Background: Monitoring of plasma drug levels is mandatory in patients receiving high-dose methotrexate. This study evaluated the analytical performance of the novel Architect and the established ARK (TM) methotrexate immunoassay (running on the Roche Cobas (c) c502 analyzer) in comparison with liquid chromatography-mass spectrometry (LC-MS) and the TDx/TDxFLx Methotrexate II assay. Methods: Imprecision and linearity were verified for the Architect and ARK assay according to CLSI EP15-A3 and EP6-A guidelines, respectively. The reported limit of -quantitation (0.04 mu mol/L) was tested for both assays according to the CLSI EP17-A2 guideline. Correlation and agreement between the different assays were evaluated using residual plasma samples (n = 153). Results: Total imprecision was <6.3% and <9.5% for the Architect and ARK assay, respectively. The claimed linearity and limit of quantitation were confirmed for the Architect assay. For the ARK assay, imprecision at the limit of quantitation was <18% with a positive bias resulting in a high total error up to 58%, and hence the linearity could not be confirmed. Both assays showed strong correlations with the TDX assay and LC-MS but a positive bias of 12.2% and 20.5% in comparison to LC-MS for the Architect and ARK assay, respectively. For the ARK assay this bias increased dramatically for samples with concentrations towards the limit of quantitation. Conclusions: The Architect assay is suitable for monitoring plasma methotrexate, but the ARK assay showed unsatisfactory performance in the analysis of low concentrated samples. Unlike the TDX assay, both assays require manual dilution of samples at higher concentrations, which delays sample processing in clinical routine.
