Background: A detectable HIV viral load after antiretroviral therapy (ART) initiation is associated with poor outcomes; however simple measures of detectable viremia may miss the cumulative effects of exposure to circulating virus over time. Viremia copy-years has been demonstrated to have prognostic value for mortality in resource-rich cohorts. We evaluated the relationship between viremia copy-years and mortality versus dichotomous measures of viremia amongst patients on ART in Johannesburg, South Africa. Methods: We conducted a prospective cohort analysis among treatment naïve patients initiating ART from April 2004-April 2010. Patients included had a CD4 at 6-months and viral loads at 6- 12- and 18-months after ART initiation. Viremia copy-years/mL was calculated as the mean viral load between two measures (6-12 and 12-18-months) times the total days between measures. We used log-binomial regression with Poisson distribution to predict mortality after 18-months on treatment as a function of log10 viremia copy-years and having a detectable viral load at 12- or 18-months. Results: The 4,785 included patients were predominately female (64.1%), median age of 36.3 (IQR: 31.4-42.3), on stavudine-lamivudine-efavirenz (89.5%) and median CD4 count of 89 cells/mm3 (IQR: 34-156) at ART initiation. 170 (3.6%) died and 669 (14.0%) were lost to follow-up. 94.2% achieved viral suppression by 6-months on ART. Median viremia copy-years was 9,175 (IQR: 9,175-26,148.8), equivalent to a viral load of 25 copies/mL/day for one year. In crude models, a detectable viral load at 12- or 18-months was a stronger predictor of death than one unit increase in log10 viremia copy-years (Table). In adjusted models including both measures of viremia, for every log10 increase in viremia copy-years/mL, mortality increased by over 50% (RR: 1.52; 95% CI: 1.28-1.82), while the dichotomous measure of detectable virus showed no association (RR: 1.10; 95% CI: 0.70-1.76). Conclusion: Our results show that higher cumulative viral load (viremia copy-years) is associated with poor treatment outcomes even after accounting for detectable virus. Viremia copy-years should be considered alongside other predictors when assessing risk of death.
Conference: CROI conference 2013, Atlanta, USA