A total of 1,659 patients were enrolled in PETTS strictly according to protocol. For 1,254 (75.6%) of these, a baseline M. tuberculosis isolate was shipped to, and DST results were available from, CDC. A total of 64.1% of patients were male, and the median age of all patients was 37 (interquartile range [IQR] 28-47) years. HIV co-infection was confirmed in 159 (12.7%) patients, of whom 135 (85%) were in 1 country. A total of 168 (13.4%) patients had new MDR TB; 879 (70.1%) previously received first-line anti-TB drugs; 185 (14.8%) received second-line drugs in the past; and 22 (1.8%) had unknown treatment history. DST was performed in local laboratories in countries to a median of 6 (IQR 4–8) drugs. Overall treatment outcomes were as follows: cure, 659 (52.5%); completion of treatment, 66 (5.2%); treatment failure, 80 (6.4%); death, 170 (13.6%); loss to follow-up, 235 (18.7%); and unknown, 44 (3.5%).

Of the 1,254 patients, 952 (75.9%) had no resistance to fluoroquinolones and second-line injectable drugs and would qualify as candidates for the short regimen on the basis of the drug resistance indication. On the other hand, 302 (24.1%) patients had resistance to a fluoroquinolone or second-line injectable drug and would qualify as candidates for a bedaquiline-containing regimen. The proportion of patients with no baseline resistance to fluoroquinolones and second-line injectable drugs varied among countries from 47.2% to 92.5%; the proportion of patients with pre-XDR or XDR varied from 7.5% to 52.8% (Table 1).

We assessed baseline resistance to individual drugs among potential candidates for the 9-month short regimen and for a bedaquiline-containing regimen and stratified the results by resistance to fluoroquinolones and second-line injectable drugs (Table 2). Susceptibility to high-dose isoniazid was observed in 8.6% of candidates for the short regimen and 3.0% of candidates for a bedaquiline-containing regimen (p<0.001), susceptibility to ethambutol in 41.9% and 27.8% (p<0.001), susceptibility to ethionamide in 81.9% and 80.1% (p = 0.48), and susceptibility to para-aminosalicylic acid in 95.0% and 76.2% (p<0.001), respectively.

Overall, a median of 5 (IQR 5–6) drugs remained effective in candidates for the short regimen and 3 (IQR 2–4) drugs in candidates for a bedaquiline-containing regimen (Table 3). However, in XDR TB cases, a median of only 2 (IQR 1–2) drugs were effective, whereas in pre-XDR cases, a median of 4 (IQR 2–4) drugs were effective (p<0.001). The initial isolate was susceptible to >3 drugs in all of the candidates for the short regimen and in 73.8% (223/302) of candidates for a bedaquiline-containing regimen. Among candidates for a bedaquiline-containing regimen, 90.4% (206/228) of pre-XDR patients had susceptibility to >3 drugs, compared with 23.0% (17/74) of XDR patients (p<0.001).

Several patient characteristics were significantly associated with higher probability of resistance to fluoroquinolones or second-line injectable drugs (and thus candidacy for bedaquiline-containing regimen). These characteristics were unemployment, history of imprisonment, alcohol abuse, history of treatment with second-line drugs, and pulmonary cavities (Table 4).

Overall treatment success among candidates for a short regimen was 66.1% (95% CI 63.0%–69.1%), varying from 90% to 50% among countries. Among candidates for a bedaquiline-containing regimen, treatment success was 39.9% (95% CI 34.3%–45.6%), varying from 90% to 10% among countries.

Data from our large prospective observational cohort study of 1,254 MDR TB patients showed that about three fourths would qualify as candidates for the short regimen and about one fourth would qualify as candidates for a bedaquiline-containing regimen on the basis of the drug resistance indication. DST demonstrated susceptibility to a median of 5 (IQR 5–6) drugs among candidates for the 9-month regimen, whereas among candidates for bedaquiline, a median of 3 (IQR 2–4) drugs remained effective. Overall treatment success was 66% for candidates for the short regimen and 40% for candidates for a bedaquiline-containing regimen. This analysis has several practical implications for national TB control programs that plan to start using short regimens or bedaquiline for programmatic management of drug-resistant tuberculosis.

First, data from this report may assist national TB control programs in planning the number of patients to be enrolled for short or bedaquiline-containing regimen. Not all countries have representative data from routine drug resistance surveillance or surveys; thus, data from this report can be used to estimate numbers of candidate patients for certain regimens. The 2013 WHO report included combined representative data from 75 countries and 4 territories showing that the 32% of persons had MDR TB resistant to a fluoroquinolone, a second-line injectable drug, or both, and thus these persons might be eligible to receive bedaquiline (1). Our data showed a relatively similar proportion (26%). The remaining 74% of patients could be considered as candidates for the short regimen. However, the proportions of patients in our study who might qualify for the 9-month or bedaquiline-containing regimen varied widely among countries: 47%–93% and 8%–53%, respectively. This variation in baseline drug resistance probably reflected previous use of anti-TB second-line drugs in these countries and the consequent epidemiology of MDR TB (13).

Second, we report on the prevalence of resistance to first- and second-line drugs by the status of resistance to fluoroquinolones or second-line injectable drugs. Candidates for the short regimen had, on average, susceptibility to 5 anti-TB drugs. We did not have DST for pyrazinamide and clofazimine (which are part of the 9-month regimen). Susceptibility to high-dose isoniazid, also included in the short regimen, was found in only 8.6% of candidates for that regimen. To our knowledge, only 1 randomized clinical trial has documented improved interim treatment outcomes (reduced time to sputum culture conversion and higher proportion with sputum culture negative 6 months after treatment started) among patients receiving high-dose isoniazid (16–18 mg/kg) as an adjuvant to second-line drugs in documented MDR TB (15). More clinical research is needed about use of high-dose isoniazid to treat MDR TB (16). Susceptibility to ethambutol was found in 42% and to thioamides in 82% of candidates for the short regimen. On the other hand, candidates for a bedaquiline-containing regimen frequently had resistance to other first- and second-line drugs with a median of 3 drugs that remained effective; the number of effective drugs progressively decreased from 3 or 4 in patients with pre-XDR to 2 drugs in XDR TB. Therefore, countries planning to implement bedaquiline for programmatic use should include plans to procure third-line drugs, such as linezolid and clofazimine. These additional drugs are needed to build regimens with the minimum 4 effective drugs because of the extent of resistance to other second-line drugs among patients who might qualify for bedaquiline. Alternatively, this study’s results can be used to infer what might happen if bedaquiline were introduced solely on the basis of the result of fluoroquinolone or second-line injectable drug resistance. Without the capacity for full-spectrum DST, widespread acquired bedaquiline resistance would be expected quickly.

Finally, when countries implement the short regimen or bedaquiline under programmatic conditions, a comparator arm is unlikely to be evaluated to determine the effectiveness of the regimens. Patients placed on a standard WHO MDR TB regimen will not be an appropriate control group to make valid comparisons. In this study, overall treatment success was 66% among candidates for the short regimen and 40% among candidates for a bedaquiline-containing regimen. A meta-analysis of the 18- to 24-month regimen outcomes reported by Falzon et al. (17) included data from 26 treatment centers and reported treatment success of 64% among 4,763 patients with MDR TB and no additional resistance to fluoroquinolones or second-line injectable drugs (this group of patients would be most comparable to candidates for the short regimen) and 40%–56% among patients with additional resistance to second-line injectable drugs or fluoroquinolones and with XDR TB (similar to candidates for bedaquiline). Thus, the proportions of patients with treatment success in the respective groups defined by drug resistance status can be used as historical comparators to assess the effectiveness of treatment in cohorts on the short regimen or the bedaquiline-containing regimen. Also, these studies clearly demonstrate the tremendous capacity for improvement of MDR TB treatment outcomes; new drugs and regimens are desperately needed.

Our report emphasizes the crucial need for strong laboratory capacity to manage drug-resistant TB. A few years ago, the Xpert MTB/RIF assay (Cepheid Inc., Sunnyvale, CA, USA) for detecting M. tuberculosis and rifampin resistance became widely available. Many countries are adopting this diagnostic tool and thus increasingly diagnosing MDR TB. However, for implementation and programmatic use of short and bedaquiline-containing regimens, national TB control programs would need capacity for rapid, accurate DST for both first- and second-line drugs to appropriately select candidates for the respective regimens and ensure the availability of at least 3 other effective drugs. This need should be considered during planning of the implementation of these new regimens.

Our study findings are subject to several major limitations in interpretation and applicability to programs. This study did not evaluate the frequency of drug-related toxicities leading to permanent discontinuation of fluoroquinolones or second-line injectable drugs. These patients would be candidates for bedaquiline-containing regimens even if their isolates remained susceptible to these drugs. Thus, we might have underestimated the proportion of patients eligible for bedaquiline. DST results were not available for pyrazinamide by the time of this analysis. On the basis of published data, pyrazinamide resistance in patients with MDR TB in countries included in PETTS varied from 49% in Thailand, 52% in South Africa, and 55% in Taiwan to 85% in South Korea (18–21). Thus, our study might overestimate the proportion of patients with MDR TB who would respond to the 9-month regimen because >50% of isolates could be resistant to pyrazinamide. CDC did not test for moxifloxacin when this work was conducted because standardized procedures had not yet been established. CDC does not test for clofazimine because standardized procedures using currently available technology have not been established. CDC does not test for cycloserine, which requires Lowenstein-Jensen medium. The intrinsic reproducibility of phenotypic DST for several drugs tested routinely including ethambutol, thioamides, cycloserine, and para-aminosalicylic acid is poor. All patients in the PETTS study were treated under WHO-recommended regimens during 2005–2010, when new treatment options were not available; thus, treatment outcomes might not be able to be extrapolated to new regimens.

Current standards for treating patients with MDR TB pose at least 2 major problems that reduce the effectiveness of treatment and treatment success rates. First, the long duration of treatment—a total of at least 20 months—places a major burden on patients and health care systems. Second, common serious adverse drug reactions contribute to reduced adherence or suspension of treatment. Given the global prevalence of MDR TB and low treatment success rates, better TB drugs and shorter regimens are urgently needed to enable more effective, less toxic, and less expensive treatment for persons with MDR TB.