High-throughput serum 25-hydroxy vitamin D testing with automated sample preparation

Abstract

Serum from bar-coded tubes, and then internal standard, are pipetted to 96-well plates with an 8-channel automated liquid handler (ALH). The first precipitation reagent (methanol:ZnSO4) is added and mixed with the 8-channel ALH. A second protein precipitating agent, 1% formic acid in acetonitrile, is added and mixed with a 96-channel ALH. After a 4-min delay for larger precipitates to settle to the bottom of the plate, the upper 36% of the precipitate/supernatant mix is transferred with the 96-channel ALH to a Sigma Hybrid SPE® plate and vacuumed through for removal of phospholipids and precipitated proteins. The filtrate is collected in a second 96-well plate (collection plate) which is foil-sealed, placed in the autosampler (ALS), and injected into a multiplexed LC-MS/MS system running AB Sciex Cliquid® and MPX® software. Two Shimadzu LC stacks, with multiplex timing controlled by MPX® software, inject alternately to one AB Sciex API-5000 MS/MS using positive atmospheric pressure chemical ionization (APCI) and a 1.87 min water/acetonitrile LC gradient with a 2.1 × 20 mm, 2.7μm, C18 fused core particle column (Sigma Ascentis Express). LC-MS/MS through put is ~44 samples/h/LC-MS/MS system with dual-LC channel multiplexing. Plate maps are transferred electronically from the ALH and reformatted into LC-MS/MS sample table format using the Data Innovations LLC (DI) Instrument Manager middleware application. Before collection plates are loaded into the ALS, the plate bar code is manually scanned to download the sample table from the DI middleware to the LC-MS/MS. After acquisition—LC-MS/MS data is analyzed with AB Sciex Multiquant® software using customized queries, and then results are transferred electronically via a DI interface to the LIS. 2500 samples/day can be extracted by two analysts using four ALHs in 4–6 h. LC-MS/MS analysis of those samples on three dual-channel LC multiplexed LC-MS/MS systems requires 19–21 h and data analysis can be done by two analysts in 4–6 h.