Contributions of culture and antimicrobial susceptibility tests to the retreatment of patients with pulmonary tuberculosis

Introduction: This study evaluated the effi cacy of retreatment of pulmonary tuberculosis (TB) with regard to treatment outcomes and antimicrobial susceptibility testing (ST) profi les. Methods: This retrospective cohort study analyzed 144 patients treated at a referral hospital in Brazil. All of them had undergone prior treatment, were smear-positive for TB and received a standardized retreatment regimen. Fisher’s 2-tailed exact test and the χ2 test were used; RRs and 95% CIs were calculated using univariate and multivariate binary logistic regression. Results: The patients were cured in 84 (58.3%) cases. Failure was associated with relapsed treatment and abandonment (n=34). Culture tests were obtained for 103 (71.5%) cases; 70 (48.6%) had positive results. ST results were available for 67 (46.5%) cases; the prevalence of acquired resistance was 53.7%. There were no signifi cant differences between those who achieved or not therapeutic success (p=0.988), despite being sensitive or resistant to 1 or more drugs. Rifampicin resistance was independently associated with therapeutic failure (OR: 4.4, 95% CI:1.12-17.37, p=0.034). For those cases in which cultures were unavailable, a 2nd model without this information was built. In this, return after abandonment was signifi cantly associated with retreatment failure (OR: 3.59, 95% CI:1.17-11.06, p=0.026). Conclusions: In this cohort, the general resistance profi le appeared to have no infl uence on treatment outcome, except in cases of rifampicin resistance. The form of reentry was another independent predictor of failure. The use of bacterial culture identifi cation and ST in TB management must be re-evaluated. The recommendations for different susceptibility profi les must also be improved.

Although tuberculosis (TB) is among the oldest and most prevalent of human diseases, specific and effective treatment only became available in 1944 with the discovery of streptomycin [1][2][3] .Within a few years after this discovery, antimicrobial resistance could be recognized from the observation of reduced effectiveness [3][4][5] .Although there has been little progress in the development of new drugs since the introduction of isoniazid and rifampicin, TB remains a curable and preventable disease 6 .However, the management of resistant cases is complex and presents therapeutic limitations.
In this context, the most concerning situation is the retreatment of patients who are often exposed to conditions associated with future failures that are attributable to microbial resistance.The management of the most resistant forms requires the use of more complementary tests, as well as more expensive and toxic drugs, which are accompanied by diminished effectiveness 2,7,8 .In this sense, ongoing global efforts aim to reduce and control the disease.Special attention is being paid to reducing disease transmission and developing preventive measures, such as vaccines, and cures through new drugs.
Meanwhile, the occurrence of resistant forms of TB has increased signifi cantly since the 1980s 2 .In the 1990s, outbreaks of multidrug-resistant (MDR) TB due to nosocomial transmission were characterized by late diagnosis, the use of inadequate treatment regimens and high transmission and mortality rates 2,9 .In the following decade, extensively drugresistant TB (XDR-TB) was described.XDR-TB is underscored by a high mortality rate (98%) and shorter median survival period, suggesting poor TB control programs and potential adaptation of the bacteria 3,[10][11][12][13] .Current studies acknowledge that patients infected with MDR strains are more prone to treatment failure, progression to chronic forms of the disease and death 14 .
The latest Brazilian guidelines recommend that the retreatment of TB cases include a standard regimen of 4 drugs,

RESULTS
as well as culture and susceptibility tests, although the benefi t of this practice has yet not been suffi ciently evaluated in Brazil 15,16 .
This study's primary aim was to evaluate the effi cacy of a retreatment regimen for patients referred to a referral hospital in Minas Gerais, Brazil, with regard to the measurements of cure (i.e., success) or abandonment and failure and these measurements' relationship with the bacterial susceptibility profi le to the drugs used.The secondary aim was to determine possible exploratory associations between clinical variables and treatment outcomes.
This was a retrospective cohort study of patients referred for retreatment of pulmonary TB at Hospital Júlia Kubitschek (HJK), a TB referral center located in Belo Horizonte, State of Minas Gerais, Brazil, from January 2004 to December 2007.The study enrolled patients with diagnoses of pulmonary TB, based on smear-positive results and a history of previous TB treatment.Patients were selected from the medical records of the hospital archives (Department of Medical Archives, SAME) and from the Hospital Admission System (SIH), Information System for Notifi able Diseases (SINAN) and Mortality Information System (SIM) databases.
Patients in retreatment with regimens other than the IR regimen (i.e., rifampicin, isoniazid, ethambutol and pyrazinamide) (n=912), cases of death from other causes (n=6) and cases with no information about treatment outcome (n=45) were excluded.From the database, 1,124 treatments were selected, and those cases with clinical indications for use of the 2RHZE/4RHE (two months of daily administration of rifampicin [R], isoniazid [H], pyrazinamide [Z] and ethambutol [E], followed by four months of daily R, H, E), standardized regimen were reviewed.After exclusions, 144 patients remained (Figure 1).In the case of multiple treatments, only the data for the later treatment were considered for analysis.
Mycobacterium tuberculosis was isolated using Lowenstein-Jensen (LJ) culture medium, and susceptibility testing (ST) was performed using the proportion method 17 .All of the tests were performed at the Central Laboratory (LACEN) of the State of Minas Gerais, which is internationally certifi ed.
Outcomes were considered successful (i.e., cured) or treatment failures (i.e., death, failure or treatment abandonment), based on the medical records and review of the databases (SIH, SINAN and SIM).Both proven cures and unproven cures (2 negative smears at the end of treatment and when the patient could not perform the smear) were considered because the designations were made by the attending physician.Categorical variables were compared using the χ 2 test or Fisher's exact test when applicable.Multiple regression analysis was performed to estimate possible independent associations between variables and outcomes, including their relative risks (RRs) and 95% confi dence intervals (CIs).Variables with a p-value less than 0.20 in univariate binary logistic regression were entered into the model.The database was created using Epidata for Windows, Entry 3.1 (EpiData Association, Copenhagen, Denmark); inconsistencies were checked by double entry.The analysis was performed using SPSS for Windows (version 13.0; SPSS Inc., Chicago, IL, USA).

This study was approved by the Ethics in Research
Committees of HJK and the Federal University of Minas Gerais (COEP-UFMG), and a waiver of informed consent was obtained.
The most prevalent cause of retreatment was the abandonment of treatment ([71.8%],mean of 1.9 prior treatments [range: 1-9]).Almost 90% of the patients had their initial TB therapy in a hospital setting (89.6%), with an average hospital stay of approximately 2.5 months, which was followed by outpatient management.Fifteen (10.4%) patients were not admitted to the hospital (Table 1).

Susceptibility test results
The results of culturing were available for 103 (71.5%) of the 144 patients in retreatment: 70 (48.6%)were positive, and 33 (22.9%) were negative.Therefore, ST could be performed in 67 (46.5%) patients of the sample or 65% of the available results).The mean interval between the collection and provision of results was 3 months (range: 20 days to 8.9 months).In 25 (37.3%)patients, ST results were made available 4 months after collection.There was no significant difference in treatment outcomes with regard to the availability of ST results (Table 1).
Univariate analysis revealed that the variables signifi cantly associated with treatment failure were alcohol use, the presence of concomitant diseases, history of 5 or more treatments, retreatment due to abandonment, retreatment due to failure and rifampicin resistance (Table 3).These variables and the length of hospital stay were included in the multivariate logistic regression analysis.For this analysis, the variable of return due to failure replaced that of return due to abandonment because both represent the same phenomenon.In contrast, in the multivariate analysis, only rifampicin resistance was independently associated with therapeutic failure (OR: 4.4; 95% CI: 1.12-17.37;p=0.034;Table 3).In the second model in which information from ST was omitted, only return due to abandonment was associated with retreatment failure (OR: 3.59; 95% CI: 1.17-11.06;p=0.026;Table 3).

DISCUSSION
The primary aim of this study was to evaluate the effi cacy of the retreatment regimen available in Brazil for pulmonary TB patients who were managed at a referral center of a public hospital in Minas Gerais, according to the measurement of cure (i.e., success) or relapse and failure, and the regimen's possible association with the microbacterial susceptibility profi le of the drugs.The secondary objective was to determine exploratory associations among clinical variables and treatment outcomes.
Despite the low availability of ST results, the high rate of acquired resistance and multiple forms of resistance, the main results showed a cure rate of 58% among retreatment cases and no associations between this outcome and the bacterial susceptibility profi les.
The results of the culture test, which represents a propedeutic breakthrough and which has been included in the national guidelines for all retreatment cases since 2004, were present in 71.5% of records 15 .Cultures were negative in 33 (32%) cases, while ST could be performed in only 67 cases, corresponding to 65% of available culture results.
The lack of information about culture exams in approximately 25% of the cases can be attributed to the absence of a request by the attending physician at the time of admission.This low yield of culture testing can also be explained by the limitations of the test itself and the scarcity of clinical specimens, which are often insuffi cient.However, these cases were considered representative of the sample because of their randomness.
In addition, the long period until the release of the ST results (approximately 3 months) can be attributed at least in part to the need to send materials to a laboratory not associated with the hospital and to the use of the proportion method.
A large number of patients exhibited polyresistance to 2 or more drugs (24/36, 66.6%).There were 18 patients resistant to MDR-TB, and 6 presented other combinations of multiple resistance.This fi nding could be due to the use of a state-level reference center, as is the case with HJK, which receives more referrals for complex cases.This fi nding might also have been a determining factor in the recognition of rifampicin as an independent predictor of failure because 18 of the 19 patients resistant to rifampicin exhibited MDR-TB (Table 2).
Secondary resistance was quite high, i.e., 53.7% of the sample.However, this rate is lower than the rate of secondary resistance of 69% reported by Fiuza de Melo et al. in São Paulo 18 .
The delay in the recognition and identifi cation of resistant forms of the disease or a satisfactory clinical response to unique forms of resistance, as reported in some studies, might also have been instrumental in this aspect 19 .Among the patients who obtained their ST results, 19 (28.1%) were resistant to rifampicin, 32 (48.5%) were resistant to isoniazid, and 10 (17.2%) were resistant to both simultaneously.According to Kritsky et al. 20 , patients with dual rifampicin and isoniazid resistance receiving the 2RHZE/4RHE regimen for a long time until the release of ST results present a risk for the use of ethambutol as monotherapy in the maintenance phase of treatment 20 .
In Belo Horizonte, Minas Gerais, 54% of all TB cases are diagnosed in hospitals, emergency services and outpatient referral clinics.According to unpublished data from the Municipal Health Department, the cure rate with retreatment was 51.8% between 2004 and 2007 with 13% abandonment.In the present study, the cure rate was 58.3%, which is very similar to the 51.8% mentioned above but higher than the cure rates recently published by the WHO, i.e., 26% and 38% for Brazil and the Americas, respectively, in 2005 21 ; this difference might be due to the high proportion of patients not evaluated in those samples, i.e., 25% and 21%, respectively.Moreover, most (89.6%) of the patients in the present study initiated their treatment in a hospital setting, with an average that was stay longer than the fi rst phase of treatment, which might have contributed to better results.
In 2007, Bierrenbach et al. reported a cure rate of 58.9% among retreatment cases throughout Brazil 22 .Regarding regional differences, the southeast region had the lowest (40.5%) cure rate 22 .
In this study, the cure rate was 76.4% when patients who abandoned treatment were excluded.When MDR forms (e.g., RH) were excluded, the cure rate was 66.6%.
There was no information regarding the use of supervised treatment strategies for patients treated as outpatients.No resistance profiles were significantly altered because of treatment (Table 2).
The prevalence of HIV was 6.9% among patients in this study, a rate that could be considered low compared to the HIV prevalence in Belo Horizonte in 2010 (12.5%) and compared to the most recent data published by the WHO for Brazil (16%) 22,23 .This fi nding might have been due to the patients with this diagnosis being managed at local HIV/AIDS referral hospitals.
This study had limitations associated with its retrospective design.There were many losses attributed to misinformation regarding the outcomes of many treatments and the yield of ST.Other limitations that are worth mentioning include only patients from one referral center being included, and the recurrence of resistant forms after cure was not evaluated.
In summary, the resistance profi le presented by ST appears to be dissociated from the treatment outcome, namely cure or failure, of retreated TB pulmonary patients.The fi nding of resistance to rifampicin or the method of retreatment in the absence of ST should be analyzed as possible predictors of treatment failure.
The appropriate management of these patients regarding the use of the different microbacterial drug susceptibility profi les requires further research.

FIGURE 1 -
FIGURE 1 -Selection of retreatment cases, Hospital Júlia Kubitschek , from 2004 to 2007.MDR TB: multidrug-resistant tuberculosis; CNS TB: central nervous system tuberculosis.Note: *only the data for later treatments were considered for analysis.

TABLE 1 -
Demographic and epidemiological characteristics of 144 patients with pretreatment with regard to current treatment outcomes.

TABLE 3 -
Univariate and multivariate analyses (Cox model), including odds ratios and 95% confi dence intervals, of the tuberculosis retreatment outcomes.

TABLE 2 -
Profi le of the drug susceptibility of 67 bacterial isolates from patients undergoing retreatment for tuberculosis.