An assessment of nucleic acid amplification testing for active mycobacterial infection
Yüklə 3,88 Mb.
|
AFB = acid-fast bacilli; NTM = non-tuberculous mycobacteria; NTM-NAAT = NAAT designed to detect all NTMs; MAC-NAAT = NAAT designed to detect M. avian complex; NAAT = nucleic acid amplification testing The sensitivity and specificity of AFB, NTM-NAAT, MAC-NAAT and culture compared with either culture or a clinical reference standard for the individual studies are shown in Figure and Figure (Appendix ), and the pooled values for various subgroups are shown in Figure . When NTM-NAAT was compared with MAC-NAAT using culture as the reference standard, NTM-NAAT was more sensitive than MAC-NAAT (84%; 95%CI 49, 97 versus 59%; 95%CI 35, 79), but this difference did not reach statistical significance due to the wide CIs (Figure ). The difference in sensitivity between MAC-NAAT and NTM-NAAT may be due to the restricted mycobacterial species detectable using MAC-NAATs. Two studies that used commercial MAC-NAATs (Gamboa et al. 1997; Ninet et al. 1997) identified 13% (5/38) and 9% (6/68) culture-positive specimens, respectively, that grew NTMs not detectable by MAC-NAAT that could have been detected by NTM-NAAT. These specimens were treated as falsely negative in the analysis presented in this report. Conversely, 2 other studies included specimens with cultures positive for other NTMs as culture-negative results in the data presented (Bogner et al. 1997; Tran et al. 2014). Thus, 4% (18/494) and 7% (25/361) of culture-negative specimens, respectively, were actually NTM culture-positive, thus overestimating the sensitivity of MAC-NAAT in the detection of patients with NTM infections in these 2 studies. The study by Matsumoto et al. (1998) did not report the presence of any other NTMS.
Figure Forest plot showing the pooled sensitivity and specificity values for AFB and NAAT compared with culture or a clinical reference standard in diagnosing NTM infections in various types of specimens Comparisons using culture as the reference standard are shown in blue and those using a clinical reference standard in red. When there were 4 or more studies, pooled values were obtained using the ‘midas’ command in Stata; when there were less than 4 studies the pooled values were estimated using the ‘metan’ command. AFB = acid-fast bacilli; CRS = clinical reference standard; NTM-NAAT = NAAT designed to detect all non-tuberculous mycobacteria; MAC-NAAT = NAAT designed to detect M. avian complex; NAAT = nucleic acid amplification testing AFB microscopy was not very useful in identifying patients who did not have NTM infections. The pooled sensitivity for AFB microscopy versus culture was 47% (95%CI 24, 72), indicating that 53% of patients with a positive culture would have a false-negative result. When compared with a clinical diagnosis based on symptoms, histopathology and culture results, 77% of those diagnosed had a false-negative AFB result (pooled sensitivity 31%; 95%CI 4, 58). However, the pooled specificity was 97–100%, indicating that few patients would have a false-positive AFB result (Figure ). The LR scattergram in Figure A shows that the summary LR+ and LR– estimates for AFB microscopy are in the upper right quadrant, indicating that patients with a positive AFB microscopy result are indeed very likely to have an NTM infection detectable by culture. However, a negative AFB result does not exclude the possibility of having an NTM infection and has no diagnostic value. When comparing AFB microscopy with NTM-NAAT using culture as the reference standard, the pooled estimates suggested that NTM-NAAT was more sensitive (47% versus 84%; Figure ), but this was not statistically significant as the CIs overlapped. There was a smaller difference in the pooled sensitivity for AFB microscopy compared with MAC-NAAT, with the CIs almost completely overlapping (47% versus 59%; Figure ). Nevertheless, fewer patients would receive a false-negative result with NAAT compared with AFB microscopy, with up to 23% of NTM culture-positive patients being AFB-negative and NTM-NAAT-positive.
Figure LR scattergram for diagnosis of NTM infection by AFB microscopy (A) and NAAT (B) compared with culture AFB = acid-fast bacilli; NTM-NAAT = NAAT designed to detect all non-tuberculous mycobacteria; MAC-NAAT = NAAT designed to detect M. avian complex; NAAT = nucleic acid amplification testing NTM-NAAT was less specific than MAC-NAAT using culture as the reference standard (90%; 95%CI 46, 99 versus 100%; 95%CI 99, 100), but did not differ significantly to that for AFB microscopy (96%; 95%CI 68, 100). It should be noted that culture is an imperfect reference standard. When compared with a clinical reference standard, only 46% (95%CI 27, 66) of those clinically diagnosed were culture-positive, and only 31% (95%CI 4, 58) were AFB-positive. The median sensitivity for NTM-NAAT (99%, range 98–100, k=2; Figure ) was higher than for culture and the specificity ranged from 87–100%. This suggested that most patients who were NTM-NAAT-positive and culture-negative probably had clinical disease. In Figure , the LR scattergram showed that the summary LR+ and LR– values were in the top right quadrant, suggesting that a patient with a positive MAC-NAAT most likely had a MAC infection detectable by culture. However, a negative MAC-NAAT result does not eliminate the possibility of being culture-positive. On the other hand, the summary LR+ and LR– values for NTM-NAAT were within the green shaded areas, indicating that a patient with a positive NTM-NAAT is more likely to be culture-positive than not, and that a negative NTM-NAAT may be suggestive of not having a culture-positive infection. Thus, NAAT performed similarly to AFB microscopy in the ability to confirm the presence of culture-positive NTM infections, but NTM-NAAT was more likely to correctly predict the absence of disease. The SROC curve, which depicts the relative trade-off between true-positive and false-positive results, shows a trend indicating that there may be a threshold effect between MAC and NTM-NAATs (Figure ). The AUCs for NTM-NAAT (0.92; 95%CI 0.90, 0.94) and MAC-NAAT (1.00; 95%CI 0.98, 1.00) indicate that the NAATs perform well in predicting culture positivity (AUC > 0.9), whereas AFB microscopy (0.76; 95%CI 0.72, 0.79) performs only moderately (AUC 0.7–0.9). There was also a threshold effect between NTM-NAAT and MAC-NAAT indicating that NTM-NAAT is more sensitive and less specific. Figure SROC curve for all studies investigating the sensitivity and specificity of AFB and NAAT versus culture in the diagnosis of NTM AUC = area under curve; SROC = summary receiver–operator characteristic; NTM-NAAT = NAAT designed to detect all non-tuberculous mycobacteria; MAC-NAAT = NAAT designed to detect M. avian complex; NAAT = nucleic acid amplification testing Although the pooled sensitivity of the 3 MAC-NAAT studies that included HIV-positive patients was only 49% compared with 69% for the other 2 MAC-NAAT studies, this difference was not significant as the wide CIs almost completely overlapped (Figure in Appendix ). Thus, there is no obvious difference in test performance in specimens from HIV-positive patients compared with those that are HIV-negative. No conclusions could be reached about the accuracy of NAAT in the diagnosis of NTM in AFB-positive or -negative specimens. Not surprisingly, many specimens included in the analysis were AFB-negative (mean 16%, range 0-39), largely due to the paucibacillary nature of many specimen types tested for NTM infections. As a result, only 3 studies provided any data on AFB-positive specimens, 2 of which used NTM-NAAT and 1 MAC-NAAT. When these studies were compared with the 6 studies (3 NTM-NAAT and 3 MAC-NAAT) that provided data on AFB-negative specimens, the variability between studies was so great that no conclusions could be reached (Figure in Appendix ). Together, these results suggest that NAAT may be a better diagnostic test for diagnosing NTM infections than either AFB microscopy or culture. However, the results should be viewed with caution due to the small number of studies included and the wide 95%CIs for many of the analyses. Yüklə 3,88 Mb. Dostları ilə paylaş:
|