Background: Commencing antiretroviral therapy (ART) in human immunodeficiency virus-infected patients with treated or unrecognized Mycobacterium tuberculosis disease may trigger tuberculosis-associated immune reconstitution inflammatory syndrome (TB-IRIS) or ART–associated tuberculosis (ART-TB). We have shown that whole-blood interferon-gamma release assays may aid in the prediction and diagnosis of ART-TB. Here, we investigate interferon-gamma–inducible chemokines CXCL9 and CXCL10.
Methods: CXCL9 and CXCL10 responses to region of difference 1 (RD1) antigens and purified protein derivative (PPD) were assayed in plasma from whole-blood cultures collected before and after 4, 12, and 24 weeks of ART from 15 TB-IRIS cases, 11 ART-TB cases, and matched controls.
Results: Relative to matched controls, ART-TB cases had elevated CXCL10 responses to RD1 antigens pre-ART (P = 0.02) and to PPD and RD1 antigens over 24 weeks of ART (P ≤ 0.02 and P ≤ 0.03). In contrast, TB-IRIS cases had higher CXCL10 responses to RD1 antigens before and after 4 weeks of ART only (P = 0.04 for both). CXCL9 responses to PPD and RD1 antigens were similar but less pronounced in ART-TB cases and did not differ between TB-IRIS cases and controls. CXCL10 responses to RD1 antigens performed as well as, or better than, IFN-γ responses in the prediction and diagnosis of ART-TB.
Conclusions: Tuberculosis after commencing ART is associated with increased CXCL10 and, to a lesser extent, CXCL9 responses to M. tuberculosis antigens. Assessment of antigen-induced CXCL10 responses to RD1 antigens may assist in the prediction and diagnosis of ART-TB.
*School of Pathology and Laboratory Medicine, University of Western Australia, Perth, Australia
†Infectious Diseases Unit, Alfred Hospital, Melbourne, Australia
‡Department of Infectious Diseases, Monash University, Melbourne, Australia
§Centre for Population Health, Burnet Institute, Melbourne, Australia
‖Western Australia Institute for Medical Research, University of Western Australia, Perth, Australia
¶Kirby Institute, University of New South Wales, Sydney, Australia.
Correspondence to: Martyn A. French, MB ChB, MD, School of Pathology and Laboratory Medicine, Level 2 MRF Building, Rear 50 Murray Street, Perth, 6000, Australia (e-mail: firstname.lastname@example.org).
Supported by the Australian Agency for International Development (AusAID), the Australian National Centre in HIV Epidemiology and Clinical Research and an Australian Government National Health and Medical Research Council program grant (grant number 510448).
This work has been presented in part at the 23rd Annual Conference of the Australasian Society for HIV Medicine, Canberra, September 26–28, 2011 (Paper Number: 503).
The authors have no conflicts of interest to disclose.
Received January 25, 2012
Accepted June 8, 2012
Mycobacterium tuberculosis disease is the leading cause of illness and death in patients with human immunodeficiency virus (HIV) infection.1 M. tuberculosis–specific Th1 cells are depleted during HIV infection2,3 but are partly restored by antiretroviral therapy (ART).4,5 In some HIV-infected patients, the restored immune response to M. tuberculosis may cause immunopathology known as immune restoration disease (IRD).6 Paradoxical worsening of treated tuberculosis (TB) after ART is a form of IRD known as TB-associated immune reconstitution inflammatory syndrome (TB-IRIS).7 Tuberculosis presenting during the first 3 months of ART might also be a form of IRD. Many cases present as ‘unmasking TB-IRIS’ whereby previously subclinical TB is ‘unmasked’ after starting ART. However, newly acquired TB may also present after ART initiation, so cases of TB that present after ART are referred to as ART-associated tuberculosis (ART-TB).7 ART-TB has been associated with increased interferon-gamma (IFN-γ) responses to mycobacterial antigens compared with HIV patients with no history of TB.8
Data from large prospective studies indicate that mortality associated with TB-IRIS is low but morbidity is substantial in HIV/TB-infected patients who have pre-ART CD4+ T cell counts <50 cells per microliter and initiate ART within 4 weeks of starting TB treatment.9,10 Morbidity and mortality from ART-TB are also significant, as HIV-infected patients who present with TB during the first 3 months of ART are 3.25 times more likely to die than HIV-infected patients presenting with TB at other times.11
Prediction and diagnosis of TB-IRIS and ART-TB might be improved by a better understanding of their immunopathogenesis. TB-IRIS has been associated with increased proinflammatory and Th1 cytokine responses to mycobacterial antigens,8,12–14 but these are also observed in coinfected patients who do not develop TB-IRIS after commencing ART,8,15,16 so other immune responses may also contribute to the immunopathology.17,18 ART-TB has also been associated with a prominent Th1 response to mycobacterial antigens, particularly those encoded by the region of difference 1 (RD1) domain of the M. tuberculosis genome.8
We have used a whole-blood IFN-γ release assay to examine IFN-γ responses to mycobacterial antigens [purified protein derivative (PPD) and RD1 antigens] in a cohort of Cambodian HIV patients with or without treated TB.8 To identify novel biomarkers of TB-IRIS and ART-TB, we examined antigen-induced production of CXCL9 (monokine induced by gamma interferon, Mig) and CXCL10 (IFN-γ-inducible protein-10, IP-10) in plasma from these assays. These chemokines are induced by IFN-γ and are important in several inflammatory conditions.19–25
Cases and Controls
Patients were recruited from the National Centre for HIV/AIDS, Dermatology and Sexually Transmitted Diseases Social Health Clinic in Phnom Penh, Cambodia, and were a subgroup of those reported previously.8 Seventy-five patients were being treated for active TB on starting ART and 15 (20%) developed TB-IRIS after a median time of 10 (range 7–89) days on ART. Eleven out of 231 patients (4.8%) with no history of TB developed ART-TB after a median time of 10 (range 1–28) days on ART. We used case definitions of TB-IRIS and ART-TB proposed by the International Network for the Study of HIV-associated IRIS.7 This subgroup differed from those reported previously8 as each TB-IRIS and ART-TB case was matched with 2 controls from the cohort by sex, pre-ART CD4+ T cell count, and TB history. The study was approved by the Cambodian National Ethics Committee and the human research ethics committees of the University of New South Wales and Royal Perth Hospital. All the patients provided written and informed consent.
Assay of Plasma Levels of CXCL10 and CXCL9 From Whole-Blood Cultures
Plasma samples from unstimulated and PPD and RD1-stimulated tubes of Quantiferon-TB Gold in-tube (QFTGIT) assays (Cellestis, Carnegie, Australia) were collected pre-ART, after 4, 12, and 24 weeks of ART and at the suspected time of TB-IRIS or ART-TB and cryopreserved at −80°C. Levels of CXCL9 and CXCL10 were assayed using BD Cytometric Bead Array Flex Sets (BD Biosciences, San Jose, CA). Three hundred events were collected per analyte using a BD FACSArray machine and BD FACSArray System Software v1.0.3 (BD Immunocytometry Systems). Analysis was performed using FCAP Array Software v1.0.1 (BD Biosciences). All samples were diluted at least one-fifth and the lowest limit of detection was 5 pg/mL. To analyze the effect of PPD and RD1 antigens on CXCL10 and CXCL9 production, levels of CXCL10 and CXCL9 in unstimulated tubes were subtracted from the levels measured in PPD- and RD1-stimulated tubes.
Demographic characteristics of TB-IRIS and ART-TB cases and controls were assessed using the Mann–Whitney U test and Fisher exact test. Data were assessed for normality using the Shapiro–Wilk W test. As the data were not normally distributed, median levels of CXCL9 and CXCL10 were compared between TB-IRIS or ART-TB cases and matched controls pre-ART and after 4, 12, and 24 weeks of ART using the Mann–Whitney U test. Receiver operating characteristic (ROC) curves were used to determine if levels of CXCL10 and CXCL9 pre-ART and at the time of TB-IRIS or ART-TB (or an equivalent time post-ART for controls) alone or in combination may aid in prediction or diagnosis of TB-IRIS or ART-TB. Comparisons were also made with data for IFN-γ. Correlations were performed using the Spearman rank test. Analyses were performed using STATA version 11 (Chicago, IL) and Prism v5.02 (San Diego, CA, USA). Statistical significance was defined as P < 0.05.
Demographic characteristics of the TB-IRIS and ART-TB cases and controls are summarized in Table 1 and are similar to those of the entire cohort.8 Clinical information about patients with TB-IRIS and ART-TB has been reported previously.8 The patients who developed TB-IRIS had a shorter interval between starting antitubercular therapy (ATT) and ART compared with those who did not (P = 0.034). The use of corticosteroid therapy for TB-IRIS was very infrequent.
CXCL10 Responses Were Most Clearly Elevated in ART-TB Cases
Before commencing ART, CXCL10 responses to RD1 antigens alone were higher in both ART-TB cases (P = 0.02) and TB-IRIS cases (P = 0.04) compared with their respective controls (Figs. 1C, D). In contrast, CXCL10 responses to both PPD and RD1 antigens were higher during 24 weeks of ART in HIV-infected patients who developed ART-TB compared with controls (PPD: for week 4: P = 0.0009, for week 12: P = 0.002, for week 24: P = 0.02; RD1: for week 4: P = 0.001, for week 12: P = 0.008, for week 24: P = 0.03) (Figs. 1A, C), whereas TB-IRIS cases demonstrated a significantly higher CXCL10 response to RD1 antigens at week 4 only (P = 0.04) (Fig. 1D). CXCL10 responses to PPD did not differ between TB-IRIS cases and controls before or during 24 weeks of ART (Fig. 1B).
IFN-γ responses to PPD and RD1 antigens have been reported in these patients.8 In general, CXCL10 responses to PPD and RD1 antigens correlated with IFN-γ responses in TB-IRIS and ART-TB cases and their controls over 24 weeks of ART (data not shown). A strong correlation between IFN-γ and CXCL10 responses to RD1 antigens was evident in patients who developed ART-TB over 24 weeks of ART (for weeks 0, 4, 12, and 24 of ART; r ≥ 0.87, P < 0.05) (data not shown).
CXCL9 Responses to RD1 Antigens Were Higher in ART-TB Cases Compared With Controls
Before ART, CXCL9 responses to PPD and RD1 antigens were not significantly different between ART-TB or TB-IRIS cases and their controls (Fig. 2). In ART-TB cases, CXCL9 responses to RD1 antigens were significantly higher compared with controls after 4, 12, and 24 weeks of ART (for week 4: P = 0.02, for week 12: P = 0.048, for week 24: P = 0.002) (Fig. 2C), whereas CXCL9 responses to PPD were higher only after 4 weeks of ART compared with controls (P = 0.02) (Fig. 2A). CXCL9 responses to PPD and RD1 antigens were similar in TB-IRIS cases and controls (Figs. 2B, D).
CXCL10 Responses to RD1 Antigens May Aid in the Prediction and Diagnosis of ART-TB
We have previously shown that pre-ART levels of IL-18, CXCL10, and CCL2 in combination with IFN-γ responses to PPD and RD1 antigens in plasma from QFTGIT assays are predictive of TB-IRIS, whereas IFN-γ responses to PPD and RD1 antigens are predictive of ART-TB.8,17 Here, ROC analyses demonstrated that CXCL10 responses to RD1 antigens were moderately predictive of ART-TB [area under the curve (AUC), 0.75] and were a better predictor than IFN-γ responses (AUC, 0.61) (Table 2). In contrast, CXCL9 and CXCL10 responses to PPD and RD1 antigens were poorly predictive of TB-IRIS (Table 2).
We have also reported that IFN-γ responses to PPD and RD1 antigens in plasma from QFTGIT assays may aid in the diagnosis of ART-TB.8 Here, CXCL10 responses to PPD and RD1 antigens were assessed at the time of disease in TB-IRIS and ART-TB cases and at an equivalent time post-ART in matched controls. CXCL10 responses to PPD and RD1 antigens were strongly diagnostic of ART-TB (AUC, 0.86; 0.85) but not TB-IRIS (AUC, 0.53; 0.56). However, the diagnostic ability of CXCL10 was not better than IFN-γ alone or when combined with CXCL9 and IFN-γ (Table 3).
CXCL9 and CXCL10 responses to PPD and RD1 antigens of M. tuberculosis were assessed in plasma from the QFTGIT assays undertaken in HIV-infected patients who developed ART-TB or TB-IRIS and matched controls, to elucidate mechanisms of immunopathogenesis and seek biomarkers of IRD associated with M. tuberculosis infection after commencing ART. Patients who developed ART-TB had higher CXCL10 responses to RD1 antigens pre-ART and higher CXCL10 responses to both RD1 antigens and PPD during 24 weeks of ART compared with controls. In addition, the patients who developed ART-TB had higher CXCL9 responses to RD1 antigens during 24 weeks of ART and higher responses to PPD at week 4 compared with controls. In contrast, TB IRIS cases displayed higher CXCL10 responses to RD1 antigens than controls pre-ART and at week 4 only. Neither CXCL9 nor CXCL10 responses to PPD differed between TB-IRIS cases and controls.
As RD1 antigens are secreted by M. tuberculosis,26 the increase in CXCL9 and CXCL10 responses to RD1 antigens in ART-TB cases may reflect an elevated antigen-specific T cell response against active M. tuberculosis infection. These findings therefore provide further evidence that ART-TB is associated with increased T cell responses to mycobacterial antigens, as observed by the measurement of IFN-γ,8 suggesting that most, if not all, cases are likely to be a form of IRD. Differences in antigen-induced CXCL9 or CXCL10 responses were less pronounced in patients who developed TB-IRIS compared to controls. This could be interpreted as evidence that T-cells mediate ART-TB but not TB-IRIS. However, 2 other possible explanations should be considered. First, differences in T-cell responses between ART-TB cases and controls may be clearer because ART-TB controls were not infected by M. tuberculosis so their T cell responses to PPD and RD1 antigens are low. Second, assays of CXCL9 and CXL10 in antigen-stimulated whole-blood cultures may not be sensitive enough to distinguish T cell responses of TB-IRIS cases from controls.
In this patient cohort pre-ART IFN-γ responses to PPD and RD1 antigens, when corrected for pre-ART CD4+ T cell count, are predictive of ART-TB.8 Here, pre-ART CXCL10 responses to RD1 antigens were higher in patients who subsequently developed TB-IRIS or ART-TB compared with matched controls. This would be expected given that IFN-γ is a potent inducer of CXCL10 and CXCL9. However, analysis of the data using ROC analyses demonstrated that CXCL10 responses to RD1 antigens were more predictive of ART-TB than IFN-γ responses (Table 2). It must be stressed, however, that the number of patients analyzed in our study was small and further studies on larger numbers of patients are required to determine if assessment of CXCL10 responses to RD1 antigens has clinical utility in predicting ART-TB.
These findings indicate that T cell responses against M. tuberculosis are present before ART is commenced in patients who develop TB-IRIS or ART-TB and that assay of CXCL10 in whole-blood cultures stimulated with RD1 antigens may be more informative than assays of IFN-γ. They also raise the possibility that demonstration of these responses might be used to detect ‘subclinical’ M. tuberculosis and predict the unmasking of infection by restoration of an immune response against mycobacterial antigens. Recent data suggest that HIV-infected patients residing in areas endemic for TB should be screened for TB by sputum culture even in the absence of TB symptoms, as smear-positive TB cultures are indicative of subclinical disease that progresses to symptomatic disease in the majority of cases.27,28 Assessment of CXCL10 responses to RD1 antigens in plasma from the QFTGIT assay may be a more effective screening tool for the detection of asymptomatic TB in HIV-infected patients, as many HIV/TB patients have smear–negative pulmonary disease.29 CXCL10 may be superior to IFN-γ in diagnosing M. tuberculosis infection in HIV patients as CXCL10 detects a greater number of HIV-infected patients with TB compared with IFN-γ.30,31
We have previously shown that IFN-γ responses to PPD and RD1 antigens in the QFTGIT assay may aid in the diagnosis of ART-TB.8 Analysis of the data presented here demonstrated that CXCL10 responses to PPD and RD1 antigens were strongly diagnostic of ART-TB using ROC analyses (AUC = 0.86 and 0.85) and performed as well as or better than IFN-γ responses. This finding should be investigated further in a larger cohort of patients.
In patients with ART-TB, CXCL9 responses to PPD and/or RD1 antigens were higher than in controls after ART. These findings are similar to those for CXCL10 and further suggest that ART-TB is associated with increased Th1 responses against mycobacterial antigens.8 In contrast, CXCL9 responses to PPD and RD1 antigens did not differ between TB-IRIS cases and controls over 24 weeks of ART. These findings are similar to patterns seen with CXCL10.
This study has some limitations which should be considered. The number of patients studied was small so larger cohorts of patients are required to validate the clinical utility of assaying CXCL10 in antigen-stimulated QFTGIT assays. Levels of CXCL9 and CXCL10 measured in TB-IRIS cases may have also been affected by treatment with corticosteroids, but very few patients received corticosteroids before sample collection. Mycobacterial load pre-ART was not examined and might have influenced pre-ART immune responses against mycobacteria,32 especially as patients who developed TB-IRIS had a shorter interval between starting ATT and ART compared with patients who did not. Due to the difficulties in defining latent M. tuberculosis infection, some ART-TB control group may have not been exposed to M. tuberculosis, some may have resolved previous infection, and some may harbor latent infection. Future studies should include careful analysis of clinical histories.
In summary, patients who developed TB-IRIS and ART-TB exhibited higher CXCL10 responses to RD1 antigens of M. tuberculosis than controls before the initiation of ART. In patients who developed ART-TB, CXCL10 responses to RD1 antigens rose further on ART, whereas CXCL9 responses to RD1 antigens also remained higher than controls over 24 weeks. In contrast, TB-IRIS was characterized by a higher CXCL10 response to RD1 antigens pre-ART and after 4 weeks of ART only. Assessment of T cell responses to RD1 antigens by assaying CXCL10 in whole-blood assays performs as well as, or better than, IFN-γ and may have clinical utility in the detection of subclinical M. tuberculosis infection and the prediction and diagnosis of ART-TB.
The authors thank all study participants for providing samples for the study and staff at the Social Health Clinic in Phnom Penh.
2. Geldmacher C, Schuetz A, Ngwenyama N, et al.. Early depletion of Mycobacterium tuberculosis
-specific T helper 1 cell responses after HIV-1 infection. J Infect Dis. 2008;198:1590–1598.
3. Geldmacher C, Ngwenyama N, Schuetz A, et al.. Preferential infection and depletion of Mycobacterium tuberculosis
-specific CD4 T cells after HIV-1 infection. J Exp Med. 2010;207:2869–2881.
4. Schluger NW, Perez D, Liu YM. Reconstitution of immune responses to tuberculosis in patients with HIV infection who receive antiretroviral therapy. Chest. 2002;122:597–602.
5. Kampmann B, Tena-Coki GN, Nicol MP, et al.. Reconstitution of antimycobacterial immune responses in HIV-infected children receiving HAART. AIDS. 2006;20:1011–1018.
6. French MA. HIV/AIDS: immune reconstitution inflammatory syndrome: a reappraisal. Clin Infect Dis. 2009;48:101–107.
7. Meintjes G, Lawn SD, Scano F, et al.. Tuberculosis-associated immune reconstitution inflammatory syndrome: case definitions for use in resource-limited settings. Lancet Infect Dis. 2008;8:516–523.
8. Elliott JH, Vohith K, Saramony S, et al.. Immunopathogenesis and diagnosis of tuberculosis and tuberculosis-associated immune reconstitution inflammatory syndrome during early antiretroviral therapy. J Infect Dis. 2009;200:1736–1745.
9. Abdool Karim SS, Naidoo K, Grobler A, et al.. Integration of antiretroviral therapy with tuberculosis treatment. N Engl J Med. 2011;365:1492–1501.
10. Blanc FX, Sok T, Laureillard D, et al.. Earlier versus later start of antiretroviral therapy in HIV-infected adults with tuberculosis. N Engl J Med. 2011;365:1471–1481.
11. Koenig SP, Riviere C, Leger P, et al.. High mortality among patients with AIDS who received a diagnosis of tuberculosis in the first 3 months of antiretroviral therapy. Clin Infect Dis. 2009;48:829–831.
12. Bourgarit A, Carcelain G, Martinez V, et al.. Explosion of tuberculin-specific Th1-responses induces immune restoration syndrome in tuberculosis and HIV co-infected patients. AIDS. 2006;20:F1–F7.
13. Tan DB, Yong YK, Tan HY, et al.. Immunological profiles of immune restoration disease presenting as mycobacterial lymphadenitis and cryptococcal meningitis. HIV Med. 2008;9:307–316.
14. Tadokera R, Meintjes G, Skolimowska KH, et al.. Hypercytokinaemia accompanies HIV-tuberculosis immune reconstitution inflammatory syndrome. Eur Respir J. 2011;37:1248–1259.
15. Meintjes G, Wilkinson KA, Rangaka MX, et al.. Type 1 helper T cells and FoxP3-positive T cells in HIV-tuberculosis-associated immune reconstitution inflammatory syndrome. Am J Respir Crit Care Med. 2008;178:1083–1089.
16. Tieu HV, Ananworanich J, Avihingsanon A, et al.. Immunologic markers as predictors of tuberculosis-associated immune reconstitution inflammatory syndrome in HIV and tuberculosis coinfected persons in Thailand. AIDS Res Hum Retroviruses. 2009;25:1083–1089.
17. Oliver BG, Elliott JH, Price P, et al.. Mediators of innate and adaptive immune responses differentially affect immune restoration disease associated with Mycobacterium tuberculosis
in HIV patients beginning antiretroviral therapy. J Infect Dis. 2010;202:1728–1737.
18. Haddow LJ, Dibben O, Moosa MY, et al.. Circulating inflammatory biomarkers can predict and characterize tuberculosis-associated immune reconstitution inflammatory syndrome. AIDS. 2011;25:1163–1174.
19. Agostini C, Facco M, Siviero M, et al.. CXC chemokines IP-10 and MIG expression and direct migration of pulmonary CD8+/CXCR3+ T cells in the lungs of patients with HIV infection and T-cell alveolitis. Am J Respir Crit Care Med. 2000;162:1466–1473.
20. Khan IA, MacLean JA, Lee FS, et al.. IP-10 is critical for effector T cell trafficking and host survival in Toxoplasma gondii
infection. Immunity. 2000;12:483–494.
21. Liu MT, Chen BP, Oertel P, et al.. The T cell chemoattractant IFN-inducible protein 10 is essential in host defense against viral-induced neurologic disease. J Immunol. 2000;165:2327–2330.
22. Harvey CE, Post JJ, Palladinetti P, et al.. Expression of the chemokine IP-10 (CXCL10) by hepatocytes in chronic hepatitis C virus infection correlates with histological severity and lobular inflammation. J Leukoc Biol. 2003;74:360–369.
23. Campanella GS, Tager AM, El Khoury JK, et al.. Chemokine receptor CXCR3 and its ligands CXCL9 and CXCL10 are required for the development of murine cerebral malaria. Proc Natl Acad Sci U S A. 2008;105:4814–4819.
24. Nie CQ, Bernard NJ, Norman MU, et al.. IP-10-mediated T cell homing promotes cerebral inflammation over splenic immunity to malaria infection. PLoS Pathog. 2009;5:e1000369.
25. Amin DN, Rottenberg ME, Thomsen AR, et al.. Expression and role of CXCL10 during the encephalitic stage of experimental and clinical African trypanosomiasis. J Infect Dis. 2009;200:1556–1565.
26. 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). Microbiology. 1998;144(pt 11):3195–3203.
27. Oni T, Burke R, Tsekela R, et al.. High prevalence of subclinical tuberculosis in HIV-1-infected persons without advanced immunodeficiency: implications for TB screening. Thorax. 2011;66:669–673.
28. Lawn SD, Kerkhoff AD, Wood R. Progression of subclinical culture-positive tuberculosis to symptomatic disease in HIV-infected individuals. AIDS. 2011;25:2190–2191.
29. Reid MJ, Shah NS. Approaches to tuberculosis screening and diagnosis in people with HIV in resource-limited settings. Lancet Infect Dis. 2009;9:173–184.
30. Aabye MG, Ruhwald M, Praygod G, et al.. Potential of interferon-gamma-inducible protein 10 in improving tuberculosis diagnosis in HIV-infected patients. Eur Respir J. 2010;36:1488–1490.
31. Kabeer BS, Sikhamani R, Raja A. Comparison of interferon gamma and interferon gamma-inducible protein-10 secretion in HIV-tuberculosis patients. AIDS. 2010;24:323–325.
32. Conesa-Botella A, Loembé MM, Manabe YC, et al.. Urinary lipoarabinomannan as predictor for the tuberculosis immune reconstitution inflammatory syndrome. J Acquir Immune Defic Syndr. 2011;58:463–468.