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.

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.

Development of an amino acid sequence-dependent analytical method for peptides using near-infrared spectroscopy.

We aimed to develop an amino acid sequence-dependent analytical method using near-infrared (NIR) spectroscopy. The detailed analysis of the NIR spectra of eight different amino acid aqueous solutions (glycine, alanine, serine, glutamine, lysine, phenylalanine, tyrosine, and proline) revealed different spectral patterns characteristic of different amino acid residues in the 6200-5700 and 5000-4200 cm-1 regions, and the amino acids were identified based on the patterns. The spectra in the region of 5000-4500 cm-1 for tripeptide organic solutions that were composed of the aforementioned eight amino acids clearly showed the spectral differences depending on the amino acid species and amino acid sequences. Namely, tripeptide species were clearly differentiated from each other based on the spectral pattern of NIR bands due to the combinations of N-H stretching and amide II/III modes and those derived from the first overtones of amide II and amide I. The quantitative evaluation of changes in the concentrations of dipeptides and tripeptides composed of two different amino acids, glycine and proline was performed using partial least squares regression (PLSR) analysis and a combination of bands for amide modes. The calibration and validation results with high determination coefficients (R2 ≥ 0.99) were successfully obtained based on the amino acid sequences. The results not only revealed the usefulness of NIR spectroscopy as a process analytical technology (PAT) tool for synthesizing peptides in a micro flow reactor but also proposed a general method for quantitatively analyzing NIR spectra obtained in the course of chemical synthesis.

Method of Monitoring the Number of Amide Bonds in Peptides Using Near-Infrared Spectroscopy.

Using near-infrared (NIR) spectroscopy, we aimed to develop a method of monitoring the increasing number of amide bonds with the elongation of the chain length of peptides. Because peptide synthesis can be monitored by evaluating the increasing number of amide bonds with dehydration occurring between amino acids, polyglycine, which has the simplest structure among polyamino acids, was studied, and the key bands whose absorption intensities increased with the elongation of the chain length, such as the bands attributed to glycine, diglycine, triglycine, and tetraglycine, were searched. The bands due to the combinations of the amide A and amide II/III modes in the region of 5000-4500 cm-1 were revealed to be good candidates for key bands, their second derivative intensities increased as the number of amide bonds increased, regardless of pH, solvent species, and the presence of protecting groups. The number of amide bonds was evaluated by a partial least square regression using the abovementioned combination bands, and a calibration model with a high determination coefficient (≥0.99) was constructed. These results not only have demonstrated the usefulness of NIR spectroscopy as a process analytical technology tool for the process of synthesizing the peptide in a microflow reactor but also have provided basic knowledge for analyzing amide bonds in the NIR spectra of proteins, polyamino acids, polypeptides, and polyamides.

An Application for the Quantitative Analysis of Pharmaceutical Tablets Using a Rapid Switching System Between a Near-Infrared Spectrometer and a Portable Near-Infrared Imaging System Equipped with Fiber Optics.

We present a rapid switching system between a newly developed near-infrared (NIR) spectrometer and its imaging system to select the spot size of a diffuse reflectance (DR) probe. In a previous study, we developed a portable NIR imaging system, known as D-NIRs, which has significant advantages over other systems. Its high speed, high spectral resolution, and portability are particularly useful in the process of monitoring pharmaceutical tablets. However, the spectral accuracies relating to the changes in the formulation of the pharmaceutical tablets have not been fully discussed. Therefore, we improved the rapid optical switching system and present a new model of D-NIRs (ND-NIRs) here. This system can automatically switch the optical paths of the DR and NIR imaging probes, greatly contributing to the simultaneous measurement of both the imaging and spot. The NIR spectra of the model tablets, including 0-10% ascorbic acid, were measured and simultaneous NIR images of the tablets were obtained. The predicted results using spot sizes for the DR probe of 1 and 5 mm diameter, resulted in concentrations of R2 = 0.79 and 0.94, with root mean square errors (RMSE) of 1.78 and 0.89, respectively. For tablets with a high concentration of ascorbic acid, the NIR imaging results showed inhomogeneity in concentration. However, the predicted values for the low concentration samples appeared higher than the known concentration of the tablets, although the homogeneity of the concentration was confirmed. In addition, the optimal spot size using NIR imaging data was estimated to be 5-7 mm. The results obtained in this study show that the spot size of the fiber probe, attached to a spectrometer, is important in developing a highly reliable model to determine the component concentration of a tablet.

Baseline Correction of Diffuse Reflection Near-Infrared Spectra Using Searching Region Standard Normal Variate (SRSNV).

An alternative baseline correction method for diffuse reflection near-infrared (NIR) spectra, searching region standard normal variate (SRSNV), was proposed. Standard normal variate (SNV) is an effective pretreatment method for baseline correction of diffuse reflection NIR spectra of powder and granular samples; however, its baseline correction performance depends on the NIR region used for SNV calculation. To search for an optimal NIR region for baseline correction using SNV, SRSNV employs moving window partial least squares regression (MWPLSR), and an optimal NIR region is identified based on the root mean square error (RMSE) of cross-validation of the partial least squares regression (PLSR) models with the first latent variable (LV). The performance of SRSNV was evaluated using diffuse reflection NIR spectra of mixture samples consisting of wheat flour and granular glucose (0-100% glucose at 5% intervals). From the obtained NIR spectra of the mixture in the 10 000-4000 cm(-1) region at 4 cm intervals (1501 spectral channels), a series of spectral windows consisting of 80 spectral channels was constructed, and then SNV spectra were calculated for each spectral window. Using these SNV spectra, a series of PLSR models with the first LV for glucose concentration was built. A plot of RMSE versus the spectral window position obtained using the PLSR models revealed that the 8680­8364 cm(-1) region was optimal for baseline correction using SNV. In the SNV spectra calculated using the 8680­8364 cm(-1) region (SRSNV spectra), a remarkable relative intensity change between a band due to wheat flour at 8500 cm(-1) and that due to glucose at 8364 cm(-1) was observed owing to successful baseline correction using SNV. A PLSR model with the first LV based on the SRSNV spectra yielded a determination coefficient (R2) of 0.999 and an RMSE of 0.70%, while a PLSR model with three LVs based on SNV spectra calculated in the full spectral region gave an R2 of 0.995 and an RMSE of 2.29%. Additional evaluation of SRSNV was carried out using diffuse reflection NIR spectra of marzipan and corn samples, and PLSR models based on SRSNV spectra showed good prediction results. These evaluation results indicate that SRSNV is effective in baseline correction of diffuse reflection NIR spectra and provides regression models with good prediction accuracy.

Image monitoring of pharmaceutical blending processes and the determination of an end point by using a portable near-infrared imaging device based on a polychromator-type near-infrared spectrometer with a high-speed and high-resolution photo diode array detector.

In the present study we have developed a new version (ND-NIRs) of a polychromator-type near-infrared (NIR) spectrometer with a high-resolution photo diode array detector, which we built before (D-NIRs). The new version has four 5 W halogen lamps compared with the three lamps for the older version. The new version also has a condenser lens with a shorter focal point length. The increase in the number of the lamps and the shortening of the focal point of the condenser lens realize high signal-to-noise ratio and high-speed NIR imaging measurement. By using the ND-NIRs we carried out the in-line monitoring of pharmaceutical blending and determined an end point of the blending process. Moreover, to determinate a more accurate end point, a NIR image of the blending sample was acquired by means of a portable NIR imaging device based on ND-NIRs. The imaging result has demonstrated that the mixing time of 8 min is enough for homogeneous mixing. In this way the present study has demonstrated that ND-NIRs and the imaging system based on a ND-NIRs hold considerable promise for process analysis.

Application of a newly developed portable NIR imaging device to monitor the dissolution process of tablets.

We have recently developed a novel portable NIR imaging device (D-NIRs), which has a high speed and high wavelength resolution. This NIR imaging approach has been developed by utilizing D-NIRs for studying the dissolution of a model tablet containing 20 % ascorbic acid (AsA) as an active pharmaceutical ingredient and 80 % hydroxypropyl methylcellulose, where the tablet is sealed by a special cell. Diffuse reflectance NIR spectra in the 1,000 to 1,600 nm region were measured during the dissolution of the tablet. A unique band at around 1,361 nm of AsA was identified by the second derivative spectra of tablet and used for AsA distribution NIR imaging. Two-dimensional change of AsA concentration of the tablet due to water penetration is clearly shown by using the band-based image at 1,361 nm in NIR spectra obtained with high speed. Moreover, it is significantly enhanced by using the intensity ratio of two bands at 1,361 and 1,354 nm corresponding to AsA and water absorption, respectively, showing the dissolution process. The imaging results suggest that the amount of AsA in the imaged area decreases with increasing water penetration. The proposed NIR imaging approach using the intensity of a specific band or the ratio of two bands combined with the developed portable NIR imaging instrument, is a potentially useful practical way to evaluate the tablet at every moment during dissolution and to monitor the concentration distribution of each drug component in the tablet.

A polychromator-type near-infrared spectrometer with a high-sensitivity and high-resolution photodiode array detector for pharmaceutical process monitoring on the millisecond time scale.

In the fine chemicals industry, particularly in the pharmaceutical industry, advanced sensing technologies have recently begun being incorporated into the process line in order to improve safety and quality in accordance with process analytical technology. For estimating the quality of powders without preparation during drug formulation, near-infrared (NIR) spectroscopy has been considered the most promising sensing approach. In this study, we have developed a compact polychromator-type NIR spectrometer equipped with a photodiode (PD) array detector. This detector is consisting of 640 InGaAs-PD elements with 20-µm pitch. Some high-specification spectrometers, which use InGaAs-PD with 512 elements, have a wavelength resolution of about 1.56 nm when covering 900-1700 nm range. On the other hand, the newly developed detector, having the PD with one of the world's highest density, enables wavelength resolution of below 1.25 nm. Moreover, thanks to the combination with a highly integrated charge amplifier array circuit, measurement speed of the detector is higher by two orders than that of existing PD array detectors. The developed spectrometer is small (120 mm × 220 mm × 200 mm) and light (6 kg), and it contains various key devices including the high-density and high-sensitivity PD array detector, NIR technology, and spectroscopy technology for a spectroscopic analyzer that has the required detection mechanism and high sensitivity for powder measurement, as well as a high-speed measuring function for blenders. Moreover, we have evaluated the characteristics of the developed NIR spectrometer, and the measurement of powder samples confirmed that it has high functionality.