Investigation of near-infrared spectroscopy for periodic determination of glucose in cell culture media in situ

Citation
Cb. Lewis et al., Investigation of near-infrared spectroscopy for periodic determination of glucose in cell culture media in situ, APPL SPECTR, 54(10), 2000, pp. 1453-1457
Citations number
12
Categorie Soggetti
Spectroscopy /Instrumentation/Analytical Sciences
Journal title
APPLIED SPECTROSCOPY
ISSN journal
00037028 → ACNP
Volume
54
Issue
10
Year of publication
2000
Pages
1453 - 1457
Database
ISI
SICI code
0003-7028(200010)54:10<1453:IONSFP>2.0.ZU;2-H
Abstract
Current techniques used to monitor glucose concentration of cell culture me dia (CCM) require invasive and tedious handling of the sample for sterile m edia removal and nutrient replacement. In order to optimize cell culture gr owth in bioreactors, biosensors must be developed that are capable of monit oring the cell culture processes noninvasively and continuously, In this st udy, on-line, noninvasive determinations of glucose in cell culture media w ere investigated via near-infrared spectroscopy (NIR) across the 2.0-2.5 mu m combination region. A system was developed, using a unique fiberoptic cou pling method and a commercial Fourier transform infrared (FT-LR) spectromet er, to characterize glucose single-beam spectra collected from cell culture media. This novel system is the first of its kind and integrates a complet ely noninvasive optical probe to measure glucose concentrations within cell culture media, in situ. Spectra recorded from a four-day fibroblast cultur e with this fiber coupled system and an FT-IR spectrometer have been analyz ed and compared with standard clinical chemistry techniques. Partial least- squares (PLS) regression has been used to extract the analyte-dependent inf ormation and to build a successful multivariate calibration model. A combin ation of spectra from cell culture media and prepared media mixtures was us ed to eliminate unwanted correlations in the calibration data, The combined use of this unique fiberoptic system, PLS, and uncorrelated spectra result ed in a true glucose prediction error of 14.8 mg/dL in an independent valid ation set.