Same sample, different result: Variation in clozapine assay measurement in four UK laboratories
Matthew Atkins, James Fritz, Mary Rose Hilaire, David TaylorBackground
Laboratory measurement of clozapine blood concentrations is central to safe and effective treatment and is commonly assumed to be accurate and reproducible across providers. Chromatographic assays require batch-specific setup and are performed using different equipment and techniques, which may introduce error or variability. However, systematic evaluation of inter-provider consistency is limited.
Objectives
To assess the analytical accuracy and reproducibility of clozapine and norclozapine blood concentration measurements across four UK laboratories providing routine therapeutic drug monitoring.
Design
Laboratory-based analytical accuracy and repeatability study.
Method
A panel of 15 spiked plasma samples covering a clinically relevant concentration range for clozapine and norclozapine (74–1443 ng/mL) was distributed to four UK laboratories providing routine therapeutic drug monitoring. Each laboratory analysed the same samples on two separate occasions. Analytical accuracy relative to known values and intra-laboratory consistency were assessed using regression and agreement analyses.
Results
All laboratories demonstrated strong linearity between measured and known concentrations (r = 0.932–0.999). Laboratories A and C showed slopes close to unity (0.94–0.99), minimal mean bias (–28 to +22 ng/mL), and narrow limits of agreement for clozapine (–89 to +51 ng/mL) and norclozapine (–71 to +71 ng/mL) across both rounds. In contrast, laboratories B and D showed reduced agreement on repeat testing, with increased slopes, positive bias, and widened limits of agreement in the second round. For laboratory B, the clozapine slope increased from 0.99 to 1.33, with limits of agreement widening to –320 to +673 ng/mL. Laboratory D showed greater deviation, with the clozapine slope increasing from 1.18 to 1.59 and mean bias rising to +349 ng/mL, with very wide limits of agreement (–354 to +1053 ng/mL).
Conclusion
Laboratory assays of clozapine using LC-MS and HPLC are not always accurate or reproducible over time. Some laboratories show a significant bias, reporting measured concentrations much higher than the true value.