Downloadable Content

Download PDF
Last Modified
  • March 19, 2019
  • Srinivas, Nithya
    • Affiliation: Eshelman School of Pharmacy, Pharmaceutical Sciences Program
  • As of the year 2015, 36.7 million people worldwide were living with HIV infection. While the introduction of highly active antiretroviral therapy (HAART) has greatly reduced the morbidity and mortality of HIV infection, there is still no cure for this disease. In the central nervous system (CNS), HIV RNA in the cerebrospinal fluid (CSF) has been found even in patients who otherwise have viral suppression in the plasma. Further, HIV infection in the brain may lead to the development of HIV-associated neurocognitive disorders (HAND). Milder forms of cognitive decline in HAND remain highly prevalent in people taking HAART and this may be a function of ineffective antiretroviral (ARV) distribution in the brain tissue. However, existing methods only measure ARV pharmacokinetics (PK) in the CSF and are insufficient to explain brain distribution of ARVs. Therefore, the goal of this project was to conduct a comprehensive analysis of ARV penetration into the brain tissue in preclinical models, evaluate the role of drug transporters in modulating ARV brain tissue disposition across species, and develop a model to predict disposition of one ARV (efavirenz [EFV]) in human brain tissue using PK data from preclinical models and determine the relationship between model-predicted drug exposure in the brain and neurocognitive impairment in a cohort of HIV-positive participants. In the first aim of the study, the brain tissue concentration and brain tissue:plasma penetration ratio of six ARVs were determined across two humanized mouse models and one nonhuman primate (NHP) model by LC-MS/MS. ARV brain tissue:plasma concentrations were only preserved across all three species for raltegravir, and showed no differences based on infection status or sex (in the NHPs). In the NHPs, ARV concentrations in the CSF were >6-fold lower than brain tissue. The CSF concentrations were poorly predictive of the brain tissue concentrations for all ARVs except EFV (r=0.91, p<0.001). Mass-spectrometry imaging could only detect EFV distribution within the brain, and greater accumulation was noted in the white matter vs. gray matter. The total colocalization to HIV target cell (microglia and CD4+ T-cells) area in the brain ranged from 45-80%. EFV was the only ARV to achieve concentrations >IC90 in the brain tissue across all NHPs, however, MSI showed that <3% of HIV-target cells contained EFV at concentrations>IC50 of viral replication. In the second aim of the study, we noted significant differences in the gene expression of drug-transporters in the brain tissue across all three species. For example, the gene expression of Abcb1 was 10-fold higher in the hu-HSC-RAG mice compared to the BLT mice. Only BCRP and P-gp proteins were quantified in the majority of the brain tissue samples and there was 16-fold higher BCRP protein in the NHPs relative to the humanized mouse models. There were no differences in the expression of drug transporters due to infection status, but female macaques showed >two-fold higher protein expression of BCRP and P-gp compared to male animals. The protein concentration of transporters in the brain tissue did not predict the brain tissue:plasma concentration of any of the ARVs. In the final aim of the study, we developed an eight-compartment PK model to describe the distribution of EFV into the CSF and brain tissue in rhesus macaques. Using the preclinical model structure and human PK data from a small clinical trial, the brain tissue distribution of EFV was predicted in our cohort of HIV positive participants. At steady state, EFV profile in the brain tissue was predicted to be flat with a median concentration of 8,000 ng/mL. Model-predicted brain tissue exposure showed good agreement with plasma (r=0.99, p<0.001). However, due to the high variability in the CSF measurements, the correlation between brain tissue and CSF concentrations in humans was poor (r=0.34, p=0.11). A 1,000-replicate Monte Carlo simulation of the final clinical model was able to capture observed EFV brain tissue concentration data available from three participants in the National Neuro-AIDS Tissue Consortium (NNTC) repository, indicating the biological plausibility of our model predictions. The model-predicted brain tissue exposures did not show any correlation with neurocognitive scores in the study participants (rho<0.05, p>0.05). Through these experiments, it was determined that ARV penetration into the brain tissue is highly variable across preclinical models. With the limited ARV concentration data that are available from the human brain tissue, drug concentrations achieved in the brain tissue of NHPs closely approximate what is seen clinically. The CSF is not an appropriate surrogate for brain tissue PK of all ARVs investigated except EFV, and our surrogate measures of efficacy for EFV indicate that although the drug achieves high concentrations in the brain tissue, a lack of adequate spatial coverage over HIV-target cells may lead to reduced efficacy. There are several inter-species differences in drug transporter expression in the brain tissue; however, brain tissue transporter expression was not predictive of ARV brain tissue penetration. Finally, our data demonstrate that sparse preclinical and clinical data can be leveraged to predict human brain tissue exposure of ARVs by the use of novel Bayesian models. Our small study suggests that factors other than ARV brain tissue PK may influence HAND persistence.
Date of publication
Resource type
Rights statement
  • In Copyright
  • Forrest, Alan
  • Robertson, Kevin
  • Batrakova, Elena
  • Kashuba, Angela
  • Corbett, Amanda
  • Brouwer, Kim
  • Doctor of Philosophy
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2018

This work has no parents.