Effect of serum albumin on in vitro displacement of valproic acid by ceftriaxone and nafcillin.

CLINICAL CHEMISTRY, Vol. 36, No. 5, 1990 825 trations of citrate. This effect is similar to, but not as marked as, that reported by Gawoski and Walsh. The total calcium concentration measured with the Synchron CX-5 in a sample with citrate concentration of 6.6 mmol/L was only 0.15 mmoIIL less than that in the plasma pool without added citrate. The level of interference due to citrate in all the total calcium methods tested was the same whether disodium citrate or citric acid was added. Both the CX-5 and the Ektachem 700 use arsenazo III dye at pH 5.6 to measure total calcium concentration. The most obvious difference between the two methods is the sample dilution used. The CX-5 method operates with a 100-fold sample dilution with buffered dye reagent, whereas the plasma sample provides the only liquid for the reaction in the dry-slide method. Therefore, the calcium-citrate complex in the sample will be in much higher concentration in the dry chemistry method than in the wet chemistry method. Arsenazo IH dye forms a 1:1 molar complex with calcium, and the affinity constant, pK , is affected by the pH, the concentration of sodium chloride, and the buffer strength used in the reaction (2). Increasing the concentration of sodium chloride from 0 to 100 mmol/L decreases the binding constant by a factor of 5.6 at pH 8.0. Also, increasing the concentration of 4-(2hydroxyethyl)-1-piperazineethane sulfonic acid (HEPES) buffer from 10 to 200 mmol/L at pH 7.4 decreases the binding constant by a factor of 12(2). This effect was also seen with imidazole HC1 and Tris HC1 buffers. Because the exact details of the composition of the arsenazo ifi dye reagent are not available from the manufacturers, possible differences between the composition of the two arsenazo ifi reagent systems may also affect the levels of citrate interference nytoin and salicylic acid have also been reported (3, 4). As we continue our studies of the interactions of antiepileptics and antibiotics, begun with phenytoin vs ceftriaxone and nafcillin (5), we report our findings on the displacement of valproic acid by these two antibiotics. Serum from patients with normal renal function receiving valproic acid was pooled to yield concentrations of serum albumin of 26 and 36 g/L, respectively. Ten to 15 L of pure ceftriaxone (Hoechst-Roussel Pharmaceuticals, Inc., Somerville, NJ) and nafcillin (Wyeth Laboratories, Inc., Philadelphia, PA), dissolved in de-ionized water, was added to 1.0-mL allquots of the pooled serum to give ceftriaxone concentrations of 150 and 200 mgfL and nafcillin concentrations of 50 and 75 mgfL. These concentrations approximate the therapeutic and toxic concentrations, respectively, in serum of patients receiving these two antibiotics. After incubation at 37 #{176}C for 12 h, protein-free filtrates were prepared by centrifugal ultrafiltration (Centrifree Micropartition; Amicon, Danvers, MA) at 1168 x g for 30 mm. To quantifSr the free valproic acid in the protein-free filtrate, we used a fluorescence polarization assay (Abbott Laboratories, North Chicago, IL). We determined the total valproic acid concentration in the original sample simJohn Davis ilarly. We observed a statistically significant increase in the free fraction of valproic acid in the ceftriaxoneenhanced samples at low concentrations of albumin, indicating drug displacement (Table 1); displacement of valproic acid by nafcillin was not statistically significant. The presence of decreased concentrations of albumin was accompanied by relative increases in free valproic acid with both ceftriaxone and nafcillin. Albumin concentrations thus seem to affect the ratio of free to total valproic acid in a manner similar to that described for phenytoin, where ceftriaxone and nafcillin both produced statistically significant