Letter to the Editor Probenecid-Associated Alterations in Valproic Acid Pharmacokinetics in Rats: Can in Vivo Disposition of Valproate Glucuronide Be Predicted from in Vitro Formation Data?

I read with interest the recent article by Ward et al. (2000) on the interactions between probenecid and valproic acid in rats. The authors attempted to predict the in vivo disposition of valproate glucuronide in the presence of probenecid using in vitro metabolism experiments and concluded that the in vitro prediction approach was not successful in this case. Although the authors provided two hypotheses as possible explanations (effects of probenecid on translocation of valproate glucuronide from the hepatocyte into blood and on the oxidative metabolism of valproate), it may be important to consider a few other explanations. In the in vivo infusion studies, probenecid coadministration caused a significant decrease in the systemic clearance of valproate and this was attributed to competitive inhibition of valproate glucuronide formation by probenecid, based solely on an in vitro study in rat liver S9. There are two caveats in drawing this inference. First, data from these two studies by themselves do not necessarily establish a causeeffect relationship between valproate glucuronidation rate, in vitro, and the systemic clearance of valproic and/or its glucuronide metabolite, in vivo, because probenecid cannot only affect glucuronidation but also oxidative metabolism and is known to be a substrate for many transporters, as the authors have acknowledged. Second, although the Ki for the inhibition of valproate glucuronide formation, in vitro, by probenecid in this study was 876 mg/ml, the target total plasma concentration of probenecid in the in vivo study was only 50 to 75 mg/ml (plasma probenecid concentrations were not determined in this study), i.e., less than one-tenth the Ki value. The unbound probenecid concentrations in plasma and presumably in the cytosol were probably less than 1% of the in vitro Ki, because probenecid is highly bound to serum proteins (Vree et al., 1993). Inhibition of valproate glucuronide formation in vivo under those conditions would therefore be unlikely, i.e., lack of any change in the valproate glucuronide biliary excretion by probenecid in this study is only to be expected. Furthermore, the calculated valproate glucuronide biliary excretion rate in the absence of probenecid treatment does not show clear concentration dependence (see Fig. 4 in Ward et al., 2000), which may suggest that the processes involved in the hepatobiliary disposition of this substrate are nearly saturated at the concentrations of valproate glucuronide employed in this study and in turn would preclude the possibility of observing a discernible effect of any inhibitor of those processes. This study serves to underline the difficulties involved in studying the in vitro-in vivo correlation for drug interactions when the substrate or the inhibitor modulate multiple ADME (absorption, distribution, metabolism, and excretion) processes and the need to design such studies with agents that exhibit a high degree of selectivity to the pathways in question.