Background:  ART with protease inhibitors (PI) alone has been proposed in ART-naive patients for avoiding drug exposure to other classes which increases the risk of side effects and viral resistance. Experience on dual-PI regimens in naive patients remains limited.
Methods:  ANRS 127 was a randomized pilot trial of fosamprenavir (fAPV; 700 mg twice daily) + atazanavir (ATV, 300 mg once daily)/ritonavir (r) (100 mg twice daily) (arm 1) and saquinavir (SQV, 1500 mg once daily) + ATV (300 mg once daily)/r (100 mg once daily) (arm 2) in ART-naive patients with at baseline, viral load 10,000 to 750,000 copies/mL, CD4>200 cells/mm³ and no evidence of PI resistance. Primary efficacy endpoint was the early virologic success defined as a viral load<50 copies/mL at week 16. Any ART modification before week 16 was analyzed as a failure. To challenge conventional HAART, each regimen had to warrant (1-sided 95%CI) at least 50% early successes. Data are presented as medians (range). Viral load reduction from baseline was estimated using a censored method.

Results:  A total of 61 patients were enrolled, 30 in arm 1 and 31 in arm 2. At baseline, both arms were well matched:  viral load was 4.8 log copies/mL (4.0 to 5.7) and CD4 was 271 cells/mm³ (197 to 740). Therapy was modified before week 16 for 8 of 61 patients:  2 in each arm for adverse events, 2 in arm 2 for unprotocolled SQV dose increase, and 2 in arm 1 for compliance or insufficient virologic reduction. At week 16, 12 of 30 (40.0% 95%CI 22.5 to 57.5) and 13 of 31 (41.9% 95%CI 24.6 to 59.3) were viral load<50 ccopies/mL in arm 1 and arm 2 respectively, with a 3.1 and 2.9 log copies/mL drop in viral load in arm 1 and arm 2, respectively. At week 24, 14 of 30 (46.7% 95%CI 28.8 to 64.5) and 13 of 31 (41.9% 95%CI 24.6 to 59.3) were Viral load<50 copies/mL in arm 1 and arm 2, respectively. Therapy was changed for patients with viral load >400 copies/mL at week 16 or viral load >50 copies/mL at week 24. CD4 count increased by 149 cells (–134 to 464) at week 24. At week 4, expected PI plasma C_min were obtained. Baseline viral load was the only significant predictor of a viral load <50 copies/mL at week 16 under treatment; response rates were 63.6% (95%CI 47.2 to 80.0) below the 50,000 copies/mL viral load cut-off (4.7 log), and 15.8% above (p = 0.003). At week 16 bilirubinemia was higher in arm 2 (42 vs 26 mmol/L, p = 0.007), and triglyceride and LDL cholesterol in arm 1 (1.4 vs 1.1 mmol/L, p = 0.036 and 3.7 vs 2.8 mmol/L, p = 0.026). No major genotypic changes in protease occurred.

Conclusions:  Despite good tolerability and adequate drug levels, neither dual-PI regimen reached the specified threshold of early efficacy. This insufficient antiviral potency restricts the potential use of these regimens in ART-naive patients to those with low to intermediate viral load.