By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Christoph Lange MD PhD, Division of Clinical Infectious Diseases, Research Center Borstel, Parkallee 35, 23845 Borstel, Germany. (fax: +49-4537-188-313; e-mail: firstname.lastname@example.org).
Abstract. Jafari C, Kessler P, Sotgiu G, Ernst M, Lange C (Research Center Borstel, Borstel, Germany; Hygiene and Preventive Medicine Institute, University Sassari, Sassari, Italy; Research Center Borstel, Borstel, Germany). Impact of a Mycobacterium tuberculosis-specific interferon-γ release assay in bronchoalveolar lavage fluid for a rapid diagnosis of tuberculosis. J Intern Med 2011; 270: 254–262.
Objectives. Evaluation of different methods for an initial treatment decision in individuals with suspected pulmonary tuberculosis.
Background. Recently, important advances regarding the diagnosis of pulmonary tuberculosis have been introduced, which influence the decision to initiate anti-tuberculosis treatment.
Methods. To evaluate the impact of different methods for the presumed diagnosis of tuberculosis, individuals with suspected tuberculosis were prospectively enrolled following a specific algorithm including initial smear microscopy and Mycobacterium tuberculosis-specific nucleic acid amplification (NAAT) from sputum. In cases of negative initial test results, tuberculin skin testing, bronchoscopy with transbronchial biopsies and interferon-γ release assays (IGRAs) in peripheral blood and bronchoalveolar lavage (BAL) fluid were performed.
Results. Amongst 135 individuals with suspected tuberculosis, 42 had tuberculosis, 10 had nontuberculous mycobacteria pulmonary infection/colonization (one had both tuberculosis and nontuberculous mycobacteria pulmonary infection/colonization) and 84 had an alternative final diagnosis. The sensitivity and specificity were 41% and 99% [positive likelihood ratio (LR+) = 40] for sputum microscopy and 31% and 98% (LR+ = 16) for BAL nucleic acid amplification, respectively. In patients with acid-fast bacilli smear-negative tuberculosis (25/42, 59.5%), M. tuberculosis-specific BAL fluid IGRA was 92% sensitive and 87% specific (LR+ = 7) for the diagnosis of tuberculosis.
Conclusion. None of the microbiological or immunological methods that aim to provide a rapid diagnosis of tuberculosis whilst waiting the confirmation of the M. tuberculosis culture results is on its own accurate enough to diagnose or exclude pulmonary tuberculosis. Negative sputum microscopy and M. tuberculosis-specific NAAT results should prompt bronchoscopy including BAL for M. tuberculosis-specific IGRA in individuals with suspected pulmonary tuberculosis.
Tuberculosis is one of the leading causes of morbidity and mortality worldwide . The World Health Organization (WHO) has stated that one-third of the world’s population is infected with Mycobacterium tuberculosis, the bacterium that causes tuberculosis . However, this is an estimate as true infection with M. tuberculosis can only be diagnosed in the minority of individuals who develop active tuberculosis. By contrast, latent infection with M. tuberculosis is diagnosed indirectly by a positive immune reaction in the in vivo tuberculin skin test or, more recently, in an ex vivo interferon-γ release assay (IGRA) .
Detection of M. tuberculosis in liquid cultures or on solid media remains the gold standard for the diagnosis of tuberculosis. In routine clinical practice, initial treatment decisions cannot rely on this method as culture growth of M. tuberculosis takes several weeks before detection is possible . In addition, approximately 20% of patients with tuberculosis are not identified by culture growth of M. tuberculosis .
In clinical practice, the initial decision to treat a patient for tuberculosis relies on methods that mostly yield nondefinitive results. These include sputum or bronchoalveolar lavage (BAL) fluid microscopy, M. tuberculosis-specific nucleic acid amplification techniques (NAATs) from biological specimens or immunodiagnostic procedures.
To date, the value of immunological methods for the diagnosis of tuberculosis has been limited, as the tuberculin skin test and IGRAs performed on peripheral blood alone cannot differentiate active tuberculosis from latent infection with M. tuberculosis . Identification of local antigen-specific lymphocytes at the site of the infection, e.g. amongst cells of the BAL fluid in pulmonary tuberculosis, improves the rapid diagnosis of tuberculosis in patients with negative acid-fast bacilli sputum smears and can influence greatly the decision of whether or not to initiate treatment for tuberculosis [7–10].
We have evaluated the relative impact of different methods for the rapid diagnosis of tuberculosis, including sputum microscopy for the presence of acid-fast bacilli, M. tuberculosis-specific NAATs, tuberculin skin testing, bronchoscopy and IGRA performed on cells from the peripheral blood as well as from BAL fluid in routine clinical practice at a clinical tuberculosis centre in Germany, a country with a low incidence of tuberculosis (5.5/100,000) .
Patients admitted between March 2005 and May 2007 for the diagnostic evaluation of presumed tuberculosis at the Medical Clinic of the Research Center Borstel, Germany, were included in the study. Patients with presumed tuberculosis were referred to the clinic if they had abnormalities on chest imaging and symptoms suggestive of active pulmonary tuberculosis. Tuberculosis was diagnosed or excluded according to a recently suggested algorithm : sputum samples were collected on three consecutive days and analysed for the presence of acid-fast bacilli by microscopy. If a positive microscopy result was obtained, a M. tuberculosis-specific NAAT (BDProbeTec ET assay, Becton Dickinson, Sparks, MD, USA) was performed subsequently for the rapid identification of M. tuberculosis.
All sputum samples were cultured for up to 6 weeks in fluid media (MGIT, BD, Heidelberg), and up to 8 weeks in solid media (Loewenstein-Jensen), before concluding that results were negative. Culture of M. tuberculosis was considered the gold standard method for the diagnosis of tuberculosis. A diagnosis of clinical tuberculosis was made after alternative diseases had been ruled out and after patients had received anti-tuberculosis therapy, which was prescribed by the treating physician on clinical grounds following no clinical response to antibiotic therapy.
Tuberculin skin testing was performed on all patients with the exception of those with a past medical history of tuberculosis, because of an increased risk of ulcerative skin lesions. Tuberculin skin testing was performed by intracutaneous injection of 0.1-mL tuberculin RT-23 (Statens Serum Institute, Copenhagen, Denmark). The induration was measured 48–72 h later by the ballpoint pen technique .
The M. tuberculosis-specific IGRA ELISpot (T-SPOT.TB Test, Oxford Immunotec, Abingdon, UK) was performed. Briefly, 250 000 peripheral blood mononuclear cells (PBMCs) were prepared by Ficoll-Hypaque density-gradient centrifugation from heparinized blood. PBMCs were plated overnight on 96-well plates precoated with a mouse anti-human interferon-γ antibody. The cells were (i) left unstimulated (negative control), (ii) stimulated with 10 ng ml−1 anti-CD3 monoclonal antibody (clone X35, Beckman-Coulter, Krefeld, Germany; positive control) or stimulated with 5 μg ml−1 of early secretory antigenic target (ESAT)-6 and culture filtrate protein (CFP)-10 peptides (kindly provided by Oxford Immunotec, Abingdon, UK). Culturing of the cells, washing, counterstaining, visualization and analysis of the spots were performed according to the manufacturer’s guidelines. The response of stimulated cultures was considered positive if the test well contained more than five spots and had twice the number of spots compared to the negative control well. The response was considered negative if it did not meet the criteria for a positive result and if the positive control well contained at least twice the number of spots compared to the negative control well. Results that met neither of these definitions were considered to be indeterminate and excluded from further analysis.
Patients with three negative microscopy results for the presence of acid-fast bacilli underwent bronchoscopy, which was performed according to German guidelines  and included a transbronchial biopsy for histology and BAL with 200–300 mL of normal saline from an affected lung segment for microscopy, M. tuberculosis-specific NAAT, M. tuberculosis culture and IGRA.
BAL fluid mononuclear cells (BALMCs) were obtained by passing the BAL through a stainless steel sieve (WMF, Geislingen, Germany) with a mesh aperture of 0.2 mm as described previously [7–9]. Elispot of BALMCs was otherwise performed and analysed in the same way as Elispot using PBMCs as described previously [7–9, 15]. Positive likelihood ratios were calculated by dividing sensitivity by (1-specificity) whilst negative likelihood ratios were calculated by dividing (1-sensitivity) by specificity. The study was approved by the ethical committee of the University of Lübeck, Germany.
In total, 135 patients who were referred for further diagnosis because of clinical suspicion of tuberculosis were included in the study (Fig. 1, Table 1).
Table 1. Demographic characteristics and Mycobacterium tuberculosis-specific immune responses in 135 individuals with suspected tuberculosis at a referral centre in Germany
Acid-fast bacilli were identified in the sputum smears of 18/135 (13.3%) individuals, 17 of whom were diagnosed with tuberculosis by positive M. tuberculosis-specific NAAT and/or culture. Nontuberculous mycobacteria (M. avium) were identified by culture in 1/18 (5.6%) individuals with a positive acid-fast bacilli sputum smear stain. There was no statistical difference amongst the prevalence of chronic obstructive pulmonary disease (COPD), smoking, alcohol abuse or diabetes mellitus in patients with tuberculosis with or without detectable acid-fast bacilli in sputum smears (data not shown).
Tuberculin skin test results
Tuberculin skin test results were available in 22 tuberculosis cases with negative acid-fast bacilli sputum smears and in 61 nontuberculosis patients with negative acid-fast bacilli sputum smears. Amongst these 83 patients, 43 (51.8%) had a tuberculin skin test reading of <5 mm, 10 (12.1%) had a reading between 5 and 10 mm and 30 (36.2%) had a reading ≥15 mm. Sensitivity and specificity were calculated for cut-off values of 5 and 15 mm (Table 2).
Table 2. Sensitivities, specificities, positive and negative likelihood ratios of different methods for the diagnosis of tuberculosis
*BAL was only performed in patients with negative acid-fast bacilly sputum smear results.
Sputum microscopy (n = 135)
BAL microscopy* (n = 101)
TST >5 mm (n = 83)
TST >15 mm (n = 83)
NAAT (BAL)* (n = 71)
NAAT (Sputum) (n = 44)
IGRA blood (n = 115)
IGRA BAL* (n = 111)
Blood IGRA results
Amongst 117 patients with a negative acid-fast bacilli sputum smear result [117/135 (86.7%)], 70 (59.8%) had a positive blood IGRA result and 45 (38.5%) had a negative result. One patient had an indeterminate IGRA result and for one nontuberculosis patient, the result was not available. Of 24 tuberculosis patients, 22 (91.7%) had a positive blood IGRA result and 76/87 (87.4%) nontuberculosis patients had a negative result. The sensitivity and specificity of the blood IGRA were 96.0% and 48.9%, respectively.
All 117 patients with suspected tuberculosis and with negative sputum smears underwent bronchoscopy with BAL.
Microscopy for the presence of acid-fast bacilli in BAL fluid was positive in 1/117 (0.9%) and negative in the remaining 116 (99.2%) subjects.
BAL IGRA results
ELISpot in BALMCs was positive in 33/117 (28.2%) cases and negative in 78/117 (66.7%) cases; indeterminate results were obtained in six (5.1%) patients. Twenty-two of 24 (91.7%) tuberculosis patients had a positive BAL IGRA result, and 76/87 (87.4%) nontuberculosis patients had a negative BAL IGRA result. The sensitivity and specificity of the test were 91.7% and 87.4%, respectively.
In patients with tuberculosis, median numbers of ESAT-6-/CFP-10-specific spot-forming cells (SFCs)/250 000 PBMCs and BALMCs were 11/20 and 34/45, respectively (Table 1). By contrast, in patients from the nontuberculosis group, the median number of ESAT-6-/CFP-10-SFC/250 000 PBMCs and BALMCs were 2/5 and 1/1, respectively. Thus, in contrast to patients with latent infection with M. tuberculosis, ESAT-6- and CFP-10-specific SFCs were concentrated in the BAL fluid compared to the peripheral blood in patients with active tuberculosis (Table 1). When the production of interferon-γ was attributed only to lymphocytes amongst mononuclear cells in blood and BAL, the concentration of ESAT-6- and CFP-10-specific lymphocytes at the site of infection was of the order of magnitude of 9.1 times and 11.4 times, respectively (data not shown).
The M. tuberculosis-specific NAAT using BAL specimens was positive in 6/71 cases (8.5%) and negative in the remaining 65 (91.6%). Of the six patients with a positive BAL NAAT result, five had a positive M. tuberculosis culture. One patient had pulmonary infiltration of unknown origin and developed bronchogenic carcinoma 2 years later. Five of 16 tuberculosis patients had a positive NAAT result using BAL specimens and 54/55 nontuberculosis patients had a negative result. The sensitivity and specificity of M. tuberculosis-specific NAAT using BAL specimens were 31.3% and 98.2%, respectively.
In addition, in 44 of the 117 patients with negative acid-fast bacilli sputum smears, NAAT was performed on sputum samples. Seven of 44 (15.9%) NAAT examinations were positive and the remaining 37 (84.1%) were negative. Amongst the seven patients with a positive sputum NAAT result, four were finally diagnosed with active tuberculosis whereas three had an alternative diagnosis.
Overall, 42/135 (31.1%) patients had a final diagnosis of tuberculosis. M. tuberculosis was cultured from the sputum, BAL or, if available, from a biopsy in 33/135 (24.4%) cases [in 17/17 (100%) patients with and in 16/25 (64.0%) without the detection of acid-fast bacilli on sputum smear stains]. Thus, in 33/42 (78.6%) patients, the diagnosis of tuberculosis was confirmed by M. tuberculosis culture. Of the 9/42 (21.4%) patients with a clinical diagnosis of tuberculosis, one had a positive NAAT result from sputum. M. tuberculosis cultures were positive from sputum in 28/135 (20.7%) patients, from BAL in 14/117 (11.9%) patients, and from histological specimens in 6/31 (19.4%) patients. In addition, growth of both M. tuberculosis and M. avium was observed in one patient with a negative acid-fast bacilli sputum smear culture. In 10/135 (7.4%) patients, nontuberculous mycobacteria could be cultured from bronchopulmonary specimens (six M. avium, two M. fortuitum, one M. gordonae and one M. tilburgii). Only 1/10 (10.0%) patients with a positive culture for nontuberculous mycobacteria had a positive acid-fast bacilli sputum smear result.
Comparative evaluation of different tests
The positive likelihood ratio was calculated and compared for all methods (Table 2). Sputum microscopy has a high specificity (98.9%) and therefore a high positive likelihood ratio (LR+ = 37.6) whereas the tuberculin skin test has a somewhat lower specificity (63–82%), depending on the cut-off (>5 mm or >15 mm), and a low positive likelihood ratio (LR+ = 2–4). It is interesting that if performed using BAL fluid, the ELISpot IGRA had a higher positive likelihood ratio than if performed on blood: LR+ = 7 for BAL ELISpot IGRA versus LR+ = 2 for blood ELISpot IGRA. To estimate the diagnostic value of a method, sensitivity also has to be taken into consideration. In Table 2, sensitivity and specificity/likelihood ratio of the relevant methods for tuberculosis diagnosis are shown. Microscopy showed the highest positive likelihood ratio but a low sensitivity (41%). Blood ELISpot IGRA had the highest sensitivity (96%) but because of a low specificity (49%), its positive likelihood ratio was only about 2.
In this study, the clinical impact of different diagnostic methods for a rapid diagnosis of tuberculosis, including IGRA (ELISpot) performed on cells from the BAL fluid, have been analysed under routine clinical conditions. The key finding of the study was that none of the evaluated methods alone was able to reliably diagnose or exclude tuberculosis. Microbiological methods (sputum/BAL fluid smear microscopy, M. tuberculosis-specific NAAT with sputum/BAL) had a superior positive likelihood ratio for the diagnosis of tuberculosis compared to the immunological methods (tuberculin skin test and IGRA). However, the low sensitivity of the microbiological methods did not allow the exclusion of tuberculosis in cases with a negative test result. In this study, 25/42 (59.5%) patients with active tuberculosis would have been missed by sputum microscopy alone. IGRA (ELISpot) performed on cells from the BAL fluid, a procedure that is not yet widely available, was the best method to distinguish active tuberculosis from latent infection with M. tuberculosis, whereas the impact of IGRA performed on cells from the peripheral blood was not greater than that of the tuberculin skin test. In fact, BAL IGRA (ELISpot) had a very high negative likelihood ratio and, according to the results from this study, to a great extend allows the exclusion of active tuberculosis. This will help to improve cost efficiency as unnecessary patient isolation can be avoided.
The clinical diagnosis of tuberculosis may be difficult , especially in countries of low tuberculosis incidence, as most individuals who are suspected of having tuberculosis receive a different final diagnosis after clinical evaluation . In this study, less than 25% of patients referred to a clinical centre for tuberculosis in Germany had a final diagnosis of pulmonary tuberculosis confirmed by a positive culture of M. tuberculosis from bronchopulmonary specimens and only 13% of suspected tuberculosis cases had positive acid-fast bacilli sputum smears. By contrast, the proportion of suspected cases with positive acid-fast bacilli sputum smears is generally higher in high-incidence countries, ranging from 17.3% and 18.6% in Malawi and Senegal to 20.2% and 32.1% in Uganda and Benin [17, 18], respectively.
Consistent with previous findings  and the report from the WHO , less than 50% of patients with active tuberculosis had acid-fast bacilli in sputum smears. Detection of acid-fast bacilli might increase by sputum induction, although the effect of sputum induction may be greatest on the rates of sputum culture conversion .
In those with suspected tuberculosis and negative acid-fast bacilli sputum smears, bronchoscopy is usually indicated . However, the identification of additional cases of tuberculosis by BAL fluid acid-fast bacilli stain analysis is limited. In a recent study from Switzerland , 4/91 (4.4%) tuberculosis patients were identified by BAL fluid stain in addition to the sputum smear examination, compared to only 1/101 (1.0%) patients in the present study.
Of importance, nontuberculous mycobacterial infection must be considered in patients with positive acid-fast bacilli staining. Because of the high sensitivity of acid-fast bacilli-positive specimens, M. tuberculosis-specific NAAT has been used for the rapid discrimination of M. tuberculosis from nontuberculous mycobacteria [22, 23]. However, the sensitivity for tuberculosis of NAAT in sputum or BAL fluid in acid-fast bacilli smear-negative cases is limited . A decision to treat for active tuberculosis based on the result of the NAAT in sputum or BAL fluid alone would have missed 56–69% of patients with a culture-confirmed diagnosis in this study. Of note, the specificity of M. tuberculosis-specific NAAT was only 91% in this study and thus lower than previously reported . Two of three individuals with a positive M. tuberculosis-specific NAAT result, who did not have active tuberculosis, had a history of tuberculosis, possibly explaining this finding.
M. tuberculosis-specific IGRA in cells from the peripheral blood had a sensitivity of 96% and a specificity of 49% for the diagnosis of active tuberculosis. The specificity of the IGRA for the diagnosis of active tuberculosis in this study is considerably lower than reported in a meta-analysis by Diel et al. . This discrepancy is most probably related to the different groups that were evaluated to determine the test specificity. In the present study, those suspected of tuberculosis who eventually received a different final diagnosis were included in the evaluation for specificity. This group included migrants from countries of high tuberculosis incidence as well as patients with a past medical history of tuberculosis. These patients were likely to test positive with the blood IGRA. By contrast, individuals with a low risk of M. tuberculosis infection are commonly included within control groups to evaluate the test specificity of IGRAs . This may lead to an incorrect diagnosis of active tuberculosis when based on IGRAs performed on cells from the peripheral blood. However, in active tuberculosis, antigen-specific cells clonally expand and are concentrated at the site of infection [7, 26–29].
In this study, patients with active tuberculosis had a roughly 10-fold concentration of antigen-specific lymphocytes in the BAL fluid versus the peripheral blood when compared to individuals with latent M. tuberculosis infection.
When compared to other immunodiagnostic methods, M. tuberculosis-specific IGRA (ELISpot) in cells from the BAL fluid was the best method for rapid diagnosis of active tuberculosis with a sensitivity of 92% and a specificity of 87%. These findings are supported by the latest meta-analysis of the role of IGRAs in the diagnosis of active tuberculosis .
The limitations of the study need to be addressed. This study was performed in a single centre in a country with a low incidence of tuberculosis and a low prevalence of HIV infection. Results from this study cannot be generalized to other settings, e.g. to developing countries with an intermediate or high incidence of tuberculosis and/or with a higher prevalence of HIV infection. Sensitivity results of sputum smear microscopy and sputum NAAT could have been expected to be higher when using sputum induction methods [31–33]. Culture confirmation of tuberculosis was not possible in 9/42 patients and, although these patients responded well to antituberculosis therapy, they could have been misclassified. Finally, not all individuals with negative acid-fast bacilli sputum smear results had M. tuberculosis-specific NAAT performed on sputum as it is not routinely recommended in Germany . Although M. tuberculosis-specific ELISpot IGRA in cells from the BAL fluid has been evaluated in countries of high  and low [7–9, 15, 35] tuberculosis incidence, the procedure is not widely available yet and needs to be evaluated by the regulatory authorities before it can be recommended for general diagnostic application.
In conclusion, identification of M. tuberculosis from culture remains the gold standard for the diagnosis of pulmonary tuberculosis and is currently the only reliable method for antibiotic resistance testing in the majority of cases with negative acid-fast bacilli sputum smears. None of the microbiological or immunological methods that aim to provide a rapid diagnosis of tuberculosis whilst awaiting the confirmation of the culture result is on its own accurate enough to diagnose or exclude tuberculosis. Local immunodiagnosis by M. tuberculosis-specific IGRA (ELISpot) in cells from the BAL fluid is possibly the best available immunological method to distinguish active tuberculosis from latent M. tuberculosis infection; however, further studies will be needed to recommend this method for routine clinical practice. On average, more than one in every two patients evaluated in this study has benefited from M. tuberculosis-specific ELISpot IGRA in cells from the BAL fluid for the rapid diagnosis of acid-fast bacilli smear-negative pulmonary tuberculosis. Bronchoscopy should be indicated for all cases of suspected active tuberculosis with negative acid-fast bacilli sputum smears, as the majority will have a different final diagnosis. Alternative diagnoses to tuberculosis, such as sarcoidosis, bronchogenic carcinoma or cryptogenic organizing pneumonia, need to be considered and identified by bronchoscopy. A stepwise diagnostic approach, as suggested by the algorithm described in this study, might yield the best results for a rapid preliminary diagnosis of tuberculosis, justifying treatment initiation whilst the results of M. tuberculosis cultures are pending.