Diagnosis of Alzheimer's Disease Based on Deeply-Fused Nets.

AIM AND OBJECTIVE: Fast and accurate diagnosis of Alzheimer's disease is very important for the care and further treatment of patients. Along with the development of deep learning, impressive progress has also been made in the automatic diagnosis of AD. Most existing studies on automatic diagnosis are concerned with a single base network, whose accuracy for disease diagnosis still needs to be improved. This study was undertaken to propose a method to improve the accuracy of the automatic diagnosis of AD. MATERIALS AND METHODS: MRI image data from the Alzheimer's Disease Neuroimaging Initiative were used to train a deep learning model to achieve a computer-aided diagnosis of Alzheimer's disease. The data consisted of 138 with AD, 280 with mild cognitive impairment, and 138 normal controls. Here, a new deeply-fused net is proposed, which combines several deep convolutional neural networks, thereby avoiding the error of a single base network and increasing the classification accuracy and generalization capacity. RESULTS: Experiments show that when differentiating between patients with AD, mild cognitive impairment, and normal controls on a subset of the ADNI database without data leakage, the new architecture improves the accuracy by about 4 percentage points as compared to a single standard based network. CONCLUSION: This new approach exhibits better performance, but there is still much to be done before its clinical application. In the future, greater research effort will be devoted to improving the performance of the deeply-fused net.

Epileptic Seizure Detection in EEG Signals Using a Unified Temporal-Spectral Squeeze-and-Excitation Network.

The intelligent recognition of epileptic electro-encephalogram (EEG) signals is a valuable tool for the epileptic seizure detection. Recent deep learning models fail to fully consider both spectral and temporal domain representations simultaneously, which may lead to omitting the nonstationary or nonlinear property in epileptic EEGs and further produce a suboptimal recognition performance consequently. In this paper, an end-to-end EEG seizure detection framework is proposed by using a novel channel-embedding spectral-temporal squeeze-and-excitation network (CE-stSENet) with a maximum mean discrepancy-based information maximizing loss. Specifically, the CE-stSENet firstly integrates both multi-level spectral and multi-scale temporal analysis simultaneously. Hierarchical multi-domain representations are then captured in a unified manner with a variant of squeeze-and-excitation block. The classification net is finally implemented for epileptic EEG recognition based on features extracted in previous subnetworks. Particularly, to address the fact that the scarcity of seizure events results in finite data distribution and the severe overfitting problem in seizure detection, the CE-stSENet is coordinated with a maximum mean discrepancy-based information maximizing loss for mitigating the overfitting problem. Competitive experimental results on three EEG datasets against the state-of-the-art methods demonstrate the effectiveness of the proposed framework in recognizing epileptic EEGs, indicating its powerful capability in the automatic seizure detection.

[Methodology research and preliminary assessment of Mycobacterium tuberculosis detection by immunomagnetic beads combined with functionalized fluorescent quantum dots].

OBJECTIVE: To establish a detection method for Mycobacterium tuberculosis (MTB) by immunomagnetic beads combined with functionalized fluorescent quantum dots technology, and to investigate the optimal test condition and the diagnostic value of this method. METHODS: MTB standard strain H37Rv was used as detection object. Nanobeads and quantum dots were prepared by using wet chemical method, and conjugated separately with MTB binding peptide H8 to obtain immunomagnetic beads and functionalized fluorescent quantum dots, which could react with H37Rv simultaneously and form a ternary complex structure. Based on measurement of the fluorescence value and observation under fluorescence microscopy to determine if MTB existed in the sample, a new detection method of MTB using nanotechnology was established. The optimal detection concentration and reaction time of immunomagnetic beads and quantum dots were investigated, and the detection limit and specificity of this detection method were evaluated by using bacterial suspension and simulation sputum samples. RESULTS: By fluorescence microscopy examination, it was found that conjugated immunomagnetic beads and functionalized fluorescent quantum dots both bound with H37Rv and formed the ternary complex structure. The fluorescent value ratio of the experimental group and the control group could be 4:1. The best detection concentration of immunomagnetic beads and functionalized fluorescent quantum dots was 100 mg/L and the optimal incubation time was 2 h. The detection limit of H37Rv bacterial suspension and simulation sputum sample were both 10(3) CFU/ml. The detection results for 3 non-mycobacteria were all negative, while for the 12 types of NTM, only Mycobacterium parafortuitum, Mycobacterium aurum, Mycobacterium smegmatis and Mycobacterium fortuitum were positive, and others were all negative. CONCLUSION: The detection method of immunomagnetic beads combined with fluorescence quantum dots can be a new detection method for MTB, but the clinical value needs to be evaluated further.

[The study of inter-simple sequences repeat genotyping based on one pentanucleotide repeat sequences in mycobacteria].

OBJECTIVE: To establish inter-simple sequences repeat (ISSR) molecular makers based on (CAGCG)n repeat sequence in mycobacteria. METHODS: The distribution of pentanucleotide repeat sequence (CAGCG)n in mycobacterial genomes was analyzed by MICdb 2.0 software in the microsatellite database. ISSR primer MISP6 based on (CAGCG)n sequences was designed and tested in mycobacterial strains, which included 17 mycobacterial strains and 41 Mycobacterium tuberculosis clinical strains. RESULTS: The abundances of pentanucleotide repeat sequences (CAGCG)n were high in most of the mycobacterial genomes and they were mainly located in the coding regions. The results of ISSR analysis in mycobacteria showed that 15 reference strains from mycobacteria were clustered into 2 major clusters. The first cluster contained 2 subtypes and the second cluster contained 4 subtypes. Forty-one clinical strains from Mycobacterium tuberculosis were divided into 2 major clusters by the analysis of MISP6 primer, and each cluster had 2 subtypes. CONCLUSION: ISSR primer MISP6 based on (CAGCG)n sequences can be used as a genetic marker to genotype mycobacterial strains.

[Cross-resistance between rifampin and rifabutin in multidrug resistant Mycobacterium tuberculosis complex strains].

OBJECTIVE: To study the cross-resistance between rifampin and rifabutin in multidrug resistant Mycobacterium tuberculosis complex strains, and therefore to provide laboratory data for using rifabutin in the treatment of multidrug resistant tuberculosis. METHODS: The MIC(90) of rifabutin and rifampin against 99 multidrug resistant Mycobacterium tuberculosis clinical strains were determined by microplate assays. Statistical analysis was performed by using the χ(2) test and the t test. RESULTS: The cross-resistance rate between rifampicin and rifabutin was 85.9% (85/99), but the MIC(90) of rifabutin (≤ 16 mg/L, median 2 mg/L) was significantly lower than that of rifampicin (≥ 2 mg/L, median > 32 mg/L). The cross-resistance rate increased with the resistance level of rifampicin. The cross-resistance strains in the lower and the medium groups were 0/9 and 5/9 respectively, while the strains of the high rifampicin-resistant group were almost all cross-resistant (98.8%, 80/81). CONCLUSION: Rifabutin had activities against rifampin resistant Mycobacterium tuberculosis complex strains in vitro, and therefore may be used as an alternative for the treatment of multidrug resistant tuberculosis.

Variation of Mycobacterium tuberculosis antigen-specific IFN-γ and IL-17 responses in healthy tuberculin skin test (TST)-positive human subjects.

OBJECTIVE: To determine the variation of IFN-γ and IL-17 responses to M. tuberculosis antigens in healthy TST+ humans. METHODS: We isolated peripheral blood mononuclear cells from 21 TST+ healthy adults, stimulated them with phytohemagglutinin (PHA), PPD, Ag85B, ESAT-6, and live M. bovis BCG, and assayed IFN-γ and IL-17 secretion by ELISA in supernatants after 24 or 72 hours of incubation respectively. RESULTS: As in other studies, we found a wide range of IFN-γ responses to M. tuberculosis antigens; the variation significantly exceeded that observed in the same donors to the polyclonal T cell stimulus, phytohemagglutinin (PHA). In addition, we assayed IL-17 secretion in response to the same stimuli, and found less subject-to-subject variation. Analysis of the ratio of IFN-γ to IL-17 secretion on a subject-to-subject basis also revealed a wide range, with the majority of results distributed in a narrow range, and a minority with extreme results all of which were greater than that in the majority of subjects. The data suggest that study of exceptional responses to M. tuberculosis antigens may reveal immunologic correlates with specific outcomes of M. tuberculosis infection. CONCLUSION: Variation of IFNγ and IFN-γ/IL-17 responses to mycobacterial antigens exceeds that of responses to the polyclonal stimulus, PHA, in TST positive healthy humans. This indicates a quantitative spectrum of human immune responses to infection with M. tuberculosis. Since the outcome of human infection with M. tuberculosis varies greatly, systematic study of multiple immune responses to multiple antigens is likely to reveal correlations between selected immune responses and the outcomes of infection.

Interferon-gamma release assays for the diagnosis of extrapulmonary tuberculosis: a systematic review and meta-analysis.

Interferon -gamma release assays (IGRAs) provide a new diagnostic method for Mycobacterium tuberculosis (TB) infection. However, the diagnostic value of IGRAs for extrapulmonary TB (EPTB) has not been clarified. We searched several databases and selected papers with strict inclusion criteria, evaluated the evidence of commercially available IGRAs (QuantiFERON(®) -TB Gold QFT-G or QFT-GIT and T-SPOT(®) .TB) on blood and the tuberculin skin test (TST) using random effects models. Twenty studies with 1711 patients were included. After excluding indeterminate results, pooled sensitivity for the diagnosis of EPTB was 72% [95% confidence interval (CI) 65-79%] for QFT-G or GIT and 90% (95% CI, 86-93%) for T-SPOT; in high-income countries the sensitivity of QFT-G or GIT (79%, 95% CI 72-86%) was much higher than that (29%, 95% CI 14-48%) in low/middle-income countries. Pooled specificity for EPTB was 82% (95% CI 78-87%) for QFT-G or GIT and 68% (95% CI 64-73%) for T-SPOT. Pooled sensitivity of TST from four studies in high-income countries was lower than that of IGRAs. T-SPOT was more sensitive in detecting EPTB than QFT-G or GIT and TST. However, both IGRAs and TST have similar specificity for EPTB. IGRAs have limited value as diagnostic tools to screen and rule out EPTB, especially in low/middle-income countries. The immune status of patients does not affect the diagnostic accuracy of IGRAs for EPTB.

[Identification and evaluation of a new nucleic acid amplification test target for specific detection of Mycobacterium tuberculosis complex].

OBJECTIVE: To identify and evaluate a new nucleic acid amplification (NAA) test target for specific detection of Mycobacterium tuberculosis (MTB) complex (MTC). METHODS: MTC-specific fragment was obtained by ISSR genotyping technology. Primer pairs were designed based on the sequences of MTC-specific fragment and tested in 211 mycobacterial strains including 107 MTC strains and 104 nontuberculous mycobacteria (NTM) strains. IS6110 element (specific identification of MTC strains) and 16s rRNA gene (specific identification of Mycobacterium) amplification were used as a control to evaluate the efficacy of the NAA test target in the detection of MTC strains. RESULTS: One MTC-specific fragment with the length of 588 bp, located in 315947 - 316534 of the genome from MTB reference strain H(37) Rv, were obtained, cloned and sequenced. MTC-specific primer pairs MTCF/R were designed based on these sequences. All 211 mycobacterial strains accurately produced the genus-specific 16s rRNA amplicon. All MTC strains were positive in the MTCF/R PCR amplification while 99% MTC strains (106/107) were positive in the amplification of IS6110 sequences. All NTM strains were negative in both IS6110 and MTCF/R PCR amplification. CONCLUSIONS: The MTC-specific fragment developed in this study can be used as a new NAA test target to correctly distinguish MTC from NTM.

[In vitro activities of clofazimine against different drug-resistant types of Mycobacterium tuberculosis].

OBJECTIVE: To study the in vitro antituberculous activities of clofazimine (CLF) to different drug-resistant types of Mycobacterium tuberculosis. METHODS: The minimal inhibitory concentration (MIC) of CLF and isoniazid (INH), rifampicin (RFP), ofloxacin (OFLX), amikacin (AK), and capreomycin (CPM) against sensitive, single-drug resistant (SDR), poly-drug resistant (PDR), multi-drug resistant (MDR), and extensive-drug resistant (XDR) Mycobacterium tuberculosis strains isolated clinically were determined by microplate assays. RESULTS: The MICs of CLF for sensitive, SDR, PDR, MDR and XDR strains of clinically isolated Mycobacterium tuberculosis were 0.06 - 4.00 mg/L, 0.03 - 4.00 mg/L, 0.06 - 8.00 mg/L, 0.06 - 8.00 mg/L, 0.03 - 8.00 mg/L. For the sensitive group, the MIC of CLF (0.06 - 4.00 mg/L) was higher than that of INH (0.06 - 0.25 mg/L) and RFP (0.06 - 0.25 mg/L), while there was no significant difference among OFLX (0.06 - 2.00 mg/L), AK (0.06 - 4.00 mg/L), and CPM (0.50 - 4.00 mg/L). For the single-drug resistant group, there was no significant difference among CLF (0.03 - 4.00 mg/L), INH (0.06 - 0.25 mg/L), and RFP (0.06 - 0.25 mg/L), but the MIC of CLF was lower than that of OFLX (0.25 - 8.00 mg/L), AK (0.06 - 4.00 mg/L), and CPM (0.50 - 8.00 mg/L). For the MDR group, there was no significant difference between CLF (0.06 - 8.00 mg/L) and AK (0.25 - 8.00 mg/L), but the MIC of CLF was lower than that of OFLX (0.125 - 8.00 mg/L) and CPM (0.50 - 8.00 mg/L). For the XDR group, the MIC of CLF (0.03 - 8.00 mg/L) was lower than that of others. CONCLUSION: CLF showed good in vitro activity against Mycobacterium tuberculosis, especially MDR and XDR strains.

[Screening of the binding peptides of Mycobacterium tuberculosis H(37)Rv].

OBJECTIVE: To screen the Mycobacterium tuberculosis H(37)Rv binding peptide using phage-displayed random peptide libraries, and to analyze the binding capacity of the peptide with Mycobacterium tuberculosis. METHODS: Inactive Mycobacterium tuberculosis H(37)Rv was used for screening of the binding peptide from the Ph.D.-7 peptide library, and Mycobacterium smegmatis was used for reverse screening during the 2(nd) to 4(th) rounds of screening. After 4 rounds of screening, single phages were randomly selected for DNA sequencing. The selected clones were tested by indirect enzyme linked immunosorbent assay (ELISA). The peptide of positive clone, which showed the highest affinity, was synthesized in vitro with fluorescent markers. The specific combination of the peptide with 16 mycobacterium standard strains and 3 other microbes (Pseudomonas aeruginosa, Staphylococcus aureus and Candida albicans) were observed by fluorescence microscopy. RESULTS: After 4 rounds of biopanning, remarkable enrichment of the phages that specifically bind with target molecules were observed. Single phages were randomly selected for sequencing analysis and 5 sequences were obtained. Five phages with different sequences were detected using indirect ELISA and all of them were found to be positive clones. Phage 8 showed the highest affinity with target molecule. The peptide of phage H8 was synthesized in vitro with fluorescent markers, and it was confirmed that the peptide could bind with H(37)Rv and other 15 mycobacterium including Mycobacterium smegmatis, but not with 3 other microbes. CONCLUSIONS: By using phage-displayed random peptide libraries, we obtained the binding peptide of H(37)Rv. It was shown that the peptide could bind with Mycobacterium tuberculosis specifically, which provided a new way for the detection of Mycobacterium tuberculosis in vitro.

[Screening of tuberculosis specific antibody binding peptides].

OBJECTIVE: To screen the specific antibody binding peptides of tuberculosis from phage-displayed random phage display (Ph.D.)-7 peptide libraries with purified IgG from tuberculosis serum, and to provide the basis for the development of serological detection reagent of tuberculosis. METHODS: Purified IgG of tuberculosis serum was used as solid ligand to screen the binding peptide from the Ph.D.-7 peptide library, according to the biopanning process of absorption, elution and amplification. Purified IgG of health people serum was used as the molecule of counter selection during the second and third selection. Phages were enriched after 3 rounds of screening, then 20 single phages separately eluted by IgG of tuberculosis and health people were randomly selected on each direction of the determination plates and amplified. The single chains DNA were extracted as template for sequencing. The combination abilities of selected clones to IgG of tuberculosis and health people were tested by indirect enzyme linked immunosorbent assay (ELISA), and the positive clone was identified. Serum samples, from 47 patients with tuberculosis and 37 healthy people vaccinated with BCG, were verified by positive phage clones using phage-ELISA. The verifying results of 24 serum samples used for panning were separately analyzed statistically. RESULTS: After 3 rounds of panning, remarkable enrichment of phages that could specifically bind with target molecules were observed. Single phages were randomly selected for sequencing analysis and 12 sequences were obtained. 12 phage clones with different sequences were amplified and detected with indirect ELISA and single phage H12 showed higher affinity with IgG of tuberculosis (S/N ≥ 2.1) and was identified as the positive clone. It was found that, in indirect ELISA with single Phage H12, the A(450) value of tuberculosis patients (0.105 ± 0.010) was significantly higher than that of healthy individuals (0.070 ± 0.005), and the t value was 2.912 (P < 0.0001). The A(450) value of 12 serum samples of tuberculosis patients and 12 samples of health individuals used for panning were 0.144 ± 0.016 and 0.052 ± 0.004, and the t value was 5.69 (P < 0.0001). CONCLUSION: By using phage-displayed random peptide libraries, we obtained the specific antibody binding peptides of tuberculosis, which showed specific binding activity with IgG of tuberculosis. It can provide a basis for the establishment of a new serological detection method of tuberculosis with peptide.

[Fast identification of mycobacteria in microtiter liquid culture].

OBJECTIVE: This research was to establish a method for fast identification of mycobacteria in microtiter liquid culture and to evaluate its clinical value. METHODS: 2-thiophenecarboxylic acid hydrazide (TCH) and paranitrobenzoic acid (PNB) at different concentrations were added into liquid culture in 96-well plate. Different mycobacterium standard strains were incubated in liquid culture with PNB and TCH for 7 to 10 days. According to the growth assay for 15 mycobacterium strains and 30 mycobacterium tuberculosis strains, the best PNB and TCH concentration were determined. A total of 424 clinical mycobacterium isolates were identified by microtiter liquid culture at the best PNB and TCH concentration. The results of microtiter liquid culture were compared with those of PCR and DNA sequencing. RESULTS: The best concentration of PNB was 200 µg/ml in microtiter liquid culture. Compared with the results of PCR, the sensitivity and specificity for identification of mycobacterium tuberculosis complex in microtiter liquid culture were 97.8% (306/313) and 100.0% (107/107) respectively and those for non-tuberculosis mycobacteria in microtiter liquid culture were 100.0% (107/107) and 96.5% (306/317) respectively. The best concentration of TCH was 0.5 µg/ml. Compared with the results of PCR, the sensitivity of mycobacterium tuberculosis in microtiter liquid culture was 100.0% (305/305). The specificity remained under and more studies were needed. CONCLUSION: In microtiter liquid culture with PNB and TCH, mycobacteria can be identified in 7 to 10 days. The results were accurate and the process was simple without expensive equipments. This method meets clinical needs and can be used in all level hospitals in China.

[Evaluation of microscopic observation drug susceptibility for drug susceptibility testing of mycobacterium tuberculosis in smear-positive sputum].

OBJECTIVE: To evaluate microscopic observation drug susceptibility (MODS) for mycobacterium tuberculosis drug susceptibility in smear-positive sputum. METHODS: Drug susceptibility of mycobacterium tuberculosis in 275 smear-positive sputum samples collected from TB patients were detected directly by MODS. The susceptibility of seven antimicrobials including streptomycin, isoniazid, rifampicin, ethambutol, levofloxacin, amikacin and capromycin were detected MODS. At the same time the sputum sample were cultured in MGIT 960 tube and the positive isolates were tested for drug susceptibility by MGIT 960 system. The results of MODS were analyzed and compared with that of MGIT 960. RESULTS: Of 275 smear-positive sputum, MODS detected 235 (85.45%). Results of MODS were obtained in a median time of 18 days (5 - 39 d). For the 235 MODS-positive samples, the compliance rates of MODS to MGIT of 7 drugs were 90.21% (212/235), 88.09% (207/235), 93.62% (220/235), 87.23% (205/235), 92.34% (217/235), 88.51% (208/235) and 86.81% (204/235) respectively. The sensitivity of MODS method were 83.33% (90/108), 85.11% (120/141), 90.74% (98/108), 85.71% (78/91), 86.73% (85/98), 76.92% (40/52) and 77.08% (37/48). The specificities of MODS method were 96.06% (122/127), 92.55% (87/94), 96.06% (122/127), 88.19% (127/144), 96.35% (132/137), 91.80% (168/183) and 89.30% (167/187) respectively. CONCLUSION: MODS is an optimal alternative method for direct and rapid drug susceptibility of sputum with high accuracy in a timely and affordable way in resource-limited settings.

Selection and application of peptide mimotopes of MPT64 protein in Mycobacterium tuberculosis.

Antibody responses can be useful markers of tuberculosis (TB) infection, especially in the screening of extra-pulmonary TB. MPT64 is an important antigen in Mycobacterium tuberculosis (MTB) infection and is used in serological diagnosis. However, large variability in the diagnostic accuracy of MPT64 as a serological tool has limited its application. Phage-displayed random peptide libraries have emerged as a powerful technique to select peptides (epitopes) or mimotopes that may serve as surrogate diagnostic markers in serological tests. In the present study, this method was employed to identify mimotopes of the MPT64 protein of MTB by screening a linear heptapeptide library with rabbit antibodies raised against MPT64 protein. Two antigenic mimotopes (M2 and M6) resembling B-cell epitopes of MPT64 were identified that bound the affinity purified anti-MPT64 polyclonal antibodies and competed with MPT64 for antibody binding. From the results of sequence alignment and a structure modelling figure of MPT64, the sequence of the 2nd to 5th amino acids (DSML) of M2 was totally consistent with the sequence of the 224th to 227th amino acids of MPT64 and the peptide is located on the surface of the space structure of MPT64, suggesting that it might be a linear epitope of MPT64. The recognition of both phage-displayed and synthetic peptides of M2 by the anti-MPT64 polyclonal antibodies also supported this. Although no recurring sequence and no analogue to MPT64 of M6 were found for sequence alignment, the recognition of both phage-displayed and synthetic peptides of M6 by the anti-MPT64 polyclonal antibodies indicated that it might be a mimotope of a conformational epitope of MPT64. According to the results of the reactivity of human sera with synthetic M2 and M6 peptides and MPT64, M2 showed a significantly higher AUC and sensitivity than M6 and MPT64, especially for the sera from sputum-negative TB patients, suggesting that the M2 mimotope may be useful in serological diagnostic testing for TB.

[Evaluation of the isothermal RNA amplification assay for Mycobacterium tuberculosis in sputum samples].

OBJECTIVE: To evaluate the use of isothermal RNA amplification assay for detection of Mycobacterium tuberculosis (SAT-TB) in sputum samples. METHODS: A total of 244 sputum samples from patients with pulmonary tuberculosis and those with other lung diseases were detected by SAT-TB and Lowenstein-Jensen (L-J) culture. The samples with different results between SAT-TB and L-J culture were tested by Mycobacterium tuberculosis PCR fluorescence diagnostic kits. The sensitivity and specificity of SAT-TB were calculated according to the results of L-J culture. The detection rates of SAT-TB and L-J culture were analyzed according to the clinical diagnosis and the difference of the 2 methods were analyzed by chi-square test. RESULTS: With the result of L-J culture as the reference, the sensitivity and the specificity of SAT-TB were 92% (71/77) and 86% (143/167), respectively. The accordance rate of SAT-TB and L-J culture was 88% (214/244). For tuberculosis patients, the detection rate of L-J culture and SAT-TB was 42% (75/177) and 54% (95/177) respectively. The difference between the detection rates of SAT-TB and L-J was significant by chi-square test (χ² = 4.527, P < 0.05). CONCLUSIONS: SAT-TB is a rapid, sensitive and specific method for detection of Mycobacterium tuberculosis in clinical sputum samples.

[The effects of gene mutation related to drug resistance and drug efflux pump in extensively drug-resistant tuberculosis clinical isolates].

OBJECTIVE: To explore the effects of 2 major drug-resistant mechanisms in clinically isolated strains of extensively drug-resistant tuberculosis (XDR-MTB). METHODS: Genomic DNA of 10 XDR-MTB strains isolated from Shanghai Pulmonary Hospital were extracted. The main gene mutations related to drug resistance and 15 SNPs unique to XDR-MTB clinical isolate KZN605 reported by the Broad Institute in USA were detected by sequencing. The changes of minimal inhibition concentration (MIC) of XDR-MTB isolates were detected before and after the addition of efflux pump inhibitors verapamil, CCCP and reserpine in liquid cultures. RESULTS: The mutation of rpoB, katG and rpsL occurred in all XDR-MTB strains. The mutation of gyrA, gyrB and rrs occurred in 9 strains, 2 strains and 6 strains respectively. There was no mutation of tlyA in all the strains. Most of the SNPs in KZN 605 strains were not detected in the clinical strains. The clinical strains showed no significant changes of MICs, except 1 strain for which the MIC of ofloxacin decreased by 16 times after addition of the efflux pump inhibitors. CONCLUSIONS: The gene mutations related to drug resistance are the key mechanism for the clinical XDR-MTB strains, while the efflux pumps partly play a role in the drug resistance to fluoroquinolones. The detailed mechanism of efflux pump mediated drug resistance to other anti-TB drugs needs further study.