INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

  If rifampin resistance became widespread, it would threaten   the success of modern short-course treatment of tuberculosis.

D. A. MITCHESON (1)

Rifampin, along with isoniazid, is the mainstay of chemotherapy for treatment of tuberculosis. Rifampin is widely used today in most countries in the world, if not for the entire course of treatment, for at least the initial 2 mo of therapy. Without rifampin, 18 to 24 mo of therapy is necessary to treat active disease (2). Rifampin is essential to short-course chemotherapy for tuberculosis (i.e., a 6-mo regimen). Rifampin resistance will impact upon the ability of tuberculosis control programs to provide short-course chemotherapy, leading to higher program costs and a larger burden of patients.

Rifampin resistance has been observed most frequently in association with resistance to isoniazid (3). Resistance to rifampin alone (rifampin mono-resistance) is rare (1, 4, 5). In treatment trials of 8,212 patients conducted by the British Medical Research Council, primary resistance to isoniazid alone was seen in 447 (5%) patients, but resistance to rifampin alone in pretreatment isolates was seen in only two persons (0.02%) (1). Resistance to rifampin exists at a rate of approximately 1 in 10⁸ bacilli (6), whereas resistance to isoniazid exists more commonly at a rate of 1 in 10⁶ bacilli (7).

In a 1991 survey in the United States of susceptibility patterns of Mycobacterium tuberculosis isolates, isoniazid resistance alone was seen in 125 (3.7%) of 3,313 isolates, and rifampin mono-resistance was reported in only 13 (0.4%) isolates (3). In that study, rifampin resistance was seen more commonly in association with resistance to isoniazid (111 [3.5%] of 3,313 isolates).

In 1993, the Centers for Disease Control and Prevention (CDC) conducted a match between national tuberculosis and acquired immune deficiency syndrome (AIDS) registries in selected states in the United States (8). Rates of drug-resistant tuberculosis were compared between persons with and without reported cases of AIDS. There were no differences in rates of resistance to isoniazid alone, 7.8 and 10.0% for persons with and without AIDS, respectively. However, a significantly higher proportion of persons with AIDS had rifampin mono-resistant tuberculosis when compared with those without AIDS, 20/583 (3.4%) versus 19/3,813 (0.5%), (RR, 6.9, 95% confidence interval, 3.7 to 12.8) (8). Other published reports of patients with rifampin mono-resistant tuberculosis, the majority of whom were also infected with human immunodeficiency virus (HIV), have led to speculation that rifabutin prophylaxis for Mycobacterium avium complex disease in persons with unrecognized tuberculosis could result in a selection of rifampin mono-resistant M. tuberculosis (9). We conducted a multicenter matched case-control study to investigate the relationship between rifampin mono-resistant tuberculosis and use of rifabutin and to identify other risk factors for rifampin mono-resistant tuberculosis among patients with and without HIV infection.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study sites included Houston, Dallas, and San Antonio, Texas; San Francisco, San Diego, and Los Angeles, California; Chicago, Illinois; and New York State, excluding New York City. Study cases were selected from cases of tuberculosis reported to CDC's national surveillance database from 1993 to 1995 with a first M. tuberculosis isolate resistant to rifampin and without resistance to isoniazid, ethambutol, or streptomycin. Additional case finding was conducted through search of local health department tuberculosis registries and laboratory logs. Cases younger than 20 yr of age at the time that tuberculosis was reported were excluded from the study.

Two matched controls 20 yr of age or older were chosen per case from tuberculosis cases with isolates demonstrating susceptibility to isoniazid, rifampin, ethambutol, and, if tested, streptomycin. Controls were the two next reported patients who matched the case by city of residence (or county, if from a rural area), HIV status, age within 5 yr and start of tuberculosis treatment within 3 mo. Cases and controls for whom HIV status was unknown were analyzed in the HIV negative group.

Medical records on all cases and controls were reviewed, including tuberculosis clinic records, hospital records, private provider records, and HIV clinic records (when applicable). Information collected included demographics, history of homelessness, incarceration, prior tuberculosis, diarrhea, and medications taken at the time of and prior to the current diagnosis of tuberculosis. Antifungal therapy refers to the use of systemic treatment. Information on the presence of diarrhea was obtained from records of patient reports and provider notes. For those with HIV infection, we obtained date of HIV diagnosis, CD4 T-lymphocyte count at the time of diagnosis of tuberculosis, and history of prior opportunistic infections. We looked for epidemiologic and laboratory evidence of transmission of M. tuberculosis among cases through review of contact investigations and DNA fingerprinting of available isolates (13, 14). DNA fingerprint patterns were considered matched if the patterns were identical. Susceptibility to rifabutin at 2 µg/ml was performed on available isolates (15). Polymerase chain reaction products (147 base pairs) encompassing a 81 base-pair region frequently associated with rifampin resistance in the mycobacterial rpoB gene encoding the -subunit of DNA-dependent RNA polymerase were sequenced from these isolates (7, 16).

Data were analyzed using Epi Info Version 5 (17), and matched odds ratios were calculated. Continuous variables were compared using Student's t and Kruskal-Wallis tests when appropriate. To adjust for the influence of potential confounding and effect modification between correlated variables, multivariable analysis was performed with stepwise, conditional logistic regression using EGRET (18). Variables associated with p values of 0.2 or less in the univariate analysis were included in the regression analysis.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 77 cases of tuberculosis with rifampin mono-resistance were identified. Of these, three were determined to be laboratory contaminants based upon lack of clinical signs or symptoms consistent with tuberculosis or response to antituberculosis therapy. Of the remaining 74 cases, 10 cases had an identified epidemiologic link to another case with rifampin mono-resistant tuberculosis. Nine were a part of a cluster in Chicago. In addition, one case in San Francisco was a coworker of an earlier case from San Diego. Isolates from both patients had identical 11-band DNA fingerprints. These 13 cases were excluded from further analyses.

The remaining 64 cases were from Houston (n = 6), Dallas-Fort Worth (n = 6), San Antonio (n = 1), San Francisco (n = 5), San Diego (n = 6), Los Angeles (n = 22), Chicago (n = 15), and New York State (n = 3). Demographic data are shown in Table 1. The mean age of cases was 42 yr. Fifty-two (81%) were male and 28 (44%) were black, 19 (30%) were Hispanic, 13 (20%) were white, and 4 (6%) were Asian. Twenty-four (38%) of the cases were foreign-born. HIV status was unknown for nine cases. Of these, eight had no known risk factors for HIV infection, and one had a history of injecting drug use; 17 cases were HIV negative. Thirty-eight (59%) cases were reported to have HIV infection. In 30 of these, HIV infection was diagnosed prior to the diagnosis of tuberculosis, and in eight, HIV infection was diagnosed at the time of the tuberculosis diagnosis. At the time of the medical record review, 22 (34%) patients were known dead; 15 of these had had AIDS.

These 64 cases and 126 controls were included in the matched analysis. Cases and controls were not statistically different with regard to sex, race/ethnicity, foreign birth, reported homelessness, or a known history of prior incarceration. Cases were significantly more likely than controls to have a history of prior tuberculosis (Table 1). Of the 12 cases with prior tuberculosis, seven were in persons without reported HIV infection.

In a separate analysis of persons with HIV infection (38 cases and 74 controls), there were no statistically significant differences between cases and controls with respect to race/ ethnicity, sex, foreign birth, homelessness, or a history of prior incarceration. Only five cases had a history of prior tuberculosis, and most of these 38 cases probably represent primary rifampin mono-resistant tuberculosis. Cases were significantly more likely than controls to have a history of prior opportunistic infection, diarrhea prior to diagnosis of tuberculosis, and lower CD4 T-lymphocyte count at the time of tuberculosis diagnosis (Table 2). Cases were also more likely than controls to have a history of rifabutin use, antifungal therapy, or antiretroviral therapy prior to the diagnosis of tuberculosis. In a stepwise logistic regression analysis, antifungal therapy (adjusted odds ratio, 4.7; p < 0.01), diarrhea (adjusted odds ratio, 3.0; p = 0.05), and rifabutin (adjusted odds ratio, undefined, p < 0.001) remained statistically significant. For the 27 matched sets where the case did not use rifabutin, prior antiretroviral therapy appeared to be an independent risk factor (adjusted odds ratio, 4.7; p = 0.01). For the 11 matched sets in which the case had a history of rifabutin use, diarrhea was also strongly associated with rifampin mono-resistant tuberculosis (adjusted odds ratio, undefined; p < 0.01). Cases with a history of rifabutin use were not clustered in a single geographic location (one from New York State, two from San Diego, five from Los Angeles, one from Dallas, and two from San Francisco). Ten cases received rifabutin as prophylaxis for M. avium complex, and one received rifampin and rifabutin for treatment of M. avium complex infection. Because no controls had used rifabutin, the interaction of rifabutin in relation to other variables could not be assessed. No cases who took rifabutin had a history of prior tuberculosis.

Laboratory

Of the 15 rifampin-resistant M. tuberculosis isolates with susceptibility to rifabutin performed, four (27%) were susceptible and the remaining 11 (73%) were resistant. Of the 37 available case isolates for DNA fingerprinting, all had unique DNA fingerprints except for two pairs of patients who had matching fingerprints. One pair had a pattern with 11 bands and the other with 15 bands. Both pairs were patients from Chicago; however, no epidemiologic links were identified. Results of polymerase chain reaction products of the 81 base pair segment of the rpoB gene of the DNA-dependant RNA polymerase in 24 M. tuberculosis isolates tested showed six different point mutations present (His 526 Tyr, Ser 531 Leu, His 526 Leu, His 526 Asp, Asp 516 Val, and Gln 513 Leu).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study identifies several risk factors associated with rifampin mono-resistant tuberculosis. First, recent transmission of rifampin mono-resistant M. tuberculosis strains probably accounted for the occurrence of 10 (13%) of the 74 cases identified, as suggested by an epidemiologic link to another person with rifampin mono-resistant tuberculosis, however, this may represent an underestimation because of limitations of contact investigations. Two other pairs with matching fingerprints may also represent unidentified recent transmission of rifampin mono-resistant tuberculosis. Second, inadequate treatment of an earlier episode of drug-susceptible tuberculosis probably led to acquired rifampin resistance in another 12 (16%) cases. A prior history of tuberculosis and prior rifampin treatment was the factor most strongly associated with rifampin mono-resistance, particularly among those without a known history of HIV infection. In a study from Saudi Arabia, widespread use of rifamycins for treatment of bacterial infections has been postulated as the cause of increasing rates of rifampin-resistant tuberculosis (19). Acquired mono-rifampin resistance versus reinfection with a rifampin mono-resistant M. tuberculosis strain cannot be ruled out in our patient population because prior isolates were not available for DNA fingerprint comparison. The majority of rifampin mono-resistant tuberculosis cases in this analysis did not have any of the risks mentioned above.

An association between HIV infection and primary rifampin mono-resistant tuberculosis has been previously reported (20, 21), but the factors contributing to this association had not been systematically elucidated. By matching cases and controls on HIV status, we sought to define specific risks for rifampin mono-resistance in HIV positive persons. In this group, rifampin mono-resistance was associated with use of rifabutin for prevention or treatment of M. avium complex infection. With the licensure of rifabutin, concern was expressed about the possibility that its increased use would lead to rifampin-resistant tuberculosis in persons receiving the drug (22), and two case reports describe a person who appears to have developed rifampin mono-resistant tuberculosis while receiving rifabutin prophylaxis for M. avium complex, although in one of these case reports the patient had prior rifampin therapy as well (9, 11). We found rifabutin use in 11 (29%) of 38 HIV positive patients, and in none of the controls.

In addition to rifabutin therapy, prior antifungal therapy and diarrhea were independently associated with rifampin mono-resistance. A strong association with diarrhea was seen among patients who had also received rifabutin therapy. Malabsorption of antituberculosis agents has been described in patients with AIDS (23), and it may have contributed to the occurrence of rifampin mono-resistant tuberculosis in some of these patients. The exact mechanism by which this would lead to rifampin resistance remains unclear. It has been demonstrated that the appropriate treatment of active tuberculosis requires multidrug therapy (2). Conditions that preferentially alter the metabolism of any of these drugs could lead to the emergence of drug-resistant isolates. Malabsorption of rifabutin used alone may potentially have a role in the selection for rifampin mono-resistant M. tuberculosis in infected persons, and such malabsorption could result from HIV enteropathy or gut infection with opportunistic pathogens. Drug-drug interactions between rifamycins and azole compounds have been described (28, 29). This interaction may also have played a role in the development of isolated rifampin resistance in some of these patients, again by unknown mechanisms.

The emergence of this unusual pattern of drug resistance appears to be multifactorial. The near universal use of rifampin-containing regimens for the treatment of tuberculosis in this country for more than 20 yr may have provided the selective pressure for the emergence of rifampin mono-resistant clones. It is possible that these existed previously but are now detected because of increased availability of drug susceptibility testing and reporting of results. In addition, the increased rates observed in HIV positive patients may be partially explained by the accelerated rate of progression to disease in patients infected with M. tuberculosis (30). It remains to be determined if the future occurrence of this drug pattern may be seen with similar frequency in persons without HIV infection.

Of the cases included in the analysis, there was no laboratory evidence of spread of a single clone of M. tuberculosis. Multiple strains were detected by both DNA fingerprinting and sequencing of the region of the rpoB gene conferring rifampin resistance. Finally, previous case reports have suggested a possible link between the occurrence of rifampin mono-resistant tuberculosis and prior rifabutin use (9, 11); the size of the present investigation allowed us to document a definite association between prior rifabutin use by HIV-infected persons and rifampin mono-resistant tuberculosis. Given this risk, alternative therapies for the prevention of M. avium complex disease such as the use of azithromycin or clarithromycin may be preferred in HIV infected persons who are at high risk for coinfection with M. tuberculosis (31). Although prior tuberculosis and rifabutin use were identified as risk factors for mono-rifampin resistance, these factors could not explain the majority of mono-rifampin resistant cases in this study. Our observations suggest that the emergence of this unusual pattern of drug resistance may also be related to additional factors not assessed in this study. Further prospective studies with attention to evidence for primary acquisition of rifampin mono-resistant strains of M. tuberculosis and to drug interactions and serum drug levels may offer additional answers to the etiology of this emerging problem of drug resistant tuberculosis.