Toward point-of-care diagnostic metabolic fingerprinting: quantification of plasma creatinine by infrared spectroscopy of microfluidic-preprocessed samples

• Published in: Analyst (London) Vol. 134; no. 6; pp. 1224 - 1231
• Main Authors: Shaw, R Anthony, Rigatto, Claudio, Reslerova, Martina, Ying, Sarah Low, Man, Angela, Schattka, Bernhard, Battrell, C Fred, Matthewson, John, Mansfield, Colin
• Format: Journal Article
• Language: English
• Published: England 01.06.2009
• Subjects: Analytic Sample Preparation Methods; Blood Chemical Analysis - instrumentation; Blood Chemical Analysis - methods; Creatinine - blood; Diffusion; Humans; Least-Squares Analysis; Metabolomics - instrumentation; Metabolomics - methods; Microfluidic Analytical Techniques; Point-of-Care Systems; Reproducibility of Results; Serum Albumin - metabolism; Spectrophotometry, Infrared; Systems Integration
• ISSN: 0003-2654; 1364-5528
• DOI: 10.1039/b821442e

Infrared (IR) spectroscopy has previously been established as a means to accurately quantify several serum and urine metabolites, based upon spectroscopy of dry films. The same technique has also provided the basis to develop certain diagnostic tests, developed in the 'metabolomics' spirit. Here, we report on the further development of an integrated microfluidic-IR technology and technique, customized with the aim of dramatically extending the capabilities of IR spectroscopy in both analytical and diagnostic (metabolomic) applications. By exploiting the laminar fluid diffusion interface (LFDI), serum specimens are processed to yield product streams that are better suited for metabolic fingerprinting; metabolites are captured within the aqueous product stream, while proteins (which otherwise dominate the spectra of films dried from serum) are present in much reduced concentration. Spectroscopy of films dried from the aqueous stream then provides enhanced diagnostic and analytical sensitivity. The manuscript introduces an LFDI card design that is customized for integration with IR spectroscopy, and details the development of a quantitative assay for serum creatinine--based upon LFDI-processed serum samples--that is substantially more accurate (standard error of calibration, SEC = 43 micromol/L) than the corresponding assay based upon unprocessed serum specimens (SEC = 138 micromol/L). Preliminary results of diffusion modeling are reported, and the prospects for further optimization of the technique, guided by accurate modeling, are discussed.