Published ahead of print on May 29, 2008, doi:10.1164/rccm.200802-240OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200802-240OC
Influence of Multidrug Resistance on Tuberculosis Treatment Outcomes with Standardized Regimens1 Respiratory Epidemiology and Clinical Research Unit, Montreal Chest Institute, McGill University, Montreal, Canada; 2 Global Drug Facility, World Health Organization, Geneva, Switzerland; 3 Pan American Health Organization, Washington, DC; 4 European Regional Office, World Health Organization, Copenhagen, Denmark; and 5 South-East Asian Regional Office, World Health Organization, New Delhi, India Correspondence and requests for reprints should be addressed to Dick Menzies, M.D., Montreal Chest Institute, Room K1.24, 3650 St. Urbain, Montreal, PQ, Canada H2X 2P4. E-mail: dick.menzies{at}mcgill.ca
Rationale: Multidrug-resistant tuberculosis (TB) poses a major challenge to global TB control. We analyzed the association between estimated prevalence of initial or acquired MDR-TB, and treatment outcomes reported nationally. Objectives: We analyzed the estimated prevalence of initial or acquired MDR-TB and treatment outcomes reported nationally. Methods: Countries were analyzed if multidrug resistance prevalence estimates were available, and if they reported outcomes for more than 250 cases treated using standardized regimens in 2003 and/or 2004. Data sources were the World Health Organization for treatment regimens, prevalence of initial multidrug resistance, and reported cases and treatment outcomes in 2003 and 2004; the Joint United Nations Programme on HIV/AIDS for HIV seroprevalence; and the World Bank for income per capita. The adjusted impact of initial multidrug resistance on initial and retreatment outcomes was estimated with weighted multivariate linear regression. Measurements and Main Results: Among countries using one of two standardized initial regimens, failure rates averaged 5.0%, and relapse rates averaged 12.8% in the 20 countries where prevalence of initial multidrug resistance exceeded 3%, compared with an average of 1.6% (P < 0.0001) and 8.1% (P = 0.0002), respectively, in 83 countries where initial multidrug resistance prevalence was less than 3%. In 92 countries using one standardized retreatment regimen, failure rates were 2.7%, 3.8%, 6.2%, and 8.1% in quartiles of increasing prevalence of acquired multidrug resistance (P < 0.0001). When stratified by initial multidrug resistance prevalence, initial and retreatment outcomes in the 79 countries using the 6-month rifampin initial regimen were not significantly different from the 24 countries using the 2-month rifampin initial regimen. Conclusions: Currently recommended standardized TB initial and retreatment regimens should be reevaluated in countries where prevalence of initial multidrug resistance exceeds 3%, in view of poor treatment outcomes.
Key Words: tuberculosis • tuberculosis treatment • drug resistance • multidrug resistance • treatment outcomes
Multidrug-resistant tuberculosis (MDR-TB), denoting bacillary resistance to at least isoniazid and rifampin (or rifampicin), presents a major challenge for clinical care (1, 2) and TB control (3). Care is difficult because treatment must use second-line drugs that are much less effective and more toxic (2). Challenges to TB control include the need for prolonged, closely supervised therapy (2), greater requirement for training of clinical personnel, and very high drug costs (4). The World Health Organization (WHO) has estimated that in 2004 approximately 424,000 new cases of MDR-TB occurred, of which 57% had never been treated before, accounting for 2.7% of all new cases (5). The WHO has also estimated that there may be more than a million persons alive with active MDR-TB, 134,000 of whom the Stop TB partnership has estimated will receive treatment in 2007 and 2008 (6). The recent emergence of extensively drug-resistant TB (7) further emphasizes the need for new strategies to prevent MDR-TB. Most countries of the world have now adopted the DOTS (directly observed therapy, short course) strategy, as recommended by the WHO. This strategy recommends one of two standard initial regimens for new cases and a reinforced regimen for previously treated cases (8). The two regimens recommended for new cases are as follows:
The regimen recommended for patients who require retreatment (streptomycin, isoniazid, rifampin, pyrazinamide, and ethambutol for 2 mo; isoniazid, rifampin, pyrazinamide, and ethambutol for 1 mo; then isoniazid and rifampin plus ethambutol for 5 mo) has been criticized as inadequate (11, 12), particularly for patients who failed initial treatment (13). These standardized regimens have proven efficacy in patients with drug-sensitive TB (9), but in patients with drug-resistant strains their efficacy is much less, both in trials (14–17) and under program conditions (11). In most low- and middle-income countries that have adopted the DOTS strategy, drug sensitivity testing is not performed for new cases, nor for most patients requiring retreatment (8, 18). Therefore, patients with underlying drug resistance will receive the same standardized treatments, and can be predicted to have higher rates of failure and relapse. We hypothesized that, in countries using standardized initial and retreatment regimens, the proportion of patients with poor treatment outcomes would be correlated with prevalence of initial and acquired multidrug resistance. Some of the results of this study have been previously reported in the form of an abstract (19).
Countries Selected Countries were selected for analysis if they (1) reported at least 250 cases annually in 2003 and/or 2004, (2) reported at least initial treatment outcomes using internationally accepted definitions (20), (3) used one of two standardized initial treatment regimens (for the analysis of initial therapy), and (4) had information regarding prevalence of initial and acquired multidrug resistance. For the analysis of retreatment outcomes, countries had to use one standardized retreatment regimen.
Data Sources
Data Analysis Rates of treatment success, dropout, failure, and death with initial therapy were calculated by country. Rates of these outcomes were also calculated overall for retreatment, and by indication for retreatment—relapsed after cure, returned after default, and failed initial treatment. Differences in these rates between countries grouped by rates of initial and acquired multidrug resistance were tested for significance using analysis of variance and t tests. To assess the impact of treatment regimen, a stratified analysis was performed in which the relationship between initial and acquired multidrug resistance rates and initial and retreatment outcomes was estimated in countries using 6-month rifampin regimens separately from countries using the 2-month rifampin regimens. To control for potential confounding of initial multidrug resistance prevalence with other country-level characteristics, such as age, HIV seroprevalence, per capita income, or treatment regimen, multivariate weighted linear regression was performed to assess the independent effect of initial multidrug resistance on treatment outcomes, adjusted for these potential confounding factors.
Role of the Study Sponsor
In total, 155 countries reported treatment outcomes for new patients with smear-positive pulmonary TB in 2003 and/or 2004, of which 121 reported at least 250 cases annually. Of these, 103 used the 2-month or 6-month rifampin standardized regimens for new cases, reported initial treatment outcomes, and had estimated multidrug resistance prevalence available. A total of 92 countries also used the standardized retreatment regimen recommended by the WHO, and had estimates of acquired multidrug resistance available; only 74 countries reported retreatment outcomes. Estimated prevalence of other drug resistance (non–multidrug resistance) was available for only 42 of the 155 countries reporting treatment outcomes, 26 of the 103 countries using standardized initial regimens, and 22 of the 92 countries using standardized initial and retreatment regimens. In view of this limited information, this was not analyzed further. As seen in Table 1, with increasing initial multidrug resistance prevalence the failure rate of initial therapy and the overall relapse rate increased very significantly. Interestingly, drop-out and mortality rates in 2003/2004 were not associated with initial multidrug resistance prevalence. As seen in Figure 1, the prevalence of initial multidrug resistance was strongly associated with failure or relapse with initial treatment, albeit with considerable variation. Of all patients treated in each country, the proportion requiring retreatment increased from 11.9% in countries with prevalence of initial multidrug resistance of less than 1%, to 21.4% in the countries with multidrug resistance prevalence of more than 3%.
In the 92 countries that used standardized initial and retreatment regimens, retreatment success rates were highest among patients who had relapsed, as shown in Table 2. Drop-out rates were highest among patients who had previously defaulted from therapy, averaging more than 30%. Failure rates were highest among patients who had failed initial therapy, and were significantly higher—overall and in all subcategories of retreatment—in countries with higher prevalence of acquired multidrug resistance. Interestingly, the relationship between retreatment outcomes and acquired multidrug resistance was not appreciably different from that with initial multidrug resistance, as shown in Figure 2.
Compared with the 24 countries using the 2-month rifampin initial regimen, in the 79 countries using the 6-month rifampin initial treatment regimen the population was older, initial multidrug resistance prevalence was higher, HIV seroprevalence was lower, and average per capita income much higher. (These characteristics are summarized in Table E1 in the online supplement). However, as seen in Figure 3, despite these population differences, treatment outcomes were remarkably similar, once the results were stratified by prevalence of initial or acquired multidrug resistance. In the 59 countries with HIV seroprevalence in the general population of less than 1%, the population was older, with much higher income, and prevalence of initial drug resistance was also higher. (The inverse association between HIV and multidrug resistance prevalence is largely a reflection of the epidemiologic situation in sub-Saharan Africa, where HIV seroprevalence is extremely high, but drug resistance levels historically have been low, reflecting limited access to therapy). As shown in Table E2, treatment success rates of initial therapy were significantly higher in countries with low HIV prevalence, largely due to lower default rates, but retreatment success rates were similar in both groups of countries.
In multivariate weighted regression (Table 3), after adjustment for age, HIV seroprevalence, per capita income, and treatment regimen, every 1% increase in initial multidrug resistance prevalence was associated with a 0.3% increase in failure among new cases (P < 0.0001), a 1.1% increase in failure among retreatment cases (P < 0.0001), and a 1% increase in relapse (P < 0.0001). Conversely, for every 1% increase in multidrug resistance, there was a 0.8% decrease in treatment success among new cases and a 1.2% decrease in success with retreatment (Table 3). Acquired multidrug resistance was highly correlated with initial multidrug resistance; nevertheless, initial multidrug resistance was more strongly and consistently associated with treatment outcomes than was acquired multidrug resistance—even for retreatment (data not shown).
In this ecologic analysis, we found that nationally reported outcomes with standardized TB treatment regimens were strongly associated with estimated prevalence of initial and acquired MDR-TB. Failure and relapse with initial therapy, and the proportion of all patients treated who required retreatment, were all significantly higher in countries where the prevalence of initial multidrug resistance exceeds 3%, the WHO criterion for designation as an MDR-TB "hot zone" (22). In all countries, retreatment outcomes were poor—with high rates of default, failure, and mortality—which are important given the large number of patients requiring retreatment in all countries. These results suggest that the currently recommended regimens for previously untreated or new cases are adequate in countries where the prevalence of initial multidrug resistance is less than 3%. But the findings raise concerns about using the same regimens in countries where initial multidrug resistance prevalence exceeded 3%. Much has been written about the limitations of the WHO standard retreatment regimen in these settings (12, 26, 27). Our finding that initial failure and relapse rates, plus the proportion requiring retreatment, are much higher in these countries suggests that the standardized initial treatment for new cases should also be critically reexamined for these settings. The higher relapse rates may reflect underlying multidrug resistance that was declared cured, as noted previously (28). In the short term, the higher failure and relapse rates mean greater morbidity and mortality for patients, with greater economic and social harm for their families and communities. In the long term, these standardized regimens may be contributing to amplification of multidrug resistance in these countries. This problem is compounded by the lack of adequate laboratory infrastructure in these countries, making drug sensitivity testing inaccessible for almost all cases. Also worthy of comment are the poor treatment outcomes with standardized retreatment. This has been shown previously in individual countries (13, 26, 27, 29), but the consistency of the finding in all countries is sobering. Even in countries with initial multidrug resistance prevalence of less than 3%, retreatment success was less than 66%. This was because of high default rates, and rates of failure and mortality that were consistently higher with retreatment than with initial treatment in the same countries. The high rates of retreatment failure in countries with a more than 3% rate of initial multidrug resistance, and extremely high rates among patients who failed initial therapy in countries with high rates of acquired multidrug resistance, presumably reflect underlying multidrug resistance in those patients, as shown in earlier cohorts who were retreated after failing initial therapy (26, 27). However, the relationship of drug resistance levels with worse outcomes in patients who had previously defaulted or relapsed has not been previously noted. The somewhat better retreatment outcomes in patients who had relapsed may reflect the fact that some patients with relapse actually have reinfection (30). The association of retreatment failure rates with underlying multidrug resistance prevalence helps to explain the discrepant results of previous publications. Failure rates with this same regimen were as low as 1 to 4% in several sub-Saharan African countries in the 1980s (31), and 3 to 4% in Nicaragua in 1992–1996 (32), yet were as high as 35% in Russian prisons (27). Levels of drug resistance were low in Africa (22) and Nicaragua (32), and very high in Russian prisons (27) supporting our findings. Our findings, and those of others, suggest that continued use of the same regimen will result in unacceptably high rates of failure (9, 26), with amplification of resistance (12, 13). An interesting finding was the similarity of retreatment outcomes in the countries that used different initial regimens. It has been believed that retreatment results should be better in countries that restricted use of rifampin to the first 2 months only (32). However, there appeared to be little advantage with this approach, once differences in prevalence of multidrug resistance were accounted for in stratified analyses. Interestingly, the most important factor reducing the success of retreatment was drop out from therapy. This was much more frequent than with initial therapy, and particularly high (averaging 25–30%) among persons who had previously defaulted. The finding that observed noncompliance with treatment predicts future noncompliance has been noted in program evaluations (29) and individual patients (33). The finding reinforces that this high-risk group for default needs closer follow-up (29). In this regard, a shorter and simpler regimen might be helpful, to replace the current longer and more complex retreatment regimen. This study had a number of strengths. Information on treatment regimens and outcomes, multidrug resistance prevalence, and other covariates was complete for 103 countries, which used a total of three standardized regimens. This provided considerable power for analysis of these regimens. This also allowed stratified and multivariate analyses to control for potential confounding from differences between countries in median age, income, treatment regimens, and HIV seroprevalence, all of which affect treatment outcomes. Nevertheless, this study had several weaknesses. Most important is the ecologic design because the unit of analysis was at the level of each country. Although the relationships found were biologically plausible, individual patient outcomes may have been unrelated to underlying drug resistance. The information on treatment regimens, obtained from the Stop TB department, should have accurately reflected policy, but patients may have actually received other treatment. This would most likely occur if patients were treated in nongovernment sectors. However, treatment outcomes of such patients are usually not reported to national TB programs, and therefore would not have affected the analysis, which was based entirely on national program reports. The analysis of retreatment outcomes had less power because only 94 countries used internationally recommended standardized retreatment regimens, of which 2 of these countries did not have information on prevalence of acquired multidrug resistance, and only 52–74 countries reported outcomes by indication for retreatment. Thus, differences in outcomes between these subgroups were less often statistically significant. A further weakness was the limited information on other forms of drug resistance (other than multidrug resistance). In the few countries with this information, prevalence of these other forms of resistance was highly correlated with prevalence of multidrug resistance. However, differences in the prevalence of these forms of resistance may have accounted for some of the unexplained variation seen (shown in Figures 1 and 2).
Conclusions
The Fonds de la Recherche en Santé du Québec provided salary support for Dr. Menzies. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200802-240OC on May 29, 2008 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form February 3, 2008; accepted in final form May 19, 2008
Related articles in AJRCCM:
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
