79
It is important to remember that occasionally, cultures can be falsely positive for MTBC, primarily because of cross-contamination within the laboratory, although “mix-up” by the submitter has also been documented.80,81 When clinical suspicion is low, a report of a single positive culture, especially with a negative smear and a long detection time, should raise the possibility of a false-positive result. The laboratory reporting this culture should investigate and review all positive cultures initially processed on the same day or within proximity to the culture. If there are other positive cultures that could be the source of cross-contamination, genetic analyses on the isolate alongside other isolates from the same lab processed around the same date should be done.82
A.3.5. Identification of mycobacterial species from culture
Mycobacterial identification based on biochemical and/or physical characteristics is labor-intensive and slow, and may not adequately identify the organism. Rapid identification of a growing culture as MTBC (vs NTM) and further subspeciation of the MTBC members is necessary for clinical and public health purposes.
Rapid culture identification can take different approaches, largely based on either genotypic characteristics or antigen detection/protein spectrometry. Immunochromatographic tests targeting DNA or protein of M. tuberculosis have been shown to be a sensitive and specific way to identify MTBC members. Similarly, MALDI-ToF analysis can be used to conclusively place an organism within either the MTBC or the NTM group. Neither of these approaches will allow sub-speciation of complex members.
Genotypic-based techniques, such as NAAT targeting specific differentiating gene targets (eg, the regions-of-difference), targeted sequencing (eg, of gyrB gene) or whole genome sequencing (WGS) can allow sub-speciation between complex members. Confirmation of a growing organism as M. bovis (or M. bovis Bacille Calmette-Guérin [BCG]) is particularly important, given the intrinsic resistance of these organisms to pyrazinamide and the need to look for an underlying cause of BCG infection (bladder cancer therapy or young infant post-vaccination).
A.3.6. Susceptibility testing for antituberculous drugs
Agar proportion is still considered the gold standard for MTBC drug susceptibility testing (DST).83 However, because of the labor-intensive nature and lengthy incubation time for the assay, the more rapid liquid media detection methods using continuous monitoring systems are now recommended and DST results are often available within 10-14 days from the time of receipt of the culture. The most current CLSI guidelines should be consulted for the following testing parameters.84
Laboratories that perform DST should generally be accredited reference laboratories, to ensure adequate volume of activity to maintain expertise.
For all new M. tuberculosis isolates, susceptibility to first-line antibiotics should be tested. First-line antibiotics are isoniazid (INH), rifampin (RMP), ethambutol (EMB) and pyrazinamide (PZA).
DST of second-line antibiotics should be set up when resistance to first-line anti-tuberculous drugs is detected, regardless of whether the DST on those first-line drugs is repeated.
Second-line drugs for which testing is currently available (at the time of writing) in reference labs in Canada include amikacin, fluoroquinolones (levofloxacin and/or moxifloxacin), rifabutin, ethionamide, p-aminosalicylic acid and linezolid.
Laboratories should test at least one drug from each class; in particular, at least one fluoroquinolone should be tested; the selection of which fluoroquinolone to test should be based on consultation with physicians who manage patients with drug-resistant TB.
Other drugs that are used for the treatment of MDR TB include bedaquiline, clofazimine, cycloserine, delamanid, and imipenem/meropenem. For bedaquiline, delamanid and clofazimine, testing is not available in Canada but the authors urge Canadian reference labs to meet this need, given that the drugs are being used and standards are available. Although cycloserine and imipenem/meropenem are viable treatment options, the CLSI does not recommend testing of cycloserine or imipenem/meropenem.
A.3.7. Molecular prediction of anti-tuberculous drug resistance
The molecular prediction of anti-tuberculous drug resistance in M. tuberculosis has become an important tool in the rapid identification of MDR TB. These molecular methods can decrease the time it takes to detect resistance using phenotypic methods and accelerate the time to adjustment of therapy. Molecular testing for determinants of drug resistance provides presumptive results and the use of these tests does not eliminate the need for conventional DST.
These methods should be validated just as any other method would be, and used only in conjunction with phenotypic susceptibility testing. However, until now, there are no defined positive controls to represent the presence of MDR-strains of M. tuberculosis. One recently developed option is safe BCG (a tuberculosis vaccine) strains marked with known mutations that confer resistance to either INH, RMP, moxifloxacin or bedaquiline.85 BCG is already resistant to PZA.
The methods to predict resistance include laboratory-developed PCR, approved commercial line-probe assays, real-time PCR-based assays, targeted sequencing and WGS. With the exception of WGS, all are limited to specific, predetermined targets in the genome; as a result, resistance-associated mutations and/or insertions/deletions outside these targets can be missed. Regarding targeted methods, two genotypic methods are endorsed by the WHO: 1) line-probe assays (LPAs) and 2) the GeneXpert MTB/RIF test.
LPAs have been developed and evaluated to perform DST from smear-positive sputum samples directly or to perform rapid DST on culture isolates. Two LPA tests are commercially available: the Inno-LiPARif.TB line probe assay (Innogenetics, Belgium) and the GenoType MTBDRplus assay (Hain Lifescience, Germany). The GenoTypeMTBDRplus assay is approved by Health Canada.86 It can be used for testing sputum or for testing an M. tuberculosis culture. A meta-analysis showed that the GenoType MTBDRplus assay has a pooled sensitivity of 98.1% and specificity of 98.7%.87 The accuracy for INH was variable, with lower and inconsistent sensitivity (84.3%) and high specificity (99.5%). LPAs are endorsed by the WHO for rapid diagnosis of INH and RMP resistance from sputum smear-positive samples. However, the use of LPAs does not eliminate the need for phenotypic DST.
As previously described, the Xpert MTB/RIF assay can provide a rapid diagnosis of TB and can also detect rpoB mutations, providing a sensitivity of about 94% and a specificity of 98% for RMP resistance.52 However, these estimates are from high-burden settings. The predictive value for RMP resistance will depend on the prevalence of drug-resistant TB in a given setting. In the aforementioned meta-analysis of the test for diagnosis of TB, an analysis was also performed for RMP resistance detection (11 studies, 2,340 participants). The pooled median sensitivity and specificity were 94% (87%, 97%) and 98% (97%, 99%) respectively for RMP resistance. Although the specificity is high, the prevalence of RMP resistance is low in Canada. The management of patients with an isolated RMP resistance report in an individual with low risk of MDR-TB is discussed in Chapter 8: Drug-Resistant Tuberculosis. This issue likely also applies with the latest generation of the test, the Xpert Ultra, where the pooled sensitivity and specificity for RMP resistance were 94.9% (88.9 to 97.9) and 99.1% (97.7 to 99.8).52
In contrast to targeted approaches, WGS provides information on the entire genome, which can then be used to not only predict resistance, but also predict susceptibility to different antibiotics, based on known genotype-phenotype associations.88
WGS has been demonstrated to have high sensitivity and specificity overall for prediction of phenotypic drug resistance for first-line drugs in particular.89 An analysis of more than 10,000 genomes with associated phenotypic data from 16 countries conducted by the CRyPTIC consortium found sensitivities of WGS-based predictions for INH, RMP, EMB and PZA were 97.5, 97.1, 94.6, and 91.3%, with specificities of 98.8, 99.0, 93.6, and 96.8%, respectively.89 Importantly for a low-resistance setting such as Canada, the modeled negative predictive value of this approach remained above 95% even with resistance prevalence levels reaching 34% (PZA) to 57% (RMP). A sub-analysis was also done, restricted to datasets not enriched for resistance; when capitalizing on the association between INH resistance and resistance to other drugs, pan-susceptibility to first-line drugs was 99.7% concordant with phenotypic DST. Similar findings have subsequently been reported in studies from the Netherlands90 and New York State.91 Discordant results in these studies were largely attributed to clerical errors/mislabeling, mutations where the minimum inhibitory concentration was close to threshold (eg, often seen with EMB), or inability to reproduce results on Mycobacteria Growth Indicator Tube (MGIT).91
In the Netherlands, a change to WGS for sensitivity prediction was estimated to reduce phenotypic testing for TB by 90%,90 as isolates predicted to be pan-sensitive no longer undergo phenotypic testing.89 To catch the aforementioned discrepancies, as well as novel mutations and/or mutations that have not yet been catalogued; however, it would be good practice for Canadian laboratories considering the transition to WGS-based predictions to continue to test a minimum percent of isolates (eg, 5%) predicted to be pan-susceptible on WGS, using phenotypic DST.
In addition to reducing phenotypic testing demands, another advantage of using WGS for resistance prediction is a potentially reduced turnaround time. This is seen in 2 scenarios. If the WGS result predicts resistance, the laboratory can proceed immediately to both first- and second-line phenotypic DST, without needing to do these tests sequentially. If the WGS results predicts pan-susceptibility to front-line drugs, the result can be emitted immediately, with no need to do phenotypic DST.
Ideally, to reduce turnaround time, WGS would be conducted directly on sputum; however, obtaining sufficient M. tuberculosis genomic DNA92,93 and sequencing depth for resistance analysis remains challenging. To enable identification of most low-frequency variants, a target of ∼100x depth of coverage is recommended for positive cultures. In New York State, sequencing is done on early-positive-MGIT culture, where a minimum concentration of 0.2 ng/microL was required to achieve <7% failure on initial WGS; under these conditions, mean depth was 133x with results reported on average 9 days earlier for first-line drugs and 32 days earlier for second-line drugs compared to DST, with a turnaround time of 15 days from first positive and including weekly batching of specimens.
Switch To: Français