Background: Treatment simplification in antiretroviral-experienced patients receiving protease inhibitor (PI)-containing antiretroviral regimens seems safe, but randomized trials have limited power to detect differences in virological rebound (VR) between different switch strategies.
Methods: From the French Hospital Database on HIV, we selected 2462 patients with undetectable viral load (VL) who switched from a first PI-containing antiretroviral combination (cART) to a combination containing efavirenz (EFV), nevirapine (NVP) or abacavir (ABC). Factors associated with VR and with immunological efficacy (gain of > or = 50 CD4(+) cells/microl) were identified by using Cox models.
Results: The 12-month Kaplan-Meier probabilities of VR were 6.8, 13.7 and 12.3% in patients switching to EFV-cART, NVP-cART and ABC-cART, respectively. Factors associated with VR were female sex, younger age, antiretroviral exposure before the first cART, time on first cART, higher VL at first cART initiation, a stavudine/didanosine backbone (rather than zidovudine/lamivudine) after the switch, and a switch to NVP or ABC [respective adjusted hazard ratio versus EFV: 1.53; 95% confidence interval (CI), 1.21-1.94; and 1.53; 95% CI, 1.12-2.08]. When the analyses were restricted to patients who were antiretroviral-naive before their first cART, NVP (but not ABC) was associated with VR. Immunological outcomes did not differ among the three switch regimens.
Conclusion: When VL is undetectable on a first PI-cART regimen, switching to an EFV-containing regimen is more likely to avoid VR than switching to an ABC or NVP-containing regimen. ABC may be an alternative to EFV for patients who were not exposed to antiretroviral before their first cART regimen, after checking for ABC resistance mutations.