553    

Background:  Despite the success of HAART regimens in antiretroviral naïve study subjects, virologic failure is seen in 10% to 30% of subjects over 3 years of follow-up. Understanding factors that influence the risk of virologic failure is important to improve virologic outcome in these subjects.

Methods:  We analyzed baseline and on-treatment factors associated with virologic failure in subjects receiving zidovudine and lamivudine plus either indinavir (IDV), efavirenz+indinavir (EFV+IDV) or nelfinavir+indinavir (NFV+IDV) in ACTG 388, a randomized phase 3 study with a median follow-up of 108 weeks. Virologic failure in the study was defined as confirmed plasma HIV RNA viral load ³ baseline or 1.0 log above nadir, >200 copies/mL at week 24 (early virologic failure) or relapse (confirmed viral load >200 following confirmed viral load <200) (late virologic failure). Factors in proportional hazards analyses were age, race, gender, baseline viral load, and treatment regimen and as time-dependent covariates, patterns of viral load responses, recent toxicity, drug holds or permanent treatment discontinuation.

Results:  Of 517 subjects randomized to ACTG 388, 172 had virologic failure. In multivariate analyses, non-white subjects were at a higher risk of virologic failure (p = 0.015, hazard ratio (HR) = 1.47) as were younger subjects (p = 0.001, HR = 1.36 for each 10 years). Compared to continuous therapy, treatment discontinuation gave a 4.2-fold increase in the risk of virologic failure (p <0.0001); temporary drug holds of 1 to 2 and >2 weeks gave 4.2- and 9.3-fold increases, respectively (p <0.0001). These factors explained some of the treatment difference observed between the NFV+IDV and IDV arms. Although many of the treatment interruptions were due to toxicity, virologic failure was not associated with recent toxicity (p = 0.55). Consistent results were seen with early and late failure. In late failure, never achieving viral load ≤50 copies/mL and a most recent viral load 51-200 copies/mL were both associated with an increased risk of subsequent virologic failure (p <0.0001, HR = 5.6, p <0.0001, HR = 5.4); there was no evidence that a history of intermittent viremia (viral load >50 copies/mL with subsequent VL≤50 copies/mL) was associated with an increased risk of virologic failure (p = 0.49).

Conclusions:  Non-white and younger subjects were at a greater risk for virologic failure. Therapy discontinuation and temporary interruption of therapy, even of short duration, were highly associated with virologic failure overall and with both early and late failure. Failure to achieve viral load ≤50 copies/mL and current viral load between 51 and 200 copies/mL were both associated with an increased risk of late virologic failure. Recent toxicity was not a risk factor for virologic failure.

Keywords: virologic failure; treatment discontinuation; intermittent viremia