The study population included HIV-1–infected patients who were at least 16 years of age, who had received at most 7 days of antiretroviral therapy previously, and who had acceptable laboratory values. Further details about the entry criteria are described in the study protocol (see the Supplementary Appendix
, available with the full text of this article at NEJM.org
). The human subjects committee at each participating center approved the study protocol, and written informed consent was obtained from all participants in compliance with the human experimentation guidelines of the U.S. Department of Health and Human Services.
The AIDS Clinical Trials Group Study A5202 is an ongoing phase 3B, randomized, partially blinded study comparing four antiretroviral regimens for the initial treatment of HIV-1 infection. The planned study duration was 96 weeks after enrollment of the last patient. Baseline evaluations included a medical history, physical examination, CD4 cell count, and HIV-1 RNA level. At screening, a genotypic resistance test was required in patients with recent HIV-1 acquisition. Testing for the HLA-B*5701 allele was permitted but not required. Patients were randomly assigned to receive one of four oral once-daily regimens: 600 mg of efavirenz (Sustiva, Bristol-Myers Squibb) or 300 mg of atazanavir (Reyataz, Bristol-Myers Squibb) plus 100 mg of ritonavir (Norvir, Abbott Laboratories) given with either 600 mg of abacavir plus 300 mg of lamivudine (Epzicom, GlaxoSmithKline) or 300 mg of tenofovir DF plus 200 mg of emtricitabine (Truvada, Gilead Sciences). The study was double-blinded with regard to the NRTIs.
Randomization was stratified according to the screening HIV-1 RNA level obtained before study entry (≥100,000 vs. <100,000 copies per milliliter), with the use of a permuted-block design with dynamic balancing according to the main institution. Screening of HIV-1 RNA levels was performed at any laboratory certified under the Clinical Laboratory Improvement Amendments. Study evaluations were completed before entry, at entry, at weeks 4, 8, 16, and 24, and every 12 weeks thereafter for the duration of the study in all patients, regardless of any treatment modification. After screening, the level of HIV-1 RNA was measured (Roche Amplicor Monitor assay, version 1.5) at Johns Hopkins University. At the time of protocol-defined virologic failure, geno-typing for drug resistance was performed at Stanford University; the baseline samples obtained from the patients were genotyped retrospectively.
Abbott Pharmaceuticals, Bristol-Myers Squibb, Gilead Sciences, and GlaxoSmithKline provided the study medications and had input into the protocol development and review of the manuscript. All the authors participated in the trial design, data analysis, and preparation of the manuscript, and all the authors vouch for the completeness and accuracy of the reported data.
HIV-1 DRUG-RESISTANCE TESTING
Since this is an ongoing study, the data and safety monitoring board recommended a resistance analysis that was restricted to the frequency of major mutations at baseline and at the time of virologic failure; the board also recommended that specific mutations not be disclosed. Major resistance mutations were defined as those listed by the International AIDS Society–USA,6
as well as L74I and G190C/E/Q/T/V for reverse transcriptase, and L24I, F53L, I54V/A/T/S, and G73C/S/T/A for protease.
The primary efficacy end point was the time from randomization to virologic failure (defined as a confirmed HIV-1 RNA level ≥1000 copies per milliliter at or after 16 weeks and before 24 weeks, or ≥200 copies per milliliter at or after 24 weeks). The primary hypotheses were that for each of the regimens that included ritonavir-boosted atazanavir and efavirenz, abacavir–lamivudine was equivalent to tenofovir DF–emtricitabine, and for each NRTI regimen, ritonavir-boosted atazanavir was equivalent to efavirenz. Regimens were considered equivalent if the two-sided 95% confidence interval for the hazard ratio was between 0.71 and 1.40. A planned sample size of 1800 subjects (450 per group) would provide an 89.8% probability of declaring equivalence if two regimens were the same, assuming uniform accrual, exponential virologic failure, and lost-to-follow-up time distributions among the four groups, with event probabilities of 17.46% and 10.00%, respectively, at 48 weeks.
Study conduct and safety data were reviewed yearly by the data and safety monitoring board. Efficacy data were reviewed annually starting with the second review of study data. Early-stopping guidelines for inferiority were prespecified, with a regimen considered to be inferior if the 99.95% two-sided confidence interval for the hazard ratio for virologic failure did not include 1.0.
Analyses of efficacy data followed the intention-to-treat principle and were stratified according to the screening HIV-1 RNA level. Time-to-event distributions were estimated with the use of the Kaplan–Meier method and compared by means of two-sided log-rank tests. Hazard ratios were estimated with the use of Cox proportional-hazards models.
The primary safety end point was the time from the initiation of treatment to the first grade 3 or 4 sign, symptom, or laboratory abnormality that was at least one grade higher than that at baseline, excluding isolated unconjugated hyper-bilirubinemia and elevations in the creatine kinase level, while the patient was receiving the randomly assigned treatment. Adverse events were graded ranging from 1 to 4, with 1 indicating mild events and 4 indicating potentially life-threatening events, according to a severity scale as adopted in December 2004 by the Division of AIDS at the National Institutes of Health.
The data and safety monitoring board met on January 29, 2008, for the first efficacy review. Protocol prespecified time-to-event distributions were presented overall and within each screening HIV-1 RNA stratum. The data and safety monitoring board noted excess virologic failures in both groups of patients who received regimens containing abacavir–lamivudine; additional requested analyses showed that these excess failures associated with abacavir–lamivudine occurred within the higher screening HIV-1 RNA stratum. When data in the four groups were combined and analyzed as two groups (i.e., the group receiving regimens with abacavir–lamivudine and the group receiving regimens without abacavir–lamivudine), the difference between these two groups was determined to be highly statistically significant. The data and safety monitoring board found the strength and validity of these findings sufficient to warrant stopping the further study of abacavir–lamivudine among participants with a screening HIV-1 RNA level of at least 100,000 copies per milliliter. The board specified that the remainder of the study should continue without change. Further details of board’s findings and recommendations are provided in the Supplementary Appendix
On release of these findings from the data and safety monitoring board, the study team completed additional analyses based on a previous analysis plan. Treatment-effect modification was assessed for six prespecified baseline covariates: sex, race or ethnic group, age, HIV-1 RNA level, CD4 cell count, and available or unavailable test results for HIV-1 genotype at screening.
For the safety end point, data were censored at the first discontinuation of a randomly assigned active NRTI. Changes in the CD4 count, fasting lipid level, and calculated creatinine clearance from baseline to week 48 were compared among patients for whom data were available with the use of a Wilcoxon–Mann–Whitney test. The binary end point of an HIV-1 RNA level of less than 50 copies per milliliter was compared at week 48 with the use of a chi-square test. Reported P values are two-sided.
Analyses were performed with the use of SAS software, version 9 (SAS) and with S-Plus software, version 6 (Insightful).