Twenty clinical trials (n=7,970 patients) were included in the review; 4,560 patients received NNRTI-based regimens and 3,410 patients received bPI-based regimens. Fifteen trials were open label and five were double blind.
At 48 weeks, virologic failure was reported in 4.9% (95% CI 3.9% to 6.1%) of NNRTI patients compared with 5.3 % (95% CI 4.4% to 6.4%, p=0.50) of bPI patients. Where failures were successfully genotyped (80% in NNRTI-based and 83% of patients in bPI-based regimens), the M184V mutation in the HIV reverse transcriptase (lamivudine resistance) was reported in 35.3% (95% CI 29.3% to 41.6%) of patients who started NNRTI-based HAART compared with 21.0% (95% CI 14.4% to 28.8%, p<0.001) of those treated with a bPI.
Incidences of virologic failure for the K65R mutation in the HIV reverse transcriptase (multi-nucleoside resistance) were 5.3% (95% CI 2.4% to 9,9%) for NNRTI and 0.0% (95% CI 0.0% to 3.6%, p=0.01) for BPI where patients were treated with non-zidovudine-based regimens. Resistance to the third agent was reported in 53% (95% CI 46% to 60%) of NNRTI patients and 0.9% (95% CI 0.0% to 6.2%, p<0.001) of bPI patients.
Newer NRTIs combined with an NNRTI were not associated with lower rates of NNRTI resistance than older thymidine analogues, nor was the rate of lamivudine resistance different.