Relevance of Commercial Amplification Methods for Direct Detection of Mycobacterium tuberculosis Complex in Clinical Samples

Mycobacteria are a group of acid-fast, aerobic, slow-growing organisms whose genus includes more than 90 different species. The causative agents of tuberculosis (TB), which is currently considered a global emergency, with more than 2 million people dying every year and 8 million new cases, belong to the Mycobacterium tuberculosis complex (MTB). Moreover, although of lesser public health importance, many other species referred to as nontuberculous mycobacteria (NTM) have also been associated with human disease with increasing frequency worldwide (65). As MTB is highly infectious for humans, it is of paramount importance that TB be diagnosed as early as possible to stop the spread of the disease. Active TB is currently diagnosed by conventional laboratory procedures including specimen digestion and decontamination, microscopic examination for the presence of acid-fast bacilli (AFB), isolation by culture on solid and/or liquid media, and identification and drug susceptibility testing of the recovered isolate. Because of the slow growth of mycobacteria, the above-reported laboratory procedures may require turnaround times of 3 to 4 weeks or longer.

During the last decade, several molecular methods have been developed for direct detection and identification of MTB in clinical specimens. These methods, being able to potentially reduce the diagnostic time from weeks to days, have been acquiring greater and greater relevance in the field of laboratory TB diagnosis. The basic principle of any molecular diagnostic test is the detection of a specific nucleic acid sequence by hybridization to a complementary sequence, a probe, followed by detection of the hybrid. However, the sensitivity of nucleic acid probe tests that do not involve amplification is much lower than that of amplified ones. Any portion of nucleic acid can be copied by using the specific polymerase, provided that some sequence data are known for the setup of appropriate primers. In general, amplification of target nucleic acid sequences is composed of three parts: denaturation, primer annealing, and primer extension. Discovery of PCRs in 1986 made this process reiterative, leading to an exponential increase in the production of the amplified target. Soon after, alternative amplification techniques were developed and patented by companies, which used different enzymes and strategies, but they are all based on reiterative reactions. Many different amplification targets including both DNA or RNA fragments have been proposed. The target most frequently amplified in MTB is the IS6110 (31) repetitive element, of which 10 to 16 copies are present in most clinical isolates. Numerous techniques for nucleic acid extraction have been proposed, as have different types of controls for monitoring the efficacy of nucleic acid extraction and amplification procedures. Currently, the U.S. Food and Drug Administration (FDA) requires that culture (still considered the “gold standard” for TB diagnosis) must be done in conjunction with the performance of each amplification-based test. In this paper, we review and discuss the currently available commercial methods which are capable of detecting MTB directly from clinical samples.

CURRENTLY AVAILABLE COMMERCIAL DIRECT AMPLIFICATION TESTS (CDATS)

AMPLICOR M. TUBERCULOSIS assay.

(i) Manufacturer. The AMPLICOR M. tuberculosis assay is produced by Roche Molecular System, Branchburg, N.J.

(ii) Description. The AMPLICOR MTB assay is a PCR-amplified qualitative test whose target is represented by a 584-bp segment of the 16S rRNA gene shared by all the members of the genus Mycobacterium. The procedure, starting from a 100-μl aliquot of decontaminated specimen, consists of four steps: (i) specimen preparation, (ii) target amplification by PCR, (iii) hybridization of amplified products to oligonucleotide probes, and (iv) detection of the probe-bound amplified products. The whole process, with the exception of the sample preparation, is automatically performed on the COBAS AMPLICOR. An internal amplification control (IAC) (a short fragment of synthetic DNA) is introduced into each amplification reaction mixture and coamplified with target DNA to detect inhibiting substances. Finally, hybrid detection is accomplished by a colorimetric reaction. Assay results are available within 6 to 7 h. This method is approved by the FDA for testing on smear-positive respiratory samples. A manual version (lacking the IAC) is also commercially available.

(iii) Literature review. At present, many reports have been published evaluating the performance of the Amplicor assay both in respiratory and extrapulmonary samples (Table 1). Results are often difficult to compare because of the different bias in design as well as in analysis. Published data include a high prevalence of smear-positive respiratory samples with a ratio between MTB culture-positive and smear-positive specimens ranging from 1.09 to 1.49 (mean, 1.27). The above ratio was reported ranging from 2.25 to 4.40 in two studies (3, 19) analyzing patient rather than specimen data (Table 2). Sensitivities in respiratory specimens (compared with culture and clinical diagnosis) ranged from 83 to 96.7%, from 90 to 100%, and from 50 to 95.9% for overall, smear-positive, and smear-negative specimens, respectively. Data from Cohen et al. (19) and Rajalahti et al. (51) showed a strong correlation between test sensitivity (within all the categories of respiratory specimens) and the number of tested specimens for each patient, thus supporting the hypothesis that CDAT performance is critically affected by the mycobacterial burden and its distribution in the sample. When extrapulmonary specimens or mixtures of respiratory and extrapulmonary specimens were evaluated, sensitivities ranged from 27.3 to 85%, from 87.5 to 100% and from 17.2 and 70.8% for overall, smear-positive and smear-negative specimens, respectively. Disappointing results were obtained (11, 29, 43, 53) when testing pleural fluids, gastric aspirates, lymph nodes, and cerebrospinal fluids. On the other hand, Gamboa et al. (27) did not find any significant difference in sensitivities and specificities between the COBAS Amplicor with the IAC and the manual version, as none of the 755 specimens showed inhibition. The nature of CDAT inhibition is still unclear, and with the exception of stool samples (the majority of which show strong inhibition), inhibiting substances have been detected in clinical samples from less than 1% to about 20% (10, 22, 51, 52, 53). Unfortunately, several specimens showing inhibition were found to be MTB culture positive, and the inhibition rate was shown to be significantly higher in extrapulmonary specimens (57) (Table 3). In this context, the combination of fully automated amplification and detection procedures with the availability of IAC may be regarded as the major advantages of the COBAS AMPLICOR system. Overall Amplicor specificity ranged from 91.3 to 100%. False-positive results in comparison with culture were observed for specimens collected from patients receiving anti-TB chemotherapy (11, 27, 52, 59) or were related to cross-reactions with NTM (19, 52, 62). Finally, literature data confirm excellent performances with smear-positive respiratory specimens, thus supporting FDA approval. However, because of the lower sensitivity with extrapulmonary and smear-negative respiratory samples, Amplicor may still be of value, provided that it is used on the basis of a sound clinical suspicion.

TABLE 1.

Evaluation of the AMPLICOR (PCR) assay for detection of MTB in clinical samples^a

Study (reference)	No. of specimens/ specimen type	No. of MTB cultures/smear- positive specimens	Gold standard used for assay's evaluation	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Sensitivity (%) for:
								Smear-positive specimens	Smear-negative specimens
Mitarai et al. (43)	75/Eç	22/4	Culture, clinical correlation	27.3	97.6	NA	NA	100	17.2
Gomez-Pastrana et al. (29)	251/R+Eç	21/6	Culture, clinical correlation	44	93.7	73.3	80.8	NA	NA
Rajalahti et al. (51)	324/R	76/51	Culture, clinical correlation	83	99	97	95	90	68
Eing et al. (22)	1,527/R+E	65/32	Culture, clinical correlation	66.3	99.7	94.4	97.7	87.5	56.5
Reischl et al. (52)	643/R	56/44	Culture, clinical correlation	84.2	99.1	NA	NA	95.4	50
	506/E	39/25	Culture, clinical correlation	82.3	98	NA	NA	100	61.5
Rimek et al. (53)	43/E	15/2	Culture, clinical correlation	45.5	91.3	NA	NA	NA	NA
Bogard et al. (10)	5,077/R	333/249	Culture, clinical correlation	85.2	99.7	96.4	98.8	96.1	71.7
Shah et al. (59)	1,090/Eç	32/10	Culture, clinical correlation	76.4	99.8	92.8	99.2	90	70.8
Gamboa et al. (27)	755/Rç	223/176	Culture, clinical correlation	90.8	100	100	95.8	100	51
	755/R	223/176	Culture, clinical correlation	92.4	100	100	96.5	100	59.6

^aAbbreviations: R, respiratory specimens; E, extrapulmonary specimens; ç, manual version not including IAC; NA, not available.

TABLE 2.

Patient-based evaluation of the AMPLICOR assay for detection of MTB^a

Study (reference)	No. of patients/specimen type	No. of MTB cultures/smear-positive patients	Gold standard used for assay's evaluation	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Sensitivity (%) for:
								Smear-positive patients	Smear-negative patients
Cohen et al. (19)	85/Rç	27/12	Culture, clinical correlation	74	93	NA	NA	100	53
Al Zahrani et al. (3)	487/Rç	44/10	Culture, clinical correlation	42	100	NA	NA	NA	NA
Bonington et al. (11)	35/Eç	3/3	Culture, clinical correlation	28.6	100	100	53.7	NA	NA

^aAbbreviations: R, respiratory specimens; E, extrapulmonary specimens; ç, manual version not including IAC; NA, not available.

TABLE 3.

Comparative studies of available CDATs for detection of MTB in clinical samples^a

Study (reference)	Method	No. of specimens/ specimen type	No. of MTB cultures/smear- positive specimens	Gold standard used for assay's evaluation	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Sensitivity (%) for:
									Smear-positive specimens	Smear-negative specimens
Piersimoni et al. (48)	LCx	273/R	61/36	Culture, clinical correlation	75.7	98.8	96.4	90.5	91.6	58.8
		184/E	24/11	Culture, clinical correlation	53.6	99.3	93.7	92.1	81.8	35.3
	AMTD2	273/R	61/36	Culture, clinical correlation	92.8b	99.4	98.5	97	100	85.3b
		184/E	24/11	Culture, clinical correlation	78.6b	99.3	95.6	96.2	100	64.7b
Wang and Tay (64)	LCx	230/R	72/66	Culture, clinical correlation	100	99.3	98.7	100	100	100
	PCR	230/R	72/66	Culture, clinical correlation	96.1	100	100	98.1	96.9	91.7
	AMTD	230/R	72/66	Culture, clinical correlation	98.6	99.4	98.6	99.4	100	85.7
Brown et al. (12)	LCx	42/R	NA	Culture, clinical correlation	79.2	100	100	77.2	84.2	55.6
		21/E	NA	Culture, clinical correlation	60	100	100	76.5
	PCRc	42/R	NA	Culture, clinical correlation	75.0	100	100	77.5	84.2	33.3
		21/E	NA	Culture, clinical correlation	50	100	100	74
Tortoli et al. (62)	LCx	697/R+E	110/84	Culture, clinical correlation	85.7d	99.4	NA	NA	93	70.2d
	PCRc	697/R+E	110/84	Culture, clinical correlation	75.5	99.8	NA	NA	94	36.1
Scarparo et al. (57)	AMTD2	296/R	114/97	Culture, clinical correlation	85.7	100	100	90.4	91.7	65.5
		190/E	33/25	Culture, clinical correlation	82.9	100	100	95.5	88	75
	PCR	296/R	114/97	Culture, clinical correlation	94.2	100	100	96	98.9	75
		190/E	33/25	Culture, clinical correlation	85	100	100	96.1	95.8	68.7
Della-Latta and Whittier (20)	AMTD2	1385/R	62/NA	Culture, clinical correlation	97.1	99.5	NA	NA	100	92.9
	PCRc	1380/R	62/NA	Culture, clinical correlation	96.7	100	NA	NA	97.4	95.9
Piersimoni et al. (49)	AMTD2	331/R	91/76	Culture, clinical correlation	88	99.2	110e	0.11e	93.4	62.5
		184/E	30/22	Culture, clinical correlation	74.3	100	235e	0.05e	95.4	47
	DTB	331/R	91/76	Culture, clinical correlation	94.5	99.6	740e	0.26e	98.7	75
		184/E	30/22	Culture, clinical correlation	92.3f	100	920e	0.07e	100	82.3

^aAbbreviations: R, respiratory specimens; E, extrapulmonary specimens; NA, not available.

^bStatistically significant with both respiratory (P = 0.005) and extrapulmonary (P = 0.048) specimens.

^cManual version not including IAC.

^dStatistically significant with either all specimens (P = 0.026) or smear-negative (P = 0.020) specimens.

^ePositive and negative likelihood ratios.

^fStatistically significant with extrapulmonary (P = 0.03) specimens.

Amplified M. tuberculosis Direct (AMTD2) assay. (i) Manufacturer. The AMTD2 assay is produced by Gen-Probe, Inc., San Diego, Calif.

(ii) Description. The AMTD2 assay is an isothermal transcription-mediated amplification method in which the target (mycobacterial 16S rRNA) is amplified by DNA intermediates. The RNA amplicons produced are then identified by the hybridization protection assay with an acridinium ester-labeled MTB complex-specific DNA probe. The entire process is autocatalytic and is performed at 42°C with a heat block. Although the reaction is performed in a single tube to reduce carryover contamination, no IAC is included in the assay. Test results, expressed as relative light units (RLUs), are available within 2.5 h from specimen submission. This method is approved by the U.S. FDA for testing on smear-positive and smear-negative respiratory samples.

(iii) Literature review. At present, several studies have been published in the literature evaluating the performance of this technique both on respiratory and extrapulmonary specimens (Table 4). Respiratory specimens were prevalent, showing a ratio between MTB culture-positive and smear-positive specimens that ranged from 1.17 to 1.97 (mean, 1.5). The overall test sensitivity in respiratory specimens (compared with culture and clinical diagnosis) ranged from 85.7 to 97.8% and was shown to be higher for smear-positive specimens (91.7 to 100%) while dropping to between 65.5 and 92.9% in smear-negative specimens. Bergmann et al. (8) evaluated the AMTD2 assay on respiratory specimens collected from 486 prison inmate patients, reporting (after resolution of discrepancies) by patient an overall sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 90.9, 99.1, 83.3, and 99.6% (Table 5). The overall values changed to 100, 100, 100, and 100% and 83.3, 99.1, 71.4, and 99.6% when evaluating smear-positive and smear-negative patients, respectively. These results, similar to those reported by other authors (26, 45, 48), seem to support the diagnostic efficacy of the AMTD2 assay on all pulmonary specimens, regardless of the smear microscopy result. This is not surprising, as the test was applied on the right specimens (high-quality specimens) collected from the right patients (soundly suspected of having TB). In this context, Catanzaro et al. (13) evaluated the performance of the AMTD2 assay with specimens from different patients stratified by level of clinical suspicion. They reported sensitivities of 83, 75, and 87% and corresponding specificities of 97, 100, and 100 for low, intermediate, and high clinical suspicion patients, respectively. PPVs were 59% (low), 100% (intermediate), and 100% (high) while the corresponding NPVs were 99, 91, and 91%. They concluded that although the AMTD2 performed well with specimens from patients with an intermediate or high clinical suspicion of TB, it lacked predictivity, whereas clinicians were more likely to depend on results from the laboratory. When extrapulmonary specimens were evaluated, overall sensitivity ranged from 74.3 to 100%. Smear-positive specimens showed sensitivities ranging from 88 to 100%, whereas for smear-negative specimens, it dropped to between 63.6 and 100%. Selecting 311 cerebrospinal fluid specimens on the basis of a sound clinical suspicion and testing at least 2 samples for each patient, Chedore and Jamieson (18) reported sensitivity, specificity, PPV, and NPV of 93.8, 99.3, 88.2, and 99.7%, respectively. A major disadvantage of AMTD2 was the absence of the IAC, thus preventing the test from detecting inhibitors. These substances have been detected from less than 1 to 5% of clinical samples (8, 26, 45, 49, 57, 67) and represent a serious drawback, as some inhibited specimens were not only found to be MTB culture positive but a minority of them were also smear positive. The overall AMTD2 test specificity ranged from 92.1 to 100%. Some papers showed specificity close to 100% (26, 57) while others reported false-positive results, ranging from about 1 to 7.1% (1, 20, 34, 45, 49). In this context, most false-positive results exhibited low RLUs close to the traditional cutoff of 30,000 RLUs. As a consequence, the manufacturer has recently changed the cutoff values and recommended retesting all the samples that yielded RLUs between 30,000 and 500,000. This approach, supported by recent data aiming at determining the optimal cutoff value (300,000 RLUs) or to establish an equivocal zone in the interpretation of results appeared to improve test specificity without reducing sensitivity (35, 41). Cross-reactions for the presence of mycobacteria other than MTB in the specimen have also been reported (1, 45, 49, 61). Finally, literature reports based on a large number of extrapulmonary specimens (1, 26, 45) seem to encourage the judicious use of AMTD2 to include this category of samples provided that the test is required on the basis of clinical suspicion and interpreted according to patient data.

TABLE 4.

Evaluation of the AMTD2 assay for detection of MTB in clinical samples^a

Study (reference)	No. of specimens/ specimen type	No. of MTB cultures/smear- positive specimens	Gold standard used for assay's evaluation	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Sensitivity (%) for:
								Smear- positive specimens	Smear- negative specimens
Gamboa et al. (26)	410/R	95/48	Culture, clinical correlation	94.7	100	100	98.4	100	83
	272/E	68/21	Culture, clinical correlation	86.8	100	100	98.4	100	88.9
Woods et al. (67)	175/E	23/21	Culture, clinical correlation	89.3	100	100	98	NA	NA
O'Sullivan et al. (45)	391/R	46/30	Culture, clinical correlation	97.8	99.1	93.9	99.7	100	75
	164/E	22/9	Culture, clinical correlation	77.3	98.5	91.7	96.4	90	63.6
Alcalà et al. (1)	663/R	115/79	Culture, clinical correlation	90.8	93.2	74.5	97.9	97.5	77.5
	238/E	33/13	Culture, clinical correlation	88.9	92.1	66.7	97.9	92.3	87
Chedore and Jamieson (17)	823/R+E	245/230	Culture, clinical correlation	100	99.6	97.4	100	100	100
Chedore and Jamieson (18)	311/E	21/10	Culture, clinical correlation	93.8	99.3	88.2	99.7	NA	NA

^aAbbreviations: R, respiratory specimens; E, extrapulmonary specimens; NA, not available.

TABLE 5.

Patient-based evaluation of the AMTD2 assay for detection of MTB^a

Study (reference)	No. of patients/specimen type	No. of MTB cultures/smear-positive patients	Gold standard used for assay's evaluation	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Sensitivity (%) for:
								Smear-positive patients	Smear-negative patients
Catanzaro et al. (13)	338/R	65/43	Culture, clinical correlation	83	97	88	95	NA	NA
Bergmann et al. (8)	486/R	22/10	Culture, clinical correlation	90.9	99.1	83.3	99.6	100	83.3

^aAbbreviations: R, respiratory specimens; E, extrapulmonary specimens; NA, not available.

LCx MTB assay, ABBOTT LCx probe system. (i) Manufacturer. The LCx MTB assay, ABBOTT LCx probe system is produced by Abbott Laboratories, Abbott Park, Ill.

(ii) Description. The ligase chain reaction is a DNA amplification method in which the probe molecule is amplified in place of the target. After DNA denaturation, two pairs of primers anneal to each strand of the target, leaving a gap of 1 to 3 bases. The gap is then filled by the action of the DNA polymerase, and the primers are covalently linked by a ligase. Once ligated together, the first pair of oligonucleotides serves as a template to direct the ligation of new complementary oligonucleotides. The target is a single-copy chromosomal gene encoding the MTB protein antigen b. Sample preparation and DNA extraction are obtained by 95°C heat inactivation followed by mechanical lysis. Amplification occurs in a thermal cycler, and detection is performed by a microparticle enzyme immune assay with the LCx fluorimetric analyzer. The assay is recommended by the manufacturer for use with respiratory specimens, it does not include an IAC, and starting from a digested, decontaminated specimen, it can be completed within 5 to 6 h. This method is not approved by the FDA for use in the United States, and in 2002, it was also withdrawn from the European market.

(iii) Literature review. To date, thousands of specimens have been evaluated by this technique at different sites worldwide by using the kit according to and without modifying the manufacturer's instructions. Most of the specimens used were collected in laboratories handling many MTB-positive samples (such as TB reference labs), with a prevalence of respiratory (even when a mixture of respiratory and extrapulmonary specimens were included in the study) and smear-positive samples (the ratio between MTB culture-positive and smear-positive samples ranged from 1.09 to 3.9; mean, 1.8). In this context, the test performed well it was shown to be one of the easiest systems to use in a nonspecialist laboratory. In published studies (Table 6), sensitivity (compared with liquid culture and clinical correlation) ranged from 69.7 to 96.8% overall and from 81.8 to 100% for smear-positive samples but drastically dropped to between 35.3 and 79.2% for smear-negative specimens and is even more disappointing in extrapulmonary samples. As the LCx Probe System MTB is a commercial kit recommended for respiratory specimens, Palacios et al. (46) appropriately modified the procedure to improve test sensitivity in extrapulmonary samples by increasing both the concentration of extracted target DNA and the number of amplification cycles. This upgrading seemed to work well, as sensitivity was very good (90.4%) despite no smear-positive specimen being included in the study. A similar approach was adopted by Lumb et al. (37) who lowered the cutoff value set by the manufacturer at 1.0 S/CO (the ratio of the sample fluorescent rate to the cutoff value calculated by reading positive and negative controls included in each run). A cutoff grey zone was set, ranging from 0.2 to 0.99 S/CO. Specimens falling within this zone were retested, and new specimens were collected from the patients to confirm clinical suspicion. Although the manufacturer affirmed that washing during the extraction phase eliminates inhibitory substances, Leckie et al. (36) did in fact discover an inhibitory substance (a calcium phosphate precipitate that forms during N-acetyl-l-cysteine NaOH decontamination and digestion) which required additional washing procedures to make it disappear. Piersimoni et al. (48) compared the performance of the LCx and the AMTD2 assays with 456 respiratory and extrapulmonary specimens obtained from 356 patients. Statistically significant differences in sensitivities were found for both overall and smear-negative specimens (Table 3). Finally, since the lack of sensitivity seems to be the main shortcoming of this system and specificity is much more satisfactory, the test can be routinely used with smear-positive samples when rapid differentiation between MTB and NTM infections is necessary.

TABLE 6.

Evaluation of the LCx assay for detection of MTB in clinical samples^a

Study (reference)	No. of specimens/specimen type	No. of MTB cultures/smear- positive specimens	Gold standard used for assay's evaluation	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Sensitivity (%) for:
								Smear-positive specimens	Smear-negative specimens
Alonso et al. (2)	322/R+E	11/8	Culture, clinical correlation	76.5	95.8	68.4	97.2	100	69.2
Palacios et al. (46)	235/E	18/0	Culture, clinical correlation	90.4	98.5	86.3	99	NA	90.4
Fadda et al. (23)	622/R+E	124/71	Culture, clinical correlation	96.8	100	100	99.2	100	79.2
Moore and Curry (44)	493/R	34/13	Culture, clinical correlation	77	99	91	98	100	56
Gamboa et al. (25)	526/E	130/33	Culture, clinical correlation	78.5	100	100	93.1	100	71.1
Garrino et al. (28)	737/R+E	61/46	Culture, clinical correlation	78	100	95	98	97.9	37.5
Lumb et al. (37)	2,347/R+E	152/79	Culture, clinical correlation	69.7	99.9	99.1	97.7	98.5	41.5
Viinanen et al. (63)	247/R	31/24	Culture, clinical correlation	83.9	97.7	83.9	97.7	96	42.8

^aAbbreviations: R, respiratory specimens; E, extrapulmonary specimens; NA, not available.

BD ProbeTec energy transfer (ET) system (DTB). (i) Manufacturer. The BD ProbeTec ET Direct TB System (DTB) is produced by Becton Dickinson Biosciences Microbiology Products, Sparks, Md.

(ii) Description. Strand displacement amplification (SDA) is an isothermal enzymatic process that amplifies DNA exponentially. Target sequences of IS6110 (specific to MTB) and 16S rRNA gene (common to most mycobacterial species) are coamplified. The process is based on the nicking of the recognition sequence in double-stranded DNA by a restriction endonuclease and further extension of that site from the 3′ end by the Klenow fragment of Escherichia coli DNA polymerase which synthesizes a new strand of DNA while displacing the existing one (54). The replicated DNA and the displaced strands are then substrates for repeated rounds of oligonucleotide annealing, nicking, and strand displacement. This entire process occurs at 52.5°C. Detection probes consist of the target-specific sequence together with a fluorescent dye and a quencher placed in such close proximity that they are unable to release any fluorescent signal. As the concentration of the amplification product increases, these probes hybridize to the product and are converted from stem-loop to double-stranded molecules which are cleaved by a restriction enzyme. This cleavage is characterized by an increase in fluorescence polarization (ET) detected and monitored through a kinetic fluorimetric reading. Sample preparation and DNA extraction are obtained by 105°C heat inactivation followed by ultrasonic lysis. Simultaneous amplification and real-time detection by fluorescent ET is performed by the ProbeTec instrument at a single temperature. The assay is recommended by the manufacturer for use with respiratory specimens and can be completed within 3.5 to 4 h starting from a digested, decontaminated specimen. An IAC, designed to detect the presence of inhibiting substances, is run with each sample. Currently, this method is not approved by the FDA for use in the United States.

(iii) Literature review. The first kit-based non-commercially available DTB system was evaluated in two major studies by Bergmann and Woods (7) and Pfyffer et al. (47), who tested the system on respiratory specimens (Table 7), reporting sensitivities and specificities ranging from 100 to 97.9% and 99.2 to 96.5%, respectively. As the system was only half automated, featuring a low throughput and an extensive hands-on time, a new thermophilic form of SDA in combination with fluorescence polarization (ET) was developed. In recent years, some published studies have evaluated the DTB system at different sites worldwide by using the kit according to and without modifying the manufacturer's instructions. Although the number of specimens collected from TB patients was large enough to permit statistical analysis, the composition of each study included a relatively high percentage of smear-positive, culture-positive samples, thus explaining the overall higher sensitivities than those obtained from smear-negative samples (the ratio between MTB culture-positive and smear-positive samples ranged from 1.0 to 2.8; mean, 1.6) In general, the test performed well and offers several advantages for clinical laboratories performing routine CDATs. The most important one is the inclusion of an IAC in the same well as the patient specimen. Moreover, amplicon contamination is minimized, as the sealed microwell in which amplification occurs does not undergo any reopening. After mycobacterial inactivation, which occurs during the initial processing of the specimen, the remaining procedure does not need to be performed in a biological safety cabinet. Finally, initial specimen processing and amplification may be carried out in the same room (as suggested by the manufacturer), and all the reagents can be stored at room temperature. Sample preparation is the most labor intensive and represents the main shortcoming of the system; thereafter, the assay is almost completely automated. Data taken from the literature (Table 7) give a rate of sensitivity (compared with liquid culture and clinical correlation) ranging from 60.7 to 100% overall and from 98.5 to 100% for smear-positive samples. Although all published studies reported lower sensitivities for smear-negative and extrapulmonary samples (ranging from 33.3 to 85.7%), this difference seems to be less significant than that observed with other methods. The improved performance in sensitivity was believed to be due mainly to IAC availability, which permits the easy detection of inhibitory samples. Specificity was shown to be very good, ranging from 98.9 to 100%, and no cross-reaction between MTB and NTM in both respiratory and extrapulmonary specimens was reported. Inhibition rates ranged from 0.3 to 14%, showing higher percentages in extrapulmonary samples (33, 39). Though retesting inhibitory samples, once diluted or after undergoing freeze-thaw treatment, are often successful, Barrett et al. (6) reported that a considerable number of samples showed irreversible inhibition (10.5%). Comparing the performance of the DTB and AMTD2 assays with 515 respiratory and extrapulmonary specimens obtained from 402 patients, Piersimoni et al. (49) reported statistically significant differences in sensitivities for both extrapulmonary specimens (P = 0.03) and patients with a conclusive TB diagnosis (P = 0.04). These differences were associated with the presence of inhibitory samples (even smear positive) which the AMTD2 assay was unable to detect in the absence of an IAC.

TABLE 7.

Evaluation of the BD ProbeTec assay (DTB) for detection of MTB in clinical samples^a

Study (reference)	No. of specimens/ specimen type	No. of MTB cultures/smear-positive specimens	Gold standard used for assay's evaluation	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Sensitivity (%) for:
								Smear-positive specimens	Smear-negative specimens
Bergmann and Woods (7)	523/R	24/15	Culture, clinical correlation	100	99.2	85.7	100	100	100
Pfyffer et al. (47)	799/R	41/28	Culture, clinical correlation	97.9	96.5	63.9	99.9	100	92.3
Bergmann et al. (9)	600/R	16/12	Culture, clinical correlation	93.8	99.8	93.8	99.8	100	75
Johansen et al. (33)	351/R	150/85	Culture, clinical correlation	82.7	99	NA	NA	100	60
	372/E	192/67	Culture, clinical correlation	60.7	98.9	NA	NA	98.5	40.3
Barrett et al. (6)	200/R	104/101	Culture, clinical correlation	97.1	96	97	96	99	33.3
Maugein et al. (39)	547/R	69/43	Culture, clinical correlation	89.5	98.2	88.5	98.3	100	76.4
	74/E	8/4	Culture, clinical correlation					100	85.7
Mazzarelli et al. (40)	537/R	184/135	Culture, clinical correlation	91.3	98.1	49.2b	0.1b	99.2	70.6
	294/E	68/28	Culture, clinical correlation	77.8	97.7	33.3b	0.2b	90	69

^aAbbreviations: R, respiratory specimens; E, extrapulmonary specimens; NA, not available.

^bPositive and negative likelihood ratios.

INNO-LiPA RIF.TB assay.

(i) Manufacturer. The INNO-LIPA RIF.TB assay is produced by Innogenetics NV, Zwijndrecht, Belgium.

(ii) Description. The INNO-LiPA RIF.TB line probe assay (LiPA) adopts a reverse hybridization method in which the labeled probe is represented by the amplicon of a short region (about 70 bp) of the gene encoding the β-subunit of the RNA polymerase (rpoB gene) shared by all members of the MTB. The label (biotin) is incorporated in the amplicon during amplification. Starting from 500 μl of a decontaminated specimen, a nested PCR is performed by adding 5 μl of the extracted DNA to 45 μl of a master mix containing the outer LiPA primers. After the first round of amplification (30 cycles), 1 μl of the amplified product is transferred to 40 μl of the second PCR master mix containing the inner LiPA biotinylated primers. A second round of amplification (30 cycles) then occurs. After gel analysis, the amplified biotinylated products are denatured and hybridized with 10 oligonucleotide immobilized probes as parallel lines on a membrane-based strip. The LiPA strip contains (i) a line for conjugate control, (ii) a specific probe for the MTB, (iii) five partially overlapping probes (S1 to S5) which span the entire rpoB core region and hybridize to the wild-type sequence, and (iv) four probes (R2, R4a, R4b, and R5) which identify some of the most frequent rpoB mutations (more than 20 mutations have been described so far). Finally, hybrids are detected through a colorimetric reaction as purple-brown lines (55). The LiPA can detect the presence of both MTB and its resistance to rifampin, does not include IAC, and can be completed in about 12 h. This method is not, however, approved by the FDA for use in the United States.

(iii) Literature review. Although several reports have been published in the literature, most of them were performed to evaluate the ability of the test to detect rifampin resistance from cultures at an early stage of growth rather than directly from clinical specimens (14). Drobniewski et al. (21), Gamboa et al. (24), and Watterson et al. (65) evaluated the LiPA on 59 and 91 selected respiratory and extrapulmonary specimens, respectively, and on 36 selected respiratory specimens, all of which yielded a positive MTB culture (many of these specimens were from patients on anti-TB drugs). The overall sensitivities were 98.3, 89, and 94.7%, respectively. A similar study by Marttila et al. (38), evaluating the assay on 75 respiratory and extrapulmonary specimens from 70 patients with sound TB clinical suspicion, showed a sensitivity of 58.8%. Finally, although this assay offered a valuable advantage, owing to its ability to detect the simultaneous presence of MTB and rifampin resistance, further studies are necessary to evaluate the performance of this assay and how it can fit into the clinical laboratory workflow.

DISCUSSION

FDA-approved and non-FDA-approved uses of CDATs.

In 1999, the U.S. FDA approved a reformulated AMTD (also known as enhanced AMTD or AMTD2) for the detection of MTB in AFB smear-positive and smear-negative respiratory specimens from patients suspected of having TB. AMTD and AMPLICOR tests had been previously approved for direct detection of MTB with regard to AFB smear-positive respiratory specimens only (15). Moreover, it was stated that both assays could not be used on specimens collected from patients who had been receiving anti-TB medication for more than 7 days or had been treated for TB within the last 2 months. In 2000, the Centers for Disease Control and Prevention updated its recommendations for use of CDATs for the diagnosis of active TB. In brief, it is recommended that sputum specimens be collected on three different days for AFB microscopy and culture. Should AFB smears be negative, CDAT is performed on the first specimen collected, otherwise it is performed on smear-positive specimen(s). If smear and CDAT are both positive, this confirms TB. If the smear is positive and the CDAT is negative, a test for inhibitors needs to be performed. If inhibiting substances are not detected on at least two consecutive smear-positive, CDAT-negative samples, the patient is presumed to have NTM. If a specimen is smear negative and CDAT positive and the same result is obtained from an additional specimen, the patient can be presumed to have TB. Finally, if both the smear and CDAT are negative in two consecutive specimens, the patient can be presumed to be active TB free (16). Nevertheless, Centers for Disease Control and Prevention recommendations conclude that the latter does not completely exclude TB, and the decision regarding therapy ultimately rests on clinical judgment. Although, in published studies, the specificity of all CDATs has been shown to be extremely high while the sensitivity varied widely, the clinical value of such tests depends largely on their PPVs and NPVs or their more recently introduced equivalents: positive and negative likelihood ratios. These values vary considerably with the pretest probability of TB. In other words, to determine the clinical utility of any CDAT result in individual patients, it is essential to consider the degree of clinical suspicion. Over the last few years, investigators have started seriously taking into account clinical factors in their research design, but unfortunately, it is evident that clinical judgement and clinical skills are not uniform entities, as are the analytic performances of standardized CDAT assays. Given the additional expense of CDATs, their use with patients for whom the likelihood of TB is either very high or very low may represent an improper use of healthcare resources (5, 42, 58). CDATs should be used primarily when they are more likely to influence decisions regarding further diagnostic evaluation and anti-TB therapy, such as cases when the likelihood of TB is neither very high nor very low. This particularly includes those patients (unfortunately the majority in industrialized countries) for whom AFB smears are negative but clinical suspicion is considerable or very high. Moreover, even though body fluids have been sterilized by prior empirical TB treatment, correct diagnosis must necessarily be made by CDATs on dead bacilli (4, 42, 58, 60). In this context, FDA licensing seems to have placed restrictions on CDAT use and clinical usefulness (42).

Detection of false-negative results.

Clinical specimens collected from patients with active TB can have false-negative results when tested by the currently available CDATs. The majority of such samples are smear negative and represent an important challenge, owing to their paucibacillary nature and uneven AFB distribution. In this case, an inappropriate specimen dilution during decontamination procedures or a sampling error will reduce the test sensitivity, thus generating a false-negative result. Other possible causes of false-negative results include the presence of inhibitors or a suboptimal target extraction. Sample sediments sporadically contain inhibiting substances such as nucleases and proteases that impair amplification, causing false-negative results. Usually, sediments are smear negative, but sometimes inhibition may also occur in smear-positive sediments. The introduction of a preformatted IAC in some kit-based assay seems to be the most desirable option compared with the manual incorporation of spiked controls or any of the many described inhibitor-removing procedures. Usually, inhibition is not found in all specimens obtained from the same patient; therefore, testing multiple specimens can solve the problem. In addition, currently available CDATs adopt target extraction methods based on physical agents such as heat, mechanical lysis, or sonication rather than conventional phenol-chloroform or other chemical-based extractions. Such methods have been developed to enhance access to amplification-based techniques, but they have probably oversimplified the target extraction step. The literature data suggest that extraction methods can be improved to allow more sensitive detection of the target than is now accomplished (32, 50).

Detection of false-positive results.

As currently available CDATs do not distinguish between live and dead organisms, they may remain positive for a long time after the institution or completion of therapy, even facing clinical evidence of good therapeutic response and outcome. To avoid this kind of false-positive result, it is well established that CDATs must not be used with patients on anti-TB chemotherapy. Another considerable aspect of some pseudo false-positive results comes from the widely used definition of positivity (the so-called gold standard), based on microbiological rather than clinical evidence. Due to the paucibacillarity of most smear-negative samples and the uneven AFB excretion and distribution within the specimen, culture cannot be considered the yardstick for the measurement of CDAT performance. In these cases, the two tests complement each other and CDAT results should be carefully interpreted alongside the clinical data. Moreover, it is suggested that high-quality specimens (66) be tested while taking into account that bronchial samples seem to be associated with a higher diagnostic accuracy than sputa (56). Finally, well-trained personnel, proper equipment, and adequate space are essential to avoid false-positive results associated with amplicon contamination. This problem, unfortunately very common with homemade assays, seems to be much less relevant with CDATs, whose design and procedures have been developed while keeping in mind how to drastically reduce amplicon contamination. Nevertheless, frequent environmental disinfection with 10% bleach is required along with random incorporation of sporadic negative controls to monitor amplicon contamination.

Laboratory characteristics for performance of CDATS.

It is current opinion that use and/or implementation of CDATs should take into account the level of laboratory service and should also be based on cost-effective analysis. Such tests should be restricted to laboratories that have already achieved the highest standards in performance of conventional procedures as reported for American Thoracic Society levels II and III (30). The frequency of testing needs to be carefully examined; a once-per-week run is likely to offer a suboptimal support to patient care, thus requiring performance of CDATs on a two-or three-times-per-week basis. In this context, as most laboratories do not have enough personnel to add a new labor-intensive procedure to the daily workload, automated systems should be preferred. Finally, test expense is very important and may be prohibitive, especially when comparing the average cost of CDATs with in-house tests. However, as it is well established that selective testing enables cost saving mainly by early detection and treatment of patients and appropriate use of costly isolation facilities, a judicious use according to well-defined clinical algorithms is likely to be cost effective (4). Patients whose specimens are smear positive and CDAT negative can be promptly released from airborne precautions, do not need contact investigation, and require a different therapy addressed toward the most frequently encountered NTM. Similarly, in those patients with smear-negative, CDAT-positive samples, TB diagnosis can be confirmed early in the course of disease and proper treatment can be started. This benefits the patients and their contacts (TB transmission from smear-negative individuals has been repeatedly documented), preventing them from undergoing further invasive diagnostic procedures (68).

CONCLUSION

Due to the theoretical ability of amplification methods to detect a single copy of genomic sequence, the introduction of CDATs was felt to be a true milestone in diagnostic mycobacteriology. Unfortunately, about 10 years later, this putative promise is still largely unmet, and a more pragmatic attitude has developed toward their use. In fact, although CDAT specificity was shown to be high, sensitivity is considerably less than that of culture. In addition, there is the need to evaluate sensitivity and specificity from the clinical point of view rather than from that of the analytical. Most often, excellent results we read in the literature are about analytical performance rather than clinical diagnosis. At present, CDATs are time-consuming and expensive and still demand a high level of expertise from experienced technologists who tend to be in short supply in all the industrialized countries. Since CDATs offer means to detect a pathogen for which conventional detection and identification procedures can take several weeks even in those laboratories featuring proper competence and expertise, we firmly believe that this kind of testing not only should be done in larger and reference laboratories but must also be grouped by discipline, rather than by technique. Taking into account that the choice of a CDAT may depend on many different issues, such as the load of samples processed, the laboratory size, its engineering and technical sophistication, and lastly, prevalence of TB or other NTM-related diseases, no single system may tout court fit all these requirements. In this context, the main features of the above-discussed CDATs as well as their major strengths and weaknesses are tentatively summarized in Table 8. As a rule of thumb, we suggest the adoption of an automated system featuring an IAC so that inhibiting substances present in the sample can be absolutely detected. Finally, it is recommended that CDATs should always be performed in conjunction with microscopy and culture and that the results should be interpreted alongside the patient's clinical data.

TABLE 8.

Comparison of different CDATs for detection of MTB in clinical samples^a

CDAT	Amplification method	Amplification target	Sample vol (μl)	Detection	Assay time (h)	Automation	IAC	Sensitivityb	Specificityb	FDA approval
AMTD2	TMA	16S RNA	450	Chemiluminescence	2.5	No	No	++++	+++	Yes
AMPLICOR	PCR	16S DNA	100	Colorimetric	6	Yes	Yes	+++	++++	Yes
LCx	LCR	PAB	500	Fluorimetric	6	Yes	No	+++	++++	No
DTB	SDA	IS6110	500	Fluorimetric (ET)	3	Yes	Yes	++++	++++	No
LiPA	Nested PCR	RpoB gene	500	Colorimetric	12	Yes	No	+++	++++	No

^aAbbreviations: TMA, transcription-mediated amplification; LCR, ligase chain reaction; PAB, protein antigen b.

^b+++, good; ++++, very good.