UC San Diego

BACKGROUND:Clinical LC-MS/MS assays traditionally require that samples be run in batches with calibration curves in each batch. This approach is inefficient and presents a barrier to random access analysis. We developed an alternative approach called multipoint internal calibration (MPIC) that eliminated the need for batch-mode analysis. METHODS:The new approach used 4 variants of 13C-labeled methotrexate (0.026-10.3 µM) as an internal calibration curve within each sample. One site carried out a comprehensive validation, which included an evaluation of interferences and matrix effects, lower limit of quantification (LLOQ), and 20-day precision. Three sites evaluated assay precision and linearity. MPIC was also compared with traditional LC-MS/MS and an immunoassay. RESULTS:Recovery of spiked analyte was 93%-102%. The LLOQ was validated to be 0.017 µM. Total variability, determined in a 20-day experiment, was 11.5%CV. In a 5-day variability study performed at each site, total imprecision was 3.4 to 16.8%CV. Linearity was validated throughout the calibrator range (r2 > 0.995, slopes = 0.996-1.01). In comparing 40 samples run in each laboratory, the median interlaboratory imprecision was 6.55%CV. MPIC quantification was comparable to both traditional LC-MS/MS and immunoassay (r2 = 0.96-0.98, slopes = 1.04-1.06). Bland-Altman analysis of all comparisons showed biases rarely exceeding 20% when MTX concentrations were >0.4 µM. CONCLUSION:The MPIC method for serum methotrexate quantification was validated in a multisite proof-of-concept study and represents a big step toward random-access LC-MS/MS analysis, which could change the paradigm of mass spectrometry in the clinical laboratory.