A Randomized, Double-Blind Trial Comparing Azithromycin and Clarithromycin in the Treatment of Disseminated Mycobacterium avium Infection in Patients with Human Immunodeficiency Virus

Abstract

Two hundred and forty-six patients infected with human immunodeficiency virus (HIV) who also had disseminated Mycobacterium avium complex received either azithromycin 250 mg every day, azithromycin 600 mg every day, or clarithromycin 500 mg twice a day, each combined with ethambutol, for 24 weeks. Samples drawn from patients were cultured and clinically assessed every 3 weeks up to week 12, then monthly thereafter through week 24 of double-blind therapy and every 3 months while on open-label therapy through the conclusion of the trial. The azithromycin 250 mg arm of the study was dropped after an interim analysis showed a lower rate of clearance of bacteremia. At 24 weeks of therapy, the likelihood of patients' developing 2 consecutive negative cultures (46% vs. 56%, P = .24) or 1 negative culture (59% vs. 61%, P = .80) was similar for azithromycin 600 mg (n = 68) and clarithromycin (n = 57), respectively. The likelihood of relapse was 39% versus 27% (P = .21) on azithromycin compared with clarithromycin, respectively. Of the 6 patients who experienced relapse, those randomized to azithromycin developed isolates resistant to macrolides, compared with 2 of 3 patients randomized to clarithromycin. Mortality was similar in patients comprising each arm of the study (69% vs. 63%; hazard, 95.1% confidence interval, 1.1 [0.7, 1.7]). Azithromycin 600 mg, when given in combination with ethambutol, is an effective agent for the treatment of disseminated M. avium disease in patients infected with HIV.

Effective therapy for disseminated Mycobacterium avium complex (MAC) infection has progressed significantly over the last 10 years. Initial skepticism regarding the pathogenicity of this organism gave way to respect for the impact that disseminated disease had on morbidity and mortality [1, 2]. Therapeutic options had been limited and have included numerous second-line antimycobacterial agents that were of questionable potency and are associated with a variety of side effects when delivered as part of a complicated polypharmacy [3, 4].

The arrival of advanced-generation macrolides significantly improved the available therapeutic options. These compounds were potent in vitro and in animal models, and on the basis of experience in other settings, they had the potential to be acceptably tolerated and easily administered. After a number of clinical trials, clarithromycin, in combination with other antimycobacterial medications, has been accepted as a cornerstone of therapy for treatment of disseminated MAC in patients infected with HIV [3, 5–8].

In vitro testing and data from animal models has confirmed the potential activity of azithromycin in treatment of MAC [9, 10]. The results of a pilot study, and numerous anecdotal experiences, advanced these early impressions [11]. A dose-ranging study of 600 mg compared with 1200 mg concluded that there was a similar degree of microbiologic efficacy, though the 1200-mg daily dose was associated with a higher rate of gastrointestinal side effects [12]. No clinical data were available regarding the efficacy of azithromycin in lower doses, though steady-state WBC levels at a 250-mg daily dose were predicted to achieve concentrations well in excess of the 90% minimal inhibitory concentration (MIC₉₀) for M. avium [11]. On the basis of this background, a comparative efficacy trial was undertaken to assess the relative merits of 2 doses of azithromycin with clarithromycin, each regimen combined with ethambutol, in treatment of disseminated disease due to MAC.

Materials and Methods

Study design. This study was a randomized, double-blind, double-dummy study conducted at 55 sites in the United States, Brazil, Argentina, Chile, and The Netherlands. Patients were considered for enrollment if a local blood culture was positive for MAC within the previous 2 months and if they were infected with HIV, were ⩾13 years old, were expected to survive at least 2 months, had not received therapy for treatment of MAC since the positive culture, had alanine aminotransferase and aspartate aminotransferase <5 times the upper limits of normal, a creatinine level of <3.0 mg/dL, and a neutrophil count of >500 cells/mm³. Patients were excluded from consideration if they had a hypersensitivity to macrolides, were pregnant or lactating, were unable to take oral medications, had been previously treated for MAC, or had a condition likely to interfere with drug absorption.

The institutional review board of each participating center reviewed and approved participation in the trial. After written informed consent was obtained, patients were randomized to receive one of the following: azithromycin 250 mg every day administered orally, azithromycin 600 mg every day administered orally, or clarithromycin 500 mg twice a day administered orally. Patients were assigned on the basis of a randomization list created by a computer-generated algorithm provided by the sponsor (Central Pharmacy, Pfizer Central Research, Groton, CT). Each macrolide regimen also contained ethambutol at doses of 800 mg or 1200 mg every day administered orally, depending on the patient's weight. In addition to active therapy, each patient received matched placebos for the alternative macrolide regimen. Sites were provided medication in block sizes of 6, with an equal proportion of each regimen in each block. Sealed envelopes containing the treatment regimen were provided to each site, to be opened only in the case of an emergency. The sponsor monitored the integrity of these envelopes at each monitoring visit. The study remained blinded to patients, investigators, and monitors through its completion.

Before receiving the first dose of study drug, patients provided a medical history and underwent thorough physical examination and laboratory testing, which included mycobacterial blood cultures. Blood samples (9 mL in Isolator 10 tubes, Wampole Lab [Cranbury, NJ]) were sent for culture to a central microbiology lab (Kuzell Institute for Arthritis and Infectious Diseases, San Francisco, CA) for quantification of mycobacteria by use of the lysis-centrifugation methods described elsewhere [13]. Susceptibility testing was performed on all isolates at baseline and in association with treatment failure or relapse [14]. Organisms with an MIC of ⩾256 µg/mL for azithromycin and ⩾16 µg/mL for clarithromycin were considered resistant. Patients were to record any episode of fever or night sweats as well as any antipyretic use in a diary. Follow-up visits occurred at weeks 3, 6, 9, 12, 16, 20, and 24, at which time detailed clinical assessments were made, adverse events were recorded, and visual acuity testing was performed; repeat safety testing and mycobacterial functional health status questionnaire were completed at baseline and at week 24. Specimens for cultures were obtained from the patients. Audiograms were performed if there was a suspicion of any hearing loss. After 24 weeks of treatment, patients were switched to open-label therapy at the investigator's discretion but continued to be followed every 3 months either by clinic visits or through phone contact. Investigators were instructed to obtain isolates from any patient who relapsed after treatment or, at a minimum, the local laboratory culture report, including the results of any sensitivity testing.

Statistical analysis. The sample size of the 300 randomized subjects was determined by assuming that if 100 patients were enrolled in each of the 3 arms, there would be 75 patients with positive baseline cultures. The level of significance for the final analysis was adjusted for the interim analysis to .049, and if 75% of the patients on clarithromycin had a positive response, then for any azithromycin response that has the same true rate, the probability would be 0.81 that the lower limit of the 2-sided 95.1% confidence bound for the difference in response rates would remain above −20%. The primary end point was sterilization, defined as 2 consecutive negative blood cultures for MAC from the central laboratory at week 24. Missing data were carried forward from the previous visit. Analyses were performed on the modified intent-to-treat population, defined as all randomized patients who had a positive baseline culture for MAC. Secondary end points included the following: time to sterilization that used all known mycobacterial culture data, change from baseline in level of mycobacteremia, durability of sterilization, mortality, clinical response as judged by the investigator, change in quality of life, and patient tolerance for each regimen.

CIs for the difference in success rates were based on the normal approximation. Time-to-event analyses were compared on the basis of the log-rank test and the Cox proportional hazards model. Event-free (survival) curves were plotted with the product-limit (Kaplan-Meier) method. A covariate analysis was performed on the assessments of sterilization at week 24, time to sterilization, and time to death. Each covariate was examined for unbalanced allocation at baseline, significant effects on the response variable, and interaction effects specific to a treatment group. Quality-of-life assessments were derived from the General Health and Functioning Questionnaire. The Perceived Health Index (PHI) is an overall measure of health and is derived from 6 of the subscales included within the questionnaire [15]. The PHI and the overall quality-of-life index were compared by an analysis of covariance model. An interim analysis, undertaken when 50% of the patients completed the first 12 weeks of study, compared sterilization at week 12 for the purpose of determining if either or both azithromycin doses were inadequately efficacious.

Results

Randomized patients. The study was conducted between 26 August 1994 and 28 September 1998. Two hundred forty-six patients were randomized, of whom 239 received therapy. Treatment arms were balanced with respect to age, sex, prior opportunistic disease, hemoglobin, alkaline phosphatase, protease inhibitor use, and prophylaxis for Pneumocystis carinii pneumonia; however, subjects randomized to the clarithromycin treatment arm were more likely to have received previous macrolide prophylaxis than those in the azithromycin 250-mg treatment arm (table 1).

Table 1

Baseline demographic characteristics of patients infected with HIV randomized to receive azithromycin or clarithromycin to treat disseminated infection with Mycobacterium avium complex (MAC).

Interim analysis. An interim analysis was performed in July 1996 that included 150 patients (table 2). The results of this analysis demonstrate that the patients treated with azithromycin 250 mg were less likely to have 2 consecutive sterile blood cultures at week 12 than to patients who received either azithromycin 600 mg or clarithromycin. The proportion of patients who had died, (taking into account all information available up to that date) was higher among patients on the azithromycin 250-mg regimen than among patients who received either other regimen, although this difference was not statistically significant. As a consequence of meeting the criteria stated in a predefined stopping rule, enrollment in the azithromycin 250-mg group was closed. Patients who remained on that regimen were removed from the trial and subsequent enrollment was divided between the other regimens.

Table 2

Interim analysis of patients infected with HIV given azithromycin and clarithromycin to treat disseminated Mycobacterium avium infection who were included in the intent-to-treat population.

Baseline comparisons. Enrollment into the remaining 2 regimen groups (azithromycin 600 mg and clarithromycin) was stopped at 181 subjects, 90% of the projected target, because of the difficulty of identifying patients with disseminated M. avium disease coincident with the availability of highly active antiretroviral therapy (HAART) and the broader use of mycobacterial prophylaxis. In the subset of patients that were also included in the intent to treat (ITT) analysis of this comparison, specifically those with a positive baseline culture for MAC, each treatment regimen was balanced with regard to baseline demographic characteristics, including antiretroviral use and CD4 count (table 1), except among clinical signs and symptoms associated with MAC. Patients randomized to receive the 600-mg dose of azithromycin were more likely, in the week before randomization, to have had daily night sweats (24 [36%] vs. 11 [20%]) and nausea (17 [25%] vs. 4 [7%]) (P < .05) than were patients randomized received to clarithromycin. Trends (P < .10) toward a difference were also seen in the likelihood of vomiting, daily fever, daily diarrhea, and the overall perceived health index.

For the 119 baseline isolates of M. avium, the MIC₉₀ was 32 µg/mL for azithromycin and 2 µg/mL for clarithromycin. There was 1 resistant isolate at baseline in the azithromycin 600-mg treatment arm (MIC ⩾ 256 µg/mL) and 2 in the clarithromycin treatment arm (MIC ⩾ 16 µg/mL).

Microbiologic efficacy. The mean duration of treatment was 86 days and 69 days for patients who received azithromycin 600 mg and clarithromycin, respectively. The mean duration of follow-up was 355 days (median, 266 days). Mean and median baseline cfu per milliliter were similar at baseline. The proportion of subjects with a 1-log reduction or sterilization of cultures, and the time to that reduction were similar for the 2 groups. The overall mean and median reduction in bacteremia, as well as quantitative measurements at each visit, were also similar for the 2 groups (table 3; figure 1). Rates of sterilization, defined as either 2 consecutive negative cultures (the primary end point) or 1 negative culture occurring at the week 24 visit, demonstrated that azithromycin and clarithromycin had similar treatment effects, within the limits defined by the 95.1% CI on the difference in rates. An analysis of the hazard ratio on time to sterilization produced similar findings. There were too few subjects receiving protease inhibitors in this trial to make valid estimates of their effect on microbiologic response.

Table 3

Response to therapy of patients infected with HIV who received azithromycin 600 mg or clarithromycin to treat disseminated Mycobacterium avium infection.

Figure 1

Reduction in cfu per milliliter of Mycobacterium avium in blood from patients treated with ethambutol and either azithromycin (AZ, 600 mg q.d., dashed line) or clarithromycin (CL, 500 mg b.i.d., solid line). MAC, M. avium complex.

Relapse, as defined by a single positive culture after sterilization, was more common among patients in the azithromycin group than among those in the clarithromycin group, although the difference did not reach statistical significance. In many of these patients, cultures reverted to sterile without a change in therapy. Defining relapse as 2 consecutive positive cultures reduced any nominal difference between the treatment arms. Resistance to azithromycin or clarithromycin in breakthrough isolates from the first 24 weeks of therapy was not observed in the 6 patients in the azithromycin group who relapsed but was noted in 2 of 3 patients in the clarithromycin group who relapsed. These 2 isolates were cross-resistant to both macrolides.

Clinical resolution of signs and symptoms. The investigator assessment of clinical response was rated on a 7-point scale, from marked deterioration through marked improvement. With similar reasons for missing data on each regimen, ∼60% of the modified intent to treat (MITT) population of subjects at week 12 and 30% at week 24 were assessed for clinical response. At week 12, 28 (68.3%) of 41 of those in the azithromycin group were felt to have had a satisfactory response, compared with 29 (91%) of 32 of those in the clarithromycin group (P = .02). By week 24, 17 (71%) of 24 in the azithromycin group and 17 (73%) of 23 subjects in the clarithromycin group had improved (P = .81). Weight gain was observed in both groups (0.91 kg and 2.99 kg on azithromycin and clarithromycin, respectively; P = .11). Overall, the proportion of patients with fever and night sweats declined in both groups, though persistence of these symptoms at week 24 was observed in patients who had reported experiencing these symptoms on a daily basis at baseline. Patients in both groups had lower mean PHI scores at week 24 than at baseline (change from baseline of −3 for azithromycin vs. -8 for clarithromycin; P = .81).

Safety. Treatment-related side effects, primarily gastrointestinal, were observed in 53 (63%) of subjects in the azithromycin group 600 mg and 56 (66%) of those in the clarithromycin group. Discontinuations from therapy for reasons related to study drug occurred in 8 (10%) of patients who received azithromycin and 5 (6%) of those who received clarithromycin.

Survival. Mortality was similar in the 2 groups, both by the end of 24 weeks of study and at the time of last observation (table 3; figure 2). Covariate analyses revealed that use of protease inhibitors, lower baseline colony counts of MAC, lower alkaline phosphatase, higher hemoglobin levels, and fewer previous opportunistic infections were associated with a survival advantage. No impact of these variables on the relative effects of azithromycin and clarithromycin was identified.

Figure 2

Survival of patients with disseminated Mycobacterium avium complex assigned to ethambutol and either azithromycin (AZ, 600 mg q.d., dashed line) or clarithromycin (CL, 500 mg b.i.d., solid line).

Discussion

The purpose of this trial was to assess the safety and efficacy of 2 doses of azithromycin relative to each other and to clarithromycin for treatment of disseminated MAC in HIV-infected patients. We noted no significant differences in clinical and microbiologic responses between patients treated with the 600-mg azithromycin dose and those treated with clarithromycin. Although the azithromycin 250-mg dose did reduce circulating levels of mycobacteremia, it did not do so as rapidly as the 600-mg dose. The development of mycobacteremia is associated with an increase in viral load, and the level of viral load has been associated with increase in mortality [16, 17]. Although the mortality rates at the interim analysis were not noted to be different, a more sensitive assessment of time to death, performed for all subjects randomized to the azithromycin 250-mg arm, did reveal a significant effect on mortality compared with patients in the 600-mg arm (hazard, 1.84; P = .011; data not shown). This finding emphasizes the effect on survival of clearing mycobacteremia and supports results from previous studies in the value of optimizing antimycobacterial therapy [1].

Although the MIC₉₀ of M. avium at baseline for clarithromycin (2 µg/mL) was lower than that for azithromycin (32 µg/mL), the rate of reduction in bacteremia was similar for patients who received each regimen. This finding may reflect the importance of pharmacokinetic exposure relative to MIC, rather than MIC alone, in predicting in vivo potency [18].

Differences in the point estimate of sterilization made on the basis of a definition of 1 or 2 sterile cultures have been seen before [18, 19]. In this trial, some of the difference in sterilization rates is explained by more cultures in the azithromycin treatment arm with 1 cfu per milliliter of blood. The definition of relapse was similarly affected, and in sensitivity analyses performed on these data, when cultures with 1 cfu/mL were treated as sterile, relapse rates overall were reduced and relative differences in relapse rates between the azithromycin and the clarithromycin groups disappeared (data not shown). This observation underscores the difficulty in defining a sterilization end point that represents true clearance of bacterial burden, because a negative blood culture does not necessarily reflect sterility at all sites [20].

Resistance emergence in breakthrough isolates on study drug therapy was not observed in patients who received the azithromycin 600-mg dose but was identified in those who received clarithromycin. This finding is consistent with observations from other macrolide trials of either MAC prophylaxis or treatment. It remains unclear why such as difference should consistently be observed. Given the similar clinical antimycobacterial effects of the 2 compounds at the doses employed, any difference in in vitro potency is unlikely to explain any relative difference in resistance rates.

No difference in survival was found between the azithromycin 600 mg group and the clarithromycin group, either during the initial double-blind phase or in long-term follow-up. This finding provides the most objective measure of equivalence of the 2 regimens, given the impact of disseminated MAC on mortality [1].

The most significant limitation of this trial was the inability to enroll all of the anticipated subjects. As a consequence of not achieving the target enrollment of 200 patients, and because the anticipated sterilization rate of the clarithromycin arm on the basis of 2 consecutive negative cultures did not reach 75%, the power of the study to conclude equivalence between the 2 treatment arms was, ultimately, 61%. Nonetheless, a broader review of analyses that include other measures of sterility, rate of bacterial clearance, and mortality supports the similar treatment effects of each regimen.

The conclusions drawn from this study stand in contrast to those derived from the open-label comparison of treatment with azithromycin 600 mg and clarithromycin performed by the Veterans Administration cooperative research group, which found that patients randomized to receive azithromycin were less likely to have a sterile culture at week 16 [21]. With no other end point was there found to be significant difference between the 2 regimens. The VA study and the present study generate different conclusions, most likely due to the difference in sample size. Although both studies could benefit from even larger study populations, the larger sample size in this trial (n = 125) compared with the VA study (n = 49) may better distribute any differences in baseline characteristics, such as baseline bacteremia and underlying, sometimes undiagnosed, opportunistic infections. Beyond the narrow microbiologic measure of rates of sterility at different visits, other microbiologic assessments, rates of relapse, resistance patterns, safety, and impact on mortality were actually similar in both studies.

The possibility that additional agents could improve the efficacy of a macrolide combined with ethambutol should be explored in future trials. It has recently been observed that the addition of rifabutin to a combination of clarithromycin and ethambutol did not add significantly to efficacy [20]. However, as was suggested by the authors of that report [20], perhaps the addition of rifabutin to an azithromycin and ethambutol combination would provide an added measure of efficacy given the lower breakthrough rate of that combination in prophylaxis of MAC and the lack of pharmacokinetic interactions between the 2 drugs [22, 23].

Analysis of data from observational studies has suggested a low rate of disseminated MAC among patients whose CD4 counts have recovered to>100 cells/µL [24]. The majority of people living with HIV, however, do not have timely access either to HAART or primary prophylaxis of opportunistic infections. In this light, to uncover new advances in the treatment of opportunistic disease remains an important priority for HIV research. Although clarithromycin remains a well-established component of effective therapy for treatment of disseminated MAC, there remain clinically relevant concerns regarding its use [5, 8, 25]. On the basis of the collection of analyses performed on the data from this trial, azithromycin 600 mg taken every day, when given in combination with ethambutol, is an effective agent for the treatment of disseminated M. avium disease in HIV-infected patients.

Study Group Members

H. Barry Baker (Baptist Hospital, Miami, FL); Jeff Beal (St. John Medical Center, Tulsa, OK); Jorge Benetucci (Hospital FJ Muñiz, Buenos Aires, Argentina); Paul Berry (Pacific Oaks Medical Group, Sherman Oaks, CA); Jack Bissett (Austin Infectious Disease Consultants, Austin, TX); Marcos Boulos (Hospital das Clinicas, DA, São Paulo, Brazil); Gary Brewton (Houston Clinical Research Network, Houston, TX); Carol Brosgart (East Bay AIDS Center, Berkeley, CA); Alfred Burnside (Burnside Clinic, Columbia, SC); Pedro Cahn (HUESPED-Center of Infectology and Immunology, Buenos Aires, Argentina); J. Robert Cantey (Medical University of South Carolina, Charleston, SC); Lidia Isabel Cassetti (Infectologic Studies [FUNCEI], Buenos Aires, Argentina); Philip Craven (Infections Limited, P.S., Tacoma, WA); Iris Davis (Chase Braxton Clinic, Baltimore, MD); Gordon Dickinson (Veterans Affairs Medical Center, Miami, FL); Milton Estes (Mill Valley, CA); W. Jeffrey Fessel (Kaiser Permanente Medical Center, San Francisco, CA); Joseph Gathe (Houston Clinical Research Network, Houston, TX); Alvan Fisher (Stratogen Health, Providence, RI); Neel French (Weiss Memorial Hospital, Chicago, IL); Hernando Garcia (Jackson Memorial Medical Center, Miami, FL); Stephen Green (Hampton Roads Medical Specialists, Hampton, VA); Peter Hawley (Whitman-Walker Clinic, Washington, DC); Charles Hicks (Duke University Medical Center, Durham, NC); Jeffrey Jacobson (Department of Veterans Affairs Medical Center, Bronx, NY); Wilbert Jordan (King/Drew Medical Center, Los Angeles County, CA); Philip Keiser (Parkland Memorial Hospital, Dallas, TX); Carol Kemper (Santa Clara Valley Medical Center, San Jose, CA); Princy Kumar (Georgetown University Medical Center, Washington, DC); Christopher Lahart (Houston Veterans Affairs Medical Center, Houston, TX); Michael L. Levin (Chase Braxton Health Services, Inc., Pikeville, MD); David Lewi (School Paulista of Medicine, São Paulo, Brazil); Steven Marlowe (West Paces Clinical Research, Inc., Atlanta, GA); Joseph Marzouk (Adult Immunology Clinic, Oakland, CA); Earl Matthew (Central Texas Medical Foundation, Austin, TX); Pieter Meenhorst (Slotervaart Hospital, Amsterdam, The Netherlands); David McKinsey (Antibiotic Research Associates Inc., Kansas City, MO); Miguel Mogyoros (Franklin Medical Center, Denver, CO); Elizabeth Mokulis (Wilford Hall Medical Center, Lackland AFB, TX); Robert Murphy (Northwestern Memorial Hospital, Chicago, IL); Mary O'Connor (Trinity Lutheran Hospital, Kansas City, MO); George Pankey (Alton Oschner Medical Foundation, New Orleans, LA); David M. Parenti (George Washington University Medical Center, Washington, DC); Juan Perez-Morales (Baptist Hospital, Miami, FL); Owen Pickus ( Maine Centers for Health Care, Portland, ME); Barry Rodwick (Clinical Research of West Florida, Clearwater, FL); Margaretha Schneider (University Hospital Utrecht, Utrecht, The Netherlands); Daniel Seekins (Comprehensive Research Institute, Tampa, FL); Michael Spence (Hahnemann University, Philadelphia, PA); Malcolm Sperling (Fountain Valley Hospital, Fountain Valley, CA); C. H. H. Ten Napel (Medisch Spectrum Twente, Enschede [The Netherlands]); Jeremiah Tilles (University of California, Irvine Medical Center, Orange, CA); Marchina van der Ende (Academisch Ziekenhuis Rotterdam, Internal Medicine Department, The Netherlands); Sergio Vargas (Hospital Clinico de la Universidad, Santiago, Chile); Robert Wallace (Bay Area AIDS Consortium, Tampa, FL); A. Clinton White, Jr. (Baylor College of Medicine, Houston, TX); Judith Wolf (Graduate Hospital, Philadelphia, PA); and Bienvenido C. Yangco (Infectious Disease Research Institute, Inc., Tampa, FL).

Acknowledgments

We acknowledge the contribution of the patients participating in this trial, the study coordinators at each site, and Michael Zelasky, Juanita Brown, and Amapola O'Brien.