Cole–Cole, linear and multivariate modeling of capacitance data for on-line monitoring of biomass

This work evaluates three techniques of calibrating capacitance (dielectric) spectrometers used for on-line monitoring of biomass: modeling of cell properties using the theoretical Cole–Cole equation, linear regression of dual-frequency capacitance measurements on biomass concentration, and multivariate (PLS) modeling of scanning dielectric spectra. The performance and robustness of each technique is assessed during a sequence of validation batches in two experimental settings of differing signal noise. In more noisy conditions, the Cole–Cole model had significantly higher biomass concentration prediction errors than the linear and multivariate models. The PLS model was the most robust in handling signal noise. In less noisy conditions, the three models performed similarly. Estimates of the mean cell size were done additionally using the Cole–Cole and PLS models, the latter technique giving more satisfactory results.     

Quantitative modeling of viable cell density,cell size,intracellular conductivity,and membrane capacitance in batch and fed‐batch CHO processes using dielectric spectroscopy

Dielectric spectroscopy was used to analyze typical batch and fed‐batch CHO cell culture processes. Three methods of analysis (linear modeling, Cole–Cole modeling, and partial least squares regression), were used to correlate the spectroscopic data with routine biomass measurements [viable packed cell volume, viable cell concentration (VCC), cell size, and oxygen uptake rate (OUR)]. All three models predicted offline biomass measurements accurately during the growth phase of the cultures. However, during the stationary and decline phases of the cultures, the models decreased in accuracy to varying degrees. Offline cell radius measurements were unsuccessfully used to correct for the deviations from the linear model, indicating that physiological changes affecting permittivity were occurring. The β‐dispersion was analyzed using the Cole–Cole distribution parameters Δε (magnitude of the permittivity drop), f_c (critical frequency), and α (Cole–Cole parameter). Furthermore, the dielectric parameters static internal conductivity (σ_i) and membrane capacitance per area (C_m) were calculated for the cultures. Finally, the relationship between permittivity, OUR, and VCC was examined, demonstrating how the definition of viability is critical when analyzing biomass online. The results indicate that the common assumptions of constant size and dielectric properties used in dielectric analysis are not always valid during later phases of cell culture processes. The findings also demonstrate that dielectric spectroscopy, while not a substitute for VCC, is a complementary measurement of viable biomass, providing useful auxiliary information about the physiological state of a culture. © 2010 American Institute of Chemical Engineers Biotechnol. Prog., 2010     

Analysis of chemometric models applied to Raman spectroscopy for monitoring key metabolites of cell culture

The Food and Drug Administration (FDA) initiative of Process Analytical Technology (PAT) encourages the monitoring of biopharmaceutical manufacturing processes by innovative solutions. Raman spectroscopy and the chemometric modeling tool partial least squares (PLS) have been applied to this aim for monitoring cell culture process variables. This study compares the chemometric modeling methods of Support Vector Machine radial (SVMr), Random Forests (RF), and Cubist to the commonly used linear PLS model for predicting cell culture components—glucose, lactate, and ammonia. This research is performed to assess whether the use of PLS as standard practice is justified for chemometric modeling of Raman spectroscopy and cell culture data. Model development data from five small-scale bioreactors (2 × 1 L and 3 × 5 L) using two Chinese hamster ovary (CHO) cell lines were used to predict against a manufacturing scale bioreactor (2,000 L). Analysis demonstrated that Cubist predictive models were better for average performance over PLS, SVMr, and RF for glucose, lactate, and ammonia. The root mean square error of prediction (RMSEP) of Cubist modeling was acceptable for the process concentration ranges of glucose (1.437 mM), lactate (2.0 mM), and ammonia (0.819 mM). Interpretation of variable importance (VI) results theorizes the potential advantages of Cubist modeling in avoiding interference of Raman spectral peaks. Predictors/Raman wavenumbers (cm⁻¹) of interest for individual variables are X1139–X1141 for glucose, X846–X849 for lactate, and X2941–X2943 for ammonia. These results demonstrate that other beneficial chemometric models are available for use in monitoring cell culture with Raman spectroscopy.     

On-line multi-analyzer monitoring of biomass, glucose and acetate for growth rate control of a Vibrio cholerae fed-batch cultivation

In situ near-infrared (NIR) spectroscopy and in-line electronic nose (EN) mapping were used to monitor and control a cholera-toxin producing Vibrio cholerae fed-batch cultivation carried out with a laboratory method as well as with a production method. Prediction models for biomass, glucose and acetate using NIR spectroscopy were developed based on spectral identification and partial-least squares (PLS) regression resulting in high correlation to reference data (standard errors of prediction for biomass, glucose and acetate were 0.20 gl(-1), 0.26 gl(-1) and 0.28 gl(-1)). A compensation algorithm for aerated bioreactor disturbances was integrated in the model computation, which in particular improved the prediction by the biomass model. First, the NIR data were applied together with EN in-line data selected by principal component analysis (PCA) for generating a trajectory representation of the fed-batch cultivation. A correlation between the culture progression and EN signals was demonstrated, which proved to be beneficial in monitoring the culture quality. It was shown that a deviation from a normal cultivation behavior could easily be recognized and that the trajectory was able to alarm a bacterial contamination. Second, the NIR data indicated the potential of predicting the concentration of formed cholera toxin with a model prediction error of 0.020 gl(-1). Third, the on-line biomass prediction based on the NIR model was used to control the overflow metabolism acetate formation of the V. cholerae culture. The controller compared actual specific growth rate as estimated from the prediction with the critical acetate formation growth rate, and from that difference adjusted the glucose feed rate.     

Online- and offline- monitoring of stem cell expansion on microcarrier

In the biopharmaceutical industry, adherent growing stem cell cultures gain worldwide importance as cell products. The cultivation process of these cells, such as in stirred tank reactors or in fixed bed reactors, is highly sophisticated. Cultivations need to be monitored and controlled to guarantee product quality and to satisfy GMP requirements. With the process analytical technology (PAT) initiative, requirements regarding process monitoring and control have changed and real-time on-line monitoring tools are recommended. A tool meeting the new requirements may be the dielectric spectroscopy for online viable cell mass determination by measurement of the permittivity. To establish these tools, proper offline methods for data correlation are required. The cell number determination of adherent cells on microcarrier is difficult, as it requires cell detachment from the carrier, which highly increases the statistical error. As an offline method, a fluorescence assay based on SYBR^®GreenI was developed allowing fast and easy total cell concentration determination without the need to detach the cells from the carrier. The assay is suitable for glass carriers used in stirred tank reactor systems or in fixed bed systems, may be suitable for different cell lines and can be applied to high sample numbers easily. The linear dependency of permittivity to cell concentration of suspended stem cells with the dielectric spectroscopy is shown for even very small cell concentrations. With this offline-method, a correlation of the cell concentration grown on carrier to the permittivity data measured by the dielectric spectroscopy was done successfully.     

In situ Raman spectroscopy for simultaneous monitoring of multiple process parameters in mammalian cell culture bioreactors

In this study, the application of Raman spectroscopy to the simultaneous quantitative determination of glucose, glutamine, lactate, ammonia, glutamate, total cell density (TCD), and viable cell density (VCD) in a CHO fed‐batch process was demonstrated in situ in 3 L and 15 L bioreactors. Spectral preprocessing and partial least squares (PLS) regression were used to correlate spectral data with off‐line reference data. Separate PLS calibration models were developed for each analyte at the 3 L laboratory bioreactor scale before assessing its transferability to the same bioprocess conducted at the 15 L pilot scale. PLS calibration models were successfully developed for all analytes bar VCD and transferred to the 15 L scale. © 2012 American Institute of Chemical Engineers Biotechnol. Prog., 2012     

Enhanced discrimination and calibration of biomass NIR spectral data using non-linear kernel methods

Rapid methods for the characterization of biomass for energy purpose utilization are fundamental. In this work, near infrared spectroscopy is used to measure ash and char content of various types of biomass. Very strong models were developed, independently of the type of biomass, to predict ash and char content by near infrared spectroscopy and multivariate analysis. Several statistical approaches such as principal component analysis (PCA), orthogonal signal correction (OSC) treated PCA and partial least squares (PLS), Kernel PCA and PLS were tested in order to find the best method to deal with near infrared data to classify and predict these biomass characteristics. The model with the highest coefficient of correlation and the lowest RMSEP was obtained with OSC-treated Kernel PLS method.     

On-line biomass measurements in bioreactor cultivations: comparison study of two on-line probes

Two on-line probes for biomass measurement in bioreactor cultivations were evaluated. One probe is based on near infrared (NIR) light absorption and the other on dielectric spectroscopy. The probes were used to monitor biomass production in cultivations of several different microorganisms. Differences in NIR probe response compared to off-line measurement methods revealed that the most significant factor affecting the response was cell shape. The NIR light absorption method is more developed and reliable for on-line in situ biomass estimation than dielectric spectroscopy. The NIR light absorption method is, however, of no significant use, when the cultivation medium is not clear, and especially in processes using adsorbents or solid matrix for the microorganism to grow on. The possibilities offered by dielectric spectroscopy are impressive, but the on-line probe technology needs to be improved.     

Bioprocess control from a multivariate process trajectory

A multivariate bioprocess control approach, capable of tracking a pre-set process trajectory correlated to the biomass or product concentration in the bioprocess is described. The trajectory was either a latent variable derived from multivariate statistical process monitoring (MSPC) based on partial least squares (PLS) modeling, or the absolute value of the process variable. In the control algorithm the substrate feed pump rate was calculated from on-line analyzer data. The only parameters needed were the substrate feed concentration and the substrate yield of the growth-limiting substrate. On-line near-infrared spectroscopy data were used to demonstrate the performance of the control algorithm on an Escherichia coli fed-batch cultivation for tryptophan production. The controller showed good ability to track a defined biomass trajectory during varying process dynamics. The robustness of the control was high, despite significant external disturbances on the cultivation and control parameters.     

Rapid monitoring of recombinant antibody production by mammalian cell cultures using fourier transform infrared spectroscopy and chemometrics

Fourier transform infrared (FT‐IR) spectroscopy combined with multivariate statistical analyses was investigated as a physicochemical tool for monitoring secreted recombinant antibody production in cultures of Chinese hamster ovary (CHO) and murine myeloma non‐secreting 0 (NS0) cell lines. Medium samples were taken during culture of CHO and NS0 cells lines, which included both antibody‐producing and non‐producing cell lines, and analyzed by FT‐IR spectroscopy. Principal components analysis (PCA) alone, and combined with discriminant function analysis (PC‐DFA), were applied to normalized FT‐IR spectroscopy datasets and showed a linear trend with respect to recombinant protein production. Loadings plots of the most significant spectral components showed a decrease in the C–O stretch from polysaccharides and an increase in the amide I band during culture, respectively, indicating a decrease in sugar concentration and an increase in protein concentration in the medium. Partial least squares regression (PLSR) analysis was used to predict antibody titers, and these regression models were able to predict antibody titers accurately with low error when compared to ELISA data. PLSR was also able to predict glucose and lactate amounts in the medium samples accurately. This work demonstrates that FT‐IR spectroscopy has great potential as a tool for monitoring cell cultures for recombinant protein production and offers a starting point for the application of spectroscopic techniques for the on‐line measurement of antibody production in industrial scale bioreactors. Biotechnol. Bioeng. 2010; 106: 432–442. © 2010 Wiley Periodicals, Inc.     

A multivariate calibration procedure for UV/VIS spectrometric monitoring of BHK‐21 cell metabolism and growth

Monitoring mammalian cell culture with UV–vis spectroscopy has not been widely explored. The aim of this work was to calibrate Partial Least Squares (PLS) models from off‐line UV–vis spectral data in order to predict some nutrients and metabolites, as well as viable cell concentrations for mammalian cell bioprocess using phenol red in culture medium. The BHK‐21 cell line was used as a mammalian cell model. Spectra of samples taken from batches performed at different dissolved oxygen concentrations (10, 30, 50, and 70% air saturation), in two bioreactor configurations and with two strategies to control pH were used to calibrate and validate PLS models. Glutamine, glutamate, glucose, and lactate concentrations were suitably predicted by means of this strategy. Especially for glutamine and glucose concentrations, the prediction error averages were lower than 0.50 ± 0.10 mM and 2.21 ± 0.16 mM, respectively. These values are comparable with those previously reported using near infrared and Raman spectroscopy in conjunction with PLS. However, viable cell concentration models need to be improved. The present work allows for UV–vis at‐line sensor development, decrease cost related to nutrients and metabolite quantifications and establishment of fed‐batch feeding schemes. © 2013 American Institute of Chemical Engineers Biotechnol. Prog., 30:241–248, 2014     

On-line process monitoring and chemometric modeling with 2D fluorescence spectra obtained in recombinant E. coli fermentations

2D spectrofluorometry produces a large volume of spectral data during fermentation processes with recombinant E. coli, which can be analyzed using chemometric methods such as principal component analysis (PCA), principal component regression (PCR) and partial least square regression (PLS). An analysis of the spectral data by PCA results in scores and loadings that are not only visualized in the score-loading plots but are also used to monitor the fermentation processes on-line. The score plots provided useful qualitative information on four fermentation processes for the production of extracellular 5-aminolevulinic acid (ALA). Two chemometric models (PCR and PLS) were used to examine the correlation between the 2D fluorescence spectra and a few parameters of the fermentation processes. The results showed that PLS had slightly better calibration and prediction performance than PCR.     

Determination of cell mass and polymyxin using multi-wavelength fluorescence

Multi-wavelength fluorescence was applied for on-line monitoring of cell mass and the antibiotic polymyxin B in Bacillus polymyxa cultivations. By varying the phosphate and nitrogen content of the medium different polymyxin-cell mass ratios could be obtained. Using this strategy, it was possible to investigate if multi-wavelength fluorescence is able to give independent prediction of the two parameters. Partial least square (PLS) regression was applied to establish mathematical relationships between off-line determined cell mass and polymyxin concentrations and on-line collected fluorescence data. For polymyxin one universal PLS model, with a correlation of 0.95 and a root mean square error of cross validation (RMSECV) of 35 mgl(-1), could be constructed. However, correlation between fluorescence and cell mass dry weight could not be established including data from all three types of cultivations. For data from each type of cultivation, separate models with high correlation and low RMSECV values were built. A large variation in cellular composition as a result of the different levels of nitrogen and phosphorus in the cultivations was the probable reason to the necessity of building three models. The results of the present investigation indicate that production of polymyxin is concomitantly regulated by phosphate and nitrogen as the highest polymyxin yield on cell mass, 0.17+/-0.01 gg(-1), was reached in the cultivations where both nitrogen and phosphate concentrations were kept low.     

Cross‐scale predictive modeling of CHO cell culture growth and metabolites using Raman spectroscopy and multivariate analysis

Multi‐component, multi‐scale Raman spectroscopy modeling results from a monoclonal antibody producing CHO cell culture process including data from two development scales (3 L, 200 L) and a clinical manufacturing scale environment (2,000 L) are presented. Multivariate analysis principles are a critical component to partial least squares (PLS) modeling but can quickly turn into an overly iterative process, thus a simplified protocol is proposed for addressing necessary steps including spectral preprocessing, spectral region selection, and outlier removal to create models exclusively from cell culture process data without the inclusion of spectral data from chemically defined nutrient solutions or targeted component spiking studies. An array of single‐scale and combination‐scale modeling iterations were generated to evaluate technology capabilities and model scalability. Analysis of prediction errors across models suggests that glucose, lactate, and osmolality are well modeled. Model strength was confirmed via predictive validation and by examining performance similarity across single‐scale and combination‐scale models. Additionally, accurate predictive models were attained in most cases for viable cell density and total cell density; however, these components exhibited some scale‐dependencies that hindered model quality in cross‐scale predictions where only development data was used in calibration. Glutamate and ammonium models were also able to achieve accurate predictions in most cases. However, there are differences in the absolute concentration ranges of these components across the datasets of individual bioreactor scales. Thus, glutamate and ammonium PLS models were forced to extrapolate in cases where models were derived from small scale data only but used in cross‐scale applications predicting against manufacturing scale batches. © 2014 American Institute of Chemical Engineers Biotechnol. Prog., 31:566–577, 2015     

Comparison of viable cell concentration estimation methods for a mammalian cell cultivation process

Various mechanistic and black-box models were applied for on-line estimations of viable cell concentrations in fed-batch cultivation processes for CHO cells. Data from six fed-batch cultivation experiments were used to identify the underlying models and further six independent data sets were used to determine the performance of the estimators. The performances were quantified by means of the root mean square error (RMSE) between the estimates and the corresponding off-line measured validation data sets. It is shown that even simple techniques based on empirical and linear model approaches provide a fairly good on-line estimation performance. Best results with respect to the validation data sets were obtained with hybrid models, multivariate linear regression technique and support vector regression. Hybrid models provide additional important information about the specific cellular growth rates during the cultivation.     

Enhanced process understanding and multivariate prediction of the relationship between cell culture process and monoclonal antibody quality

This work investigates the insights and understanding which can be deduced from predictive process models for the product quality of a monoclonal antibody based on designed high‐throughput cell culture experiments performed at milliliter (ambr‐15^®) scale. The investigated process conditions include various media supplements as well as pH and temperature shifts applied during the process. First, principal component analysis (PCA) is used to show the strong correlation characteristics among the product quality attributes including aggregates, fragments, charge variants, and glycans. Then, partial least square regression (PLS1 and PLS2) is applied to predict the product quality variables based on process information (one by one or simultaneously). The comparison of those two modeling techniques shows that a single (PLS2) model is capable of revealing the interrelationship of the process characteristics to the large set product quality variables. In order to show the dynamic evolution of the process predictability separate models are defined at different time points showing that several product quality attributes are mainly driven by the media composition and, hence, can be decently predicted from early on in the process, while others are strongly affected by process parameter changes during the process. Finally, by coupling the PLS2 models with a genetic algorithm first the model performance can be further improved and, most importantly, the interpretation of the large‐dimensioned process–product‐interrelationship can be significantly simplified. The generally applicable toolset presented in this case study provides a solid basis for decision making and process optimization throughout process development. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:1368–1380, 2017     

Advances in on-line monitoring and control of mammalian cell cultures: Supporting the PAT initiative

In recent years, much attention has been directed towards the development of global methods for on-line process monitoring, especially since the Food and Drug Administration (FDA) launched the Process Analytical Technology (PAT) guidance, stimulating biopharmaceutical companies to update their monitoring tools to ensure a pre-defined final product quality. The ideal technologies for biopharmaceutical processes should operate in situ, be non-invasive and generate on-line information about multiple key bioprocess and/or metabolic variables. A wide range of spectroscopic techniques based on in situ probes have already been tested in mammalian cell cultures, such as near infrared (NIR), mid infrared (MIR), 2D fluorescence and dielectric capacitance spectroscopy; similarly, the electronic nose technique based on chemical array sensors has been tested for in situ off-gas analysis of mammalian cell cultures. All these methods provide series of spectra, from which meaningful information must be extracted. In this sense, data mining techniques such as principal components regression (PCR), partial least squares (PLS) or artificial neural networks (ANN) have been applied to handle the dense flow of data generated from the real-time process analyzers. Furthermore, the implementation of feedback control methods would help to improve process performance and ultimately ensure reproducibility. This review discusses the suitability of several spectroscopic techniques coupled with chemometric methods for improved monitoring and control of mammalian cell processes.     

Visualizing Solvent Mediated Phase Transformation Behavior of Carbamazepine Polymorphs by Principal Component Analysis

The purpose of this research is to gain a greater insight into the hydrate formation processes of different carbamazepine (CBZ) anhydrate forms in aqueous suspension, where principal component analysis (PCA) was applied for data analysis. The capability of PCA to visualize and to reveal simplified structures that often underlie large data sets are explored. Different CBZ polymorphs were dispersed separately in aqueous solution, and then recovered and measured by FT-Raman spectroscopy. PCA was employed for visualizing the dynamics of the phase transformation from each CBZ polymorph to the dihydrate (DH). As a comparison to PCA visualization, the transformation process of each CBZ polymorph was quantified using PLS modeling. The results demonstrated that PCA has advantages in presenting the original data in terms of the differences and similarities, and also directly identify the statistical patterns in the data even when the data set is large. These advantages provided greater insight into the measured Raman spectra as well as the phase transformation process of CBZ polymorphs to the DH in aqueous environment.     

Vibrational spectroscopy and chemometrics to characterize and quantitate trehalose crystallization

Investigating the phase behavior of sugars in ice and lyophilized solids is of significant interest in the pharmaceutical industry. In this study, Raman and near infrared (NIR) spectroscopy are used to characterize and quantitate trehalose crystallization using several chemometric models. The predictive behaviors of partial least squares (PLS), principal component analysis (PCA), and multiple linear regression (MLR) models are compared. In general, PLS and PCA outperform linear and MLR models. Changes in specific vibrational modes associated with several coupled motions are described and assigned as a function of crystal content. In addition to characterization and quantitation, our method may be used to localize gradients of amorphous and/or crystallized trehalose within a sample.     

A novel approach for using dielectric spectroscopy to predict viable cell volume (VCV) in early process development

Online monitoring of viable cell volume (VCV) is essential to the development, monitoring, and control of bioprocesses. The commercial availability of steam‐sterilizable dielectric‐spectroscopy probes has enabled successful adoption of this technology as a key noninvasive method to measure VCV for cell‐culture processes. Technological challenges still exist, however. For some cell lines, the technique's accuracy in predicting the VCV from probe‐permittivity measurements declines as the viability of the cell culture decreases. To investigate the cause of this decrease in accuracy, divergences in predicted vs. actual VCV measurements were directly related to the shape of dielectric frequency scans collected during a cell culture. The changes in the shape of the beta dispersion, which are associated with changes in cell state, are quantified by applying a novel “area ratio” (AR) metric to frequency‐scanning data from the dielectric‐spectroscopy probes. The AR metric is then used to relate the shape of the beta dispersion to single‐frequency permittivity measurements to accurately predict the offline VCV throughout an entire fed‐batch run, regardless of cell state. This work demonstrates the possible feasibility of quantifying the shape of the beta dispersion, determined from frequency‐scanning data, for enhanced measurement of VCV in mammalian cell cultures by applying a novel shape‐characterization technique. In addition, this work demonstrates the utility of using changes in the shape of the beta dispersion to quantify cell health. © 2013 American Institute of Chemical Engineers Biotechnol. Prog., 30:479–487, 2014