Analytical Issues Regarding Psychotropic Agents and Weighted Regression for Data Interpretation

Most important for the analysis of psychotropic agents in human plasma or serum are chromatographic methods (1). They use to a lower part gas chromatography, mainly liquid chromatography with ultraviolet, fluorescence or electrochemical detection. Sample preparation can be done by liquid liquid extraction, solid phase extraction and column switching. Today we see a considerable increase of LCMS and LCMSMS methods. LCMS(MS) has high selectivity and high speed of analysis due to simplification of the prepurification. Many compounds can be analysed simultaneously. MS procedures require highly qualified staff. Immunoassays are easy to use and have high throughput, but they show the problem of crossreactivity. Radioreceptor assays give a surrogate of the pharmacological activity instead of plasma concentration.

Only after calibration of the instrumental method it is possible to estimate the plasma concentrations. The basis of calibration and data interpretation in chromatography is linear regression, for which precondition is homogenity of variances. In chromatography this is seldom. In quality control including a working range of factor 2 this does not matter. For use of a working range of 10, 100 or 1000 the heteroscedasticity must be balanced by contrary weighting of residuals (2). This can be done by nearly each integration software, however some of the validation programs do not use any or insufficiently weighting. Creating the guidelines and recommendations, probably jurists and statisticians had too much influence. Therefore „calibration“ is unnecessarily complicated and hardly to understand. Weighted regression is 1). sensible, 2). easily comprehensible, 3). simply to use (3), 4). it demonstrates a convenient way out of the described problems and 5). it allows to estimate measurement uncertainty.

References: (1) Baumann P, Hiemke C, Ulrich S, Eckermann, Gaertner I, Gerlach M, Kuss HJ, Laux G, Müller-Oerlinghausen B, Rao ML, Riederer P, Zernig G; The AGNP-TDM expert group consensus guidelines: Therapeutic Drug Monitoring in Psychiatry 37 (2004) 1–23. (2) Kuss HJ; in: Handbuch der Validierung in der Analytik (Ed.: Kromidas S); Wiley-VCH 2000, p.182. (3) Kuss HJ; Weighted least Squares Regression in Practice: Selection of the Weighting Exponent; LCGCEurope 16 (2003).