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

Different modeling approaches for inline biochemical monitoring over the VLP-making upstream stages using Raman spectroscopy

This work aimed to set inline Raman spectroscopy models to monitor biochemically (viable cell density, cell viability, glucose, lactate, glutamine, glutamate, and ammonium) all upstream stages of a virus-like particle-making process. Linear (Partial least squares, PLS; Principal components regression, PCR) and nonlinear (Artificial neural networks, ANN; supported vector machine, SVM) modeling approaches were assessed. The nonlinear models, ANN and SVM, were the more suitable models with the lowest absolute errors. The mean absolute error of the best models within the assessed parameter ranges for viable cell density (0.01-8.83 × 106 cells/mL), cell viability (1.3-100.0 %), glucose (5.22-10.93 g/L), lactate (18.6-152.7 mg/L), glutamine (158-1761 mg/L), glutamate (807.6-2159.7 mg/L), and ammonium (62.8-117.8 mg/L) were 1.55 ± 1.37 × 106 cells/mL (ANN), 5.01 ± 4.93 % (ANN), 0.27 ± 0.22 g/L (SVM), 4.7 ± 2.6 mg/L (SVM), 51 ± 49 mg/L (ANN), 57 ± 39 mg/L (SVM) and 2.0 ± 1.8 mg/L (ANN), respectively. The errors achieved, and best-fitted models were like those for the same bioprocess using offline data and others, which utilized inline spectra for mammalian cell lines as a host.

Raman spectroscopy as an alternative rapid microbial bioburden test method for continuous, automated detection of contamination in biopharmaceutical drug substance manufacturing

AIMS

To investigate an in-line Raman method capable of detecting accidental microbial contamination in pharmaceutical vessels, such as bioreactors producing monoclonal antibodies via cell culture.

METHODS AND RESULTS

The Raman method consists of a multivariate model built from Raman spectra collected in-line during reduced-scale bioreactor batches producing a monoclonal antibody, as well as a reduced-scale process with intentional spiking of representative compendial method microorganisms (n = 4). The orthogonal partial least squares regression discriminant analysis model (OPLS-DA) area under the curve (AUC), specificity and sensitivity were 0.96, 0.99, and 0.95, respectively. Furthermore, the model successfully detected contamination in an accidentally contaminated manufacturing-scale batch. In all cases, the time to detection (TTD) for Raman was superior compared to offline, traditional microbiological culturing.

CONCLUSIONS

The Raman OPLS-DA method met acceptance criteria for equivalent decision making to be considered a viable alternative to the compendial method for in-process bioburden testing. The in-line method is automated, non-destructive, and provides a continuous assessment of bioburden compared to an offline compendial method, which is manual, results in loss of product, and in practice is only collected once daily and requires 3-5 days for enumeration.

Single-Use, Metabolite Absorbing, Resonant Transducer (SMART) Culture Vessels for Label-Free, Continuous Cell Culture Progression Monitoring

Secreted metabolites are an important class of bio-process analytical technology (PAT) targets that can correlate to cell conditions. However, current strategies for measuring metabolites are limited to discrete measurements, resulting in limited understanding and ability for feedback control strategies. Herein, a continuous metabolite monitoring strategy is demonstrated using a single-use metabolite absorbing resonant transducer (SMART) to correlate with cell growth. Polyacrylate is shown to absorb secreted metabolites from living cells containing hydroxyl and alkenyl groups such as terpenoids, that act as a plasticizer. Upon softening, the polyacrylate irreversibly conformed into engineered voids above a resonant sensor, changing the local permittivity which is interrogated, contact-free, with a vector network analyzer. Compared to sensing using the intrinsic permittivity of cells, the SMART approach yields a 20-fold improvement in sensitivity. Tracking growth of many cell types such as Chinese hamster ovary, HEK293, K562, HeLa, and E. coli cells as well as perturbations in cell proliferation during drug screening assays are demonstrated. The sensor is benchmarked to show continuous measurement over six days, ability to track different growth conditions, selectivity to transducing active cell growth metabolites against other components found in the media, and feasibility to scale out for high throughput campaigns.

Integration of just-in-time learning with variational autoencoder for cell culture process monitoring based on Raman spectroscopy

Protein production in the biopharmaceutical industry necessitates the utilization of multiple analytical techniques and control methodologies to ensure both safety and consistency. To facilitate real-time monitoring and control of cell culture processes, Raman spectroscopy has emerged as a versatile analytical technology. This technique, categorized as a Process Analytical Technology, employs chemometric models to establish correlations between Raman signals and key variables of interest. One notable approach for achieving real-time monitoring is through the application of just-in-time learning (JITL), an industrial soft sensor modeling technique that utilizes Raman signals to estimate process variables promptly. The conventional Raman-based JITL method relies on the K-nearest neighbor (KNN) algorithm with Euclidean distance as the similarity measure. However, it falls short of addressing the impact of data uncertainties. To rectify this limitation, this study endeavors to integrate JITL with a variational autoencoder (VAE). This integration aims to extract dominant Raman features in a nonlinear fashion, which are expressed as multivariate Gaussian distributions. Three experimental runs using different cell lines were chosen to compare the performance of the proposed algorithm with commonly utilized methods in the literature. The findings indicate that the VAE-JITL approach consistently outperforms partial least squares, convolutional neural network, and JITL with KNN similarity measure in accurately predicting key process variables.

Development of generic metabolic Raman calibration models using solution titration in aqueous phase and data augmentation for in-line cell culture analysis

This study presents a novel approach for developing generic metabolic Raman calibration models for in-line cell culture analysis using glucose and lactate stock solution titration in an aqueous phase and data augmentation techniques. First, a successful set-up of the titration method was achieved by adding glucose or lactate solution at several different constant rates into the aqueous phase of a bench-top bioreactor. Subsequently, the in-line glucose and lactate concentration were calculated and interpolated based on the rate of glucose and lactate addition, enabling data augmentation and enhancing the robustness of the metabolic calibration model. Nine different combinations of spectra pretreatment, wavenumber range selection, and number of latent variables were evaluated and optimized using aqueous titration data as training set and a historical cell culture data set as validation and prediction set. Finally, Raman spectroscopy data collected from 11 historical cell culture batches (spanning four culture modes and scales ranging from 3 to 200 L) were utilized to predict the corresponding glucose and lactate values. The results demonstrated a high prediction accuracy, with an average root mean square errors of prediction of 0.65 g/L for glucose, and 0.48 g/L for lactate. This innovative method establishes a generic metabolic calibration model, and its applicability can be extended to other metabolites, reducing the cost of deploying real-time cell culture monitoring using Raman spectroscopy in bioprocesses.

Enhancing real-time cell culture monitoring: Automated Raman model optimization with Taguchi method

Raman spectroscopy has found widespread usage in monitoring cell culture processes both in research and practical applications. However, commonly, preprocessing methods, spectral regions, and modeling parameters have been chosen based on experience or trial-and-error strategies. These choices can significantly impact the performance of the models. There is an urgent need for a simple, effective, and automated approach to determine a suitable procedure for constructing accurate models. This paper introduces the adoption of a design of experiment (DoE) method to optimize partial least squares models for measuring the concentration of different components in cell culture bioreactors. The experimental implementation utilized the orthogonal test table L25(56). Within this framework, five factors were identified as control variables for the DoE method: the window width of Savitzky-Golay smoothing, the baseline correction method, the order of preprocessing steps, spectral regions, and the number of latent variables. The evaluation method for the model was considered as a factor subject to noise. The optimal combination of levels was determined through the signal-to-noise ratio response table employing Taguchi analysis. The effectiveness of this approach was validated through two cases, involving different cultivation scales, different Raman spectrometers, and different analytical components. The results consistently demonstrated that the proposed approach closely approximated the global optimum, regardless of data set size, predictive components, or the brand of Raman spectrometer. The performance of models recommended by the DoE strategy consistently surpassed those built using raw data, underscoring the reliability of models generated through this approach. When compared to exhaustive all-combination experiments, the DoE approach significantly reduces calculation times, making it highly practical for the implementation of Raman spectroscopy in bioprocess monitoring.

Multivariate data analysis of process parameters affecting the growth and productivity of stable Chinese hamster ovary cell pools expressing SARS-CoV-2 spike protein as vaccine antigen in early process development

The recent COVID-19 pandemic revealed an urgent need to develop robust cell culture platforms which can react rapidly to respond to this kind of global health issue. Chinese hamster ovary (CHO) stable pools can be a vital alternative to quickly provide gram amounts of recombinant proteins required for early-phase clinical assays. In this study, we analyze early process development data of recombinant trimeric spike protein Cumate-inducible manufacturing platform utilizing CHO stable pool as a preferred production host across three different stirred-tank bioreactor scales (0.75, 1, and 10 L). The impact of cell passage number as an indicator of cell age, methionine sulfoximine (MSX) concentration as a selection pressure, and cell seeding density was investigated using stable pools expressing three variants of concern. Multivariate data analysis with principal component analysis and batch-wise unfolding technique was applied to evaluate the effect of critical process parameters on production variability and a random forest (RF) model was developed to forecast protein production. In order to further improve process understanding, the RF model was analyzed with Shapley value dependency plots so as to determine what ranges of variables were most associated with increased protein production. Increasing longevity, controlling lactate build-up, and altering pH deadband are considered promising approaches to improve overall culture outcomes. The results also demonstrated that these pools are in general stable expressing similar level of spike proteins up to cell passage 11 (~31 cell generations). This enables to expand enough cells required to seed large volume of 200-2000 L bioreactor.

What should next-generation analytical platforms for biopharmaceutical production look like?

Biotherapeutic products, particularly complex products such as monoclonal antibodies (mAbs), have as many as 20-30 critical quality attributes (CQAs), thereby requiring a collection of orthogonal, high-resolution analytical tools for characterization and making characterization a resource-intensive task. As discussed in this Opinion, the need to reduce the cost of developing biotherapeutic products and the need to adopt Industry 4.0 and eventually Industry 5.0 paradigms are driving a reappraisal of existing analytical platforms. Next-generation platforms will have reduced offline testing, renewed focus on online testing and real-time monitoring, multiattribute monitoring, and extensive use of advanced data analytics and automation. They will be more complex, more sensitive, resource lean, and more responsive compared with existing platforms.

Pioneering Just-in-Time (JIT) Strategy for Accelerating Raman Method Development and Implementation for Biologic Continuous Manufacturing

Raman spectroscopy is a popular process analytical technology (PAT) tool that has been increasingly used to monitor and control the monoclonal antibody (mAb) manufacturing process. Although it allows the characterization of a variety of quality attributes by developing chemometric models, a large quantity of representative data is required, and hence, the model development process can be time-consuming. In recent years, the pharmaceutical industry has been expediting new drug development in order to achieve faster delivery of life-changing drugs to patients. The shortened development timelines have impacted the Raman application, as less time is allowed for data collection. To address this problem, an innovative Just-in-Time (JIT) strategy is proposed with the goal of reducing the time needed for Raman model development and ensuring its implementation. To demonstrate its capabilities, a proof-of-concept study was performed by applying the JIT strategy to a biologic continuous process for producing monoclonal antibody products. Raman spectroscopy and online two-dimensional liquid chromatography (2D-LC) were integrated as a PAT analyzer system. Raman models of antibody titer and aggregate percentage were calibrated by chemometric modeling in real-time. The models were also updated in real-time using new data collected during process monitoring. Initial Raman models with adequate performance were established using data collected from two lab-scale cell culture batches and subsequently updated using one scale-up batch. The JIT strategy is capable of accelerating Raman method development to monitor and guide the expedited biologics process development.

Development of an in-line Raman analytical method for commercial-scale CHO cell culture process monitoring: Influence of measurement channels and batch number on model performance

The mammalian cell culture process is a key step in commercial therapeutic protein production and needs to be monitored and controlled due to its complexity. Raman spectroscopy has been reported for cell culture process monitoring by analysis of many important parameters. However, studies on in-line Raman monitoring of the cell culture process were mainly conducted on small or pilot scale. Developing in-line Raman analytical methods for commercial-scale cell culture process monitoring is more challenging. In this study, an in-line Raman analytical method was developed for monitoring glucose, lactate, and viable cell density (VCD) in the Chinese hamster ovary (CHO) cell culture process during commercial production of biosimilar adalimumab (1500 L). The influence of different Raman measurement channels was considered to determine whether to merge data from different channels for model development. Raman calibration models were developed and optimized, with minimum root mean square error of prediction of 0.22 g L-1 for glucose in the range of 1.66-3.53 g L-1 , 0.08 g L-1 for lactate in the range of 0.15-1.19 g L-1 , 0.31 E6 cells mL-1 for VCD in the range of 0.96-5.68 E6 cells mL-1 on test sets. The developed analytical method can be used for cell culture process monitoring during manufacturing and meets the analytical purpose of this study. Further, the influence of the number of batches used for model calibration on model performance was also studied to determine how many batches are needed basically for method development. The proposed Raman analytical method development strategy and considerations will be useful for monitoring of similar bioprocesses.

Validation of the cell culture monitoring using a Raman spectroscopy calibration model developed with artificially mixed samples and investigation of model learning methods using initial batch data

The development of calibration models using Raman spectra data has long been challenged owing to the substantial time and cost required for robust data acquisition. To reduce the number of experiments involving actual incubation, a calibration model development method was investigated by measuring artificially mixed samples. In this method, calibration datasets were prepared using spectra from artificially mixed samples with adjusted concentrations based on design of experiments. The precision of these calibration models was validated using the actual cell culture sample. The results showed that when the culture conditions were unchanged, the root mean square error of prediction (RMSEP) of glucose, lactate, and antibody concentrations was 0.34, 0.33, and 0.25 g/L, respectively. Even when variables such as cell line or culture media were changed, the RMSEPs of glucose, lactate, and antibody concentrations remained within acceptable limits, demonstrating the robustness of the calibration models with artificially mixed samples. To further improve accuracy, a model training method for small datasets was also investigated. The spectral pretreatment conditions were optimized using error heat maps based on the first batch of each cell culture condition and applied these settings to the second and third batches. The RMSEPs improved for glucose, lactate, and antibody concentration, with values of 0.44, 0.19, and 0.18 g/L under constant culture conditions, 0.37, 0.12, and 0.12 g/L for different cell lines, and 0.26, 0.40, and 0.12 g/L when the culture media was changed. These results indicated the efficacy of calibration modeling with artificially mixed samples for actual incubations under various conditions.

Current and future technologies for monitoring cultured meat: A review

The high population growth rate, massive animal food consumption, fast economic progress, and limited food resources could lead to a food crisis in the future. There is a huge requirement for dietary proteins including cultured meat is being progressed to fulfill the need for meat-derived proteins in the diet. However, production of cultured meat requires monitoring numerous bioprocess parameters. This review presents a comprehensive overview of various widely adopted techniques (optical, spectroscopic, electrochemical, capacitive, FETs, resistive, microscopy, and ultrasound) for monitoring physical, chemical, and biological parameters that can improve the bioprocess control in cultured meat. The methods, operating principle, merits/demerits, and the main open challenges are reviewed with the aim to support the readers in advancing knowledge on novel sensing systems for cultured meat applications.

An assessment of the impact of Raman based glucose feedback control on CHO cell bioreactor process development

Process analytical technology (PAT) tools such as Raman Spectroscopy have become established tools for real time measurement of CHO cell bioreactor process variables and are aligned with the QbD approach to manufacturing. These tools can have a significant impact on process development if adopted early, creating an end-to-end PAT/QbD focused process. This study assessed the impact of Raman based feedback control on early and late phase development bioreactors by using a Raman based PLS model and PAT management system to control glucose in two CHO cell line bioreactor processes. The impact was then compared to bioreactor processes which used manual bolus fed methods for glucose feed delivery. Process improvements were observed in terms of overall bioreactor health, product output and product quality. Raman controlled batches for Cell Line 1 showed a reduction in glycation of 43.4% and 57.9%, respectively. Cell Line 2 batches with Raman based feedback control showed an improved growth profile with higher VCD and viability and a resulting 25% increase in overall product titer with an improved glycation profile. The results presented here demonstrate that Raman spectroscopy can be used in both early and late-stage process development and design for consistent and controlled glucose feed delivery.

Development of Raman Calibration Model Without Culture Data for In-Line Analysis of Metabolites in Cell Culture Media

In this study, we developed a method to build Raman calibration models without culture data for cell culture monitoring. First, Raman spectra were collected and then analyzed for the signals of all the mentioned analytes: glucose, lactate, glutamine, glutamate, ammonia, antibody, viable cells, media, and feed agent. Using these spectral data, the specific peak positions and intensities for each factor were detected. Next, according to the design of the experiment method, samples were prepared by mixing the above-mentioned factors. Raman spectra of these samples were collected and were used to build calibration models. Several combinations of spectral pretreatments and wavenumber regions were compared to optimize the calibration model for cell culture monitoring without culture data. The accuracy of the developed calibration model was evaluated by performing actual cell culture and fitting the in-line measured spectra to the developed calibration model. As a result, the calibration model achieved sufficiently good accuracy for the three components, glucose, lactate, and antibody (root mean square errors of prediction, or RMSEP = 0.23, 0.29, and 0.20 g/L, respectively). This study has presented innovative results in developing a culture monitoring method without using culture data, while using a basic conventional method of investigating the Raman spectra of each component in the culture media and then utilizing a design of experiment approach.

Fast-Training Deep Learning Algorithm for Multiplex Quantification of Mammalian Bioproduction Metabolites via Contactless Short-Wave Infrared Hyperspectral Sensing

Within the biopharmaceutical sector, there exists the need for a contactless multiplex sensor, which can accurately detect metabolite levels in real time for precise feedback control of a bioreactor environment. Reported spectral sensors in the literature only work when fully submerged in the bioreactor and are subject to probe fouling due to a cell debris buildup. The use of a short-wave infrared (SWIR) hyperspectral (HS) cam era allows for efficient, fully contactless collection of large spectral datasets for metabolite quantification. Here, we report the development of an interpretable deep learning system, a convolution metabolite regression (CMR) approach that detects glucose and lactate concentrations using label-free contactless HS images of cell-free spent media samples from Chinese hamster ovary (CHO) cell growth flasks. Using a dataset of <500 HS images, these CMR algorithms achieved a competitive test root-mean-square error (RMSE) performance of glucose quantification within 27 mg/dL and lactate quantification within 20 mg/dL. Conventional Raman spectroscopy probes report a validation performance of 26 and 18 mg/dL for glucose and lactate, respectively. The CMR system trains within 10 epochs and uses a convolution encoder with a sparse bottleneck regression layer to pick the best-performing filters learned by CMR. Each of these filters is combined with existing interpretable models to produce a metabolite sensing system that automatically removes spurious predictions. Collectively, this work will advance the safe and efficient adoption of contactless deep learning sensing systems for fine control of a variety of bioreactor environments.

Advances in bioreactor control for production of biotherapeutic products

Advanced control strategies are well established in chemical, pharmaceutical, and food processing industries. Over the past decade, the application of these strategies is being explored for control of bioreactors for manufacturing of biotherapeutics. Most of the industrial bioreactor control strategies apply classical control techniques, with the control system designed for the facility at hand. However, with the recent progress in sensors, machinery, and industrial internet of things, and advancements in deeper understanding of the biological processes, coupled with the requirement of flexible production, the need to develop a robust and advanced process control system that can ease process intensification has emerged. This has further fuelled the development of advanced monitoring approaches, modeling techniques, process analytical technologies, and soft sensors. It is seen that proper application of these concepts can significantly improve bioreactor process performance, productivity, and reproducibility. This review is on the recent advancements in bioreactor control and its related aspects along with the associated challenges. This study also offers an insight into the future prospects for development of control strategies that can be designed for industrial-scale production of biotherapeutic products.

Open-sourced Raman spectroscopy data processing package implementing a baseline removal algorithm validated from multiple datasets acquired in human tissue and biofluids

SIGNIFICANCE

Standardized data processing approaches are required in the field of bio-Raman spectroscopy to ensure information associated with spectral data acquired by different research groups, and with different systems, can be compared on an equal footing.

AIM

An open-sourced data processing software package was developed, implementing algorithms associated with all steps required to isolate the inelastic scattering component from signals acquired using Raman spectroscopy devices. The package includes a novel morphological baseline removal technique (BubbleFill) that provides increased adaptability to complex baseline shapes compared to current gold standard techniques. Also incorporated in the package is a versatile tool simulating spectroscopic data with varying levels of Raman signal-to-background ratios, baselines with different morphologies, and varying levels of stochastic noise.

RESULTS

Application of the BubbleFill technique to simulated data demonstrated superior baseline removal performance compared to standard algorithms, including iModPoly and MorphBR. The data processing workflow of the open-sourced package was validated in four independent in-human datasets, demonstrating it leads to inter-systems data compatibility.

CONCLUSIONS

A new open-sourced spectroscopic data pre-processing package was validated on simulated and real-world in-human data and is now available to researchers and clinicians for the development of new clinical applications using Raman spectroscopy.

Advancing Raman model calibration for perfusion bioprocesses using spiked harvest libraries

BACKGROUND

Raman spectroscopy has gained popularity to monitor multiple process indicators simultaneously in biopharmaceutical processes. However, robust and specific model calibration remains a challenge due to insufficient analyte variability to train the models and high cross-correlation of various media components and artifacts throughout the process.

MAIN METHODS

A systematic Raman calibration workflow for perfusion processes enabling highly specific and fast model calibration was developed. Harvest libraries consisting of frozen harvest samples from multiple CHO cell culture bioreactors collected at different process times were established. Model calibration was subsequently performed in an offline setup using a flow cell by spiking process harvest with glucose, raffinose, galactose, mannose, and fructose.

MAJOR RESULTS

In a screening phase, Raman spectroscopy was proven capable not only to distinguish sugars with similar chemical structures in perfusion harvest but also to quantify them independently in process-relevant concentrations. In a second phase, a robust and highly specific calibration model for simultaneous glucose (RMSEP = 0.32 g/L) and raffinose (RMSEP = 0.17 g/L) real-time monitoring was generated and verified in a third phase during a perfusion process.

IMPLICATION

The proposed novel offline calibration workflow allowed proper Raman peak decoupling, reduced calibration time from months down to days, and can be applied to other analytes of interest including lactate, ammonia, amino acids, or product titer. This article is protected by copyright. All rights reserved.

Raman based chemometric model development for glycation and glycosylation real time monitoring in a manufacturing scale CHO cell bioreactor process

The Quality by Design (QbD) approach to the production of therapeutic monoclonal antibodies (mAbs) emphasizes an understanding of the production process ensuring product quality is maintained throughout. Current methods for measuring critical quality attributes (CQAs) such as glycation and glycosylation are time and resource intensive, often, only tested offline once per batch process. Process analytical technology (PAT) tools such as Raman spectroscopy combined with chemometric modeling can provide real time measurements process variables and are aligned with the QbD approach. This study utilizes these tools to build partial least squares (PLS) regression models to provide real time monitoring of glycation and glycosylation profiles. In total, seven cell line specific chemometric PLS models; % mono-glycated, % non-glycated, % G0F-GlcNac, % G0, % G0F, % G1F, and % G2F were considered. PLS models were initially developed using small scale data to verify the capability of Raman to measure these CQAs effectively. Accurate PLS model predictions were observed at small scale (5 L). At manufacturing scale (2000 L) some glycosylation models showed higher error, indicating that scale may be a key consideration in glycosylation profile PLS model development. Model robustness was then considered by supplementing models with a single batch of manufacturing scale data. This data addition had a significant impact on the predictive capability of each model, with an improvement of 77.5% in the case of the G2F. The finalized models show the capability of Raman as a PAT tool to deliver real time monitoring of glycation and glycosylation profiles at manufacturing scale.

The role of Raman spectroscopy in biopharmaceuticals from development to manufacturing

Biopharmaceuticals have revolutionized the field of medicine in the types of active ingredient molecules and treatable indications. Adoption of Quality by Design and Process Analytical Technology (PAT) frameworks has helped the biopharmaceutical field to realize consistent product quality, process intensification, and real-time control. As part of the PAT strategy, Raman spectroscopy offers many benefits and is used successfully in bioprocessing from single-cell analysis to cGMP process control. Since first introduced in 2011 for industrial bioprocessing applications, Raman has become a first-choice PAT for monitoring and controlling upstream bioprocesses because it facilitates advanced process control and enables consistent process quality. This paper will discuss new frontiers in extending these successes in upstream from scale-down to commercial manufacturing. New reports concerning the use of Raman spectroscopy in the basic science of single cells and downstream process monitoring illustrate industrial recognition of Raman’s value throughout a biopharmaceutical product’s lifecycle. Finally, we draw upon a nearly 90-year history in biological Raman spectroscopy to provide the basis for laboratory and in-line measurements of protein quality, including higher-order structure and composition modifications, to support formulation development.

About Model Validation in Bioprocessing

In bioprocess engineering the Qualtiy by Design (QbD) initiative encourages the use of models to define design spaces. However, clear guidelines on how models for QbD are validated are still missing. In this review we provide a comprehensive overview of the validation methods, mathematical approaches, and metrics currently applied in bioprocess modeling. The methods cover analytics for data used for modeling, model training and selection, measures for predictiveness, and model uncertainties. We point out the general issues in model validation and calibration for different types of models and put this into the context of existing health authority recommendations. This review provides a starting point for developing a guide for model validation approaches. There is no one-fits-all approach, but this review should help to identify the best fitting validation method, or combination of methods, for the specific task and the type of bioprocess model that is being developed.

Prospective inference of bioprocess cell viability through chemometric modeling of fluorescence multiway data

In this investigation, the fermentation step of a standard mammalian cell-based industrial bioprocess for the production of a therapeutic protein was studied, with particular emphasis on the evolution of cell viability. This parameter constitutes one of the critical variables for bioprocess monitoring since it can affect downstream operations and the quality of the final product. In addition, when the cells experiment an unpredictable drop in viability, the assessment of this variable through classic off-line methods may not provide information sufficiently in advance to take corrective actions. In this context, Process Analytical Technology (PAT) framework aims to develop novel strategies for more efficient monitoring of critical variables, in order to improve the bioprocess performance. Thus, in this work, a set of chemometric tools were integrated to establish a PAT strategy to monitor cell viability, based on fluorescence multiway data obtained from fermentation samples of a particular bioprocess, in two different scales of operation. The spectral information, together with data regarding process variables, was integrated through chemometric exploratory tools to characterize the bioprocess and stablish novel criteria for the monitoring of cell viability. These findings motivated the development of a multivariate classification model, aiming to obtain predictive tools for the monitoring of future lots of the same bioprocess. The model could be satisfactorily fitted, showing the non-error rate of prediction of 100%.

Flow virometry for process monitoring of live virus vaccines-lessons learned from ERVEBO

Direct at line monitoring of live virus particles in commercial manufacturing of vaccines is challenging due to their small size. Detection of malformed or damaged virions with reduced potency is rate-limited by release potency assays with long turnaround times. Thus, preempting batch failures caused by out of specification potency results is almost impossible. Much needed are in-process tools that can monitor and detect compromised viral particles in live-virus vaccines (LVVs) manufacturing based on changes in their biophysical properties to provide timely measures to rectify process stresses leading to such damage. Using ERVEBO, MSD's Ebola virus vaccine as an example, here we describe a flow virometry assay that can quickly detect damaged virus particles and provide mechanistic insight into process parameters contributing to the damage. Furthermore, we describe a 24-h high throughput infectivity assay that can be used to correlate damaged particles directly to loss in viral infectivity (potency) in-process. Collectively, we provide a set of innovative tools to enable rapid process development, process monitoring, and control strategy implementation in large scale LVV manufacturing.

Monitoring E. coli Cell Integrity by ATR-FTIR Spectroscopy and Chemometrics: Opportunities and Caveats

During recombinant protein production with E. coli, the integrity of the inner and outer membrane changes, which leads to product leakage (loss of outer membrane integrity) or lysis (loss of inner membrane integrity). Motivated by current Quality by Design guidelines, there is a need for monitoring tools to determine leakiness and lysis in real-time. In this work, we assessed a novel approach to monitoring E. coli cell integrity by attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy. Various preprocessing strategies were tested in combination with regression (partial least squares, random forest) or classification models (partial least squares discriminant analysis, linear discriminant analysis, random forest, artificial neural network). Models were validated using standard procedures, and well-performing methods were additionally scrutinized by removing putatively important features and assessing the decrease in performance. Whereas the prediction of target compound concentration via regression was unsuccessful, possibly due to a lack of samples and low sensitivity, random forest classifiers achieved prediction accuracies of over 90% within the datasets tested in this study. However, strong correlations with untargeted spectral regions were revealed by feature selection, thereby demonstrating the need to rigorously validate chemometric models for bioprocesses, including the evaluation of feature importance.

A Chemometric Tool to Monitor and Predict Cell Viability in Filamentous Fungi Bioprocesses Using UV Chromatogram Fingerprints

Monitoring process variables in bioprocesses with complex expression systems, such as filamentous fungi, requires a vast number of offline methods or sophisticated inline sensors. In this respect, cell viability is a crucial process variable determining the overall process performance. Thus, fast and precise tools for identification of key process deviations or transitions are needed. However, such reliable monitoring tools are still scarce to date or require sophisticated equipment. In this study, we used the commonly available size exclusion chromatography (SEC) HPLC technique to capture impurity release information in Penicillium chrysogenum bioprocesses. We exploited the impurity release information contained in UV chromatograms as fingerprints for development of principal component analysis (PCA) models to descriptively analyze the process trends. Prediction models using well established approaches, such as partial least squares (PLS), orthogonal PLS (OPLS) and principal component regression (PCR), were made to predict the viability with model accuracies of 90% or higher. Furthermore, we demonstrated the platform applicability of our method by monitoring viability in a Trichoderma reesei process for cellulase production. We are convinced that this method will not only facilitate monitoring viability of complex bioprocesses but could also be used for enhanced process control with hybrid models in the future.