This study was a retrospective study of MDR-TB and XDR-TB M. tuberculosis isolates involving TB patients during 2005-2012 at the Office of Disease Prevention and Control Region 10 (DPC 10) in the north of Thailand as shown in Figure 1. The DPC 10 laboratory is a Regional TB Laboratory covering TB patient treatment from eight provinces in the upper north of Thailand, which can provide M. tuberculosis (MTB) cultures, identification, and Drug Susceptibility Tests (DST) for first- and second-line drugs. The MTB isolates were subcultured, then tested for phenotypes for first- and second-line drug resistance to isoniazid (INH), rifampicin (RIF), ofloxacin (OFX), and kanamycin (KAN) at DPC 10. Further, genetic site mutation for drug resistance in the corresponding resistant gene (katG, inhA, ahpC, rpoB, gyrA and rrs) was performed at Macrogen in Korea. Medical records were retrospectively reviewed for demographic data, diagnosis, and laboratory identification and DST results.

161 MTB multidrug and extensive drug resistant strain isolates from TB patients during 2005-2012 were subcultured from collections at DPC 10 received from 8 hospitals in the upper north of Thailand. Only 34 isolates were able to grow in 5 ml of 7H9 broth supplemented with PANTA and 3% Ogawa. Samples were collected from individual TB patients who presented with the initial treatment status.

Phenotypes testing on first and second line anti-tuberculous drugs (INH, RIF, OFX, and KAN) were performed on 34 isolates of M. tuberculosis, using the proportion method on LJ medium [11] . DST was completed according to the WHO guideline for DST testing for first- and second-line anti-tuberculosis drugs for DOTS-plus [12] . DST was determined using the indirect agar proportion method which, was performed on an LJ medium supplemented individually with anti-TB drugs, which included INH (0.2 µg/ml), RIF (40.0 µg/ml), OFX (2.0 µg/ml), and KAN (30 µg/ml).

34 isolates were grown on solid media (Löwenstein-Jensen and OGAWA), and chromosomal DNA was extracted using the commercial kit method with MolecuTech REBA MTB-MDR 2011. The purified DNA pellet was stored at 4˚C until use.

Six loci were amplified by PCR: katG, inhA, and ahpC (INH); rpoB (RIF); gyrA (OFX); and rrs (KAN) at Macrogen. Genetic site mutations in the corresponding resistance gene (katG, inhA, ahpC, rpoB, gyrA, and rrs) were performed using Macrogen molecular laboratory outsources. The primers are presented in Table 1 [4] [13] [14] . The result of sequencing was then subjected to comparison and analysis.

The sequencing data produced by the ABI 3730xl DNA analyzer were reviewed for confidence levels with an ABI sequence scanner, and chromatograms were analyzed for the presence or absence of mutations by comparison with published sequences of H37Rv. The data on clinical patients, resistant genes, genetic site mutation, and phenotype were compiled using the Excel 2010 database. Statistical analysis was performed using a statistical software package. The baseline characteristics of demographic data, treatment outcome and the genetic site mutation were presented using frequency and percentage. The associations between the demographic data, treatment outcome, resistant gene and MDR, Pre-XDR, and XDR were evaluated using exact probability tests. Univariable logistic regression was used to quantify the strength of the association between the demographic data, the gene mutation with Mycobacterium tuberculosis and the prevalence of Pre-XDR and XDR-TB among the MDR-TB patients.

There was no statistically significant difference (p < 0.05) between the characteristic of MDR-TB and Pre-XDR/XDR-TB cases (Table 2).

The majority of treatments in of the MDR-TB and Pre-XDR/XDR-TB patients were similar that found combination directly observed and self-administered for therapy type, CAT V (I) for treatment pattern, during on treatment more than 24 months in Table 3. The majority of side effects were different in two groups that found minor side effect (75.0%) in MDR-TB patients but found major side effect (40.0%) in Pre-XDR or XDR-TB patients in Table 3.

The treatment outcome resulting as “cure” was observed mainly in MDR-TB (50%). Cure/successful treatment was found 30% in Pre-XDR/XDR-TB group with defaulted (30%) and dead (30%) as shown in Table 4. However, it is found that 20% of deaths in Pre-XDR/XDR-TB patients occurred before the initiation of TB treatment.

The analysis was conducted on 34 isolates of which 24 was MDR-TB, 9 Pre XDR and 1 XDR-TB. DNA sequencing was tested following six resistant genes: katG, inhA, ahpC, rpoB, gyrA, and rrs. The katG, inhA, ahpC, and rpoB indicated resistance to the first-line antibiotic treatment, while gyrA and rrs indicated resistance to the second-line antibiotic treatment. The distribution of MDR, Pre-XDR and XDR-TB by mutation site in katG, inhA, ahpC, rpoB, gyrA, and rrs gene can be seen in Table 5 and Table 6. The isoniazid (INH) resistant isolates had genetic site mutations within the katG gene, inhA gene, and ahpC gene, which had mutated in many codons (Table 5). The majority of the katG gene mutations in MDR-TB had a genetic site mutation in codon 315 (Table 5). There was no mutation in any katG codon of the 14 cases in MDR, Pre-XDR/XDR-TB (Table 5). There were two cases of isoniazid (INH) drug resistance that exhibited no mutation to any genetic site on the katG, inhA, and ahpC gene in the MDR-TB patients. Two cases found mutation only a katG gene in the Pre-XDR-TB patients in our study. Mutation of rpoB 445 codon was significantly found in Pre-XDR/XDR-TB isolates (50%) than in MDR-TB isolates (23.5%) with the p-value of 0.031 (Table 6).

Our study found that Pre-XDR/XDR-TB patients significantly presented a mutation in the region of the rpoB gene at codon 445 (C445T) 20.6 times than the MDR-TB patients (P = 0.026) (Table 7). The results also showed that the prevalence inhA gene mutation at codon 114 (T114G) was significantly higher (8.1 times) in the Pre-XDR/XDR-TB patients than in the MDR-TB patients (p = 0.034) (Table 7). Also the data presented that minor side effect was significantly lower (0.14 times) in the Pre-XDR/XDR-TB patients than in the MDR-TB patients (p = 0.020) (Table 7).

The predictive markers in a logistic model (the mutation of the inhA gene at codon 114, the rpoB gene at codon 445, the rrs gene at codon 414 and minor

side effect can be explained 89.6% the probability of Pre-XDR/XDR-TB among MDR-TB (Figure 2).

Our study showed that the inhA gene mutation position at 114 (T114G) and the rpoB gene mutation position at 445 (C445T) maybe used as a tool to predict the Pre-XDR/XDR-TB patients. The mutation of T114G or C445T was more likely to be associated with the development to Pre-XDR-TB and XDR-TB among MDR-TB, with the chance of 8.1 and 20.6 times, respectively. In our study, gene mutation in inhA 114 was detected in 82.6% (19/23) of the MDR and in 17.4% (4/23) of the Pre-XDR or XDR-TB strains. There have been no previous reports of inhA 114 mutation in MDR-TB and XDR-TB strains; however this could be a case of silent mutation. Mutations of inhA are also commonly found at (−15) [17] [18] [19] [20] among the Mycobacterium tuberculosis drug resistant strains that can be found among TB and MDR-TB patients.

Many studies have documented that rpoB 445 is very specific to rifampicin resistance, which has been used to detect MDR-TB [21] [22] [23] . In our study, gene mutation in rpoB was detected in 85.3% (29/34) of the MDR and XDR-TB strains and was more likely to be found in Pre-XDR and XDR-TB patients by about 20 times when compared with the MDR-TB patients. One study showed that rpoB 445 was a very strong factor in predicting rifampicin resistance [24] . A previous study in Swaziland showed rpoB 445 mutation in MDR-TB patients (79.17%) [22] . The rpoB 445 mutation was also found during the outbreak of MDR-TB in Argentina in 1973 [23] . Previous studies have shown that rpoB445 could predict MDR with high specificity but low sensitivity [19] .

The attention paid for treatment of ADR with minimum modification of treatment regimen that was increased cure rate [25] . Also the previous study in MDR-TB without co-infection with HIV showed ADR was not effect to stop treatment [26] . Minor side effects appeared to have little impact on treatment completion and the conversion to Pre-XDR/XDR-TB because patients tended to visit healthcare providers more often to discuss their side effect concerns, resulting in better continuation of care and treatment which indirectly lowering the conversion to Pre-XDR/XDR-TB.