Objective: To assess the impact of baseline HIV-1 substitutions, individual pharmacokinetic (PK) parameters (Cmin, Cmax, area under the curve [AUC0-->24 h]) and genotype-inhibitory quotient (GIQ) on virological responses (VR) to atazanavir-ritonavir (300 mg/100 mg)-based highly active antiretroviral therapy (HAART) in 71 antiretroviral-experienced, atazanavir-naive patients in virological failure (VF) on HAART.
Methodology: VR was defined as HIV RNA <1.7 log10 copies/ml at week 12 (W12). A clinically relevant genotype-substitutions score for atazanavir-ritonavir was developed and validated (Reyaphar substitutions score). Previously published substitutions scores were also tested.
Results: Patients had a median (Q1; O3) of 6 (3; 8) previous treatment lines during 9 (7; 11) years. Baseline (WO) values were as follows: 262 (187; 435) CD4+/microl, 3.9 (2.6; 4.9) log10 HIV-1 RNA copies/ml, 4 (2; 6) protease substitutions and 3 (1; 4) NRTI-related substitutions. Respective steady-state Cmin, Cmax and AUC0-->24 h were 300 (200; 700) ng/ml, 620 (430; 750) ng/ml and 78,000 (61,000; 94,000) ng.h/ml. At W12, 49% of the patients had VR with a median decrease of -1.2 (-0.5; -2.3) log10 HIV-1 RNA copies/ml. The Reyaphar score included 12 baseline protease substitutions from the International AIDS Society USA list that were associated with poorer VR: L10I/F/R/V, K20I/M/R, L241, M461/L, 154L/M/T/V, L63P, A71I/L/V/T, G73A/C/F/T, V771, V82A/F/S/T, 184V, L90M and the polymorphism substitution Q58E. Comparing <5 versus > or =5 Reyaphar substitutions, the W12-W0 HIV-1 RNA decrease was - 1.4 (-0.7; -2.3) versus -0.5 (-1.2; +0.5) log10 copies/ml (P=0.009) with VR in 63% versus 110% (P<10(-4)), respectively. This score predicted VF at W12 with 46% sensitivity, compared to 33% and 28% for the ANRS 2004 and 2005 scores. PK parameters alone were not associated with VR, but GIQ was associated with virological outcome (P=0.04). 150L, known to be correlated with atazanavir-specific resistance, emerged in 2 (8%) of the 24 failing patients with paired genotypes at WO and VF.
Conclusions: These findings highlight the need to cross-validate genotype-based algorithms to interpret substitution impact on virological outcome using different patient databases before their implementation in routine clinical practice.