CSF antiretroviral drug penetrance and the treatment of HIV-associated psychomotor slowing Article abstract—The authors evaluated whether highly active antiretrovi- ral therapy (HAART) with multiple CSF-penetrating drugs results in greater improvement in HIV-associated psychomotor slowing than HAART with a single CSF-penetrating drug. Both groups had improvement in CD4 count, plasma viral load, as well as two tests of psychomotor speed. Comparing the two groups, there were no differences in the mean change for CD4 count, viral load, or any of the neuropsychological tests. Multiple and single CSF-penetrating HAART may be equivalent for treating HIV-associated psychomotor slowing. NEUROLOGY 2001;57:542–544 N. Sacktor, MD; P.M. Tarwater, PhD; R.L. Skolasky, MA; J.C. McArthur, MB, BS, MPH; O.A. Selnes, PhD; J. Becker, PhD; B. Cohen, MD; and E.N. Miller, PhD; for the Multicenter AIDS Cohort Study (MACS)* Psychomotor slowing is a sensitive index of HIV-1- associated dementia complex (HIV dementia), 1 and highly active antiretroviral therapy (HAART) is as- sociated with improved psychomotor speed perfor- mance in HIV-seropositive patients with cognitive impairment. 2 However, CSF penetrance of antiretro- viral drugs is variable. In addition, CSF drug levels may bear little relationship to brain extracellular fluid drug levels. 3 Zidovudine, stavudine, abacavir, and nevirapine have the best CSF penetration among the nucleoside and non-nucleoside reverse transcriptase inhibitors. 4 The protease inhibitors have uniformly poor CSF penetration, with indinavir having the best penetra- tion. 4 It is assumed that a combination antiretroviral regimen with better CSF penetration would be more effective in treating HIV dementia. However, it is unknown whether HAART with multiple CSF- penetrating drugs would be more useful for treating HIV-associated cognitive impairment than HAART with a single CSF-penetrating drug. We evaluated whether psychomotor slowing has greater improve- ment with multiple CSF-penetrating HAART com- pared with single CSF-penetrating HAART in HIV- seropositive patients with psychomotor slowing. Methods. The study included 73 HIV-seropositive gay/ bisexual men on HAART with psychomotor slowing in the Multicenter AIDS Cohort Study (MACS) studied from Oc- tober 1995 to September 1998. Psychomotor slowing was defined as performance on one of several tests of psy- chomotor speed that was at least 1 SD below the mean for age- and education-matched HIV-seronegative men. 5 These tests included the Grooved Pegboard Test (GP) (nondomi- nant or dominant hand), 6 Trail Making Test Part B, 7 and the Symbol Digit Modalities (SDMT). 8 Longitudinal neuro- psychological testing data from six study visits in HIV- seropositive participants with psychomotor slowing initiating HAART were examined. Participants with CNS opportunistic infections or lymphoma were removed from the analysis. Participants were classified by treatment into two groups: multiple CSF-penetrating HAART (n = 55) and single CSF- penetrating HAART (n = 18). Multiple CSF-penetrating HAART contained two or more antiretroviral drugs with good CSF penetration. Single CSF-penetrating HAART con- tained only one antiretroviral drug with good CSF penetra- tion. CSF-penetrating drugs were defined as zidovudine, stavudine, abacavir, nevirapine, and indinavir, based on studies of brain tissue penetration and CSF drug concentra- tions after chronic oral dosing. 4 During this study period, all HAART regimens contained at least one drug with good CSF penetration. Differences in demographic characteristics across ther- apy groups were assessed using t-tests for continuous data and  2 tests for categorical data. Normalized Z-scores were used to evaluate performance for each neuropsychological test. Covariates and potential confounders of the relation- ship between therapy and change in neuropsychological performance were assessed using regression models. Gen- eralized estimating equations were used to account for re- peated neuropsychological test measurements from all six study visits. Variables included in these models were the fixed covariates of age, years of education, race, and neuro- psychological test score at baseline visit (defined as the visit before initiation of HAART) and the time-dependent covariates of CD4 cell count, plasma viral load, and HIV disease state (Centers for Disease Control and Prevention stage; i.e., asymptomatic, symptomatic, or AIDS). Details of this model were described previously in a similar analy- sis that demonstrated improved psychomotor speed perfor- mance in participants on HAART compared with participants on no treatment or monotherapy. 2 Results. Fifty-five participants were on a multiple CSF- penetrating HAART regimen, and 18 participants were on a single CSF-penetrating HAART regimen. Baseline demo- graphic characteristics were similar among the two treat- ment groups, as shown in table 1. Subjects in both groups had moderate immunosuppression and an unsuppressed *See Appendix on page 544 for a list of the members of the Multicenter AIDS Cohort Study (MACS). From the Department of Neurology (Drs. Sacktor and Selnes and J.C. McArthur), Johns Hopkins University School of Medicine, and the Depart- ment of Epidemiology (Dr. Tarwater and J.C. McArthur and R.L. Skolasky), Johns Hopkins School of Hygiene and Public Health, Baltimore, MD; Neu- ropsychology Research Program (Dr. Becker), University of Pittsburgh Medical Center, Pittsburgh, PA; Department of Neurology (Dr. Cohen), Northwestern University, Chicago, IL; and Neuropsychiatric Institute (Dr. Miller), University of California, Los Angeles, CA. Supported by NIH grants AI 35039, AI 35040, AI 35041, AI 35042, AI 35043, RR 00722, and NS 26643. Received October 12, 2000. Accepted in final form March 27, 2001. Address correspondence and reprint requests to Dr. N. Sacktor, Depart- ment of Neurology, Johns Hopkins Bayview Medical Center, 4940 East- ern Ave., B-Building, Room 122, Baltimore, MD 21224; e-mail: sacktor@jhmi.edu 542 Copyright © 2001 by AAN Enterprises, Inc.