Abstract
A widely used metric of substrate exposure in brain is the brain-to-serum partition coefficient (K(p,brain); C(brain)/C(serum)), most appropriately determined at distribution equilibrium between brain tissue and serum. In some cases, C(brain)/C(serum) can peak and then decrease, as opposed to monotonically increasing to a plateau, precluding accurate estimation of partitioning. This "overshoot" has been observed with compounds that undergo enterohepatic recycling (ER), such as valproic acid (VPA). Previous simulation experiments identified a relationship between overshoot in the C(brain)/C(serum) versus time profile and distribution into a peripheral "compartment" (e.g., the ER loop). This study was conducted to evaluate model predictions of that relationship. Initial experiments tested the ability of activated charcoal, antibiotics, or Mrp2 deficiency to impair VPA ER in rats, thereby limiting the apparent volume of distribution associated with ER. Mrp2 deficiency (significantly) and antibiotics (moderately) interrupted VPA ER. Subsequently, brain partitioning was evaluated in the presence versus absence of ER modulation. Although overshoot was not eliminated completely, deconvolution revealed that overshoot was reduced in Mrp2-deficient and antibiotic-treated rats. Consistent with model predictions, overshoot was higher after antibiotic treatment (moderate ER interruption) than in Mrp2 deficiency (substantial ER interruption). Steady-state K(p,brain) was unaffected by experimental manipulation, also consistent with model predictions. These data support the hypothesis that C(brain)/C(serum) may overshoot K(p,brain) based on the extent of peripheral sequestration. Consideration of this information, particularly for compounds that undergo significant extravascular distribution, may be necessary to avoid erroneous estimation of K(p,brain).