Background: Darunavir (DRV, TMC-114) is the most recently approved protease inhibitor (PI). Mutations correlated with reduced clinical response to DRV are similar to those known to confer reduced susceptibility to amprenavir (APV). We examined PI cross-resistance patterns and predictive accuracy of a simple DRV mutation score genotype interpretation among samples tested in the Monogram Clinical Reference Laboratory.

Methods:  Fold-change in IC₅₀ data from 1631 samples tested since May 2006 (PhenoSense HIV) with at least 1 major PI mutation were analyzed. A DRV mutation score was calculated as the number of mutations present among V11I, V32I, L33F, I47V, I50V, I54L or M, G73S, L76V, I84V, and L89V. Samples with mixtures at any of these positions were excluded. The presence of 3 or more of these mutations is considered to indicate reduced susceptibility. Samples were classified as drug sensitive (S), partially sensitive (PS), or resistant (R) based on lower and upper clinical cut-offs. Mutations in gag or protease associated with reduced susceptibility (RS, fold change >2) were determined using Fisher’s exact test.

Results:  The coefficient of correlation R² for log-transformed PI fold-change data was highest between DRV and APV (0.90), intermediate for lopinavir (0.60), and relatively low for other PI (range 0.25 to 0.51). However, all 305 APV PS samples retained DRV susceptibility, and only 11% of 301 APV R samples were DRV R; 52% were PS to DRV. The percentage of samples with 1 (n = 400) or 2 (n = 355) DRV mutations, which were not DRV S was 12.5% and 50%, respectively. Conversely, 32% of samples with 3 mutations (n = 165) were DRV S. When present as the only major PI mutation, D30N (n = 9), I50L (n = 5), and N88S (n = 5) were associated with increased susceptibility to DRV (median fold change 0.52, 0.36, 0.25, respectively; corresponding median APV fold change 0.79, 0.67, 0.13). There was extensive overlap between mutations associated with RS to both APV and DRV (38 of 65 mutations shared), including all DRV mutations.

Conclusions:  DRV and APV in vitro susceptibility patterns are very similar. Predicted incidence of clinically meaningful cross-resistance is low, due to differences in clinical cut-offs, which are higher for DRV. The expected increased efficacy of DRV compared to APV in PI-experienced patients is most likely a result of higher potency (16-fold lower IC₅₀ in the PhenoSense assay) and ~2-fold higher free drug levels in plasma, rather than a unique cross-resistance profile.