Determinants of HIV Progression and Assessment of the Optimal Time to Initiate Highly Active Antiretroviral Therapy

Abstract

We analyze the factors related to progression to AIDS or death in HIV-infected patients from the Proyecto para la Informatización del Seguimiento Clínico epidemiológico de los pacientes con Infección por VIH/SIDA (PISCIS) Cohort and we assess the optimal time to initiate highly active antiretroviral therapy (HAART) taking lead time into account. We selected naive patients who were AIDS-free and initiated HAART after January 1998. Statistical analyses were performed using Cox proportional hazards models. Lead time was defined as the time it took the deferred group with an early disease stage to reach the later stage. The analysis accounting for lead time was performed using multiple imputation methods based on estimates from the pre-HAART period as described elsewhere. Multivariate analysis on 2035 patients (median follow-up = 34.3 months) showed significantly higher hazard ratios (HRs) for a CD4 count <200 cells/microL (HR = 3.79, 95% confidence interval [CI]: 2.18 to 6.57), HIV-1 RNA level >100,000 copies/mL (HR = 1.84, 95% CI: 1.26 to 2.69), and hepatitis C virus (HCV) coinfection (HR = 2.40, 95% CI: 1.65 to 3.49), whereas a lower risk was found for those who started HAART between January 2001 and June 2004 (HR = 0.55, 95% CI: 0.30 to 0.90). When lead time and unseen events were included, we found a higher risk of progression to AIDS among patients who deferred treatment when the CD4 count reached <200 cells/microL (HR = 2.97, 95% CI: 1.91 to 4.63) and 200 to 350 cells/microL (HR = 1.85, 95% CI: 1.03 to 3.33) compared with those who started treatment with CD4 counts from 200 to 350 cells/microL and >350 cells/microL, respectively. Advanced HIV disease, HCV coinfection, and early HAART period were determinants of AIDS progression or death. Lead-time analysis in asymptomatic HIV-infected patients suggests that the best time to start HAART is before the CD4 count falls to lower than 350 cells/microL.