Multiple Product Quality Research Articles

Rapid Mass Spectrometry-Based Multiattribute Method for Glycation Analysis with Integrated Afucosylation Detection Capability.

The multiattribute method (MAM) has emerged as a powerful tool for simultaneously screening multiple product quality attributes of therapeutic antibodies. One such potential critical quality attribute (CQA) is glycation, a common modification that can impact the heterogeneity, functional activity, and immunogenicity of therapeutic antibodies. However, current methods for monitoring glycation levels in MAM are rare and not sufficiently rapid and accurate. In this study, an improved mass spectrometry (MS)-based MAM was developed to simultaneously monitor glycation and other quality attributes including afucosylation. The method was evaluated using two therapeutic antibodies with different glycosylation site numbers. Treatment with IdeS, Endo F2, and dithiothreitol generated three distinct subunits, and the glycation results obtained were similar to those treated with PNGase F, which is routinely used to release glycans; the sample processing time was greatly reduced while providing additional quality attribute information. The MS-based MAM was also employed to assess the glycation progression following forced glycation in various buffer solutions. A significant increase in oxidation was observed when forced glycation was conducted in an ammonium bicarbonate buffer solution, and a total of 23 potential glycation sites and 4 significantly oxidized sites were identified. Notably, we found that ammonium bicarbonate was found to specifically stimulate oxidation, while glycation had a synergistic effect on oxidation. These findings establish this study as a novel methodology for achieving a technologically advanced platform and concept that enhances the efficacy of product development and quality control, characterized by its broad-spectrum, rapid, and accurate nature.

Simultaneous prediction of 16 quality attributes during protein A chromatography using machine learning based Raman spectroscopy models.

Several key technologies for advancing biopharmaceutical manufacturing depend on the successful implementation of process analytical technologies that can monitor multiple product quality attributes in a continuous in-line setting. Raman spectroscopy is an emerging technology in the biopharma industry that promises to fit this strategic need, yet its application is not widespread due to limited success for predicting a meaningful number of quality attributes. In this study, we addressed this very problem by demonstrating new capabilities for preprocessing Raman spectra using a series of Butterworth filters. The resulting increase in the number of spectral features is paired with a machine learning algorithm and laboratory automation hardware to drive the automated collection and training of a calibration model that allows for the prediction of 16 different product quality attributes in an in-line mode. The demonstrated ability to generate these Raman-based models for in-process product quality monitoring is the breakthrough to increase process understanding by delivering product quality data in a continuous manner. The implementation of this multiattribute in-line technology will create new workflows within process development, characterization, validation, and control.

Digital design of an integrated purification system for continuous pharmaceutical manufacturing

In this work dynamic models of the continuous crystallization, filtration, deliquoring, washing, and drying steps are introduced, which are developed in the open-source pharmaceutical modeling tool PharmaPy. These models enable the simulation and digital design of an integrated continuous two-stage crystallization and filtration-drying carousel system. The carousel offers an intensified process that can manufacture products with tailored properties through optimal design and control. Results show that improved crystallization design enhances overall process efficiency by improving critical material attributes of the crystal slurry for downstream filtration and drying operations. The digital design of the integrated process achieves enhanced productivity while satisfying multiple design and product quality constraints. Additionally, the impact of model uncertainty on the optimal operating conditions is investigated. The findings demonstrate the systematic process development potential of PharmaPy, providing improved process understanding, design space identification, and optimized robust operation.

A modeling method of wide random forest multi-output soft sensor with attention mechanism for quality prediction of complex industrial processes

Complex industrial production processes often involve multiple product quality indicators that are interrelated. There exists a complex nonlinear mapping relationship between the operational input feature variables and multiple output target quality variables, making it difficult to accurately model through first-principle models. In order to fully capture the complex relationship between measurable variables and difficult-to-measure quality variables, and achieve accurate prediction of multiple output variables to meet the needs of practical industrial sites, this paper proposes a broad random forest-based multi-output soft sensor modeling method based on the idea of attention mechanism derived from the concept of broad learning systems. This method comprehensively considers the dynamic impact of different feature variables on the target quality indicators in actual production processes. The attention mechanism assists the soft sensor model in capturing contextual information better when dealing with long sequences, with a focus on the relevant parts related to the current task. Additionally, the interpretable random forest algorithm is employed as the weight estimator for the basic feature learning unit of Broad-based learning, enabling regression modeling of multiple target quality variables. The use of Broad-based random forest improves the model’s learning ability, interpretability, and generalization capability. To validate the reliability of the proposed method, it was applied to real industrial cases. The results demonstrated that the multi-output quality variable prediction performance of the proposed soft sensor outperforms existing soft sensors in terms of prediction accuracy. This indicates promising industrial application prospects for the proposed method.

A multitask encoder–decoder model for quality prediction in injection moulding

It is desirable in injection moulding that every yielded product is of high quality and precision. Several studies are conducted on modelling the relationship between the process parameters e.g., holding pressure, mould temperature, etc. that are either time series or non-time series, and the output product quality characteristics e.g., weight, dimensions, and warpage. To build a more comprehensive/generalized quality prediction model with online learning, this study for the first time takes time series process parameters from 2 pairs of in-mould pressure and temperature sensors simultaneously with 18 non-time series parameters and introduces a single deep learning architecture to perform multitask learning i.e., to leverage information from the related sources to simultaneously predict multiple product quality characteristics on a real injection moulding machine. This along with the discussed modern containerization practices develops a fully automated software part of the quality control process. The non-time series process parameters are learned by fully connected neural network layers of the model while the time series process parameters are related through an attention-based encoder–decoder architecture. This model is ablated as well as compared with a benchmark model to validate the performance advantage in terms of the reduced error asymptote. The proposed model offers an order of magnitude lower average mean squared error than the benchmark model, and 1.2 to 3 times smaller mean squared errors than its ablated versions.

Method validation and new peak detection for the liquid chromatography-mass spectrometry multi-attribute method

The multi-attribute method (MAM) is a liquid chromatography-mass spectrometry (LC-MS) peptide mapping technique that has been proposed as a replacement for several conventional quality control (QC) methods for therapeutic proteins. In addition to quantification of multiple product quality attributes (PQAs), MAM can also monitor impurities using a new peak detection (NPD) feature. Here, results are provided from method validation and NPD studies of an MAM approach applied to rituximab as a model monoclonal antibody (mAb). Twenty-one rituximab PQAs were monitored, including oxidation, pyroglutamination, deamidation, lysine clipping, and glycosylation. The PQA monitoring aspect of the method was validated according to ICH Guidance. Accuracy, precision, specificity, detection and quantitation limits, linearity, range, and robustness were demonstrated for this MAM approach with minimal issues. All PQAs were successfully validated except for several oxidation sites, which did not pass intermediate precision criteria. The variability found in oxidation measurements was attributed to artificial oxidation during sample preparation and could likely be alleviated through several approaches. The NPD aspect of the method was also evaluated. A spike-in approach was used to assess the limits of detection and quantitation (LOD/LOQ) of the NPD feature of MAM. For NPD, the peak intensity threshold was found to be the most critical parameter for accurate detection of impurities since a low threshold can result in false positives while a high threshold can obscure the detection of true peaks. Overall, the MAM approach presented and validated here has been demonstrated to be suitable for both targeted monitoring of rituximab PQAs and non-targeted detection of new peaks that represent impurities.

Rapid analysis of titer, aggregate, and intact mass of antibody therapeutics using automated multi-dimensional liquid chromatography coupled with native mass spectroscopy.

Monoclonal antibodies are tetrameric complex proteins, primarily produced using mammalian cell culture. Attributes such as titer, aggregates, and intact mass analysis are monitored during process development/optimization. In the present study, a novel workflow such that the Protein-A affinity chromatography is performed in the first dimension for purification and titer estimation, whereas size exclusion chromatography is employed in the second dimension to characterize size variants using native mass spectrometry. The present workflow offers a significant advantage over the traditionally used standalone Protein-A affinity chromatography followed by size exclusion chromatography analysis in that it can monitor these four attributes in 8 min while requiring a minimal sample size (10-15μg) and not requiring any manual peak collection. In contrast, the traditional standalone approach requires manual collection of eluted peaks in Protein-A affinity chromatography followed by buffer exchange to a mass-compatible buffer, which can take up to 2-3h with considerable risk of sample loss, degradation, and induced modifications. As the biopharma industry moves to make analytical testing efficient, we believe that the approach proposed here would be of significant interest due to its ability to monitor multiple process and product quality attributes in a single workflow and via rapid analysis.

Improvements on sample preparation and peptide separation for reduced peptide mapping based multi-attribute method analysis of therapeutic monoclonal antibodies using lysyl endopeptidase digestion

The mass spectrometry based multi-attribute method (MAM) has gained popularity in the field of biopharmaceutical analysis as it promises a single method for comprehensive monitoring of multiple product quality attributes (PQAs) and product purity. Sample preparation for protein digestion and peptide separation are critical considerations for a reduced peptide mapping-based MAM. To avoid desalting steps required in most tryptic protein digestion and in order to improve peptide separation for hydrophilic peptides, we developed an improved robust sample preparation using lysyl endopeptidase (Lys-C) for high-throughput MAM testing. Additionally, this method optimizes the peptide retention and separation of a stability-indicating VSNK peptide using a HSS T3 column for comprehensive PQA monitoring. A fully automated sample preparation had similar assay variations for PQAs monitoring compared to manual sample preparation. To the best of our knowledge, this is the first report of a high-resolution MS-based MAM using a streamlined Lys-C digestion without desalting with enhanced PQA monitoring for hydrophilic peptides. The improved, robust MAM workflow for protein digestion and peptide separation will pave the way for broader MAM qualification and its applications for the characterization and quality control of therapeutic monoclonal antibodies.

Simultaneous monitoring of multiple attributes of pyrrolobenzodiazepine antibody-drug conjugates by size exclusion chromatography – high resolution mass spectrometry

Antibody-drug conjugates (ADCs) are an emerging class of oncology treatments combining the unique specificity of monoclonal antibodies with the highly cytotoxic properties of small molecule compounds. Pyrrolobenzodiazepines (PBDs) are highly potent agents capable of inhibiting cellular DNA replication which leads to apoptosis. To ensure efficacy and patient safety upon administration of such toxic and heterogeneous molecules, their structure and quality attributes must be closely monitored. Size exclusion chromatography (SEC) is a powerful, fast and robust tool for the separation of compounds varying in molecular weight. When using volatile components in the chromatographic mobile phase, SEC has also been shown to be amenable for interfacing to mass spectrometry, providing potential for reliable identification of protein isoforms across the size variants present. Here, we present a SEC-MS method developed for the characterisation of PBD-based ADCs on the intact molecular level. We demonstrate that information on ADC monomers such as the glycoform distribution and the average drug-antibody ratio (DAR) can be obtained in 15 minutes of analysis time. Qualitative and quantitative information on low and high molecular weight impurities such as aggregates and fragments, fundamental for critical quality attribute analysis of biopharmaceuticals, can be generated simultaneously. SEC-MS enables the characterisation of multiple product quality attributes of complex biotherapeutics at the same time.

Fully automated peptide mapping multi-attribute method by liquid chromatography–mass spectrometry with robotic liquid handling system

The multi-attribute method (MAM) based on liquid chromatography (LC)–tandem mass spectrometry is emerging as a powerful tool to directly monitor multiple product quality attributes simultaneously. To better implement MAM, either for product characterization or for quality control (QC), there is a need for a robust, universal, and high-throughput workflow that can be broadly adopted in different laboratories with minimal barriers to implementation. Manual preparation of samples for MAM, however, is labor intensive and produces nontrivial variations across analysts and laboratories. We describe the development of a fully automated peptide mapping procedure with a high-throughput robotic liquid handling system to improve sample handling capacity and outcome reproducibility while saving analyst hands-on time. Our procedure features the automation of a “microdialysis” step, an efficient desalting approach prior to proteolytic digestion that optimizes digestion completeness and consistency each time. The workflow is completely hands-free and requires the analyst only to pre-normalize the sample concentrations and to load buffers and reagents at their designated positions on the robotic deck. The robotic liquid handler performs all the subsequent preparation steps and stores the digested samples on a chiller unit to await retrieval for further analysis. We also demonstrate that the manual and automated procedures are comparable with regard to protein sequence coverage, digestion completeness and consistency, and quantification of posttranslational modifications. Notably, in contrast to a previously reported automated sample preparation protocol that relied on customized accessories, all components in our automation procedure are commercial products that are readily available. In addition, we also present the high-throughput data analysis workflow by using Protein Metrics. The automation procedure can be applied cross-functionally in the biopharmaceutical industry and, given its practicality and reproducibility, can pave the way for MAM implementation in QC laboratories.

A soft-computing framework for automated optimization of multiple product quality criteria with application to micro-fluidic chip production

We describe a general strategy for optimizing the quality of products of industrial batch processes that comprise multiple production stages. We focus on the particularities of applying this strategy in the field of micro-fluidic chip production. Our approach is based on three interacting components: (i) a new hybrid design of experiments (DoE) strategy that combines expert- and distribution-based space exploration with model-based uncertainty criteria to obtain a representative set of initial samples (i.e., settings of essential machining process parameters), (ii) construction of linear and non-linear predictive mappings from these samples to describe the relation between machining process parameters and resulting quality control (QC) values and (iii) incorporation of these mappings as surrogate fitness estimators into a multi-objective optimization process to discover settings that outperform those routinely used by operators. These optimized settings lead to final products with better quality and/or higher functionality for the clients. The optimization module employs a co-evolutionary strategy we developed that is able to deliver better Pareto non-dominated solutions than the renowned NSGA-II multi-objective solver. We applied our proposed high-level surrogate-based multi-objective strategy both in a single/late-stage optimization scenario and in a more challenging multi-stage scenario, yielding final optimization results that improved parameter settings and thus product quality compared to standard expert-based production process parameterizations.

Simultaneous Monitoring and Comparison of Multiple Product Quality Attributes for Cell Culture Processes at Different Scales Using a LC/MS/MS Based Multi-Attribute Method

In response to the FDA's call for applying Quality by Design (QbD) to the manufacturing process, the biopharmaceutical industry has invested extensively into the monitoring and controlling of product quality attributes for bioprocesses. To assure the safety and efficacy of the drug product, defining critical quality attributes (CQA) and understanding their correlation with critical process parameters (CPP) becomes vitally important. In this work, a liquid chromatography-mass spectrometry based multi-attribute method (MAM) has been applied to the monitoring and trending of multiple CQAs of a monoclonal antibody product. To the best of our knowledge, this is the first demonstration of applying MAM to both a 3-liter development mini-bioreactor (3 L bioreactor) and a 2000-liter GMP single use bioreactor (2000L SUB). MAM was proven not only to be a great analytical tool for monitoring product quality attributes throughout the time course of the cell culture process, it could also provide critical product quality information in order to understand any potential process performance differences during scale-up and/or technology transfer. The successful monitoring and trending of the multiple CQAs throughout the 17-day cell culture process lays a solid foundation for possible real time in-process control and release of biotherapeutics using MAM in the future.

Tool Replacement for Photovoltaic Industry Processes With Multiple Characteristics Based on Capability Index

Abstract In the photovoltaic industry, measuring efficiency is an important process for manufacturing silicon wafers. The cutting tool gradually wears during the measuring efficiency process. Consequently, the fraction of defectives would gradually become significant. When the fraction defective reaches a certain level, the tool must be replaced. Therefore, it is essential to find out the optimal time to replace the cutting tool for minimizing the fraction defective and reducing the manufacturing cost. In this research, we consider the replacement problem in the photovoltaic industry to obtain the optimal replacement time for the measuring efficiency process. The measuring efficiency process involves multiple product quality characteristics, including the Isc, Voc, and Rsh. In order to maintain the high wafer process quality and to minimize the production cost, we first develop a method for measuring the manufacturing capability of measuring efficiency process. Then, a tool replacement policy is proposed to assess the true capability at each time period of measuring efficiency process and to find out the optimal time for tool replacement.

A modified multivariate process capability vector

Process capability indices have been widely used in industries to assess the performance of the manufacturing processes. Various different multivariate capability indices have been introduced. In this paper, a new multivariate capability vector is proposed under the assumption of multivariate normality, to assess the production capability of the processes that involve multiple product quality characteristics. Also, we investigate the relation between this index and process centering, as well as the relation between this index and the lower and upper bounds of percentage of non-conforming items manufactured. Two real manufacturing data set are used to demonstrate the effectiveness of the proposed index.

Modified multivariate process capability index using principal component analysis

The existing research of process capability indices of multiple quality characteristics mainly focuses on nonconforming of process output, the concept development of univariate process capability indices, quality loss function and various comprehensive evaluation methods. The multivariate complexity increases the computation difficulty of multivariate process capability indices(MPCI), which makes them hard to be used in practice. In this paper, a new PCA-based MPCI approach is proposed to assess the production capability of the processes that involve multiple product quality characteristics. This approach first transforms the original quality variables into standardized normal variables. MPCI measures are then provided based on the Taam index. Moreover, the statistical properties of these MPCIs, such as confidence intervals and lower confidence bound, are given to let the practitioners understand the capability indices as random variables instead of deterministic variables. A real manufacturing data set and a synthetic data set are used to demonstrate the effectiveness of the proposed method. An implementation procedure is also provided for quality engineers to apply our MPCI approach in their manufacturing processes. The case studies demonstrate the effectiveness and feasibility of this new kind of MPCI, which is easier to be used in production practice. The proposed research provides a novel approach of MPCI calculation.

Determination of optimum washing conditions for a lignite coal based on ash and sulfur content

This study deals with the maximization of composite yield subjected to multiple product constraints simultaneously based on float–sink data of a lignite coal from Yozgat-Ayrıdam district of Turkey by using equal incremental product quality approach. Solver and Mathcad were used for determination of optimum separation densities for the coarser size fraction (130×18mm) and the finer size fraction (18×0.5mm) by Dense Medium Bath (DMB) and Dense Medium Cyclone (DMC), respectively. Four different cases were analyzed for the incremental product quality approach. In the first case, composite yield was maximized to 87.45% for the target composite ash of 18.26% with sulfur content of 0.98% when the incremental ashes were equalized by Solver. In the second case, yield maximized to 76.55% for the target sulfur of 0.83% with an ash content of 14.65% by Solver equalization of the incremental sulfur. In the third case, multiple product quality (ash and sulfur) was employed simultaneously by Solver. Yield was maximized only to 66.87% with ash and sulfur content of 11.12% and 0.74%, respectively. In the fourth case, multiple constraints were subjected to the maximization of the yield simultaneously using Mathcad. Yield was maximized to 66.33% with ash and sulfur content of 11.00% and 0.74%, respectively. The yield maximization was successfully achieved using Solver and Mathcad based on multiple product quality (ash and sulfur) constraints simultaneously as well as single product quality constraint.

A model of strategic product quality and process improvement incentives

In many production firms it is common practice to financially reward managers for firm performance improvement. The use of financial incentives for improvement has been widely researched in several analytical and empirical studies. Literature has also addressed the strategic effect of incentives, in particular what the effect of certain incentive structures would be on the behavior of a firm's competitor(s). Most of these studies, however, focus on sales incentives. In this paper we investigate the effects of strategic incentives for product quality and process improvement using a game theoretic model that considers two owner–manager pairs in competition. We find that if one of the managers is told to only maximize firm profits (which in fact is similar to profit incentives), the other manager will be offered positive incentives for product quality and process improvement. These product quality and process improvement incentives result in increased profits, at the expense of the profits of the other firm. Also we find that if both firm owners have the possibility to offer incentives for product quality and process improvement, they will both do so. However, this equilibrium essentially entails a prisoner's dilemma, in which the two firms earn lower profits compared to a situation in which the owners instruct their managers only to maximize firm profits. Insights into the normalization of the problem and the aggregation of multiple product quality and process improvement variables are also discussed.

Aircraft Turbine Engine Manufacturing with Multiple Specifications

Abstract The blade is a key part of an aircraft engine because its geometry and manufacturing quality affect engine performance and the lift time of the airplane. Therefore, blade inspection is a key aspect of engine quality. Measuring and analyzing airfoil section parameters and geometric tolerance are essential but complex tasks in blade inspection. This study focuses on the airfoil geometry. Dimensional inspection of complex geometry parts by coordinate measuring machines is a common practice in both the automotive and aerospace industries. The conventional approach is to accept or reject parts according to inspection data. This study used a method of evaluating the capability of the blade manufacturing process to produce blades that conform to the profile specifications. The blade is possessed of multiple characteristics. All these characteristics influence the blade manufacturing process yield. An overall yield measure index CpkT is used for performing accurate yield assessments of processes with multiple characteristics. That is, the index is a generalization of the index Cpk. The applied method measures the process capability to produce a blade that meets the manufacturing specifications considering multiple product quality characteristics, including the segments of the airfoil section.

Genetic algorithms — A novel technique to optimize coal preparation plants

A coal preparation plant typically operates with multiple cleaning circuits based on the particle size distribution of run-of-mine coal. Clean coal product from a plant commonly has to satisfy multiple product quality constraints, including product ash, product sulfur, heating value, moisture content, etc. Numerous studies in the past illustrate that the optimal yield of the plant can be obtained by operating each circuit to produce the same incremental product quality. This equal incremental product quality approach optimizes the plant yield considering only one product quality at a time. Thus, when required to simultaneously satisfy multiple product quality constraints, the process not only becomes increasingly complex and cumbersome, but also may lead to erroneous conclusions in many cases. A novel plant optimization technique was developed using genetic algorithms (GA) to maximize the overall revenue generated by a coal preparation plant by searching the best possible combination of overall yield and multiple product quality constraints. This approach is based on an evolutionary algorithm that maximizes the overall plant revenue based on a single objective function, which was developed by incorporating clean coal yield, targeted product ash content, product heating value, and product SO 2 emission potential. Comparative results discussed in this publication indicate the suitability of the proposed GA-based plant optimization approach.

Coal preparation plant optimization: A critical review of the existing methods

A coal preparation plant typically operates with multiple cleaning circuits to clean individual size fractions of run-of-mine coal. Coal preparation plants are traditionally optimized using the equalization of incremental product quality approach. Individual cleaning circuits are operated at the same specific incremental product quality so that the targeted overall plant product quality is achieved. Over the years, it has been well established that equal incremental product quality approach maximizes plant-yield for a given product quality constraint. However, while dealing with multiple quality constraints, the incremental quality approach may not provide a complete solution to the optimization problem. It may be intuitive to realize that the dirtiest particle(s) in a coal product, with respect to ash content, may not be the same particle(s) with respect to sulfur content. Therefore, with increasing number of product quality constraints, which may include (but not limited to) limiting ash, sulfur and trace element contents, the plant has to be optimized based on each incremental product quality. Understandably, the operating points selected for each circuit to maximize plant-yield based on incremental ash content, may not be suitable for obtaining maximum plant yield based on incremental sulfur content. These limitations of the equalization of incremental product quality approach to satisfy multiple product quality constraints have been reviewed in detail in this publication with an example of ash and sulfur data collected from an operating coal preparation plant.