Does plasma HIV RNA predict outcome in a cohort of treated HIV-infected individuals followed over 3 years?

Background: Despite reductions in AIDS illness and mortality, it is increasingly apparent that a significant proportion of individuals treated with combination antiretroviral (cARV) therapy have continuing or recrudescent HIV RNA in plasma. The predictive value of plasma HIV RNA in treated individual remains uncertain and rates of and risk factors for adverse outcomes such as hospitalisation, opportunistic infections and deaths are needed. Objectives: The objectives of this study were to establish a retrospective cohort of individuals treated with cARVs, to assess factors associated with detectable HIV RNA and to determine rates of and risk factors for hospitalisation, opportunistic infection and mortality over 3 years of follow-up. Study design: All individuals treated at The Alfred Hospital, Melbourne, Victoria between January and June 1997 who had had plasma HIV RNA measured were included in the retrospective cohort. Clinical, virological and hospitalisation data were recorded and validated by cross-reference with electronically stored laboratory, hospital activity and state notification databases. Outcome was assessed at October 2000. Results: Amongst the 555 individuals tested, 438 (60.7%) had detectable (> 500 copies/ml) HIV RNA (bDNA assay, version 2) at baseline. The overall mortality rate was 5.5 per 100 person years; the AIDS rate 1.99 per 100 person years and hospitalisation rate 16.4 per 100 person years. Risk factors for death in this population identified by univariate analysis were HIV RNA concentration at baseline and at follow-up October 2000, nadir and most recent CD4 lymphocyte number, not receiving cARV as initial treatment, total number of ARV agents and number of changes in ARV per year, developing AIDS and being hospitalised during follow-up. In a multivariate model, the most recent CD4 lymphocyte number, the number of different ARVs per year and having more than one hospitalisation remained predictive of death. Conclusions: HIV RNA remained detectable in the majority (60.7%) of this treatment-experienced population over 3 years, yet mortality rate remained relatively low at 5.5 per 100 person years. Factors associated with death were immunological (CD4 lymphocyte number) and treatment related (numbers of changes of ARV and hospitalisation) rather than virological (HIV RNA) in this cohort. We believe hospitalisation rates may be a useful marker of HIV disease in cARV treated populations and may identify groups at risk of poorer outcome and in need of intervention.