The study prospectively enrolled 2 cohorts of consecutively seen, consenting, nonimprisoned adult patients in Arkhangelsk Oblast, Russia, who had confirmed pulmonary MDR TB and were starting treatment with second-line drugs. MDR TB was confirmed by sputum culture and drug-susceptibility testing (DST) at the regional TB laboratory. Patients in cohort 1 were enrolled from January 1, 2005, through December 31, 2006; patients in cohort 2 were enrolled from January 1, 2007, through December 31, 2008. The 2 cohorts were approved in 2 separate applications to the GLC. Most patients in cohort 1 had been on a waiting list for MDR TB treatment for an extended amount of time at the time of treatment initiation, whereas most patients in cohort 2 had a recent diagnosis of MDR TB at the time of treatment initiation. Study inclusion criteria required having >1 M. tuberculosis–positive culture result within 1 month (before or after) of starting second-line drugs for the treatment for MDR TB (baseline isolate) and >1 month of treatment with second-line drugs.

Standardized forms were used to prospectively collect sociodemographic, clinical, and laboratory data from patients’ medical charts. Chest radiographs were read by experienced chest physicians and radiologists, and results were recorded in a standardized manner. Sputum specimens were collected from each patient at the start of second-line drug treatment (baseline isolate) and then monthly until treatment outcome was known.

The study protocol was approved by ethics committees at the US Centers for Disease Control and Prevention (CDC), Northern State Medical University in Arkhangelsk, and the State Research Center for Applied Microbiology and Biotechnology (SRCAMB) in Obolensk, Russia. All patients provided written informed consent.

Baseline and follow-up sputum specimens were cultured on Lowenstein-Jensen solid media in the Arkhangelsk Regional TB Dispensary Laboratory. Frozen M. tuberculosis isolates were shipped to SRCAMB in Obolensk, Russia, for first- and second-line DST, genotyping, and DNA sequencing. The analysis reported in this article is based on the SRCAMB results. Testing at SRCAMB was conducted months to years after patients were enrolled, and results were not available in real time.

At SRCAMB, each isolate was cultured in 6 mL of Middlebrook 7H9 broth to an optical density of >1.0 McFarland standard and on Lowenstein-Jensen medium. Susceptibility testing for the baseline and the last follow-up (final) isolates from each patient were determined for isoniazid, rifampin, ethambutol, streptomycin, kanamycin, amikacin, capreomycin, ofloxacin, ethionamide, and para-aminosalicylic acid. Drug susceptibility was determined by the proportion method according to CDC protocol (19). When drug susceptibility of a patient’s baseline and final isolates differed, isolates were genotyped by mycobacterial interspersed repetitive units–variable number of tandem repeats analysis and by restriction fragment-length polymorphism–IS6110 analysis to determine whether the isolates were the same strain.

Definitions of pulmonary TB, MDR TB, and treatment outcomes were based on WHO guidelines (20). Third-line drugs refer to drugs classified by WHO as group 5 drugs (1). “Effective treatment” was defined as treatment with a drug or combination of drugs to which baseline DST reported susceptibility. “Ineffective treatment” was considered use of said drug(s) despite reported resistance. Effectiveness of treatment was considered unknown and the patient was not included in the analysis when the patient never received said drugs or baseline DST results were not available. “Acquired resistance” was defined as occurring when baseline DST result showed susceptibility in vitro, the final DST result showed resistance in vitro, and genotypes matched for the initial and final isolates. Acquired resistance was considered absent in each of the following 3 scenarios: 1) baseline and final isolates were susceptible to a drug; 2) DST result changed from susceptible to resistant, but genotyping indicated different strains; or 3) no follow-up positive culture results were available because the patient’s sputum culture results sustainably converted to negative after the baseline DST, the patient died, or the patient was lost to follow-up. In each instance, the denominator for each group refers to the number of isolates with baseline DST results indicating susceptibility to the given drug. Patients whose baseline isolate was resistant to a given drug were excluded from the acquired resistance analysis for that drug. Successful treatment outcome was defined as cure and treatment completion (20). Poor treatment outcome was defined as treatment failure or death. The second-line companion drugs were para-aminosalicylic acid or ethionamide. Being underweight was defined as having a body mass index <18.5 kg/m².

The data from standardized forms were double-entered into an Epi Info (CDC, Atlanta, GA, USA) database in Arkhangelsk and sent to CDC for checking and analysis. Laboratory data were sent directly from the SRCAMB laboratory to CDC.

Statistical analyses were performed by using SAS version 9.3 (SAS Institute Inc., Cary, NC, USA). The main outcome of interest was acquired resistance to specific second-line drugs. The secondary outcome of interest was end-of-treatment outcome. Bivariate associations between the potential risk factors and the outcome variable for each respective analysis were examined by using the Fisher exact test with a significance level of 0.05. Multivariate logistic regression was used to assess associations of acquired resistance and treatment with treatment outcomes while controlling for the potential confounding effects of extent of drug resistance at baseline, disease severity, and previous treatment for MDR TB.

A total of 202 MDR TB patients were enrolled in the study: 81 in cohort 1 and 121 in cohort 2. Median patient age was 42 years, 171 (84.7%) patients were male, and none were HIV infected (HIV test results were available for all patients). Most patients had previously received treatment for TB: 69 (34.7%) had received first-line drugs and 73 (36.7%) had received additional second-line drugs. Almost all patients (189 [93.6%]) had pulmonary cavities, 162 (80.2%) had bilateral lung involvement, 161 (80.9%) had sputum smears with acid-fast bacilli seen with microscopy, and 45 (22.3%) had a body mass index <18.5 kg/m² at the start of MDR TB treatment.

A total of 740 M. tuberculosis isolates from 202 patients were shipped to SRCAMB. Of these 202 patients, baseline DST results from SRCAMB were available for 171 (84.7%) and were included in analysis of baseline drug resistance (Figure). Of the 171 patients for whom baseline DST results were available, follow-up DST results at SRCAMB were available for 117. Among the other 54 patients (without a final DST result), cultures converted after the initial isolate for 45, follow-up cultures were contaminated or did not grow for 5, and a reason was not documented for 4.

Among 171 patients for whom baseline isolate DST results were available (Table 1), MDR TB was not confirmed by DST for 4 (2.3%) isolates, and 130 (76.0%) baseline isolates were resistant to 4 first-line drugs tested (rifampin, isoniazid, ethambutol, and streptomycin). In addition, 74 isolates were resistant to >1 of the 3 second-line injectable agents: 72 (42.1%) to kanamycin, 30 (17.5%) to amikacin, 13 (7.6%) to capreomycin, and 10 (5.8%) to all 3. A total of 10 (5.8%) isolates were resistant to ofloxacin, and 7 (4.1%) were XDR. Of the second-line companion drugs tested, resistance to ethionamide was found for 46 (26.9%) isolates and to para-aminosalicylic acid for 54 (31.6%) isolates.

Among 117 patients for whom final DST results from SRCAMB were available, results for the baseline and final isolates differed for 32 (27.3%) and the isolates were successfully genotyped. Of these, genotype results for isolate pairs matched for 17 (53.1%) patients and were eligible for inclusion in numerators for respective analyses of acquired drug resistance (Figure). Of 41 paired isolates with baseline susceptibility to >1 first-line drug, resistance to a first-line drug was acquired in 3 (7.3%) (Table 1). Of 161 paired isolates with baseline susceptibility to >1 second-line injectable agents, resistance to >1 second-line injectable agents was acquired in 7 (4.4%): resistance to kanamycin by 4 (4.0%) of 99, to amikacin by 1 (0.7%) of 141, and to capreomycin by 3 (1.9%) of 158. Resistance to ofloxacin was acquired by 6 (3.7%) of 161, and XDR TB was acquired by 4 (2.4%) of 164 over the course of treatment for MDR TB.

Acquired resistance to capreomycin was significantly associated with receiving the following drugs or drug groups: <3 effective drugs (p = 0.008), an ineffective fluoroquinolone (p = 0.009), or ineffective para-aminosalicylic acid (p = 0.02). Furthermore, acquired resistance to capreomycin was associated with not having received ofloxacin (regardless of baseline DST results) (p = 0.003), baseline resistance to ofloxacin (p = 0.008), and baseline resistance to para-aminosalicylic acid (p = 0.03) (Table 2). In addition, patients whose isolates acquired resistance to capreomycin were more likely to have received moxifloxacin (instead of ofloxacin) in the treatment regimen.

Acquired resistance to ofloxacin was significantly more common among patients who were underweight (p = 0.02) (Table 3). Patients with acquired ofloxacin resistance were more likely to have received moxifloxacin (p = 0.006), to have had fluoroquinolones switched during treatment (p = 0.05), and to be receiving a third-line drug during the current episode (p = 0.01).

Acquired XDR TB was more frequent among those receiving <3 effective drugs than among those receiving >3 effective drugs (p = 0.03) and among those who were underweight (p = 0.03) (Table 4). Those who acquired XDR TB were more likely to be receiving moxifloxacin during the current episode (p = 0.02). Patients in whom isolates acquired resistance to any second-line companion drug (ethionamide or para-aminosalicylic acid) were less likely to have received ofloxacin (p = 0.003) and more likely to be receiving moxifloxacin (p<0.001) during the current episode (Table 5).

Of 171 patients for whom baseline DST results were available, treatment was successfully completed for 94 (55.0%), treatment failed for 18 (10.5%), 20 (11.7%) died, and 39 (22.8%) defaulted from treatment. Poor treatment outcomes (treatment failure or death) were more likely among patients whose MDR TB acquired resistance to capreomycin (100% vs. 25.9%; p = 0.02) or ofloxacin (83.3% vs. 22.7%; p = 0.004) or became XDR TB (100% vs. 24.4%; p = 0.004) than among those in whom the respective resistance was not acquired (Table 6). Patients who received an effective fluoroquinolone were statistically less likely to have poor treatment outcomes than were those who received an ineffective fluoroquinolone (25.6% vs. 85.7%; p = 0.002). Patients who received any third-line drug were more likely to have previously received treatment for MDR TB (21.6% vs. 8.4%; p = 0.02), have resistance to >4 drugs at baseline (72.7% vs. 47.0%; p<0.001), and experience treatment failure or die (42.0% vs. 20.6%; p = 0.01) than those who did not receive any third-line drug. According to multivariable analysis, compared with no acquired resistance, acquired resistance to ofloxacin was associated with 10.2-fold (95% CI 1.1–95.1) increased odds of poor outcome when confounding was controlled for (Table 6). Compared with not receiving a third-line drug, treatment with a third-line drug was associated with 2.7-fold (95% CI 1.2–5.7) increased odds of poor treatment outcome when confounding was controlled for. Compared with not receiving effective fluoroquinolone treatment, effective treatment with a fluoroquinolone was associated with 16.7-fold (95% CI 1.9–100.0) increased the odds of successful treatment outcome when confounding was controlled for.