Current issues on molecular and immunological diagnosis of tuberculosis.

Laboratory diagnosis of tuberculosis (TB) traditionally relies on smear microscopy and culture of Mycobacterium tuberculosis from clinical samples. With recent advances in technology, there have been numerous efforts to develop new diagnostic tests for TB that overcome the low sensitivity and specificity and long turnover time associated with current diagnostic tests. Molecular biological tests based on nucleic acid amplification have brought an unprecedented opportunity for the rapid and specific detection of M. tuberculosis from clinical specimens. With automated sequencing analysis, species identification of mycobacteria is now easier and more accurate than with conventional methods, and rapid detection of mutations in the genes associated with resistance to TB drugs provides early information on the potential drug resistance for each clinical isolate or for clinical samples. In addition, immunological tests for the detection of M. tuberculosis antigens and antibodies to the antigens have been explored to identify individuals at risk of developing TB or with latent TB infection (LTBI). The recent introduction of commercial IFN-gamma assay kits for the detection of LTBI provides a new approach for TB control even in areas with a high incidence of TB. However, these molecular and immunological tools still require further evaluation using large scale cohort studies before implementation in TB control programs.

MOLECULAR DIAGNOSIS OF TUBERCULOSIS

Molecular tests have brought unprecedented opportunities for the rapid diagnosis of tuberculosis (TB) and can be incorporated into control programs for the disease. Most research on molecular tests has focused on the following three areas: (a) detection of Mycobacterium tuberculosis in clinical specimens, (b) identification of Mycobacterium species, and (c) detection of mutations in genes associated with resistance to drugs against TB.

Detection of M. tuberculosis in clinical samples

Over the last 20 years, nucleic acid amplification (NAA) tests have been studied for the ability to rapidly detect nucleic acids of pathogens in clinical samples for the diagnosis of infectious diseases including TB. NAA tests were particularly attractive for the diagnosis of TB because of the slow growth of TB, and these tests have advantages such as greater sensitivity and specificity and faster results than conventional laboratory diagnostics tests. Among the NAA tests, the polymerase chain reaction (PCR) has been most widely used for the detection of M. tuberculosis in clinical specimens including sputum, blood, bone marrow, and biopsy samples. Subsequently, several commercial kits have been developed using different targets from the M. tuberculosis genome and amplification formats, and these kits have been evaluated extensively in various clinical settings and samples with periodic reviews of their accuracy.1-5 Despite numerous claims that commercial kits have greater sensitivity than do culture methods, a series of technical problems have been encountered when PCR is employed in clinical mycobacteriology laboratories. A recent meta-analysis on commercially-based NAA tests including Amplicor-MTB (PCR), Cobas Amplicor-MTB (PCR), BDProbeTecET (strand displacement amplification: SDR), E-MTD (transcription mediated amplification; TMD), LCx (ligase chain reaction: LCR) showed marked variation in the diagnostic accuracy of the tests with regard to sensitivity and specificity.1 In the case of AFB smear-positive samples, the mean sensitivity ranged from 0.96 for the lowest kit to 0.98 for the highest, but the mean specificity varied markedly from 0.71 for the lowest kit to 0.96 for the highest. With AFB smear-negative samples, however, the mean sensitivity was only 0.57 for the lowest kit and 0.76 for the highest, and the mean specificity was 0.97 for the lowest and 0.99 for the highest. Overall, the pooled sensitivity and specificity were 0.96 and 0.85 for AFB smear-positive samples and 0.66 and 0.98 for AFB-negative samples, respectively. The high variability in sensitivity and specificity was also noted in a previous meta-analysis of commercially available kits.2-5

There have also been tremendous efforts to develop and evaluate the use of in-house PCR particularly in a research-based laboratory, where commercial kits are not affordable. In most cases, the variability in sensitivity and specificity for in-house PCR is even higher than that for commercial kits.2-4 With in-house PCR, the sensitivity varied from 9.4% to 100%, and the specificity varied from 5.6% to 100%.2 Therefore, careful interpretation of the NAA test results is important due to potentially confounding factors that may influence the results, such as the targets of amplification, the nature and amount of specimens, the storage method and duration before experiment, the specimen processing procedure, DNA extraction procedures, the presence of inhibitors, amplification methods, detection methods of amplified products, the contamination by amplicons, and cross-contamination between samples, etc.6,7

Besides the problems in accuracy and confounding factors associated with NAA tests, the relatively high expense is one of the major drawbacks for implementing these tests in field settings where TB is a major public health problem. Most NAA tests require expensive equipment and trained personnel as well as an expensive supply of materials. In order to overcome these drawbacks, another commercial kit, the loop-mediated isothermal amplification (LAMP) test,8 was recently developed to detect M. tuberculosis.9 The major advantages of the LAMP test include a room temperature reaction mode, obviating the need for thermocyclers, and visual reading of the results, obviating the need for electrophoresis steps or hybridization steps. A recent multi-center study involving countries with a high burden of TB including Tanzania, Peru, and Bangladesh was performed with technicians who had one week of LAMP test training, and the sensitivity of the LAMP test was 97.7% for smear- and culture- positive samples and 48.8% for smear-negative and culture-positive samples, and the specificity was 99%.10 Thus, the LAMP test looks promising for application in effective control programs in field settings with a high burden of TB.

Molecular tests for species identification of mycobacteria

Species identification of Mycobacterium species has become more important recently because of the steady increase in non-tuberculosis mycobacteria infections, particularly among HIV infected and elderly people and because of the wide use of liquid culture systems. Traditionally, species identification of mycobacteria relies on biochemical tests such as growth characteristics, pigmentation, the niacin test, the nitrate reduction test, the Tween 80 hydrolysis test, among others. Because these test procedures are often cumbersome, time consuming and inaccurate, several other methods such as HPLC analysis of mycolic acids,11,12 DNA probes,13,14 and sequence-based species identification systems have been developed and are widely used. HPLC analysis of mycolic acids provides a discrimination power of more than 50 species and can be run semi-automatically with minimum processing of cultures. One drawback of HPLC is the expensive instrumentation and the cost of machine and system maintenance. The AccuProbeR has been most widely used for species identification in laboratories that run a liquid culture system. At lease five species, including M. tuberculosis complex, M. avium, M. intracellulare, M. kansasii, and M. gordonae, constitute the majority of isolates from clinical samples.

With a marked increase in HIV infections, NTM infection has also increased as well and the number of Mycobacterium species has grown substantially. To cope with the increased number of Mycobacterium species, sequence information from the hypervariable region of the 16S rRNA of mycobacteria has been utilized for species identification.15,16 The polymorphic sites of the rpoB gene have been also used to differentiate among Mycobacterium species, and simple restriction enzyme analysis of the PCR products of the polymorphic sites makes it possible to differentiate more than 30 species.17,18 Oligonucleotide probes from the polymorphic site of the rpoB gene can also be used with a dot-blot hybridization assay for species identification.18 In addition, the polymorphic sequences of hsp65 can be successfully used for Mycobacterium species identification.19,20 Therefore, molecular techniques are more often used for Mycobacterium species identification and can be used to develop microarray-based assays in the near future.

One advantage of PCR-based methods over HPLC or AccuProbe is its direct applicability to clinical specimens, such as sputum samples, where the samples have enough bacilli to be AFB smear-positive. In the case of smear-negative and culture-positive specimens, the amplified products may not be enough to perform sequencing or hybridization with Mycobacterium species-specific probes. Since smear-positive specimens are indicative of transmitting agents, the rapid and correct identification of Mycobacterium species is very important.

Detection of mutations in the genes associated with resistance to TB drugs

The recent emergence of multidrug-resistant tuberculosis (MDR-TB)21 and extremely drug-resistant tuberculosis (XDR-TB)22 requires the rapid and accurate determination of drug-resistant M. tuberculosis. The recent emergence of MDR-TB may be due to easy access to the first-line TB drugs in most countries with a high prevalence of TB, including 22 high burden countries. The conventional methods for DST include proportion methods, absolute concentration methods, the drug resistance ratio, and the use of LJ and Middlebrook 7H10 agar. In addition, liquid culture systems, notably BACTEC MGIT960 and MB/BacT systems, have been evaluated extensively and seem to provide comparable results to the solid media-based tests with the advantage of rapid results.23 However, these culture-based methods still require 3-4 weeks including the original culture of organisms from clinical specimens. In order to reduce culture time, therefore, various molecular tests have been explored. Considering that the majority of drug resistance is due to mutations in genes encoding drug targets, one can develop molecular tests to identify gene mutations that are associated with drug resistance.24 The best examples are mutations in the rpoB gene, which are responsible for more than 95% of rifampin (RIF) resistance in M. tuberculosis. Some commercial hybridization assay kits, such as the INNO-LiPA Rif.TB kit25 and the GenoTyper MTBDR assay,26 can provide rapid results on RIF-resistance. A meta-analysis of the commercial probe assays gave a sensitivity of over 95% and specificity of 100% in 12 of 14 studies when cultured M. tuberculosis was used.27 However, there was a lower sensitivity of 82% to 92% and a specificity of 92% to 94% in the other two studies using the same commercial kits, indicating that careful interpretation of the results is required.

Likewise, resistance to isoniazid (INH) has been attributed to mutations in the katG, inhA, kasA, and/or ahpC genes of M. tuberculosis.24,28,29 However, the molecular mechanisms behind some INH- resistant M. tuberculosis isolates are not known. Other mutations in genes associated with resistance to other first line drugs and some second line drugs are listed in the Table 1. For EMB, while mutations at the embB codon 306 site have been implicated in causing resistance to the drug, only 40 to 70% of EMB-resistant isolates show mutations. Interestingly, mutations at the pncA gene match well with resistance to PZA because over 90% of PZA resistant isolates had mutations. The molecular mechanisms of drug resistance for most second-line drugs are only partially understood and the mutations that are responsible for the drug resistance have not been identified.

Table 1

Mutations in the Genes Associated with Resistance to TB Drugs

IMMUNODIAGNOSTIC TESTS

In general, immunodiagnostic tests can provide indirect evidence of current or past infections in organisms of interest. With the exception of the tuberculin skin test (TST), immunodiagnosis has not been widely used for chronic TB infections because of the low sensitivity and specificity. However, since immunological tests have the advantages of simplicity and rapid turnover over time, there have been tremendous efforts to develop serological tests for identifying people at risk of developing overt TB in the near future. In addition, since latent TB infection has become a hot issue for TB control programs, T cell-based assays such as the QuantiFERON test and T-SPOT.TB assay were developed as commercial kits and have been evaluated in field settings.

Antibody detection

Antibody-based immunodiagnostic tests have been widely explored for detection of TB. With the whole genome sequence of M. tuberculosis available, it is now possible to identify M. tuberculosis-specific antigens that can be used to detect antibodies. In the literature, there are numerous M. tuberculosis antigens, including native, semi-synthetic, and recombinant ones, that have been shown to be useful for serologic diagnosis of pulmonary and/or extrapulmonary TB. Some of the antigens such as lipoarabinomman, cord factor, A60, 38kDa, 16kDa have been used to develop commercial kits, and the kits have been evaluated in areas with a low and high burden of TB (Table 2). As noted in Table 2, the results for all commercial kits varied markedly with sensitivity ranging from 15.7% to 89.2% and specificity ranging from 50% to 100%. In general, higher sensitivity comes at the cost of lower specificity. One of the major problems in the serodiagnosis of TB has been heterologous antibody responses to various M. tuberculosis antigens. It seems, therefore, that more than one antigen is required for increased sensitivity. This approach may result in a lower specificity despite the use of M. tuberculosis specific antigens.

Table 2

Evaluation of Serological Test Kits for Diagnosis of Tuberculosis

ICT, immunochromatography test; TBGL, tuberculous glycolipid, cord factor.

The antibody detection method also evolved into various formats, although enzyme-linked immunosorbent assay (ELISA) is still the predominant one. Lateral flow tests or rapid tests, developed as commercial kits, have the advantages of rapidness and simplicity. However, subjective reading of the flow test results is a major drawback. Recently, a fluorescence polarization assay74 and surface plasmon resonance systems75 have been introduced for the detection of antibodies, which can be objectively read. However, the sensitivity and specificity have not been determined in serologic tests for TB. Likewise, protein chip-based assays are being developed that use a battery of M. tuberculosis antigens.76 Despite all the new technical developments in detecting antibodies for M. tuberculosis antigens, however, research should be conducted about the interpretation of the results for diagnosis or monitoring chemotherapy. A series of cohort studies is necessary to address these issues in the near future.

Antigen detection

Unlike antibody detection, M. tuberculosis-specific antigen detection provides direct evidence of current M. tuberculosis infection, thus warranting immediate chemotherapy against TB. However, since M. tuberculosis antigen-specific antibodies are used for the detection of antigens, there may be some intrinsic problems with non-specific positive results, even if for only a small fraction of them. Another limitation of antigen detection is the low sensitivity due to the scarcity of M. tuberculosis antigens in body fluids such as cerebrospinal fluid (CSF), pleural fluid, blood, and urine. Antigen detection formats include sandwich ELISA (sELISA) and dot-ELISA.

Target antigens vary markedly depending on the nature of the clinical samples and the test format. Lipoarabinomanna (LAM) has been most frequently used as the target antigen, and its monoclonal and polyclonal antibodies have been used to detect antigen in sputum samples (Table 3). One problem with using LAM as the target antigen is the non-specificity due to the presence in other mycobacterial species. Other antigens, such as whole cells, PPD, culture filtrate protein (CFP), 65kDa, 14kDa, among others, have been used as target antigens for antigen detection in CSF and sputum samples. Selection criteria for target antigens may include abundance, secretion, stability during specimen processing, and specificity to M. tuberculosis.

Table 3

Detection of M. tuberculosis Antigens in Clinical Samples for Diagnosis of Tuberculosis

CSF, cerebrospinal fluid; LAM, lipoarabinomannan; sELISA, sandwich ELISA; dELISA, direct ELISA; RPHA, reverse passive haemagglutination; RIA, radioimmunoassay; PPD, purified protein derivatives; CFP, culture filtrate protein.

Sensitivity and specificity of antigen detection varies depending on the clinical specimens and the target antigen. In CSF samples, the sensitivity ranged from 50%81 to 90%79 with the specificity ranging from 80%81 to 100%77 (Table 3). Interestingly, LAM antigen was detectable in 74-93% of urine samples from TB patients and in 4-13% of those from healthy controls in the study areas. Since urine samples are easy to obtain and minimal processing is required, the antigen detection assay is promising for the rapid and simple diagnosis of tuberculosis in areas with a high prevalence of TB. However, urine antigen detection methods need to be evaluated further before they can be fully implemented in TB control programs. In sputum samples, the sensitivity of antigen detection varied from 71% to 91% with a specificity ranging from 91% to 100% (Table 3). Due to the complexity of sELISA and sputum processing steps, however, further studies are necessary to develop simple and highly specific antigen detection methods. In order to overcome the limitation of non-specificity, analytical tools, such as liquid chromatography-mass spectrometry, MRI, and NMR, among others, need to be explored to determine if they can directly detect mycobacterial antigens in clinical samples.

IFN-γ assays

The tuberculin skin test (TST) has been widely used to detect latent infection of M. tuberculosis to justify preventive therapy.90 The major drawback of the TST is the positive reaction due to recent BCG immunization or exposure to NTM that are present in the environment. To overcome such non-specific positive reactions with TST, ex vivo T cell-based assays have been explored in many laboratories using M. tuberculosis-specific antigens such as ESAT-6, CFP-10, and TB7.7 antigens whose genes are located in the RD1 or RD11 region, which is deleted in BCG vaccines.91,92 There are two commercial kits available in the market that claim to provide efficient and specific detection of latent M. tuberculosis infection: the QuantiFERON tests (QFT), including QuantiFERON-Gold (QFT-G), and QuantiFERON-Gold In Tube (QFT-GIT), and the T-SPOT.TB (SPOT-TB) assays.93,94 QFT quantitatively measures the amount of IFN-γ released by effector T cells after a 16-24 hour exposure to M. tuberculosis-specific antigens or peptides derived from ESAT-6, CFP-10, and TB7.7 proteins. The SPOT-TB assay was designed to count the number of effector T cells producing IFN-γ after stimulating T cells with the ESAT-6 and CFP-10 antigens overnight.

Both assays were excellent in detecting latent infection of M. tuberculosis.93,95 However, there was a subtle difference in the results due to the nature of testing formats. Since QFT measures the total amount of IFN-γ in the culture supernatant, certain T cells producing a large amount of IFN-γ can influence the total amount, while other T cells may produce little IFN-γ. In contrast, the SPOT-TB assay counts the effector T cells producing any level of IFN-γ without knowing the total amount of IFN-γ. It is not known yet which measurements are most relevant for the diagnosis of active or latent M. tuberculosis infection (LTBI) in terms of sensitivity and specificity, and which measurement is the best indicator of effective chemotherapy in the early phase of treatment.

Both the QFT and the SPOT-TB assay have been evaluated for the diagnosis of active TB and detection of LTBI among TB patients and healthy controls from areas with low, medium, and high prevalence of the disease.93,95-97 In a meta-analysis of numerous studies on IFN-γ assays, the sensitivity was 0.76 (95% confidence interval, CI: 0.7-0.83) by QFT and 0.88 (95% CI: 0.81-0.95) by the SPOT-TB assay when active TB patients were used as the gold standard for LTBI, indicating that the sensitivity of the SPOT assay is greater than that of QFT.95 On the other hand, the specificity of the IFN-γ assays was very high at 0.97 (95% CI: 0.95-0.99) by QFT and 0.92 (95% CI: 0.88-0.95) by the SPOT-TB assay when healthy controls with a low risk of LTBI were used as the negative control, indicating a greater specificity with QFT than with the SPOT-TB assay.95 In comparison, TST showed a sensitivity of 0.71 (95% CI: 0.65-0.74) and a specificity of 0.66 (95% CI: 0.41-0.84). The results indicate that, despite the high specificity, the sensitivity of IFN-γ assays is suboptimal and may be due to immune suppression among TB patients, which was evident in patients with cavity formation.98 The specificity of TST was very low, largely due to BCG vaccination in the study populations.95

When healthy controls (n = 3,216) with a varying risk for LTBI were examined, the overall QFT positive rate was 26.6%, ranging from 5% in the U.S. to 56% in South Africa, while TST gave a positive rate of 45.6%, ranging from 9% in the U.S. to 81% in South Africa.95,99,100 In a Korean population comprised of close contacts of TB patients and healthy controls, the QFT positive rate was 23% compared with a TST positive rate of 64%.96 In comparison, the SPOT-TB assay gave an overall positive rate of 31.8% among healthy controls from various geographical areas (n = 2,916), while the positive rate by TST was 37.4%.95 When the two IFN-γ assays were compared using the same study subjects (n = 633), the overall agreement rate was 83%, ranging from 76% to 93%, with a discordant rate of 6-21%.95,101,102 In a contact tracing of TB patients, the overall agreement rate was 90%, but this could be increased to 94% if different cut-off values are used.103

Test failure and indeterminate results have been noted with IFN-γ assays. In a prospective study on 393 subjects who were exposed to TB patients or who were suspected of having TB, five samples failed with QFT and another five failed with the SPOT-TB assay, indicating a failure rate of 1.3% in each assay.102 Of 383 subjects with satisfactory results, 144 (38%) tested positive on the SPOT-TB assay and 100 (26%) by QFT. The lower positive rate with QFT may be due to a higher rate of indeterminate results, which occurred more frequently with QFT (11%) than with the SPOT-TB assay (3%).102 The indeterminate results were due to immunosuppression from underlying diseases such as cancer, diabetes, HIV infection, among others, and were found more frequently when subjects were TST negative.104

The IFN-γ assays have been employed for diagnosis of active TB and for monitoring patients undergoing chemotherapy. When both IFN-γ assays were employed for the detection of TB among people suspected of having the disease, the results lead to other medical examinations such as CT, which in turn verified the diagnosis of TB.105-107 In addition, IFN-γ assays have been analyzed for use in monitoring TB patient progress with chemotherapy treatment. While QFT failed to show a difference in values before and after chemotherapy in TB patients,108,109 the SPOT-TB assay showed a significant decrease in counts after the completion of chemotherapy in TB patients, indicating that the SPOT-TB assay can be used as a biomarker for treatment efficacy.110 In addition, QFT values decreased when individuals with LTBI were given INH for six months for chemoprophylaxis, indicating that QFT can be also be used to monitor the efficacy of prophylactic treatment of LTBI.111 Likewise, IFN-γ assays were employed to monitor subjects at high risk of M. tuberculosis infection in order to identify newly infected persons between test intervals. In a cohort study on 216 health care workers, Pai et al.112 showed that 12% of the cohort subjects were converted to positive by QFT in 18 months. In comparison, only 4.1% had TST positive conversion, indicating that QFT is more sensitive than TST in detecting new M. tuberculosis infection.

More research on IFN-γ assays is necessary before they can be fully applied to detect LTBI. Future research topics should include: (a) reproducibility or validation of the assays, including variation in test results due to the delay between blood collection and assays, intra-laboratory reproducibility, inter-laboratory reproducibility, endotoxin contamination during assays, antigen types (protein vs. peptides), among others; (b) influence of repeated skin testing113 and exposure to non-tuberculosis mycobacteria such as M. kansasii, M. marinum, M. szulgai,114 which share the antigens and peptides used in the IFN-γ assays; (c) effects of immunosuppression by HIV infection115,116 and other underlying diseases such as cancer, diabetes and other concurrent infections, like pneumonia;117 and (d) the positive predictive values of IFN-γ assay results for the progression of IFN-γ positive subjects to overt TB, which will require a large scale cohort study. In the end, the interpretation of results from IFN-γ assays will vary depending on several confounding factors associated with the individuals or populations of interest.