Background: Patients showing virologic failure to available antiretroviral classes have few residual treatment options. Lopinavir (LPV)/r and amprenavir (APV) show less cross resistance to other protease inhibitors (PI) but preliminary retrospective studies have shown negative pharmacokinetic (PK) interactions between these 2 agents.

Methods: Patients with virologic failure to all 3 available antiretroviral classes were enrolled in a prospective non-controlled, 24-weeks pilot study of NRTI(s) + LPV/r (400/100 mg BID) + APV 600 mg BID. Baseline  genotypic (TruGene ,VGI) and phenotypic (Antivirogram, Virco) resistance was assessed but not used to guide treatment. Drug levels of RTV, LPV, and APV were assessed at week 2 at morning pre-dose and 1, 2, and 3 hours post-dose by LC-MS/MS (VircoPlasmagram, Tibotec-Virco). Inhibitory quotient (IQ) was calculated by Cmin/phenotypic fold-resistance. ITT-LOCF was used for all analyses. Linear and logistic regression were used to identify predictors of virologic response.

Results. 22 patients were enrolled, all have completed follow-up (17 male, median age 37 years). At entry, median time on HAART was 45 months (12-61), median CD4 177 cells/mL and median viral load (VL) 4.8 log copies/mL. Most frequently associated NRTIs were ABC+d4T or ddI (16). 11 (50%) patients presented grade 3-4 laboratory adverse events (AE) (most hypertriglyceridemia) and 6 (27%) AE-related treatment discontinuations. Mean (SD) changes from baseline VL were -1.18 (0.92) and -1.13 (1.30) log at week 8 and week 24. Mean (SD) CD4 change at week 24 was +88 (87) cells/L. Mean (SD) drug levels at pre-dose and 1, 2, and 3 hours post dose were for RTV: 262 (194), 435 (281), 520 (281), and 522 (348) ng/mL; for LPV: 5298 (2795), 5520 (2521), 7311 (2854), and 7289 (2898) ng/mL; for APV: 1180 (711), 3226 (1909), 2853 (1523), and 2157 (952) ng/mL. While RTV and LPV levels were those expected, APV levels were lower (-37% at pre-dose, -25% at 1 hour, -18% at 2 hours, and -29% at 3 hours) than those from reference curves of same dose RTV+APV. 14/22 patients isolates, showed a mean 3.5 (95% CI 2.1-4.9)-fold resistance to APV, and mean 27.7 (11.2-44.2) fold resistance to LPV. Higher levels of LPV at 3 hours post dose were independently predictive of the maximal VL reduction between weeks 2 and 16 (for each 100 mg/mL increase: beta -0.54, slope -0.17 log copies/mL, p<0.01; model R² =0.45), while more thymidine analogue mutations (TAMs) were associated with worse 24-week VL response (ordinal variable) after adjusting for other mutations (for each more: beta -0.48, p=0.01; model R² =0.56). More TAMs (OR 0.3, 95% CI 0.1-1.0) and more LPV mutations (11 codons considered) (OR 0.5, 0.2-1.0) were negative predictors of 1 log VL reduction from baseline at week 24. Preliminary analysis showed higher log IQ for APV (p=0.09) and LPV (p=0.05) associated with >1 log VL reduction from baseline to week 2-16.

Conlusions: When APV is co-administered with LPV/r, plasma APV levels are lower than expected. Salvage therapy with this combination + NRTIs in 3-class experienced patients yields significant CD4 response and partial virologic response. LPV drug levels predict 2-16 week, while LPV-resistance mutations and TAMs predict 24-week virologic responses.