Drug Susceptibility Patterns in MDR-TB Patients: Challenges for Future Regimen Design. A Cross-Sectional Study

Globally, there is substantial concern regarding the challenges of treating complex drug resistance patterns in multidrug resistant tuberculosis cases. Utilising data from three different settings (Estonia, Latvia, Romania) we sought to contrast drug susceptibility profiles for multidrug resistant tuberculosis cases, highlight the difficulties in designing universal regimen, and inform future regimen selection. Demographic and microbiological surveillance data for multidrug resistant tuberculosis cases from 2004–13 were analysed. High levels of additional resistance to currently recommended second line drugs were seen in all settings, with extensive variability between countries. Accurate drug susceptibility testing and drug susceptibility testing data are vital to inform the development of comprehensive, flexible, multidrug resistant tuberculosis guidance.


Introduction
The World Health Organization (WHO) estimates that 480,000 people globally had incident multidrug resistant tuberculosis (MDR-TB) in 2013; 97,000 started MDR-TB treatment [1]. The global treatment success rate for the 2011 cohort of patients was 48% [1]. 20 month regimens with an intensive phase of 8 months are recommended for most patients that contain at least four second-line drugs likely to be effective in the intensive phase ("a fluoroquinolone, a parenteral agent, ethionamide (or prothionamide), and either cycloserine or p-aminosalicylic acid (PAS)"), as well as the first-line drug pyrazinamide [2]. Tailoring treatment regimens based on individual resistance patterns, which may be highly complex in MDR-TB patients, improves the chance of cure, but may not always be feasible.
Estonia, Latvia and Romania are three central European countries within the WHO European region with contrasting challenges for MDR-TB control. Estonia and Latvia are both ranked as high MDR-TB burden countries, with an estimated 17% of new tuberculosis (TB) cases and 48% of retreatment cases MDR-TB in Estonia, and 9% and 26%, respectively, in Latvia, in 2013 [1]. Estonia is classified as having a high HIV burden, with 13% of tested TB patients HIV positive [1]. In Romania the percentage MDR-TB was estimated to only be 3% among new tuberculosis (TB) cases and 11% among retreatment cases in 2013, but Romania's high TB incidence and population size mean absolute numbers of pulmonary MDR-TB cases in 2013 were 4.5 times greater than in Estonia and Latvia combined [1].
By describing and contrasting drug resistance patterns among MDR-TB patients in three nations with differing epidemiological profiles and for whom MDR-TB is of concern, as well as reflecting on these patterns in the light of WHO MDR-TB treatment guidance, we sought to inform the development of future treatment recommendations applicable to a variety of settings, without being adversely prescriptive.

Methods
For this cross-sectional study demographic and microbiological data for diagnosed and notified MDR-TB cases were extracted from the three countries' surveillance, clinical and laboratory systems for the years 2004-2013. An effort was made to access all cases and case data, in order to avoid bias. This time frame represented the most recent decade for which data were available from all three countries.
Drug susceptibility testing (DST) results for the first line drugs ethambutol, isoniazid, pyrazinamide and rifampicin; the injectables amikacin, capreomycin and kanamycin; the fluoroquinolones ofloxacin and moxifloxacin; the oral second line bacteriostatic drugs cycloserine, paminosalicylic acid (PAS) and prothionamide; and the second/third line drug linezolid were available for 2013. Streptomycin is not recommended for use in MDR-TB cases by WHO and hence was not included [2].
During the time period studied Estonia used BACTEC MGIT as its gold standard DST for first line drugs and capreomycin, amikacin, kanamycin, prothionamide, ofloxacin, moxifloxacin, and linezolid. (HAIN GenoType MTBDRplus is also used as a preliminary diagnostic for isoniazid and rifampicin and MTBDRsl for fluoroquinolones and capreomycin, amikacin and kanamycin, but if there is a discrepancy with the BACTEC MGIT results the latter are taken.) Latvia used the indirect proportion method on Löwenstein-Jensen solid media for isoniazid, rifampicin, ethambutol, kanamycin, amikacin, capreomycin, ofloxacin, ethionamide (as representative of both ethionamide and prothionamide sensitivity testing), cycloserine, and PAS and indirect DST in the BACTEC MGIT system for isoniazid, rifampicin, pyrazinamide, amikacin, capreomycin, and ofloxacin. The choice of method for isoniazid and rifampicin depended upon whether the patient was deemed high risk for MDR-TB; if this was the case BACTEC MGIT is chosen. In (rare) instances of disagreement for ofloxacin sensitivity the worst case scenario, i.e. resistance, was assumed. GeneXpert was introduced in Estonia in 2009 and Latvia in 2010; rifampicin sensitivity results are always confirmed by culture in both countries. Romania used Löwenstein-Jensen solid media for isoniazid, rifampicin, ethambutol, kanamycin, amikacin, capreomycin, ofloxacin, ethionamide (as representative of both ethionamide and prothionamide sensitivity testing), cycloserine, and PAS. Genetic tests and DST on liquid media have been used in few laboratories for selected cases (with a high probability of MDR-TB). In case of discordant results, a decision was taken based on DST on the solid media results. New (a 'patient who has never been treated for TB or has taken anti-TB drugs for less than one month') and retreatment (a 'patient who has been treated for one month or more with anti-TB drugs in the past') cases, defined as per WHO guidance [1], were analysed separately. The percentage of MDR-TB cases known to be fluoroquinolone resistant, fluoroquinolone and/or second-line injectable resistant, and extensively drug resistant (XDR; both fluoroquinolone and second-line injectable resistant) was examined over time. For these calculations ofloxacin was the only fluoroquinolone considered in order to ensure consistency between the countries, as Latvia and Romania did not test moxifloxacin sensitivity.
Data were additionally broken down by population group-children under 14 years of age, individuals with HIV, and vulnerable individuals (unemployed, drug users, alcoholics, and/or previous or current imprisonment).
This study did not undertake primary research on human subjects; it utilised aggregate, precollected, surveillance data that was anonymised and de-identified in-country prior to analysis and thus did not require informed consent. All countries have ethical permission to collect surveillance data and analyse it under such conditions without further study-specific ethical approvals being required (Estonia: Government Regulation Act for Tuberculosis Registry No 70, 26.05.2011; Romania: Ministry of Health for Romania 'Technical norms for national public health programs implementation'; Latvia: approval for this project from the ethics committee of Riga Stradins University).

Results
In  Romania more MDR-TB cases are resistant to second-line injectables than fluoroquinolones (57% in 2013); the reverse is true in Estonia (24% in 2013). For all countries in 2013 versus 2004 resistance to injectables proportionally accounted for less of the additional resistance among MDR-TB cases than resistance to fluoroquinolones. Drug resistance patterns for cases diagnosed in 2013 were broken down further, as would be required to inform treatment decisions ( Table 2). Resistance to the first-line drugs pyrazinamide and ethambutol was high (70-96% resistant to ethambutol across all cases and countries; 69-75% resistant to pyrazinamide in Estonia and Latvia (not tested for in Romania)).
Only Estonia undertook DST to moxifloxacin as well as ofloxacin among the fluoroquinolones (Table 2); of 19 ofloxacin resistant cases 16 were tested for susceptibility to moxifloxacin and four found to be resistant. The percentage of MDR-TB cases resistant to ofloxacin was high in Estonia and for retreatment cases in Latvia and Romania.
Resistance patterns to the injectables capreomycin and amikacin were relatively uniform between new and retreatment cases, apart from in Romania, with Latvian cases and Romanian retreatment cases most likely to be resistant (Table 2). In all three countries kanamycin resistance was more common in retreatment cases than in new cases. In Estonia kanamycin resistance was more common than resistance to other injectables; of the 14 cases resistant to kanamycin six were also resistance to amikacin and capreomycin. Romania and Latvia also undertook DST to cycloserine and PAS (Table 2). PAS and cycloserine resistance are of particular concern for Romanian retreatment cases, and PAS resistance for all Latvian cases. Prothionamide/ethionamide resistance was high among Latvian cases and Estonian retreatment cases, but not elsewhere.
Only Estonia undertook DST to the group 5 drug linezolid [3]; resistant cases were few in number (Table 2).
Latvia sees the highest percentage of MDR-TB cases (24% in 2013) co-infected with HIV, with 96%, 91% and 75% of MDR-TB cases tested in Estonia, Latvia and Romania, respectively. Paediatric MDR-TB was rare in all countries. Vulnerable population groups comprised 74% of MDR-TB cases in Estonia, and 54% in Romania, in 2013 (data unavailable for Latvia).

Discussion
Extensive additional drug resistance exists in MDR-TB cases from Estonia, Latvia and Romania, with considerable between-country variability. It is useful to evaluate such patterns in relation to WHO treatment recommendations for MDR-TB: the high percentage of cases pyrazinamide resistant means that this drug is unlikely to benefit many patients; the increase in the proportion of cases with additional resistance to fluoroquinolones; high levels of resistance to prothionamide/ethionamide in Latvia cases and Estonian retreatment cases. The difference in the resistance profiles of MDR-TB cases in Estonia and Latvia is especially interesting, given their geographical proximity.
Our greatest uncertainty surrounds the estimates for Romania, given the low percentage of cases with a reported 2 nd line DST and concerns about their representativeness. It should also be borne in mind that notified cases in Romania excluded those that were not treated e.g. due to broad drug resistance. Reporting fold changes may conceal more complex trends; however, except where noted, these figures were consistently decreasing over the study period.
It is critical in the context of any report that utilises phenotypic or genotypic DST data to consider the reliability and reproducibility of the tests utilised. [4] On the basis of available evidence the WHO classifies DST for moxifloxacin, cycloserine, PAS, and prothionamide as category four (of a possible five, with routine DST not being recommended for category four or five); linezolid DST to category five and ofloxacin DST to category three. The BACTEC MGIT testing system may overestimate the likelihood of pyrazinamide resistance. [5] All three countries used phenotypic DST, which provides reassurance about the comparability of the data, however the assessed levels of additional drug resistance in the described MDR-TB patients may be inaccurate for certain drugs due to the quality of the tests available. Thus the usefulness of such testing data for clinical decision making should be carefully evaluated. Dalton [6,7]. Although both papers examined additional drug resistance in MDR-TB cases (although not reporting separately for new and retreatment cases), Kurbatova pooled data across continents. The most recent entry to the cohort was five years previous to our dataset and our population-based dataset may be more representative than PETTS participants, which may explain the striking contrasts in terms of a much lower proportion of cases resistant to capreomycin in Latvia, and a suggestion of lower levels of ethambutol and fluoroquinolone resistance in the latter [6]. Zignol et al. previously analysed surveillance data of drug resistance patterns across Europe 1997-2012, broadly examining resistance to fluoroquinolones and/or injectables for Estonia and Latvia, but not in the detail provided here [8].
The implementation of an individualised regimen policy within a country requires high quality DST data and a fast turnaround time. Given the relatively small number of cases in Estonia and Latvia it may be feasible to individualise treatment regimens. In Romania individualisation was more challenging as capreomycin and PAS were only available through the Global Fund for certain patients. Between-country differences complicate future guideline development for treatment regimens and drug susceptibility surveillance. The complexity of the resistance patterns reported makes treating cases using programmatic guidelines in the absence of 2 nd line DST results very difficult and risks cases accumulating further resistance if access to second line or new drugs increases without improved DST access.
By describing and comparing the patterns of additional drug resistance of MDR-TB cases in three epidemiologically distinct nations we illuminate the difficulties surrounding the development of future treatment guidelines, and highlight some particular areas of concern for MDR-TB control in Estonia, Latvia and Romania for the coming years. Accurate DST and DST data for second line drugs from a variety of countries is vital for comparison to WHO treatment recommendations in order to inform the development of future MDR-TB treatment regimens that meet the needs of all regions and patient groups.

Disclaimer
The views expressed are those of the authors and not necessarily those of the UK Medical Research Council, National Health Service, National Institute for Health Research, Department of Health, or Public Health England.