Two-months-off, four-months-on antiretroviral regimen increases the risk of resistance, compared with continuous therapy: A randomized trial involving west African adults

Background. A randomized trial was launched in Côte d'Ivoire in 2002 to compare continuous antiretroviral treatment (hereafter, "C-ART") to an ART regimen of 2 months off and 4 months on therapy (hereafter, "2/4-ART"). We report the final analysis. Methods. A total of 435 adults who were receiving successful ART ((median CD4 cell count prior to ART, 272 cells/mm3; 88% were receiving a zidovudine-lamivudine-efavirenz regimen) were randomized to receive C-ART or 2/4-ART. The main primary end point was the percentage of patients with <350 CD4 cells/mm3 at 24 months. The sample size ensured 80% power to demonstrate noninferiority (noninferiority bound, -15%), assuming that 30% of the patients in the C-ART arm would have <350 CD4 cells/mm3. Other end points were mortality, morbidity, cost of care, genotypic resistance, adherence, and toxicity. Results. The percentage of patients with <350 CD4 cells/mm3 at 24 months was 5.6% (6 of 107) in the C-ART arm and 14.6% (46 of 315) in the 2/4-ART arm (lower bound of the 95% CI for the difference, -14%). Cost was 18% higher in the C-ART arm, and resistance to nonnucleoside reverse-transcriptase inhibitors (NNRTIs) was 20% higher in the 2/4-ART arm. Other end points were nonconclusive. Conclusions. Although 2/4-ART met the predetermined criteria for noninferiority, the percentage of patients with <350 CD4 cells/mm3 in the C-ART arm was lower than anticipated, which makes the clinical significance of this noninferiority uncertain. In addition, 2/4-ART led to an unacceptable additional risk of selecting for drug-resistant virus. This new argument against episodic ART strategies is also a caveat against any unplanned ART interruptions in Africa, where most patients receive NNRTIs. Trial registration. ClinicalTrials.gov identifier: NCT00158405. © 2008 by the Infectious Diseases Society of America. All rights reserved.