Micropatterned Adhesion Sites for Spheroid Cultivation Under Flow

Abstract

The biopharmaceutical industry faces challenges relating to batch quality control and performance of cell growth media for bioreactors. There is an express need for rapid, sensitive analyses and a non-destructive optical technique to prevent costly losses in productivity. Major issues pertain to cell media preparation, including component stoichiometry, as well as time-, light-, oxidation-and temperature-dependent storage issues. This study reports on cell media analyses using patented A-TEEM spectroscopy, which provide rapid (s-min) analyses of all chromophoric and fluorescent compounds in the UV-VIS range with micro-to sub-microgram/L sensitivity to many aromatic compounds. A-TEEM analysis primarily based on the fluorescence EEMs that are corrected for inner-filter effects (IFE) to eliminate or reduce concentration-dependent spectral distortion. Importantly IFE correction facilitates resolution of low-concentration compounds from components at higher concentrations or intensities in the sample. Additionally, A-TEEM analysis can use 'multi-block' data organization to leverage the resolution capacity of the simultaneously acquired, independent absorbance and EEM data sets. The above factors are important because cell media is complex with many overlapping spectral components with widely varying extinction and or fluorescence properties. A-TEEM data analysis in this study focuses on a machine-learning technique known as Extreme Gradient Boosting (XGB) for purposes of both discrimination (XGBDA) and regression (XGBR). We compare results of XGBDA and XGBDR with controlled data sets to other comparable methods including Partial-Least Squares and Support Vector Machine. Key results include comparisons of cell media from different vendors as well time-and light-dependent effects on media storage. We show that XGB analyses of the A-TEEM data can lead to more effective validation for both discrimination and regression models to facilitate cell media characterization important to quality control including the effects of photo-oxidation.