The accuracy and reliability of nucleic acid amplification tests in the diagnosis of tuberculosis.

Accurate and early diagnosis of tuberculosis is a critical part of the management and control of the disease. Diagnostic work up for tuberculosis involves the detection of acid-fast bacilli in clinical samples by microscopy (smear) and culture. These conventional tests are not always helpful in making the diagnosis. Microscopy, although rapid and inexpensive, has only modest sensitivity and specificity. Mycobacterial cultures, although very specific, might be negative in 10%–20% of cases, and the results are often not available for weeks. In the context of these limitations, nucleic acid amplification (NAA) tests have emerged with the intended goal of enabling clinicians to make a rapid and accurate diagnosis. Polymerase chain reaction (PCR) is the best known and most widely used NAA test. All NAA tests amplify target nucleic acid regions (DNA or RNA) that uniquely identify the Mycobacterium tuberculosis (M. tuberculosis ) complex. Because NAA tests can be used directly on clinical specimens (such as sputum), they are also called ‘direct amplification tests’ (DAT). In theory, PCR tests are exquisitely sensitive—they can amplify even a single copy of the target genomic sequence. Their specificity is also expected to be high because they amplify genomic targets that are highly specific to the M. tuberculosis complex. NAA tests are rapid—results are usually obtained within 6–12 hours. Because of these potential advantages, the introduction of NAA tests was hailed as a major breakthrough in the diagnosis of tuberculosis. Have these tests lived up to their reputation? How accurate and reliable is their clinical performance? It is important to examine the evidence from recently published meta-analyses and systematic reviews on the validity and role of NAA tests for the diagnosis of tuberculosis. NAA tests are categorized as commercial or in-house (‘home-brew’). Commercial kits include the Amplicor® MTB tests (Roche Molecular Systems), the Amplified Mycobacterium tuberculosis Direct Test® (MTD) (Gen-Probe Inc), the LCx® kit (Abbott Laboratories), and the BD ProbeTec ET assay (BD Diagnostic Systems). The costs of commercial NAA tests vary (list price US$ 25–50 per test). In-house tests are laboratory-developed PCR assays where the investigators put together their own PCR protocols. In-house assays, therefore, vary greatly in their design and laboratory methods. The cost of an in-house PCR is about US$ 15–20 per test. These cost estimates do not include the cost of buying and maintaining PCR equipment. All NAA tests require a specialized and sophisticated laboratory infrastructure, and skilled personnel for optimal performance. The accuracy and reliability of NAA tests for tuberculosis have been extensively studied since the early 1990s. Accuracy refers to test performance characteristics such as sensitivity and specificity. Reliability refers to repeatability (variability when the test is repeated). As hundreds of studies have evaluated NAA tests, it is now possible to determine their overall performance using meta-analyses and systematic reviews. Table I presents the results of some recent meta-analyses and reviews on the accuracy of NAA tests. Because these meta-analyses and reviews synthesize data