Virologic failure occurs in a large proportion of individuals receiving treatment with combination antiretroviral therapy. Studies suggest that treatment interruption prior to initiation of a multiple-drug rescue regimen may improve virologic response in individuals who have failed several prior antiretroviral regimens. Other studies suggest there is a virologic benefit derived from using genotypic or phenotypic resistance testing in selecting salvage therapy regimens for patients failing antiretroviral therapy. This study tests the hypothesis that salvage regimens selected on the basis of HIV-1 resistance genotype, phenotype [AS PER AMENDMENT 02/19/02: virtual phenotype], and treatment history will be more effective if there is a period of treatment interruption before initiating that regimen.

Patients continue their antiretroviral therapy until randomization. Based on the results of the pre-entry genotype and phenotype [AS PER AMENDMENT 02/19/02: virtual phenotype] tests and treatment history, an individualized salvage therapy regimen (not provided by the study) is selected by the site investigator(s). Additionally, patients start or continue maintenance therapy (not provided by the study) for opportunistic infections (OIs). Patients are randomized to 1 of 2 treatment arms. In Arm A, patients have antiretroviral treatment interruption for a period of 16 weeks (Step 1), followed by initiation of the [AS PER AMENDMENT 02/19/02: best available] salvage therapy regimen (Step 2). [AS PER AMENDMENT 02/19/02: Patients in Arm A will be placed immediately on their individualized salvage regimen before the end of the 16-week period of treatment interruption if their CD4 count falls below a defined threshold, or if they develop a new OI]. In Arm B, patients switch immediately to the salvage therapy regimen. [AS PER AMENDMENT 02/15/01: Patients who become pregnant during Step 1 of Arm A must be advised to begin their selected, individualized salvage therapy regimen or a modified salvage regimen. Patients who become pregnant during Step 2 of Arm A or Arm B have therapy evaluated and undergo any changes required by their pregnancy.] Patients in both arms are monitored for plasma HIV-1 RNA levels, CD4+ and CD8+ cell counts, and HIV drug resistance genotypes and phenotypes for a duration of 64 weeks from randomization. Patients in Arm A are also monitored for immune reactivation by measurement of T-cell subsets and plasma cytokines during treatment interruption. Patients may participate in a virology substudy (A5100s) and an immunology substudy (A5104s). [AS PER AMENDMENT 02/19/02: Patients who volunteer to participate in the substudies must be registered to the main study at the same time they are registered to a substudy.]