Marta B. Lopes

Near-infrared hyperspectral unmixing based on a minimum volume criterion for fast and accurate chemometric characterization of counterfeit tablets.

A rapid detection of the nonauthenticity of suspect tablets is a key first step in the fight against pharmaceutical counterfeiting. The chemical characterization of these tablets is the logical next step to evaluate their impact on patient health and help authorities in tracking their source. Hyperspectral unmixing of near-infrared (NIR) image data is an emerging effective technology to infer the number of compounds, their spectral signatures, and the mixing fractions in a given tablet, with a resolution of a few tens of micrometers. In a linear mixing scenario, hyperspectral vectors belong to a simplex whose vertices correspond to the spectra of the compounds present in the sample. SISAL (simplex identification via split augmented Lagrangian), MVSA (minimum volume simplex analysis), and MVES (minimum-volume enclosing simplex) are recent algorithms designed to identify the vertices of the minimum volume simplex containing the spectral vectors and the mixing fractions at each pixel (vector). This work demonstrates the usefulness of these techniques, based on minimum volume criteria, for unmixing NIR hyperspectral data of tablets. The experiments herein reported show that SISAL/MVSA and MVES largely outperform MCR-ALS (multivariate curve resolution-alternating least-squares), which is considered the state-of-the-art in spectral unmixing for analytical chemistry. These experiments are based on synthetic data (studying the effect of noise and the presence/absence of pure pixels) and on a real data set composed of NIR images of counterfeit tablets.

Determination of the composition of counterfeit Heptodin™ tablets by near infrared chemical imaging and classical least squares estimation

According to the WHO definition for counterfeit medicines, several categories can be established, e.g., medicines containing the correct active pharmaceutical ingredient (API) but different excipients, medicines containing low levels of API, no API or even a substitute API. Obviously, these different scenarios will have different detrimental effects on a patients health. Establishing the degree of risk to the patient through determination of the composition of counterfeit medicines found in the market place is thus of paramount importance. In this work, classical least squares was used for predicting the composition of counterfeit Heptodin tablets found in a market survey. Near infrared chemical imaging (NIR-CI) was used as a non-destructive measurement technique. No prior knowledge about the origin and composition of the tablets was available. Good API (i.e., lamivudine) predictions were obtained, especially for tablets containing a high API (close to the authentic) dose. Concentration maps of each pure material, i.e., the API (lamivudine) and the excipients microcrystalline cellulose, sodium starch glycollate, rice starch and talc, were estimated. Below 1% of the energy was not explained by the model (residuals percentage) for every pixel in all 12 counterfeit tablets. The similarities among tablets with respect to the total API percentage determined, as well as the corresponding concentration maps, support the classification of the tablets into the different groups obtained in previous work.

Investigation into classification/sourcing of suspect counterfeit Heptodin™ tablets by near infrared chemical imaging

Near infrared chemical imaging (NIR-CI) analysis was performed on 55 counterfeit Heptodin tablets obtained from a market survey and an additional 11 authentic Heptodin tablets for comparison. The aim of the study was to investigate whether NIR-CI can be used to detect the counterfeit tablets and to classify/source them so as to understand the possible number of origins to aid investigators and authorities to shut down counterfeiting operations. NIR-CI combined with multivariate analysis is particularly suited to compare chemical and physical properties of samples, since it is a quick and non-destructive method of analysis. Counterfeit tablets were easily distinguished from the authentic ones. Principal component analysis (PCA) and k-means clustering were performed on the data set. The results from both analyses grouped the counterfeit tablets in 13 main groups. The main groups found with both methods were quite consistent. Out of the 55 tablets only 18% contained the correct active pharmaceutical ingredient (API), i.e., the anti-viral drug lamivudine. The remaining 82% of counterfeit tablets contained talc and starch as main excipients. The API containing tablets classified into three main groups, based mainly on the amount of lamivudine present in the tablet. The group which had close to the correct amount of lamivudine sub-classified into three groups. From the analysis carried out, it is likely that the counterfeit tablets originate from as many as 15 different sources.

Identification of Polymer Materials Using Laser-Induced Breakdown Spectroscopy Combined with Artificial Neural Networks

Many pharmaceutical problems require chemical identification of the ingredients present in a drug product, e.g., a tablet. Examples include the identification of the compounds present in many steps of the manufacturing process and the chemical characterization of counterfeit and third-party tablets. Hyperspectral unmixing of near-infrared images is a key method for solving the above problems, as it provides estimates of the number of pure compounds present in a mixture, their spectral signatures, and the corresponding spatially mapped abundance fractions. The performance of hyperspectral unmixing depends upon the degree of homogeneity of the tablets, as well as the pixel resolution used for image acquisition. This work explores the use of the recent simplex identification via split augmented Lagrangian (SISAL) algorithm to unmix near-infrared images of tablets under different homogeneity and pixel resolution conditions. SISAL is known to solve complex problems beyond the reach of previous hyperspectral unmixing methods. The tablets used in this study are 4- and 5-compound model pharmaceutical mixtures, produced with good and poor blending processes, and the acquisition was performed at three pixel resolutions: 8.1, 27.9, and 40.3 μm/pixel. Heterogeneity proved to increase SISALs accuracy, as did increased pixel resolution in homogeneous tablets. Given the fast image acquisition and algorithm execution times, low- and high-resolution images should always be acquired; combined with the homogeneity grade of the samples, this may be determinant to a case-by-case decision on the proper action to be taken next.A combination of laser-induced breakdown spectroscopy (LIBS) and artificial neural networks (ANNs) has been used for the identification of polymer materials, including polypropylene (PP), polyvinyl chloride (PVC), polytetrafluoroethylene (PTFE), polyoxymethylene (POM), polyethylene (PE), polyamide or nylon (PA), polycarbonate (PC) and poly(methyl methacrylate) (PMMA). After optimization of the experimental setup and the spectrum acquisition protocol, successful identification rates between 81 and 100% were achieved using spectral features gathered from single spectra without averaging (1 second acquisition time) over a wide spectral range (240–820 nm). Furthermore, ten different materials based on PVC were tested using the identification procedure. Correct identifications were obtained as well. Sorting of the materials into sub-categories of PVC materials according to their charges (concentration in trace elements such as Ca) was performed. The demonstrated capacities fit, in practice, the needs of plastic-waste sorting and of producing high-grade recycled plastic materials.

In situ near infrared spectroscopy monitoring of cyprosin production by recombinant Saccharomyces cerevisiae strains

Near infrared (NIR) spectroscopy was used to in situ monitoring the cultivation of two recombinant Saccharomyces cerevisiae strains producing heterologous cyprosin B. NIR spectroscopy is a fast and non-destructive technique, that by being based on overtones and combinations of molecular vibrations requires chemometrics tools, such as partial least squares (PLS) regression models, to extract quantitative information concerning the variables of interest from the spectral data. In the present work, good PLS calibration models based on specific regions of the NIR spectral data were built for estimating the critical variables of the cyprosin production process: biomass concentration, cyprosin activity, cyprosin specific activity, the carbon sources glucose and galactose concentration and the by-products acetic acid and ethanol concentration. The PLS models developed are valid for both recombinant S. cerevisiae strains, presenting distinct cyprosin production capacities, and therefore can be used, not only for the real-time control of both processes, but also in optimization protocols. The PLS model for biomass yielded a R(2)=0.98 and a RMSEP=0.46 g dcw l(-1), representing an error of 4% for a calibration range between 0.44 and 13.75 g dcw l(-1). A R(2)=0.94 and a RMSEP=167 Um l(-1) were obtained for the cyprosin activity, corresponding to an error of 6.7% of the experimental data range (0-2509 Um l(-1)), whereas a R(2)=0.93 and RMSEP=672 U mg(-1) were obtained for the cyprosin specific activity, corresponding to an error of 7% of the experimental data range (0-11,690 Um g(-1)). For the carbon sources glucose and galactose, a R(2)=0.96 and a RMSECV of 1.26 and 0.55 g l(-1), respectively, were obtained, showing high predictive capabilities within the range of 0-20 g l(-1). For the metabolites resulting from the cell growth, the PLS model for acetate was characterized by a R(2)=0.92 and a RMSEP=0.06 g l (-1), which corresponds to a 6.1% error within the range of 0.41-1.23 g l(-1); for the ethanol, a high accuracy PLS model with a R(2)=0.97 and a RMSEP=1.08 g l(-1) was obtained, representing an error of 9% within the range of 0.18-21.76 g l(-1). The present study shows that it is possible the in situ monitoring and prediction of the critical variables of the recombinant cyprosin B production process by NIR spectroscopy, which can be applied in process control in real-time and in optimization protocols. From the above, NIR spectroscopy appears as a valuable analytical tool for online monitoring of cultivation processes, in a fast, accurate and reproducible operation mode.

In Situ Near-Infrared (NIR) versus High-Throughput Mid-Infrared (MIR) Spectroscopy to Monitor Biopharmaceutical Production

The development of biopharmaceutical manufacturing processes presents critical constraints, with the major constraint being that living cells synthesize these molecules, presenting inherent behavior variability due to their high sensitivity to small fluctuations in the cultivation environment. To speed up the development process and to control this critical manufacturing step, it is relevant to develop high-throughput and in situ monitoring techniques, respectively. Here, high-throughput mid-infrared (MIR) spectral analysis of dehydrated cell pellets and in situ near-infrared (NIR) spectral analysis of the whole culture broth were compared to monitor plasmid production in recombinant Escherichia coli cultures. Good partial least squares (PLS) regression models were built, either based on MIR or NIR spectral data, yielding high coefficients of determination (R2) and low predictive errors (root mean square error, or RMSE) to estimate host cell growth, plasmid production, carbon source consumption (glucose and glycerol), and by-product acetate production and consumption. The predictive errors for biomass, plasmid, glucose, glycerol, and acetate based on MIR data were 0.7 g/L, 9 mg/L, 0.3 g/L, 0.4 g/L, and 0.4 g/L, respectively, whereas for NIR data the predictive errors obtained were 0.4 g/L, 8 mg/L, 0.3 g/L, 0.2 g/L, and 0.4 g/L, respectively. The models obtained are robust as they are valid for cultivations conducted with different media compositions and with different cultivation strategies (batch and fed-batch). Besides being conducted in situ with a sterilized fiber optic probe, NIR spectroscopy allows building PLS models for estimating plasmid, glucose, and acetate that are as accurate as those obtained from the high-throughput MIR setup, and better models for estimating biomass and glycerol, yielding a decrease in 57 and 50% of the RMSE, respectively, compared to the MIR setup. However, MIR spectroscopy could be a valid alternative in the case of optimization protocols, due to possible space constraints or high costs associated with the use of multi-fiber optic probes for multi-bioreactors. In this case, MIR could be conducted in a high-throughput manner, analyzing hundreds of culture samples in a rapid and automatic mode.

Kinetic modeling of plasmid bioproduction in Escherichia coli DH5α cultures over different carbon-source compositions.

The need for the development of economic high plasmid production in Escherichia coli cultures is emerging, as a result of the latest advances in DNA vaccination and gene therapy. In order to contribute to achieve that, a model describing the kinetics involved in the bioproduction of plasmid by recombinant E. coli DH5α is presented, as an attempt to understand the complex and non-linear metabolic relationships and the plasmid production occurring in dynamic batch culture environments, run under different media compositions of glucose and glycerol, that result in distinct maximum biomass growths (between 8.2 and 12.8 g DCW/L) and specific plasmid productions (between 1.1 and 7.4 mg/g DCW). The model based on mass balance equations for biomass, glucose, glycerol, acetate and plasmid accurately described different culture behaviors, using either glucose or glycerol as carbon source, or mixtures of both. From the 17 parameters obtained after model simplification, the following 10 parameters were found to be independent of the carbon source composition: the substrate affinity constants, the inhibitory constants of biomass growth on glycerol by glucose, of biomass growth on acetate by glycerol and the global biomass growth by acetate, and the yields of biomass on acetate, acetate on glucose and glycerol, and plasmid on glucose. The parameters that depend on the culture composition, and that might explain the differences found between cultures, were: maximum specific growth rates on glucose, glycerol and acetate; biomass yield on glucose and glycerol; and plasmid yield on glycerol and acetate. Moreover, a crucial role of acetate in the plasmid production was revealed by the model, with most of plasmid production being associated to the acetate consumption. The model provides meaningful insight on the E. coli dynamic cell behavior concerning the plasmid bioproduction, which might lead to important guidelines for culture optimization and process scale-up and control.

Spectral unmixing via minimum volume simplices: Application to near infrared spectra of counterfeit tablets

Counterfeit pharmaceutical products pose a serious public health problem. It is thus important not only to detect them, but also to identify their composition and assess the risk for the patient. Identifying the spectral signatures of the pure compounds present in a (maybe counterfeit) tablet of unknown origin is clearly a hyperspectral unmixing problem. In fact, under a linear mixing model, the hyperspectral vectors belong to a simplex whose vertices are the spectra of the pure compounds in the mixture. Minimum volume simplex analysis (MVSA) and minimum-volume enclosing simplex (MVES) are recently proposed algorithms, exploiting the idea of finding a simplex of minimum volume fitting the observed data. This work gives evidence of the usefulness of MVES and MVSA for unmixing near infrared (NIR) hyperspectral data of tablets of unknown composition. Experiments reported in this paper show that MVES and MVSA strongly outperform the state-of-the-art method in analytical chemistry for spectral unmixing: multivariate curve resolution - alternating least squares (MCR-ALS). These experiments are based on synthetic data (studying the effect of noise and of the presence/absence of pure pixels) and on a real dataset composed of NIR hyperspectral images of counterfeit tablets.

Monitoring the ex-vivo expansion of human mesenchymal stem/stromal cells in xeno-free microcarrier-based reactor systems by MIR spectroscopy.

Human mesenchymal stem/stromal cells (MSCs) have received considerable attention in the field of cell‐based therapies due to their high differentiation potential and ability to modulate immune responses. However, since these cells can only be isolated in very low quantities, successful realization of these therapies requires MSCs ex‐vivo expansion to achieve relevant cell doses. The metabolic activity is one of the parameters often monitored during MSCs cultivation by using expensive multi‐analytical methods, some of them time‐consuming. The present work evaluates the use of mid‐infrared (MIR) spectroscopy, through rapid and economic high‐throughput analyses associated to multivariate data analysis, to monitor three different MSCs cultivation runs conducted in spinner flasks, under xeno‐free culture conditions, which differ in the type of microcarriers used and the culture feeding strategy applied. After evaluating diverse spectral preprocessing techniques, the optimized partial least square (PLS) regression models based on the MIR spectra to estimate the glucose, lactate and ammonia concentrations yielded high coefficients of determination (R2 ≥ 0.98, ≥0.98, and ≥0.94, respectively) and low prediction errors (RMSECV ≤ 4.7%, ≤4.4% and ≤5.7%, respectively). Besides PLS models valid for specific expansion protocols, a robust model simultaneously valid for the three processes was also built for predicting glucose, lactate and ammonia, yielding a R2 of 0.95, 0.97 and 0.86, and a RMSECV of 0.33, 0.57, and 0.09 mM, respectively. Therefore, MIR spectroscopy combined with multivariate data analysis represents a promising tool for both optimization and control of MSCs expansion processes.

Does Nonlinear Modeling Play a Role in Plasmid Bioprocess Monitoring Using Fourier Transform Infrared Spectra

The monitoring of biopharmaceutical products using Fourier transform infrared (FT-IR) spectroscopy relies on calibration techniques involving the acquisition of spectra of bioprocess samples along the process. The most commonly used method for that purpose is partial least squares (PLS) regression, under the assumption that a linear model is valid. Despite being successful in the presence of small nonlinearities, linear methods may fail in the presence of strong nonlinearities. This paper studies the potential usefulness of nonlinear regression methods for predicting, from in situ near-infrared (NIR) and mid-infrared (MIR) spectra acquired in high-throughput mode, biomass and plasmid concentrations in Escherichia coli DH5-α cultures producing the plasmid model pVAX-LacZ. The linear methods PLS and ridge regression (RR) are compared with their kernel (nonlinear) versions, kPLS and kRR, as well as with the (also nonlinear) relevance vector machine (RVM) and Gaussian process regression (GPR). For the systems studied, RR provided better predictive performances compared to the remaining methods. Moreover, the results point to further investigation based on larger data sets whenever differences in predictive accuracy between a linear method and its kernelized version could not be found. The use of nonlinear methods, however, shall be judged regarding the additional computational cost required to tune their additional parameters, especially when the less computationally demanding linear methods herein studied are able to successfully monitor the variables under study.