You are on page 1of 9

This article was published in an Elsevier journal.

The attached copy
is furnished to the author for non-commercial research and
education use, including for instruction at the author’s institution,
sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:

ac. All rights reserved.0132). London.mcilleron@uct. Watkinsb.R. 557–564 Available at www.d a Department of Medicine.1016/j. UK Received 14 May 2007. Wright Fleming Institute. All rights reserved. In multivariate analysis of parameters at 2 months the strongest positive associations with disease free survival were IFN-g response to PPD (p ¼ Rifampin levels. Tel.001). Division of Medicine. E-mail addresses: helen. pyrazinamide and ethambutol were determined by high-performance liquid chromatography. Folb).see front matter & 2007 Elsevier Ltd.2007. Division of Clinical Immunology. Institute of Infectious Diseases and Molecular Medicine. Peter I. Cape Town.wilkinson@imperial. Pharmacokinetics Summary Factors that relate to medium-term outcome in patients with pulmonary tuberculosis (PTB) who have completed the 2-month intensive phase of treatment are incompletely understood. peter. Groote Schuur Hospital. fax: +27 21 448 1989.40–0. 7925.002) and serum creatinine (p ¼ 0. Imperial College London. 1472-9792/$ (P. to outcome at 24 months was studied prospectively.: +27 21 406 6292. University of Cape Town. received in revised form 27 July 2007. interferon-gamma release and outcome in complicated pulmonary tuberculosis Helen McIllerona. Robert J. r.folb@mrc. South Africa d Wellcome Trust Center for Research in Clinical Tropical Medicine. University of Cape Wilkinsonc. South Africa b Department of Medicine. & 2007 Elsevier Ltd.54 (95% CI: 0. isoniazid.elsevierhealth. The relationship between in vitro production of interferon-gamma (IFN-g). interleukins (ILs)-5 and -10 and drug levels determined after 2 months of drug therapy. tuberculosis may influence treatment outcome. Drug concentrations were not associated with outcome although rifampin exposure correlated with IFN-g response to PPD (p ¼ 0. Cytokine concentrations were determined from culture supernatants after stimulation of whole blood with purified protein derivative (PPD) of Mycobacterium tuberculosis. Faculty of Health (R. Observatory.sciencedirect. Cytokines. Marcia L. McIlleron). marcia. Ress). University of Cape Town.watkins@uct. K-45 Old Main Buliding. Watkins). South Africa c Department of (M. Plasma concentrations of rifampin. stan. Corresponding author. The treatment failure and relapse free survival probability was ( accepted 10 August 2007 KEYWORDS Mycobacterium tuberculosis.J.67) at 24 (H. These data suggest that the ability to mount a recall immune response to M.L. Wilkinson). Ressb. Western journal homepage: http://intl. Stanley R. Groote Schuur Hospital. The data support the idea to identify persons at risk of a poor treatment outcome by monitoring of the in vitro response to tuberculosis antigens.002 . Division of Clinical Pharmacology..Author's personal copy ARTICLE IN PRESS Tuberculosis (2007) Folba.

pyrazinamide and ethambutol. overcrowded living conditions. . debility. The principal control measure is detection and treatment of infectious cases of pulmonary tuberculosis (PTB). A rapid method to predict poor prognosis could contribute to increased vigilance thus improving cure rates. the results of which may not become available until many weeks after the patient has left the inpatient facility and possibly lost to follow-up. including antibody. Daily antitubercular treatment doses based on body weight were directly observed by hospital staff.11 There is considerable interest in the possibility of assisting diagnosis and monitoring therapy via in vitro detection of interferon-gamma (IFN-g) released in response to Mycobacterium tuberculosis antigens. Germany). cytokine. serum creatinine and plasma concentrations of rifampicin. cellular proliferative. change during successful treatment of tuberculosis. Participants provided written consent for inclusion. immunological and pharmacological determinations at the 2-month time point to outcome at 24 months. participants were actively followed-up through the regional clinics using 3.Author's personal copy ARTICLE IN PRESS 558 H.1 drug sensitive disease also presents treatment difficulties requiring inpatient management. We related clinical. Patients with disease relapse had a positive sputum by microscopy or culture after treatment completion. severe or complicated disease and poor socioeconomic circumstances. and advanced age or immunosuppression can contribute. They were referred to hospital for reasons including poor response to treatment. However. The significance of. suspected non-adherence. Rifampinbased multi-drug regimens administered as part of a directly observed therapy (DOTS) strategy to ambulatory patients are cost effective albeit with the need to be administered for at least 6 months. these observations are discussed. Tuberculosis is a globally important cause of mortality. the latter is not always accompanied by culture conversion. until 24 months after their admission to the hospital. tuberculosis. Following discharge after 2 months of treatment in hospital.11–13 Materials and methods The study was approved by the University of Cape Town Research Ethics Committee (REC 122/99). 142 patients ↓ cytokine levels were evaluated in 112 patients ↓ 4 failed treatment hiv-infection retreatment male 0 4 2 ↓ 12 died ↓ ↓ ↓ 4 died early (≤ 6 months) 8 died later (7 to 24 months) 12 relapsed 1 4 2 3 5 1 3 6 7 ↓ 22 had a good outcome at 24 months 3 13 7 ↓ 62 lost to follow up 4 39 31 Figure 1 Summary of recruitment and of cohort outcome. cytokine responses to PPD. serum albumin. Determinations at the 2 month time point included demographic and treatment factors. isoniazid. although the subsequent relapse rate is higher. and acute phase responses. was conducted in accordance with the Helsinki Declaration of 1975 as revised in 1983 and was performed in collaboration with the regional health authorities. Introduction We evaluated factors that may influence the medium term outcome of PTB in a cohort of 142 patients requiring hospitalisation for drug sensitive PTB in an area of South Africa with a very high incidence of tuberculosis. some subgroups do badly and require hospitalisation at least for the initial 4-drug phase of therapy. HIV status (AxSYM HIV Ag/Ab Combo. The strongest predictor of outcome was the level of IFN-g released into culture in response to purified protein derivative (PPD) of M. and practical possibilities raised by. Abbott Diagnostics. McIlleron et al.8. This level in turn correlated with the simultaneously determined area under the curve (AUC) for rifampin. Unemployment. Patients from whom M.and then 6-monthly sputum smear microscopy and culture of a single sputum specimen. and clinic records. extensive or recurrent disease. underlying lung disease. but no other antitubercular drug. A variety of immune parameters. Whilst the increasing incidence of drug resistant disease is a major concern. The cure rate determined by sputum clearance is a key indicator of the performance of treatment programmes. relapse or death and who were not lost to follow-up at the end of the observation period.2–5 The bioavailability of drugs during therapy is important but not routinely ascertained in patients. tuberculosis resistant to rifampin or isoniazid had been isolated were excluded. This assessment is classically based upon sputum smear conversion.10 The extent to which the changes represent phenotypic change in the immune response or recirculation of cells from disease sites is also unclear. However. Patients (n ¼ 142) admitted to the regional tuberculosis hospital with PTB 2 months previously were sequentially enrolled (Figure 1). In these circumstances it is useful to know whether therapy is effective such that the patient may be discharged to ambulant DOTS care after completing 2 months of treatment.6–9 Whether the immune response thereby contributes to the success of therapy is conjectural: an equal cure rate for uncomplicated tuberculosis is observed in HIV-infected persons. lymphocyte subset counts. Patients with a good clinical outcome at 24 months were defined as those without treatment failure. smoking.

relapse or death and for those lost to follow-up during the study period. Figure 3). Antitubercular drug concentrations were not associated with the 24-month outcome. UK) at a final concentration of 5. Hialeah.050).0000.75  103 mitogenic units/ml. and 0. The lower limit of detection of the assays was 8 pg/ml. Immunology. San Diego.001. p ¼ 0. using CD3RD1 and either CD4-FITC or CD8-FITC markers [Coulter. it assumed a value of 0 pg/ml. Sixty-two observations (55%) were censored as the patients were alive and did not have either positive sputum microscopy or culture results at the time of their last study visit.002. Supernatant concentrations of IFN-g.1 g/l. pyrazinamide and ethambutol were determined over 8 h to generate the AUC08 as previously described. 33. but not the CD8+ lymphocyte count or the IL-10 response to PPD. To determine the cytokine response.14 Dual parameter flow cytometric analysis (performed on an Epics Profile II [Coulter]. but not the CD8+ count. whole blood diluted at a final concentration of 1:5 in RPMI 1640 (BioWittaker. especially IL-5 (Table 2). For the multivariate analysis. drug concentrations and serum albumin and creatinine as biochemical markers of nutrition and disease severity) with the 24-month outcome. Table 3. The response to PPD was correlated with the response to mitogen for all three cytokines (Spearman’s rho ¼ 0. relapse or death 12. tuberculosis (Central Veterinary Laboratory. The evaluated patients tended to be slightly older than those not included in the analysis (median 36 years.7 for albumin. 31. p ¼ 0.2587. the IL-5 response to PPD correlated with the CD4+ lymphocyte count (Spearman’s rho ¼ 0.54 (95% CI: 0.08). 67% retreated and 45% vs. Higher creatinine levels were also associated with more favourable outcome (p ¼ 0.56–92. variable selection was initially by an automated backwards stepwise process (variables with pX0.82) times greater than those with IFN-g responses in the highest quartile (44554 pg/ml). with low or absent responses in the plasma of many patients. Analysis Stata version 8.03 (95% CI: 1.3378.5 pg/ml) had a risk of treatment failure. . Cox proportional hazards models were used to report associations of the covariate factors (age. 0. 47% male. p ¼ 0. Kaplan–Meier survival analysis described tuberculosis-free survival from 2 until 24 months after hospital admission using time to the first positive sputum smear or culture (in cases diagnosed with treatment failure or relapse).7096. Figure 5).9996 reflects the risk reduction associated with a 1 pg/ml elevation in the IFN-g response. respectively). p ¼ 0. lymphocyte subset counts. Rifampin levels and IFN-g responses were positively correlated (Figure 4).4620. likewise albumin and creatinine concentrations were closely similar between the two groups (mean 34. isoniazid. Variables that were related to outcome are summarised in Table 1 for patients with tuberculosis-free survival. The model assumptions of proportionality and the appropriate form of the covariates in the model were checked using Schoenfeld residuals’ distributions.5 years. Male sex was associated with unfavourable outcome (Figure 4). death or relapse by 24 months. p ¼ 0. interleukin (IL)-5 and IL-10 were determined by ELISA using commercially available antibody pairs (BD PharMingen. The in study survival probability at the end of 24 months was 0.. and median 74 mmol/l. Results Recruitment and clinical outcome data are summarised in Figure 1. IQR: 26–40. The PPD stimulated increments were calculated by deduction of the levels detected in the unstimulated blood samples.0002). IQR: 65–83 vs. A creatinine concentration elevation of 5 mmol/l was associated with a 26% reduction in the risk of treatment failure.2291.Author's personal copy ARTICLE IN PRESS Rifampin levels. computed levels below this were assigned 8 pg/ml. for patients with treatment failure. and retained if po0. SD: 6. There was. Observations were censored if the patient was alive and not sputum-positive at the time of loss to follow-up. CA). although in contrast to the associations of the IFN-g response and serum creatinine. p ¼ 0.0 g/l.2535. A positive control for the whole blood culture was provided by phytohaemaglutinin (PHA: Wellcome Research Laboratories.0000 for the IFN-g. IFN-g release and outcome in complicated pulmonary tuberculosis Blood assays The steady-state plasma concentrations of rifampin. thus the relative hazard reduction for treatment failure. Florida]) and PPD stimulated cytokine levels were determined in 112 patients.0004) and the CD3+ lymphocyte count (Spearman’s rho ¼ 0.3514. demonstrated that those with IFN-g responses in the lowest quartile (p447.40–0. death or relapse associated with a 1000 pg/ml elevation is 31%. HIV status. or death. College Station. SD: 4.0070) and the IL-5 response to PPD (Spearman’s rho ¼ 0. The IL-10 response did not correlate with any T-cell subset. If the result was negative. the CD3+ lymphocyte count (Spearman’s rho ¼ 0. IL-5 and IL-10 responses. Beckenham. sex. and 14 HIV-infected patients were evaluated compared to none in the group not evaluated. p ¼ 0. Similarly.0 vs. the strongest predictor of a good outcome was the whole blood IFN-g response to PPD stimulation at 2 months (p ¼ 0.5 mmol/l.67). correlation between rifampicin exposure and the IFN-g response to PPD stimulation (Spearman’s rho ¼ 0. The hazard ratio of 0. The associations of IFN-g response to PPD stimulation and creatinine were preserved if the data were reanalysed using the assumption that the 62 patients lost to follow-up had a good clinical outcome at 24 months.055 being removed from the model.4989. p ¼ 0. TX 77845) was used to compute summary statistics and for statistical modelling. Cambrex) was cultured with PPD from M. The proportions of 559 retreatment and male patients were similar between the evaluated and unevaluated groups (63% vs. Logistical constraints prevented cytokine evaluation in 30 patients. ‘new’ or ‘retreatment’ category. UK) at a final concentration of 3 mg/ml for 5 days at 37 1C in a humidified atmosphere containing 5% CO2.0182). Figure 2).235.013. this finding was not supported by the univariate analysis. p ¼ 0. respectively). IQR: 65–84). followed by a forwards stepwise procedure based on the contribution of individual variables to the overall fit of the model. In multivariate analysis. A univariate analysis including only those patients with the smallest and biggest IFN-g responses to PPD.0059). p ¼ 0. median 73. The distribution of cytokine responses to PPD stimulation were skewed. cytokine responses to PPD.0000. however. IQR: 29–46 vs.2 (Stata Corp. The IFN-g response to PPD was correlated with CD4+ lymphocyte count (Spearman’s rho ¼ 0. Weybridge.

3524(0. 1534) (43%) (68%) (25%) (28.688) 0.35.9966–1.9 298. 20) (587.99 vs. 1.5 32 70 15. 35. 0.2 1828 299 0 706 437 1369 31 39 4 37 36 74 20.3) (584. 690) (1121. 28. 335. (0. 38) (73.64–22. 31.0070) 1.0038) (0. 83) 19.040.4647 (0. 310. Table 1 The patient characteristics for those with tuberculosis-free survival at 24 months (good outcome). 35) (61.85) (238.0452).0000) (0. 906) (305.8257–0.9984–1. 31. relapse or death during the study period using Cox proportional hazards models.04 (9.27–19.79) 292.8 24. 38) (65. 960) (216.519.1 25. 4554) 242 (68.56 (245.40) 1.9926) 1. 909) (332.9994 0. 23. 1723) (50%) (63%) (6%) (29. relapse or death at 24 months (poor outcome) and for those lost to follow-up during the study period.2) (15. The median AUC08 for rifampin was significantly reduced amongst those with IFN-g responses in the lowest quartile in comparison to the remaining patients (13.80.299. The strongest .5 (31. All patients underwent voluntary counselling and testing. 786) (1046.9) (22.9410–1.9801–1.8 (493.21.7225) 0. None of the patients were treated with antiretroviral drugs.8860 (0. Discussion We have investigated factors determined at 2 months that associate with subsequent 24-month outcome amongst hospitalised patients with PTB in South Africa.4 19. 41.0835) 0.7) (20.112 0.3. 4736) (127.27) 18.009732 (0.3)  Expressed as median (interquartile range) unless indicated otherwise.7. 95% CI: 17.9951–1.346 0. 46) 34.690 ( respectively. 95% CI: 6. 1729) (68.64. McIlleron et al. 28. p ¼ 0.9998 (0.434 0.3 293. 421) 0 (0.6621–3. y z Previously treated for tuberculosis or received 41 month of antitubercular treatment prior to admission.9644 (0.15) (18.736 0.0011) 0.9994–0.0289) 1. 20.58) 26. 0. 11) (352.6350–2. 456) (0.9850 (0. 47) (33.8) (247. 5505) (47.9997 0.002 0.0015 (0. 82) (14.8805) 1.055 0.9999) (0.409 (0. Covariate Poor outcome (n ¼ 28) IFN-g response to PPD (pg/ml) IL-10 response to PPD (pg/ml) IL-5 response to PPD (pg/ml) CD4+ lymphocyte count (  109 l1) CD8+ lymphocyte count (  109 l1) CD3+ lymphocyte count (  109 l1) Male sex Retreatmentmy HIV infectionz Age (years) Albumin (g/l) Creatinine (mmol/l) Rifampicin AUC (mg h/l) Isoniazid AUC (mg h/l) Pyrazinamide AUC (mg h/l) Ethambutol AUC (mg h/l) 488 131 0 624 387 1227 12 19 7 34.3 28.38.8505–4.5. 90) (4. 24. Covariate n Summary statistics IFN-g response to PPD (pg/ml) IL-10 response to PPD (pg/ml) IL-5 response to PPD (pg/ml) CD4+ lymphocyte count (  109 l1) CD8+ lymphocyte count (  109 l1) CD3+ lymphocyte count (  109 l1) Male sex Retreatment HIV infection Age (years) Albumin (g/l) Creatinine (mol/l) Rifampicin AUC (mg h/l) Isoniazid AUC (mg h/l) Pyrazinamide AUC (mg h/l) Ethambutol AUC (mg h/l) 112 112 106 112 1092 (448. 35.5) (26.5%) 36 (29.663) 0.3 19.0484) 1.503 112 0.49) (33.580 0. 17) 0.394 112 112 112 112 112 110 111 111 111 100 50 males (45%) 71 (63%) 14 (12.340 0. 335. 627) (803.788 Hazard ratio (95% confidence interval) for univariate cox proportional hazards models p-value  Expressed as median (interquartile range) unless indicated otherwise.795 0.1.1) (253.004 (0. 4) (522.9985–1.0053 (0.805 0.0.6) Good outcome (n ¼ 22) Lost to follow-up (n ¼ 62) 2642 275 0 668 433 1276 7 13 3 37 35 81 11.731 112 1. 38) 74 (65.49 (20.9549–1.9996 (0.8.51 mg h/l. 1742) (32%) (59%) (14%) (26.9371–0.1.6) (15. 379.9506) 0.6. 236) (0.9 280.924) 0.85 mg h/l.9997 (0.84 (15.Author's personal copy ARTICLE IN PRESS 560 H. 518) (0.0004) 0. Table 2 Summary of covariates in the study participants and the univariate associations of each with tuberculosis treatment failure. 76) (7. 21.014 0.9989–1.9640–1.1 18.0007) 0.001 0. 22.298 (1.9983 0.0160) 0. 36.9961–1.8 (168. for patients with treatment failure.3) (16.2404) 2.

IFN-g release and outcome in complicated pulmonary tuberculosis association was between the IFN-g response and PPD in whole blood. o75th centile [74.00 0 5 10 15 20 25 analysis time Figure 3 Kaplan–Meier plot describing the proportion of patients remaining free of tuberculosis treatment failure.001 0.9081. o83 mmol/l].05) with tuberculosis treatment failure. by creatinine 1.2466.75 0. Kaplan-Meier survival estimates.9994.9999) (pg/ml) Creatinine (mmol/l) 0.25 0.25 IFN_gamma = 50th to < 75th centile IFN_gamma = 25th to < 50th centile IFN_gamma = />75th centile 0.9742) Male sex 3. Although drug levels were not associated with outcome. interferon-g response X50th centile.Author's personal copy ARTICLE IN PRESS Rifampin levels. disease relapse or death during the study period. tuberculosis and obviously contains a wider range of antigens than the region of difference 1 antigens (ESAT-6 and CFP-10) upon which the commercial assays rely. o4554 pg/ ml].9406 (0. commercial detection of Table 3 The multivariate Cox regression model best describing the proportional hazards of covariate factors associated (at the significance level of 0.9996 (0. disease relapse or death during the study period. o1092 pg/ml]. Firstly. Secondly. o74 mmol/l].015 Kaplan-Meier survival estimates. the stimulus we employed was PPD which is not specific for M. pyrazinamide and ethambutol were not associated with this variable. and interferon-g response X75th centile [4554 pg/ml]) after 2 months of inpatient treatment with rifampin based regimens. . o50th centile [448. o75th centile [1092. o50th centile [65. 0. by IFN_gamma 1. relapse or death during the study period.75 0.002 0.7007) 561 0. the AUC08 for rifampicin correlated with the IFNg response. and serum creatinine X75th centile [83 mmol/l]) after 2 months of inpatient treatment with rifampin based regimens.00 0. TB and QuantiFERON-TB assays that are being extensively researched. 7. serum creatinine X25th centile. 0. whereas the levels of isoniazid. serum creatinine X50th centile.00 0 5 10 15 20 25 analysis time Figure 2 Kaplan–Meier plot describing the proportion of patients remaining free of tuberculosis treatment failure.50 creatinine = < 25th centile creatinine = 50th to < 75th centile creatinine = 25th to < 50th centile creatinine = />75th centile 0. Covariate factor Hazard ratio (95% confidence interval) pvalue IFN-g response to PPD 0. The form of IFN-g release assay that we employed differs in important respects from the commercially available T-Spot.00 0.0983 (1. by the serum creatinine concentration (serum creatinine below 25th centile [65 mmol/l].50 IFN_gamma = < 25th centile 0. interferon-g response X25th centile. by the interferon-g response (interferon-g response o25th centile [448 pg/ml].

50 0. by sex. p ¼ 0.00 0.25 sex = Female 0.Author's personal copy ARTICLE IN PRESS 562 H.8.235.75 0.013) in the cohort of 112 patients after 2 months of inpatient treatment for pulmonary tuberculosis.h/l) 60 40 20 0 0 5000 10000 IFN-gamma response to PPD (pg/ml) 15000 Figure 5 Scatter plot illustrating the association between rifampicin AUC08 and the IFN-g response to PPD stimulation (Spearman’s rho 0. by sex 1. We also determined the level of IL-5.16 It is difficult therefore to extrapolate our findings to what might be expected in a study where overnight assays were employed.15 although there are exceptions. as type 2 T cell activation has been postulated to be a determinant of tissue damage in tuberculosis.8. there is a risk that relying solely on the overnight production of IFN-g to a highly restricted panel of species-specific antigens might poorly inform studies that aim to understand immunity to tuberculosis. McIlleron et al. disease relapse or death during the study period.17 In the majority of patients no response was detected (Figure 3) and the level of IL-5 did not associate with an adverse outcome. theoretically allowing the differentiation of memory cells to produce cytokines. IFN-g is based upon overnight stimulation of effector cells either in the form of whole blood or PBMC. 80 rifampin AUC (mg. Although large studies are few. By contrast the proliferative and IFN-g response in longer term assays tends to increase.16. Whilst there is an understandable desire amongst the research community to standardise on assays that are comparable between studies and subject to rigorous quality control.00 0 5 sex = Male 10 15 20 25 analysis time Figure 4 Kaplan–Meier plot describing the proportion of patients remaining free of tuberculosis treatment failure. it appears that the rapid effector response that is detected by overnight assays tends to fall during successful treatment of tuberculosis. Our assay was of 5 days duration. IL-10 has also been implicated in anergic responses18 and PPD stimulated levels of this cytokine tend to fall during treatment. Kaplan-Meier surviva lestimates.19 The lack of relationship between the level of IL-10 secretion and CD4 or CD8 lymphocyte count tends to suggest that a significant .

failure to respond to clinic-based treatment). Wright A. The analysis may also have been underpowered for some covariates: for example highly variable continuous covariates (e. Most participants had low creatinine levels (90% of participants had creatinine in the range 52–97 mmol/l. An important role for IFN-g. variability attributable to 563 adherence was not accounted for. Banavalikar JN. EuroTB Correspondents. the rates of relapse and death might be exaggerated due to the substantial number of censored individuals (62 or 55%) who may have had a good treatment outcome. Acknowledgements We thank the staff and patients of Brewelskloof Hospital. The overall outcome for the region (Boland/Overberg) of patients at 24 months after treatment initiation is not known. et al. for performing flow cytometric analysis. 4. Coker R. Int J Tuberc Lung Dis 2005.21 The radiographic extent of disease was not measured. Lai KN.8:70–3. being approximately equal to the number of patients receiving treatment for the first time (Table 2). Dr. relapsed or died within the 24-month period following their admission to the hospital. the immune response may be more important than the sterilising activity of the antitubercular regimen for prevention of disease recurrence in the months following treatment in a community with high rates of re-infection. Belghiti F. Wilkinson RJ. normal range: 75–115 mmol/l) and few patients had chronic renal impairment which one would expect to be associated with a poor rather than good outcome. Infuso A.70:171–8.170: 1124–30. Peripheral blood T lymphocyte subpopulations in patients with tuberculosis and the effect of chemotherapy. Funding: The Medical Research Council of South Africa supported this project through a Self-initiated Research Grant. Vordermeier HM. Zignol M. Am J Respir Crit Care Med 2004. Atun R. An interesting association between serum creatinine and positive outcome was observed. concomitant illness and malnutrition.g. 8. We therefore ascribe the relationship to the higher creatinine values reflecting better nutritional status. 5. was conducted in accordance with the Helsinki Declaration of 1975 as revised in 1983 and was performed in collaboration with the regional health authorities. Therefore. rifampicin levels). Whether greater bioavailability of rifampin promotes a stronger IFN-g response is not clear from this study. 6. The region has a very high incidence of tuberculosis (41%) and it is also possible that re-infection with a new strain of M.g. Competing interests: None of the authors are aware of any potential conflict of interest. Notwithstanding this. A nested case–control study on treatment-related risk factors for early relapse of tuberculosis. Risk factors for pulmonary tuberculosis in Russia: case–control study. It is difficult therefore to quantitatively dissect cause from effect in human studies but our data support the idea that protein supplementation during the course of therapy of tuberculosis may improve the outcome. McKee M. Leung CC. Saha K. but the proportion of retreatment patients in our study was high.Author's personal copy ARTICLE IN PRESS Rifampin levels. 3. Leung CC. Tubercle 1989. The study was underpowered to detect any effect of rifampin concentration on outcome at 24 months. The possibility that the cytokine could be used as a surrogate marker to select persons at risk of a poor outcome for further intensive intervention is also raised and deserves investigation. Li T. Protein energy malnutrition predisposes to experimental tuberculosis20 and tuberculosis is a well-recognised associate of wasting syndromes in humans. the results should not be generalised without caution. Eur Respir J 1995. Smoking and tuberculosis among the elderly in Hong Kong. Ho AS. Falzon D. University of Cape Town for assistance with this work and helpful discussions. 25% of the 112 patients are known to have failed treatment.332:85–7. such as poor treatment adherence. Mustapha Keraan of the Division of Clinical Immunology. Groote Schuur Hospital and University of Cape Town. Tam CM. et al. Global incidence of multidrug-resistant tuberculosis. 7. However. Lai CK. Department of Medicine. Patients able to mount a stronger IFN-g response after 2 months of treatment also had higher AUC08 values for rifampin.194:479–85. those who were admitted to Brewelskloof Hospital for a variety of reasons which might predispose to a poor prognosis. RJW holds a Wellcome Trust Senior Fellowship (072070).9: 1224–9. Hosseini MS. Yew WW. HIVinfection status). Chan CH. Peptidespecific T cell response to Mycobacterium tuberculosis: clinical . Singhal M. Moreover. Chang KC. appears to underpin this. our study supports and extends observations that suggest that an immune component may contribute to successful resolution of both latent and active tuberculosis. References 1. Wilkinson KA. 2. Ethical approval: The study was approved by the University of Cape Town Research Ethics Committee (REC 122/99). Leung JC. Factors therefore contributing to the poor outcome we observed undoubtedly included selection of patients intrinsically at high risk of treatment failure (i. J Infect Dis 2006. Professors Gary Maartens and Peter Smith of the Division of Clinical Pharmacology.e. 170:1027–33. BMJ 2006. et al. et al. The extents to which IFN-g response and serum creatinine were associated with outcome independently of disease extent were therefore not addressed. in particular protein balance. Am J Respir Crit Care Med 2004. Ho SC. Lam TH. Our study has a number of limitations. tuberculosis could have occurred in a substantial proportion of patients as has been shown in another Western Cape population with a lower overall incidence of tuberculosis. or dichotomous covariates with only a few subjects in one group (e. Exploring the determinants of treatment success for tuberculosis cases in Europe. Sharma S. IFN-g release and outcome in complicated pulmonary tuberculosis component of the PPD stimulated response arises from innate cells. As drug concentrations were measured during a single dosing interval. severe or complicated disease. Le Strat Y. Elevated interleukin2 receptor level in patients with active pulmonary tuberculosis and the changes following anti-tuberculosis chemotherapy. a cytokine that is recognised to be crucial to defence against mycobacteria. Firstly.

Toossi Z. Depressed T-cell interferon-gamma responses in pulmonary tuberculosis: analysis of underlying mechanisms and modulation with therapy. Chan J. J Clin Invest 2000. J Infect Dis 2000. 18. spectrum. Rook GA. Udwadia Z. Effects of protein calorie malnutrition on tuberculosis in mice. Scott GM.171:1430–5. J Infect Dis 2001.193: 354–9. Use of a T cell-based assay for monitoring efficacy of antituberculosis therapy. Type 2 cytokine gene activation and its relationship to extent of disease in patients with tuberculosis. Israelski DM. et al.183:469–77. Am J Respir Crit Care Med 2001. Am J Respir Crit Care Med 2005. Proc Natl Acad Sci USA 1996. Clin Infect Dis 2004.50:1170–7. Deeks JJ. McIlleron H. Girardi E. Determinants of rifampicin. J Infect Dis 1999. 17. 14. Am J Respir Crit Care Med 2006. et al. Annu Rev Med 2004. 10. Newton SM. Lalvani A. Burger A. 21. Boussiotis VA. McIlleron et al.105:1317–25. 13. H. Antimicrob Agents Chemother 2006.180:225–8. 16. Bryant G. et al. Lalvani A. Small PM. 12. Carrara S. Amicosante M. pyrazinamide and ethambutol pharmacokinetics in a cohort of tuberculosis patients. Wilkinson KA. Corbett EL. et al.163:824–8. 19. et al. Clinical management of tuberculosis in the context of HIV infection. Warren RM. Lalvani A. Alvarez L. Millington KA. Tanaka KE. 11. Ewer K.174:831–9. Seah GT. Rapid detection of Mycobacterium tuberculosis infection by enumeration of antigen-specific T cells. Petrosillo N. and effect of chemotherapy. isoniazid. Di Sano C. McShane H. Sequestration of T lymphocytes to body fluids in tuberculosis: reversal of anergy following chemotherapy. IL-10-producing T cells suppress immune responses in anergic tuberculosis patients.178:760–8. J Infect Dis 1998. Tsai EY. compartmentalization. Dieli F. Kon OM. J Infect Dis 2006.55:283–301. 20. . Yunis EJ. Vincenti D. Hirsch CS. Tian Y. Dynamic antigen-specific T cell responses after point-source exposure to Mycobacterium tuberculosis. Verver S. Enumeration of T cells specific for RD1-encoded antigens suggests a high prevalence of latent Mycobacterium tuberculosis infection in healthy urban Indians.181:385–9. De Jong BC. Othieno C. Folb PI. Friscia G. et al.93:14857–61. Beyers N. 15. et al. Rate of reinfection tuberculosis after successful treatment is higher than rate of new tuberculosis. et al. Smith P. Nagvenkar P.38:754–6.Author's personal copy ARTICLE IN PRESS 564 9. Effect of treatment of latent tuberculosis infection on the T cell response to Mycobacterium tuberculosis antigens. Norman J. J Infect Dis 1999. Goletti D.180:2069–73. Pathan AA. Wash P.