Background . We sought to determine the safety of treatment interruption (TI) and to identify parameters that would define patients with human immunodeficiency virus (HIV) for whom TI is safer.

Methods . AIDS Clinical Trials Group 5170 was a multicenter, 96-week-long, prospective study of HIV-infected patients receiving antiretroviral therapy (ART) who had CD4⁺ cell counts >350 cells/mm³ and who underwent TI.

Results . A total of 167 patients were enrolled. The median nadir in CD4⁺ cell count was 436 cells/mm³. The initial decrease (i.e., during the first 8 weeks) in CD4⁺ cell count after ART interruption was 20 cells/mm³/week; the subsequent decrease was 2.0 cells/mm³/week until week 96. Both the CD4⁺ cell count before enrollment and the increase in CD4⁺ cell count during ART predicted early decrease; later decrease was predicted by the level of interleukin- 7 at enrollment. A Centers for Disease Control and Prevention (CDC) diagnosis of a category B or C event was made for 2 and 2 patients, respectively (all had CD4⁺ cell counts >350 cells/mm³). At week 96, 17 patients had CD4⁺ cell counts ≤250 cells/mm³, and 46 patients had resumed ART; 5 patients died (unrelated to HIV or acquired immunodeficiency syndrome). In a multivariate analysis, a higher nadir in CD4⁺ cell count (>400 cells/mm³), a lower HIV load (<50 copies/mL) at the time of TI, and an HIV load ≤22,000 copies/mL before ART predicted a longer time to the primary end point (CDC category B or C event, death, CD4⁺ cell count ≤250 cells/mm³, or resumption of ART).

Conclusion . Disease progression after TI was low in this cohort. A higher nadir in CD4⁺ cell count, a lower HIV load before ART, and an HIV load ≤50 copies/mL at the time of TI predicted a longer time to the primary end point.