Evaluation of chromatographic and kit immunoassay techniques for the measurement of theophylline in serum: a study based on external quality assurance measurements.

The accuracy and precision of assay techniques used to measure theophylline concentrations in human serum were compared using data from 96 samples from the Heathcontrol external quality assurance scheme. Abbott TDX had the highest precision, with a mean coefficient of variation (CV) of measurements of 6.4%, and Ames Fluorostat was most accurate, with a mean bias of +0.1%. Differences between the better techniques, however, were not significant. The Beckman ICS assay gave the lowest precision (CV 9.3%) and, in addition, produced the highest proportion (7.5%) of rejected observations greater than 3 standard deviations from the sample mean. The least accurate method was radioimmunoassay, with a -9.9% bias. Measurements from two samples spiked with paraxanthine demonstrated that 76% of laboratories using high-pressure liquid chromatography (HPLC) were unable to distinguish between paraxanthine and theophylline. The +4% bias observed at low concentrations and the high variability (CV 8.3%) of measurements made by HPLC were thus attributed to interference by paraxanthine present in the sample matrix. Discriminant analysis of a range of HPLC column and mobile phase parameters indicated that separation of theophylline and paraxanthine was achieved by the use of lower flow rates with mobile phases of solvent mixtures with lower proton donator selectivity.