The HIV-infected population is heterogeneous in terms of drug disposition and drug response to HIV treatment. Identifying subgroups of patients at risk of suboptimal therapeutic effects or toxicity is critical to maintain drug efficacy and safety. The objectives of this dissertation were to apply nonlinear mixed-effects modeling (NLME) methods to investigate the potential discrepancies in the pharmacokinetic (PK) and pharmacodynamic (PD) profiles of two special populations, pregnant women and older patients, compared to the general HIV-infected population, and inform appropriate HIV treatment for these patients.

In the investigation of lopinavir/ritonavir (LPV/RTV) unbound drug PK in twelve HIV-infected pregnant women, clearances of LPV and RTV during pregnancy demonstrated an exponential increase between 20-32 weeks of gestation. The unbound fractions of LPV/RTV were not significantly different between pregnancy and postpartum. Despite the increased LPV/RTV elimination in pregnancy, >90% effective LPV inhibitory quotient (IQ) values were predicted under the standard dosing (400/100 mg B.I.D.) with ≤4-fold increase in viral half inhibitory concentration (IC50), supporting the use of standard dosing in pregnant women with low-resistance virus. As viral susceptibility decreases (>10-fold increase in viral IC50), higher doses or alternative regimens should be considered.

The effects of aging on ART PK were studied in the context of different aging surrogates and pharmacogenetics in ninety-one HIV-infected patients aged 18-72 years receiving either efavirenz (EFV)- or atazanavir/ritonavir (ATV/RTV)- containing regimens. A cellular aging biomarker, the expression of p16INK4a, was shown to be negatively associated with the unbound drug clearance of EFV, implying that cellular aging might predict slower EFV metabolism. No alterations in drug clearance or unbound fraction with aging were observed for ATV or RTV.

Regarding the recovery of CD4+ T-cells, analyses of long-term data (up to 15 years on treatment) of 102 patients and short-term data (within 2 years) of 20 patients indicated that advanced age was significantly associated with a greater apparent elimination rate constant of CD4+ naive T-cells. Simulation showed that the differences of median total CD4+ T-cell counts between younger (18-35 years) and older (≥50 years) age groups were 93, 137 and 145 cells/microL at year 1, 4 years and steady state after treatment, respectively, suggesting that advanced age may have a greater negative effect on the long-term CD4+ T-cell recovery.

In summary, this dissertation demonstrates the power of NLME approaches in studying antiretroviral PK and PD, and provides insights for the optimization of HIV intervention in pregnant women and older patients.