Treatment Outcomes and Survival Based on Drug Resistance Patterns in Multidrug-resistant Tuberculosis

Doh Hyung Kim1, Hee Jin Kim2, Seung-Kyu Park3, Suck-Jun Kong4, Young Sam Kim5, Tae-Hyung Kim6, Eun Kyung Kim7, Ki Man Lee8, Sung-Soon Lee9, Jae Seuk Park1, Won-Jung Koh10, Chang-Hoon Lee11, and Tae Sun Shim12
1 Division of Allergy and Respiratory Medicine, Department of Internal Medicine, Dankook University College of Medicine, Dankook University Hospital, Cheonan, South Korea; 2Department of Epidemiology, The Korean Institute of Tuberculosis, Seoul, South Korea; 3Clinical Research Center, National Masan Hospital, Masan, South Korea; 4Department of Thoracic Surgery, Mokpo National Hospital, Mokpo, South Korea; 5Department of Internal Medicine, Yonsei University College of Medicine, Severance Hospital, Seoul, South Korea; 6Department of Internal Medicine, Hanyang University College of Medicine, Seoul, South Korea; 7Division of Pulmonary and Critical Care Medicine, Department of Internal Medicine, College of Medicine, Pochon CHA University, Seongnam, South Korea; 8Department of Internal Medicine, Chungbuk National University College of Medicine, Chungbuk National University Hospital, Cheongju, South Korea; 9Department of Internal Medicine, Inje University School of Medicine, Ilsan Paik Hospital, Goyang, South Korea; 10Division of Pulmonary and Critical Care Medicine, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea; 11Division of HIV and TB control, Department of Disease Prevention, Korea Centers for Disease Control and Prevention, Seoul, South Korea; 12 Division of Pulmonary and Critical Care Medicine, University of Ulsan College of Medicine, Asan Medical Center, Seoul, South Korea

Rationale: Few large-scale studies have investigated multidrugresistant tuberculosis (MDR-TB) treatment outcomes relative to drug-resistance patterns. Objectives: To assess the impact of additional drug resistances on treatment outcomes and long-term survival in a large HIV-negative MDR-TB cohort. Methods: Treatment outcomes and long-term survival of patients with MDR-TB newly diagnosed or retreated in 2000 to 2002 were retrospectively analyzed based on drug-resistance patterns after 58 years of follow-up. Measurements and Main Results: Of 1,407 patients with MDR-TB, 75 (5.3%) had extensively drug-resistant TB (XDR-TBre) by the revised denition; 159 (11.3%) had ooxacin-resistant preXDR-TB (preXDRTBo); and 117 (8.3%) had second-line injectable drug (SLID)resistant preXDR-TB (preXDR-TBs). Patients with XDR-TBre showed the lowest treatment success rate (29.3%) and the poorest long-term survival, and XDR-TBre was more strongly associated with long-term mortality than XDR-TB as originally dened (hazards ratio [HR], 3.15; 95% condence interval [CI], 2.064.83; P , 0.001 vs. HR, 2.15; 95% CI, 1.493.09; P , 0.001). Patients with either form of preXDR-TB showed poorer cumulative survival than those with ooxacin-susceptible/SLID-susceptible MDR-TB (P , 0.05 for each comparison). Although streptomycin susceptibility did not affect the treatment outcomes of patients with preXDR-TB, streptomycin-resistant preXDR-TB was more strongly associated with long-term mortality than ooxacin-susceptible/SLIDsusceptible MDR-TB (HR, 2.17; 95% CI, 1.223.84; P , 0.008 for pre XDR-TBo; and HR, 2.69; 95% CI, 1.405.16; P 5 0.003 for preXDR-TBs). Conclusions: The revised XDR-TB denition is appropriate for dening patients with MDR-TB with the poorest outcomes. Both preXDR-TBo and preXDR-TBs were independently associated with poor long-term survival in patients with MDR-TB. SM susceptibility was linked to better survival in patients with preXDR-TB. Keywords: tuberculosis; multidrug-resistant; survival rate; treatment outcome; drug susceptibility

AT A GLANCE COMMENTARY
Scientic Knowledge on the Subject

Treatment outcomes and long-term survival with respect to the original and revised denitions of extensively drugresistant (XDR-TB) have not been compared in previous studies, and the impact of preXDR-TB (multidrug-resistant tuberculosis [MDR-TB] resistant to either any uoroquinolones or at least one second-line injectable drug, but not to both) on MDR-TB treatment outcomes has also not received attention.
What This Study Adds to the Field

Our results conrmed that the revised XDR-TB denition was adequate to dene a subset of MDR-TB patients with the poorest treatment outcome and the worst long-term survival. Moreover, the presence of preXDR-TB was an independent prognostic factor of poor outcomes and survival in patients with MDR-TB. Streptomycin susceptibility was an important predictor of favorable long-term survival in patients with preXDR-TB.

(Received in original form November 3, 2009; accepted in nal form March 9, 2010 ) Supported by the International Tuberculosis Research Center (ITRC-2006010 and 2008002). Correspondence and requests for reprints should be addressed to Tae Sun Shim, M.D., Division of Pulmonary and Critical Care Medicine, Department of Internal Medicine, University of Ulsan College of Medicine, Asan Medical Center, 388-1 Pungnap-dong, Songpa-gu, Seoul, 138-736, South Korea. E-mail: shimts@amc.seoul.kr This article has an online supplement, which is accessible from this issues table of contents at www.atsjournals.org
Am J Respir Crit Care Med Vol 182. pp 113119, 2010 Originally Published in Press as DOI: 10.1164/rccm.200911-1656OC on March 11, 2010 Internet address: www.atsjournals.org

Drug-resistant tuberculosis (TB), especially multidrugresistant (MDR) TB, dened as TB resistant to both isoniazid and rifampicin, remains a serious threat to public health and TB control programs worldwide (1). In addition, the emergence of extensively drug-resistant (XDR) TB, a subset of MDR-TB, has complicated treatment and control of this disease. XDR-TB was originally dened as MDR-TB resistant to at least three of six main classes of second-line drugs (aminoglycosides, polypeptides, uoroquinolones [FQNs], thioamides, cycloserine, and p-aminosalicylic acid) (2), but the most recent denition of the World Health Organization (WHO), dened in 2006 (3), considers MDR-TB to be present if resistance is demonstrated to any FQN and to at least one second-line injectable drug (SLID), including amikacin, capreomycin, and kanamycin. Treatment outcomes with respect to additional drug-resistance patterns in patients with MDRTB have varied among reports. Several recent studies showed that resistance to additional rst-line drugs, other than isoniazid and rifampicin (4), FQNs (5, 6), kanamycin (7),

114

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 182

2010

capreomycin (8), and streptomycin (SM) (9), was independently associated with unfavorable treatment outcomes. Although the presence of XDR-TB has become the most serious negative prognostic factor in MDR-TB treatment (1015), treatment outcomes with respect to the original and revised denitions of XDR-TB have not been compared, and the impact of the division of patients with preXDR-TB (dened as those with MDR-TB resistant to either any FQN or at least one SLID, but not to both) into two subsets on MDR-TB treatment outcomes has also not received attention. Although SM is included in the class of aminoglycoside drugs, along with kanamycin and amikacin, SM susceptibility is not included in the denition of XDR-TB. Thus, the impact of SM susceptibility on treatment outcomes should be evaluated separately from the denition of XDR-TB. To investigate the impact of these drug-resistance patterns on treatment outcomes and survival in patients with MDR-TB, we analyzed outcomes in a large cohort of patients with MDR-TB, most of who were not coinfected with HIV, with respect to drug-resistance patterns. Preliminary results were presented at the 107th Conference of the Korean Academy of Tuberculosis and Respiratory Diseases held in 2008 in Seoul, South Korea (16).

(10). Adequate treatment duration was dened as 18 months, including 12 months after culture conversion. Short-term treatment completion was dened as follows: (1) inadequate treatment duration, but more than 6 months; (2) three consecutive negative cultures before treatment completion; and (3) treatment completion assessed by a physician based on favorable response. Treatment success was dened as the sum of cure, treatment completion, and short-term treatment completion. All outcomes were the initial treatment outcomes during the study period, and were analyzed based on AFB culture results.

Statistical Analysis
Chi-square and one-way analysis of variance tests were used in univariate analyses. Bonferroni correction was used if multiple comparisons were performed, and all P values evaluated in chi-square tests were Bonferroni-corrected. Cumulative survival was estimated by the Kaplan-Meier method and compared by the log-rank test. Cox regression was used to dene the predictors of long-term survival by multivariate analysis using the backward elimination procedure. SPSS software, version 13.0 (SPSS Inc., Chicago, IL), was used for all analyses, with P , 0.05 indicating signicance.

RESULTS
Demographics

METHODS
Study Participants and Data Collection
A total of 1,407 Korean patients with MDR-TB were enrolled from January 2000 to December 2002; their details have been described previously (10). Medical records were retrospectively reviewed for demography, TB treatment history, comorbidities, acid-fast bacilli (AFB) cultures and drug susceptibility test (DST) results, chest radiographs, treatment modalities, and outcomes. All patients were followed-up for 58 years after commencement of treatment, and information on vital status as of December 2007 was obtained from the death registry of the Korea National Statistical Ofce; however, this registry does not provide information on cause of death. This study was approved by the Institutional Review Board of the Asan Medical Center, Seoul, South Korea, and other institutions.

AFB Cultures and DST

AFB were examined by Ziehl-Neelsen staining, and were cultured on egg-based Ogawa medium. DSTs for isoniazid, rifampicin, ethambutol, pyrazinamide, SM, kanamycin, cycloserine, p-aminosalicylic acid, prothionamide, and ooxacin (OFX) were performed at all participating laboratories. Enviomycin-capreomycin susceptibility was determined at the Korean Institute of Tuberculosis, the Supranational TB Reference Laboratory, Seoul, South Korea, and one private laboratory. The proportion method and the pyrazinamidase test were used for DST at the Korean Institute of Tuberculosis. Of the FQNs and second-line aminoglycosides, only OFX and kanamycin susceptibility tests, respectively, were routinely performed during the study period. Thus, DSTs of amikacin and other FQNs were not evaluated.

Of 1,407 patients with MDR-TB, 75 (5.3%) were diagnosed with XDR-TBre; 159 (11.3%) with preXDR-TBo; 117 (8.3%) with preXDR-TBs; and 1,056 (75.1%) with the other form of MDR-TB (susceptible to both OFX and SLID). The mean age of the study patients was 42.9 years (range, 1389 years); the male to female ratio was 2.8:1; and the mean body mass index was 19.2 kg/m2 (range, 1232 kg/m2). HIV-ELISA tests were conducted on 66 (4.7%) patients, with only one found to be seropositive. Patient demographic characteristics are shown in Table 1. A previous history of treatment with second-line TB drugs was more common in patients with XDR-TBre (35.6%) and preXDR-TBo (32%) than in those with preXDR-TBs (17.5%) or the other form of MDR-TB (8.8%) (P , 0.05 for each comparison), and a history of at least two previous TB treatments was more common in patients with XDR-TBre (47.7%), preXDR-TBo (38.4%), and preXDR-TBs (27%) than in those with the other form of MDR-TB (24.9%) (P , 0.05 for each comparison). The proportion of patients with faradvanced disease on simple chest radiograph examination was higher in patients with XDR-TBre and preXDR-TBo than in the other patients with MDR-TB (50.7%, 45.5%, and 29%, respectively; P , 0.05 for each comparison).
Treatment Outcomes and Survival Based on the Denitions of XDR-TB

Classication of Patients
Four groups were dened as follows: (1) XDR-TBre (revised denition), MDR-TB with bacillary resistance to OFX and at least one of the SLIDs (kanamycin, capreomycin, or enviomycin); (2) XDR-TBor (original denition), MDR-TB with bacillary resistance to at least three of the six main classes of second-line drugs (2); (3) XDR-TB(or-re), XDR-TBor excluding XDR-TBre; and (4) preXDR-TB, MDR-TB resistant to either OFX (preXDR-TBo) or at least one SLID (pre XDR-TBs), but not to both drugs.

Denitions of Treatment Outcomes

The seven categories of outcome denitions included the six WHOrecommended categories (cure, treatment completion, transfer out, default, death, and failure) (17) and short-term treatment completion

The overall treatment success rate of the 1,407 study participants was 45.3% (cure 30.2%, treatment completion 6.6%, and short-term treatment completion 8.5%). Treatment success rates for patients with XDR-TBre (n 5 75), XDR-TBor (n 5 149, including 65 patients with XDR-TBre), and XDRTB(or-re) (n 5 84) were 29.3, 38.9, and 44%, respectively, and these rates did not signicantly differ among groups (P . 0.05 for each comparison). However, the success rate of the group with XDR-TBre was signicantly higher than that of those with non-XDR MDR-TB (n 5 1,248; success rate 46.3%) (P 5 0.025) (Table 2). Mean survival times were 61.7, 68, and 72.2 months in patients with XDR-TBre, XDR-TBor, and XDR-TB(or-re), respectively, and were thus lower than that (89.2 months) in patients with non-XDR MDR-TB (P , 0.001 for each compar-

Kim, Kim, Park, et al.: Outcomes and Survival in MDR-TB Patients TABLE 1. CLINICAL CHARACTERISTICS OF 1,407 PATIENTS WITH MDR-TB
XDR-TBre Characteristics Mean age, y 6 SD (range)* Male sex Mean BMI 6 SD (range)

115

PreXDR-TBo (n 5 159) 44.8 6 14.4 (1779) 112 (70.4) 19.1 6 3.3 (12.727) (n 5 150) 40 (26.7) 62 (41.3) 48 (32) 58/151 (38.4)

PreXDR-TBs (n 5 117) 41.7 6 16 (1380) 75 (64.1) 18.7 6 3 (13.925.2) (n 5 114) 29 (25.4) 65 (57) 20 (17.5) 30/111 (27)

Other MDR-TB (n 5 1,056) 42.4 6 14.7 (1389) 799 (75.7) 19.4 6 3.3 (1232) (n 5 1,022) 302 (29.5) 631 (61.7) 89 (8.7) 251/1,008 (24.9)

Total (n 5 1,407) 42.9 6 14.9 (1389) 1,039 (73.8) 19.2 6 3.2 (1232) (n 5 1,359) 385 (28.3) 791 (58.2) 183 (13.5) 370/1,335 (27.7)

(n 5 75) 47.8 6 15.5 (1680) 53 (70.7) 18.5 6 2.7 (13.223.7) (n 5 73)

Previous history of TB treatment (n 5 1,359) No history of previous TB treatment 14 (19.2) 33 (45.2) History of TB treatment with x rst-line drugs only 26 (35.6) History of TB treatment with second-line drugs*xk 31/65 (47.7) No. of previous TB treatment: >2 times (n 5 1,335)*x Underlying diseases Diabetes mellitus 14 (18.7) Chronic liver disease 1 (1.3) Malignancy 1 (1.3) { 4 (5.4) Others HIV seropositive (n 5 66) 0/6 (0) Extrapulmonary TB 1 (1.3) Positive AFB smear at treatment 52 (69.3) initiation Radiologic severity (n 5 1,251)** (n 5 67) Minimal 2 (3) Moderately advanced 31 (46.3) Far advancedx 34 (50.7) Cavitary disease 37 (49.3) Bilateral disease (n 5 1,190) 51/67 (76.1)

27 5 1 6 1/14 6 107

(17) (3.1) (0.6) (3.8) (7.1) (3.8) (67.3)

17 4 1 4 0/8 5 82

(14.5) (3.4) (0.9) (3.4) (0) (4.3) (70.1) (n 5 107) (5.6) (55.1) (39.3) (41) (68.3)

181 16 10 26 0/38 41 717

(17.1) (1.5) (0.9) (2.5) (0) (3.9) (67.9) (n 5 934) (8.2) (62.7) (29) (42.2) (73.7)

239 26 13 40 1/66 53 958

(17) (1.8) (0.9) (2.8) (1.5) (3.8) (68.1) (n 5 1,251) (7.6) (59.5) (32.9) (42.6) (73.9)

10 68 65 68 103/130

(n 5 143) (7) 6 (47.6) 59 (45.5) 42 (42.8) 48 (79.2) 71/104

77 586 271 446 655/889

95 744 412 599 880/1,190

Denition of abbreviations: AFB 5 acid-fast bacilli; BMI 5 body mass index; HIV 5 human immunodeciency virus; MDR 5 multidrug-resistant; TB 5 tuberculosis; XDR 5 extensively drug-resistant; XDR-TBre 5 revised denition of XDR-TB; preXDR-TBo 5 ooxacin-resistant non-XDR MDR-TB; preXDR-TBs 5 second-line injectable drug-resistant non-XDR MDR-TB. Values are n (%), means 6 SD, and (ranges), as shown. Statistical analyses were performed using chi-square test with Bonferroni correction and one-way analysis of variance test. P , 0.05. * XDR-TBre vs. preXDR-TBs. XDR-TBre vs. other MDR-TB. preXDR-TBs vs. other MDR-TB. x preXDR-TBo vs. other MDR-TB. k preXDR-TBo vs. preXDR-TBs. { Others included chronic lung disease (n 5 9), chronic renal disease (n 5 4), hypothyroidism or hyperthyroidism (n 5 5), connective tissue disease (n 5 3), psychiatric disease (n 5 5), malnutrition and postgastrectomy state (n 5 14). ** Radiographic severity was adopted from the recommendation of the National Tuberculosis Association of America.

ison) (Figure 1). Among the three groups with XDR-TB, presence of XDR-TBre was the strongest predictor of poor long-term survival in multivariate analyses. Compared with the presence of non-XDR MDR-TB, the hazard ratio (HR) was

3.15 (95% condence interval [CI], 2.064.83; P , 0.001) in patients with XDR-TBre; 2.15 (95% CI, 1.493.09; P , 0.001) in those with XDR-TBor; and 1.38 (95% CI, 0.802.39; P 5 0.252) in patients with XDR-TB(or-re), respectively (Table 3).

TABLE 2. TREATMENT OUTCOME BY XDR-TB DEFINITION

XDR-TBre Treatment outcomes Cure* Treatment completion Short-term treatment completion Failure* Transfer out Default* Death* Treatment success*

XDR-TBor (n 5 149) 31 10 17 24 15 23 29 58 (20.8) (6.7) (11.4) (16.1) (10.1) (15.4) (19.5) (38.9)

XDR-TB(or-re) (n 5 84) 22 7 8 12 8 14 13 37 (26.2) (8.3) (9.5) (14.3) (9.5) (16.7) (15.5) (44)

Non-XDR MDR-TB (n 5 1,248) 394 82 102 41 91 427 111 578 (31.6) (6.6) (8.2) (3.3) (7.3) (34.2) (8.9) (46.3) P Value ,0.001 0.895 0.408 ,0.001 0.304 ,0.001 ,0.001 0.015

(n 5 75) 9 4 9 12 9 12 20 22 (12) (5.3) (12) (16) (12) (16) (26.7) (29.3)

Denition of abbreviations: MDR 5 multidrug-resistant; TB 5 tuberculosis; XDR 5 extensively drug-resistant; XDR-TBre 5 revised denition of XDR-TB; XDR-TBor 5 original denition of XDR-TB; XDR-TB(or-re) 5 XDR-TBor excluding XDR-TBre. Values are n (%). Statistical analyses were performed using chi-square test. Bonferroni-corrected P , 0.05. * XDR-TBre vs. non-XDR MDR-TB. XDR-TBor vs. non-XDR MDR-TB. XDR-TB(or-re) vs. non-XDR MDR-TB.

116

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 182

2010

spectively, and success rates in patients with XDR-TBre and preXDR-TBo were signicantly lower than in patients with the other form of MDR-TB (P , 0.001 and P 5 0.018, respectively) (Table 4). With respect to long-term survival, patients with XDR-TBre (61.7 months) showed the worst survival compared with patients with preXDR-TBo (75.7 months), patients with preXDR-TBs (78.3 months), and patients with the other form of MDR-TB (91.1 months) (P , 0.001, P 5 0.032, and P , 0.001, respectively). Although the survival of patients with preXDRTBo seemed to be worse than that of those with preXDR-TBs, the difference was not signicant (P 5 0.062) (Figure 2). In multivariate analyses, the presence of any of XDR-TBre, pre XDR-TBo, or preXDR-TBs was also a signicant predictor of poor long-term survival. Compared with the survival of patients with the other form of MDR-TB, HR values were 3.76 (95% CI, 2.435.83; P , 0.001), 1.62 (95% CI, 1.092.40; P 5 0.018), and 1.57 (95% CI, 1.012.44; P 5 0.048), respectively.
Treatment Outcomes and Survival Based on SM Susceptibility in Patients with PreXDR-TB

Figure 1. Kaplan-Meier survival analysis estimates of long-term survival based on various denitions of XDR-TB. Mean survival time was 61.7, 68, 72.2, and 89.2 months in patients with XDR-TBre, XDR-TBor, XDRTB(or-re), and non-XDR MDR-TB, respectively. All three XDR-TB groups showed poorer long-term survival than did the non-XDR MDR-TB group (P , 0.05 for each comparison), but there was no signicant difference among the three XDR-TB groups (XDR-TBre vs. XDR-TBor, P 5 0.240; XDR-TBre vs. XDR-TB(or-re), P 5 0.066; XDR-TBor vs. XDR-TB(or-re), P 5 0.347). Black solid line 5 XDR-TBre, black dashed line 5 XDR-TBor, gray dashed line 5 XDR-TB(or-re), gray solid line 5 nonXDR MDR-TB. MDR 5 multidrug-resistant; TB 5 tuberculosis; XDR 5 extensively drug-resistant; XDR-TBre 5 revised denition of XDR-TB; XDR-TBor 5 original denition of XDR-TB; XDR-TB(or-re) 5 XDR-TBor excluding XDR-TBre.

Treatment Outcomes and Survival Based on Resistance to OFX and SLID

Treatment success rates were 29.3, 35.8, 47, and 47.6% in the groups with XDR-TBre, preXDR-TBo, preXDR-TBs, and the other form of MDR-TB (non-XDR and nonpre-XDR), re-

In the group with XDR-TBre, neither treatment success rate (nor other outcomes) differed between patients who had SMsusceptible disease (15/48; 31.2%) and those with SM-resistant disease (7/27; 25.9%) (P 5 0.627). When treatment outcome categories were compared among the four subgroups with pre XDR-TB, the treatment failure rate was signicantly lower in patients with SM-susceptible preXDR-TBs (4/80; 5%) than in those with SM-resistant preXDR-TBo (13/56; 23.2%) (P 5 0.009) (Table 5). However, cumulative survival was signicantly higher in the group with SM-susceptible preXDR-TBs than in the other three groups with preXDR-TB by univariate analysis (P , 0.05 for each comparison) (Figure 3), and SM-resistance was associated with relatively poor long-term survival in patients with preXDR-TB irrespective of OFX- or SLIDresistance status by multivariate analyses. Compared with patients with the other form of MDR-TB (those with nonXDR and nonpre-XDR disease), the HR was 1.22 (95% CI, 0.692.17; P 5 0.497) in patients with SM-susceptible pre XDR-TBs, 1.43 (95% CI, 0.872.35; P 5 0.158) in those with SM-susceptible preXDR-TBo, 2.17 (95% CI, 1.223.84; P 5

TABLE 3. PREDICTORS OF LONG-TERM SURVIVAL BY XDR-TB DEFINITION

Univariate Analysis Variables Age .40 years Male sex BMI ,18.5 kg/m2 Previous TB treatment with second-line drugs or more previous TB treatments Far advanced disease Hb ,10 g/dl Positive AFB smear at treatment initiation Mean No. of TB drugs used before 6 SD Mean No. of potentially effective TB drugs 6 SD Surgical treatment Denitions of XDR-TB XDR-TBre XDR-TBor XDR-TB(or-re) Non-XDR MDR-TB No. of Deaths, n (%) 329/833 (39.5) 362/1,139 (31.8) 115/224 (51.3) 113/231 (57.6) 182/424 (42.9%) 250/479 (52.2) 48/84 (57.1) 338/1,060 (31.9) 3.75 6 2.90 (04) 3.00 6 1.75 (24) 9/76 (11.8) 43/75 72/149 35/84 315/1,248 (57.3) (48.3) (41.7) (25.2) HR* (95% CI) 2.45 1.36 2.46 2.65 2.03 3.33 3.11 1.29 1.12 0.82 0.35 (2.012.99) (1.091.69) (1.993.04) (2.103.35) (1.682.46) (2.754.04) (2.294.21) (1.051.58) (1.081.17) (0.780.87) (0.180.67) P Value ,0.001 0.006 ,0.001 ,0.001 ,0.001 ,0.001 ,0.001 0.015 ,0.001 ,0.001 0.002 ,0.001 ,0.001 ,0.001 Multivariate Analysis HR* (95% CI) 2.49 (1.893.28) 2.08 (1.492.91) 2.29 (1.683.11) 1.98 (1.522.58) 1.58 (1.042.41) 0.43 (0.171.05) 3.15 (2.064.83) 2.15 (1.493.09) P Value ,0.001 ,0.001 ,0.001 ,0.001 0.032 0.062 ,0.001 ,0.001

2.87 (2.083.94) 2.29 (1.772.95) 1.89 (1.332.68) 1

Denition of abbreviations: AFB 5 acid-fast bacilli; BMI 5 body mass index; CI 5 condence interval; Hb 5 hemoglobin; HR 5 hazard ratio; TB 5 tuberculosis; XDR 5 extensively drug-resistant; XDR-TBre 5 revised denition of XDR-TB; XDR-TBor 5 original denition of XDR-TB; XDR-TB(or-re) 5 XDR-TBor excluding XDR-TBre. Values are n (%), means 6 SDs, and ranges. * Hazard ratio for all-cause mortality.

Kim, Kim, Park, et al.: Outcomes and Survival in MDR-TB Patients TABLE 4. TREATMENT OUTCOMES OF PATIENTS WITH XDR-TBRE AND preXDR-TB
XDR-TBre Treatment Outcomes Cure* Treatment completion Short-term treatment completion Failure Transfer out Default Death* Treatment success

117

PreXDR-TBo (n 5 159) 34 9 14 22 12 36 32 57 (21.4) (5.7) (8.8) (13.8) (7.5) (22.6) (20.1) (35.8)

PreXDR-TBs (n 5 117) 34 13 8 7 11 31 13 55 (29.1) (11.1) (6.8) (6) (9.4) (26.5) (11.1) (47)

Other MDR-TB (n 5 1,056) 348 67 88 24 76 374 79 503 (33) (6.3) (8.3) (2.3) (7.2) (35.4) (7.5) (47.6)

Total (n 5 1,407) 425 93 119 65 108 453 144 637 (30.2) (6.6) (8.5) (4.6) (7.7) (32.2) (10.2) (45.3) P Value ,0.001 0.222 0.646 ,0.001 0.421 ,0.001 ,0.001 0.001

(n 5 75) 9 4 9 12 9 12 20 22 (12) (5.3) (12) (16) (12) (16) (26.7) (29.3)

For denition of abbreviations, see Table 2. Values are n (%). Statistical analyses were performed using chi-square test. Bonferroni-corrected P , 0.05. * XDR-TBre vs. preXDR-TBs. XDR-TBre vs. other MDR-TB. preXDR-TBo vs. other MDR-TB.

0.008) in patients with SM-resistant preXDR-TBo, and 2.69 (95% CI, 1.405.16; P 5 0.003) in those with SM-resistant pre XDR-TBs (see Table 5E in the online supplement).

DISCUSSION
The effects of drug-resistance on MDR-TB treatment outcomes have varied in previous studies (49, 18), and it is often difcult to compare data because of relatively small patient numbers in previous studies. Moreover, no study has compared treatment outcomes among patients classied using both the revised and original denitions of XDR-TB. In the present work, both longterm survival after 5 to 8 years of follow-up and initial treatment outcomes were analyzed using the second-largest MDR-TB cohort ever studied, exceeded only by the 1,989 patients with MDR-TB examined in Peru (19). Outcomes were compared with respect to both the original and revised denitions of XDR-TB to investigate the appropriateness of the two classications, and were also analyzed among groups of patients with XDR-TBre, preXDR-TBo, preXDR-TBs, and the other form (OFX-susceptible/SLID-susceptible) of MDR-TB to determine the impact of preXDR-TB status on outcomes in patients with MDR-TB. In addition, results were separately examined with respect to SM susceptibility to determine the impact of SM status on outcomes of patients with XDR-TBre and preXDRTB. Our results conrmed that the revised XDR-TB denition was optimal to dene a subset of patients with MDR-TB with the poorest treatment outcome and the worst long-term survival. Moreover, the results showed that the denition of preXDR-TB was an independent prognostic factor of poor outcomes and survival in patients with MDR-TB and also showed that SM susceptibility was an important predictor of favorable long-term survival in patients with preXDR-TB. In previous studies, the presence of XDR-TB was found to be the most serious obstacle to successful treatment of MDRTB (4, 1013, 15). In the present work, 57.3% (43/75) of patients with XDR-TBre eventually died during follow-up, and this group had the poorest long-term survival rate. Since publication of the revised denition of XDR-TB (3), no report has compared treatment outcomes and long-term survival relative to the new and old criteria. As a new category of drug-resistant TB, XDR-TB should be clearly distinguished from non-XDR MDRTB because patients with the former condition have particularly poor prognosis. Although treatment success rates and survival of patients with XDR-TBre and XDR-TBor did not differ signicantly in the present study, patients with XDR-TBre

tended to have a poorer treatment success rate than those with XDR-TB(or-re) (29.3 vs. 44%; P 5 0.331), and the success rate of patients with XDR-TB(or-re) was about the same as the 46.3% success rate of those with non-XDR MDR-TB. These ndings indicate that XDR-TBor includes a subgroup of patients with a relatively favorable prognosis (XDR-TB(or-re)). In addition, patients with XDR-TBre showed the worst cumulative survival by Kaplan-Meier analysis, and compared with patients with

Figure 2. Kaplan-Meier survival analysis estimates of long-term survival based on the resistance to either ooxacin or at least one of second-line injectable drugs. Mean survival time was 61.7, 75.7, 78.3, and 91.1 months in patients with XDR-TBre, preXDR-TBo, preXDR-TBs, and the other form of MDR-TB, respectively. Patients with XDR-TBre showed the worst long-term survival (XDR-TBre vs. preXDR-TBo, P 5 0.032; XDRTBre vs. preXDR-TBs, P , 0.001; XDR-TBre vs. other MDR-TB, P , 0.001). Long-term survival of patients with preXDR-TBo and preXDRTBs did not differ (P 5 0.062), but was poorer than that of those with the other form of MDR-TB (preXDR-TBo vs. other MDR-TB, P , 0.001; preXDR-TBs vs. other MDR-TB, P 5 0.041). Black solid line 5 XDR-TBre, black dashed line 5 preXDR-TBo, gray dashed line 5 preXDR-TBs, gray solid line 5 other MDR-TB. MDR 5 multidrug-resistant; TB 5 tuberculosis; XDR 5 extensively drug-resistant; XDR-TBre 5 revised denition of XDR-TB; XDR-TBor 5 original denition of XDR-TB; XDRTB(or-re) 5 XDR-TBor excluding XDR-TBre.

118

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 182

2010

TABLE 5. TREATMENT OUTCOMES BY SM SUSCEPTIBILITY IN PATIENTS WITH pre-XDR-TB

PreXDR-TBo SM-Resistant Treatment Outcomes Cure Treatment completion Short-term treatment completion Failure* Transfer out Default Death Treatment success (n 5 56) 12 5 4 13 4 8 10 21 (21.4) (8.9) (7.1) (23.2) (7.1) (14.3) (17.9) (37.5) SM-Susceptible (n 5 103) 22 4 10 9 8 28 22 36 (21.4) (3.9) (9.7) (8.7) (7.8) (27.2) (21.4) (35) SM-Resistant (n 5 37) 8 3 4 3 5 10 4 15 (21.6) (8.1) (10.8) (8.1) (13.5) (27) (10.8) (40.5) PreXDR-TBs SM-Susceptible (n 5 80) 26 10 4 4 6 21 9 40 (32.5) (12.5) (5) (5) (7.5) (26.2) (11.2) (50) P Value 0.290 0.199 0.605 0.005 0.678 0.280 0.227 0.211

Denition of abbreviations: SM 5 streptomycin; TB 5 tuberculosis; XDR 5 extensively drug-resistant; preXDR-TBo 5 ooxacin-resistant non-XDR MDR-TB; preXDRTBs 5 second-line injectable drug-resistant non-XDR MDR-TB. Values are n (%). Statistical analyses were performed using chi-square test. Bonferroni-corrected P , 0.05. * SM-resistant preXDR-TBo vs. SM-susceptible preXDR-TBs.

non-XDR MDR-TB, the HR for all-cause mortality of those with XDR-TBre was higher than that of patients with XDRTB(or-re) (3.15 vs. 1.38, respectively). Thus, the revised denition of XDR-TB seems to be more appropriate than the original denition in identifying the subgroup with MDR-TB with the worst prognosis. Various drug-resistance patterns, such as resistance to FQNs (5, 6), kanamycin (7), capreomycin (8), SM (9), and additional rst-line drugs (4), have been previously analyzed to identify factors predictive of unsuccessful MDR-TB treatment. We found that treatment success rate and long-term survival did not differ between patients with preXDR-TBo and pre XDR-TBs, and that the long-term survival of both groups with preXDR-TB was poorer than those with OFX-susceptible/ SLID-susceptible MDR-TB. In addition, the high HR values (1.62 and 1.57, respectively) for all-cause mortality in patients with preXDR-TBo and preXDR-TBs showed that such disease status was independently important in determining poor long-term survival in patients with MDR-TB. Thus, the current denition of preXDR-TB also seems to be appropriate for dening an intermediate risk group lying between XDR-TBre and OFX-susceptible/SLID-susceptible patients with MDR-TB in terms of disease severity. Although SM is a rst-line injectable anti-TB drug with a mechanism of action similar to that of other SLIDs, SMresistance has not been included in the current denition of XDR-TB or preXDR-TB (3), and few studies have investigated the impact of SM susceptibility on the outcomes of patients with XDR-TB and preXDR-TB. Although one study found that SM-resistance was associated with adverse treatment outcomes in HIV-negative patients with XDR-TB (9), another study showed that patients with MDR-TB, resistant to both FQN and SM but susceptible to all three SLIDs, had better treatment outcomes than did patients with XDR-TB (18). In the present study, neither treatment outcomes nor long-term survival was affected by SM susceptibility within the group with XDR-TBre. However, when that group was compared with patients with preXDR-TB, SM susceptibility was an important predictor of favorable long-term survival, irrespective of the presence of either OFX- (preXDR-TBo) or SLID-resistance (preXDR-TBs). Collectively, these ndings indicate that SMresistance does not need to be included in the current denition of XDR-TB and SM use may be helpful to improve treatment outcomes in patients with preXDR-TB. Considering the importance of SM susceptibility in MDR-TB treatment, judi-

cious use of SM seems to be prerequisite in improving treatment outcomes in patients with MDR-TB. In accordance with this nding, recently revised WHO guidelines restricted the Category II regimen (inclusion of SM) only for recurrent patients with TB at medium to low likelihood of MDR-TB, because the Category II regimen has a possibility to amplify drug-resistance including SM-resistance in patients with high risk of MDR-TB (21).

Figure 3. Kaplan-Meier survival analysis estimates of long-term survival based on SM-resistance in patients with preXDR-TB. Mean survival time was 73.4, 74.8, 67, and 82.4 months in SM-resistant pre XDR-TBo, SM-susceptible preXDR-TBo, SM-resistant preXDR-TBs, and SM-susceptible preXDR-TBs patients, respectively. Patients with SM-susceptible preXDR-TBs showed more favorable long-term survival than did those with SM-resistant preXDR-TBo (P 5 0.041), SM-susceptible preXDR-TBo (P 5 0.013), or SM-resistant preXDR-TBs (P 5 0.026), respectively. Black solid line 5 SM-resistant preXDR-TBo, black dashed line 5 SM-susceptible preXDR-TBo , gray solid line 5 SM-resistant preXDR-TBs, gray dashed line 5 SM-susceptible pre XDR-TBs. SM 5 streptomycin; TB 5 tuberculosis; XDR 5 extensively drug-resistant; preXDR-TBo 5 ooxacin-resistant non-XDR MDR-TB; preXDR-TBs 5 second-line injectable drug-resistant non-XDR MDR-TB.

Kim, Kim, Park, et al.: Outcomes and Survival in MDR-TB Patients

119
5. Chan ED, Laurel V, Strand MJ, Chan JF, Huynh ML, Goble M, Iseman MD. Treatment and outcome analysis of 205 patients with multidrugresistant tuberculosis. Am J Respir Crit Care Med 2004;169:1103 1109. 6. Migliori GB, Lange C, Girardi E, Centis R, Besozzi G, Kliiman K, Codecasa LR, Spanevello A, Cirillo DM. Fluoroquinolones: are they essential to treat multidrug-resistant tuberculosis? Eur Respir J 2008; 31:904905. 7. Jeon CY, Hwang SH, Min JH, Prevots DR, Goldfeder LC, Lee H, Eum SY, Jeon DS, Kang HS, Kim JH, et al. Extensively drug-resistant tuberculosis in South Korea: risk factors and treatment outcomes among patients at a tertiary referral hospital. Clin Infect Dis 2008;46: 4249. tterlein R, Hoffmann H, 8. Migliori GB, Lange C, Centis R, Sotgiu G, Mu Kliiman K, De Iaco G, Lauria FN, Richardson MD, et al. Resistance to second-line injectables and treatment outcomes in multidrugresistant and extensively drug-resistant tuberculosis cases. Eur Respir J 2008;31:11551159. 9. Hwang SS, Kim HR, Kim HJ, Kim MJ, Lee SM, Yoo CG, Kim YW, Han SK, Shim YS, Yim JJ. Impact of resistance to rst-line and injectable drugs on treatment outcomes in MDR-TB. Eur Respir J 2009;33:581 585. 10. Kim DH, Kim HJ, Park SK, Kong SJ, Kim YS, Kim TH, Kim EK, Lee KM, Lee SS, Park JS, et al. Treatment outcomes and long-term survival in patients with extensively drug-resistant tuberculosis. Am J Respir Crit Care Med 2008;178:10751082. 11. Kim HR, Hwang SS, Kim HJ, Lee SM, Yoo CG, Kim YW, Han SK, Shim YS, Yim JJ. Impact of extensive drug resistance on treatment outcomes in non-HIV-infected patients with multidrug-resistant tuberculosis. Clin Infect Dis 2007;45:12901295. 12. Keshavjee S, Gelmanova IY, Farmer PE, Mishustin SP, Strelis AK, Andreev YG, Pasechnikov AD, Atwood S, Mukherjee JS, Rich ML, et al. Treatment of extensively drug-resistant tuberculosis in Tomsk, Russia: a retrospective cohort study. Lancet 2008;372: 14031409. 13. Kwon YS, Kim YH, Suh GY, Chung MP, Kim H, Kwon OJ, Choi YS, Kim K, Kim J, Shim YM, et al. Treatment outcomes for HIVuninfected patients with multidrug-resistant and extensively drugresistant tuberculosis. Clin Infect Dis 2008;47:496502. 14. Migliori GB, Ortmann J, Girardi E, Besozzi G, Lange C, Cirillo DM, Ferrarese M, De Iaco G, Gori A, Raviglione MC. Extensively drugresistant tuberculosis, Italy and Germany. Emerg Infect Dis 2007;13: 780782. 15. Shah NS, Pratt R, Armstrong L, Robison V, Castro KG, Cegielski JP. Extensively drug-resistant tuberculosis in the United States, 1993 2007. JAMA 2008;300:21532160. 16. Kim DH, Kim HJ, Park SK, Kong SJ, Kim YS, Kim TH, Kim EK, Lee KM, Lee SS, Park JS, et al. Treatment outcomes and survival based on drug resistance patterns in multidrug-resistant tuberculosis [abstract 26]. The 107th conference of the Korean Academy of Tuberculosis and Respiratory Diseases, Seoul, South Korea, Nov. 13, 2008. 17. World Health Organization. Guidelines for the programmatic management of drug-resistant tuberculosis. Geneva, Switzerland: WHO; 2008. 18. Chan ED, Strand MJ, Iseman MD. Multidrug-resistant tuberculosis (TB) resistant to uoroquinolones and streptomycin but susceptible to second-line injection therapy has a better prognosis than extensively drug-resistant TB. Clin Infect Dis 2009;48:e50e52. 19. Bonilla CA, Crossa A, Jave HO, Mitnick CD, Jamanca RB, Herrera C, Asencios L, Mendoza A, Bayona J, Zignol M, et al. Management of extensively drug-resistant tuberculosis in Peru: cure is possible. PLoS ONE 2008;3:e2957. 20. World Heath Organization. Treatment of tuberculosis: guidelines, 4th ed. Geneva, Switzerland: WHO; 2009.

This study has several limitations inherent to retrospective studies, as already described (10). Because DST results for amikacin and FQNs other than OFX were not obtained, the denition of XDR-TB used in this study was suboptimal and the impact of FQN cross-resistance on treatment outcomes was not evaluated. Long-term survival was analyzed only for allcause mortality, not for TB-related mortality, because of the recently amended rules of the Korea National Statistical Ofce. In addition, the high default rate may have affected the accuracy of treatment outcome analysis. However, despite these limitations, the large number of patients studied and the careful assessment of fatal outcomes facilitate accurate subset analysis.
Conclusion

We found that the revised denition of XDR-TB was optimal for dening the subset of patients with MDR-TB with the worst treatment outcomes and long-term survival. PreXDR-TB, irrespective of OFX- or SLID-resistance, was independently associated with poor outcomes in patients with MDR-TB, and SM-resistance affected treatment outcomes of patients with preXDR-TB.
Conict of Interest Statement: D.H.K. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. H.J.K. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. S-K.P. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. S-J.K. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. Y.S.K. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. T-H.K. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. E.K.K. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. K.M.L. received up to $1,000 from Pzer Korea, up to $1,000 from Bayer Korea, and up to $1,000 from LG Life Sciences in lecture fees. S-S.L. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. J.S.P. received up to $1,000 from Pzer Korea, up to $1,000 from Bayer Korea, and up to $1,000 from LG Life Sciences in lecture fees. W.J.K. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. C-H.L. does not have a nancial relationship with a commercial entity that has an interest in the subject of this manuscript. T.S.S. received up to $1,000 from Pzer Korea, up to $1,000 from Bayer Korea, and up to $1,000 from LG Life Sciences in lecture fees.

References
1. Wright A, Zignol M, Van Deun A, Falzon D, Gerdes SR, Feldman K, Hoffner S, Drobniewski F, Barrera L, van Soolingen D, et al. Epidemiology of antituberculosis drug resistance 200207: an updated analysis of the global project on anti-tuberculosis drug resistance surveillance. Lancet 2009;373:18611873. 2. Centers for Disease Control and Prevention. Emergence of Mycobacterium tuberculosis with extensive resistance to second-line drugsworldwide, 20002004. MMWR Morb Mortal Wkly Rep 2006;55:301305. 3. World Health Organization. Extensively drug-resistant tuberculosis (XDR-TB): recommendations for prevention and control. Wkly Epidemiol Rec 2006;81:430432. 4. Migliori GB, Besozzi G, Girardi E, Kliiman K, Lange C, Toungoussova OS, Ferrara G, Cirillo DM, Gori A, Matteelli A, et al. Clinical and operational value of the extensively drug-resistant tuberculosis denition. Eur Respir J 2007;30:623626.