Comparison of Tuberculin Skin Test and New Specific Blood Test in Tuberculosis Contacts

Comparison of Tuberculin Skin Test and New Specific Blood Test in Tuberculosis Contacts Inger Brock, Karin Weldingh, Troels Lillebaek, Frank Follmann, and Peter Andersen Department of Infectious Disease Immunology, Statens Serum Institute; and International Reference Laboratory of Mycobacteriology, Statens Serum Institute, Copenhagen, Denmark The tuberculin skin test used to detect latent Mycobacterium tuberculosis infection has many drawbacks, and a new diagnostic test for latent tuberculosis (QuantiFERON-TB [QTF-TB]) has recently been introduced. This test measures the production of IFN- in whole blood upon stimulation with purified protein derivative (PPD). The QTF-TB test addresses the operational problems with the tuberculin skin test, but, as the test is based on PPD, it still has a low specificity in populations vaccinated with the Bacile Calmette-Guérin (BCG) vaccine. We have modified the test to include the antigens ESAT-6 and CFP-10, which are not present in BCG vaccine strains or the vast majority of nontuberculous mycobacteria. This test was used to detect infection in contacts in a tuberculosis outbreak at a Danish high school. The majority of the contacts were BCG-unvaccinated, which allowed a direct comparison of the skin test and the novel blood test in individuals whose skin test was not confounded by vaccination. An excellent agreement between the two tests was found (94%, value 0.866), and in contrast to the blood test based on PPD, the novel blood test was not influenced by the vaccination status of the subjects tested. Keywords: tuberculosis; latent infection; whole blood diagnostic; Bacile Calmette-Guérin vaccine; tuberculin skin test Tuberculosis (TB) is a major cause of morbidity and mortality throughout the world. Although 95% of cases and 97% of all deaths occur in high-endemic countries, the disease continues to be a problem in industrialized countries as well, mostly in immigrant populations, in elderly individuals with reactivating latent infection, and in local outbreaks (1). Targeted tuberculin skin testing (TST) and chemotherapy to prevent latent Mycobacterium tuberculosis infection from progressing to overt disease are important for tuberculosis elimination strategies in low-incidence countries. TST has many drawbacks, such as the need for patients to return for test reading, as well as variability and subjectivity in test application and reading. Most importantly, TST has low specificity as purified protein derivative (PPD), the antigen used for the test, is a mixture of mycobacterial antigens also present in nontuberculous mycobacteria and in the Bacille Calmette- Guérin (BCG) vaccine strains (2). Identification of genes in the M. tuberculosis genome that are absent from BCG vaccine strains and nontuberculous mycobacteria, has enabled the development of more specific tests for M. tuberculosis infection. ESAT-6 and CFP-10 are deleted from BCG Region 1 (RD1), and are not present in most nontuberculous mycobacteria (exceptions are M. kansasii, M. szulgai, and (Received in original form February 24, 2004; accepted in final form April 12, 2004) Supported by the Danish Lung Association and the European Community Program (Quality of Life and the Management of Living Resources) grant 2000 00630 (to T.L.). Correspondence and reprint requests should be addressed to Peter Andersen, D.V.M., D.M.Sc., Department of Infectious Disease Immunology, Statens Serum Institut, Artillerivej 5, DK-2300 Copenhagen S, Denmark. E-mail: pa@ssi.dk Am J Respir Crit Care Med Vol 170. pp 65 69, 2004 Originally Published in Press as DOI: 10.1164/rccm.200402-232OC on April 15, 2004 Internet address: www.atsjournals.org M. marinum). These antigens are highly specific indicators of M. tuberculosis infection (3), and have allowed precise diagnosis of active as well as latent TB in several studies with BCGvaccinated individuals (4, 5). In contrast, BCG vaccination significantly increases the likelihood of a positive TST in subjects without latent TB infection (6). Recently, the RD1 antigens used in an enzyme-linked immunospot assay were evaluated in a TB outbreak in Leicester, UK, and found to correlate significantly more closely to the level of exposure than did the TST (7). These studies have fully demonstrated the specificity of the RD1 antigens, but the widespread use of the BCG vaccine has prevented a direct comparison of the performance of these novel reagents and the TST that serves as the current gold standard for detection of latent TB. In the present study, we have compared the performance of TST and a whole-blood test based on ESAT-6 and CFP-10 (referred to here as QFT-RD1, but now commercially available under the trademark QuantiFERON- TB Gold) for detection of latent TB in individuals whose TST responses had not been confounded by prior BCG vaccination. In 2002, a case of TB in a community high school in Denmark was identified. A 17-year-old student had symptoms for 6 months before admission to hospital. The patient was TST-negative, but sputum microscopy and culture-positive. The Danish health authorities immediately initiated a contact-tracing investigation among 700 contacts and, in all, 37 TST-positive individuals were identified. In Denmark, BCG vaccination was stopped as standard procedure in the vaccination program during the late 1970s and early 1980s, and the majority of the subjects were, therefore, not BCG-vaccinated. We show that for contact tracing, the QFT- RD1 test has excellent agreement (94%) with TST in identifying contacts with latent TB infection among BCG-nonvaccinated individuals. In contrast to the blood test based on PPD, the QFT-RD1 test was not confounded by BCG vaccination. METHODS Subjects The 125 participating contacts were comprised of 85 BCG-unvaccinated (male/female, 45/40; mean age, 17 years) contacts and 40 contacts who had been vaccinated with BCG in childhood (male/female, 19/21; mean age, 45 years). The subjects were recruited as follows: all the nearest contacts of the index case were asked to participate. This high-exposure group contained individuals with close contact to the index case, either through the household, the school class, or the local choir that the index case regularly attended. The low-exposure group was comprised of 40 students from 2 other classes at the high school who had no connection to the index case. These classes were chosen by the head of the high school based on the schedules of the different classes. Furthermore, the low-exposure group contained 32 adults (including teachers from the other classes) with infrequent contact to the index case. They all volunteered on their own account. The remaining 700 contacts investigated by TST were all Danish students from other classes at high school with a similar age and sex distribution as the other students participating in the study. In all, 29 of the 37 contacts with a positive TST were studied.

66 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 170 2004 The division into the low- and high-exposure groups was in accordance with the division made by the local clinician before the contact investigation. TST was performed according to standard Danish procedures by a trained nurse from the Department of Lung Diseases at the local hospital: 2 TU of PPD (RT23; Statens Serum Institut, Copenhagen, Denmark) were injected intradermally into the dorsal aspect of the forearm and induration measured after 72 hours. Induration of greater than 10 mm was considered indicative of TB infection. TST guidelines vary in different countries, and, according to Danish policy, TST was not performed on BCG-vaccinated subjects. These vaccinated subjects were only screened for TB infection by chest X-ray and clinical examination. All unvaccinated subjects had their blood drawn for the QFT-RD1 72 hours after applying the TST. All participating subjects gave written, informed consent and were interviewed about exposure to the index case and BCG vaccination status. The study was approved by the local ethics committee for Copenhagen and Frederiksberg (KF-11 108/02). Antigens Recombinant CFP-10 and ESAT-6 were produced as previously described (8, 9). PPD was obtained from the Statens Serum Institut (tuberculin RT23), and phytohemagglutinin was provided with the QuantiFERON- TB Gold kit (Cellestis Limited, Carnegie, Australia). QFT-RD1 The whole blood assay was performed per the QuantiFERON-TB Gold kit instructions, but instead of using synthetic peptides, recombinant antigens and PPD were used. Briefly, heparinized blood (1 ml) was incubated (24 hours at 37 C) with 3 drops of saline (nil control), phytohemagglutinin (mitogenic-positive control), 5 g/ml PPD, 5 g/ml of ESAT-6, or 5 g/ml CFP-10. Plasma was collected and frozen until analysis, and IFN- concentrations were determined by ELISA (detection limit: 0.05 IU/ml) provided with the kit. For each subject, the nil control value was subtracted from the values of antigen-stimulated plasma samples. The cut-off value for a positive response to the specific antigens in the QFT-RD1 test and for PPD was established using data from 39 non M. tuberculosis exposed control donors and 26 culture confirmed TB patients (Pernille Ravn, unpublished data). From a receiver operating characteristic curve analysis based on the highest IFN- level produced in response to ESAT-6 and CFP-10, a cut-off value of 0.35 IU/ml of IFN- was established for use in the QFT-RD1 test, giving a specificity of 97%. This established cut-off value is the same as suggested in the QuantiFERON-TB Gold kit instructions when using peptides instead of recombinant antigens. From a similar receiver operating characteristic curve analysis on IFN- in response to PPD, an optimal cut-off value of 7.4 IU/ml of IFN- was selected. Statistical Analysis The utility of the QFT-RD1 assay was assessed using the proportion correctly classified (agreement) and measure of agreement, when compared with the TST results in BCG-unvaccinated individuals. RESULTS QFT-RD1 Analysis of Subjects The 125 participating subjects were divided into groups with high and low levels of exposure. Both the high- and low-exposure group contained unvaccinated as well as BCG-vaccinated individuals. All subjects were tested with the QFT-RD1 and QFT- PPD, and the IFN- levels given in IU/ml (Figure 1). For the QFT-RD1, many subjects in the high-exposure group gave strong responses to the RD1 antigens, and these responders were found both among vaccinated (4 of 8) and unvaccinated (24 of 45) subjects. In the low-exposure group, only four responders to the RD1 antigens were found (2 of 32 BCG-vaccinated and 2 of 40 unvaccinated). Among the BCG-vaccinated subjects, 3 of 8 (38%) in the high-exposure group, and 14 of 32 (44%) in the low-exposure group responded to PPD in the QFT assay. Therefore, in contrast Figure 1. QuantiFERON (QFT)-RD1 and QFT purified protein derivative (PPD) analysis of tuberculosis (TB) contacts. Bacile Calmette-Guérin (BCG)-vaccinated subjects (n 40, closed symbols) and unvaccinated subjects (n 85, open symbols) were divided into high- and low-exposure groups. Upper panel, whole blood was tested in the QFT-RD1 assay based on the specific antigens ESAT-6 and CFP-10. The highest response of the two antigens is shown for each contact. Cut-off value for a positive response to the specific antigens in the QFT-RD1 test was 0.35 IU/ml. Lower panel, whole blood was tested in the QFT-PPD assay based on the specific antigens ESAT-6 and CFP-10. Cut-off: 7.4 IU/ml. Cut-off is depicted as broken horizontal lines. to the QFT-RD1 assay, QFT-PPD did not discriminate between the high- and low-exposure groups in BCG-vaccinated subjects. For unvaccinated individuals, QFT-PPD showed the expected discrimination between the high- and low-exposure groups, with 47% (21 of 45) positive from the high-exposure group and only 5% (2 of 40) positive from the low-exposure group. All contacts whose TST or QFT-RD1 test results were positive were offered treatment of latent TB. Comparison between TST and the QFT-RD1 Test in Unvaccinated Individuals For the 85 BCG-unvaccinated subjects, the QFT-RD1 test results were compared with the size of the TST induration. A reaction of 10 mm or greater was used as an indication of infection. In the high-exposure group, both the TST (25 of 45) and the QFT-RD1 (24 of 45) identified more than half of the subjects as infected. Twenty-three subjects (51%) were positive in both tests, and 19 (42%) were negative in both tests. Two subjects were TST-positive and QFT-RD1 negative, and one subject had the inverse profile. In the high-exposure group, this gave an agreement between the two tests of 93% (95% confidence interval, 86 100%) (Figure 2B). In the low-exposure group, the majority (90% [36 of 40] were both TST- and QFT-RD1 negative). Two subjects (5%) were both TST- and QFT-RD1 positive, and two subjects (5%) were TSTpositive (10 15 mm) but QFT-RD1 negative (Figure 2C). This gave an agreement between the two tests in the low-exposure group of 95% (95% confidence interval, 88 102%), and an overall agreement between the two tests of 94% (95% confidence

Brock, Weldingh, Lillebaek, et al.: New Specific TB Blood Test 67 Figure 2. Comparison of tuberculin skin testing (TST) and QFT-RD1 in unvaccinated subjects. Unvaccinated subjects (n 85) were tested with QFT-RD1 and the values compared with those of TST (A). The subjects were classified into (B) high-exposure (n 45) and (C ) lowexposure (n 40) groups. Bars (y-axis) represent the number of subjects with different TST sizes, giving either a positive (black) or a negative (gray) QFT-RD1 test. The TST results were grouped for each fifth millimeter and the size presented on the x-axis. For each subject the highest response to ESAT-6 or CFP-10 was used and scored positive if 0.35 IU/ml or greater of IFN-. Two by two tables show the breakdown of the positive and negative subjects for the two tests. interval, 89 99%) in all subjects tested ( 0.866), indicating excellent agreement between the two tests (Figure 2A). Five subjects had discrepant QFT-RD1 and TST results. In the high-exposure group, two subjects were TST-positive and QFT-RD1 negative. One of these had a negative QFT-PPD response, and one had a positive QFT-PPD response. One subject had a negative TST and a positive QFT-RD1, and this subject also had a positive QFT-PPD response. In the low-exposure group, two subjects had a positive TST response and negative QFT-RD1 result. Both of these individuals had negative QFT- PPD test results. QFT-RD1 for the Detection of Infection in BCG-vaccinated Subjects Subjects in the BCG-vaccinated group were not skin-tested, but subjects from the high-exposure group were offered chest X-ray and clinical examination. This evaluation did not result in the detection of clinically recognizable TB in any of the subjects. For the high-exposure group, 50% of vaccinated subjects were positive by QFT-RD1, which gives the same overall prevalence as that of the unvaccinated component of this group (53%). For the low-exposure group, only 6% were positive in the QFT- RD1 test, almost in complete agreement with the prevalence of 5% found in the unvaccinated component of the low-exposure group (Table 1). The QFT-RD1 test, therefore, detected six infected subjects among the BCG-vaccinated subjects. Four of these QFT-RD1 positive subjects were found in the high-exposure group of close contacts. DISCUSSION With the exception of the TST and the PPD-based QFT test, current diagnostic assays for detecting M. tuberculosis infection are all based on the identification of the bacterium, which makes them inapplicable for diagnosis of latent infection. The development of the QFT-RD1 test to detect T cells specific for M. tuberculosis antigens, as described in the present study, addresses this important problem. Our study investigated a population of young BCG-unvaccinated individuals, all coming from a lowendemic area. Therefore, the TST is likely to be a good indicator of latent infection in recently exposed contacts. The school TB outbreak in a nonvaccinated population presented an excellent opportunity to test the performance of the QFT-RD1 test for detecting latent M. tuberculosis infection, using the TST as a proxy standard. We found that the agreement between TST and QFT-RD1 results was very high, with discordance in only 5 of 85 (6%) of non BCG-vaccinated subjects. The present study, therefore, demonstrates that the QFT-RD1 test is similar in performance to TST for detecting latent infection in young non- BCG vaccinated individuals, a population in which the TST clearly is a useful test. However, as the antigens used in the QFT-RD1 test are absent from all strains of BCG and from most nontuberculous mycobacteria (CFP-10 and ESAT-6 are deleted from all tested nontuberculous mycobacterial strains except M. kansasii, M. szulgai and M. marinum), the test is very accurate for detecting M. tuberculosis infection. TST, on the other hand, is confounded by BCG vaccination and exposure to nontuberculous mycobacteria (10 14). The limited specificity of a test based on PPD also characterizes in vitro assays, as demonstrated by attempts to develop serodiagnostic tests based

68 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 170 2004 TABLE 1. QUANTIFERON-TUBERCULOSIS RD1 AND TUBERCULIN SKIN TESTING RESULTS FOR VACCINATED AND UNVACCINATED SUBJECTS TST Positive* QFT-RD1 Positive Exposure Group BCG Status Total n n (% ) n (% ) High 45 25 (56) 24 (53) 8 ND 4 (50) Low 40 4 (10) 2 (5) 32 ND 2 (6) Definition of abbreviations: BCG Bacile Calmette-Guérin; ND not determined; TST tuberculin skin testing; QFT QuantiFERON-tuberculosis. * TST 10 mm. IFN- release 0.35 IU of IFN- /ml for at least one of the two antigens (ESAT-6 and CFP-10). on PPD (15), and, more recently, from the first generation QuantiFERON-TB test in which PPD is used to induce the secretion of IFN- from sensitized T cells in whole blood (16). This lack of specificity of PPD used in the blood test is also clearly demonstrated in the present study in which a large proportion of the BCG-vaccinated donors in the low-exposure group are positive for PPD, but negative to the RD1 antigens (Figure 1). Importantly, the test based on the RD1 antigens is at least as sensitive for detection of latent TB as the PPD-based test, as demonstrated by the similar number of responders (21 and 24 of 45, respectively) in the nonvaccinated high-exposure group. In addition to precise detection of M. tuberculosis infection, the QFT-RD1 has many other advantages over the TST. Objective quantitative results can be obtained the day after blood sampling, and time spent on return visits to have the TST read is spared. Furthermore, since no antigen is injected, the problem of a booster effect on sequential skin tests is avoided. The test is simple to perform, and can be used even in countries with less-developed infrastructure. Only 400 to 500 cases of TB occur annually in Denmark, and two thirds of these are found in immigrants from high-endemic areas (17). TB in a young Danish-born individual without any known risk factors is thus very rare, and this could, in part, be the explanation for the lag time between onset of symptoms and diagnosis. In the present study, the index case was left untreated with active pulmonary TB for more than 6 months, and transmitted the infection to at least 37 contacts during that period. This number is much higher than the previously estimated transmission rate of 10 15 contacts each year from an untreated individual with pulmonary TB (18). The high transmission rate may relate to the fact that this outbreak occurred in an institutional setting in which transmission to larger numbers of individuals could be expected. That the unrecognized transmission in institutional settings may result in rapid spreading of TB beyond what had previously been expected was recently emphasized by a TB outbreak in Leicester, UK, in which 69 active TB cases and 254 cases of latent TB were all tracked from a single index case left untreated for 9 months (7). In this study, T cell IFN- responses to ESAT-6 and CFP-10 were measured using an ELISPOT test and the results compared with the tuberculin Heaf test. The lack of a gold standard for detection of latent TB in this population, which contained a majority of BCG-vaccinated individuals, was addressed by correlating the performance of the tests with the degree of exposure to the index case expressed as proximity and duration of exposure to the index case (7). The conclusion was that the ESAT-6/CFP-10 based assay correlated significantly more closely to the level of exposure than did the TST, thereby confirming earlier reports of a low specificity of TST in a BCGvaccinated population (6). Together, these studies establish the RD1 antigens, ESAT-6 and CFP-10, as powerful reagents for the precise detection of latent M. tuberculosis infection, both in BCG-vaccinated and unvaccinated populations. To specifically detect and treat individuals with latent M. tuberculosis infection and thereby prevent progression to overt disease is an increasingly important part of the tuberculosis elimination strategy in low incidence countries (19). Our data suggest that QFT-RD1 may be a very valuable test for precisely this purpose. Conflict of Interest Statement : I.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this article; K.W. does not have a financial relationship with a commercial entity that has an interest in the subject of this article; T.L. does not have a financial relationship with a commercial entity that has an interest in the subject of this article; F.F. does not have a financial relationship with a commercial entity that has an interest in the subject of this article; P.A. does not have a financial relationship with a commercial entity that has a financial interest in the subject of this article; Statens Serum Institut is a government research institution that owns patents relating to the ESAT-6 antigen, and has out-licensed the use of ESAT-6 and CFP10 antigens for TB diagnosis. Acknowledgment : The authors thank Vita Elleby Skov, Thomas Okkels Thomasen, Lene Rasmussen, and Jolanta Kobusch for dedicated technical assistance; Kirsten Stax Jacobsen from Skive Hospital, Lene Skadholm, Anette Skovsted and Aase Hovmand from Thisted Hospital, and Medical Officer of Health, Flemming Stenz, for input and help throughout this study; and Timothy Mark Doherthy for critically reading the manuscript. References 1. Broekmans JF, Migliori GB, Rieder HL, Lees J, Ruutu P, Loddenkemper R, Raviglione MC. European framework for tuberculosis control and elimination in countries with a low incidence: recommendations of the World Health Organization (WHO), International Union Against Tuberculosis and Lung Disease (IUATLD) and Royal Netherlands Tuberculosis Association (KNCV) Working Group. Eur Respir J 2002; 19:765 775. 2. Harboe M. Antigens of PPD, old tuberculin, and autoclaved Mycobacterium bovis BCG studied by crossed immunoelectrophoresis. Am Rev Respir Dis 1981;124:80 87. 3. Andersen P, Munk ME, Pollock JM, Doherty TM. Specific immunebased diagnosis of tuberculosis. Lancet 2000;356:1099 1104. 4. Arend SM, Andersen P, van Meijgaarden KE, Skjot RL, Subronto YW, van Dissel JT, Ottenhoff TH. Detection of active tuberculosis infection by T cell responses to early-secreted antigenic target 6-kDa protein and culture filtrate protein 10. J Infect Dis 2000;181:1850 1854. 5. Lalvani A, Pathan AA, Durkan H, Wilkinson KA, Whelan A, Deeks JJ, Reece WH, Latif M, Pasvol G, Hill AV. Enhanced contact tracing and spatial tracking of Mycobacterium tuberculosis infection by enumeration of antigen-specific T cells. Lancet 2001;357:2017 2021. 6. Wang L, Turner MO, Elwood RK, Schulzer M, FitzGerald JM. A metaanalysis of the effect of Bacille Calmette Guerin vaccination on tuberculin skin test measurements. Thorax 2002;57:804 809. 7. Ewer K, Deeks J, Alvarez L, Bryant G, Waller S, Andersen P, Monk P, Lalvani A. Comparison of T-cell based assay with tuberculin skin test for diagnosis of Mycobacterium tuberculosis infection in a school tuberculosis outbreak. Lancet 2003;361:1168 1173.

Brock, Weldingh, Lillebaek, et al.: New Specific TB Blood Test 69 8. Berthet FX, Rasmussen PB, Rosenkrands I, Andersen P, Gicquel B. A Mycobacterium tuberculosis operon encoding ESAT-6 and a novel low molecular-mass culture filtrate protein (CFP-10). Microbiol 1998; 144:3195 3203. 9. Pollock JM, McNair J, Bassett H, Cassidy JP, Costello E, Aggerbeck H, Rosenkrands I, Andersen P. Specific delayed-type hypersensitivity responses to ESAT-6 identify tuberculosis-infected cattle. J Clin Microbiol 2003;41:1856 1860. 10. Edwards LB, Acquaviva FA, Livesay VT, Cross FW, Palmer CE. An atlas of sensitivity to tuberculin, PPD-B, and histoplasmin in the United States. Am Rev Respir Dis 1969; 99(Suppl):1 132. 11. Felten MK, van-der-merwe CA. Random variation in tuberculin sensitivity in schoolchildren: serial skin testing before and after preventive treatment for tuberculosis. Am Rev Respir Dis 1989;140:1001 1006. 12. Huebner RE, Schein MF, Bass JB. The tuberculin skin test. Clin Infect Dis 1993;17:968 975. 13. Lind A, Larsson LO, Bentzon MW, Magnusson M, Olofson J, Sjogren I, Strannegard IL, Skoogh BE. Sensitivity to sensitins and tuberculin in Swedish children: a study of schoolchildren in an urban area. Tubercle 1991;72:29 36. 14. von Reyn CF, Horsburgh CR, Olivier KN, Barnes PF, Waddell R, Warren C, Tvaroha S, Jaeger AS, Lein AD, Alexander LN, et al. Skin test reactions to Mycobacterium tuberculosis purified protein derivative and Mycobacterium avium sensitin among health care workers and medical students in the United States. Int J Tuberc Lung Dis 2001;5: 1122 1128. 15. Daniel TM. Rapid diagnosis of tuberculosis: laboratory techniques applicable in developing countries. Rev Infect Dis 1989;11:S471 S478. 16. Brock I, Munk ME, Kok-Jensen A, Andersen P. Performance of whole blood IFN-gamma test for tuberculosis diagnosis based on PPD or the specific antigens ESAT-6 and CFP-10. Int J Tuberc Lung Dis 2001; 5:462 467. 17. Lillebaek T, Andersen AB, Bauer J, Dirksen A, Glismann S, de Haas P, Kok-Jensen A. Risk of Mycobacterium tuberculosis transmission in a low-incidence country due to immigration from high-incidence areas. J Clin Microbiol 2001;39:855 861. 18. World Health Organization. Tuberculosis. Fact Sheet No. 104. 2000. Available from http://www.who.int/mediacentre/factsheets/who104/en/ (accessed January 2004). 19. Gomez JE, McKinney JD. M. tuberculosis persistence, latency, and drug tolerance. Tuberculosis (Edinb) 2004;84(1 2):29 44.