Interferon-γ Release Assays Improve the Diagnosis of Tuberculosis and Nontuberculous Mycobacterial Disease in Children in a Country with a Low Incidence of Tuberculosis
- 1Department of Pediatric Pneumology and Immunology, Charité University Medical Center, Berlin, Germany
- 2Institute for Social Medicine, Epidemiology and Health Economics, Charité University Medical Center, Berlin, Germany
- 3Department of Pediatric Pneumology and Allergy, Berlin, Germany
- 4Department of Microbiology, Helios Klinikum Emil von Behring, Department Heckeshorn, Berlin, Germany
- 5German Central Committee against Tuberculosis, Berlin, Germany
- Reprints or correspondence: Dr. Anne Detjen, Dept. of Pediatric Pneumology and Immunology, Charité Campus Virchow Klinikum, Augustenburger Platz 1, 13353 Berlin, Germany (anne.detjen{at}charite.de).
Abstract
Background. Diagnosis of childhood tuberculosis (TB) is challenging. The widely used tuberculin skin test (TST) may produce -positive results because of cross-reactivity with nontuberculous mycobacteria or bacille Calmette-Guérin vaccination, resulting in unnecessary treatment. Two recently developed interferon-γ release assays (IGRAs) show good diagnostic accuracy for active TB in adults; pediatric data are limited, particularly in areas with a low incidence of TB. We assessed the diagnostic accuracy of IGRAs for TB in children in an area with a low incidence of TB.
Methods. In a hospital-based study, the diagnostic accuracy of the TST and 2 IGRAs (T SPOT-TB [T-SPOT; Oxford Immunotec] and QuantiFERON-TB Gold In-Tube [QFT-IT; Cellestis]) were assessed in a cohort of 73 children (median age, 39 months); 28 children with bacteriologically confirmed TB were compared with children without TB (23 with bacteriologically confirmed nontuberculous mycobacterial lymphadenitis and 22 with other nonmycobacterial respiratory tract infections).
Results. The specificity for TB of QFT-IT was 100% (95% confidence interval [CI], 91%–100%), and the specificity of T-SPOT was 98% (95% CI, 87%–100%), both of which were considerably higher than the specificity of TST (58%; 95% CI, 42%–73%). The specificity of the TST was 10.5% (95% CI, 1%–33%) in children with nontuberculous mycobacterial lymphadenitis and was 100% (95% CI, 83%–100%) in children with other nonmycobacterial respiratory tract infections. The sensitivity of both QFT-IT and T-SPOT was 93% (95% CI, 77%–99%), and the sensitivity of the TST was 100% (95% CI, 88%–100%). Agreement between the IGRAs was 95.6% (κ = 0.91); 6.8% of the IGRAs showed indeterminate results.
Conclusions. Both IGRAs showed high diagnostic value in bacteriologically confirmed childhood TB. Their advantage in this study, when performed in addition to the TST, was the ability to distinguish -positive TST results caused by nontuberculous mycobacterial disease, thereby reducing overdiagnosis of TB and guiding clinical management.
The diagnosis of childhood tuberculosis (TB) is complicated by nonspecific clinical and radiological presentation and bacteriological confirmation in the minority of children. The specificity of the tuberculin skin test (TST) is low, because of cross-reactions with bacille Calmette-Guérin (BCG) vaccination or infection with nontuberculous mycobacteria (NTM). In Germany, routine BCG vaccination was discontinued in 1998; therefore, BCG cross-reactivity only plays a role in immigrant children. Although a continuing decrease in TB is evident in countries with a low incidence of disease, there is an apparent increase in NTM infection in such countries. The most common NTM that causes clinical disease in children in central Europe is Mycobacterium avium, which mainly causes cervical lymphadenitis [1].
A new approach for the immunodiagnosis of TB has recently been introduced that uses IFN-γ release assays (IGRAs), which measure the IFN-γ release of sensitized T cells after in vitro stimulation with antigens that are specific for Mycobacterium tuberculosis (early secreted antigenic target 6 [ESAT-6] and culture filtrate protein 10 [CFP-10]; the QuantiFERON-TB Gold In-Tube [QFT-IT; Cellestis] also incorporates TB7.7). Two IGRAs are commercially available: the PBMC-based T SPOT-TB (T-SPOT; Oxford Immunotec), which is an enzyme-linked immunospot assay, and the whole blood–based QFT-IT, which is an ELISA. The QFT-IT is the latest QuantiFERON test generation in which the blood collection tubes contain the antigens for stimulation. The antigens used are encoded within the region of difference 1 of the genome of M. tuberculosis, which is deleted from all BCG strains, and most NTM species, with the exception of Mycobacterium kansasii, Mycobacterium marinum, Mycobacterium flavescens, and Mycobacterium szulgai. Therefore, these antigens are highly specific for M. tuberculosis [2].
Recent data on the use of IGRAs for the diagnosis of active TB demonstrate sensitivities similar to those for the TST, with values ranging from 71%–97% for the QuantiFERON test (mostly using the predecessor of the QFT–IT version) and 80%–97% for the T-SPOT [3, 4]. Specificities of 90%–98% and 91%–100% were found for the QuantiFERON test and the T-SPOT, respectively [3]. In immunocompromised patients, the sensitivity of these tests seems to be superior to the sensitivity of the TST [5, 6]. Diagnosis of latent TB infection (LTBI) is difficult because of the lack of a gold standard, but IGRAs seem to be at least as sensitive as the TST, with a higher specificity in low-risk populations and a better correlation with TB exposure [7,8,9,10–11]. To our knowledge, only 3 studies to date have directly compared the 2 commercially available IGRAs [12,13–14].
In the United Kingdom, the National Institute of Health and Clinical Excellence (NICE) recently recommended the routine use of IGRAs in the diagnosis of LTBI for confirming positive TST results [15]. This approach has not been prospectively validated. The US Centers for Disease Control and Prevention recommend the use of QuantiFERON-TB Gold in all cases in which TST is indicated [16]. Both guidelines imply the use of IGRAs in children; however, data on the routine use of IGRAs in clinical settings are limited.
We examined the diagnostic value of both commercially available IGRAs, compared with the TST, in the diagnosis of TB, and we assessed their ability to distinguish TB from NTM disease in a region where the incidence of TB is low.
Methods
Participants. In this hospital-based study, 91 children aged 4 months to 15 years were enrolled from December 2004 through March 2006. of these children, 64 were seen at the Pediatric Department of the Chest Hospital Heckeshorn (Berlin, Germany) for suspected TB, NTM disease, or other respiratory tract infection. In addition, clinical data and samples from 27 children were sent from referral hospitals. The study protocol was approved by the Ethics Committee of the University Hospital Charité, Berlin (study number EA2/28/04). Parents gave informed consent.
Eligibility criteria and study groups. Children were eligible for study participation if they were classified into 1 of 3 predefined groups: group A (confirmed TB), which comprised children with culture-confirmed TB and chest radiograph findings typical for TB; group B (NTM lymphadenitis), which comprised children with culture-confirmed NTM lymphadenitis without chest radiograph findings typical for TB, with no history of exposure to TB, and not vaccinated with BCG vaccine; and group C (other respiratory tract infections), which comprised children with other, nonmycobacterial respiratory tract infections that responded well to adequate treatment, without chest radiograph findings typical for TB, with no history of exposure to infectious TB, and not vaccinated with BCG vaccine. Children who had been receiving antituberculous treatment for ⩾4 weeks, those with immunosuppression, and those with incomplete clinical data or referrals with a sample transport period >24 h were excluded.
Radiological findings typical for pulmonary TB were defined as Ghon focus, mediastinal or hilar lymphadenopathy with or without concomitant atelectasis, pneumonic infiltration, or pleural effusion. Chest radiographs were reviewed by an experienced pediatric pulmonologist. Mycobacterial culture was routinely obtained from induced sputum samples and gastric or bronchial aspirate samples. Children with extrathoracic lymphadenitis underwent excision biopsy; mycobacterial culture was performed on biopsy tissue.
The TST was performed by experienced pediatricians, according to standard national guidelines, using intradermal injection of 10 U of PPD (GT10; Chiron Behring), with an estimated bioequivalence to 5 IU PPD-S. The transverse induration was read at 72 h. In Germany, intervention cut-off values for the TST are based on recommendations from the American Thoracic Society and the Centers for Disease Control and Prevention, considering a history of TB exposure and individual risk factors, including age and immune status [17]. In this study, to include all children with possible TB, a positive TST result was defined as an induration >5 mm in diameter. Phlebotomy for IGRAs was performed after the TST.
IGRAs were performed at the Department of Microbiology, Chest Hospital Heckeshorn, according to the manufacturers' guidelines. T-SPOT was performed on 2 mL of heparinized blood [18]. Spots were counted with a hand lens; a positive test result was defined as >5 spots. QFT-IT was performed on 2 mL of heparinized blood [19]; a positive test result was defined as >0.35 IU/mL.
General T cell reactivity was confirmed by positive mitogen control (phytohemaglutinin) for T-SPOT. The QFT-IT kits used did not contain a positive control; therefore, results were classed as indeterminate for both tests if the positive T-SPOT control failed (i.e., if <t;20 spots were counted, according to the manufacturer's instructions) and the response to specific antigens was negative.
Statistical methods. Concordance between IGRAs in the different groups was assessed using percentage agreement and κ coefficients. Sensitivity, specificity, likelihood ratios (LRs), and predictive values with confidence intervals for the TST, T-SPOT, and QFT-IT (for each test separately and for T-SPOT and QFT-IT in combination) were calculated for children with TB (group A) versus children without TB (groups B and C). McNemar's test was used to compare sensitivity and specificity between groups.
LRs were defined as ratios of the probabilities; positive LR was calculated as sensitivity/(1 - specificity), and negative LR was calculated as (1 - sensitivity)/specificity. To enable the calculation of LR in cases in which the sensitivity or specificity was 100% (thus dividing by zero), we classified 1 child with a correct diagnosis of TB as having received a false-negative result or 1 child without TB as having received a false-positive result. Statistical analysis was performed using SAS software, version 9.1 (SAS Institute). We also calculated diagnostic accuracy using a stepwise approach based on the algorithm recommended by NICE (i.e., confirmation of a positive TST result by IGRAs) [15].
Results
Study population. of 91 children screened, 73 were included and assigned to groups A–C, regardless of the TST or IGRA results; 18 children were excluded (figure 1). Patient characteristics and TST results are shown in table 1. The main differences were with respect to age and the proportion of German-born children versus children born in non–European Union countries (mainly countries in Africa or the Middle East, Russia, and Turkey). Children with TB were younger (median age, 28 months) and mostly of non-German origin; children with NTM lymphadenitis (median age, 44 months) were mostly German-born, as were the children in group C, who reflect the typical age distribution of patients with respiratory tract infection who are admitted to our hospital (median age, 52.5 months).
Figure 1
Flow diagram summarizing patient recruitment, exclusion criteria, and the patient groups. NTM, nontuberculous mycobacteria; TB, tuberculosis.
Table 1
Clinical characteristics, vaccination status, and distribution of tuberculin skin test results in children with tuberculosis (group A), nontuberculous mycobacterial lymphadenitis (group B), and other respiratory tract infections (group C).
Of 28 children classified as having bacteriologically confirmed TB (group A), 27 had pulmonary disease, and 1 had extrapulmonary disease. All children with pulmonary TB had classical findings for TB on chest radiographs. Three children were receiving antituberculous treatment for <t;4 weeks; 25 had received no treatment or <t;7 days of treatment at enrollment.
Of the children with bacteriologically confirmed NTM lymphadenitis (group B; n = 23), 22 had cervical lymphadenitis and 1 had inguinal lymphadenitis. The children assigned to group C (n = 22) had other respiratory diagnoses, including bronchitis, bacterial or viral pneumonia, cystic fibrosis, or airway hyperreactivity, that were responsive to non-TB treatment.
All children with confirmed TB (group A) had a TST result ⩾10 mm in diameter. of those with confirmed NTM lymphadenitis (group B), 18 (78.3%) of 23 had TST results of ⩾10 mm in diameter; 2 children had negative TST results (no induration). All children from group C had negative TST results.
Test agreement and indeterminate results. Statistical analysis was performed for 68 children: 28 with TB (group A) versus 40 without TB (groups B and C combined). Five children (6.8%) had indeterminate IGRA results (4 children from group B and 1 child from group C), defined as a negative test result (on T-SPOT and QFT-IT) and a negative mitogen response for T-SPOT. Cases with indeterminate results were excluded from statistical analysis.
Agreement and κ statistics between the tests are shown in table 2. Overall agreement of the TST and either T-SPOT or QFT-IT was 70%–72% (κ = 0.45 and κ = 0.48, respectively), mainly caused by the low agreement in the children with NTM lymphadenitis (5%–10%). The overall agreement of the IGRAs was 95.6% (κ = 0.91). Divergent IGRA results occurred in 3 (4.4%) of the children (2 children from group A and 1 child from group B), indicating that one of the test results was either a false-positive or false-negative result.
Table 2
Agreement and κ statistics between QFT-IT and the tuberculin skin test (TST), T-SPOT and TST, and QFT-IT and T-SPOT in children with tuberculosis (group A), nontuberculous mycobacterial lymphadenitis (group B), and other respiratory tract infections (group C).
Diagnostic accuracy of IGRAs and the TST. As demonstrated in table 3, accurate diagnosis of disease in children with active TB was seen more frequently with the TST (sensitivity, 100%; 95% CI, 88%–100%) than with either of the IGRAs used alone (sensitivity for each IGRA, 93%; 95% CI, 77%–99%) or in combination. However, the differences in sensitivity between the 3 tests were not statistically significant (data not shown). In contrast, specificity for excluding TB was significantly better using the IGRAs alone (QFT-IT specificity, 100%; 95% CI, 91%–100%; P <t; .001; T-SPOT specificity, 98%; 95% CI, 87%–100%; P <t; .001) or in combination than using the TST only (TST specificity, 58%; 95% CI, 42%–73%). The low specificity of the TST was a consequence of false-positive results among children with NTM lymphadenitis, with a specificity of only 10.5% (95% CI, 1%–33%) in group B. The specificity of the TST in group C was 100% (95% CI, 83%–100%).
Table 3
Indicators of diagnostic accuracy for the tuberculin skin test (TST), QFT-IT, T-SPOT, and the combination of QFT-IT and T-SPOT.
Using 1 of the IGRAs, the likelihood of a positive test result was 37 times higher for a child with TB than for a child without TB. The TST had a low positive LR of 2.4. For a child without TB, the likelihood of a negative result was 10 times higher than for a child with TB if the IGRAs or TST were used. The positive predictive value of the TST in this population was 62%, compared with 96%–100% for IGRAs.
If either IGRA was used in a stepwise approach as confirmation of a positive TST result, the specificity was significantly increased from 58% to 100%; sensitivity was not significantly affected (sensitivity decreased from 100% to 93%; P = .157).
False-negative results in children with TB (group A). One child in group A with extrapulmonary TB (TST induration, 20 mm) had both a negative QFT-IT result and a negative T-SPOT result; additional IGRAs continued to have negative results. In 2 other children with confirmed TB, IGRA results differed; 1 child had a positive QFT-IT result and a negative T-SPOT result (TST induration, 30 mm), and 1 child had a negative QFT-IT result and a positive T-SPOT result (TST induration, 20 mm).
Samples sent from referral hospitals. of the 27 samples sent from referral hospitals, only 16 were included in the analysis; 2 of these had indeterminate IGRA results. The 14 other samples sent by express mail within 24 h showed good mitogen response and, therefore, good T cell reactivity for T-SPOT (for most samples, >100 spots were counted).
Discussion
The use of IGRAs to diagnose TB and LTBI in children has been recommended in recent guidelines [15, 16]; however, the characteristics, accuracy, and reliability of IGRAs in children are not yet well defined [3, 20]. Studies examining IGRAs in children have mostly been conducted in countries with high burdens of disease, and the effect of infection with NTM on the specificity of the IGRAs has not been addressed [5, 21,22,23,24–25]. This study was performed to compare IGRAs with the TST for diagnosing or excluding TB in children. We demonstrate a high diagnostic accuracy of QFT-IT and T-SPOT for the immunologic diagnosis of childhood TB in a country with a low incidence of disease. These findings confirm the value of a stepwise approach (i.e., confirmation of positive TST results by IGRA), which has been suggested in recent diagnostic guidelines [15].
The incidence of TB among children in Germany was 2.0 cases per 100,000 population in 2005; immigrant children are affected ∼10-fold more often than German-born children [26]. This is reflected in our study population, in which twice as many foreign-born children as German-born children received a diagnosis of TB.
Since the discontinuation of BCG mass vaccination in countries with a low incidence of TB, there has been an increase in NTM infection [27, 28]. Germany does not routinely report cases of NTM disease; however, in The Netherlands, the estimated annual incidence of NTM infection in children aged <t;4 years was 2.3 cases per 100,000 population during the period 2001–2003, and similar estimates exist for Germany [1, 29]. This corresponds to the incidence of childhood TB in Germany and highlights the need for tests that are able to differentiate between disease caused by M. tuberculosis and disease caused by NTM. We have demonstrated a significant confounding effect of NTM infection on TST responses. It has been shown elsewhere that TST indurations can exceed 15 mm in 50% of children with NTM disease, confirming that the TST is not a reliable tool for differentiation of mycobacterial infections [30]. In the present study, nearly all children with NTM lymphadenitis had positive TST results (14.3% had indurations of 5–9 mm, 61.9% had indurations of 10–15 mm, and 23.8% had indurations >15 mm), whereas IGRAs clearly discriminated TB from NTM disease. A recent study also confirmed the ability of T-SPOT to exclude TB in 11 adults with NTM pulmonary disease [31].
In our study, the agreement between IGRAs and the TST was high in children with culture-confirmed TB, as well as in children who had not received BCG vaccination who had other respiratory tract infections (93% and 100%, respectively).
The overall agreement between QFT-IT and T-SPOT in our study was 95.6%. In 2 studies that compared T-SPOT and the QuantiFERON-TB Gold (the predecessor of the QFT-IT), overall agreement between IGRAs ranged from 79% to 84%; more individuals had positive results according to the T-SPOT, and this method gave fewer indeterminate results [12, 13]. More recently, Arend et al. [14] compared the TST, T-SPOT, and QFT-IT in a cohort of 785 adults exposed to sputum smear–positive TB, showing better correlation of IGRAs, compared with the TST, to the duration of exposure. IGRA test agreement was 89.6%. Discordant IGRAs, with positive T-SPOT results and negative QFT-IT results, were associated with negative TST results and also with immunosuppression [14].
Few studies have been conducted involving children with TB, but these data indicate a higher sensitivity of IGRAs, compared with the TST, especially in malnourished and HIV-infected children [5, 21]. The high sensitivity and specificity of IGRAs found in our study population of immunocompetent children suggest that IGRAs can reliably be used in children aged <t;3 years for diagnosing active TB (54% of the children with TB were aged <t;3 years). However, subgroups, including immunocompromised children, were not included in this study and need to be further investigated.
The Centers for Disease Control and Prevention has recommended that QuantiFERON-TB Gold can be used instead of the TST for the diagnosis of active TB and LTBI [16]. In contrast, the German Central Committee against Tuberculosis recently recommended a stepwise approach for LTBI diagnosis, as recommended by NICE [15, 32]. However, the interpretation of IGRAs versus the TST for the diagnosis of LTBI is controversial because of the lack of a diagnostic gold standard [23,24–25, 33,34–35]. Long-term follow-up studies involving people treated or not treated according to NICE guidelines are needed to confirm the value of IGRAs in diagnosing LTBI.
It is tempting to use bacteriologically confirmed TB as a reference standard for TB infection; therefore, by applying the recommended stepwise approach in our statistical calculations, we were able to show that the use of either T-SPOT or QFT-IT reduces overdiagnosis of TB in countries with a low incidence of disease. These findings have implications for the management of cases that would otherwise be misdiagnosed, because NTM lymphadenitis is usually treated by lymph node excision instead of antituberculous treatment.
It remains important to interpret TST results, as well as IGRA results, in the context of a history of exposure to TB, as well as other diagnostic measures. If we applied the NICE guidelines, which recommend a stepwise diagnostic approach, few children with TB and a positive TST result would have received a false-negative diagnosis on the basis of IGRA results. None of the children with other respiratory tract infections (group C) had a history of TB exposure, and all of them had negative TST results. However, NTM infection could play a role in this group of children and would be excluded if IGRAs were used to confirm the diagnosis in the case of a positive TST result.
We were not able to compare indeterminate results between QFT-IT and T-SPOT. Indeterminate results for both tests were defined by a negative mitogen control in the T-SPOT test. We have no explanation for the indeterminate results obtained for 5 (6.8%) of 73 patients; elsewhere, indeterminate results have been related to young age or to the extent of immunosuppression [12]. An association between young age and low mitogen response has also been shown [33]. However, the 5 children with indeterminate results in our study were aged between 22 and 68 months and were not immunosuppressed. Two of the samples with indeterminate results were sent by mail within 24 h after they were obtained. Delays before analysis may influence T cell response [36, 37]. However, all other samples sent to us by mail had good mitogen response, indicating functioning T cell reactivity.
Our study found both QFT-IT and T-SPOT to be valuable tests for diagnosing active TB in children. In addition, the IGRAs demonstrated a high specificity and, in contrast with the TST, could exclude TB in children with lymphadenitis caused by NTM. Stepwise testing with IGRAs for confirmation of a positive TST result can, therefore, be recommended for the diagnosis of TB in children in areas with a low incidence of disease.
Acknowledgments
We thank A. Hesseling and R. Gie of the Childhood TB Group at Desmond Tutu TB Centre, Stellenbosch University (Tygerberg, South Africa), for helpful discussions.
Financial support. Department of Pediatric Pneumology and Immunology, Charité Berlin, and the German Society of Pediatric Infectious Diseases.
Potential conflicts of interest. All authors: no conflicts.
- Received January 11, 2007.
- Accepted April 11, 2007.
- © 2007 of the Infectious Diseases Society of America
References
This Article
- AbstractFree
- » Full Text (HTML)Free
- Full Text (PDF)Free
Classifications
-
- Articles and Commentaries
Services
Citing Articles
Google Scholar
PubMed
Related Content
Share
- Email this article
-
CiteULike
Connotea
Delicious
Facebook
Google+
Mendeley
Twitter
Published on behalf of
IDSA Members: For your free access to this journal, log in via the IDSA members area.
The Journal
Impact Factor: 8.186
5-Yr impact factor: 7.898
For The Media
For Authors
- Services for authors
- Self-archiving Policy
- IDSA Conflict of Interest policy
- Instructions for Authors
- ICMJE Form
- Submit Now!
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy
Corporate Services
Alerting Services
Related IDSA Publications
