In vitro models to predict the in vivo mechanism, rate, and extent of placental transfer of dideoxynucleoside drugs against human immunodeficiency virus

OBJECTIVE: We tested the hypothesis that the mechanism, rate, and extent of in vivo placental transfer of dideoxynucleoside drugs against human immuno deficiency virus can be predicted by the in vitro perfused human placenta a nd the drug octanol-water partition coefficient. STUDY DESIGN: Near-term pregnant macaques (Macaca nemestrina) underwent lon g-term catheterization for the administration of 4 dideoxynucleosides again st human immunodeficiency virus: zidovudine, didanosine, zalcitabine, and s tavudine. Maternal plasma, fetal plasma, and amniotic fluid concentrations were determined frequently after intravenous bolus and/or infusion of the d rugs administered into the maternal or fetal circulation on separate occasi ons. Antipyrine was included in all experiments as a marker of placental bl ood flow. The mechanism, rate, and extent of placental transfer of the 4 di deoxynucleosides in perfused human placenta were determined and compared wi th the findings obtained by others. RESULTS: The mechanism and rate of the antipyrine-normalized placental tran sfer of the 4 dideoxynucleosides in perfused human placenta were highly cor related with those observed in vivo. The extent of placental transfer (feta l/maternal steady-state plasma concentration ratio) was also highly correla ted with both the antipyrine-normalized placental transfer clearance (clear ance index) determined in the in vitro perfused human placenta model (r(2) = 0.95, in vitro clearance-index model) and the drug octanol-water partitio n coefficient (r(2) = 0.99, in vitro partition-coefficient model). To deter mine the predictive capacity of these correlative models, we predicted the fetal/maternal steady-state plasma concentration ratio of each drug after e xcluding the data on that drug from the model fit. Both in vitro models to predict in vivo placental transfer of drug models resulted in good predicti ons of the observed fetal/maternal steady-state plasma concentration ratio (mean error: in vitro clearance-index model = -1.2%; in vitro partition-coe fficient model = 3.9%). CONCLUSIONS: We propose that our models will accurately predict the extent of placental transfer of dideoxynucleoside drugs against human immunodefici ency virus. The models may also be applicable to other classes of drugs, re gardless of therapeutic category, provided that these drugs passively diffu se across the placenta. Such a result will expedite phase 1 clinical trials of drugs in pregnant women.