Abstract
Commentary
The transport of AEDs by P-gp and related proteins has been the subject of much investigation and mounting confusion. Initial in vitro studies, employing a neuroectodermal cell line and transfected porcine kidney epithelial cells, suggested that phenytoin was subject to P-gp-mediated efflux (2,3), albeit to a greater extent in cells expressing the rodent form of the protein than its human equivalent. Investigations using genetic knockout mice that are reportedly devoid of functional P-gp at the blood–brain barrier failed to demonstrate enhanced penetration of radiolabeled phenytoin in the brain of mdr1a knockout mice (which most likely is due to the confounding influence of similarly radiolabeled metabolites) but did reveal an intriguing regional selectivity in phenytoin transport, with accumulation in hippocampus but not cerebellum, of the mdr1a/1b double knockout (3,4). A further pharmacokinetic study employing mdr1a knockouts concluded that only topiramate, of seven common AEDs under examination, was potentially susceptible to P-gp-mediated transport (5). Finally, a detailed investigation using multiple experimental techniques, including mdr1a knockouts, Caco-2 cell monolayers, and rhodamine-123 efflux from human lymphocytes, failed to provide any indication that carbamazepine was a substrate for P-gp (6). Thus, the evidence that commonly used AEDs are substrates for transporter-mediated efflux is modest, at best.
These individual studies were somewhat overshadowed, however, by a series of elegant investigations in experimental animals that employed bilateral intracerebral microdialysis and appeared to suggest that several AEDs, including phenobarbital, phenytoin, carbamazepine, lamotrigine, felbamate, and oxcarbazepine (but interestingly not levetiracetam), are transported to some extent by P-gp and/or MRPs (7–9). Such was the clarity and, some might argue, convenience of this evidence that the promiscuity of efflux transporters in terms of their ability to extrude AEDs became accepted wisdom, almost overnight. Any concern regarding species differences or the capacity of these systems to transport AEDs at clinically relevant concentrations was ignored in the wave of optimism arising from the identification of a credible mechanism of drug resistance in epilepsy.
Fortunately, a degree of circumspection has returned to this area of epilepsy research. Two recent publications highlight the difficulty in extrapolating experimental findings between species as well as in relying on a single model or experimental technique to characterize a complex pharmacological relationship. These papers come from the same research group that published the microdialysis data described above but have now turned their attention to more molecular matters. The results make interesting reading, particularly in light of previous findings and the support those findings provided for the drug transporter hypothesis.
The manuscripts, reviewed here, describe a series of in vitro studies examining the interaction between commonly used AEDs and drug transporter proteins, using porcine and canine kidney epithelial cells transfected with mouse and human forms of P-gp and MRPs. In this paradigm, phenytoin and levetiracetam displayed unidirectional transport characteristics of P-gp substrates in cells transfected with the mouse form of the protein but not the human equivalent. Carbamazepine and valproic acid were not transported by cells expressing P-gp of either species, and none of the four AEDs examined in these two research articles was subject to MRP-mediated efflux. When considered alongside the promiscuity of AED transport described in the in vivo rat microdialysis model, these findings are much more conservative in their implications. They not only emphasize a hitherto under appreciated species distinction in the substrate profile of drug transporters (most strikingly exemplified by the apparent efflux of levetiracetam by mouse but not rat P-gp), but perhaps more importantly, they offer no evidence to suggest that any AED is a substrate for any of the major human efflux transporters.
Other recent in vitro studies have been similarly negative, although none have compared the substrate profile of rodent and human drug transporters in a single investigation or afforded a direct contrast with in vivo findings. In 2003, Weiss et al. reported that several AEDs including carbamazepine, phenytoin, lamotrigine, and valproate could interact with human P-gp expressed in porcine kidney epithelial cells but only at concentrations above the upper limit of the therapeutic range (10). Neither phenytoin nor carbamazepine appeared to be a substrate for P-gp-mediated efflux in an in vitro model of the blood–brain barrier employing bovine retinal endothelial cells in monolayer culture (11); and, there was similarly no evident transport of phenobarbital, carbamazepine, vigabatrin, lamotrigine or gabapentin in a subclone of the Caco-2 cell line that demonstrates extensive P-gp expression (12). Interestingly, phenytoin was transported to a modest extent in these cells, but the effect was not reversible in the presence of known P-gp inhibitors, suggesting a lack of selectivity.
This clear disparity in the findings of experimental studies designed to assess the interaction between drug transporter proteins and AEDs is perplexing. With the exception of phenytoin transport by rodent P-gp, in vitro studies are largely negative, while data derived from knockout mice are inconclusive (possibly a result of functional redundancy in transport systems and the upregulation of alternative proteins to replace those that have been deactivated), and levetiracetam aside, in vivo microdialysis results are overwhelmingly positive. Despite considerable research efforts, the fundamental question of whether AEDs are substrates for transporter-mediated efflux remains. The authors of the two manuscripts discussed here quite rightly propose that their in vitro work does not exclude the possibility that antiepileptic agents might be transported to some extent by efflux transporters, but the balance of opinion has shifted in light of their new data, and particularly in relation to the inter-species differences in substrate specificity. While previously AEDs were afforded the benefit of the doubt in order to satisfy a biologically plausible hypothesis, it may now be more appropriate to suggest that, under normal circumstances and until proven otherwise, they are not substrates for human transporter-mediated efflux.
The credence of the drug transporter hypothesis of refractory epilepsy has been diminished by these latest findings, but it may be a little premature to declare it irreparably undermined. There are still many aspects of this theory that require investigation. If the characteristics of human blood–brain barrier transporters are sufficiently distinct from those expressed in other tissues or if the expression and functionality of drug transporter proteins is altered in disease states and/or influenced by genetic factors, then these still represent avenues for potential clinical exploitation. Models and experimental systems that adequately mirror the appropriate physiological and pathological circumstances are required, and there are now clear grounds for a detailed examination of the individual properties of human and rodent efflux transporters and an evaluation of whether any observed disparity might offer further therapeutic advantage. It is possible that the drug transporter hypothesis of refractory epilepsy is one that will never be entirely provable or disprovable. If nothing else, however, it is likely to keep investigators busy for a number of years to come.