Adaptive modeling optimized by the data fusion strategy: Real-time dying cell percentage prediction using capacitance spectroscopy.

The previous research showcased a partial least squares (PLS) regression model accurately predicting cell death percentages using in-line capacitance spectra. The current study advances the model accuracy through adaptive modeling employing a data fusion approach. This strategy enhances prediction performance by incorporating variables from the Cole-Cole model, conductivity and its derivatives over time, and Mahalanobis distance into the predictor matrix (X-matrix). Firstly, the Cole-Cole model, a mechanistic model with parameters linked to early cell death onset, was integrated to enhance prediction performance. Secondly, the inclusion of conductivity and its derivatives over time in the X-matrix mitigated prediction fluctuations resulting from abrupt conductivity changes during process operations. Thirdly, Mahalanobis distance, depicting spectral changes relative to a reference spectrum from a previous time point, improved model adaptability to independent test sets, thereby enhancing performance. The final data fusion model substantially decreased root-mean squared error of prediction (RMSEP) by around 50%, which is a significant boost in prediction accuracy compared to the prior PLS model. Robustness against reference spectrum selection was confirmed by consistent performance across various time points. In conclusion, this study illustrates that the data fusion strategy substantially enhances the model accuracy compared to the previous model relying solely on capacitance spectra.

Capacitance spectroscopy enables real-time monitoring of early cell death in mammalian cell culture.

BACKGROUND/AIMS: Previous work developed a quantitative model using capacitance spectroscopy in an at-line setup to predict the dying cell percentage measured from a flow cytometer. This work aimed to transfer the at-line model to monitor lab-scale bioreactors in real-time, waiving the need for frequent sampling and enabling precise controls. METHODS AND RESULTS: Due to the difference between the at-line and in-line capacitance probes, direct application of the at-line model resulted in poor accuracy and high prediction bias. A new model with a variable range and offering similar spectral shape across all probes was first constructed, improving prediction accuracy. Moreover, the global calibration method included the variance of different probes and scales in the model, reducing prediction bias. External parameter orthogonalization, a preprocessing method, also mitigated the interference from feeding, which further improved model performance. The root-mean-square error of prediction of the final model was 6.56% (8.42% of the prediction range) with an R2 of 92.4%. CONCLUSION: The culture evolution trajectory predicted by the in-line model captured the cell death and alarmed cell death onset earlier than the trypan blue exclusion test. Additionally, the incorporation of at-line spectra following orthogonal design into the calibration set was shown to generate calibration models that are more robust than the calibration models constructed using the in-line spectra only. This is advantageous, as at-line spectral collection is easier, faster, and more material-sparing than in-line spectra collection.

Rapid at-line early cell death quantification using capacitance spectroscopy.

Cell death is one of the failure modes of mammalian cell culture. Apoptosis is a regulated cell death process mainly observed in cell culture. Timely detection of apoptosis onset allows opportunities for preventive controls that ensure high productivity and consistent product quality. Capacitance spectroscopy captures the apoptosis-related cellular properties changes and thus quantifies the percentage of dying cells. This study demonstrated a quantification model that measures the percentage of apoptotic cells using a capacitance spectrometer in an at-line setup. When predicting the independent test set collected from bench-scale bioreactors, the root-mean-squared error of prediction was 8.8% (equivalent to 9.9% of the prediction range). The predicted culture evolution trajectory aligned with measured values from the flow cytometer. Furthermore, this method alarms cell death onset earlier than the traditional viability test, that is, the trypan blue exclusion test. Compared to flow cytometry (the traditional early cell death detection method), this method is rapid, simple, and less labor-intensive. In addition, this at-line setup can be easily transferred between scales (e.g., lab-scale for development to manufacturing scale), which benefits process transfers between facilities, scale-up, and other process transitions.

Real-time monitoring of quality attributes by in-line Fourier transform infrared spectroscopic sensors at ultrafiltration and diafiltration of bioprocess.

Technologies capable of monitoring product quality attributes and process parameters in real time are becoming popular due to the endorsement of regulatory agencies and also to support the agile development of biotherapeutic pipelines. The utility of vibrational spectroscopic techniques such as Fourier transform mid-infrared (Mid-IR) and multivariate data analysis (MVDA) models allows the prediction of multiple critical attributes simultaneously in real time. This study reports the use of Mid-IR and MVDA model sensors for monitoring of multiple attributes (excipients and protein concentrations) in real time (measurement frequency of every 40 s) at ultrafiltration and diafiltration (UF/DF) unit operation of biologics manufacturing. The platform features integration of fiber optic Mid-IR probe sensors to UF/DF set up at the bulk solution and through a flow cell at the retentate line followed by automated Mid-IR data piping into a process monitoring software platform with pre-loaded partial least square regression (PLS) chemometric models. Data visualization infrastructure is also built-in to the platform so that upon automated PLS prediction of excipients and protein concentrations, the results were projected in a graphical or numerical format in real time. The Mid-IR predicted concentrations of excipients and protein show excellent correlation with the offline measurements by traditional analytical methods. Absolute percent difference values between Mid-IR predicted results and offline reference assay results were ≤5% across all the excipients and the protein of interest; which shows a great promise as a reliable process analytical technology tool.

Process Analytical Technology for High Shear Wet Granulation: Wet Mass Consistency Reported by In-Line Drag Flow Force Sensor Is Consistent With Powder Rheology Measured by At-Line FT4 Powder Rheometer.

Drag flow force (DFF) sensor that measures the force exerted by wet mass in a granulator on a thin cylindrical probe was shown as a promising process analytical technology for real-time in-line high-resolution monitoring of wet mass consistency during high shear wet granulation. Our previous studies indicated that this process analytical technology tool could be correlated to granulation end point established independently through drug product critical quality attributes. In this study, the measurements of flow force by a DFF sensor, taken during wet granulation of 3 placebo formulations with different binder content, are compared with concurrent at line FT4 Powder Rheometer characterization of wet granules collected at different time points of the processing. The wet mass consistency measured by the DFF sensor correlated well with the granulation's resistance to flow and interparticulate interactions as measured by FT4 Powder Rheometer. This indicated that the force pulse magnitude measured by the DFF sensor was indicative of fundamental material properties (e.g., shear viscosity and granule size/density), as they were changing during the granulation process. These studies indicate that DFF sensor can be a valuable tool for wet granulation formulation and process development and scale up, as well as for routine monitoring and control during manufacturing.

The use of in situ near infrared imaging and Raman mapping to study the disproportionation of a drug HCl salt during dissolution.

NIR imaging and Raman mapping of the dissolution of model pharmaceutical formulations containing the HCl salt of a developmental compound, were carried out using a custom designed flow through cell. The results of this work have shown that NIR imaging and Raman mapping are capable of monitoring the distribution of the components in a formulation during dissolution while also revealing any form changes which may occur in real time. The NIR and Raman data revealed that the drug underwent conversion to the free base when water was used as the dissolution medium. However, in 0.1M HCl this conversion was no longer seen as the medium was below the pHmax (the pH of saturation of both unionised and ionised species and above which the free base can form) of the drug. The data from both approaches broadly agreed demonstrating the applicability of these methods to studying and enhancing our understanding of the complex physical and chemical processes which occur during dissolution in real time.

Comparison of near infrared and microwave resonance sensors for at-line moisture determination in powders and tablets.

In this paper we demonstrate the feasibility of replacing KF for water content testing in bulk powders and tablets with at-line near infrared (NIR) or microwave resonance (MR) methods. Accurate NIR and MR prediction models were developed with a minimalistic approach to calibration. The NIR method can accurately predict water content in bulk powders in the range of 0.5-5% w/w. Results from this method were compared to a MR method. We demonstrated excellent agreement of both NIR and MR methods for powders vs. the reference KF method. These methods are applicable to in-process control or quality control environments. One of the aims of this study was to determine if a calibration developed for a particular product could be used to predict the water content of another product (with related composition) but containing a different active pharmaceutical ingredient (API). We demonstrated that, contrary to the NIR method, a general MR method can be used to predict water content in two different types of blends. Finally, we demonstrated that a MR method can be developed for at-line moisture determination in tablets.