Comparison of Once-Daily Atazanavir With Efavirenz, Each in Combination With Fixed-Dose Zidovudine and Lamivudine, As Initial Therapy for Patients Infected With HIV
JAIDS Journal of Acquired Immune Deficiency Syndromes: Volume 36(5) 15 August 2004
Squires, Kathleen MD*; Lazzarin, Adriano MD†; Gatell, José M. MD, PhD‡; Powderly, William G. MD§; Pokrovskiy, Vadim MD, PhD||; Delfraissy, Jean-François MD¶; Jemsek, Joseph MD#; Rivero, Antonio MD**; Rozenbaum, Willy MD††; Schrader, Shannon MD‡‡; Sension, Michael MD§§; Vibhagool, Asda MD||||; Thiry, Alexandra PhD¶¶; Giordano, Michael MD¶¶
From the *Department of Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA; †Vita Salute University San Raffaele, Milan, Italy; ‡Infectious Diseases and AIDS Units, Clinical Institute of Infectious Diseases and Immunology, Barcelona, Spain; §Division of Infectious Diseases, Washington University School of Medicine, St. Louis, MO; ||Federal AIDS Center, Moscow, Russia; ¶Service de Médecine Interne, Hôpital de Bicêtre, Paris, France; #Jemsek Clinic, Huntersville, NC; **Sección de Enfermedades Infecciosas, Hospital Universitario Reina Sofía, Córdoba, Spain; ††Service des Maladies Infectieuses et Tropicales, Hôpital Tenon, Paris, France; ‡‡Southampton Medical Group, Houston, TX; §§HIV Comprehensive Care Center, North Broward Hospital District, Ft. Lauderdale, FL; ||||Department of Medicine, Ramathibodi Hospital, Bangkok, Thailand; and ¶¶Pharmaceutical Research Institute, Bristol-Myers Squibb Company, Wallingford, CT.
Background: Atazanavir, an azapeptide protease inhibitor (PI), has pharmacokinetics that allow once-daily dosing, and it is not associated with significant PI-associated dyslipidemia.
Methods: A randomized, double-blind, double-dummy, active-controlled, 2-arm study comparing the antiviral efficacy and safety of atazanavir 400 mg administered once daily with efavirenz 600 mg administered once daily in combination with open-label fixed-dose zidovudine plus lamivudine twice daily. The 810 treatment-naive patients were stratified by HIV RNA level. The primary efficacy end point was the proportion of treated patients with HIV RNA levels <400 copies/mL through week 48.
Baseline patient characteristics
median age: 33
gender: 64% men
38% Latino/Hispanic; 34% white; 13% BlackAsian/Pacific Islander: 13%
Median HIV RNA: 4.88 log copies/ml (range: 3.12-5.88)
>30,000 copies/ml: 73%
>100,000 c/ml: 43%
Median CD4 count: 283 (range: 68-1400)
Results: At week 48, HIV RNA levels were <400 copies/mL in 70% of patients receiving atazanavir and 64% of patients receiving efavirenz (intent-to-treat, difference; 95% confidence interval: 5.2%; -1.2%, 11.7%). Median CD4+ cell counts increased at comparable magnitudes and rates in the 2 treatment arms (mean change at week 48: 176 cells/mm3 with atazanavir, 160 cells/mm3 with efavirenz). Atazanavir-treated patients relative to comparator-treated patients did not demonstrate significant increases in total cholesterol, fasting low-density lipoprotein cholesterol, or fasting triglycerides over 48 weeks of therapy. Atazanavir-linked bilirubin elevations infrequently resulted in treatment discontinuation (<1%). Atazanavir treatment did not increase fasting glucose or insulin levels.
Conclusions: For initial HIV treatment, a highly active antiretroviral therapy regimen of atazanavir/zidovudine/lamivudine is as efficacious and well tolerated as the combination of efavirenz/zidovudine/lamivudine.
Atazanavir treatment was associated with a more favorable lipid profile than was efavirenz treatment, as assessed by significant differences in the mean percent changes from baseline in total cholesterol, fasting LDL cholesterol, and fasting triglycerides at week 48 (P < 0.0001). Mean increases in these lipid parameters were 18% to 23% in the efavirenz arm, although no significant increases in total or fasting LDL cholesterol and a decrease of 9% in triglycerides were observed in the atazanavir arm. Mean HDL cholesterol levels were >40 mg/dL at week 48 on both regimens, but significantly greater mean percent increases in HDL cholesterol levels were observed in the efavirenz arm (P < 0.0001). At week 48, mean total cholesterol levels in both treatment arms remained within the desirable range (<200 mg/dL) and mean LDL cholesterol levels remained within the optimal or near-optimal range (<130 mg/dL) as defined by the National Cholesterol Education Program, Adult Treatment Panel III (NCEP ATP III).
Glucose and insulin levels were measured at baseline and week 48. Mean fasting glucose levels at week 48 were comparable to those at baseline for both the atazanavir and efavirenz regimens. In the atazanavir treatment group, mean fasting glucose levels were 90 mg/dL at baseline and 93 mg/dL at week 48; in the efavirenz treatment group, they were 90 mg/dL at baseline and 94 mg/dL at week 48. Through week 48, mean fasting insulin levels increased (but not significantly) in patients in the atazanavir arm (11.3 uU/mL at baseline vs. 12.3 uUmL at week 48). Among the efavirenz-treated patients, a significant (but not clinically relevant) mean increase of 1.4 uUmL in fasting insulin was observed (9.9 uUm/L at baseline vs 11.5 uUmL at week 48; p=0.04).
EXPLORATORY ANALYSIS AND ASSESSMENTS
Additional analyses were performed to explore unexpected discrepancies when analyzing the <400 and <50 copies/mL viral load end points for the 2 treatment regimens. To further explore the sensitivity of the LOQ of 400 copies/mL, the proportion of patients with HIV RNA levels <200 copies/mL at week 48 was also computed for each treatment group. Additional analyses were performed to explore the distribution of HIV RNA levels between 50 and 400 copies/mL in patients achieving HIV RNA levels <400 copies/mL at week 48.
In addition, a laboratory assessment was conducted to examine the effects of plasma collection procedures on observed virologic response rates. Paired plasma specimens were collected in ethylenediaminetetraacetic acid (EDTA) and PPT tubes at week 52 and were quantitated for HIV RNA using the Roche AMPLICOR HIV-1 MONITOR ultrasensitive assay, version 1.0 or 1.5. Whole blood collected in EDTA-containing tubes (no heparin) was centrifuged in a swing-out bucket rotor at 800 g to 1600 g for 20 minutes within 6 hours of collection, and the plasma recovered was stored in secondary tubes. Whole blood was collected in PPT tubes that contained a gel barrier to assist in plasma separation and was centrifuged in a swing-out bucket rotor at 1100 g for 10 minutes within 2 hours of collection; the plasma recovered was stored in situ. Plasma-containing tubes were frozen on site and shipped on dry ice to the central laboratories. The samples remained at -70°C for 3 to 6 months before HIV RNA level assessments in April 2003. HIV RNA levels were assessed for all available samples for the week 52 visit. The proportions of samples with HIV RNA levels <400 and <50 copies/mL were evaluated (as-treated analysis).
Of the 1046 patients enrolled, 810 (77%) were randomized between February 15 and June 21, 2001. A total of 805 (99% of randomized patients) began therapy (atazanavir [N = 404], efavirenz [N = 401]). Of the patients included in the protocol, 144 (18%) discontinued treatment before the week 48 visit. More patients on the efavirenz regimen than on the atazanavir regimen discontinued treatment during this time (20% vs. 16%), consistent with a slightly longer mean time on therapy in the atazanavir group compared with the efavirenz group (47 weeks vs. 44 weeks). The higher rates of discontinuation in the efavirenz arm were contributed to by slightly higher incidences of adverse events and patient withdrawal. Discontinuation due to AE were 6% in ATV & 8% in EFV.
Baseline characteristics of the treated patients were evenly distributed across the treatment regimens. Overall, 35% of the study population was female and 67% was nonwhite. Geographic representation included South America (34%), Europe (28%), Asia (16%), North America (14%), and Africa (9%).
In the primary analysis, 70% of the patients treated with atazanavir (281 of 404 patients) had HIV RNA levels <400 copies/mL at week 48 compared with 64% of patients treated with efavirenz (258 of 401 patients). The atazanavir-efavirenz difference estimate (95% confidence interval) was 5.2 (-1.2, 11.7), indicating similarity between the treatment regimens. The proportions of treated patients with HIV RNA levels <50 copies/mL through week 48 were 32% (131 of 404 patients) for the atazanavir group and 37% (150 of 401 patients) for the efavirenz group. Of patients completing 48 weeks of treatment (as-treated analysis), 81% and 84% of those receiving atazanavir and efavirenz, respectively, had HIV RNA levels <400 copies/mL at week 48; the proportions with HIV RNA <50 copies/mL were 44% and 51%, respectively. By week 12, atazanavir and efavirenz regimens had produced mean reductions in HIV RNA levels of 2.5 and 2.6 log10 copies/mL, respectively. The reductions reached 2.7 log10 copies/mL for both treatment regimens by week 24 and were maintained at that level through week 48.
Further evaluation of virologic response revealed that among patients with HIV RNA levels <400 copies/mL at week 48, most had HIV RNA levels <200 copies/mL. In the intent-to-treat analysis, week 48 response rates for the LOQ of 200 copies/mL were 62% (251 of 404 patients) in the atazanavir treatment group and 58% (234 of 401 patients) in the efavirenz treatment group. In the as-treated analysis, response rates were 70% in the atazanavir treatment group and 75% in the efavirenz treatment group. Evaluation of the number of patients whose HIV RNA level fell within specific viral load ranges <400 copies/mL showed that HIV RNA levels tended to concentrate near 50 copies/mL. Of the patients with HIV RNA between 50 and <400 copies/mL, 71% of patients in each treatment group had HIV RNA <200 copies/mL and 33% had HIV RNA <100 copies/mL (data not shown; as-treated analysis).
Analysis of paired specimens at week 52 confirmed the comparability in antiretroviral efficacy between atazanavir and efavirenz. Plasma collected in PPT tubes consistently yielded higher HIV RNA readings than plasma collected in EDTA tubes in both treatment groups, however. A total of 584 patients (300 administered atazanavir and 284 administered efavirenz) were evaluable in this analysis. PPT tubes consistently yielded higher HIV RNA levels than EDTA tubes between 50 and 400 copies/mL. In the overall assessment of HIV RNA by tube type, 95% of EDTA tube specimens compared with 84% of PPT tube specimens were below the LOQ of 400 copies/mL. Eighty-nine percent of EDTA tube specimens compared with 55% of PPT tube specimens were below the LOQ of 50 copies/mL. The degree of difference between tube types in the number of specimens calculated to contain HIV RNA levels less than a given LOQ was comparable for the atazanavir and efavirenz treatment arms.
| ||RESPONDERS/EVALUABLE (%) |
| ||LOQ=400 copies/ml ||LOQ=50 copies/ml |
| ||EDTA ||PPT ||EDTA ||PPT |
|Overall N=584 ||95% ||84% ||89% ||55% |
|ATVN=300 ||93% ||83% ||86% ||53% |
|EFVN=284 ||96% ||85% ||93% ||57% |
The regimens were comparable with respect to the magnitude and rate of CD4+ cell count increase, which reached median levels of 455 and 446 cells/mm3 for the atazanavir and efavirenz treatment groups, respectively, at week 48. Although the TAD between the regimens in the change in CD4+ cell count was statistically significant (P < 0.001), favoring atazanavir, the magnitude of the difference (16 cells/mm3) did not reach the predefined criterion for similarity that would indicate clinical significance (50 cells/mm3).
The overall incidence of adverse events as well as the incidence of grade 3 to 4 adverse events was comparable between the treatment regimens. The incidence of grade 2 to 4 adverse events that were possibly, probably, likely, or certainly related to the study regimen was also comparable between the treatment arms. The most common treatment-related grade 2 to 4 adverse event was nausea, which occurred with comparable frequency in both study arms. Rash and dizziness, characteristic of efavirenz therapy, were significantly more frequent in the efavirenz arm (P < 0.05), whereas jaundice and scleral icterus, mainly associated with atazanavir therapy, were significantly more common in the atazanavir arm (P < 0.05). One patient in the efavirenz/zidovudine/lamivudine arm experienced lactic acidosis syndrome. Serious adverse events were infrequent (10% for both regimens). Four patients (<1%) in each treatment group experienced a newly diagnosed Centers for Disease Control and Prevention Class C AIDS event during the study. Five deaths, all unrelated to study therapy, were reported during the trial; 3 of the deaths occurred after the discontinuation of study therapy. Two of the 5 deaths occurred in atazanavir-treated patients (tuberculous meningoencephalitis and central nervous system lymphoma), and 3 of the 5 deaths occurred in efavirenz-treated patients (tuberculosis, bronchial carcinoma/pneumonia, and gun shot wound).
GRADE 2-4 ADVERSE EVENTS
| ||ATV ||EFV |
| ||N=404 ||n=401 |
|Total ||41% ||45% |
|Nausea ||14% ||13% |
|Rash ||6% ||10% |
|Headache ||6% ||6% |
|Jaundice ||5% ||0% |
|Vomiting ||4% ||7% |
|Dizziness ||2% ||6% |
|Scleral icterus ||1% ||0% |
|Diarrhea ||1% ||2% |
|LAS/SHL** ||0% ||<1% |
GRADE 3-4 LAB ABNORMALITIES
|ALT/SGPT ||4% ||3% |
|AST/SGOT ||2% ||2% |
|Tot Bilirubin* ||33% ||<1% |
|ØNeurophils ||6% ||9% |
|ØHemoglobin ||4% ||3% |
*dose reductions, n=20 (5%); treatment discontinuations, n=2 (<1%)
++LAS, lactic acidosis syndrome; SHL, symptomatic hyperlactatemia.
Grade 3 to 4 elevations in total bilirubin concentration, which were mainly indirect (unconjugated), occurred more frequently in the atazanavir treatment group than in the efavirenz treatment group (33% vs. <1%; P < 0.0001). The incidence of grade 3 to 4 total bilirubin elevations was comparable between hepatitis-negative and hepatitis-positive atazanavir-treated patients (as assessed by the absence or presence of hepatitis B virus [HBV] antigen or hepatitis C virus [HCV] antibody, data not shown). Grade 4 elevations in total bilirubin were reported in 25 of the 404 atazanavir-treated patients, of whom 20 underwent dose reduction. After dose reduction, 3 patients had a subsequent worst on-study bilirubin level of grade 1 to 2, 12 patients had a grade 3 elevation, and 5 had a recurrence of grade 4. Of the patients with recurrent grade 4 elevations, 2 (<1%) discontinued atazanavir therapy and 3 (<1%) continued atazanavir therapy at a lower dose. The atazanavir dose was also reduced in a single patient with jaundice but no grade 4 bilirubin elevation. Grade 3 to 4 elevations in aminotransferases were infrequent in both treatment groups, occurring in <=4% of patients (see Table 4) and suggesting no direct correlation between elevated transaminases and total bilirubin in the atazanavir treatment group.
The development and incorporation of PIs into combination treatment regimens has been one of the most important advances in the management of HIV infection. Treatment with these drugs can significantly decrease HIV-related morbidity and mortality. Although all approved PIs show substantial antiretroviral activity in vitro, rapid metabolism, predominantly by hepatic mechanisms, often necessitates frequent dosing to maintain plasma levels high enough to effectively suppress HIV viral replication. When exposure is inadequate, a second reduced-dose PI booster may be used to inhibit cytochrome P450-related metabolism. Thus, although the incorporation of PIs into HAART regimens can substantially reduce plasma viral loads, increase CD4+ cell counts, slow HIV disease progression, and prolong life, in clinical practice, their success may be limited by pharmacokinetic issues as well as by those related to adherence and safety.
The pharmacokinetic profile of atazanavir, a potent azapeptide PI, allows for once-daily dosing without boosting as well as fewer pills, potentially leading to greater adherence. In the present study, atazanavir and efavirenz showed comparable efficacy when combined with fixed-dose zidovudine and lamivudine in treatment-naive patients over 48 weeks of therapy. Although the percentage of efavirenz-treated patients in this study achieving HIV RNA levels <400 copies/mL was consistent with that seen in another efavirenz study, the percentage of patients who reached levels <50 copies/mL in the present study was lower. 23 In the study reported here, 64% and 37% of the efavirenz-treated patients achieved these respective HIV RNA levels, whereas in the other study (a large randomized trial comprising 1266 patients), the respective response rates at 48 weeks among the 422 patients treated with the same doses of efavirenz, zidovudine, and lamivudine were 68% and 62%. Nevertheless, exploratory analyses revealed that at week 48, most patients who were classified as responders at the LOQ of 400 copies/mL but as failures at the LOQ of 50 copies/mL actually tended to have viral loads closer to 50 copies/mL than 400 copies/mL.
The apparent disparity between the efavirenz treatment regimen results at the LOQ of 50 copies/mL in this trial and other trials is of particular interest. Cross-trial comparisons of efficacy, in general, may not be valid for a number of reasons, including differences between patient baseline characteristics, trial procedures, and analysis techniques. 24 Importantly, PPT tubes were used throughout this randomized study to collect whole blood for the primary assessments of virologic efficacy using validated methodologies. A cross-sectional assessment at a single time point (week 52) conducted on all available samples identified large differences in HIV RNA levels measured in EDTA and PPT tubes influencing observed response rates, particularly at the LOQ of 50 copies/mL. The magnitude of these tube-dependent differences was comparable between treatment regimens. Higher HIV RNA values have been observed previously when using PPT tubes compared with EDTA tubes 25; however, the magnitude of the difference at 50 copies/mL has not been described in as large a sample as represented in this study. Potential explanations, such as whether virus associated with cell debris is more readily detectable in a PPT tube, require further exploration. The discrepancy in HIV RNA levels measured in EDTA and PPT tubes underscores the hazards of making cross-study comparisons but supports the integrity of the trial and the comparability in efficacy for atazanavir and efavirenz.
The overall safety profile of atazanavir was found to be comparable to that of efavirenz. The incidence of adverse events and the patterns of their occurrence were consistent with the known adverse effect profiles of atazanavir, efavirenz, and zidovudine plus lamivudine. Potentially serious adverse events were infrequent and were evenly distributed between patients in both treatment arms. Elevated total bilirubin levels were associated with atazanavir treatment; however, they were predominantly of the indirect (unconjugated) type and reversible, and they seemed to be clinically benign and unrelated to hepatic damage. Dose reductions, although not a recommended strategy for managing bilirubin levels, and bilirubin-related discontinuations were infrequent (5% and <1%, respectively). Grade 3 to 4 elevations in total bilirubin were not associated with grade 3 to 4 increases in hepatic transaminases.
Treatment with many PIs has been associated with increases in total cholesterol, LDL cholesterol, and triglycerides. These increases are sometimes large. Dyslipidemia of the magnitude observed with other PIs has the potential to worsen cardiovascular risk and frequently leads to a recommendation to use lipid-lowering therapy. Recent epidemiologic data suggest that HIV-infected patients may represent a rapidly emerging population at risk for cardiovascular disease. In addition, treatment with PIs has been associated with lipodystrophy syndrome, which includes body shape changes, fat redistribution, and other metabolic abnormalities such as glucose intolerance.
In this study, confirming the observations made in previous studies, treatment with atazanavir combined with fixed-dose zidovudine and lamivudine did not result in significant increases in total cholesterol, fasting LDL cholesterol, and fasting triglycerides through 48 weeks of therapy. The proportions of patients with fasting LDL cholesterol levels >=160 mg/dL, which is the threshold level recommended by NCEP ATP III guidelines for lipid-lowering therapy and/or dietary intervention in the absence of underlying cardiovascular risk factors, were 2% and 4% at baseline and 3% and 8% at week 48 for the atazanavir and efavirenz treatment groups, respectively. It is also noteworthy that the NCEP ATP III identifies HDL cholesterol levels <40 mg/dL as a strong independent predictor of coronary heart disease risk and that both atazanavir and efavirenz had beneficial effects on HDL cholesterol, increasing the levels to >40 mg/dL.
Treatment with atazanavir or efavirenz was also not associated with increases in plasma glucose or insulin levels, alterations that have been linked to lipodystrophy. In vitro studies show that atazanavir does not inhibit glucose transport, suppress adipocyte differentiation, or interfere with lipid metabolism, as do other PIs. These in vitro studies provide a potential explanation for the drug's favorable metabolic profile.
Atazanavir is comparable in efficacy and safety to efavirenz in the initial treatment of HIV infection. The 2 drugs are each administered once daily and have a low pill burden. Unlike other PIs, atazanavir does not produce clinically significant increases in total cholesterol, fasting LDL cholesterol, fasting triglycerides, insulin, or glucose, thus potentially reducing the use of lipid-lowering agents and the risk for future cardiovascular disease events that may be associated with treatment in HIV-infected patients. The results of this study demonstrate that atazanavir and efavirenz are agents that, in combination with nucleoside analogues, provide convenient and potent treatment regimens for initial HIV therapy.
Combination antiretroviral therapy has transformed the treatment of patients infected with HIV. Incorporation of protease inhibitors (PIs) into highly active antiretroviral therapy (HAART) has been specifically accompanied by a significant improvement in overall patient outcomes. The combination of high pill burden, complex food and fluid requirements, poor drug tolerability, drug toxicities, and pharmacokinetic issues complicates treatment, however, and may lead to virologic treatment failure. Adherence to therapy is also a critical issue because of its close association with virologic response. In the case of PIs, decreased adherence has been related to high pill burden and significant side effects. Consequently, the use of an antiretroviral regimen incorporating the nonnucleoside reverse transcriptase inhibitor (NNRTI) efavirenz has become a preferred initial therapy. Efavirenz is an effective once-daily NNRTI with a low pill burden that may produce a less deleterious effect on lipid profiles than do current PIs.
Because of their effectiveness and demonstrated clinical benefit as components of HAART, PIs continue to be developed with an aim toward designing agents with a low pill burden, more convenient dosing, better tolerability, and less toxicity. Atazanavir is an azapeptide PI with a pharmacokinetic profile that supports once-daily dosing with 2 pills per day. Phase 2 studies show that atazanavir produces rapid and durable suppression of HIV RNA and a durable increase in CD4+ cell count in treatment-naive and -experienced HIV-infected patients. Treatment with atazanavir also results in significantly less hyperlipidemia and hypertriglyceridemia than with comparator PIs as seen in phase 2 studies. Furthermore, in vitro studies suggest that atazanavir is not associated with insulin resistance, another metabolic abnormality common to treatment with current PIs.
This report describes the results of a phase 3, randomized, double-blind, controlled trial comparing the antiviral efficacy and safety of atazanavir with efavirenz, each as part of a 3-drug antiretroviral regimen in combination with fixed-dose zidovudine plus lamivudine, in treatment-naive patients.
Patients were recruited from 91 centers in North America, South America, Europe, Asia, and South Africa. Patients were required to be >=16 years old and to have plasma HIV RNA levels >=2000 copies/mL and CD4+ cell counts >=100 cells/mm3 (or >=75 cells/mm3 if the patient had no prior history of any AIDS-defining illness). The presence of a suspected primary (acute) HIV infection, a newly diagnosed HIV-related opportunistic infection, or any medical condition requiring acute therapy at the time of enrollment disqualified patients from participation. No antiretroviral therapy was permitted during the 30 days before screening. Patients were excluded if they had been treated for >30 days with a nucleoside reverse transcriptase inhibitor (NRTI) and/or >7 days with an NNRTI or a PI. Patients who were pregnant or breast-feeding were also not eligible. Laboratory abnormalities leading to exclusion included alanine aminotransferase or aspartate amino-transferase levels >=3 times the upper limit of normal or total serum bilirubin levels >=1.5 times the upper limit of normal. All patients provided written informed consent.
Study Design and Outcomes
The trial was designed as a randomized, double-blind, double-dummy, active-controlled, 2-arm comparative study. The primary efficacy outcome was the proportion of treated patients with HIV RNA levels <400 copies/mL through 48 weeks of therapy (intent-to-treat analysis). Secondary comparative efficacy outcomes included the proportion of treated patients with HIV RNA levels <50 copies/mL through week 48, the proportion of patients completing 48 weeks of treatment with HIV RNA levels <400 copies/mL and <50 copies/mL at week 48 (astreated analysis), and changes in plasma log10 HIV RNA values and CD4+ cell counts from baseline through week 48. Treatment response was defined as 2 or more sequential HIV RNA measurements below the limit of quantification (LOQ), either <400 copies/mL or <50 copies/mL. Patients were considered to have had treatment failure if they had 2 sequential HIV RNA measurements above the specified LOQ. Patients were also considered to have treatment failure if they discontinued treatment for any reason, required any substitution or dose modification in nucleoside therapy, or had missing data. Treatment failure did not include patients with a newly diagnosed Centers for Disease Control and Prevention Class C AIDS event. Safety and tolerability assessments included the frequency and severity of adverse events, incidence of laboratory abnormalities, and changes from baseline at week 48 in metabolic parameters (glucose, insulin, and serum lipid concentrations).
Randomization of patients to the treatment regimens (1:1 ratio) was performed centrally and stratified by qualifying plasma HIV RNA level (<30,000 or >=30,000 copies/mL) and investigative site using the method of Pocock and Simon. 16 Patients randomized to atazanavir received atazanavir, 400 mg, once daily as well as efavirenz placebo and fixed-dose zidovudine plus lamivudine (Combivir; GlaxoSmithKline, Research Triangle Park, NC); patients randomized to efavirenz received efavirenz (Sustiva; Bristol-Myers Squibb, New York, NY), 600 mg, once daily as well as atazanavir placebo and fixed-dose zidovudine plus lamivudine. Atazanavir and efavirenz treatments were blinded. Zidovudine plus lamivudine was administered open label as one 300- or 150-mg fixed-dose combination tablet twice daily.
Assessment and Monitoring
Patients enrolled in the study were evaluated at baseline; weeks 2, 4, 8, 12, and 16; and every 8 weeks thereafter through week 48. Interim history was obtained, targeted physical examinations were performed, and adverse events were recorded at all visits. Hematology and serum chemistry were assessed at all visits. Plasma HIV RNA levels, CD4+ cell counts, and fasting metabolic parameters were assessed at all scheduled visits except at week 2.
Samples for the HIV RNA measurements were collected using uniform techniques for all study sites. Whole blood was collected in PPT tubes (Vacutainer® Brand PPT Plasma Preparation Tube; Becton Dickinson, Franklin Lakes, NJ), frozen in situ at -70°C after centrifugation at 1100 g for 10 minutes, and sent to a central laboratory for further processing. HIV RNA levels were measured at screening and at baseline using the Roche AMPLICOR HIV-1 MONITOR™ (Roche Diagnostics, Basel, Switzerland) standard assay, version 1.0 or 1.5 (accuracy range: 400-750,000 copies/mL). HIV RNA levels were measured during the treatment phase using the Roche AMPLI-COR HIV-1 MONITOR ultrasensitive assay, version 1.0 or 1.5 (accuracy range: 50-75,000 copies/mL). Version 1.0 of the assays was used in North and South America, whereas version 1.5 was used in Europe, Asia, and Africa because of its enhanced ability to detect non-B HIV-1 clades more common to those regions.
Efficacy and safety analyses included all treated patients. A sample size of 750 patients was chosen to provide at least 90% power to demonstrate similar (noninferior) efficacy between the 2 regimens for the primary analysis comparing the proportion of treated subjects with HIV RNA levels <400 copies/mL through week 48. The sample size was based on an intent-to-treat analysis in which patients who dropped out or were lost to follow-up were counted as having failed treatment. Although all randomized patients did not remain on study at week 48, they were all included in the analysis so that the sample size was unaffected by patient dropout or loss to follow-up. Proportions were computed within the qualifying HIV RNA strata and were combined using a weighted average with weights proportional to the size of each stratum. The difference in proportions and 95% confidence interval were computed using a normal approximation. For the primary analysis, the treatment regimens were determined to be similar if the lower 95% confidence limit for the difference (atazanavir-efavirenz) in proportions exceeded -10%. The comparison of treatment regimens with respect to changes from baseline through week 48 in HIV RNA levels and CD4+ cell counts used time-averaged differences (TADs) and 95% confidence intervals for observed measurements. Week 48 mean percent changes from baseline and standard errors for lipid parameters (total cholesterol, high-density lipoprotein [HDL] cholesterol, low-density lipoprotein [LDL] cholesterol, and triglycerides) were computed on the log scale and then back-transformed.