Long-term efavirenz use is associated with worse neurocognitive functioning in HIV-infected patients

Qing Ma, Florin Vaida, Jenna Wong, Chelsea A. Sanders, Yu-ting Kao, David Croteau, David B. Clifford, Ann C. Collier, Benjamin B. Gelman, Christina M. Marra, Justin C. McArthur, Susan Morgello, David M. Simpson, Robert K. Heaton, Igor Grant, Scott L. Letendre, for the CHARTER Group

Neurocognitive (NC) complications continue to afflict a substantial proportion of HIV-infected people taking effective antiretroviral therapy (ART). One contributing mechanism for this is antiretroviral neurotoxicity. Efavirenz (EFV) is associated with short-term central nervous system (CNS) toxicity, but less is known about its long-term effects. Our objective was to compare NC functioning with long-term use of EFV to that of a comparator, lopinavir-ritonavir (LPV/r), in a cohort of well-characterized adults. Four hundred forty-five patients were selected from the CNS HIV Antiretroviral Therapy Effects Research (CHARTER) cohort based on their use of either EFV (n = 272, mean duration 17.9 months) or LPV/r (n = 173, mean duration 16.4 months) and the lack of severe NC comorbidities. All patients had undergone standardized comprehensive NC testing. Univariable and multivariable analyses to predict NC outcomes were performed. Compared with LPV/r users, EFV users were more likely to be taking their first ART regimen (p < 0.001), were less likely to have AIDS (p < 0.001) or hepatitis C virus (HCV) coinfection (p < 0.05), had higher CD4+ T cell nadirs (p < 0.001), had lower peak (p < 0.001) and current (p < 0.001) plasma HIV RNA levels, and were less likely to have detectable HIV RNA in cerebrospinal fluid (CSF) (p < 0.001). Overall, EFV users had worse speed of information processing (p = 0.04), verbal fluency (p = 0.03), and working memory (p = 0.03). An interaction with HCV serostatus was present: Overall among HCV seronegatives (n = 329), EFV users performed poorly, whereas among HCV seropositives (n = 116), LPV/r users had overall worse performance. In the subgroup with undetectable plasma HIV RNA (n = 269), EFV users had worse speed of information processing (p = 0.02) and executive functioning (p = 0.03). Substantial differences exist between EFV and LPV/r users in this observational cohort, possibly because of channeling by clinicians who may have prescribed LPV/r to more severely ill patients or as second-line therapy. Despite these differences, EFV users had worse functioning in several cognitive abilities. A potentially important interaction was identified that could indicate that the NC consequences of specific antiretroviral drugs may differ based on HCV coinfection. The complexity of these data is substantial, and findings would best be confirmed in a randomized clinical trial.