Background And Scientific Rationale

Frequency of Coexisting TB and HIV Infection and Disease in the United States
In the United States, epidemiologic evidence indicates that the HIV epidemic contributed substantially to the increased numbers of TB cases in the late 1980s and early 1990s (6,7). Overlap between the acquired immunodeficiency syndrome (AIDS) and TB epidemics continues to result in increases in TB morbidity. Analysis of national HIV-related TB surveillance data is limited by incomplete reporting of HIV status for persons with TB. As an alternative, state health department personnel have compared TB and AIDS registries to help estimate the proportion of persons reported with TB who are also infected with HIV. In the most recent comparison conducted by the 50 states and Puerto Rico, 14% of persons with TB in 1993-1994 (27% among those aged 25-44 years) also appeared in the AIDS registry (8). This proportion of TB patients with AIDS is believed to be a minimum estimate for the United States and might represent an increase in the proportion of TB patients identified as having TB and AIDS in 1990 (9%) (6). During 1993-1994, most persons with TB and AIDS (80%) were found in eight reporting areas: New York City, California, Florida, Georgia, Illinois, New Jersey, New York, and Texas (8).

In prospective epidemiologic studies, investigators have estimated that the annual rate of TB disease among untreated, tuberculin skin-test (TST)-positive, HIV-infected persons in the United States ranges from 1.7 to 7.9 TB cases per 100 person-years (Table_1) (9-11). The variability observed in these studies mirrors the differences in TB prevalence observed for different U.S. populations (i.e., the highest case rate was found in a study of a New York City population of intravenous drug users at a time when the incidence of TB was high and increasing 9{}; and the lowest case rate was evident in a community-based cohort of persons enrolled in a study of the pulmonary complications of HIV infection at a time and in a population in which the incidence of TB was relatively low 11{}). However, in all of these studies, the rate of TB disease among HIV-infected, TST-positive persons was approximately 4-26 times higher than the rate among comparable HIV-infected, TST-negative persons, and it was approximately 200-800 times higher than the rate of TB estimated for the U.S. population overall (0.01%) (12). Therefore, activities to control and eliminate TB in the United States must include aggressive efforts to identify HIV-infected persons with latent TB infection and to provide them with therapy to prevent progression to active TB disease.

Rates of Drug-Resistant TB Among HIV-Infected Persons in the United States
Resistance to antituberculosis drugs is an important consideration for some HIV-infected persons with TB. According to the results of a study of TB cases reported to CDC from 1993 through 1996, the risk of drug-resistant TB was higher among persons with known HIV infection compared with others (13). During this 4-year period, among U.S.-born persons aged 25-44 years with TB, HIV test results were reported as positive for 32% of persons, negative for 23%, and unknown for 45%. Using univariate analysis that excluded patients known to have had a previous episode of TB, investigators found that patients known to be HIV seropositive had a significantly higher rate of resistance to all first-line antituberculosis drugs, compared with HIV-seronegative patients and patients with unknown HIV serostatus (Table_2). Moreover, using a multivariate model that included age, history of previous TB, birth country, residence in New York City, and race/ethnicity, the investigators confirmed HIV-positive serostatus as a risk factor for resistance to at least isoniazid, for both isoniazid and rifampin resistance (multidrug-resistant MDR{} TB) and for rifampin monoresistance (TB resistant to rifampin only). In some areas of the United States with a low level of occurrence of MDR TB, however, differences in MDR TB related to HIV status have not been found (8). Reasons for the increased risk for TB drug resistance among HIV-seropositive persons might reflect a higher proportion of TB disease resulting from recently acquired M. tuberculosis infection (14,15) and thus an increased risk of disease caused by drug-resistant strains in areas with high community and institutional transmission of drug-resistant strains of M. tuberculosis (16). Several well-described outbreaks of nosocomially transmitted MDR TB, primarily affecting persons with AIDS, support this association (17-21).

In the past decade, reports have increased of TB caused by strains of M. tuberculosis resistant to rifampin only, and growing evidence has indicated that this rare event is associated with HIV coinfection (22-32). In retrospective studies, nonadherence with TB therapy has been associated with acquired rifampin monoresistance (22-24); and among a small number of patients, the use of rifabutin as prophylaxis for Mycobacterium avium complex was associated with the development of rifamycin resistance (31). However, the occurrence of TB relapse with acquired rifampin monoresistance also has been documented among patients with TB who initially had rifampin-susceptible isolates and who were treated with a rifampin-containing TB regimen by directly observed therapy (DOT) (30,32). The mechanisms involved in the development of acquired rifampin monoresistance are not clearly understood but could involve the persistence of actively multiplying mycobacteria in patients with severe cellular immunodeficiency, selective antituberculosis drug malabsorption, and inadequate tissue penetration of drugs.

Thus, of critical importance for HIV-infected persons is implementation of TB prevention and control strategies such as a) appropriate use of therapy for latent M. tuberculosis infection, b) early diagnosis and effective treatment of active TB (i.e., administering four-drug antituberculosis regimens by DOT to all coinfected patients), and c) prompt compliance with requirements for reporting TB cases and drug-susceptibility test results. Implementing these strategies for persons coinfected with HIV will not only help reduce new cases of TB in general; it also could decrease further transmission of drug-resistant strains and new cases of drug-resistant TB.

Copathogenicity of TB and HIV Disease
Human immunodeficiency virus type 1 (HIV-1) and M. tuberculosis are two intracellular pathogens that interact at the population, clinical, and cellular levels. Initial studies of HIV-1 and TB emphasized the impact of HIV-1 on the natural progression of TB, but mounting immunologic and virologic evidence now indicates that the host immune response to M. tuberculosis enhances HIV replication and might accelerate the natural progression of HIV infection (33). Therefore, the interaction between these two pathogens has important implications for the prevention and treatment of TB among HIV-infected persons. Studies of the immune response in persons with TB disease support the biologic plausibility of copathogenesis in dually infected persons. The initial interaction between the host immune system and M. tuberculosis occurs in the alveolar macrophages that present mycobacterial antigens to antigen-specific CD4+ T cells (34). These T cells release interferon-gamma, a cytokine that acts at the cellular level to activate macrophages and enhance their ability to contain mycobacterial infection. The activated macrophages also release proinflammatory cytokines, such as tumor necrosis factor and interleukin (IL)-1, cytokines that enhance viral replication in monocyte cell lines in vitro (35-38). The mycobacteria and their products also enhance viral replication by inducing nuclear factor kappa-B, the cellular factor that binds to promoter regions of HIV (39,40).

When TB disease develops in an HIV-infected person, the prognosis is often poor, though it depends on the person's degree of immunosuppression and response to appropriate antituberculosis therapy (41-43). The 1-year mortality rate for treated, HIV-related tuberculosis ranges from 20% to 35% and shows little variation between cohorts from industrialized and developing countries (44-49). The observed mortality rate for HIV-infected persons with TB is approximately four times greater than the rate for TB patients not infected with HIV (44,46,49,50). Although the cause of death in the initial period of therapy can be TB (46), death after the induction phase of antituberculosis therapy usually is attributed to complications of HIV other than TB (45,51,52). Epidemiologic data suggest that active TB accelerates the natural progression of HIV infection. In a retrospective cohort study of HIV-infected women from Zaire, investigators estimated the relative risk of death to be 2.7 among women with active TB compared with those without TB (53). In a retrospective cohort study of HIV-infected subjects from the United States, active TB was associated with an increased risk for opportunistic infections and death (54). The risk of death, or hazard rate, for persons with HIV-related TB follows a bimodal distribution, peaking within the first 3 months of antituberculosis therapy and then again after 1 year (48); the reasons for this distribution are not clear but might relate to the impact of TB on HIV disease progression. The observation that active TB increases deaths associated with HIV infection has been corroborated in studies of three independent cohorts in Europe (55-57).

Early in the HIV epidemic, researchers postulated that the immune activation resulting from concurrent infection with parasitic or bacterial pathogens might alter the natural progression of HIV infection (58). Subsequent observations have demonstrated that immune activation from TB enhances both systemic and local HIV replication. In some patients with active TB, the plasma HIV RNA level rises substantially before TB is diagnosed (59). Moreover, TB treatment alone leads to reductions in the viral load in these dually infected patients. TB and HIV also interact in the lungs, the site of primary infection with M. tuberculosis. In a recently published study of HIV-infected patients with TB, researchers found that the viral load was higher in the bronchoalveolar lavage fluid from the affected versus the unaffected lung and was correlated with levels of tumor necrosis factor in bronchoalveolar fluid (60). Researchers used V3 loop viral sequences to construct a phylogenetic tree and observed that the HIV quasispecies from the affected lung differed from those in the plasma within the same patient. These data suggest that pulmonary TB might act as a potent stimulus for the cellular-level replication of HIV. In summary, recent research findings have improved clinicians' understanding of how HIV affects the natural progression of TB and how TB affects the clinical course of HIV disease, and these findings support the recommendation for prevention, early recognition, and effective treatment for both diseases.

TB Therapy Outcomes Among Patients with HIV-Related TB
Among patients treated for TB, early clinical response to therapy and the time in which M. tuberculosis sputum cultures convert from positive to negative appear to be similar for those with HIV infection and those without HIV infection (30,61,62). However, the data are less clear about whether rates of TB relapse (recurrence of TB following successful completion of treatment) differ among patients with or without HIV infection (63). Current CDC and American Thoracic Society guidelines recommend a 6-month treatment regimen for drug-susceptible TB disease for patients coinfected with HIV (2) but suggest prolonged treatment for patients who have a delayed clinical and bacteriologic response to antituberculosis therapy. Some experts have suggested that to ensure an optimal antituberculosis treatment outcome, all patients with HIV-related TB should be treated with a longer course of therapy (i.e., 9 months), regardless of evidence of early response to therapy (64,65).

To make a recommendation on duration of therapy for HIV-related TB, expert consultants at the September 1997 CDC meeting considered the results of prospective studies that ascertained the posttreatment relapse rate following 6-month TB therapy regimens among patients with HIV infection (Table_3) (29,30,49,66,67). Differences in the study designs, including those pertaining to eligibility for enrollment in the study and to the definition of TB relapse, limited the analysis of combined results from the five studies. Despite this limitation, the expert consultants were able to make the following observations: a) the studies had a posttreatment follow-up duration that ranged from 8 to 22 months (median duration: 18 months); b) in three studies (30,49,67), investigators found that 6-month TB regimens were associated with a clinically acceptable (less than or equal to 5.4%) TB relapse rate; and c) in two studies (29,66), researchers found a high greater than or equal to 9%) TB relapse rate associated with the use of 6-month TB regimens. In the Zaire study (66), TB patients coinfected with HIV had almost two-fold higher posttreatment relapse rates than patients not infected with HIV who received the same TB treatment regimen; however, the authors did not investigate whether the relapses were the result of a recurrence of disease with the same strain of M. tuberculosis or reinfection (new disease) with a different strain. In the other study (29), which enrolled HIV-seropositive patients from 21 different sites in the United States, in all three patients who relapsed, the strain of M. tuberculosis isolated during the relapse episode matched, by DNA fingerprint, the strain of M. tuberculosis that was isolated during the initial episode of TB; this finding ruled out the possibility of reinfection.

The expert consultants who reviewed the available data agreed that short-course (i.e., 6-month) regimens should be used for the treatment of HIV-related pansusceptible TB (i.e., susceptible to all first-line antituberculosis drugs) in the United States, where patients are usually treated with DOT and where response to antituberculosis drugs can be monitored. This approach limits the use of lengthier multidrug antituberculosis therapies to the minimum possible number of patients with TB and HIV disease. Some experts believe the risk of TB treatment failure is increased among patients with advanced HIV-related immunosuppression and therefore advocate greater caution (or longer duration of therapy) when treating such patients for TB. The available data do not permit CDC to make a definitive recommendation regarding this issue. However, the experts recommended that clinicians treating TB in patients with HIV infection should consider the factors that increase a person's risk for a poor clinical outcome (e.g., lack of adherence to TB therapy, delayed conversion of M. tuberculosis sputum cultures from positive to negative, and delayed clinical response) when deciding the total duration of TB therapy.

Paradoxical Reactions Associated with Initiation of Antiretroviral Therapy During the Course of TB Therapy
The temporary exacerbation of TB symptoms and lesions after initiation of antituberculosis therapy -- known as a paradoxical reaction -- has been described as a rare occurrence (68-74) attributed to causes such as recovery of the patient's delayed hypersensitivity response and an increase in exposure and reaction to mycobacterial antigens after bactericidal antituberculosis therapy is initiated (75). Recently, a similar phenomenon was reported among patients with HIV-related TB (76). These reactions appear to be related more often to the concurrent administration of antiretroviral and antituberculosis therapy and occur with greater frequency than do paradoxical reactions associated primarily with the administration of antituberculosis therapy. Patients with paradoxical reactions can have hectic fevers, lymphadenopathy (sometimes severe), worsening of chest radiographic manifestations of TB (e.g., miliary infiltrates, pleural effusions), and worsening of original tuberculous lesions (e.g., cutaneous and peritoneal). However, these reactions are not associated with changes in M. tuberculosis bacteriology (i.e., no change from negative to positive culture and smear), and patients generally feel well and have no signs of toxicity.

In a prospective study, paradoxical reactions were more common among 33 patients with HIV-related TB who received TB treatment and combination antiretroviral therapy (36%) than among 55 patients not infected with HIV who received antituberculosis drugs alone (2%) and among 28 HIV-infected patients (historical control patients during pre-zidovudine era) who received antituberculosis drugs alone (7%) (76). Furthermore, among patients treated for both diseases, the paradoxical reactions were more temporally related to the initiation of combination antiretroviral therapy (mean +/- standard deviation SD{}: 15 +/- 11 days afterward) than to the initiation of antituberculosis treatment (mean SD: 109 +/- 72 days afterward). Researchers investigated potential causes for these symptoms and lesions (i.e., TB treatment failure, antituberculosis drug resistance, nonadherence with TB therapy, drug fever, development of conditions not related to TB or HIV) but considered such causes unlikely because these evaluations produced negative results, and TB was cured in patients who remained on unmodified antituberculosis regimens. Among patients in this study who received combination antiretroviral therapy, which usually included a protease inhibitor, the paradoxical reactions corresponded with a concurrent drop in HIV viral loads after antiretroviral therapy began and, in all but one patient, occurred while peripheral blood CD4+ T-cell counts were less than 200 cells/uL (76). In the historical control group (i.e., patients who were treated for TB but not for HIV), two (7%) of the 28 patients had a paradoxical reaction after antituberculosis therapy was initiated. This finding indicates that treatment of TB alone might sometimes decrease HIV viral load substantially and improve immune function (40,59,68,76).

After reviewing information about paradoxical reactions occurring during the course of TB therapy, expert consultants at the September 1997 CDC meeting concluded that exacerbation of TB signs and symptoms in patients with HIV-related TB can occur soon after combination antiretroviral therapy is initiated. Clinicians should always conduct a thorough investigation to eliminate other etiologies before making a diagnosis of paradoxical treatment reaction. For patients with paradoxical reactions, rarely are changes in antituberculosis or antiretroviral therapy needed. If the lymphadenopathy or other lesions are severe, one option is to continue with appropriate antituberculosis therapy and administer short-term steroids that suppress the enhanced immune response.

In the prospective study (76), despite having low CD4+ T-cell counts, six (86%) of seven TB patients who were initially tuberculin skin-test (TST)-negative had positive TST results after combination antiretroviral therapy was started. The reaction sizes of postantiretroviral TSTs ranged from 7 to 67 mm of induration. Clinicians must be aware of the potential public health and clinical implications of restored TST reactivity among persons who have not been diagnosed with active TB but who might be latently infected with M. tuberculosis. Persons previously known to have negative TST results might benefit from repeat tuberculin testing if they have evidence of restored immune function after antiretroviral therapy is initiated, because TB preventive therapy is recommended for TST-positive HIV-infected persons.

Considerations for TB Therapy for HIV-Infected Patients Treated with Antiretroviral Agents

Drug Interactions Between Rifamycins Used for TB Therapy and Antiretroviral Drugs Used for HIV Therapy
Widely used antiretroviral drugs available in the United States include protease inhibitors (saquinavir, indinavir, ritonavir, and nelfinavir) and nonnucleoside reverse transcriptase inhibitors (NNRTIs) (nevirapine, delavirdine, and efavirenz). Protease inhibitors and NNRTIs have substantive interactions with the rifamycins (rifampin, rifabutin, and rifapentine) used to treat mycobacterial infections (3,77). These drug interactions principally result from changes in the metabolism of the antiretroviral agents and the rifamycins secondary to induction or inhibition of the hepatic cytochrome CYP450 enzyme system (78,79). Rifamycin-related CYP450 induction decreases the blood levels of drugs metabolized by CYP450. For example, if protease inhibitors are administered with rifampin (a potent CYP450 inducer), blood concentrations of the protease inhibitors (all of which are metabolized by CYP450) decrease markedly, and most likely the antiretroviral activity of these agents declines as well. Conversely, if ritonavir (a potent CYP450 inhibitor) is administered with rifabutin, blood concentrations of rifabutin increase markedly, and most likely rifabutin toxicity increases as well.

Of the available rifamycins, rifampin is the most potent CYP450 inducer; rifabutin has substantially less activity as an inducer; and rifapentine, a newer rifamycin, has intermediate activity as an inducer (80-82). The four currently approved protease inhibitors and amprenavir (141W94, an investigational agent in Phase III clinical trials) are all, in differing degrees, inhibitors of CYP450 (83,84). The rank order of the agents in terms of potency in inhibiting CYP450 is ritonavir (the most potent); amprenavir, indinavir, and nelfinavir (with approximately equal potencies); and saquinavir (the least potent). The magnitude of the effects of coadministering rifamycins and protease inhibitors has been evaluated in limited pharmacokinetic studies (Table_4) (85-91). The three approved NNRTIs have diverse effects on CYP450: nevirapine is an inducer, delavirdine is an inhibitor, and efavirenz is both an inducer and an inhibitor. The magnitude of the effects of coadministering rifamycins and NNRTIs has also been evaluated in pharmacokinetic studies or has been predicted on the basis of what is known about their potential for inducing or inhibiting CYP450 (Table_5) (92-96).

In contrast to the protease inhibitors and the NNRTIs, the other class of antiretroviral agents available, nucleoside reverse transcriptase inhibitors (NRTIs) (zidovudine, didanosine, zalcitabine, stavudine, and lamivudine) are not metabolized by CYP450. Rifampin (and to a lesser degree, rifabutin) increases the glucuronidation of zidovudine and thus slightly decreases the serum concentration of zidovudine (97-100). The effect of this interaction probably is not clinically important, and the concurrent use of NRTIs and rifamycins is not contraindicated (77). Also, no contraindication exists for the use of NRTIs, NNRTIs, and protease inhibitors with isoniazid, pyrazinamide, ethambutol, or streptomycin. These first-line antituberculosis medications, in contrast to the rifamycins, are not CYP450 inducers.

Coadministration of Antituberculosis and Antiretroviral Therapies
According to 1998 U.S. Department of Health and Human Services guidelines on the use of antiretroviral agents among HIV-infected adults and adolescents (4), to improve the length and quality of patients' lives, all persons with symptomatic HIV infection should be offered antiretroviral therapy. HIV-infected patients with TB fall in this category. When used appropriately, combinations of potent antiretroviral agents can effect prolonged suppression of HIV replication and reduce the inherent tendency of HIV to generate drug-resistant viral strains. However, as antiretroviral therapeutic regimens have become increasingly effective, they also have become increasingly complex in themselves as well as in the problems they cause for the treatment of other HIV-associated diseases.

At present, regimens that include two NRTIs combined with a potent protease inhibitor (or, as an alternative, combined with an NNRTI) are the preferred choice for combination antiretroviral therapy for the majority of patients. Each of the antiretroviral drug combination regimens must be used according to optimum schedules and doses (4) because the potential for resistant mutations of HIV decreases if serum concentrations of the multiple antiretroviral drugs are maintained steadily. Because rifampin markedly lowers the blood levels of these drugs and is likely to result in suboptimal antiretroviral therapy, the use of rifampin to treat active TB in a patient who is taking a protease inhibitor or an NNRTI is always contraindicated. Rifabutin is a less potent inducer of the CPY450 cytochrome enzymes than is rifampin and, when used in appropriately modified doses, might not be associated with a clinically significant reduction of protease inhibitors or nevirapine (Table_6). Thus, the substitution of rifabutin for rifampin in TB treatment regimens has been proposed as a practical choice for patients who are also undergoing therapy with protease inhibitors (with the exception of ritonavir 86,87,101-103{} or hard-gel capsule saquinavir Invirase (85){}) or with the NNRTIs nevirapine or efavirenz (but not delavirdine 93,94{}). Currently, more clinical and pharmacokinetic data are available on the use of indinavir or nelfinavir with rifabutin than on the use of amprenavir or soft-gel saquinavir (Fortovase ) with rifabutin. Rifapentine is not recommended as a substitute for rifampin because its safety and effectiveness have not been established for the treatment of patients with HIV-related TB. As an alternative to the use of rifamycin for the treatment of TB, the use of streptomycin-based regimens that do not contain rifamycin can be considered for the treatment of TB in patients undergoing antiretroviral therapy with protease inhibitors or NNRTIs.

Use of Rifabutin-Based Regimens for the Treatment of HIV-Related TB
At present, TB drug regimens that include rifabutin instead of rifampin appear to offer the best alternative for the treatment of active TB among patients taking antiretroviral therapies that include protease inhibitors or NNRTIs. This recommendation is based on findings from studies of equivalent in vitro antituberculosis activity of rifabutin and rifampin (104,105) and the results of three clinical trials (106-108). These trials demonstrated that 6-month rifabutin-containing regimens (at a daily dose of either 150 mg or 300 mg) were as effective and as safe as similar control regimens containing rifampin for the treatment of TB (Table_7). The smallest (n=49) of these three trials was conducted in Uganda (108) and is the only one to include HIV-coinfected patients (who were not undergoing antiretroviral therapy at the time of the study). This study indicated that 81% of patients taking a TB treatment regimen containing daily rifabutin converted their sputum from M. tuberculosis positive to negative after 2 months of treatment, compared with a 48% sputum conversion rate among patients taking a TB regimen containing daily rifampin (pless than 0.05). However, when the researchers controlled for differences in baseline characteristics (a greater proportion of patients in the rifampin group had cavitary disease), they found no difference in the time to sputum conversion between the two study groups.

Studies are under way to evaluate the use of rifabutin administered daily (at a dose of 150 mg) or twice a week (at a dose of 300 mg) for the treatment of TB in HIV-infected patients who take protease inhibitors. Physicians at a state tuberculosis hospital in Florida have treated or consulted on the treatment of approximately 30 HIV-infected patients who received a protease inhibitor while undergoing treatment for TB with rifabutin. Patients have been treated for TB primarily with administration of rifabutin (150 mg daily) as part of four-drug therapy for 2-4 weeks, followed by rifabutin (300 mg twice weekly) as part of four-drug therapy to complete 8 weeks of induction, and then a continuation phase consisting of twice-weekly isoniazid and rifabutin (300 mg) to complete 6 months of treatment. To date, patients treated with this regimen have not experienced clinically significant increases in rifabutin serum levels, have had a minimal incidence of adverse reactions from rifabutin (one patient developed a case of uveitis), and have had a good clinical response to TB and HIV therapies. Approximately 80% of the patients attained sputum conversion by the second month of treatment, most have attained and maintained suppression of HIV replication, and no TB relapses have occurred with up to 1 year of posttreatment follow-up (David Ashkin, M.D., and Masahiro Narita, M.D., A.G. Holley State Tuberculosis Hospital, Lantana, Florida, personal communication, 1998). In previous reports, CDC and the American Thoracic Society jointly recommended the use of rifampin-containing short-course regimens for the initial treatment of HIV-related TB (2). The inclusion of rifampin in regimens to treat TB was supported by data collected from approximately 90 controlled clinical trials conducted from 1968 to 1988 (109). Excluding rifampin from the TB treatment regimen was not recommended because regimens not containing rifampin a) had not been proven to have acceptable efficacy (i.e., have been associated with higher rates of TB treatment failure and death and with slower bacteriologic responses to therapy leading to potential increases in the likelihood of M. tuberculosis transmission) and b) require prolonging duration of therapy from 6 months to 12-15 months. Presently, available data suggest that rifabutin in short-course (i.e., 6 months) multidrug regimens to treat TB provides the same benefits as the use of rifampin. Three additional reasons support the use of rifabutin for treating HIV-related TB: a) observations suggest that rifabutin might be more reliably absorbed than rifampin in patients with advanced HIV disease (110,111); b) the use of rifabutin appears to have been better tolerated in patients with rifampin-induced hepatotoxicity (David Ashkin, M.D., and Masahiro Narita, M.D., A.G. Holley State Tuberculosis Hospital, Lantana, Florida, personal communication, 1998); and c) the use of rifabutin might lessen the possibility of interactions with other medications commonly prescribed for patients with HIV infection (e.g., azole antifungal drugs, anticonvulsant agents, and methadone) (77).

Use of Alternative TB Treatment Regimens that Contain Minimal or No Rifamycin
TB treatment regimens that contain no rifamycins have been proposed as an alternative for patients who take protease inhibitors or NNRTIs. Several clinical trials conducted in Hong Kong and Africa by the British Medical Research Council and published from 1974 through 1984 provide information about nonrifamycin and minimal-rifamycin regimens for the treatment of TB in patients who were not likely to be infected with HIV (Table_8) (112-115). Most of these studies demonstrated high relapse rates when regimens not containing streptomycin were used and when the duration of therapy was less than 9 months. However, in a large (n=404) randomized controlled clinical trial in Hong Kong that evaluated the use of six TB treatment regimens consisting of streptomycin, isoniazid, and pyrazinamide either daily, three times a week, or two times a week for 6 or 9 months (112), almost all patients treated with any of the study regimens achieved rapid sputum conversions (86%-94% of patients converted within 3 months of therapy). In this study, the 30-month posttreatment follow-up relapse rates were high 18%-24%) among patients treated with 6-month regimens, but the relapse rates among patients treated with 9-month regimens (5%-6%) were similar to the relapse rates expected following the use of rifampin-based TB treatments. Thus, the expert consultants who developed these guidelines concluded that treatment of TB without rifamycin always requires longer-duration (at least 9 months) regimens that include streptomycin (or an injectable antituberculosis drug such as capreomycin, amikacin, or kanamycin) (63). However, these TB regimens have not been studied among patients with HIV infection.

Streptomycin is highly bactericidal against M. tuberculosis, but it is rarely used in the United States to treat drug-susceptible TB because of problems associated with its administration by injection that can be intensified in patients with low body mass or wasting and because of potential ototoxicity and nephrotoxicity. The associated potential toxicities and increased duration of therapy and the patient's difficulty in adhering to an injectable-drug-based TB regimen can compromise the effectiveness of streptomycin-based TB regimens, and these limitations should be considered by physicians and patients.

Treatment of Latent M. tuberculosis Infection in Patients with HIV Infection Scientific Rationale
Preventive therapy for TB is essential to controlling and eliminating TB in the United States (116,117). Treatment for HIV-infected persons who are latently infected with M. tuberculosis is an important part of this strategy and is also an important personal health intervention because of the serious complications associated with active TB in HIV-infected persons (118-120). Expert consultants attending the September 1997 CDC meeting and additional consult ants attending a September 1998 meeting sponsored by the American Thoracic Society and CDC considered findings from multiple studies (121-130) before developing recommendations about the optimal duration of isoniazid preventive therapy regimens; frequency of administration (intermittency) of preventive therapy; new short-course multidrug regimens; and preventive therapy for anergic HIV-infected adults with a high risk of M. tuberculosis infection.

A key difference in most preventive therapy trials conducted before and after the beginning of the HIV epidemic is that the earlier trials focused on 12-month regimens of isoniazid, whereas five of seven trials (122-126) conducted in HIV-infected populations assessed 6-month regimens of isoniazid (Table_9). Four of these 6-month isoniazid regimens (122-125) were chosen for study on the basis of the operational feasibility of providing therapy in countries with limited resources where preventive therapy programs were not available; the fifth study (126), a U.S. trial conducted among anergic patients, used a 6-month regimen because of the absence of previous data about optimal duration of therapy for TST-negative, HIV-infected patients. Despite these variations, the expert consultants concluded that the findings from these different preventive therapy studies should apply to most persons with latent M. tuberculosis infection, regardless of their HIV serostatus, because similar levels of protection have been observed when identical preventive therapy regimens have been administered to persons infected with HIV and those not infected.

Optimal Duration of Isoniazid Regimens for Treatment of Latent M. tuberculosis Infection
The American Thoracic Society and CDC have previously recommended a regimen of 12 months of isoniazid alone for treatment of latent M. tuberculosis infection in HIV-infected adults (2). The recommended duration of TB preventive therapy for persons not infected with HIV was a minimum of 6 months. When considering the optimal duration of isoniazid preventive therapy, the consultants reviewed findings from two studies conducted in populations not known to be infected with HIV (128,129). One of these studies, a controlled trial conducted in seven European countries, compared the efficacy of three durations (3, 6, and 12 months) of isoniazid preventive treatment for TST-positive persons with stable, fibrotic lesions on chest radiographs (128). In this study, compliant patients who received medication for 12 months had better protection against TB (93%) than those who received medication for 6 months (69%). The other study was conducted among the Inuits in the Bethel area of Alaska, where participants received 0-24 months of isoniazid preventive therapy (129). In an assessment of observed posttherapy case rates of TB relative to the amount of isoniazid ingested (expressed as a percentage of a 12-month regimen), researchers found that higher amounts of therapy corresponded with lower TB rates among participants who had received 0-9 months of isoniazid therapy; after 9 months of therapy, participants had no additional benefits in terms of decreased TB case rates.

Four studies of HIV-infected persons have evaluated 6-month and 12-month regimens of daily isoniazid (121,123,125,127). Both of the studies that evaluated a 6-month regimen included a placebo comparison group and demonstrated reductions in the incidence of TB among persons in the treatment group -- 70% in Uganda (123) and 75% in Kenya (125). A study of the 12-month regimen (121), which was conducted in Haiti and also included a placebo comparison group, demonstrated an 83% reduction in the incidence of TB among persons in the treatment group. A multicenter trial conducted in the United States, Mexico, Brazil, and Haiti (127) demonstrated that the magnitude of protection obtained from a regimen of isoniazid administered daily for 12 months was similar to that obtained from a regimen of rifampin and pyrazinamide administered daily for 2 months. Isoniazid preventive therapy regimens of 6 and 12 months' duration have not been compared with each other in the same study conducted among HIV-infected persons. In summary, these data indicate that a) the optimal duration of isoniazid preventive therapy should be greater than 6 months to provide the maximum degree of protection against TB; b) therapy for 9 months appears to be sufficient; c) therapy for greater than 12 months does not appear to provide additional protection.

Frequency of Administering Isoniazid Preventive Therapy
Two clinical trials (122,124) have evaluated 6-month twice-weekly isoniazid regimens for the prevention of active TB in HIV-infected persons (Table_9). Participants enrolled in the twice-weekly 6-month isoniazid arm of a study conducted in Zambia had a 40% reduction in the rate of TB compared with persons who took a placebo for 6 months (124). The findings of a trial conducted in Haiti (122) suggest that the magnitude of protection obtained from isoniazid administered twice a week for 6 months is equivalent to that obtained from rifampin and pyrazinamide regimens administered twice a week for 2 months. Preventive therapy trials that include twice-weekly isoniazid regimens for greater than 6 months or comparisons of the same drugs administered daily versus intermittently have not been conducted. However, in a Baltimore demonstration project in which isoniazid was administered twice a week (10-15 mg/kg, with a maximum dose of 900 mg) to a cohort of injecting-drug users under directly observed preventive therapy (DOPT), the findings support the efficacy of twice-weekly isoniazid preventive therapy (130). Twice-weekly regimens with DOPT were used in Baltimore because the project staff expected that supervised delivery of therapy would enhance adherence with and completion of the preventive therapy regimen. Thus, the available data suggest that the protection obtained from isoniazid preventive therapy regimens should be the same whether the drug is administered daily or twice a week.

Short-Course Multidrug Regimens for TB Preventive Therapy
Four clinical trials (122-124,127) conducted among HIV-infected populations have evaluated courses of preventive therapy that are shorter than 6 months and that include rifampin in combination with isoniazid or pyrazinamide (Table_9). The largest and most recent of these trials was a multicenter, randomized TB prevention study conducted from 1992 through 1998 (127). Researchers found identical rates of TB (1.2 per 100 person-years) in two groups of TST-positive, HIV-infected persons: those who primarily self-administered isoniazid daily for 12 months and those who primarily self-administered rifampin and pyrazinamide daily for 2 months. Both study groups had similar adverse events and mortality rates; persons taking rifampin and pyrazinamide for 2 months were significantly more likely (80%) to complete therapy than were persons taking isoniazid for 12 months (68%) (pless than 0.001). Two other trials conducted in Haiti and Zambia (122,124) have also evaluated regimens of rifampin and pyrazinamide for the prevention of TB but have not included comparison arms of 12-month isoniazid regimens. The study in Haiti (122) compared patients receiving rifampin and pyrazinamide administered twice a week for 2 months with patients receiving isoniazid twice a week for 6 months; in both arms of the study, one of the twice-weekly doses was administered by DOPT. Investigators observed no difference in TB risk or mortality among participants enrolled in the two treatment arms (122). The placebo-controlled trial in Zambia demonstrated comparable protection from 3 months of rifampin and pyrazinamide versus 6 months of isoniazid; both regimens were self-administered twice a week (124). In the multicenter trial (127) and in the Haiti and Zambia studies (122,124), regimens that included rifampin and pyrazinamide were well tolerated. In a study conducted in Uganda (123), investigators observed no statistically significant reduction in TB rates but a high rate of toxicity and drug intolerance among persons who took three drugs (isoniazid, rifampin, and pyrazinamide) daily for 3 months compared with persons who took a placebo (Table_9); 3 months of daily self-administered rifampin and isoniazid provided protection similar to that of 6 months of daily self-administered isoniazid. Thus, short-course multidrug regimens (i.e., two drugs for 2-3 months) have been shown to be effective for the prevention of active TB in HIV-infected persons. The use of three drugs for preventive therapy, however, can be associated with unacceptably high rates of toxicity, and the use of a 3-month regimen of rifampin and isoniazid is not being considered for use in the United States. Available data indicate that in the United States, a regimen of rifampin and pyrazinamide administered daily for 2 months is a reasonable treatment option for HIV-infected adults with latent M. tuberculosis infection. The available data do not permit CDC to make a definitive statement regarding the intermittent (i.e., twice a week) administration of a 2-month regimen of rifampin and pyrazinamide.

Preventive Therapy for Anergic HIV-Infected Adults with a High Risk of Latent M. tuberculosis Infection
Isoniazid preventive therapy has not been found to be useful or cost-effective in preventing TB when administered to anergic, HIV-infected persons (123,126) (Table_9). The anergic subjects who received isoniazid in the Uganda trial had a statistically insignificant (17%) reduction in the rate of TB (2.5 cases per 100 person-years) compared with patients in the placebo group (3.1 cases per 100 person-years) (123). Similarly, anergic HIV-infected persons with a high risk for tuberculous infection who were enrolled in a U.S. multicenter trial and treated with isoniazid daily for 6 months had a rate of TB (0.4 cases per 100 person-years) that was 50% less than, but not statistically different from, the rate observed among patients treated with placebo (0.9 cases per 100 person-years) (126). In both of these studies, HIV-infected persons with anergy tolerated isoniazid well, as suggested by the low rates of adverse reactions and high rates of therapy completion. These study findings do not support the routine use of preventive therapy in anergic, HIV-infected persons. Preventive therapy for TST-negative, HIV-infected persons also has not been proven effective (121,124, 125) (Table_9); however, some experts recommend primary preventive therapy (to prevent M. tuberculosis infection) for TST-negative or anergic HIV-infected residents of institutions that pose an ongoing high risk for exposure to M. tuberculosis (e.g., prisons, jails, homeless shelters).

Implications of Results of TB Preventive Therapy Trials
The effects of TB preventive therapy on mortality and progression of HIV infection appear to be limited, with the exception that such therapy can protect against the development of TB disease and its associated consequences. Moreover, the duration of this protective effect has not been clearly established for HIV-infected persons. Despite these limitations and uncertainties, preventive therapy is recommended because its benefits in preventing TB disease are thought to be greater than the risks of serious treatment-related adverse events, and such therapy benefits society by helping to prevent the spread of infection to other persons in the community.

The implementation of TB preventive therapy programs should be facilitated by the use of newly recommended short-course multidrug regimens and twice-weekly isoniazid regimens, especially among patients for whom DOPT is feasible. Because of the drug interactions between rifampin and protease inhibitors or NNRTIs, the use of shorter regimens containing rifampin is contraindicated for patients taking these antiretroviral drugs. Although preventive therapy trials evaluating rifabutin use among TST-positive, HIV-infected persons have not been conducted, the expert consultants reviewed available data and agreed that the use of rifabutin instead of rifampin is valid on the basis of the same scientific principles that support the use of rifabutin for the treatment of active TB.

previous page | next page

return to table of contents