Objectives: To investigate the ability of several HIV-1 drug-resistance interpretation systems, as well as the number of pre-specified combinations of abacavir-related mutations, to predict virological response to abacavir-containing regimens in antiretroviral therapy-experienced, abacavir-naive patients starting an abacavir-containing regimen in the EuroSIDA cohort.
Patients and methods: A total of 100 HIV-infected patients with viral load (VL) >500 copies/ml who had a plasma sample available at the time of starting abacavir (baseline) were included. Resistance to abacavir was interpreted by using eight different commonly used systems that consisted of rules-based algorithms or tables of mutations. Correlation between baseline abacavir-resistance mutations and month 6 virological response was performed on this population using a multivariable linear regression model accounting for censored data.
Results: The baseline VL was 4.36 log10 RNA copies/ml [interquartile range (IQR): 3.65-4.99 log10 RNA copies/ml] and the median CD4 cell count was 210 cells/microl (IQR: 67-305 cells/microl). Our patients were pre-exposed to a median of seven antiretrovirals (2-12) before starting abacavir therapy. The median (range) number of abacavir mutations (according to the International AIDS Society-USA) detected at baseline was 3.5 (0-8). Overall, the Kaplan-Meier estimate of the median month 6 VL decline was 0.86 log10 RNA copies/ml [95% confidence intervals (95% CI): 0.45-1.24]. The VL in those patients (n=31) who intensified treatment by adding only abacavir decreased by a median 0.20 log10 RNA copies/ml (95% CI: -0.18; +0.94). The proportion of patients who harboured viruses fully resistant to abacavir among the eight genotypic resistance interpretation algorithms ranged from 12% [Agence Nationale de Recherches sur le SIDA (ANRS)] to 79% [Stanford HIV RT and PR Sequence Database (HIVdb)]. Some interpretation systems showed statistically significant associations between the predicted resistance status and the virological response while others showed no consistent association. The number of active drugs in the regimen was associated with greater virological suppression (additional month 6 VL reduction per additional sensitive drug=0.51, 95% CI: 0.15-0.88, P=0.006); baseline VL was also weakly associated (additional month 6 VL reduction per log10 higher=0.30, 95% CI: -0.02; +0.62, P=0.06). In contrast, the number of drugs previously received was associated with diminished viral reduction (additional month 6 VL reduction per additional drug=-0.14, 95% CI: -0.28; 0.00, P=0.05).
Conclusions: Our results revealed a high degree of variability among several genotypic resistance interpretation algorithms currently in use for abacavir. Therefore, the interpretation of genotypic resistance for predicting response to regimens containing abacavir remains a major challenge.