Evaluation of drug-resistant tuberculosis treatment outcome in Portugal, 2000–2016

Treatment of drug-resistant tuberculosis (TB), which is usually less successful than that of drug-susceptible TB, represents a challenge for TB control and elimination. We aimed to evaluate treatment outcomes and to identify the factors associated with death among patients with MDR and XDR-TB in Portugal. We assessed MDR-TB cases reported for the period 2000–2016, using the national TB Surveillance System. Treatment outcomes were defined according to WHO recommendations. We identified the factors associated with death using logistic regression. We evaluated treatment outcomes of 294 MDR- and 142 XDR-TB patients. The treatment success rate was 73.8% among MDR- and 62.7% among XDR-TB patients (p = 0.023). The case-fatality rate was 18.4% among MDR- and 23.9% among XDR-TB patients. HIV infection (OR 4.55; 95% CI 2.31–8.99; p < 0.001) and resistance to one or more second-line injectable drugs (OR 2.73; 95% CI 1.26–5.92; p = 0.011) were independently associated with death among MDR-TB patients. HIV infection, injectable drug use, past imprisonment, comorbidities, and alcohol abuse are conditions that were associated with death early on and during treatment. Early diagnosis of MDR-TB and further monitoring of these patients are necessary to improve treatment outcome.


Introduction
Tuberculosis (TB) treatment success (the percentage of cured patients and those with treatment completed) is one indicator for monitoring implementation of the End TB Strategy. Globally, the recommended target level for 2025 is above 90% [1]. In the European Union, the treatment success rate for the 45,499 TB cases treated in 2017 was 67.6%, still standing far from the established goal. Moreover, the latest surveillance data (2018) for patients with drugresistant TB shows lower treatment success rates: 48.1% for multidrug-resistant TB (MDR-TB), defined by resistance of Mycobacterium tuberculosis (Mtb) to isoniazid and rifampicin, and 37.4% for extensively drug-resistant TB (XDR-TB), defined as MDR-TB plus resistance to at least one of the fluoroquinolones and one of the injectable drugs [2]. Death is a more frequent unfavorable outcome among those with MDR-or XDR-TB than with susceptible TB (17.1%, 21.8%, and 6.9%, respectively) [2]. The treatment of MDR/XDR-TB requires the use of bactericidal and bacteriostatic drugs for long periods. Treatment success depends not only on the choice of an effective treatment regimen but also on the patient monitorisation and the management of therapeutics adverse events and comorbidities, potential drug-drug interactions, and even the patient's tolerability to the drug regimen implemented [3].
In Portugal, TB cases are managed mainly in TB Outpatient Centres. As a specific strategy to MDR-TB control, in 2007, the National Reference Centre was created to monitor and support the treatment of MDR/XDR-TB cases, producing the national guidelines and recommendations. Later, to support the implementation of these standard procedures, to decentralize this approach and to facilitate accessibility to all, the Regional Reference Centres were created. Although these Centres currently operate in each of the seven health regions of the country, they started working at different times. While the first Centre opened in the Northern Region in 2009 [4,5], the Centre in Lisbon and Tagus Valley Region, with the highest TB burden in the country, started to work only in 2013 [6]. In Portugal, all TB patients receive free treatment from the National Health System under the National Tuberculosis Program. Hospitalization is the first choice at the start of MDR-TB treatment, and the patient remains hospitalized until smear sputum conversion. The Regional Reference Centres are responsible for the clinical management of patients during the entire treatment course, including the choice of the treatment regimen. Adequate regimens are based on current MDR-TB treatment guidelines [7,8] and adjusted according to the clinical and microbiological response along with drug susceptibility testing (DST) results. Directly Observed Therapy (DOT) is provided throughout the treatment at the primary care level [4,5].
Following our earlier analysis evaluating treatment outcomes of MDR-TB patients in Portugal [9], here we updated information to give a complete assessment of the disease along 17 years of cases reported in Portugal. Furthermore, we assessed case-fatality rate and identified the factors associated with death in MDR-and XDR-TB patient groups.

Data collection
We selected MDR-TB cases diagnosed in Portugal from January 2000 until December 2016 from the national TB Surveillance System (SVIG-TB). We evaluated patients with known treatment outcomes and second-line drug resistance profiles, dividing those into groups: MDR-and XDR-TB cases.
Information collected included demographic and clinical characteristics of each case: age, sex, country of origin, addictions (e.g., drug or alcohol abuse), HIV status, living conditions (e.g., being a prisoner, living in a community residence, homelessness), comorbidities (diabetes, silicosis, chronic obstructive pulmonary disease, liver disease or neoplasia), previous TB treatment and site of infection. We also collected dates of the onset of symptoms, diagnosis of TB, and treatment initiation.

Diagnosis and treatment
MDR-TB diagnosis requires a positive culture or detection of both acid-fast bacilli by microscopy and an Mtb-specific nucleic acid amplification testing, followed by detection of resistance to isoniazid and rifampicin by genotypic and phenotypic methods [2]. We included in our study culture-positive MDR-TB cases tested for resistance to first-and second-line anti-TB drugs by phenotypic methods, the conventional gold standard. All tests were performed in laboratories integrated into the national network, periodically certified and checked. Mtb strains that revealed resistance to isoniazid and rifampicin were tested for second-line anti-TB drugs in the TB National Reference Laboratory (Instituto Nacional de Saúde Ricardo Jorge: INSA) in Porto, which is also a World Health Organization (WHO) Supranational Reference Laboratory.
HIV testing is done routinely, using an opt-out strategy (patients were informed that an HIV test will be conducted and that they may decline or defer) [10]. The treatment regimen is designed according to WHO guidelines. Treatment begins with a standard or empirical regimen until DST for second-line drugs results are available. Afterwards, individually tailored regimens that take into account the drug resistance patterns are used. Treatment continues for at least 18 months after culture conversion [11][12][13]. The patients are hospitalized until smear conversion and then followed in one of the Regional Reference Centre. Directly observed treatment is performed during the entire treatment.

Treatment outcomes
The treatment outcomes were defined according to WHO recommendations as cured, treatment completed, treatment failed, death, and lost to follow-up [7,14], and reported to the National TB Surveillance System. The sum of cured (treatment completed and three or more consecutive cultures taken at least 30 days apart are negative after the intensive phase) and treatment completed (treatment completed but no record that three or more consecutive cultures taken at least 30 days apart are negative after the intensive phase) is considered as treatment success. That is, treatment success includes treatment completed with or without three or more consecutive cultures taken at least 30 days apart, which are negative after the intensive phase. Although in our study no cured was registered during the studied period, it will not affect estimated success rate. Treatment failure, death and lost to follow-up were considered as unfavorable treatment outcomes.
The percentage of cases for each outcome was determined considering the total number of patients who started treatment over studied period.

Statistical analysis
We describe patient's characteristics through absolute and relative frequencies for categorical variables as gender, country of origin, HIV status, alcohol abuse, injectable drug use, prisoner, community residence, homelessness, comorbidity, chest radiography, previous TB treatment, site of disease. Median with interquartile range (IQR) was used for continuous variables as age (years), delay in diagnosis and treatment (days) and duration of treatment (months). Delay in diagnosis and treatment was defined as the period from the date of the onset of symptoms until the diagnosis of TB and treatment initiation. We compared the prevalence of these characteristics between patients grouped according to second-line drug resistance profile and between MDR-TB patients that died within and after the first six months of treatment, using the Chi-squared test (or Fisher's test, if appropriate) for categorical variables. The Mann-Whitney U-test (or median test) was used to compare continuous variables. We estimated treatment success rate by year and we compared their medians before and after 2008, year that corresponds to half of the study period and when some measures of MDR-TB control were taken in Portugal. We also estimated treatment success rate over the study period and we compared it among patient groups. We used the Chi-squared test (or Fisher's test, if appropriate) to compare patient groups' treatment outcomes. Univariate and multivariate logistic regression was conducted to identify death factors during treatment among patients in each drug resistance categories. Crude and adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were determined. Statistical analyses were performed with SPSS version 18.0 (PASW Statistics 18), and p-values below 0.05 were considered statistically significant.

Ethical considerations
Ethical approval and informed consent were not required, as the patient data, collected by an official national surveillance system, were anonymized following the ethical research guidelines in Portugal.
Demographic and clinical characteristics of the patients are shown in Table 1. Compared to MDR-TB, XDR-TB patients presented higher prevalence of alcohol abuse (34.1%), injectable drug use (30.8%), past or present imprisonment (12.7%) and previous TB treatment (54.9%) ( Table 1).

Treatment and treatment outcomes
In our study, median TB diagnostic delay, which was calculated for patients with available date for the onset of symptoms (n = 333; 76.4%), was 72 days. There were no significant differences in TB diagnostic delay between MDR-and XDR-TB (73 days and 71 days; p = 0.941; Table 1).
All 436 patients started treatment on the day of diagnosis of TB. Initial treatment regimen data was available for 414 (95.0%) patients. Of them, 170 (41.1%) received second-line anti-TB drugs in the initial treatment regimen.
Duration of treatment among MDR-TB patients was 20.4 months and 23.4 months among XDR-TB patients (p < 0.001) ( Table 1). The treatment success rate tended to increase since 2000 to 2016 (S1 Fig). We compared the medians of treatment success rate before and after 2008 and we found no statistically significant difference between them (67.5% and 78.6% respectively; p = 0.153).
The treatment success rate over the study period was superior among MDR than XDR-TB patients (73.8% and 62.7%; p = 0.023; Table 2). Among unfavorable treatment outcomes, death was more frequent than treatment failure or loss to follow-up in both groups. The case-fatality rate among MDR-and XDR-TB patients was 18.4% and 23.9% respectively (p = 0.218; Table 2).
Additionally, we evaluated treatment outcomes among MDR-TB patients, dividing cases into groups: MDR-TB without additional second-line drug resistance, MDR-TB with additional resistance to one or more second-line injectable drugs (pre-XDR SLID -TB) and MDR-TB with additional resistance to one or more fluoroquinolones (pre-XDR FQ -TB) (S1 Table).
We found that pre-XDR SLID -TB patients had worse treatment outcome than patients from other drug-resistance groups: the treatment success rate was 55.6% (p = 0.002) and the casefatality rate was 31.0% (p = 0.065) among them (S1 Table). However, due to the small numbers of patients in with pre-XDR SLID -and pre-XDR FQ -TB, we decided not to go ahead with looking for factors associated with death in these groups, but to include these resistance profiles as independent variables in the analysis of factors associated with patient death among MDR-TB.

Factors associated with patient death
We assessed separately factors associated with patient death among MDR-and XDR-TB patients (Tables 3 and 4 (Table 3).
Injectable drug use (OR 2.69; 95% CI 1.15-6.33; p = 0.023) and history past or present of imprisonment (OR 3.30; 95% CI 1.10-9.87; p = 0.033) were significantly associated with death among patients with XDR-TB in univariate analysis. However, these associations were not confirmed in the multivariate analysis ( Table 4).

The elapsed time between beginning of treatment and death
In the cases for which death was reported, the treatment duration until death was 9.5 months among MDR-TB patients and 13.1 months among XDR-TB patients (Table 5). In addition, 22  (Table 5). We compared the demographic characteristics of MDR-TB patients who died within the first six months of treatment with patients who died after that period (S2 Table). Alcohol abuse (60.0%), injectable drug use (55.6%), and comorbidities (61.5%) were more frequent among MDR-TB patients that died within the first six months, although these differences were not statistically significant. The proportion of MDR-TB patients without additional secondline injectable drug resistance was significantly higher among them (56.2%; p = 0.020) (S2 Table).

Discussion
This study evaluated treatment outcomes of a cohort of 436 patients with drug-resistant TB diagnosed in Portugal within 17 years, using national TB surveillance data. We paid special attention to death during treatment and identified factors associated with it in MDR-and XDR-TB patient groups. We found that treatment was more successful for MDR-TB than XDR-TB patients. The case-fatality rate was highest among XDR-TB patients. Death during treatment occurred earlier for patients with MDR-TB than for the ones with XDR-TB; 40.7% of them died within the first six months of treatment. HIV infection and resistance to one or more second-line injectable drugs were independently associated with death among MDR-TB patients.
In our study, the overall treatment success rate among MDR/XDR-TB was 70.2%, below the 2020 target of the action plan for the WHO European Region (75%) [15]. However, treatment success rate among MDR-TB patients (77.9%) was higher than one the reported for the European Union by the European Centre for Disease Prevention and Control (ECDC) (48.1%) [2]. In this meta-analysis, 74 different studies were from several locations (60%) [16] and a study from Brazil (58,1%). Still, it was lower than reported by European countries, like Italy [17] and the Netherlands [18] (81.3% and 88%, respectively).
Death was more frequent among unfavorable treatment outcomes in both patient groups. A higher frequency of death among MDR-TB patients was also reported by ECDC [2], while lost to follow-up was more frequently reported in the meta-analysis mentioned above [16], in Brazil [20] and China [21]. The case-fatality rate in our study was 18.4% that is equal to the rate reported in a meta-analysis (18%) [16], but is higher than reported by ECDC (17.1%) [2] and shown in Brazil (14.3%) [19] and China (2.8%) [21]. Nevertheless, in Pakistan, the rate was even higher (19.8%) [22].
Among XDR-TB patients, contrary to our results, treatment failure was a more frequent unfavorable outcome reported by ECDC [2] and shown in a meta-analysis [16] and Brazil [19]. However, our case-fatality rate (23.4%) is lower than was shown in Brazil (30.0%) [19], but is higher than was reported by ECDC (21.8%) [2] and in a meta-analysis (21%) [16].
We found that HIV infection was independently associated with death among MDR-TB patients. Although HIV infection was less prevalent among MDR-TB patients than among XDR-TB patients (23.8% vs. 31.0%), HIV infected patients of this group were 4.6 times more likely to die. This finding is consistent with the results described previously in several studies. A meta-analysis noted an increase in deaths among HIV-MDRTB co-infected patients in lowincome regions compared with high-income regions [23]. In India and Tanzania, HIV infection was also associated with death [24,25]. Toxicity and adverse events from antiretroviral therapy (ART) coupled with MDR-TB, therapy's side effects can be accountable for poor treatment outcome in this patient group. According to WHO, antiretroviral therapy must be started as early as possible (within the first eight weeks) after the beginning of anti-TB treatment, irrespective of CD4 T cell count [13]. However, a study carried out in South Africa found that ART use before MDR-TB treatment was significantly associated with higher casefatality rate than when ART was initiated after the beginning of MDR-TB treatment [26]. Unfortunately, we do not assess information about antiretroviral therapy or CD4 cell counts because this information is not reported to our TB Surveillance System. Resistance to one or more second-line injectable drugs was also independently associated with death among MDR-TB patients. We included this variable in the analysis of factors associated with death after observing the low treatment success rate (55.6%) and the high casefatality rate (31.0%) among these patients that is contradictory to what has been demonstrated in previous studies [27,28].
The second-line injectable drugs (amikacin, kanamycin, and capreomycin) are some of the core second-line drugs used in the intensive phase of treatment of MDR-TB, which varied from 6 (WHO guidelines, 2008) [11] to 7-8.5 months (WHO guidelines, 2011) [12]. However, the role in treatment and the importance of resistance to each of these drugs is not the same. On the one hand, resistance to capreomycin [29] and kanamycin [9] were independently associated with unfavorable outcome in some studies. On the other hand, a recent meta-analysis showed that amikacin provided modest benefits in treatment, while kanamycin and capreomycin were associated with unfavorable outcomes [30]. Thus, according to the new WHO consolidated guidance of 2019, kanamycin and capreomycin are not included in longer treatment regimens [13].
Among XDR-TB patients, injectable drug use and being prisoner were associated with death in univariate analysis. However, all seven prisoners, who died, were also injectable drug users. Thus, a combination of these two factors should be considered as a risk situation for death in this group of patients.
Finally, we found that 40.7% of the MDR-TB patients, who died during treatment, died in the first six months. These patients represent more than half (56.2%) of the patients without additional second-line drug resistance who died during treatment. Alcohol abuse (60.0%), injectable drug use (55.6%), and comorbidities (61.5%) were also most frequent among them. These findings indicate that patients' death is not due to resistance to second-line injectable drugs but due to their addictions and comorbidities.
One of the strengths of this study is that it evaluated treatment outcomes of a cohort of drug-resistant TB patients over a significant amount of time. These patients were also previously characterised for the genetics of the pathogen [31] and for their spatial distribution in Portugal [32]. The knowledge about risk factors for death generated in this study will improve patients' clinical management, enhancing treatment success, which is another strength. One of the study's limitations is a restriction of the studied variables due to retrospective study design and using only the SVIG-TB date. Thus, we could not assess the effect of important variables on lethality, such as antiretroviral therapy and CD4 T cell count.
In conclusion, our findings suggest that factors and conditions as HIV infection, injectable drug use, alcohol abuse and comorbidities are most often associated with death early on and during treatment. This suggests the need for early diagnosis of MDR-TB and further monitoring of patients that present those characteristics from treatment initiation. Furthermore, it is also required careful assessment of the relationship between early death and delayed MDR-TB diagnosis and clinical status of the patient that should grant further investigation.