Background:  The clinical and biological characteristics that influence the outcome of structured treatment interruptions have not been fully clarified. The final analysis of the ISS PART study may help define patients potentially candidate to this therapeutic strategy.

Methods:  Were enrolled 273 subjects, 136 in arm B (5 structured treatment interruptions of 1-, 1-, 2-, 2-, and 3-month duration, each followed by 3 months of therapy) and 137 in arm A (continuous HAART). Primary end-point was the proportion of patients with >500 CD4/mm³ after 24 months. Mutations in proviral DNA and residual replication (assay cut-off:  2.5 copies HIV RNA/mL) were measured at baseline. Genotypic resistance was tested during structured treatment interruptions. Both per-protocol and intention-to-treat analyses were conducted.

Results:  Risk of protocol discontinuation was higher in arm B (66.5% vs 19.4% in A, hazard ratio = 4.6, 95%CI 3.0 to 7.3); conversely, more patients achieved primary endpoint in arm A (86.5 % vs 69.1% in B in the per-protocol analysis, p = 0.0075). Probability to complete study protocol and achieve the primary endpoint in arm B were independently predicted by pre-HAART CD4⁺ and male sex. Similar rates of virological failure were observed in A and B (24% and 26%, respectively). However, in arm B 2 factors entailed a higher risk of failure:  baseline residual replication >2.5 copies/mL (30.4% virological failure vs 11.3% in those with <2.5 copies/mL, HR =2.9, 95%CI 1.1 to 7.5) and plasma mutations during structured treatment interruptions. Indeed, of 136 subjects, 40 (29.4%) developed mutations in arm B, corresponding to a 32% (95%CI 23.5 to 40.0) cumulative risk over 24 months. Of them, 13 (32.5%) failed to achieve HIV viremia values <400 copies on ≥1 occasion at treatment resumption, compared with 14 of 96 (14.6%) patients without mutations. In a time-dependent Cox regression model this yielded a higher risk of virological failure associated with resistance (HR 2.3, 95%CI 1.0 to 5.1). The emergence of plasma mutations was independently predicted by the presence of archived mutations in proviral DNA at baseline and the use of a regimen based on unboosted protease inhibitor (PI) in a multivariate model (p = 0.001 for DNA mutations and p= 0.040 for PI-HAART).

Conclusions:  Potential candidates to structured treatment interruptions are subjects with high pre-HAART CD4 count, absence of archived mutations, and residual replication <2.5 copies HIV RNA/mL. For this therapeutic strategy NNRTI-based regimens appear preferable to unboosted PI HAART.