Rik de Greef1, Helen Kastrissios PhD, Jean-Michel Gries PharmD,PhD, Lewis B. Sheiner, MD, Terrence F. Blaschke MD.
Stanford University and University of California, San Francisco, CA, USA.1 present affiliation: LACDR, Division of Pharmacology, Leiden University, The Netherlands
Non-compliance with study medications can influence the results of clinical trials. It has been hypothesised that a model-based analysis of actual drug exposure could result in an estimate of the true effectiveness of the drug, instead of the estimate of use-effectiveness that is obtained with an intention-to-treat analysis.
To address these issues in the context of AIDS clinical trials a substudy was performed in AIDS Clinical Trials Group Protocol 175. Patients received dideoxynucleoside monotherapy (ZDV or ddI) or combination therapy (ZDV+ddI or ZDV+ddC). CD4 cell counts, plasma HIV-1 RNA copy numbers and HIV-1 titers in PBMC were the surrogate markers of drug effect. Compliance was monitored for all three drugs (or placebos) in 44 subjects, using electronic devices. The pharmacokinetics of ZDV, ddI and ddC were determined using sparse sampling (up to 8 samples/individual). We used a population analysis to obtain individual estimates for AUC, which were used to calculate individual ‘exposures’ by adjusting for individual compliance. An analysis was performed which sought to account for the variability in the effect of treatment due to factors including baseline characteristics, differences between treatments and compliance or exposure measures, using a stepwise approach. In a generalised additive model (GAM) several descriptors of the change in surrogate markers over time were regressed on linear or non-linear transformations of these factors. Those factors that showed a significant influence on the descriptor were included into the final model.
The results show that, besides several baseline characteristics, only measures of compliance (but not drug exposure) were significantly related to the descriptors of the effect on surrogate markers. The apparent absence of any exposure-response relationship can probably be attributed to the ‘prodrug’ nature of the nucleoside analogues tested in this trial. Thus, compliance with drug therapy is an important determinant of efficacy of antiretroviral therapy with nucleoside analogues.
Reference: PAGE 6 () Abstr 585 [www.page-meeting.org/?abstract=585]
Poster: oral presentation