(Chest. 2005;128:102-107.)
© 2005
American College of Chest Physicians
Effect of Nucleic Acid Amplification for Mycobacterium tuberculosis on Clinical Decision Making in Suspected Extrapulmonary Tuberculosis*
Renda Soylemez Wiener, MD;
Phyllis Della-Latta, PhD and
Neil W. Schluger, MD, FCCP
* From the Departments of Medicine (Drs. Soylemez Wiener and Schluger) and Pathology (Dr. Della-Latta), Columbia University College of Physicians and Surgeons, New York, NY.
Correspondence to: Neil W. Schluger, MD, FCCP, Division of Pulmonary, Allergy, and Critical Care Medicine, Columbia Presbyterian Medical Center, 630 West 168th St, New York, NY 10032; e-mail: ns311{at}columbia.edu
 |
Abstract
|
|---|
Background: Laboratory-based studies have suggested the nucleic acid amplification test (NAAT) for Mycobacterium tuberculosis may be useful in diagnosing extrapulmonary tuberculosis. We sought to determine how clinicians in one hospital used results of the NAAT in clinical decision making in cases of suspected extrapulmonary tuberculosis.
Methods: We performed a retrospective analysis of all patients who underwent the NAAT on at least one nonsputum sample, excluding cerebrospinal fluid, from 1999 to 2001 in one large urban hospital. For these patients, we reviewed the hospital course, with particular attention to date of the NAAT and its influence on days treated with antituberculous medications and days to final diagnosis.
Results: Thirty-five patients with suspected tuberculosis who had undergone the NAAT on extrapulmonary specimens were identified. From three patients, NAAT results were nondiagnostic because of inhibitors, and they were excluded from the analysis, leaving 32 patients. Tuberculosis was ultimately diagnosed in 14 of these 32 patients. NAAT findings were positive in specimens from 12 of 14 patients with extrapulmonary tuberculosis and in 0 of 18 cases in which tuberculosis was excluded (sensitivity, 86%; specificity, 100%; positive predictive value, 100%; negative predictive value, 90%). In only 2 of 19 patients treated with antituberculous medications was the NAAT result used to determine the onset or discontinuation of therapy. In no instance was a negative NAAT result used by clinicians as definitive evidence that a patient did not have extrapulmonary tuberculosis; in all but one case, patients were continued on antituberculous therapy until final culture results were available.
Conclusions: The NAAT proved to be a sensitive and specific test for detection of M tuberculosis in extrapulmonary specimens but did not weigh heavily in clinical decision making at our hospital. Judicious use of these tests may improve the accuracy and speed of diagnosis of extrapulmonary tuberculosis, while helping to eliminate unnecessary antituberculous treatment in patients without tuberculosis.
Key Words: decision making • extrapulmonary tuberculosis • Mycobacterium tuberculosis • nucleic acid amplification testing
|
Introduction
|
|---|
Extrapulmonary tuberculosis remains a diagnosis that is often difficult to establish immediately and conclusively.1 In many cases, it is not until culture results are available, up to 8 weeks after clinical presentation, that a diagnosis of extrapulmonary tuberculosis is definitely established or excluded. In addition, obtaining material for culture in extrapulmonary cases often requires invasive procedures. Finally, cases of extrapulmonary tuberculosis are more often culture negative than are pulmonary tuberculosis cases. At times, in order to avoid an invasive procedure, patients may be treated presumptively for extrapulmonary tuberculosis; if they appear initially to respond, efforts to confirm tuberculosis or exclude other diagnoses may be inappropriately deferred. As a result of these diagnostic challenges, the institution of appropriate antituberculous therapy is often delayed in patients who do in fact have tuberculosis, while patients without tuberculosis may be treated unnecessarily.2 Because of the morbidity and mortality associated with delaying treatment for true tuberculosis and the significant added expense and potential toxicity of antituberculous therapy in patients without tuberculosis, a diagnostic tool to expedite identification of patients with true extrapulmonary tuberculosis could both improve patient care and decrease costs of unnecessary hospitalization and medication.
The nucleic acid amplification test (NAAT) is rapid to perform and produces results within hours. The NAAT detects Mycobacterium tuberculosis with extreme accuracy in respiratory specimens that stain positive for acid-fast bacilli (AFB), and can also detect organisms in a significant number of smear-negative specimens.1 The test has gained acceptance in a number of clinical settings for the diagnosis of pulmonary tuberculosis, and guidelines are available for its use.3 In the setting of extrapulmonary disease, the clinical utility of nucleic acid amplification assays is less clear. Multiple laboratory-based studies of the NAAT for M tuberculosis suggest the assay is both sensitive (73 to 100%) and specific (93 to 100%) in a wide array of extrapulmonary specimens.4567891011 There is variability in results obtained in different studies and using different approaches to nucleic acid amplification. For example, Piersimoni and colleagues9 used strand displacement amplification and recombinant RNA amplification and found that the former approach was 74% sensitive and 100% specific for diagnosing extrapulmonary tuberculosis and the latter was 92.3% sensitive and 100% specific. Using the ligase chain reaction assay, Rantakokko-Jalava et al11 found a sensitivity of 73.3% and a specificity of 98%, using culture as a "gold standard." Despite this evidence of accuracy in diagnosis, the NAAT has not gained widespread acceptance in the clinical diagnosis and management of cases of suspected extrapulmonary tuberculosis, and no guidelines are available to offer indications for its use in cases of suspected extrapulmonary tuberculosis. Moreover, the NAAT has not been officially approved by the US Food and Drug Administration for use in nonrespiratory specimens. Perhaps one reason the test is not used more widely for the clinical diagnosis of extrapulmonary tuberculosis is that most of the studies of the NAAT for extrapulmonary tuberculosis have been performed from the perspective of the laboratory and have provided little insight into how results are integrated into clinical practice. To address the actual clinical utility of the NAAT in the diagnosis of extrapulmonary tuberculosis, we sought to determine how the NAAT affected clinical decision making in cases of suspected extrapulmonary tuberculosis in our hospital.
|
Materials and Methods
|
|---|
Study Design
We performed a retrospective analysis of all cases of suspected extrapulmonary tuberculosis in which an NAAT was performed on one or more nonsputum specimens between 1998 to 2001 at Columbia Presbyterian Medical Center of the New York-Presbyterian Hospital. For the purposes of this study, we defined extrapulmonary tuberculosis to encompass both tuberculosis isolated outside of the lung (eg, scrofula, pleural tuberculosis) as well as tuberculosis with both pulmonary as well as extrapulmonary involvement. Growth of M tuberculosis in culture and/or clinical presentation strongly suggestive of extrapulmonary tuberculosis with documented response to antituberculous therapy was used as the "gold standard" to identify cases of extrapulmonary tuberculosis. Patients were eligible for inclusion in the analysis if the NAAT was performed on at least one nonsputum specimen. We excluded from the analysis specimens for which the NAAT provided a nondiagnostic result, due to presence of inhibitors or other technical factors that did not allow the laboratory to clearly mark a test result as positive or negative. Eligible extrapulmonary specimens included ascitic fluid, pericardial fluid, pleural fluid, and tissue specimens; however, cerebrospinal fluid specimens were excluded from this study. Once patients were identified, data were abstracted from the medical charts, with particular attention to clinical presentation, date of NAAT and AFB stain, use of and timing of antituberculous therapy, culture results, and final diagnosis on hospital discharge or death.
Laboratory Methods
Specimens were digested and decontaminated using NALC/NaOH within 3 days of collection and stained for AFB using auramine 0 fluorescent stain. They were inoculated onto Lowenstein-Jensen, Middlebrook 7H11 selective biplate, chocolate agar, and BBL MGIT broth (Becton Dickinson; Sparks, MD) and incubated at 35°C in CO2 for up to 8 weeks. The NAAT used was the Amplified MTD (AMTD) [GenProbe; San Diego, CA]. Following the instructions of the manufacturer, the AMTD test on each specimen included a duplicate control that was seeded with M tuberculosis cells to detect nucleic acid amplification inhibition. It is laboratory protocol at our hospital to run the AMTD on all specimens that stain positive for AFB. In all other cases, however, the AMTD is restricted by the laboratory, requiring consultation with the Director of the Clinical Microbiology section as well as with an infectious disease or pulmonary attending physician to perform the test based on clinical suspicion of tuberculosis (Fig 1
). Our analysis included subjects from both of these groups. Once a case has been approved for the NAAT, the laboratory runs the AMTD on all specimens received for that patient.
Statistical Methods
Unless otherwise indicated, sensitivity, specificity, and positive and negative predictive values of the AMTD were calculated on a per-patient basis, rather than on a per-specimen basis. The two-tailed Fisher Exact Test was used to compare results between groups. Bayesian calculations were used to determine confidence intervals (CIs) around positive and negative predictive values based on prior probability.
|
Results
|
|---|
Patient Profile
A total of 35 eligible patients were identified for inclusion in the study: 16 patients with and 19 patients without a final diagnosis of extrapulmonary tuberculosis. Three patients were subsequently excluded from analysis because AMTD results on sputum samples were nondiagnostic due to the presence of inhibitors. The final analysis group of 32 patients consisted of 14 patients with tuberculosis and 18 without extrapulmonary tuberculosis. Characteristics of the two groups are shown in Table 1
. Of the 14 patients with tuberculosis, 5 patients (36%) had pulmonary as well as extrapulmonary involvement. Of the 18 patients without tuberculosis, 8 patients (44%) were HIV positive and presented with either lymphadenopathy (6 of 8 patients) or abscesses (2 of 8 patients) that subsequently tested AFB stain positive. Ultimately, it was believed that all eight of these patients had Mycobacterium avium complex (MAC) infection rather than tuberculosis, although two of these eight patients never grew MAC on culture.
Specimens Tested
From these 32 subjects, the AMTD was performed on 55 extrapulmonary specimens. Fluid specimens comprised 36% of the specimens tested, including ascites, pleural fluid, and pericardial fluid. The other 64% of specimens tested were tissue specimens, and the majority (61%) of these were lymph nodes.
Results of AMTD Testing
AMTD results are shown in Table 2
, stratified by final diagnosis of extrapulmonary tuberculosis. AMTD results were positive in specimens from 12 of 14 patients with extrapulmonary tuberculosis and in none of 18 cases in which tuberculosis was excluded. Overall, the AMTD for this patient sample had a sensitivity of 86%, a specificity of 100%, a positive predictive value of 100%, and a negative predictive value of 90% (95% CI, 77 to 103%). By contrast, AFB staining in the same patient population had a sensitivity of 50%, a specificity of 56%, a positive predictive value of 47% (95% CI, 21 to 72%), and a negative predictive value of 59% (95% CI, 35 to 82%) [Table 3
]. The AMTD was significantly more specific than AFB stain (100% vs 56%) in cases of suspected extrapulmonary tuberculosis (two-tailed Fisher Exact Test, p = 0.003). There was a strong trend toward increased sensitivity of the AMTD compared to AFB stain as well (86% vs 56%, p = 0.04 in a one-tailed test, p = 0.10 in a two-tailed test).
As indicated above, there were two patients who had AMTD results that were discordant with the final diagnosis; both results were finally classified as false negative. The characteristics of these patients are shown in more detail in Table 4
.
We found that in our laboratory, the AMTD was more sensitive for tissue specimens (14 of 15 specimens from true tuberculosis patients; AMTD positive, 93%) than for fluid specimens (3 of 7 specimens from true tuberculosis patients; AMTD positive, 43%; two-tailed Fisher Exact Test, p = 0.02). Of note, in one of the two patients with AMTD false-negative results (Table 4), a high-quality surgical specimen, as indicated by the multiple granulomata seen on pathologic examination, was not sent for AMTD at all, leaving only the inferior quality fluid specimen on which to perform the AMTD.
Of 14 patients with a final diagnosis of tuberculosis, 7 patients (50%) had specimens that were AFB stain negative. The AMTD proved relatively sensitive (86%) even in these patients with AFB stain-negative specimens (six of seven AMTD positive vs zero of seven AFB positive; two-tailed Fisher Exact Test, p = 0.005). The AMTD was specific for M tuberculosis; there were no false-positive AMTD results in the six patients (nine specimens) that grew MAC but not M tuberculosis in culture. There were two patients who grew both M tuberculosis and MAC in culture; these patients both had appropriately positive AMTD results.
Effect of the AMTD on Initiation and Cessation of Antituberculous Therapy
Among 14 patients with true tuberculosis, all 14 received antituberculous therapy. Only once was a positive AMTD result the deciding factor in beginning treatment. By contrast, six of these patients (43%) were begun on antituberculous therapy when the patient was noted to be AFB stain positive, and three patients (21%) were begun on treatment based on a suggestive pathology result with granulomata or AFB positive stain. Among the 18 patients in whom tuberculosis was excluded, 5 patients received antituberculous therapy; in all five cases, treatment was initiated based on clinical suspicion alone (n = 2), positive AFB stain (n = 2), or suggestive pathology (n = 1); in each of these five cases, therapy was begun before the AMTD result was even available. In these same five patients, a negative AMTD result was used only once as the primary reason to discontinue antituberculous therapy.
|
Discussion
|
|---|
In our hospital, the AMTD for M tuberculosis in cases of suspected extrapulmonary tuberculosis is being performed with increasing frequency. However, clinicians in our hospital have not routinely used the AMTD result as a key factor in deciding to initiate or discontinue antituberculous therapy. While we found the AMTD compared to AFB stain to be significantly more specific (100% vs 56%, two-tailed Fisher Exact Test, p = 0.003) and to show a trend toward increased sensitivity (86% vs 50%, two-tailed Fisher Exact Test, p = 0.10; if there had been one additional patient with AMTD-positive, AFB-negative specimens, the two-tailed Fisher Exact Test p value would change to p = 0.05) for detecting the presence of M tuberculosis in nonsputum specimens from patients with suspected extrapulmonary tuberculosis, management decisions were far more often based on the AFB stain result than on the AMTD result. This frequent use of AFB stain as a major factor in clinical decision making likely stems from clinicians’ familiarity and comfort level with the AFB stain, which has been in use for many decades, as opposed to the relatively new technology of the NAAT. An obvious flaw in relying primarily on the AFB stain result to decide whether to implement antituberculous therapy occurs when a tuberculosis patient produces AFB-negative specimens, a situation that is common in many paucibacillary extrapulmonary specimens. In our sample, 7 of 14 true tuberculosis patients were AFB stain negative (50%), but 6 of these 7 AFB stain-negative patients (86%) were AMTD positive (two-tailed Fisher Exact Test, p = 0.005). In only one of these six AFB stain-negative patients with tuberculosis was the positive AMTD result used as the deciding factor to initiate antituberculous therapy; the other five patients were started on therapy based on clinical suspicion or pathology results. Using a positive AFB stain as the deciding factor in initiating antituberculous therapy can lead to unnecessary treatment, as the AFB stain obviously does not distinguish between typical and atypical mycobacteria. In our study, this occurred in two of the five patients without tuberculosis who received antituberculous therapy; both of these two AFB stain-positive patients ultimately received a diagnosis of MAC rather than tuberculosis. Both of these patients had AMTD testing performed shortly after the positive AFB stain as per laboratory protocol (Fig 1), and had no evidence of pulmonary involvement to trigger public health concerns, yet antituberculous therapy was continued well after the negative AMTD results were available.
However, in this sample, AMTD did fail to identify 2 of 14 true tuberculosis cases (14%). Because of the public health risk of missing a case of pulmonary tuberculosis and the personal risk to the patient of leaving tuberculosis untreated, it is reasonable for clinicians to institute or continue antituberculous treatment despite a negative AMTD result in cases of strong clinical suspicion for extrapulmonary tuberculosis, particularly if there is pulmonary involvement as well. One way in which the sensitivity of NAAT could be improved on a per-patient basis is to test multiple specimens per patient, particularly high-quality tissue specimens (as by surgical biopsy) rather than fluid specimens (93% vs 43%, two-tailed Fisher Exact Test, p = 0.02). In one of the two patients who had tuberculosis but had negative AMTD test results, a surgical tissue specimen was available but was sent only to pathology, not to the microbiology laboratory for AMTD testing, leaving only a single fluid specimen for the AMTD test. It is important that the laboratory have all specimens from a patient with suspected tuberculosis available for testing, especially high-quality surgical specimens, in order to increase the sensitivity of the NAAT. As clinician familiarity with the role of the NAAT for M tuberculosis grows, these types of oversights may become less common.
Because the positive predictive value of the AMTD was high (100%), a positive AMTD result for M tuberculosis should encourage clinicians to consider strongly the diagnosis of extrapulmonary tuberculosis and subsequent antituberculous therapy. Of course, a high positive predictive value depends not only on the characteristics inherent to the test being performed (sensitivity and specificity), but also on the prevalence of the disease in the population studied. We believe that the use of specialists to screen patients before the NAAT is done (Fig 1) serves to increase the prior probability of tuberculosis in the patients to be tested and therefore raises the positive predictive value. This has been demonstrated to be the case in previous studies.12
Extrapulmonary tuberculosis remains a challenging diagnosis for clinicians to make definitively. Because the NAAT for M tuberculosis is increasingly available, more rapid than culture growth, and more sensitive and specific than AFB stain, it is a useful test to expedite identification of cases of extrapulmonary tuberculosis. Though our sample size was small with limited power and our analysis restricted by the retrospective design, we believe our study provides some insight into how the NAAT for M tuberculosis has been used in clinical decision making in cases of suspected extrapulmonary tuberculosis and how it may be applied in the future. Our experience indicates that, when positive, the AMTD result should be used as strong evidence of likely tuberculosis and should influence clinical management accordingly. Improved sensitivity, as may be achieved through testing multiple specimens per patient including all surgical tissue specimens, is needed before clinicians can use a negative AMTD result to exclude definitively extrapulmonary tuberculosis.
|
Footnotes
|
|---|
Abbreviations: AFB = acid-fast bacilli; AMTD = Amplified MTD; CI = confidence interval; MAC = Mycobacterium avium complex; NAAT = nucleic acid amplification testing
Received for publication September 13, 2004.
Accepted for publication November 22, 2004.
|
References
|
|---|
- Schluger, NW (2001) Changing approaches to the diagnosis of tuberculosis. Am J Respir Crit Care Med 164,2020-2024[Free Full Text]
- Moudgil, H, Leitch, AG Extra-pulmonary tuberculosis in Lothian 1980–1989: ethnic status and delay from onset of symptoms to diagnosis. Respir Med 1994;88,507-510[CrossRef][ISI][Medline]
- Update: nucleic acid amplification tests for tuberculosis. MMWR Morb Mortal Wkly Rep 2000;49,593-594[Medline]
- Nagesh, BS, Sehgal, S, Jindal, SK, et al Evaluation of polymerase chain reaction for detection of Mycobacterium tuberculosis in pleural fluid. Chest 2001;119,1737-1741[Abstract/Free Full Text]
- Villegas, MV, Labrada, LA, Saravia, NG Evaluation of polymerase chain reaction, adenosine deaminase, and interferon-
in pleural fluid for the differential diagnosis of pleural tuberculosis. Chest 2000;118,1355-1364[Abstract/Free Full Text]
- Portillo-Gomez, L, Morris, SL, Panduro, A Rapid and efficient detection of extra-pulmonary Mycobacterium tuberculosis by PCR analysis. Int J Tuberc Lung Dis 2000;4,361-370[Medline]
- Shah, S, Miller, A, Mastellone, A, et al Rapid diagnosis of tuberculosis in various biopsy and body fluid specimens by the AMPLICOR Mycobacterium tuberculosis polymerase chain reaction test. Chest 1998;113,1190-1194[Abstract/Free Full Text]
- Della-Latta, P, Whittier, S Comprehensive evaluation of performance, laboratory application, and clinical usefulness of two direct amplification technologies for the detection of Mycobacterium tuberculosis complex. Am J Clin Pathol 1998;110,301-310[ISI][Medline]
- Piersimoni, C, Scarparo, C, Piccoli, P, et al Performance assessment of two commercial amplification assays for direct detection of Mycobacterium tuberculosis complex from respiratory and extrapulmonary specimens. J Clin Microbiol 2002;40,4138-4142[Abstract/Free Full Text]
- Alcala, L, Ruiz-Serrano, MJ, Hernangomez, S, et al Evaluation of the upgraded amplified Mycobacterium tuberculosis direct test (Gen-probe) for direct detection of Mycobacterium tuberculosis in respiratory and non-respiratory specimens. Diagn Microbiol Infect Dis 2001;41,51-56[CrossRef][Medline]
- Rantakokko-Jalava, K, Marjamaki, M, Marttila, H, et al LCx Mycobacterium tuberculosis assay is valuable with respiratory specimens, but provides little help in the diagnosis of extrapulmonary tuberculosis. Ann Med 2001;33,55-62[ISI][Medline]
- Divinagracia, RM, Harkin, TJ, Bonk, S, et al Screening by specialists to reduce unnecessary test ordering in patients evaluated for tuberculosis. Chest 1998;114,681-684[Abstract/Free Full Text]
TOP
Abstract
Introduction
