Abstract Body

Background: Neurocognitive (NC) impairment (NCI) is common among treated HIV+ adults and can occur for several reasons, including production of viral proteins in the CNS, neuroinflammation, and antiretroviral therapy (ART) neurotoxicity. To date, clinical trials have focused on treatment of existing NCI and have been largely inconclusive. We conducted a clinical trial to determine whether one ART regimen would better prevent NCI than another.

Methods: 250 HIV+, ART-naive adults were randomized to either open-label zidovudine-lamivudine-nevirapine (ZLN) or tenofovir-lamivudine-efavirenz (TLE) at two hospitals in Beijing. All subjects were NC normal and had CD4+ T-cell counts below 350/mm3 at entry. Subjects were followed for 96 weeks with comprehensive NC testing and safety assessments. NC performance was adjusted for local normative data and summarized by the Global Deficit Score (GDS) and by the proportion of subjects who developed incident NCI. NC change was summarized using a regression-based summary change score (SCS). Intent-to-treat completer (ITT-C, n=239) and as-treated (AT, n=189) analyses were performed.

Results: The treatment groups had similar demographic and disease characteristics, including plasma HIV RNA (mean 4.2 log10 copies(c)/mL in both groups) and CD4+ T-cell counts (mean 235.1 (ZLN) vs. 222.1 (TLE), p=0.32). GDS values at entry were also similar (mean 0.12 (ZLN) vs. 0.14 (TLE), p=0.15). A high proportion of subjects attained virologic suppression: 92% had plasma HIV RNA below 50 c/mL in both groups at 48 weeks (p=.65, ITT-C). TLE was associated with greater risk of incident NCI over 96 weeks than ZLN in either ITT-C (p=.009) or AT (p=0.037) analyses. Consistent with this finding, the SCS also differed between the groups at 96 weeks (ITT-C: d=.27, p=0.016; AT: d=.28, p=0.034). Safety assessments identified that 73/128 (57.0%) subjects in the ZLN group had at least one adverse event compared with 42/122 (34.4%) in the TLE group, leading to greater discontinuation of study drugs in the ZLN group (38% vs. 2%, p<.001). The most common adverse events were liver toxicity and bone marrow suppression.

Conclusions: This is the first randomized clinical trial to demonstrate that ART regimens differ in preventing NC decline using comprehensive NC testing. Possible explanations include differences between the regimens in drug distribution into the CNS and neurotoxicity. The substantial difference in other adverse events dictates additional research to identify a less toxic alternative to ZLN.