ArticlesPerformance Characteristics of Four Free Phenytoin ImmunoassaysRoberts, William L.* ; Annesley, Thomas M.† ; De, Barun K.‡ ; Moulton, Linda§ ; Juenke, JoEtta M.§; Moyer, Thomas P.∥Author Information Department of Pathology at *University of Utah Health Sciences Center, Salt Lake City, Utah, †University of Michigan, Ann Arbor, Michigan, and ‡University of Mississippi Medical Center, Jackson, Mississippi; §ARUP Institute for Clinical and Experimental Pathology, Salt Lake City, Utah; ∥Laboratory Medicine and Pathology, Mayo Clinic, Rochester, Minnesota Received July 13, 2000; accepted November 16, 2000. Address correspondence and reprint requests to William L. Roberts, M.D., Ph.D., ARUP Laboratories, 500 Chipeta Way, Salt Lake City, UT 84108; E-mail: [email protected] Therapeutic Drug Monitoring: April 2001 - Volume 23 - Issue 2 - p 148-154 Buy Abstract The measurement of the unbound or free phenytoin concentration is indicated in several situations, including uremia. In patients with uremia, metabolites of phenytoin and other substances accumulate and can displace phenytoin from its protein binding sites, with a consequent increase in the free fraction of drug. Some of the phenytoin metabolites that accumulate in uremia can cross-react with phenytoin immunoassays. In this study the authors evaluated four free phenytoin immunoassays compared with a high-performance liquid chromatography (HPLC) method: the Roche COBAS Integra, the Syva EMIT 2000, the Opus INNOFLUOR, and the Abbott TDx. All four methods demonstrated good precision, with interday coefficients of variation of ≤5% and comparable recoveries using quality control material. Two of the methods, the EMIT 2000 and COBAS Integra, showed excellent agreement with the HPLC method using samples from patients both with normal renal function and with renal insufficiency. The other two methods, the INNOFLUOR and TDx, showed average positive biases for the therapeutic range of 3–7% and 21–22%, respectively, compared with the HPLC method for samples from patients with normal renal function, and average positive biases of 24–32% and 75–81%, respectively, with samples from patients with uremia. © 2001 Lippincott Williams & Wilkins, Inc.