Comments
Abstract
Rationale: Current immunodiagnostic tests for tuberculosis (TB), including the tuberculin skin test and IFN-γ release assay (IGRA), have significant limitations, which include their inability to distinguish between latent TB infection (LTBI) and active TB, a distinction critical for clinical management.
Objectives: To identify mycobacteria-specific cytokine biomarkers that characterize TB infection, determine their diagnostic performance characteristics, and establish whether these biomarkers can distinguish between LTBI and active TB.
Methods: A total of 149 children investigated for TB infection were recruited; all participants underwent a tuberculin skin test and QuantiFERON-TB Gold assay. In parallel, whole-blood assays using early secretory antigenic target-6, culture filtrate protein-10, and PPD as stimulatory antigens were undertaken, and cytokine responses were determined by xMAP multiplex assays.
Measurements and Main Results: IFN-γ, interferon-inducible protein-10 (IP-10), tumor necrosis factor (TNF)-α, IL-1ra, IL-2, IL-13, and MIP-1β (macrophage inflammatory protein-1β) responses were significantly higher in LTBI and active TB cases than in TB-uninfected individuals, irrespective of the stimulant. Receiver operating characteristic analyses showed that IP-10, TNF-α, and IL-2 responses achieved high sensitivity and specificity for the distinction between TB-uninfected and TB-infected individuals. TNF-α, IL-1ra, and IL-10 responses had the greatest ability to distinguish between LTBI and active TB cases; the combinations of TNF-α/IL-1ra and TNF-α/IL-10 achieved correct classification of 95.5% and 100% of cases, respectively.
Conclusions: We identified several mycobacteria-specific cytokine biomarkers with the potential to be exploited for immunodiagnosis. Incorporation of these biomarkers into future immunodiagnostic assays for TB could result in substantial gains in sensitivity and allow the distinction between LTBI and active TB based on a blood test alone.
Current immunodiagnostic tests for tuberculosis (TB), including the tuberculin skin test and IFN-γ release assays, have significant limitations. These include the inability to distinguish between latent TB infection (LTBI) and active TB, a distinction that is critical for clinical management.
This study identified mycobacteria-specific cytokine responses that allow the distinction between TB-infected and TB-uninfected individuals, as well as between LTBI and active TB. Several of the cytokine biomarkers showed better performance characteristics than IFN-γ, which forms the basis of IFN-γ release assays. Addition of these biomarkers into future immunodiagnostic assays for TB is likely to result in higher assay sensitivity while retaining high specificity, and it could potentially allow the distinction between LTBI and active TB based on a blood test alone.
Approximately one-third of the world’s population is infected with Mycobacterium tuberculosis (MTB), and there are more than 8 million new cases of active tuberculosis (TB) (i.e., TB disease) resulting in an estimated 1.3 million deaths each year (1, 2). Children account for an increasing proportion of TB cases, in both high-resource and low-resource settings (3).
Despite advances in diagnostics, microbiologic confirmation of TB in childhood remains challenging, primarily because most children have paucibacillary disease, reducing the yield of conventional microbiologic methods. Newer molecular diagnostic techniques, including the Xpert MTB/RIF assay, show promise, but also perform considerably less well in patients with paucibacillary disease (4).
Existing immunodiagnostic tests for TB also have considerable limitations (5, 6). The tuberculin skin test (TST) is operator-dependent and has limited specificity. False-positive results can occur as a result of boosting from repeated TST, prior bacillus Calmette-Guérin (BCG) vaccination, or infection with nontuberculous mycobacteria (7). The comparatively poor specificity of the TST intrinsically results from the use of PPD, a heterogeneous mixture of more than 200 mycobacterial peptides, many of which are expressed by BCG vaccine strains and nontuberculous mycobacteria (8). IFN-γ release assays (IGRAs), which rely on the detection of IFN-γ produced by sensitized T cells, are thought to have better specificity than the TST, because mycobacterial peptides that are absent from all BCG vaccine strains and most nontuberculous mycobacteria are used as the stimulatory antigens in these assays (8). Currently licensed IGRAs (the QuantiFERON-TB Gold [QFT] assay and the T-SPOT.TB assay) incorporate the MTB-specific RD1 peptide antigens early secretory antigenic target 6 (ESAT-6) and 10-kD culture filtrate protein (CFP-10). However, IGRAs have other, important limitations. Although IGRAs are solely licensed for the diagnosis of latent TB infection (LTBI), in clinical practice these assays are frequently used to support a presumptive diagnosis of active TB. However, a recent metaanalysis shows that the sensitivity of both commercial IGRAs barely exceed 80% in patients with active TB (9). Furthermore, IGRAs perform considerably worse in children, and significant rates of indeterminate results are an additional problem in this age group (10–14).
A further significant limitation of both TST and IGRAs is their inability to discriminate between LTBI and active TB, a distinction that is highly relevant in geographic regions where TB prevalence is high and a large proportion of the population has LTBI (6). This distinction is critical, because the treatment of LTBI and active TB differs. Consequently, an immunodiagnostic test that can discriminate between these two infection states would be a major advance for clinical care.
A large body of evidence supports the critical role of IFN-γ in the immune response to mycobacterial infections (15, 16). However, mounting data highlight the importance of other cytokines in the immune response to MTB, which may also have a potential to be used for immunodiagnosis (15–17). This study aimed to identify mycobacteria-specific cytokine biomarkers that characterize TB infection, to determine their diagnostic performance characteristics, and to establish whether any of the identified biomarkers can be used to distinguish between LTBI and active TB.
Children and adolescents up to 18 years of age were recruited at the Royal Children’s Hospital Melbourne between January 2010 and February 2011. Eligible for participation were all children undergoing screening for suspected LTBI or active TB, comprising the following: (1) children with symptoms and signs suggestive of active TB, (2) children with known contact with a case of active TB, and (3) children who had recently migrated from countries with a high TB prevalence (defined by incidence ≥40 TB cases per 100,000 population). The exclusion criteria comprised the following: (1) known immunodeficiency, (2) current immunosuppressive treatment (including oral corticosteroids), and (3) TST in the previous 6–52 weeks. The last criterion was chosen because at commencement of the study it was thought that a TST undertaken more than 6 weeks before an IGRA may induce boosting, thereby causing a false-positive IGRA result (18). We have subsequently shown this not to be the case (19).
Before participation, informed consent was obtained from the child’s parent and/or guardian. The study was approved by the Royal Children’s Hospital Human Research Ethics Committee (HREC 29040A). Demographics, history, and clinical findings were recorded on a standardized data collection sheet. A chest radiograph was obtained in all cases with a positive TST and/or positive IGRA result. Histological, conventional microbiological, and molecular microbiological tests were performed in all children with suspected active TB as clinically indicated.
All participants had a TST placed by specifically trained nurses by intradermal injection of 0.1 ml of Tubersol (Sanofi Pasteur, Toronto, Canada; bioequivalent to 5 Tuberculin Units PPD-S) into the volar surface of the lower arm, and any resulting induration was recorded after 48–72 hours. In addition, blood was obtained for the QFT In-Tube (QFT-GIT) assay (Cellestis/Qiagen, Carnegie, Australia), and an additional 10 ml for whole-blood assays were collected in heparinized tubes. The QFT-GIT was processed and interpreted at the Victorian Infectious Diseases Reference Laboratories in accordance with the manufacturer’s instructions. Polymerase chain reaction testing for MTB, for which a Taqman real-time polymerase chain reaction (Applied Biosystems, Waltham, MA) targeting the insertion sequence IS6110 was used, was also performed at the Victorian Infectious Diseases Reference Laboratories using previously described methods (20).
Participants were placed into seven categories according to their clinical features, TST, IGRA, and microbiologic results as detailed in Table 1. Active TB was defined as either microbiologic confirmation of infection with MTB by culture or polymerase chain reaction, or a symptomatic patient fulfilling at least two of the following three criteria in conjunction with response to treatment with antituberculous therapy: (1) symptoms and signs consistent with active TB (chronic cough, persistent fever, night sweats, unexplained weight loss), (2) radiologic findings suggestive of active TB, and (3) presence of risk factors for TB infection (known TB contact, birth or previous residence in a high TB prevalence country). These stringent criteria exceed those proposed by the American Thoracic Society and the Centers for Disease Control and Prevention (21). In this manuscript, the term “TB-infected” is used as a collective term for participants with LTBI or active TB.
Whole blood was incubated with ESAT-6, CFP-10 (each at a concentration of 10 μg/ml; JPT Peptide Technologies, Berlin, Germany), PPD (20 μg/ml; RT50; Statens Serum Institut, Copenhagen, Denmark), staphylococcal enterotoxin B (5 μg/ml; positive control; Sigma-Aldrich, St. Louis, MO), or without stimulant (negative control) in the presence of costimulatory antibodies, anti-CD28 and anti-CD49d (each 1 μg/ml; BD Biosciences, San Jose, CA). Mycobacterial antigens were added at the beginning of the assay; staphylococcal enterotoxin B was only added for the last 5 hours of the incubation period based on previous optimization experiments (data not shown). Following incubation at 37°C for 20–24 hours, supernatants were harvested and cryopreserved at −80°C for batched analysis.
Cytokine concentrations in supernatants were measured using Milliplex human cytokine/chemokine kits (Millipore Corp., Billerica, MA) according to the manufacturer’s instruction, with analyses conducted on a Luminex 200 analyzer (Luminex Corp., Austin, TX). Based on previous optimization experiments (data not shown), IFN-γ, tumor necrosis factor (TNF)-α, IL-1ra, IL-2, IL-10, IL-12(p40), IL-13, IL-15, IL-17, and GM-CSF (granulocyte–macrophage colony–stimulating factor) were analyzed in undiluted samples with a 10-plex assay, whereas interferon-inducible protein-10 (IP-10), IL-6, IL-8, monocyte chemotactic protein-1 (MCP-1), MCP-3, macrophage inflammatory protein-1β (MIP-1β), and CCL5 (RANTES [regulated upon activation, normal T-cell expressed and secreted]) were analyzed in 1:20 diluted samples with a 7-plex assay. The laboratory scientists performing the sample analyses were blinded to the clinical data and the results of the TST, the QFT-GIT, and the microbiologic investigations.
Only those participants with an unambiguous diagnosis (“uninfected,” “common discordance,” “LTBI,” and “active TB”) were included in the analysis. Participants in the “probable uninfected” and “possible discordance” categories were deliberately not included a priori to avoid potential contamination of data.
Comparisons of continuous variables between multiple groups were done using Kruskal-Wallis tests. In instances where the Kruskal-Wallis P value was less than 0.05, indicating a difference between the groups, additional two-group comparisons were done using Mann-Whitney U tests. Categorical variables were compared using two-tailed chi-square tests. A P value less than 0.05 was considered significant. Cytokine concentrations were background-corrected before analysis (i.e., by subtracting the concentration measured in the negative control sample). Analyses were done using Stata V11 (StataCorp, College Station, TX) and Prism V5 (GraphPad Software Inc., La Jolla, CA). Receiver operating characteristic analyses were performed with Prism; the optimal cut-offs for each of the stimulant/cytokine combinations were determined by tabulation of sensitivity against specificity at every threshold in the data set. The study was conducted and is reported in accordance with QUADAS (Quality Assessment of Diagnostic Accuracy Studies) criteria (22).
A total of 149 patients were recruited. Nine participants were excluded: three did not return for TST reading, insufficient blood was obtained in three, a laboratory error occurred during the QFT-GIT processing in one, and the QFT-GIT result was indeterminate in two (both failed positive controls). Therefore, a total of 140 patients were included in the analysis.
The six participants with active TB comprised three patients with microbiologically confirmed TB (one case with intrathoracic TB; two cases with lymph node TB), and three patients without microbiologic confirmation who fulfilled the study criteria for active TB (two cases with pulmonary TB; one case with spinal TB). All six patients had both a positive TST (range, 13–25 mm induration) and a positive QFT-GIT result, and resolution of symptoms with antituberculous therapy. Additional patient data are shown in Table E1 in the online supplement.
Table 2 shows the demographic and other details of the participants in the four major diagnostic categories (uninfected, common discordance, LTBI, and active TB). In the remaining three diagnostic categories there were six “probable uninfected,” five “possible discordance,” and four “reverse discordance” cases.
| Total Cohort (n = 140) | Uninfected (n = 75) | Common Discordance (n = 28) | LTBI (n = 16) | Active TB (n = 6) | |
|---|---|---|---|---|---|
| Median (IQR) age, yr | 8.3 (3.7–12.8) | 6.3 (2.8–10.9) | 12.1 (6.0–14.5) | 11.6 (6.0–14.2) | 15.0 (12.1–16.2) |
| Ethnic origin, No. (%) | |||||
| Africa | 60 (42.8) | 31 (41.3) | 9 (32.1) | 10 (62.5) | 5 (83.3) |
| Asia | 53 (37.9) | 24 (32.0) | 13 (46.4) | 5 (31.3) | 1 (16.7) |
| Middle East | 9 (6.4) | 5 (6.7) | 4 (14.3) | 0 | 0 |
| Australia/New Zealand | 18 (12.9) | 15 (20) | 2 (7.1) | 1 (6.3) | 0 |
| Country of birth, No. (%) | |||||
| Africa | 41 (29.3) | 14 (18.7) | 8 (28.6) | 10 (62.5) | 5 (83.3) |
| Asia | 29 (20.7) | 10 (13.3) | 7 (25.0) | 5 (31.3) | 1 (16.7) |
| Middle East | 9 (6.4) | 4 (5.3) | 4 (14.3) | 0 | 0 |
| Australia/New Zealand | 59 (42.1) | 47 (62.7) | 8 (28.6) | 1 (6.3) | 0 |
| Europe | 2 (1.4) | 0 | 1 (3.6) | 0 | 0 |
| Migration background,* No. (%) | 81 (57.9) | 28 (37.3) | 20 (71.4) | 15 (93.8) | 6 (100) |
| Median (IQR) duration of residence in Australia (migrants only)*, mo | 8.0 (3.5–36.0) | 5.5 (2.0–33.0) | 17.0 (6.0–36.0) | 7.0 (4.0–36.0) | 21.0 (3.0–69.0) |
| BCG vaccination history, No. (%) | |||||
| Yes | 75 (53.6) | 23 (30.7) | 21 (75.0) | 14 (87.5) | 5 (83.3) |
| No | 58 (42.4) | 49 (65.3) | 4 (14.2) | 1 (6.3) | 1 (16.7) |
| Unknown | 7 (5.0) | 3 (4.0) | 3 (10.7) | 1 (6.3) | 0 |
| BCG scar, No. (%) | |||||
| Yes | 67 (47.9) | 22 (29.3) | 18 (64.3) | 12 (75.0) | 5 (83.3) |
| No | 73 (52.1) | 53 (70.7) | 10 (35.7) | 4 (25.0) | 1 (16.7) |
| Known TB contact, No. (%) | |||||
| Yes | 89 (63.6) | 54 (72.0) | 14 (50.0) | 8 (50.0) | 1 (16.7) |
| No | 51 (36.4) | 21 (28.0) | 14 (50.0) | 8 (50.0) | 5 (83.3) |
| Type of TB contact, No. (%) | |||||
| Parent | 34 (24.3) | 14 (18.6) | 9 (32.1) | 4 (25.0) | 1 (16.7) |
| Other household member | 33 (23.6) | 23 (30.7) | 3 (10.7) | 3 (18.8) | 0 |
| Other contact | 22 (15.7) | 17 (22.7) | 2 (7.1) | 1 (6.3) | 0 |
Significant differences in the background-corrected cytokine concentrations measured in supernatants between the four major diagnostic groups in response to all three antigenic stimulants were detected for IFN-γ, IP-10, TNF-α, IL-1ra, IL-2, IL-13, and MIP-1β (Figures 1A–1C). Overall, median concentrations of these cytokines were highest in the active TB group, followed by the LTBI group. For all seven cytokines the lowest median concentrations were observed in the uninfected group, irrespective of the antigenic stimulant.
Figure 1. Box plot with Tukey whiskers showing background-corrected cytokine concentrations in uninfected participants (white) and participants with common discordance (yellow), latent tuberculosis infection (orange), and active tuberculosis (red) in ESAT-6 (early secretory antigenic target-6)– (A), CFP-10 (10-kD culture filtrate protein)– (B), and PPD- (C) stimulated samples. Statistical differences were analyzed using Kruskal-Wallis tests; the corresponding P values are shown at the top of each graph. The results of additional two-group comparisons are detailed in Table 3. GM-CSF = granulocyte-macrophage colony–stimulating factor; IP-10 = interferon-inducible protein-10; MCP = monocyte chemotactic protein; MIP-1β = macrophage inflammatory protein 1β; RANTES = regulated upon activation, normal T-cell expressed and secreted; TNF-α = tumor necrosis factor α.
Table 3 shows the results of the two-group comparisons of cytokine responses in the four diagnostic groups. Median concentrations of IFN-γ, IP-10, TNF-α, IL-1ra, IL-2, IL-13, and MIP-1β were all significantly higher in the LTBI and the active TB groups, compared with the uninfected group, with all three antigenic stimulants, with the single exception of IL-1ra in PPD-stimulated samples. Most of these comparisons were highly statistically significant with a Mann-Whitney U P value below 0.0001. In contrast, few comparisons between the LTBI and the active TB group reached statistical significance: TNF-α in ESAT-6– and CFP-10–stimulated samples; IL-1ra in ESAT-6– and PPD-stimulated samples; and IP-10, IL-6, and IL-10 in PPD-stimulated samples. In all instances median concentrations were higher in the active TB group than in the LTBI group (Figure 1), with the exception of IP-10 in PPD-stimulated samples. In addition, several significant differences in median cytokine concentrations between the uninfected group and the group with common discordance were detected (Table 3). This was the case not only in response to stimulation with PPD, but also in response to stimulation with ESAT-6 and CFP-10.
| Stimulant | Cytokine | Kruskal-Wallis P Value | Uninfected vs. CD | Uninfected vs. LTBI | Uninfected vs. Active TB | CD vs. LTBI | CD vs. Active TB | LTBI vs. Active TB |
|---|---|---|---|---|---|---|---|---|
| ESAT-6 | IFN-γ | <0.0001 | 0.0147 | <0.0001* | <0.0001* | 0.0001* | 0.0008* | 0.3020 |
| IP-10 | <0.0001 | 0.0026 | <0.0001* | <0.0001* | <0.0001* | 0.0007* | 0.8828 | |
| TNF-α | <0.0001 | 0.0504 | <0.0001* | <0.0001* | <0.0001* | 0.0004* | 0.0183 | |
| IL-1ra | <0.0001 | 0.2925 | 0.0005* | <0.0001* | 0.0180 | 0.0007* | 0.0183 | |
| IL-2 | <0.0001 | <0.0001* | <0.0001* | <0.0001* | <0.0001* | 0.0010* | 0.5070 | |
| IL-6 | 0.0642 | — | — | — | — | — | — | |
| IL-8 | 0.7564 | — | — | — | — | — | — | |
| IL-10 | 0.1042 | — | — | — | — | — | — | |
| IL-12(p40) | 0.0061 | 0.0080 | 0.0255 | 0.0540 | 0.6258 | 0.2433 | 0.3451 | |
| IL-13 | <0.0001 | 0.0269 | <0.0001* | <0.0001* | 0.0127 | 0.0006* | 0.0650 | |
| IL-15 | 0.4083 | — | — | — | — | — | — | |
| IL-17 | 0.1647 | — | — | — | — | — | — | |
| GM-CSF | <0.0001 | 0.6012 | <0.0001* | 0.0058 | <0.0001* | 0.0129 | 0.7124 | |
| MCP-1 | 0.1701 | — | — | — | — | — | — | |
| MCP-3 | 0.1038 | — | — | — | — | — | — | |
| MIP-1β | <0.0001 | 0.1883 | <0.0001* | <0.0001* | 0.0005* | 0.0006* | 0.0900 | |
| RANTES | 0.3249 | — | — | — | — | — | — | |
| CFP-10 | IFN-γ | <0.0001 | 0.1588 | <0.0001* | <0.0001* | 0.0128 | 0.0011* | 0.0900 |
| IP-10 | <0.0001 | 0.2630 | <0.0001* | <0.0001* | 0.0013* | 0.0018 | 0.6058 | |
| TNF-α | <0.0001 | 0.2908 | <0.0001* | <0.0001* | 0.0029 | 0.0005* | 0.0121 | |
| IL-1ra | <0.0001 | 0.4653 | 0.0013* | 0.0003* | 0.0338 | 0.0025 | 0.1048 | |
| IL-2 | <0.0001 | 0.0202 | <0.0001* | <0.0001* | 0.0007* | 0.0021 | 0.4174 | |
| IL-6 | 0.1354 | — | — | — | — | — | — | |
| IL-8 | 0.8776 | — | — | — | — | — | — | |
| IL-10 | 0.0367 | 0.0279 | 0.0250 | 0.2635 | 0.7325 | 0.8037 | 0.9412 | |
| IL-12(p40) | 0.1895 | — | — | — | — | — | — | |
| IL-13 | <0.0001 | 0.0556 | 0.0066 | <0.0001* | 0.1368 | 0.0004* | 0.0765 | |
| IL-15 | 0.9146 | — | — | — | — | — | — | |
| IL-17 | 0.4539 | — | — | — | — | — | — | |
| GM-CSF | 0.0013 | 0.4064 | 0.0012* | 0.0104 | 0.0192 | 0.0377 | 0.7681 | |
| MCP-1 | 0.1643 | — | — | — | — | — | — | |
| MCP-3 | 0.0155 | 0.0149 | 0.0120 | 0.2948 | 0.5310 | 1 | 0.9105 | |
| MIP-1β | <0.0001 | 0.4630 | 0.0041 | <0.0001* | 0.0404 | 0.0018 | 0.1404 | |
| RANTES | 0.1390 | — | — | — | — | — | — | |
| PPD | IFN-γ | <0.0001 | <0.0001* | <0.0001* | <0.0001* | 0.0832 | 0.0025 | 0.1845 |
| IP-10 | <0.0001 | <0.0001* | <0.0001* | 0.0007* | 0.2134 | 0.5877 | 0.0270 | |
| TNF-α | <0.0001 | <0.0001* | <0.0001* | <0.0001* | 0.0180 | 0.0016* | 0.0553 | |
| IL-1ra | 0.0040 | 0.0154 | 0.6843 | 0.0044 | 0.0570 | 0.3203 | 0.0032 | |
| IL-2 | <0.0001 | <0.0001* | <0.0001* | <0.0001* | 0.0043 | 0.0025 | 0.4610 | |
| IL-6 | <0.0001 | <0.0001* | <0.0001* | 0.0001* | 0.5747 | 0.0100 | 0.0122 | |
| IL-8 | 0.0719 | — | — | — | — | — | — | |
| IL-10 | 0.0007 | 0.3737 | 0.0022 | 0.0058 | 0.0790 | 0.0239 | 0.0004* | |
| IL-12(p40) | <0.0001 | <0.0001* | 0.0002* | <0.0001* | 0.3602 | 0.0377 | 0.5070 | |
| IL-13 | 0.0003 | 0.0007* | 0.0061 | 0.0173 | 0.8073 | 0.8922 | 0.3763 | |
| IL-15 | 0.2734 | — | — | — | — | — | — | |
| IL-17 | 0.3771 | — | — | — | — | — | — | |
| GM-CSF | <0.0001 | <0.0001* | <0.0001* | 0.0002* | 0.4068 | 0.1245 | 0.4610 | |
| MCP-1 | 0.4549 | — | — | — | — | — | — | |
| MCP-3 | 0.2129 | — | — | — | — | — | — | |
| MIP-1β | <0.0001 | 0.0004* | 0.0002* | 0.0024* | 0.3667 | 0.3907 | 0.7681 | |
| RANTES | 0.3126 | — | — | — | — | — | — |
To determine the potential for diagnostic use of the seven cytokine responses that were found to have discriminatory ability in the previous analyses, receiver operating characteristic analyses were performed. For this purpose the data from patients with LTBI and active TB were grouped together (TB-infected group; case values), and compared with the data from the uninfected group (control values). At their optimal cut-off values, IP-10, TNF-α, and IL-2 achieved sensitivity and specificity values close to, or exceeding those of IFN-γ (Table 4). The best performance characteristics were observed with IL-2 in PPD-stimulated samples (sensitivity, 100%; specificity, 96%).
| IFN-γ | IP-10 | TNF-α | IL-1ra | IL-2 | IL-13 | MIP-1β | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cut-off | Sens. (%) | Spec. (%) | Cut-off | Sens. (%) | Spec. (%) | Cut-off | Sens. (%) | Spec. (%) | Cut-off | Sens. (%) | Spec. (%) | Cut-off | Sens. (%) | Spec. (%) | Cut-off | Sens. (%) | Spec. (%) | Cut- off | Sens. (%) | Spec. (%) | |
| ESAT-6 | 10.2 | 90.9 | 97.3 | 10,820.0 | 95.5 | 96.0 | 5.99 | 95.5 | 88.0 | −28.5 | 81.8 | 69.3 | 16.7 | 95.5 | 97.3 | 0.32 | 77.3 | 89.3 | 158.6 | 90.9 | 70.7 |
| CFP-10 | 0.8 | 81.8 | 89.3 | 2,204.0 | 86.4 | 93.3 | 1.9 | 86.4 | 85.3 | −23.6 | 77.3 | 78.7 | 4.7 | 90.9 | 94.7 | 0.05 | 63.6 | 86.7 | 50.2 | 77.3 | 68.0 |
| PPD | 659.4 | 95.5 | 97.3 | 4,6571.0 | 95.5 | 81.3 | 125.3 | 95.5 | 84.0 | 279.7 | 50.0 | 46.7 | 383.7 | 100.0 | 96.0 | 103.5 | 77.3 | 69.3 | 9618.0 | 81.8 | 60.0 |
In addition, receiver operating characteristic curves were constructed for each cytokine/stimulant combination (Figure 2). The very high area under the curve values for IFN-γ, IP-10, TNF-α, and IL-2 (irrespective of the stimulant used) further support the potential of these cytokines to discriminate between TB-uninfected and TB-infected individuals. Notably, the area under the curve values of IP-10, TNF-α, and IL-2 in ESAT-6– and CFP-10–stimulated samples universally exceeded those of IFN-γ.
Figure 2. Receiver operating characteristic curves for IFN-γ, interferon-inducible protein-10 (IP-10), tumor necrosis factor α (TNF-α), IL-1ra, IL-2, IL-13, and macrophage inflammatory protein-1β (MIP-1β) responses according to antigenic stimulant (early secretory antigenic target-6 [ESAT-6], 10-kD culture filtrate protein [CFP-10], or PPD). Case values comprised background-corrected cytokine concentrations in participants with latent tuberculosis infection or active tuberculosis (tuberculosis-infected group); control values comprised cytokine concentrations in uninfected participants. AUC = area under the curve.
[More] [Minimize]Among the cytokines investigated TNF-α, IL-1ra, and IL-10 responses were found to have the greatest ability to discriminate between LTBI and active TB (Figure 3). At a cut-off of 80 pg/ml in ESAT-6–stimulated samples, and 40 pg/ml in CFP-10–stimulated samples, TNF-α responses correctly classified 81.8% and 86.4% of the participants, respectively. In PPD-stimulated samples IL-1ra responses (cut-off, 450 pg/ml) and IL-10 responses (cut-off, 100 pg/ml) correctly classified 90.9% and 100% of the participants, respectively. Figure 4 shows that combining TNF-α with either of these two cytokines results in very high levels of correct classification. The combination TNF-α/IL-1ra only classified one participant with LTBI falsely as “active TB” (95.5% correct classification); the combination TNF-α/IL-10 classified all participants correctly.
Figure 3. Background-corrected tumor necrosis factor (TNF)-α, IL-1ra, and IL-10 concentrations in participants with latent tuberculosis infection (LTBI) (circles) and active tuberculosis (TB) (squares) in early secretory antigenic target-6 (ESAT-6)–, 10-kD culture filtrate protein (CFP-10)–, and PPD-stimulated samples. The bars indicate the median concentrations. The dotted lines indicate the optimal cut-offs for the distinction between participants with LTBI and those with active TB. Statistical differences were analyzed using Mann-Whitney U tests; the corresponding P values are shown above each bracket.
[More] [Minimize]
Figure 4. Background-corrected tumor necrosis factor (TNF)-α, IL-1ra, and IL-10 concentrations in uninfected participants (gray diamonds), participants with latent tuberculosis infection (LTBI) (orange circles), and active tuberculosis (TB) (red squares) in PPD-stimulated samples. (A) Combination of TNF-α/IL-1ra. (B) Combination of TNF-α/IL-10. The dotted lines indicate the optimal cut-offs for the distinction between participants with LTBI and those with active TB (TNF-α, 800 pg/ml; IL-1ra, 450 pg/ml; IL-10, 100 pg/ml).
[More] [Minimize]Our data show that in addition to IFN-γ, which forms the basis of existing immunodiagnostic tests for TB, several other MTB-specific cytokine responses, namely IP-10, TNF-α, IL-1ra, IL-2, IL-13, and MIP-1β responses, differ significantly between individuals infected with MTB and those with no evidence of infection, indicating their potential as diagnostic biomarkers of TB infection. Notably, the performance characteristics of some of these cytokine responses, including IP-10, TNF-α, and IL-2, were similar to or exceeded those of IFN-γ.
Importantly, our data also show that certain cytokine responses, including TNF-α, IL-1ra, and IL-10, in addition to identifying TB infection, may simultaneously allow the distinction between LTBI and active TB. Furthermore, we have shown that high levels of discriminatory accuracy can be achieved by combining these biomarkers. This potentially represents a significant advance, because current immunodiagnostic tests (i.e., TST and IGRA) are unable to make this distinction (5). From a clinical perspective, the ability to discriminate between LTBI and active TB based on a blood test alone, which can provide a result within 2 days, would be an important advantage, because this would allow clinicians to make timely management decisions, rather than having to wait for culture results, which can take several weeks.
Several recent studies have aimed to identify biomarkers of TB infection, and some have also attempted to identify biomarkers that differ between cases with LTBI and active TB. Harari and coworkers (23), who used multicolor flow cytometry to analyze responses to stimulation with RD1 antigens (ESAT-6 and CFP-10), reported that MTB-specific CD4+ T cells producing only TNF-α are the hallmark of active TB, whereas polyfunctional CD4+ T cells producing IFN-γ, IL-2, and TNF-α are characteristic of LTBI. Based on their findings, the authors suggested measurement of single-positive TNF-α+ CD4+ T cells could be used in the diagnostic setting to distinguish between both infection states. However, the high cost of flow cytometry and the need for highly trained personnel limit its usefulness in resource-limited, high TB prevalence settings where better TB diagnostics are needed most. Our data suggest that phenotypic analysis of the cellular origin of TNF-α may not be necessary, and that cytokine measurement in supernatant, which could be achieved with much simpler methods, is likely to be sufficient.
Several studies have previously highlighted the diagnostic potential of IP-10 (CXCL10) (24–30). In agreement with our findings, those studies showed that MTB-specific IP-10 responses in TB-infected individuals are of greater magnitude than IFN-γ responses, and that this biomarker lacks the ability to distinguish between LTBI and active TB (24–28). This is not surprising because IP-10 production in polymorphonuclear neutrophils and monocytes is primarily induced by IFN-γ that also lacks this discriminatory ability (as reflected by IGRAs lacking this ability) (31). MTB-specific IP-10 responses likely represent an amplified read-out of IFN-γ responses, meaning they differ quantitatively, but not qualitatively.
In our study, IL-2 responses had greater sensitivity and specificity than IFN-γ responses for distinguishing between TB-uninfected and TB-infected individuals. Animal models show that IL-2 plays an important role in the antimycobacterial host immune response (32, 33), but few studies in humans have investigated the diagnostic potential of MTB-specific IL-2 responses (25, 26, 34–39). Similar to our findings, two of these studies reported that RD1 antigen-induced IL-2 responses differ between TB-uninfected and TB-infected individuals, but not between LTBI and active TB (26, 36). Interestingly, data from one study using ELISpot assays suggested that combining the measurement of IFN-γ responses with IL-2 responses results in increased sensitivity for detecting TB infection (37). However, ELISpot assays are labor-intensive and difficult to integrate into a routine diagnostic laboratory setting. Our data show that measuring IL-2 in supernatants from whole-blood assays is a suitable, more practicable alternative.
Few studies have investigated MTB-specific IL-1ra responses as a diagnostic tool (26, 40, 41). Similar to our findings, one study found that RD1 antigen-induced IL-1ra responses were significantly higher in active TB than in LTBI (26). However, there was a substantial overlap between the IL-1ra responses in both groups, suggesting that a categorical separation between infection states based on this biomarker alone would be difficult. Interestingly, we found that although there was considerable overlap between the magnitude of IL-1ra responses in LTBI and active TB cases when RD1 antigens were used for stimulation, there was very little overlap when PPD was used. The likely explanation for this is that PPD stimulation resulted in several-fold higher IL-1ra responses in cases with active TB compared with RD1 antigen stimulation (Figure 1), thereby resulting in greater differences between the two groups.
We also found that combining cytokine biomarkers resulted in accurate discrimination between LTBI and active TB. The combination of TNF-α and IL-1ra responses correctly classified 95.5% of cases, whereas the combination of TNF-α and IL-10 responses resulted in correct classification of all cases. The concept of combining two or more cytokine biomarkers to improve the distinction between infection states has been explored previously. Frahm and coworkers (26) used a model based on MCP-1 and IL-15 responses, but only achieved 86.4% correct classification with this approach. Wang and coworkers (39) used an IL-2/IFN-γ ratio, which achieved a sensitivity and specificity of 77.2% and 87.2%, respectively. A study by Sutherland and coworkers (42) that compared antimycobacterial cytokine responses in patients with LTBI (defined as TST-positive individuals) with active TB patients reported that in PPD-stimulated samples a combination of TNF-α, IL-12(p40), and IL-13 responses correctly classified 81% of cases.
This study has also produced interesting data in relation to patients with discordance, who remain a significant management dilemma in clinical practice (5, 7, 10, 11). It has been postulated that a TST+/IGRA− discordant result constellation is predominately the result of false-positive TST results, primarily resulting from prior BCG vaccination. However, although BCG vaccination can produce false-positive TST results due to cross-reactivity of antigens present in PPD and antigens produced by BCG vaccine strains, solid evidence to support that this accounts for the majority of cases with discordance is lacking. Notably, the results of a metaanalysis, involving more than 200,000 children in 24 studies, suggest that only 8.5% of individuals BCG-vaccinated in infancy develop a false-positive TST (defined as ≥10 mm induration) (43), indicating that BCG vaccination alone is unlikely to account for the large proportions of discordance reported in most pediatric IGRA studies (44–46). In our study, we observed multiple statistically significant differences between the cytokine responses in the uninfected group and the common discordance group. Importantly, these differences were also detected in samples stimulated with ESAT-6 and CFP-10, which cannot be explained by antigenic cross-reactivity, because both peptides are absent from all BCG vaccine strains (8). However, there was also considerable overlap between the responses observed in both groups, with some patients with common discordance showing no response to antigenic stimulation, highlighting that these patients are likely a heterogeneous group comprising both TB-infected and TB-uninfected individuals.
For all cytokines identified as potential TB biomarkers in this study there are convincing data indicating that they play an important role in the antimycobacterial host immune response. The central role of TNF-α in the human immune response to mycobacterial infections is undisputed (15). Multiple studies have shown that patients treated with TNF-α inhibitors for autoimmune conditions are at significant risk of TB progression (47, 48). The importance of IL-1ra in the antimycobacterial immune response, particularly related to granuloma formation, has been extensively documented in animal models (49). Recent reports of TB reactivation in patients treated with monoclonal IL-1ra further highlight the importance of this cytokine in this setting (48). Infection of murine macrophages with MTB has been shown to result in up-regulation of MIP-1β expression (50). Also, progression of TB disease is associated with increased MIP-1β expression in murine lung tissue (50). In vitro studies in humans have shown that infection of macrophages with MTB results in increased production of MIP-1β, which suppresses intracellular growth of MTB (51). IL-13 is a key cytokine in the alternative activation of macrophages, which is associated with macrophage fusion and granuloma formation, a critical event in mycobacterial infections (52, 53).
The main limitation of this study is the inclusion of a limited number of patients with active TB, a limitation shared by many other studies in this area (23, 26, 27, 29, 34, 38). Nonetheless the differences in cytokine responses were sufficiently large to enable the detection of statistical differences between the diagnostic groups. An important strength of this study is the use of unambiguous diagnostic groups in the analyses. Many previous immunodiagnostic studies have included patients with uncertain TB infection status (e.g., solely based on TST results) or uncertain active TB cases (e.g., “possible” and “probable” TB cases based on clinical features alone), and therefore have an inherent risk for data contamination to occur. One limitation of all diagnostic biomarker studies is that cut-offs and performance characteristics are based on the data from the study population, thus potentially overestimating their performance. Our study population did not include cases at the most severe end of the disease spectrum (i.e., miliary TB or TB meningitis); therefore, the performance of our cytokine biomarkers in these patients remains uncertain. In addition, most children with active TB were older than 10 years of age and it is therefore uncertain whether their performance is equally robust in infants and young children. Consequently, further evaluation of these biomarkers is required in larger cohorts with a broad range of disease manifestations. Further studies in both children and adults are currently ongoing.
In conclusion, our study shows that several MTB-specific cytokine responses, including IP-10, TNF-α, IL-1ra, IL-2, IL-13, and MIP-1β, allow the distinction between individuals infected with TB and those without TB infection. Importantly, some of these biomarkers had better performance characteristics than IFN-γ. In addition, we have identified biomarkers that distinguish between LTBI and active TB potentially with high levels of accuracy. Incorporation of these biomarkers in future immunodiagnostic assays for TB could result in substantial gains in assay sensitivity, and may allow the distinction between LTBI and active TB based on a blood test alone.
The authors thank the Victorian Department of Health Tuberculosis Control Team, in particular Lynne Brown and Lucy Cosentino, for their help with recruitment, and Dr. David Leslie and Dr. Norbert Ryan at the Victorian Infectious Diseases Reference Laboratory for their assistance with the QuantiFERON-TB Gold assays. They also thank the children and parents who kindly agreed to participate in this study.
| 1. | Dye C. Global epidemiology of tuberculosis. Lancet 2006;367:938–940. |
| 2. | World Health Organization. Global tuberculosis report 2013 [accessed 2015 Jul 22]. Available from: http://www.who.int/tb/publications/global_report/en/ |
| 3. | Marais BJ, Gie RP, Schaaf HS, Beyers N, Donald PR, Starke JR. Childhood pulmonary tuberculosis: old wisdom and new challenges. Am J Respir Crit Care Med 2006;173:1078–1090. |
| 4. | Nicol MP, Workman L, Isaacs W, Munro J, Black F, Eley B, Boehme CC, Zemanay W, Zar HJ. Accuracy of the Xpert MTB/RIF test for the diagnosis of pulmonary tuberculosis in children admitted to hospital in Cape Town, South Africa: a descriptive study. Lancet Infect Dis 2011;11:819–824. |
| 5. | Tebruegge M, Connell T, Curtis N. Tuberculosis in children. N Engl J Med 2012;367:1568. |
| 6. | Tebruegge M, Ritz N, Curtis N, Shingadia D. Diagnostic tests for childhood tuberculosis – past imperfect, present tense and future perfect? Pediatr Infect Dis J [online ahead of print] 22 Jun 2015; DOI: 10.1097/INF.0000000000000796. |
| 7. | Tebruegge M, Connell T, Ritz N, Bryant PA, Curtis N. Discordance between TSTs and IFN-gamma release assays: the role of NTM and the relevance of mycobacterial sensitins. Eur Respir J 2010;36:214–215. |
| 8. | Andersen P, Munk ME, Pollock JM, Doherty TM. Specific immune-based diagnosis of tuberculosis. Lancet 2000;356:1099–1104. |
| 9. | Sester M, Sotgiu G, Lange C, Giehl C, Girardi E, Migliori GB, Bossink A, Dheda K, Diel R, Dominguez J, et al. Interferon-γ release assays for the diagnosis of active tuberculosis: a systematic review and meta-analysis. Eur Respir J 2011;37:100–111. |
| 10. | Connell T, Tebruegge M, Ritz N, Curtis N. Interferon-gamma release assays for the diagnosis of tuberculosis. Pediatr Infect Dis J 2009;28:758–759. |
| 11. | Connell TG, Tebruegge M, Ritz N, Bryant P, Curtis N. The potential danger of a solely interferon-gamma release assay-based approach to testing for latent Mycobacterium tuberculosis infection in children. Thorax 2011;66:263–264. |
| 12. | Connell TG, Tebruegge M, Ritz N, Bryant PA, Leslie D, Curtis N. Indeterminate interferon-gamma release assay results in children. Pediatr Infect Dis J 2010;29:285–286. |
| 13. | Haustein T, Ridout DA, Hartley JC, Thaker U, Shingadia D, Klein NJ, Novelli V, Dixon GL. The likelihood of an indeterminate test result from a whole-blood interferon-gamma release assay for the diagnosis of Mycobacterium tuberculosis infection in children correlates with age and immune status. Pediatr Infect Dis J 2009;28:669–673. |
| 14. | Tebruegge M, de Graaf H, Sukhtankar P, Elkington P, Marshall B, Schuster H, Patel S, Faust SN. Extremes of age are associated with indeterminate QuantiFERON-TB gold assay results. J Clin Microbiol 2014;52:2694–2697. |
| 15. | Kaufmann SH. How can immunology contribute to the control of tuberculosis? Nat Rev Immunol 2001;1:20–30. |
| 16. | Ruhwald M, Ravn P. Biomarkers of latent TB infection. Expert Rev Respir Med 2009;3:387–401. |
| 17. | Clifford V, Zufferey C, Street A, Denholm J, Tebruegge M, Curtis N. Cytokine biomarkers for monitoring anti-tuberculous therapy: a systematic review. Tuberculosis (Edinb) 2015;95:217–228. |
| 18. | Naseer A, Naqvi S, Kampmann B. Evidence for boosting Mycobacterium tuberculosis-specific IFN-gamma responses at 6 weeks following tuberculin skin testing. Eur Respir J 2007;29:1282–1283. |
| 19. | Ritz N, Yau C, Connell TG, Tebruegge M, Leslie D, Curtis N. Absence of interferon-gamma release assay conversion following tuberculin skin testing. Int J Tuberc Lung Dis 2011;15:767–769. |
| 20. | Globan M, Fyfe J. Mycobacterium tuberculosis complex. In: Schuller M, Sloots TP, James GS, Halliday CL, Carter IWJ, editors. PCR for clinical microbiology, 1st ed. New York, NY: Springer; 2010. pp. 165–170. |
| 21. | American Thoracic Society and the Centers for Disease Control and Prevention. Diagnostic standards and classification of tuberculosis in adults and children. Official statement of the American Thoracic Society and the Centers for Disease Control and Prevention. Am J Respir Crit Care Med 2000;161:1376–1395. |
| 22. | Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003;3:25. |
| 23. | Harari A, Rozot V, Bellutti Enders F, Perreau M, Stalder JM, Nicod LP, Cavassini M, Calandra T, Blanchet CL, Jaton K, et al. Dominant TNF-α+ Mycobacterium tuberculosis-specific CD4+ T cell responses discriminate between latent infection and active disease. Nat Med 2011;17:372–376. |
| 24. | Alsleben N, Ruhwald M, Rüssmann H, Marx FM, Wahn U, Magdorf K. Interferon-gamma inducible protein 10 as a biomarker for active tuberculosis and latent tuberculosis infection in children: a case-control study. Scand J Infect Dis 2012;44:256–262. |
| 25. | Armand M, Chhor V, de Lauzanne A, Guérin-El Khourouj V, Pédron B, Jeljeli M, Gressens P, Faye A, Sterkers G. Cytokine responses to quantiferon peptides in pediatric tuberculosis: a pilot study. J Infect 2014;68:62–70. |
| 26. | Frahm M, Goswami ND, Owzar K, Hecker E, Mosher A, Cadogan E, Nahid P, Ferrari G, Stout JE. Discriminating between latent and active tuberculosis with multiple biomarker responses. Tuberculosis (Edinb) 2011;91:250–256. |
| 27. | Whittaker E, Gordon A, Kampmann B. Is IP-10 a better biomarker for active and latent tuberculosis in children than IFNgamma? PLoS One 2008;3:e3901. |
| 28. | Yassin MA, Petrucci R, Garie KT, Harper G, Arbide I, Aschalew M, Merid Y, Kebede Z, Bawazir AA, Abuamer NM, et al. Can interferon-gamma or interferon-gamma-induced-protein-10 differentiate tuberculosis infection and disease in children of high endemic areas? PLoS One 2011;6:e23733. |
| 29. | Ruhwald M, Bjerregaard-Andersen M, Rabna P, Eugen-Olsen J, Ravn P. IP-10, MCP-1, MCP-2, MCP-3, and IL-1RA hold promise as biomarkers for infection with M. tuberculosis in a whole blood based T-cell assay. BMC Res Notes 2009;2:19. |
| 30. | Ruhwald M, Bjerregaard-Andersen M, Rabna P, Kofoed K, Eugen-Olsen J, Ravn P. CXCL10/IP-10 release is induced by incubation of whole blood from tuberculosis patients with ESAT-6, CFP10 and TB7.7. Microbes Infect 2007;9:806–812. |
| 31. | Gasperini S, Marchi M, Calzetti F, Laudanna C, Vicentini L, Olsen H, Murphy M, Liao F, Farber J, Cassatella MA. Gene expression and production of the monokine induced by IFN-gamma (MIG), IFN-inducible T cell alpha chemoattractant (I-TAC), and IFN-gamma-inducible protein-10 (IP-10) chemokines by human neutrophils. J Immunol 1999;162:4928–4937. |
| 32. | Colizzi V. In vivo and in vitro administration of interleukin 2-containing preparation reverses T-cell unresponsiveness in Mycobacterium bovis BCG-infected mice. Infect Immun 1984;45:25–28. |
| 33. | Hubbard RD, Collins FM. Immunomodulation of mouse macrophage killing of Mycobacterium avium in vitro. Infect Immun 1991;59:570–574. |
| 34. | Biselli R, Mariotti S, Sargentini V, Sauzullo I, Lastilla M, Mengoni F, Vanini V, Girardi E, Goletti D, D’ Amelio R, et al. Detection of interleukin-2 in addition to interferon-gamma discriminates active tuberculosis patients, latently infected individuals, and controls. Clin Microbiol Infect 2010;16:1282–1284. |
| 35. | Casey R, Blumenkrantz D, Millington K, Montamat-Sicotte D, Kon OM, Wickremasinghe M, Bremang S, Magtoto M, Sridhar S, Connell D, et al. Enumeration of functional T-cell subsets by fluorescence-immunospot defines signatures of pathogen burden in tuberculosis. PLoS One 2010;5:e15619. |
| 36. | Chiappini E, Della Bella C, Bonsignori F, Sollai S, Amedei A, Galli L, Niccolai E, Del Prete G, Singh M, D’Elios MM, et al. Potential role of M. tuberculosis specific IFN-γ and IL-2 ELISPOT assays in discriminating children with active or latent tuberculosis. PLoS One 2012;7:e46041. |
| 37. | Krummel B, Strassburg A, Ernst M, Reiling N, Eker B, Rath H, Hoerster R, Wappler W, Glaewe A, Schoellhorn V, et al. Potential role for IL-2 ELISpot in differentiating recent and remote infection in tuberculosis contact tracing. PLoS One 2010;5:e11670. |
| 38. | Lighter-Fisher J, Peng CH, Tse DB. Cytokine responses to QuantiFERON® peptides, purified protein derivative and recombinant ESAT-6 in children with tuberculosis. Int J Tuberc Lung Dis 2010;14:1548–1555. |
| 39. | Wang S, Diao N, Lu C, Wu J, Gao Y, Chen J, Zhou Z, Huang H, Shao L, Jin J, et al. Evaluation of the diagnostic potential of IP-10 and IL-2 as biomarkers for the diagnosis of active and latent tuberculosis in a BCG-vaccinated population. PLoS One 2012;7:e51338. |
| 40. | Chegou NN, Detjen AK, Thiart L, Walters E, Mandalakas AM, Hesseling AC, Walzl G. Utility of host markers detected in Quantiferon supernatants for the diagnosis of tuberculosis in children in a high-burden setting. PLoS One 2013;8:e64226. |
| 41. | Yu Y, Zhang Y, Hu S, Jin D, Chen X, Jin Q, Liu H. Different patterns of cytokines and chemokines combined with IFN-γ production reflect Mycobacterium tuberculosis infection and disease. PLoS One 2012;7:e44944. |
| 42. | Sutherland JS, de Jong BC, Jeffries DJ, Adetifa IM, Ota MO. Production of TNF-alpha, IL-12(p40) and IL-17 can discriminate between active TB disease and latent infection in a West African cohort. PLoS One 2010;5:e12365. |
| 43. | Farhat M, Greenaway C, Pai M, Menzies D. False-positive tuberculin skin tests: what is the absolute effect of BCG and non-tuberculous mycobacteria? Int J Tuberc Lung Dis 2006;10:1192–1204. |
| 44. | Connell TG, Ritz N, Paxton GA, Buttery JP, Curtis N, Ranganathan SC. A three-way comparison of tuberculin skin testing, QuantiFERON-TB gold and T-SPOT.TB in children. PLoS One 2008;3:e2624. |
| 45. | Connell TG, Curtis N, Ranganathan SC, Buttery JP. Performance of a whole blood interferon gamma assay for detecting latent infection with Mycobacterium tuberculosis in children. Thorax 2006;61:616–620. |
| 46. | Hill PC, Brookes RH, Adetifa IM, Fox A, Jackson-Sillah D, Lugos MD, Donkor SA, Marshall RJ, Howie SR, Corrah T, et al. Comparison of enzyme-linked immunospot assay and tuberculin skin test in healthy children exposed to Mycobacterium tuberculosis. Pediatrics 2006;117:1542–1548. |
| 47. | Wallis RS. Infectious complications of tumor necrosis factor blockade. Curr Opin Infect Dis 2009;22:403–409. |
| 48. | Singh JA, Wells GA, Christensen R, Tanjong Ghogomu E, Maxwell L, Macdonald JK, Filippini G, Skoetz N, Francis D, Lopes LC, et al. Adverse effects of biologics: a network meta-analysis and Cochrane overview. Cochrane Database Syst Rev 2011;2:CD008794. |
| 49. | Ruth JH, Bienkowski M, Warmington KS, Lincoln PM, Kunkel SL, Chensue SW. IL-1 receptor antagonist (IL-1ra) expression, function, and cytokine-mediated regulation during mycobacterial and schistosomal antigen-elicited granuloma formation. J Immunol 1996;156:2503–2509. |
| 50. | Lyadova IV, Tsiganov EN, Kapina MA, Shepelkova GS, Sosunov VV, Radaeva TV, Majorov KB, Shmitova NS, van den Ham HJ, Ganusov VV, et al. In mice, tuberculosis progression is associated with intensive inflammatory response and the accumulation of Gr-1 cells in the lungs. PLoS One 2010;5:e10469. |
| 51. | Saukkonen JJ, Bazydlo B, Thomas M, Strieter RM, Keane J, Kornfeld H. Beta-chemokines are induced by Mycobacterium tuberculosis and inhibit its growth. Infect Immun 2002;70:1684–1693. |
| 52. | DeFife KM, Jenney CR, Colton E, Anderson JM. Cytoskeletal and adhesive structural polarizations accompany IL-13-induced human macrophage fusion. J Histochem Cytochem 1999;47:65–74. |
| 53. | Gordon S. Alternative activation of macrophages. Nat Rev Immunol 2003;3:23–35. |
Supported by the Australian National Health and Medical Research Council (Project Grant No. 1011200), the John Burge Trust, the Myer Foundation, the Aranday Foundation, and the Murdoch Children’s Research Institute. M.T. and N.R. were supported by Fellowship awards from the European Society for Paediatric Infectious Diseases and by International Research Scholarship awards from The University of Melbourne.
Author Contributions: Conception and design, M.T., B.D., S.D., N.R., R.R.-B., S.M.G., T.C., and N.C. Acquisition of data, M.T., B.D., B.F., and K.C.-B. Analysis and interpretation of data, M.T., B.D., S.D., N.R., B.F., K.C.-B., V.C., C.Z., R.R.-B., W.H., S.M.G., T.C., and N.C. Drafting the manuscript, M.T. and N.C. Revising the manuscript for important intellectual content, B.D., S.D., N.R., B.F., K.C.-B., V.C., C.Z., R.R.-B., W.H., S.M.G., and T.C.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.201501-0059OC on June 1, 2015
Author disclosures are available with the text of this article at www.atsjournals.org.
