Central nervous system penetration-effectiveness rank does not reliably predict neurocognitive impairment in HIV-infected individuals

摘要

IntroductionCentral nervous system (CNS) penetration-effectiveness (CPE) rank was proposed in 2008 as an estimate of penetration of ARV regimen into the CNS, and validated as predictor of CSF HIV-1 replication. Results on predictive role of CPE on neurocognitive and clinical outcome were conflicting.Materials and MethodsRetrospective, cross-sectional analysis of neurocognitive profile in HIV-infected cART-treated patients. All patients underwent neuropsychological (NP) assessment by standardized battery of 14 tests on 5 different domains. People were classified as having NCI if they scored >1 standard deviation (SD) below the normal mean in at least two tests, or >2 SD below in one test. Linear and logistic regression analyses were fitted using as outcome Npz8 and impairedot impaired respectively.ResultsA total of 660 HIV-infected cART-treated individuals from 2009 to 2014, contributing a total of 1003 tests (mean age 49 (IQR 43–56), male 82%; median current CD4 586/mm3; 18% HCV infected; HIV-RNA <40 cp/mL in 84%). Current ARV regimen was 2NRTIs+1NNRTI 50.3%, 2NRTI+1PI/r in 32.6%, NRTI sparing in 11.1%. Mean CPE of current regimens was 6.6 (95% CI 6.5–6.7). As per test multivariable analysis, higher CPE values were associated to poor NP tasks (Beta=?0,09; 95% CI ?0,14 ?0,03; p=0.002 at multivariable linear regression). The association between higher CPE and increased NCI risk was confirmed at multivariable logistic regression, with a 1.24-fold risk of NCI occurrence for each point increase of CPE of current regimen at the time of NP testing (see Table 1). In a sensitivity analysis performed only on patients at the first NP test, the association between higher CPE and poor NP tasks and enhanced NCI risk was only marginally confirmed (Beta=?0,05; [?0,12–0,02]; p=0,19; OR 1,13 [0,95–1,34]; p=0.17). Older age, longer time from HIV diagnosis, current CD4 count <350 cell/mm3 and lower education level were all associated to an increased risk of NCI. Table 1Multivariable analysis of predictors of neurocognitive impairment by logistic regression model