Background:  Most of the current resistance algorithms for fosamprenavir/ritonavir (FosAPV/r) are adapted from results obtained for boosted or unboosted amprenavir (APV). We defined a clinically relevant genotypic score and C_min cut-off to predict virological response to FosAPV/r-based regimens in protease inhibitor-experienced HIV⁺ patients.

Methods:  Patients with virological failure (HIV RNA >1.7 log₁₀ copies/mL) were included in a prospective cohort study and received a FosAPV/r -based regimen (700 mg/100 mg twice daily). Sequencing of HIV reverse transcriptase and protease was performed at baseline. Virological response at week 12 was defined by HIV RNA <400 copies/mL or a decrease in viral load of at least 1 log₁₀. APV C_min, C_max, and AUC were determined by high-performance liquid chromatography (HPLC) at week 4.

Results:  FosAPV/r-containing regimen was initiated in 121 patients. Median (Q1;Q3) prior ART exposure was 8 (5; 10) years with 9 (3; 15) previous treatment lines. At baseline, median plasma HIV RNA was 4.4 (3.6; 5.1) log₁₀ copies/mL. Number of total, major and minor protease inhibitor (PI) mutations were 6, 2, and 5, respectively, and 5 nucleoside reverse transcriptase inhibitor (NRTI)-related mutations. C_min, C_max, AUC and genotypic inhibitory quotient (GIQ) were 1400 (800; 1800) ng/mL; 4650 (3600; 6150) ng/mL, 35 (27; 45) mg·h/L and 315 (95; 1200), respectively. At week 12, 51% of patients were considered virological response. Median HIV RNA decrease was –0.8 (–2.8; 0.04) log₁₀ copies/mL. Associated with reduced virological response were 12 baseline protease mutations among the IAS list at codons 10, 33, 36, 46, 54, 62, 63, 71, 73, 82, 84, 90, and 4 polymorphism mutations at codons 13, 19, 89, and 55 (corrected for multiple testing, p <0.003) and included in the Zephir mutation score. When it was <4 vs ≥4 mutations, DM3-M0 HIV RNA was -2.4 vs –0.09 log₁₀ copies/mL (p <10^­4) and virological response occurred in 91% vs 17% (p <10^­4). This score predicted failure at week 12 with 93% sensitivity and 78% specificity. Similar results were obtained without taking into account polymorphism mutations. We cross-validated the ANRS score which did not include the L90M in Zephir patients and found a sensitivity to predict failure of 18%. C_min, C_max, AUC and GIQ were associated with virological response (p <10^-4). When C_min cut-off was <1600 vs ≥1600 ng/mL, DM3-M0 HIV RNA was -2.4 vs –0.2 log₁₀ copies/mL and virological response occurred in 79% vs 31% (p <10^­4).

Conclusions:  In experienced patients, the proposed score, pharmacokinetic parameters and GIQ were predictive of virological response at week 12. These findings highlight the need to cross-validate genotype interpretation algorithms in patients from different database before their use in routine clinical practice.