Multidrug-Resistant Tuberculosis in Europe, 2010–2011

Ongoing transmission, high levels of drug resistance, and poor diagnostic

MDR TB living in this region may be much higher because a substantial proportion of patients are never screened for drug-resistant TB before starting treatment, partly because of a lack of diagnostic capacity (3).
MDR TB is associated with poor treatment outcomes (1,2,4). The proportion of treatment success in patients with MDR TB was only 54% in an individual patient data metaanalysis of >9,000 patients from 32 observational studies (5). Results from this cohort showed that additional resistance to fluoroquinolones in patients with MDR TB reduced treatment success to 48%; patients with XDR TB were treated successfully in 40% of cases (6), which approached treatment outcomes similar to those of the preantimicrobial drug era (4). A recent surveillance report from the EU reported 32.2% treatment success for MDR TB and 19.1% treatment success for XDR TB (7).
Detailed information about characteristics, management, and outcomes of patients with MDR TB in Europe is scarce but essential to inform health policy makers and optimize disease management (8). We compared baseline characteristics and risk factors for patients with MDR TB, as well as availability and results of drug susceptibility testing (DST) for second-line drugs for treatment of TB, in a cohort of patients from 16 countries in Europe with low, intermediate, and high incidence of TB, who had started first-line or second-line TB treatment.

Participating Sites
TBNET is a European consortium for clinical research in the field of TB (9). This study was conducted at 23 TBNET sites in 16 countries in Europe: 2 with a high (>100 cases/100,000 persons) incidence of TB, 4 with an intermediate (20-100 cases/100,000 persons) incidence, and 10 with a low (<20 cases/100,000 persons) incidence (Figure). Stratification was based on WHO/European Centre for Disease Prevention and Control estimates of TB incidence during the study period (2010-2011) (10). The number of study participants per country is shown in online Technical Appendix Table 1.

Study Population
After informed consent was obtained, patients starting treatment for a new episode of culture-confirmed TB with resistance to at least rifampin and isoniazid (MDR TB cohort) were eligible for enrollment. Patients were included prospectively by using consecutive inclusion during January 2010-December 2011 at each site. In Belarus, Latvia, Moldova, and Romania, additional enrollment was stopped when a preagreed number of patients were enrolled to avoid overrepresentation of patients from these countries in the cohort.
For each MDR TB patient, 1 patient with non-MDR TB (pan drug-susceptible, monoresistant, or polydrugresistant TB [11]) was enrolled at each center at the time of enrollment of the MDR TB patient: these additional patients were denoted as controls. Controls were selected on the basis of DST results that identified non-MDR TB, and that were obtained at the closest date before enrollment of an MDR TB patient at the same site.
Because of this selection process, a limited number of patients (41, 5.4%) started treatment before the study began in January 2010, but none started treatment before Figure. TBNET study sites in the Pan European network for study and clinical management of drug-resistant tuberculosis (TBPAN-NET) project. Stratification is based on the incidence of tuberculosis (TB) reported during 2010-2011, which matched the inclusion period of the study. Data for 2011 were obtained from the European Centre for Disease Control and Prevention (10). Low TB incidence, <20 cases/100,000 persons; intermediate TB incidence, 20-100 cases/100,000 persons; high TB incidence, >100 cases/100,000 persons. January 2007. However, we maintained consecutive inclusion for MDR TB patients. This feature ensured an acceptable sample size for countries with a low incidence of TB during the inclusion period.

Data Collection
Data collection used an electronic case record form designed in Open Clinica (http://www.openclinica.com). A paper version of this form was used in Moldova, Romania, Estonia, and Belarus, where internet access was not always available. All investigators were initially trained onsite, and continuous training was ensured through annual investigator meetings, regular site visits, and newsletters.

Laboratory Testing
Routine data were obtained from local laboratory reports for sputum smear microscopy, sputum culture, and DST for first-line and second-line drugs and, when available, M. tuberculosis-specific nucleic acid amplification tests. All laboratories at study sites were subjected to quality control through the WHO Supranational Reference Laboratory Network.

Study Outcome
We analyzed characteristics of the cohort at the time of enrollment. We also assessed factors associated with MDR TB in a cross-sectional approach.

Data Management
Data management included regular data checks on key variables for missing data and inconsistencies. The study coordinator, a study monitor, and a trained study nurse performed routinely manual plausibility checks and clarified inconsistencies with the investigators.

Statistical Analysis
Descriptive statistics are reported as frequencies or medians, where appropriate. Risk factor analysis was performed by using univariable and multivariable logistic regression. We used robust SEs to adjust for clustering by country. All variables with <20% missing data were assessed for inclusion in the models. Missing data for included variables were coded as separate categories to prevent listwise deletion of observations. Age was dichotomized at 45 years to align with values in a previous study (12). The variables age and sex were purposefully included in the multivariable analysis in which other variables were included on the basis of the Wald statistic (<0.1) and the change in the pseudo R 2 (>10%) because a robust SE precludes formal use of the log-likelihood ratio test. In a sensitivity analysis, we repeated multivariable logistic regression with the inclusion of a sampling weight for the MDR TB patients (inverse of the sampling fraction with expected number of MDR TB patients in the country as denominator) (online Technical Appendix Table 5). Non-MDR TB patients were given a weight of 1. The weighted analyses assessed the potential effect of unbalanced contribution of countries in the cohort. Goodness-of-fit was assessed by using the F-adjusted mean residual test.
Drug resistance was expressed as the proportion of isolates tested and the proportion of isolates that were resistant. Corresponding frequencies when applying sampling weights and analysis by a complex survey approach (13,14) are shown in online Technical Appendix Table 4.

Ethics
Patient information and consent forms were translated into local languages when needed. The study was approved by the Ethics Committee of the University of Lübeck (Lübeck, Germany). The study protocol was approved by the local ethics committee at all participating centers. Written informed consent was obtained from all patients according to site-specific regulations. Data were collected pseudonymously and stored on a secured server.

Cohort Characteristics
The cohort consisted of 380 MDR TB patients and 376 non-MDR TB controls. Descriptive characteristics of the MDR TB cohort are shown in Table 1 (http://wwwnc.cdc.gov/EID/ article/21/3/14-1343-T1.htm) and those for the non-MDR TB cohort in online Technical Appendix Table 2. Both groups had predominantly male patients. The median age was 36 years (interquartile range 27-47 years) for the MDR TB patients and 41 years (interquartile range 31-54 years) for the controls. The proportion of foreign-born patients with MDR TB in countries of low, intermediate, and high TB incidence was 85.4%, 5.8% and 0.5%, respectively. Similar proportions were observed in controls (56.3%, 5.7% and 2.1%, respectively). Of 94 foreign-born patients, 60 (64%) were from countries of the European region of WHO, 17 (18%) from Russia, 18 (19%) from Southeast Asia, 11 (12%) from sub-Saharan Africa, 1 (1%) from North Africa, and 4 (4%) from South America. Smoking was common in both groups (50.5% for MDR TB patients and 41.5% for controls).
HIV infection and diabetes mellitus were infrequently observed: 6.6% in MDR TB patients and 4.3% in controls for HIV, and 4.2% in MDR TB patients and 5.3% in controls for diabetes mellitus. The percentage of patients with MDR TB whose episode of active TB was their first was 52.4% (59.2%, 74.4%, and 38.7% in countries with low, intermediate, and high TB incidence, respectively).

Drug Resistance Profiles
Among 380 patients with MDR TB, second-line M. tuberculosis DST profiles were available for 356 patients.
Reasons for unavailable baseline DST results were 1) an initial diagnosis of MDR TB at a peripheral hospital and subsequent patient transfer to a central hospital where M. tuberculosis could not be grown in culture (n = 6); 2) contamination of cultures (n = 12); 3) insufficient growth in cultures (n = 4); 4) patient death between the first and second cultures (n = 1), and 5) unknown reason (n = 1). Among patients with MDR TB, 6.8% of cases fulfilled the definition of XDR TB. Drug resistance profiles for first-line and second-line drugs other than rifampin and isoniazid are shown in Table 2 for the MDR TB cohort, in online Appendix Table 3 for the MDR TB cohort compared with the non-MDR TB cohort, and in online Technical Appendix Table 4 Table 2). The weighted analysis showed higher proportions of resistance to all drugs, except capreomycin, moxifloxacin, and ethionamide/prothionamide (online Technical Appendix Table 4).

Risk Factors for MDR TB
Risk factors for TB were compared between patients with MDR TB and controls. Previous treatment for TB (odds ratio 10.7, 95% CI 7.3-15.6) and age <45 years (OR 1.90, 95% CI 1.23-2.93) were identified as independent risk factors for MDR TB by multivariable analysis (Table 3). There was also a moderate association for sex and current homelessness with MDR TB by weighted analysis (online Technical Appendix Table 5). No association was found between MDR TB and abnormal body mass index (<18 or >25), employment status, birth in a foreign country, history of imprisonment, injectable drug use, co-infection with HIV, or diabetes. The role of TB contact was not evaluated because data were not sufficiently robust because of a high percentage of unknown/unreliable results for self-reporting. Weighted analyses showed similar results with only minor differences in effect size.

Discussion
We studied a multicenter cohort of patients with MDR TB at 23 referral centers across Europe and found high rates of drug resistance to second-line drugs for treatment of TB in circulating M. tuberculosis strains, and limited availability of second-line drug resistance testing in several countries with a high incidence of TB. Furthermore, we found evidence of ongoing transmission of MDR strains of M. tuberculosis in eastern Europe: 52.4% of patients with MDR TB were experiencing their first episode of TB. In countries in western Europe with a low incidence of TB, MDR TB is predominantly a disease of immigrants (15), which reflects the epidemiology of MDR TB in the country of origin. Only a few (8.9%) MDR TB patients were born outside the European region of WHO. Thus, interventions for the control of MDR TB should be specific for countries with high incidence of MDR TB, especially in eastern Europe (16). Mathematical and epidemiologic models indicate that early diagnosis, effective treatment, and improved access to laboratory infrastructure could have a strong effect on the incidence of MDR TB in high-prevalence regions (17). However, few of such programmatic requirements are met at many sites in Europe at the present time (18).
Possible active transmission of strains causing MDR TB, as reflected by the large proportion of patients never having received TB treatment before in this European cohort, is consistent with recently reported data and deserves attention. A drug resistance survey conducted in Belarus in 2011 showed that 32.3% of new TB infections and 75.6% of previously treated TB infections had an MDR strain of M. tuberculosis (19). In Moldova, for which adequate surveillance data are available, 23.7% of new TB cases involve an MDR strain (3). A recent report of surveillance data in countries with >700 estimated MDR TB cases per year indicated that more than half of the reported pulmonary MDR TB cases were new cases (20). More than 90% of strains from MDR TB patients had undergone DST for ≥1 second-line injectable drug and fluoroquinolone. The role of ethambutol and pyrazinamide for treatment of MDR TB is unclear. In our cohort, 97.6% and 45.0% of MDR TB strains were tested for resistance to ethambutol and pyrazinamide, respectively. In countries with a high incidence of TB, only 5.2% of MDR TB cases were tested for pyrazinamide resistance because of limited availability of liquid culture methods and special pH media requirements for pyrazinamide DST. Less than half of the strains tested were susceptible to these drugs. Currently, the mechanism of action of pyrazinamide in combination therapy and the relevance of in vitro DST for pyrazinamide are uncertain. Findings from this study raise questions about a universal recommendation to treat MDR TB with pyrazinamide throughout the entire course of treatment (21).
In our study cohort, 1 of 3 M. tuberculosis strains with resistance to at least rifampin and isoniazid were also resistant to protionamide/ethionamide, 1 of 4 were resistant to any second-line injectable drug, and 1 of 5 were resistant to a fluoroquinolone. Of all MDR TB cases, 6.8% fulfilled the definition of XDR TB. Surveillance data from the European Centre for Disease Prevention and Control indicated that 9.1% of cases of XDR TB in patients with MDR TB underwent second-line DST. Given the high proportion of strains that received a second-line DST, it is unlikely that these percentages are overstated because of preferred testing of patients at high risk for acquiring TB.
Our results are consistent with those from the Preserving Effective TB Treatment Study (PETTS) (22), which investigated second-line drug resistance in strains from 1,278 patients in 8 countries, including Latvia and Estonia, which were countries with study sites in this cohort. The main difference between PETTS study and ours was a high frequency of M. tuberculosis resistance to prothionamide/ethionamide in our cohort, which reflected the relatively higher frequency of treatment with this drug combination in eastern Europe than in other parts of the world (23). Recently published data from the PETTS study showed an increased risk of acquiring resistance to second-line drugs during treatment and increased baseline resistance (24). Increased resistance to second-line drugs is associated with higher proportions of treatment failures (6). It can be assumed, if one considers the findings from the PETTS study, that many of the patients in our cohort are at high risk for treatment failure.
Of particular concern is resistance to fluoroquinolones because these drugs are the core of new treatment regimens (25,26), including regimens for patients with drug-susceptible strains of M. tuberculosis (26). In our study, the capacity to perform DST for later-generation fluoroquinolones (levofloxacin and moxifloxacin) was only present for 19.2% of strains for levofloxacin and 8.4% of strains for moxifloxacin. Later generations of fluoroquinolones may still be effective for treatment of MDR TB in some patients when drug resistance to ofloxacin is documented (27). The capacity to perform DST for later generations of fluoroquinolones needs to be improved in the region.
Multivariable analysis showed that previous TB treatment and patient age <45 years showed an association with MDR TB (male sex and current homelessness showed an association in a weighted model). However, none of the other traditional risk factors for drug-resistant TB, such as HIV infection or birth in a foreign country (12), showed this association. Although previous treatment for TB and contact with persons infected with drug-resistant strains have been reported as strong risk factors for MDR TB, the role of HIV infection, young age, sex, and previous imprisonment are less clear (12,28). The high proportion of new cases and the lack of association of other traditional risk factors with drug-resistant TB suggest an increased role of ongoing transmission in the community outside established risk groups for becoming infected with drug-resistant strains of M. tuberculosis (19,20,29).
Our study has several major limitations. First, baseline data were obtained from an observational cohort study and were not derived from routine surveillance. Only 14 of 28 countries from the European Union and 2 countries outside the European Union were represented in the study. Site selection was based on voluntary participation in the study and being a center for the management of MDR TB. Because a high number of patients in Europe are being treated outside such centers, the generalizability of data might be limited. However, the included centers adhered to national policies regarding diagnosis and treatment of MDR TB patients and therefore reflect current practice. To provide a better estimation of representativeness of data for the situation in Europe, we additionally performed weighted analyses based on the sampling fraction and the expected number of reported MDR TB patients in the countries from which patients were recruited (online Technical Appendix). Results suggest that frequencies of drug resistance to second-line drugs might be underestimated by our analysis.
Second, some data collected were self-reported by patients and are prone to information bias. This limitation particularly applies to information on previous TB treatment in foreign-born patients, who might fear stigmatization in the country where treatment was provided.
Third, DST was performed at laboratories that used external quality control practices. However, quality control for testing of second-line drugs varies among sites and respective laboratories (30). Incompleteness of DST data for second-line drugs demonstrates the situation with which clinicians are confronted in making their management decisions and shows the need for scale up in laboratory testing, even in MDR TB reference centers in Europe.
Despite these limitations, our study identified 3 major concerns regarding TB in Europe. First, transmission of MDR strains of M. tuberculosis is ongoing. Second, diagnostic capacity is poor, especially for DST. Third, levels of resistance to second-line TB drugs are high. These factors must be addressed in any TB surveillance and control programs that are implemented.

The Public Health Image Library (PHIL)
The Public Health Image Library (PHIL), Centers for Disease Control and Prevention, contains thousands of public health-related images, including high-resolution (print quality) photographs, illustrations, and videos. PHIL collections illustrate current events and articles, supply visual content for health promotion brochures, document the effects of disease, and enhance instructional media.