Fluorescence polarization immunoassay: Does it always represent a reliable method to monitor treatment with teicoplanin? Comparison with data obtained by high-performance liquid chromatography

As with vancomycin, treatment with teicoplanin requires monitoring and dosage adjustment. As is often the case in dose adjustment, the decision is essentially based on measurements of trough concentrations (C-min), i.e. 10-12 mu g/ml in the case of teicoplanin. It appeared essential to validate the polarization immunoassay technique (FPIA) vs HPLC for low values as well as higher values. We compared an HPLC assay method with FPIA for the measurement of teicoplanin in the serum of 25 ICU patients (i.e. 211 samples). Equation C1 of the global correlation for HPLC vs. FPIA was: FPIA = 3.4398 + 0.8050HPLC (R = 0.9716). The Student test revealed a significant difference between the two techniques. The real difference between the two techniques was assessed by means of an accuracy test: the FPIA method appears to be inaccurate over the concentration range analyzed in the study. Further, we verified whether the FPIA technique remained constantly inaccurate over C1. A partial correlation C2 corresponding to concentrations less than or equal to 12 mu g/ml (n = 112) was constructed. Application of the accuracy test led to the same conclusion as for C1: over this range of concentrations, the FPIA method remained inaccurate. The FPIA:HPLC concentration ratio, (1.09 +/- 0.29 (0.49 - 2.15)), indicates that the assays performed by FPIA tended to be overestimated. HPLC method remains the analytical reference and should be preferred to FPIA for the assay of TCN for pharmacokinetic or pharmacological research. For treatment monitoring, FPIA represents a simple method and also an alternative solution when HPLC is not available. Correlation observed with the reference chromatographic technique remains acceptable for concentrations of the order of 12 mu g/ml and over while, for low concentrations, the FPIA method appears inaccurate. (C) 1997 Elsevier Science B.V.