The RCPA(NZ) News In this issue - Professor Peter Browett, Haematologist, University of Auckland, discusses a case of Polymyositis provides an update on Non-Hodgkin's Lymphoma. Non-Hodgkin's Lymphoma Update Key Points • lymphomas are a heterogeneous group of malignancies with variable clinical presenta-tion, response to therapy and prognosis.
Doi:10.1016/j.addr.2005.01.020Advanced Drug Delivery Reviews 57 (2005) 1109 – 1143 Near-infrared spectroscopy and imaging: Basic principles and pharmaceutical applications Institute for Pharmacy and Molecular Biotechnology, Department of Pharmaceutical Technology and Pharmacology, University of Heidelberg, Im Neuenheimer Feld 366, D-69120 Heidelberg, Germany Received 17 December 2003; accepted 19 January 2005 Near-infrared (NIR) spectroscopy and imaging are fast and nondestructive analytical techniques that provide chemical and physical information of virtually any matrix. In combination with multivariate data analysis these two methods open manyinteresting perspectives for both qualitative and quantitative analysis. This review focuses on recent pharmaceutical NIRapplications and covers (1) basic principles of NIR techniques including chemometric data processing, (2) regulatory issues, (3)raw material identification and qualification, (4) direct analysis of intact solid dosage forms, and (5) process monitoring andprocess control.
D 2005 Elsevier B.V. All rights reserved.
Keywords: Noninvasive qualitative and quantitative analysis; Calibration and validation; Chemometrics; Raw material identification andcharacterization; Quality control of intact dosage forms; Process analytical technologies (PAT); Process monitoring Basic principles of near-infrared (NIR) spectroscopy . . . . . . . . . . . . . . . .
Origin and characteristics of NIR absorption bands . . . . . . . . . . . . . .
Instrumentation and sample presentation . . . . . . . . . . . . . . . . .
Theory and practice of chemometric data processing. . . . . . . . . . . . . . . .
Reduction of variables by principal component analysis (PCA) . . . . . . . . . . .
Multivariate calibration for quantitative analysis . . . . . . . . . . . . . . .
Multivariate classification for qualitative analysis . . . . . . . . . . . . . . .
T Tel.: +49 6221 548335; fax: +49 6221 545971.
0169-409X/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.addr.2005.01.020 G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 Actual status of pharmaceutical NIR analysis . . . . . . . . . . . . . . .
NIR spectroscopy in view of the U.S.F.D.A. initiative on PAT . . . . . . . . . . .
Identification and qualification of raw materials and intermediates . . . . . . . . . .
Quantitative calibration models . . . . . . . . . . . . . . . . .
Polymeric implants and microspheres . . . . . . . . . . . . . . .
Process monitoring and process control . . . . . . . . . . . . . . . . .
Tabletting and capsule-filling . . . . . . . . . . . . . . . . .
Basic principles and instrumentation . . . . . . . . . . . . . . . . . .
Analytical targets and strengths . . . . . . . . . . . . . . . . . . .
Pharmaceutical applications . . . . . . . . . . . . . . . . . . . .
process-control tool proceeded in jumps coincidingwith the introduction of efficient chemometric data Near-infrared spectroscopy (NIRS) is a fast and processing techniques and the development of novel nondestructive technique that provides multi-constit- spectrometer configurations based on fiber optic uent analysis of virtually any matrix. It covers the wavelength range adjacent to the mid infrared and In recent years, NIR spectroscopy has gained wide extends up to the visible region. Historically, the acceptance within the pharmaceutical industry for raw discovery of the NIR region in 1800 is ascribed to material testing, product quality control and process Herschel who separated the electromagnetic spectrum monitoring. The growing pharmaceutical interest in with a prism and found out that the temperature NIR spectroscopy is probably a direct result of its increased markedly towards and beyond the red, i.e. in major advantages over other analytical techniques, the region that is now called the near-infrared.
namely, an easy sample preparation without any Although a number of NIR experiments were carried pretreatments, the possibility of separating the sample out in the early 1920s, it was not before the mid to late measurement position and spectrometer by use of 1960s that NIR spectroscopy was practically used. It fiber optic probes, and the prediction of chemical and was Karl Norris from the U.S. Department of physical sample parameters from one single spectrum.
Agriculture who recognized the potential of this This paper is dedicated to pharmaceutical applica- analytical technique and introduced bmodern NIRSQ tions of NIR spectroscopy. To fully appreciate the into industrial practice From then on, the break- analytical versatility of this spectroscopic technique, a through of the method as an industrial quality- and short introduction into the principles of the method is G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 helpful. To this end, the author provides the reader between vibrational states of Dt = 2 or 3 are possible, with a short introduction into the theoretical funda- although their probability decreases with an increase mentals of the technique (Section 2.1), the equipment in the vibrational quantum number t. These multi- it uses (Section 2.2), and the mathematical and level energy transitions are the origin of NIR overtone statistical tools that are needed to process recorded bands that occur at multiples of the fundamental signals and extract the relevant information for vibrational frequency. For most chemical bonds the qualitative or quantitative analysis (Section 3). Sec- wave numbers of overtones can be estimated from tion 4 focuses on regulatory aspects that are critical their fundamental vibrations with an anharmonicity for pharmaceutical NIR analyses. Important current constant v of 0.01–0.05 by the following equation: and possible future pharmaceutical applications of NIR spectroscopy, including raw material identifica- tion and characterization, analysis of intact dosage where mx = wave number of x overtone, m0 = wave forms and process monitoring, are discussed in number of fundamental vibration, v = anharmonicity Section 5. Section 6 briefly emphasizes the pharma- ceutical potential of NIR imaging techniques.
Combination bands appearing between 1900 nm and 2500 nm are the result of vibrational interactions,i.e. their frequencies are the sums of multiples of each 2. Basic principles of near-infrared (NIR) interacting frequency. A special type of configuration interaction, called Fermi resonance, leads to thefeature that two NIR absorption bands of a polyatomic 2.1. Origin and characteristics of NIR absorption molecule with the same frequency do not simply overlay and sum up, but split in two peaks ofsomewhat higher and lower frequencies than the The American Society of Testing and Materials expected unperturbed position. Furthermore, intermo- (ASTM) defines the NIR region of the electro- lecular hydrogen bondings and dipole interactions magnetic spectrum as the wavelength range of 780– have to be considered, since they alter vibrational 2526 nm corresponding to the wave number range energy states, thus shifting existing absorption bands 12820–3959 cm1. The most prominent absorption and/or giving rise to new ones. This effect allows bands occurring in the NIR region are related to crystal forms, for instance, to be determined by NIR overtones and combinations of fundamental vibra- tions of –CH, –NH, –OH (and –SH) functional In conclusion, NIR absorption bands are typically groups. The key issues which determine the occur- broad, overlapping and 10–100 times weaker than rence and spectral properties, i.e. frequency and their corresponding fundamental mid-IR absorption intensity of NIR absorption bands are anharmonicity bands. These characteristics severely restrict sensitiv- and Fermi resonance, the physical basis of which will ity in the classical spectroscopic sense and call for be briefly described in this section. For a more chemometric data processing to relate spectral infor- comprehensive treatise the reader is referred to some mation to sample properties (see Section 3). The low excellent textbook chapters on the subject matter absorption coefficient, however, permits high pene- Since the energy curve of an oscillating molecule is tration depth and, thus, an adjustment of sample affected by intramolecular interactions, vibrations thickness. This aspect is actually an analytical around the equilibrium position are non-symmetric advantage, since it allows direct analysis of strongly and the spacings between energy levels that the absorbing and even highly scattering samples, such as molecule can attain are not identical, but rather turbid liquids or solids in either transmittance or decrease with increasing energy. This situation reflectance mode without further pretreatments.
resembles the quantum mechanical model of an The dual dependence of the analytical signal on the anharmonic oscillator. Since quantum mechanical chemical and physical properties of the sample, selection rules do not rigorously exclude transitions resulting from absorption and scatter effects, can be with Dt N 1 for anharmonic systems, transitions favorably used to perform chemical and physical
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 analysis from one single measurement. However, if photometers or light-emitting diode (LED)-based not the analytical target, scatter effects in NIR spectra, instruments provide selected frequencies, thus, cover- resulting from physical sample variations, may also ing only a narrow spectral range of 50–100 nm.
pose more or less severe analytical problems. In these Diffraction grating, interferometer, diode-array or situations, they have to be considered in the calibra- acousto-optic tunable filter (AOTF)-based instruments tion process as dinterfering parametersT, as will be provide full spectral coverage. Selection of the discussed in Section 3. More detailed information on appropriate technology is usually based upon the the theory of absorption and scatter effects in diffuse required analyte sensitivity, reliability, ease of use, reflectance and transmittance NIR spectroscopy can calibration transferability and implementation needs.
be found elsewhere The latter aspect requires laboratory and processanalyzers to be differentiated.
2.2. Instrumentation and sample presentation Laboratory analyzers are intended for off-line or at-line measurements in quality control, research and A NIR spectrometer is generally composed of a plant laboratories, i.e. high analyte sensitivity and light source, a monochromator, a sample holder or a reliability are required, while speed is of lower sample presentation interface, and a detector, allowing importance. Optimum sample presentation to the for transmittance or reflectance measurements ( instrument, high signal-to-noise ratio, instrument The light source is usually a tungsten halogen stability, and sufficient resolution are the most lamp, since it is small and rugged Detector types important aspects for analysis. Presently, grating and include silicon, lead sulfide (PbS) and indium gallium interferometer-based instruments are mainly in use for arsenide (InGaAs) Silicon detectors are fast, low- this purpose. The appropriate NIR measuring mode noise, small and highly sensitive from the visible will be dictated by the optical properties of the region to 1100 nm. PbS detectors are slower, but very samples (Transparent materials are usually popular since they are sensitive from 1100 to 2500 nm measured in transmittance (Turbid liquids or and provide good signal-to-noise properties. The most semi-solids and solids may be measured in diffuse expensive InGaAs detector combines the speed and transmittance (diffuse reflectance ( size characteristics of the silicon detector with the or transflectance (E), depending on their wavelength range of the PbS detector.
absorption and scattering characteristics. In any case, A number of optical configurations exist that can absorbance (A) values relative to a standard reference be used to separate the polychromatic NIR spectral material are measured, with A corresponding to log 1/ region into dmonochromaticT frequencies. A detailed R and log 1/T for reflectance and transmittance description of the different principles can be found in various textbooks Here the basic principles and To measure good NIR spectra, the proper sample main differences will be shortly discussed from a presentation is of utmost importance, especially when practical point of view. Broadband, discrete filter measuring solid samples, since scatter effects and Diffuse Reflectance Fig. 1. Basic NIR spectrometer configurations.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 opaque liquids and solids is typically carried out bycontact transmission and reflectance probes, whilenon-contact reflectance measurements are performedon materials transported in hoppers or conveyor belts.
3. Theory and practice of chemometric dataprocessing Since NIR spectra are typically composed of broad overlapping and, thus, ill-defined absorption bandscontaining chemical and physical information of allsample components, the analytical information is multivariate in nature and, therefore, hardly selective.
To perform qualitative or quantitative NIR analysis,i.e. to relate spectral variables to properties of theanalyte, mathematical and statistical methods (i.e.
chemometrics) are required that extract brelevantQ Fig. 2. NIR measuring modes—(A/B) transmittance, (C) diffuse information and reduce birrelevantQ information, i.e.
reflectance and (D/E) transflectance.
In the following sections, the most frequently used mathematical data pretreatments and their specific stray light induced by variations in packing density of purpose (Section 3.1), reduction of variables with powders or sample positioning of tablets or capsules principal component analysis (Section 3.2), multi- may cause large sources of error in the spectra variate calibration methods for quantitative analysis Therefore, several types of sample cells, such as (Section 3.3), and multivariate classification techni- quartz cuvettes with defined optical path length for ques for qualitative analysis (Section 3.4) will be liquids, specifically designed sample cells with quartz discussed. Different methods for calibration transfer windows for semi-solids and powders, and adjusted between instruments, an important economic and sample holders for tablets and capsules have been regulatory issue for qualitative and quantitative developed Temperature control and sample pharmaceutical NIR analysis, have recently been movement are other options that have been realized.
commented on by Blanco et al. and will, thus, Process analyzers are intended for in-line or on- not be considered here in detail.
line measurements to provide real-time processinformation while operating in harsh conditions. This 3.1. Data pretreatments requires fast and rugged instruments with no movingparts, such as AOTF-based instruments, allowing for Interfering spectral parameters, such as light numerous readings per second without being sensitive scattering, path length variations and random noise, to vibrations. AOTF-based instruments choose wave- resulting from variable physical sample properties or lengths by using radio-frequency signals to alter the instrumental effects, call for mathematical corrections, refractive index of a birefringent crystal (usually so-called data pretreatments, prior to multivariate TeO2). Wavelength scans can, thus, be performed modeling in order to reduce, eliminate or standardize much more rapidly than with other configurations.
their impact on the spectra. Since careful selection of Since process analyzers are dedicated to performing a data pretreatments can significantly improve the particular analysis on a specific sample type, the robustness of a calibration model, the most commonly process sample interface depends on the sample type used methods are briefly discussed with respect to the and the process conditions, with NIR light being effect they are able to correct. A detailed description transferred via fiber optics. In-line analysis of clear to of the techniques can be found elsewhere G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 Mathematical treatments used to compensate for method is principal component analysis (PCA). PCA scatter-induced baseline offsets include multiplica- is a mathematical procedure that resolves the spectral tive scatter correction (MSC) and standard normal data into orthogonal components whose linear combi- variate (SNV). Both methods have originally been nations approximate the original data. The new developed to process reflectance spectra, but they variables, called principal components (PC), eigen- are also applied to transmittance spectra. Baseline vectors or factors, correspond to the largest eigenval- shifts and intensity differences resulting from vari- ues of the covariance matrix, thus, accounting for the able positioning or path length variations may be largest possible variance in the data set. The first PC reduced or eliminated by normalization algorithms.
represents maximum variance amongst all linear Derivatives can be applied to improve the resolution combinations and each successive variable accounts of overlapping bands. In addition, they are able to for as much of the remaining variability as possible.
reduce baseline offsets. Since spectral noise is also The transformation procedure is visualized schemati- amplified by derivation, derivatives are usually cally in on the basis of three original variables, combined with Taylor or Savitzky Golay smoothing i.e. three wavelengths per spectrum. For real spectra with p wavelengths the transformation leads to a p-dimensional space.
3.2. Reduction of variables by principal component In pharmaceutical NIR analysis, it is often possible to compress most of the spectral variability to only afew principal components, i.e. factors with only a Since multivariate NIR spectral data contain a huge rather small loss of information. A number of multi- number of correlated variables (= collinearity), there is variate calibration and classification methods, there- a need for reduction of variables, i.e. to describe data fore, rely on PCA data (see Sections 3.3 and 3.4). For variability by a few uncorrelated variables containing further details on PCA, interested readers are referred the relevant information for calibration modeling. The to the excellent and comprehensive treatise of Howard best known and most widely used variable-reduction Fig. 3. Transformation of a spectrum with three variables, i.e. wavelengths (a) to a new coordinate system with one axis for each wavelengththereby converting the spectrum to a single point in a three-dimensional space (b), cloud formation of several spectra (c), mean centering (d),and determination of principal components F1, F2 and F3 (e).
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 3.3. Multivariate calibration for quantitative analysis 3.4. Multivariate classification for qualitative analysis Before a NIR spectrometer can do any quantitative In qualitative analysis, sample properties that have analysis, it has to be trained, i.e. calibrated using to be related to spectral variations have discrete values multivariate methods. The calibration process basi- that represent a product identity or a product quality, cally involves the following steps: for example bgoodQ or bbadQ. To solve the selectivityand interference problems of NIR spectra, multivariate 1. Selection of a representative calibration sample set.
classification methods are used for grouping samples 2. Spectra acquisition and determination of reference with similar characteristics. Multivariate classification methods, also known as pattern-recognition methods, 3. Multivariate modeling to relate the bspectral var- are subdivided in bsupervisedQ and bnon-supervisedQ iationsQ to the breference valuesQ of the analytical learning algorithms, depending on whether or not the target property.
class to which the samples belong is known.
4. Validation of the model by cross validation, set bNon-supervisedQ methods, also known as cluster validation or external validation.
analysis, do not require any a priori knowledgeabout the group structure in the data, but instead The multivariate regression methods most fre- produces the grouping, i.e. clustering, itself. This quently used in quantitative NIR analysis are principal type of analysis is often very useful at an early stage component regression (PCR) and partial least-squares of an investigation to explore subpopulations in a (PLS) regression PCR uses the principal compo- data set, for instance different physical grades of a nents provided by PCA (see Section 3.2) to perform material. Cluster analysis can be performed with regression on the sample property to be predicted.
simple visual techniques, such as PCA (see Section PLS finds the directions of greatest variability by 3.2) or some hierarchical methods leading to so-called comparing both spectral and target property informa- tion with the new axes, called PLS components or bSupervised classificationQ methods, also known as PLS factors. Thus, the main difference between the discriminant analysis, are used to build classification two methods is that the first principal component or rules for a number of pre-specified subgroups, i.e. the factor in PCR represents the largest variations in the group structure of the training set is known. The spectrum, whereas in PLS it represents the most classification rules are later used for allocating new or relevant variations showing the best correlation with unknown samples to the most probable subgroup.
the target property values. In both cases, the optimum Identity or good/bad quality are, thus, defined as number of factors used to build the calibration model belonging to a group with known properties. Algo- depends on the sample properties and the analytical rithms of this type such as LDA (= linear discriminant target. Too many factors may lead to an boverfittedQ analysis), QDA (= quadratic discriminant analysis), model with a high regression coefficient and a low SIMCA (= Soft Independent Modelling of Class standard error of calibration (SEC), but a large Analogies) or KNN (= K nearest neighbours) are standard error of prediction (SEP). Such a model is typically used for constructing spectral libraries.
not very robust and may fail when tested with an Most of the classification methods can operate independent validation set.
either in wavelength space or in a dimension-reduced In some cases, the spectral data and the target factor space. In any case, their ultimate goal is to property may not be linearly related as a result of establish mathematical criteria for parametrizing physical sample properties or instrumental effects.
spectral similarity, thus, allowing similarity between These cases can only be addressed by non-linear samples or a sample and a class to be expressed calibration methods, such as PLS-2, locally weighted quantitatively. For this purpose, comprehensive libra- regression (LWR) or artificial neural networks ries of spectra that represent the natural variation of (ANNs). For details on these methods interested each product have to be constructed in a bcalibrationQ readers are referred to the corresponding chapters in process, with similarity being expressed by either a a recent textbook on multivariate calibration correlation coefficient, such as the spectral match G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 value (SMV) or a distance measure, such as dress the suitability of NIR instrumentation for use in Euclidian or Mahalanobis distance.
pharmaceutical analysis focussing mainly on opera- A detailed description of the different classification tional qualification and performance verification com- procedures is certainly beyond the scope of this paper.
prising wavelength scale and repeatibility, response Interested readers are, therefore, referred to a recent repeatibility, photometric linearity, and photometric textbook on the topic Worth mentioning here are noise. Only some limited guidance is provided in terms the following practical aspects: of developing and validating an application.
The general legal requirements for instrumentation ! The correlation coefficient, being defined as the qualification procedures, namely design qualification cosine of the angle between vectors for the sample (DQ), installation qualification (IQ), operational qual- spectrum and the average spectrum for each ification (OQ), and performance qualification (PQ), product in the library, is a rather robust parameter are described in the cGMP guideline title 21 CFR part that can be favorably used for chemical identity 211. For practical realization of these requirements, testing (see Section 5.1), since it relies on second the American Society for Testing and Materials derivative spectra and is, thus, not influenced by (ASTM) has provided NIR specific directions regard- spectral offsets and globalintensity variations ing appropriate methodology for establishing spec- resulting from physical differences or concentra- trophotometer performance tests including suitable tion changes.
standards and multivariate calibration Further ! Distance-based methods, on the other hand, also guidance for evaluation of a NIR spectrophotometer allow for product qualification. The conformity has been provided in a special report of the Analytical index (CI), based on the wavelength distance Methods Committee of the British Royal Society of method, is one such parameter that has been used successfully to pinpoint quality differences in raw Many pharmaceutical companies have success- materials and products by using a so-called C-plot, fully implemented NIR spectrometers in their i.e. a plot of the absolute distance at each wave- quality control laboratories for routine use in raw length as a function of the wavelength (see material identification and qualification. This is also Section 5.1).
based on the fact that major pharmacopoeias allowmanufacturers to use analytical methods other thancompendial ones for compliance testing, provided 4. Regulatory aspects they are validated according to parameters, such asspecificity, linearity, range, accuracy, precision, 4.1. Actual status of pharmaceutical NIR analysis repeatibility, reproducibility, detection limit, quanti-fication limit, and robustness, as is detailed in the NIR spectroscopy has a large number of advan- U.S.P. Chapter 1225 on Validation of Compendial tages over other analytical techniques, and, thus, Methods and the general ICH Guidelines Q2A offers many interesting perspectives in pharmaceutical and Q2B on Validation of Analytical Procedures analysis. The scientific rationale of this technology has been established for many different applications Interestingly, only few quantitative NIR methods and justified by a huge number of publications from have gained regulatory approval as yet. The main academia and industry (see Section 5). However, in reason for this is that bnon-separativeQ multivariate the highly regulated pharmaceutical world, an ana- NIR methods differ markedly from bseparativeQ uni- lytical method is only valuable for routine implemen- variate chromatographic methods for which U.S.P.
tation if it is approved by regulatory authorities.
Chapter 1225 and the general ICH Guidelines Q2A Actually, the major pharmacopoeias have generally and Q2B were written. Moffat et al. discussed adopted NIR techniques. The European and these aspects extensively in an excellent paper United States Pharmacopoeia both contain a published in 2000. Based on the example of a general chapter on near-infrared spectrometry and quantitative NIR method for the analysis of para- spectrophotometry, respectively. These chapters ad- cetamol in tablets, the authors made suggestions on G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 how NIR assays can best meet the ICH Guidelines on According to a recently published U.S.F.D.A.
Validation. The recently published Guidelines for the Guidance for Industry PATs are defined as Development and Validation of Near-Infrared Spectro- systems for real-time monitoring and control of scopic Methods in the Pharmaceutical Industry critical process parameters and material performance established by the NIR sub-group of the UK Pharma- attributes, thus, helping to improve process under- ceutical Analytical Sciences Group (PASG), cover the standing, manufacturing cycle time, and final prod- unique and specific NIR requirements whilst remain- uct quality. NIR spectroscopy and imaging may be ing complementary to ICH Q2A and Q2B, which one of the major PAT tools, since these techniques address traditional method validation requirements. It are well-suited for at-line, in-line and on-line might be expected that the PASG guidelines, compris- measurements. They can provide a wealth of ing hardware as well as software aspects, can help both chemical and physical information important for pharmaceutical industry and regulatory agencies in measuring process performance and open up oppor- evaluating future submissions of qualitative and tunities to move forward from traditional quality quantitative NIR methods. For details of the PASG control concepts to process qualification and product conformity testing. Although a number of challengesconcerning hardware design and regulatory approval 4.2. NIR spectroscopy in view of the U.S.F.D.A.
must be overcome to realize the full potential of NIR initiative on PAT spectroscopy and imaging as PAT tools, it may beexpected that parametric or even real-time release The production of pharmaceutical dosage forms is concepts may be well assisted by the use of NIR usually a multistage operation, consisting of several techniques (see Sections 5.3 and 6.3).
validated processes managed by standard operatingprocedures (SOPs). Quality assurance, includingdecisions concerning the satisfactory completion of 5. Pharmaceutical applications each unit operation, is actually based on off-linetesting to document quality of a small, nominally NIR spectroscopy combined with multivariate random product sample. This approach is often very data analysis opens many interesting perspectives time consuming and adds significantly to the manu- in pharmaceutical analysis, both qualitatively and facturing cycle time, since it requires the process to be quantitatively. Fast and nondestructive NIR measure- stopped during sample removal, data generation and ments without any sample pre-treatments may documentation. In addition, it does not assure zero increase the analytical throughput tremendously.
defect product quality, since risk assessment and risk The use of fiber optic probes offers the opportunity management are not included, e.g. critical process for in-line and on-line process monitoring. The parameters and material performance attributes may special feature of combined chemical and physical not be identified.
information allows for the assessment of a bspectral In view of this undesirable situation for industry signatureQ of raw materials, intermediates and final and public health, it has been recognized that new dosage forms, which in turn offers the possibility of testing paradigms are required to succeed in both, an a simultaneous determination of several sample increase in manufacturing efficiency and product safety. The Process Analytical Technology (PAT) Notwithstanding these advantages, pharmaceutical initiative, driven by the United States Food and industry and regulatory bodies have been slow to Drug Administration (U.S.F.D.A.) and major phar- adopt the NIR technique, most probably since it maceutical companies, is a challenging approach lacks the ability of mid-IR to identify samples by intended to assist the progression of real-time or mere inspection of spectra and involves calibration parametric release and quality-by-design concepts by sophisticated mathematical techniques (see Sec- by providing an opportunity to move from the tion 3). Although the earliest publications on phar- laboratory-based btesting to document quality para- maceutical NIR applications date back to the late digmQ to a bcontinuous quality assurance paradigmQ.
1960s, it was not until the last 20 years that NIR G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 spectroscopy has gained increasing interest in the The methods are time-consuming, as they are usually pharmaceutical industry with the real breakthrough in performed in an off-line laboratory, are often wet- the 1990s as a result of hardware and software chemical in nature, and are, therefore, not appropriate improvements. Within the last 10 years a growing to handle the enormous number of analyses of modern number of research and review articles have reported industrial material identification and qualification on the great potential of NIR spectroscopy in pharmaceutical research, production, and quality With the pharmacopoeial-based authorization to control focussing on various banalytical targetsQ, such use methods other than the compendial ones for as identity, content uniformity, moisture content, compliance testing and the GMP-based opportunity particle size, polymorphic and pseudopolymorphic of using bany appropriate procedure or measure to forms, hardness, thermal and biopharmaceutical prop- assure the identity of the contents of each container erties. These different aspects, resulting from the dual of starting materialsQ, it has been possible to take dependence of the NIR signal on chemical and advantage of multi-sensing NIR techniques based on physical sample characteristics, will be discussed in fiber optic probes for fast and nondestructive the context of raw material and intermediate identi- pharmaceutical raw material identification and qual- fication and qualification (Section 5.1), analysis of ification. Many papers have reported on the feasi- intact dosage forms (Section 5.2), and process bility of NIR identification and qualification of both monitoring (Section 5.3), with a main focus on solid active ingredients and excipients and most dosage forms.
companies have adopted some form of NIR materialtesting in their supply chain, either in the warehouse 5.1. Identification and qualification of raw materials only and/or elsewhere in a manufacturing operation, and intermediates i.e. wherever rapid assessment of identity and qualityis needed. In combination with bar-code readers, Raw materials intended for use in pharmaceutical weighing stations, and electronic batch documenta- products, i.e. active ingredients and excipients, are tion a bsmartQ system can be developed that subject to pharmaceutical quality requirements as guarantees successful manufacturing operations by prescribed by Good Manufacturing Practice (GMP) ensuring that the correct materials of the appropriate Guidelines for Medicinal Products, and pharmaco- quality are used in the manufacturing process (see poeial monographs. To guarantee maximal product also Sections 4.2 and 5.3).
safety, the GMP guidelines require special testing Using NIR techniques, the chemical identity of a procedures within the material supply chain (Directive particular material is usually confirmed with a spectral 91/355/EEC, Chapter 5.30). In addition to the routine library approach. If an appropriate library has been release testing of the substance, single container constructed, the combined chemical and physical identification has to be performed for any lot of raw information in the spectra can also be used for material material at any time of dispensal.
qualification. Moreover, with an appropriate calibra- Since modern pharmaceutical processes rely heav- tion setup, simultaneous quantitative measurements, ily on a reproducible source and grade of raw such as moisture content and particle size determi- materials to ensure consistent finished product quality, nations, can be performed or bconformityQ approaches material qualification is another analytical require- can be used to predict material performance in ment in the supply chain that has to be fulfilled.
manufacturing processes. The different approaches Qualification is supposed to confirm the grade and/or will be discussed in the following paragraphs.
source of materials including physical properties, suchas particle size, density, morphology etc., which may 5.1.1. Library approach in turn indicate its suitability for the intended use.
Chemical identification usually does not involve Traditionally, pharmaceutical raw material identifica- any conceptual problems with respect to spectral tion and qualification, known as compliance testing, library development However, exten- has been based on compendial methods and/or sion of the identification concept to material qual- alternative validated in-house testing procedures.
ification is usually more complex. The key parameters G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 for constructing a robust spectral library may, there- derivative data. For qualification of different grades of fore, be defined as follows: excipients, more sophisticated algorithms, such asSIMCA are recommended (see Section 3.4). Only 1. Definition of library scope and purpose.
recently, Kemper and Luchetta have published a 2. Selection of authentic sample spectra for calibra- comprehensive paper giving practical guidelines for tion, internal and external validation.
construction, validation and maintenance of spectral 3. Rationale of data pretreatments.
libraries for raw material identification and qualifica- 4. Selection of classification algorithm(s).
5. Determination of thresholds.
6. Maintenance and updating.
5.1.2. Conformity approach In the early 1990s, van der Vlies and co-workers The library structure may depend on the software developed a discriminating method, which limitations and the user's requirements. In the they called the bconformityQ approach, and introduced simplest case, all materials are incorporated into a new quality parameter, the Conformity Index (CI), one library . Alternatively, they may be split into to replace compendial methods for identification, sub-libraries to ensure the required level of specific- assay, and moisture content determination of ampi- ity, as for discrimination of chemically similar cillin trihydrate. It is worth mentioning that this was substances, such as close members of a homologous the first NIR method for release testing of a bulk series or different grades of microcrystalline cellulose pharmaceutical product for human consumption approved by the U.S.F.D.A.
The selection of samples is critical to the success of The CI is the largest value obtained by dividing the the application. Two sets of samples are required: one absolute difference in absorption between sample and for the construction of the library and an independent reference spectrum (first or second derivative) for one for external validation purposes to verify the each data point by the standard deviation of the performance of the data base. The number of batches absorbance of the reference spectrum at that particular required to train the system depends on the intended data point. The authors defined the bstandard qualityQ, scope, i.e. the required discriminatory power of the i.e. the specification of their material at CI of 5 or method. The training set must collectively describe lower, and achieved a high sensitivity of CI for the typical variation of the substance being analyzed.
chemical and physical deviations. With the so-called As a rule of thumb, identification normally requires a Conformity Plot (C-Plot: CI versus wavelength plot) it much smaller number of different batches (usually 3) was possible to pinpoint the sources of even very than qualification (usually 20 or more).
slight variations in chemical and physical properties, Data pretreatments (see also Section 3.1) strongly including crystallinity. The conformity approach is depend on the application. For identification purposes, well suited for industrial raw material and intermedi- second derivative and scatter correction are often used ate qualification, since it gives qualitative answers to to reduce offsets, due to variable physical material quantitative questions without the need of exhaustive characteristics. The rationale of transforms in qual- calibration work.
ification methods strongly depends on the parameterof interest and is a case by case decision. The effect of 5.1.3. Quantitative calibration models NIR data pre-processing on the pattern recognition of Quantitative calibration models in raw material pharmaceutical excipients has been discussed by qualification have been described for analytical targets, such as moisture content particle The classification model (see also Section 3.4) is size specific surface area polymor- the heart of the library. The proper choice of the phic and pseudopolymorphic forms amor- algorithm depends on the scope of the library. For phous/crystalline ratios viscosity and identification purposes, where physical parameters are gel strength Moisture content, particle size and not determined, it is usually sufficient to use a match polymorphism, also relevant to pharmaceutical inter- by wavelength correlation method based on second mediates, will be discussed in more detail.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 Since chemical, physical, technological and bio- potential of NIR spectroscopy for particle size pharmaceutical properties of active ingredients and determination has been alluded to in many review excipients may be largely affected by their water articles, only a few research papers have been content and the type of water present, evaluation of dedicated to this subject. Mean particle size batch-to-batch variability or storage effects on water or particle size distribution measurements content and water binding is usually an integral part of with NIR spectroscopy have been reported, using material qualification. NIRS is an effective alternative lactose monohydrate microcrystalline cel- to traditional methods, such as thermogravimetry and lulose NaCl, and sorbitol aspirin, Karl Fischer titration for both water content and water caffeine and paracetamol and piracetam as binding determinations. This is due to the fact that model excipients and active ingredients, respectively.
O–H bands of water are very intensive in the NIR Various chemometric approaches have been sug- region, exhibiting five absorption maxima (at 760, gested for correlating particle size with NIR spectral 970, 1190, 1450, 1940 nm), the positioning of which information and the literature data clearly reveal that depends on the hydrogen bonding intensity. The there is more than one way to model mean particle specific band to be used for water determinations size data with NIR spectra, depending on the particle depends on the desired sensitivity and selectivity level.
size range, shape of the particle size distribution, NIR quantification of moisture content is usually an materials refractive index, and absorption properties.
easy task with respect to data processing, i.e. MLR and Ciurczak et al. found an inverse relationship PLSR models have been reported. Moreover, reference between absorbance at each wavelength and mean data provided by Karl Fischer titration are reliable. It particle size, with two distinct segments below and is, therefore, not surprising that NIR moisture content above 85 Am, indicating the complicating effect of determinations in both transmittance and reflectance small particles for quantitative NIR mean particle size mode have been described extensively in the literature.
measurements. Burger and coworkers have investi- Most of the early work has been summarized and gated this aspect in detail and the interested reader is discussed by Blanco Two papers are worth referred to some excellent papers of the group dealing mentioning here, since they demonstrate the potential with radiative transfer investigations to quantify of NIRS to distinguish different states of water in raw absorption and scattering coefficients of pharmaceut- materials and intermediates. Ciurczak and coworkers ical powders From a more practical point of were among the first who demonstrated the view, Blanco et al. revealed that spectral opportunity of NIRS to differentiate between total, reproducibility was affected by sample compactness bound, and surface bulk water in pharmaceutical raw and varied in an exponential manner with particle size materials, thus, demonstrating the advantage of NIRS (in the range 175–325 Am), thus, pointing to the over traditional methods, such as KFT and LOD. Dziki importance of sample presentation for quantitative et al. detected differences in the location or particle size measurements.
orientation of the water molecules within the crystal Pharmaceutical raw materials may exist in amor- lattice of sarafloxacin with NIRS and used this phous or crystalline form, with polymorphism and approach to distinguish between acceptable and pseudopolymorphism being widely observed in crys- unacceptable batches for formulation purposes.
talline compounds. The impact of a certain poly- Mean particle size and particle size distribution of morphic or pseudopolymorphic form or the degree of solid raw materials and intermediates are key issues in crystallinity on the physicochemical and biopharma- the formulation of many pharmaceutical products, ceutical material characteristics is well known. NIR since they have a profound effect on bulk physical spectroscopy has been reported to be an alternative to properties, which in turn influence blending and flow traditional techniques, such as DSC and X-ray powder characteristics, density, compressibility, and dissolu- diffraction, for qualification and quantification of the tion rate. Particle size measurements with NIRS in crystallinity of miokamycin, lactose mono- diffuse reflectance mode rely on the particle size- hydrate, mannitol, sucrose and raffinose; of polymor- dependent scatter effect of powders resulting in non- phic or pseudopolymorphic forms of sulfathiazol, linearly sloping baselines Although the caffeine and theophylline in bulk and of G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 crystallinity upon hydration during granulation pro- ing mode is not as critical as with quantitative cesses The rationale behind this approach is applications, except for very thick, highly absorbing the sensitivity of NIR spectra to intermolecular tablets and sugar-coated tablets, for which the bondings. The magnitude of spectral differences reflectance mode is recommended to overcome between the different forms is, therefore, the key problems of low analyte signal intensity or even total issue for quantitative determinations. Patel et al. absorption in transmittance. Challenges associated demonstrated in a recent paper that NIRS can be used with the identification of placebo and verum tablets to determine polymorphs of sulfathiazol in binary of different dosage levels (2, 5, 10 and 20% w/w) mixtures in the range of 0.3% w/w. For amorphous/ within the blister packaging have been reported by crystalline mixtures of lactose monohydrate, the Dempster et al. The results of this study clearly amorphous content was accurately determined to revealed a higher discriminating ability of direct within 1% w/w. The literature data clearly reveal that measurements compared to measurements through NIR results are comparable with other techniques, the blister packaging, thus, emphasizing that the effect thus, reflecting the potential of the method for the of the packaging material on the accuracy of NIR assessment of different physical forms in bulk identification approaches may not be neglected.
materials and intermediates.
Quantitative NIR analysis of active ingredients in tablets has been widely reported and reviewed in the 5.2. Analysis of intact dosage forms literature. However, in the earliest NIR assays, tabletswere not analysed intact. The active was extracted The nondestructive and multivariate nature of NIR from the matrix or the tablets were at least pulverized techniques opens new perspectives in the pharma- prior to NIR measurements. The opportunity to ceutical analysis of intact dosage forms, including accurately measure active contents in whole tablets chemical, physical and related biopharmaceutical started in the late 1980s with the development and aspects. This section will discuss NIR applications subsequent commercialization of appropriate sample for the characterization of solid dosage forms, namely holders that allow for a proper fit of even curved tablets, capsules, lyophilized products and implants.
tablets, thereby reducing variable positioning andstray light effects. Within the last 10 years, the number of publications describing quantitative NIR measure- Most of the literature data available on NIR ments of active ingredients in intact tablets has applications for intact dosage forms focus on tablets, increased tremendously Various aspects ranging from identification and assay to physical and have been addressed, two of which will be discussed biopharmaceutical parameters, such as hardness, coat- in more detail, namely the rationale for selecting the ing thickness and dissolution rate. It is certainly appropriate measuring mode, and the practical and beyond the scope of this paper to review all the regulatory aspects to be considered in choosing the published data in these fields. This section is rather appropriate chemometric approach, including calibra- intended to provide an update of and comment on tion sample selection and data pretreatments.
some specific aspects that have not been reviewed in Selecting the measuring mode for NIR tablet detail yet. Special attention will be paid to the analysis strongly depends on tablet thickness, compo- importance of sample selection, sample presentation sition and target parameter. Considering quantitative and collection of reliable reference data for develop- analysis of active ingredients in tablets, the reflectance ing robust calibration models. Readers interested in a mode, mainly used in early work, may have some more comprehensive coverage of the topics including limitations, since it covers only a certain part of the earlier data are referred to selected review articles tablet This, in turn, can cause false results, if and a recent book chapter homogeneity within the tablet cannot be assured or is Fast and nondestructive identification of active part of the delivery concept, such as in multilayer ingredients and exipients in whole tablets, even tablets. Moreover, the assay of coated tablets may be through the blister packaging, is certainly a domain complicated in cases where the majority of spectral of NIR spectroscopy Generally, the measur- information is coming from the coating polymer. In G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 view of this, regulators have expressed their concerns it is surprising that stability issues, i.e. identification regarding reflectance measurements for content uni- and quantification of degradation products in tablets, formity testing. Transmittance spectra, representing a have only rarely been addressed. There is merely one larger volume of the scanned tablet, certainly provide early paper by Drennen and Lodder that reports a better description of a tablet matrix in bulk.
the use of NIR diffuse reflectance spectroscopy for Improved accuracy, precision, and sensitivity of monitoring the hydrolysis of acetylsalicylic acid to transmittance measurements in various tablet assays salicylic acid in tablets upon water absorption. Due to have been demonstrated in the literature the combined spectral information on water and However, it should not be neglected that a signifi- salicylic acid, the authors were able to predict both cantly narrower wavelength range is available in parameters from one single measurement, thus, bdiffuseQ transmittance mode, and limitations are emphasizing the great potential of NIRS for tablet observed with very thick tablets Recent papers stability testing. In addition to chemical stability, dealing with NIR tablet assays for content uniformity polymorphic transitions might be another target testing, therefore, clearly reveal that selection of the parameter that could be addressed in tablets appropriate measuring mode is a case by case decision The mechanical performance of tablets is of importance for bulk handling, coating, packaging, As a non-separative method, quantitative NIR removal from blister, and disintegration. Current measurements on tablets rely heavily on chemometric methods of hardness testing are destructive in nature procedures for data modelling, with sample selection and often subject to operator error. NIR spectroscopy, and data pretreatments being the most critical issues on the other hand, offers the opportunity for fast and regarding calibration development. Since process- nondestructive hardness measurements, and provides related natural variations in tablet mass and hardness additional information on structural features of the affect the optical properties and, thus, the baseline of tablet matrix. Several groups have described the the recorded spectra, derivative transformation and/or application of NIRS as an alternative method for normalization are usually required for accurate NIR tablet hardness testing Since the approaches content uniformity measurements. Sample selection are different with respect to the measuring mode, the for calibration modelling strongly depends on the range of hardness levels included in the model, and chemometric approach. For bconformityQ testing, the the chemometric data processing, they will be calibration samples should bsimplyQ cover the normal discussed in more detail.
range of tablet variability, including intra-batch and Drennen and co-workers were among the batch-to-batch variability. Out-of-specification sam- first who applied NIR spectroscopy to tablet hardness ples should be considered in the validation step. For testing. The authors used diffuse reflectance spectro- quantitative modelling, additional requirements have scopy and realized that an increase in tablet hardness to be fulfilled, namely the use of tablets with an causes a bprimaryQ effect of wavelength-dependent extended range of active concentrations in the nonlinear baseline shifting to higher absorbance calibration step. This is not an easy task in industrial values, which can be attributed to a decrease in practice since normal tablet production batches multiplicative light scattering. Various tablet formula- are manufactured with tight tolerances. In an excellent tions, including coated tablets, were investigated at and comprehensive paper, Moffat and co-workers hardness levels ranging from 1 to 7 kp and from 6 have discussed this issue and given various options to 12 kp respectively. A pressure-dependent for proper calibration sample selection In the bsecondaryQ spectral effect, namely a peak shifting at same paper, the authors provided suggestions on how higher hardness levels arising from changes in to meet the ICH Guidelines on Validation for NIR intermolecular bonding, could be observed for some quantitative analysis of active ingredients in tablets materials. In view of these observations, the authors (also see Section 4.1). Validation of quantitative NIR used different approaches for different hardness methods has also been addressed by Blanco levels to correlate spectral data with hardness values.
Considering the huge amount of literature data on For hardness values in the range of 6 to 12 kp, they NIR assays for active qualification and quantification, used PCA/PCR based models, considering mainly G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 bsecondaryQ spectral effects, while removing baseline Prediction of drug dissolution rates from whole shifts also resulting from tablet positioning variability tablet NIR spectra is another application that has been The SEP values obtained were as precise as the alluded to in many review articles. However, only a laboratory hardness test. For hardness values in the few research papers are really concerned with this range of 1 to 7 kp, where the bprimaryQ spectral effect topic, probably due to the challenge of providing was mainly observed, the authors developed a spectral tablet samples that cover the appropriate range of best-fit algorithm based on traditional statistical variability required to develop robust calibration methods The proposed approach exploits the models. The first papers, dating back to the early baseline shift and involves the determination of a best- 1990s deal with the prediction of the fit line through each spectrum, thereby reducing the dissolution rate of carbamazepine tablets following spectrum to slope and intercept values, e.g. de- exposure to high humidity. NIR diffuse reflectance weighting individual absorbance peaks and valleys.
spectra were collected periodically from whole tablets The method was found to be insensitive to slight stored in a hydrator. Dissolution rates were correlated formulation changes (1–10% w/w cimetidine) and with the spectral data using PCR and the bootstrap compared favorably to the multivariate PCA/PCR (BEST) algorithm for modelling. Although this method with SEP values of around 0.5 kp.
example clearly indicates the potential of NIRS for Morisseau and Rhodes revealed SEP values in nondestructive dissolution testing, its citation in the same range (0.3–0.6 kg) for different tablet review articles is somewhat misleading, since in this formulations, namely hydrochlorothiazide (15 and special example the most prominent parameter affect- 20% w/w) and chlorpheniramine (2 and 6% w/w) in ing dissolution rate was the moisture content. Quanti- a matrix of microcrystalline cellulose and magnesium tative modelling of drug dissolution rates of stearate, at six hardness levels ranging between 2 and commercialized tablets stored under normal condi- 12 kg. The authors used MLR and PLS to model the tions is certainly a greater challenge and requires diffuse reflectance spectra. Obviously, due to the wide exhaustive calibration work based on a priori knowl- range of hardness levels included in the calibration edge of the formulation- and process-dependent tablet model, it was not possible to develop acceptable variables, as well as their effect on both the drug bmixedQ calibrations by combining data from two dissolution profile and the spectra. A qualitative concentrations of the same drug. In a recent paper, bconformityQ approach (see Section 3.4) might be a Chen et al. described the favorable use of more practical option for modelling drug dissolution artificial neural networks (ANN) to predict tablet from fast dissolving tablets.
hardness from diffuse reflectance NIR spectral data.
Some authors have examined the Interestingly, there is only one paper that describes opportunity of predicting the drug dissolution profile the use of NIR transmittance measurements for tablet of tablets with a rate-controlling film coat from whole hardness determinations Based on the fact that tablet NIR spectra. Kirsch and Drennen used compaction of pharmaceutical powders results in theophylline tablets coated with various amounts of density variations in different directions and regions ethylcellulose and collected the spectra in diffuse of the tablet the author suggests a better reflectance mode. Reich and co-workers predictability of whole tablet hardness values from used a transmittance configuration to collect spectra transmittance than from reflectance measurements from Eudragit RL-coated theophylline tablets. In both Indeed, the data revealed a strong correlation cases, reliable quantitative calibration models could between tablet hardness and transmission spectra over be developed to predict the time required for 50% of a wide range of hardness levels (10–180 N). In the theophylline to be released. The rationale behind addition, material specific bprimaryQ and bsecondaryQ these approaches is the effect of film coat thickness spectral effects were used to study the consolidation and film coat uniformity on both drug dissolution rate characteristics of different pharmaceutical excipients and NIR spectra. It is, therefore, not surprising that the and active ingredients indicating the potential of same authors used NIR diffuse reflectance and trans- NIR transmittance applications in tablet formulation mission spectroscopy to predict film coat thickness and even film coat uniformity on G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 tablets. SEP values for the determination of film coat monitoring, for NIR measurements on intact hard thickness were comparable for transmission and capsules. The sources of variance in NIR measure- diffuse reflectance mode. However, reliable reference ments on hard capsules, being more pronounced than data were difficult to achieve and were, thus, the with tablets, has been stressed in detail by Candolfi et major source of error in the quantitative models.
al. Positioning and time of measurement were Prediction of film coat uniformity and related gastro- found to be the most important sources of variance.
resistance with a conformity approach provided much Positioning effects were attributed to the loose and better results and required less calibration work movable filling and the round, smooth, and brilliant This indeed emphasizes again that bnon-calibratingQ shell, which affected the reflection angles. The time qualitative chemometric techniques combined with factor expresses the effect of surrounding conditions, NIRS are valuable tools to answer quantitative such as temperature and relative humidity, on the sample properties, by inducing small changes in thewater content of the gelatin shell.
Taking these aspects into consideration, it is not Besides tablets, capsules are among the most surprising that only a few papers mainly focussing on prominent solid dosage forms. Since hard and soft empty capsule shell properties have been published.
capsules differ with respect to manufacturing technol- Buice et al. and Berntsson et al. described ogy and formulation, i.e. shell and fill composition, NIR moisture determinations of empty capsule shells which in turn may affect analytical target parameters using reflectance measurements with a filter and a and NIR measurements, they will be discussed grating-based instrument, respectively. Buice et al.
used the time-dependent weight gain upon water Hard capsules are a rather versatile dosage form uptake of the transparent capsule shells in a hydrator that can be filled with a variety of formulations, such at 100% relative humidity as reference data for the as powders, granules, pellets, microtablets, and even PCR model, and observed an inaccuracy of the NIR liquids or semi-solids. The empty shell, usually method at high humidities. Several possible explan- composed of gelatin and 12–16% residual moisture ations were given. However, the most obvious one, acting as a plasticizer, is purchased from a contract namely structural changes of the gelatin shell induced manufacturer and filled on automatic high speed at high moisture levels was not considered and filling machines. Identity, assay, moisture content certainly omitted in the PCR model simply based on and drug dissolution are the key parameters in hard the first PC. Berntsson et al. used loss on drying capsule quality control. At first glance, NIR spectro- reference data in the moisture range of 5.6–18% w/w scopy is actually an ideal method to simultaneously and obtained best results using MLR based on three determine these parameters from one single measure- wavelength regions for water and the gelatin back- ment, thus, replacing time-consuming compendial bone, respectively.
methods. Moreover, stability testing, aiming at the Since gelatin is susceptible to cross-linking when effect of storage conditions and shell/fill interactions, traces of aldehydes are present in the fill, non- might be facilitated. The reality is, however, some- destructive monitoring of this reaction is highly what more difficult, as will be discussed below.
valuable, since it affects the in vitro dissolution rate In 1987, Lodder and co-workers published a of the capsules. Gold et al. published a paper paper describing the use of NIR spectroscopy and a on NIR reflectance monitoring of formaldehyde- quantile-BEAST bootstrap algorithm for discriminat- induced crosslinking of hard gelatin capsules.
ing adulterated and unadulterated capsules. It is worth Although the measurements were performed with mentioning that this was the first report of NIRS empty capsules, the target parameter for the calibra- applied to the analysis of intact dosage forms tion model was the dissolution rate of amoxicillin following the deaths caused by cyanide-laced capsules used as a model drug in the fill. The NIR spectra of in the early and mid-1980s. The authors reported the stressed versus unstressed capsule shells revealed significance of shell color, which induced light changes reflecting new chemical bonds and water scattering, and sample positioning, which affected fill loss upon cross-linking.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 Within the last few years, Reich and co-workers composed of gelatin, water and one or two polyol have presented a large body of data plasticizers Analysis of soft gelatin capsu- demonstrating the potential of NIR transmittance les, i.e. identity, assay, hardness, moisture content, and reflectance spectroscopy in hard capsule shell dissolution, and stability testing, is usually a very qualification focussing on identification of the gelatin time-consuming procedure, due to the more or less type, manufacturing and storage-induced structural or complex composition of shell and fill. A non- moisture changes, and related performance problems, separative, multi-sensing method, such as NIR spec- such as brittleness. The studies revealed that the troscopy, providing combined chemical and physical spectral range between 1800 and 2500 nm is favorable information of shell and fill, would certainly be for hard gelatin capsule shell identification and desirable. However, only a few papers have been qualification purposes. Different batches of chemi- published dealing with the application of NIR to soft cally identical transparent and opaque capsules with gelatine capsule analysis Several different mechanical performance upon filling, result- reasons might be responsible for this: (1) The thick, ing from manufacturing-induced structural changes, often colored gelatin shell strongly absorbs in the NIR could be distinguished by characteristic band shifts in region, thus, more or less complicating NIR measure- this region (Moisture content evaluation was ments of target parameters in the fill. (2) Positioning found to depend strongly on the type of colorant for spectra collection can be an important source of present in the shell. Strong correlations of NIR variance, due to shape effects, e.g. variable shell spectral data with DSC and DMTA test parameters, thickness within the capsule, seam effects, and bi- e.g. differences in gelatin physical state (Tg), struc- coloring (3) Room conditioning is required tural order (enthalpy), and viscoelastic properties (EV, during NIRS measurements to reduce undesired EW) were feasible In summary, these data effects of moisture changes in the shell clearly reveal that NIR spectroscopy is a powerful tool Considering these challenges, it is not surprising for predicting hard capsule shell performance upon that NIR feasibility studies focussing on shell cross- filling, thus allowing for at-line or even on-line linking shell moisture content plasticizer control of these parameters at capsule filling machines content and related physical shell perform- (see Section 5.3.5).
ance have been performed with transparent, Soft capsules consist of a lipophilic, hydrophilic or emptied capsules and/or film formulations. Gold et al.
amphiphilic liquid or semi-solid fill enveloped by a used NIR reflectance measurements to study the one-piece, hermetically sealed outer shell. Contrary to migration of formaldehyde from a polyethylene glycol hard capsules, they are formed, filled, and sealed in (PEG) fill into the shell and its reaction with gelatin.
one continuous operation. Their shell, having a The authors used clear capsules and extracted the fill thickness in the range of about 500 Am, is usually before data collection. The spectral changes clearlyrevealed the formation of new chemical bonds and adepletion of water in the shell with increasingconcentration of formaldehyde in the PEG fill. Only Empty Hard Gelatin Capsules -3D- Loading Plot recently, Reich and co-workers presented a series ofconference proceedings demonstrating the potential ofNIRS for assessing the chemical and physical proper-ties of soft gelatine capsule shells immediately afterprocessing and upon storage Toreduce the variance associated with positioning andinterferences with the fill, the authors used transparentfilm formulations instead of soft capsules in their feasibility studies, which were performed in trans- flectance mode. The spectral data revealed that thecomplex dynamic gelatin/water/plasticizer system of a Fig. 4. NIR discrimination of elastic and brittle hard gelatin capsuleshells.
soft capsule shell that has been reported in the G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 literature requires careful selection of data Interestingly, very little data is available on the use pretreatments and data processing for modelling of NIRS for quality control of lyophilized proteins moisture and plasticizer content determinations Lin and Hsu used five different Moreover, the type of gelatin was found proteins to evaluate the accuracy of NIR moisture to be an important issue that should not be neglected.
content determinations using different chemometric However, with the appropriate chemometric approach, approaches. The results revealed differences between robust calibration models were able to reliably the proteins with respect to calibration modelling.
quantify moisture (range: 6–12% w/w; SEP = 0.3%; Reich and co-workers reported the use of Karl Fischer reference data) and plasticizer content NIR spectroscopy to evaluate stress-induced structural (range: 0–50% w/w relative to gelatin; SEP = 1.3%) in changes of proteins and stabilization effects of sugars different formulations with respect to gelatin and upon lyophilization, storage, and rehydration. Spectra plasticizer type These results clearly indicate of stressed and unstressed proteins revealed changes that understanding the NIR spectral changes of soft associated with the primary, secondary, and tertiary gelatin capsule shells associated with water and structure of the proteins. Sensitive amide I, II and III plasticizer changes is a prerequisite for future appli- bands and the water absorption band could be used for cations of NIR spectroscopy in soft capsule quality the assessment of protein structural changes and control and stability testing.
aggregation, moisture content changes, and even thephysical state (Tg) of the lyophilized product. Based 5.2.3. Lyophilized products on MIR reference data, reliable calibration models for Lyophilization is usually performed to increase the the determination of changes in the a-helical structure storage stability of hydrolytically unstable drugs that were achieved In addition, feasibility of NIR are intended to be used as injectables or to achieve an qualification and quantification of amorphous to instantly soluble oral dosage form. High cake poros- crystalline transitions as a function of storage con- ity, low residual moisture, and, in the case of proteins, ditions were shown.
an amorphous, glassy state are the most prominent Although there are still a number of challenges to quality criteria of lyophilized products.
overcome, it can be expected that in the near future Traditionally, the moisture content of lyophilized noninvasive NIR measurements will at least partly products is determined by time-consuming methods, replace mid-IR measurements for stability testing of such as Karl Fischer titration. In addition, the lyophilized proteins. Moreover, this approach is procedure requires the vial to be opened for analysis.
interesting for on-line and in-line process monitoring Moisture determination with NIR diffuse reflectance (see Section 5.3.2).
techniques can be performed in a fast and non-invasive manner through the glass vials. Due to these 5.2.4. Polymeric implants and microspheres advantages, the NIR technique has been well- Within the last 20 years, polymeric implants and adopted in the pharmaceutical industry for efficient microspheres have gained increasing interest as moisture content determination of lyophilized prod- parenteral drug delivery systems to provide sustained ucts. Early and recent scientific papers in this field release profiles. The matrix of such systems usually have focussed on the investigation of consists of a hydrophobic, non-degradable polymer parameters affecting measurement accuracy, such as and optionally a water-soluble pore-forming additive, cake dimensions particle size or a biodegradable polymer, such as polylactide-co- porosity and formulation changes glycolide (PLGA). Quantitative analysis of active Derksen et al. used the NIR approach for ingredients and/or release-controlling excipients stability testing and correlated moisture content data within these dosage forms usually involves destruc- with the concentration of the active ingredient to tive extraction procedures. Moreover, release testing calculate product shelf-lives. Only recently, Sukow- is time-consuming and often requires huge amounts of ski and Ulmschneider described high speed test samples, since these dosage forms are sometimes AOTF-based NIR measurements of lyophilized vials formulated to release the active component over for moisture compliance, i.e. release testing.
weeks or months.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 The application of NIRS as a fast and non- 133,135]. The studies revealed that release monitoring destructive alternative method for quantification of of drugs from PLGA matrices is a great challenge, excipients and actives within polymeric drug delivery since upon incubation in buffer solution the polymer systems, such as implants, films and microspheres has hydrates and slowly hydrolyses, and the matrix been reported in the literature by two different groups.
erodes. Spectral changes recorded from tablets, films Brashear et al. investigated the use of NIR or microspheres, therefore, comprise not only the reflectance measurements for quantification of an information of the decreasing drug content, but also active compound, namely lomefloxacin HCl, and a the information of the changing structure of the pore-forming excipient, namely polyethylene glycol polymer matrix. Anyhow, reliable calibration models (PEG) 600, in poly(e-caprolactone) microspheres and could be obtained for both dried and hydrated implants fabricated by a melt-mold technique. Analyte samples, thus, indicating the potential of NIRS even specific wavelength selection and second derivative for the analysis of complex matrix systems ( transformation followed by PLS modelling allowedfor excellent correlations with UV results for the 5.3. Process monitoring and process control active and weight-based theoretical values for PEG,respectively. Reich and co-workers used Noninvasive monitoring of all relevant process NIR transmittance and reflectance spectroscopy steps leading to a pharmaceutical drug product is an together with analyte specific wavelength selection, integral part of the PAT paradigm of real-time or second derivative transformation, and PLS data parametric release and quality by design (see Section processing to determine theophylline and quinine 4.2). Ideally, the pharmaceutical survey chain should content (0–20% w/w) within PLGA microparticles include raw material income (see Section 5.1), all unit and tablets and lyophilized protein/sugar operations leading to intermediates and final products, mixtures (absolute protein content: 0–2.5% w/w) in and packaging.
The noninvasive and multivariate character of NIR The same group described the application of NIR techniques provides an interesting platform for transmittance and reflectance measurements for mon- pharmaceutical process monitoring and control.
itoring matrix hydration, matrix degradation, and drug Although most of the reported applications of NIR release (theophylline and lysozyme) from biodegrad- spectroscopy in the pharmaceutical industry are off- able PLGA tablets, films and microspheres line or at-line, there are also some on-line and in-line Lysozym release from PLGA Validation Spectra f(x)=0.9601x+0.2566 r=0.974755 Calibration Spectra f(x)=0.9777x+0.0964 r=0.988771 - NIR measurement Lysozym in tablet after incubation [mg] Lysozym in tablet after incubation [mg] - Reference measurement Fig. 5. Quantitative calibration model for NIR determination of in vitro lysozyme release from poly(d,l-lactide-co-glycolide) tablets (PBS pH7.4/37 8C).
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 applications. In this section, the current state and also evenly distributed. The role of the excipients, future potential of NIR techniques in pharmaceutical which not only improves dosage form compliance, but at-line, on-line, and in-line process monitoring and also affects the technological and biopharmaceutical process control will be reviewed and discussed, with performance of the formulation, is simply neglected.
the main focus on technological unit operations that Considering these disadvantages of traditional are critical for the manufacture of solid dosage forms.
powder blend monitoring procedures, the potential A discussion on chemical reactions, crystallization value of a noninvasive NIR on-line or in-line approach and fermentation processes, or extraction and purifi- is evident. NIR monitoring of powder blending can be cation procedures, all relevant operations in the performed with fiber-optic reflectance probes, thus, production of pharmaceutical raw materials, is minimizing assay time and sampling error. Moreover, beyond the scope of this paper and will not be since most pharmaceutical active ingredients and considered. For these topics, the interested reader is excipients absorb NIR radiation, NIR measurements referred to an excellent textbook chapter dealing with can provide homogeneity information regarding all chemical reaction monitoring and some inter- mixture components. The multi-sensing property of esting papers containing a comprehensive discussion NIR diffuse reflectance spectra, resulting from absorp- of chemical reaction polymorph conver- tion and scattering, provides a bmultivariate finger- sion and bioprocess monitoring printQ of both chemical and physical sample properties.
with NIR spectroscopy.
The use of NIR spectroscopic techniques for powder blend uniformity analysis has been reported 5.3.1. Powder blending by several authors using off-line analysis of samples Mixing is a fundamental and critical process in the taken from different blender locations at various manufacturing process of solid and semisolid phar- blending times and on-line or in-line maceutical dosage forms. The ultimate goal of any monitoring of powder mixing For on-line mixing procedure is to achieve an bideal mixQ, i.e. a and in-line monitoring, two different approaches of situation where the components of a mixture are spectral data acquisition have been used, namely in a homogeneously distributed. In practice, this cannot be bstop-startQ fashion, where the blender is kept sta- achieved in many cases, in particular when dealing tionary during NIR measurements, and in a bdynamicQ with powder blends, since the nature of an boptimalQ fashion with moving samples.
powder blend may be rather diversified depending on Sekulic and co-workers were among the first the material characteristics and the blender type who reported the use of a NIR fiber-optic probe Pharmaceutical powder blending processes are, there- inserted in the axis of rotation of a tumble blender for fore, optimized during development in such a way as real on-line stop-start measurements at different times to stop the process when the mixture homogeneity is of the blending process. Only recently, El-Hagrasy within a pre-defined bspecificationQ regarding active pointed out that multiple spectral sampling content uniformity.
points in the blender are essential for accurate and Current approaches to assess powder blend homo- precise estimation of mixing end points when using geneity are time consuming and hampered by the stop–start fashion. This result was further sub- sampling errors since they involve the removal stantiated by the additional use of a NIR camera that of unit-dose samples from defined mixer locations enabled large spectral images of the blend to be using a sample thief, the extraction of the active drug obtained (see also Section 6.3).
from the sample matrix, and the drug content analysis To allow proper in situ analysis of moving powder by either HPLC or UV spectroscopy. The distribution blends, the effect of sample movement on the spectral of individual excipients is typically assumed to be response was addressed in detail by Berntsson et al.
homogeneous if the active ingredient is uniformly The authors realized that sample move- distributed. In the traditional pharmaceutical sense, ment can cause unwanted spectral artefacts when blend homogeneity obviously addresses only the heterogeneous samples are analyzed with a dispersive, distribution, i.e. the content uniformity of the active mechanically scanning grating spectrometer. The drug substance while assuming that the excipients are performance of an FT spectrometer was found to be G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 suitable for the analysis of powders moving at probes is a feasible option to optimize drying times.
moderate speeds (up to 1 m s1).
Several approaches, including microwave, vacuum, Several data processing strategies for the assess- fluid-bed and freeze drying processes, have been ment of blend homogeneity and/or optimal blending described in the literature.
times from NIR measurements have been evaluated in White published a paper in 1994 reporting the literature. Most of these reports were concerned the use of NIR for on-line moisture endpoint detection with qualitative assessments, such as dissimilarity in a microwave vacuum dryer. The calibration between the spectra of a mixture and the ideal equation used NIR absorbances of water and the spectrum of the mixture or a moving block matrix measured at 1410, 1930 and 1630 nm, standard deviation of NIR spectra respectively. For samples containing less than 6% These approaches generally revealed acceptable moisture, NIR values were within 1% of the Karl results, although Wargo and Drennen suggested Fischer reference data with a SEP of 0.6%. At that bootstrap techniques provided a greater sensitiv- moisture levels above 6%, a bias was observed, ity for blend homogeneity assessment than chi-square which was attributed to sampling limitations and the calculations. Some recent papers are also broad range of moisture contents (0.7–25.7%) con- concerned with quantitative analysis, pointing out that sidered in the calibration. Changes in drug content of quantitative analysis is a prerequisite for a complete the granules did not affect the prediction of moisture resolution of the chemical and physical properties of content, thus, demonstrating the robustness of the the mixture. Non-linearity, which was found to be a calibration model.
feature of powder blends containing coarse and fine The work of Harris and Walker involved particles, was not a problem when using a cubic PLS real-time quantification of organic solvents, water and mixtures thereof, evaporating from a vacuum dryer. A To summarize, it can be concluded that on- and in- fiber-optic coupled AOTF-NIR spectrometer was used line powder blend monitoring with NIR spectroscopy for data collection from the vapor stream and a is not an easy task, but feasible and in line with the balance was placed in the dryer to record the reference PAT paradigm of real-time release, focussing on data. PLS calibration models were built for on-line continuous process understanding and quality control prediction of optimal drying times. Morris et al. of all production steps, rather than a final product and Wildfong et al. used NIR in-line monitoring control only.
to visualize the different stages during a fluid-beddrying process and to accurately determine the endpoint of accelerated fluid-bed drying processes.
The manufacturing process of a solid pharmaceut- Only recently, Zhou et al. described the ical dosage form usually involves several steps, often advantage of NIRS for in-line monitoring of a drying including at least one blengthyQ drying process, process with concomitant distinction between bound resulting from the time required to dry the material and free water of a drug substance forming different plus the time to analytically verify the drying hydrates. The study revealed that NIRS can serve as a endpoint. Fluid-bed drying and tray drying in a large tool to ensure that the desired hydrate form is oven are the most frequently used methods for wet achieved at the end of a drying process.
granules. Microwave vacuum drying is another An interesting paper on the in situ monitoring of a option, although less popular. Freeze- and spray- freeze-drying process has recently been published by drying are the methods of choice for temperature- and Brqlls et al. A NIR fiber-optic probe fitted to a moisture-sensitive drug substances. Current methods FT spectrometer was placed in the center of a vial 1 to determine drying endpoints include indirect in-line mm above the bottom. An aqueous PVP solution was methods, such as temperature measurements, and used as a model formulation. NIR monitoring of the direct off-line moisture analysis of samples taken different stages of the process, namely freezing, from the dryer. Since O–H vibrations of water exhibit primary, and secondary drying, was able to detect a large absorption in the NIR region, on-line the freezing point, completion of ice formation, and monitoring of moisture levels using NIR fiber-optic transition from the frozen solution to an ice-free G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 material. Moreover, NIR spectra provided new infor- ments. The authors observed a significant effect of the mation about the drying process, such as the liquid flow rate and the process air temperature desorption rate and the steady-state value at which Frake et al. reported the use of in-line NIR to drying was complete. These results clearly indicate investigate granule water uptake and particle size that the application of an in situ NIR configuration changes during aqueous top-spray fluid-bed granula- offers the possibility of studying product character- tion. During the process, spectra were obtained every istics during freeze-drying, thus, increasing our 2.5 min with a mounted fiber-optic probe fitted to a understanding of important parameters in the formu- grating-based spectrometer ranging from 1100 to lation development of lyophilized products.
2500 nm. To determine moisture content quantita-tively, and, thus, allowing for exact endpoint deter- 5.3.3. Granulation mination, the second derivative absorbance changes at The production of tablets often requires a gran- 1932 nm were calibrated against LOD and Karl ulation step to improve powder flow and compaction Fischer reference data. A linear relationship was characteristics, as well as to achieve content uniform- obtained with SEC values in the order of 0.5% for ity. Wet granulation is usually performed in a high both models ranging from 1.5 to 11% w/w of speed mixer or a fluid-bed granulator and comprises moisture. For particle growth monitoring, the authors the following critical steps: wetting, granule formation tried to develop another calibration model, again and drying. At-line or in-line monitoring and endpoint based on one single wavelength only, namely 2282 determination of wet granulation processes with NIR nm. However, considering the complex full range spectroscopy offers the possibility of simultaneously spectral effects of particle size changes (see also determining particle size and moisture content. More- Section 5.1), it is not surprising that the authors failed over, water/excipient interactions, hydrate formation, to develop an acceptable quantitative calibration and/or blend segregation may be assessed easily. The model for particle size determination.
following examples taken from the literature will Goebel and Steffens presented successful illustrate the potential and limitations of granulation data for a simultaneous on-line determination of process monitoring with NIR spectroscopy in both particle size and moisture content of samples in a formulation development and in routine production.
fluid-bed granulation process using a FT spectrometer.
In 1996, List and Steffens published a paper The robustness of the PLS calibration models, based on NIR in-line monitoring of a wet granulation on Karl Fischer and laser diffraction reference data, process in a mixer granulator. The process was was evaluated by applying them to development and stopped after certain time intervals and a NIR sensor pilot-scale plants. The results clearly revealed that probe within the mixer recorded the spectra. A reliable particle size measurements are a greater challenge for quantitative PLS calibration model for moisture NIR on-line monitoring configurations than moisture determination of a placebo mixture ranging between content determination, a fact that was attributed to 6 and 15% w/w was developed and validated using sample presentation, e.g. density effects and certain Karl Fischer reference data. Best results were obtained variables of the fiber-optic probes.
with the following spectral pretreatments: wavelength Rantanen and co-workers published a series of selection (5000–5500 cm1), normalization, and first papers dealing with the evaluation of a derivative. The authors discussed the limitations of NIR sensor of only a few wavelengths for in-line transferring placebo calibrations to active products moisture monitoring of fluid-bed granulation. In one and demonstrated the feasibility of qualitative NIR of the papers the authors investigated the effect particle size monitoring during granulation.
of particle size, particle composition and binder type Watano and co-workers were among the on NIR moisture monitoring using a full range off-line first who reported the use of a NIR sensor for moisture FT spectrometer. The study revealed that wetting and monitoring and process automation of an agitation particle growth changes the reflection and refraction fluid-bed granulation process. A fixed-wavelength properties of the granules in a complex manner, NIR filter instrument was used to study the effects depending not only on the wavelength, but also on the of operational variables on the NIR moisture measure- absorption properties of the powder matrix and the G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 binder type. Calibration of in-line NIR moisture the hydrate formation of anhydrous theophylline and measurements, even with a fixed-wavelength setup, caffeine during wet granulation.
therefore, requires understanding and consideration ofthese factors affecting NIR signals.
5.3.4. Pelletization The use of spectral changes of solid powders and Interestingly, only little literature data is available granules associated with moisture uptake and/or on NIR monitoring of pelletization. In 1996, Wargo moisture loss is not limited to moisture content and Drennen developed an at-line NIR method determinations. They can help to understand the to monitor the layering of non-pareil seeds with an chemical and physical performance of active com- aqueous suspension containing diltiazem HCl, poly- pounds and excipients in wet granulation processes.
vinyl pyrrolidone, and micronized talcum. Three Buckton et al. used NIR to study the effect of independent calibration models were developed to granulation on the structure of microcrystalline determine endpoint pellet potency of 15, 30 and 55% (MCC) and silicified (SMCC) microcrystalline cellu- w/w diltiazem beads. The models were successfully lose and to explain the compressibility changes of transferred from a laboratory scale to pilot scale.
MCC after wet granulation. It was found that MCC, Radtke et al. described in- and at-line NIR SMCC and wet granulated SMCC had essentially configurations for moisture monitoring during matrix identical physical structures, while wet granulated pellet production in a rotary fluidized bed. The authors MCC exhibited structural changes in the NIR spec- found out that sample presentation is as critical in this trum related to C–H bonding. With the NIR assess- case as in granulation process monitoring.
ment of the altered physical structure, it was possibleto explain the change in compressibility of MCC after 5.3.5. Tabletting and capsule-filling wet granulation.
High speed automatic capsule filling and tabletting Derbyshire et al. used NIR together with machines require non-segregating powder blends or other analytical techniques, namely DSC, NMR, and granule mixtures with good flow characteristics to TDS, to study the molecular properties of water in work properly, and ensure content uniformity and hydrated mannitol. In accordance with the results consistent dissolution profiles of the final product. In obtained from the other methods, NIR spectral data at practice, segregation of free-flowing particulate mix- 5172 cm1 (O–H bond of water) and 5930 cm1 tures with differences in particle size and/or density is (C–H stretching peak) clearly indicated two transition likely to occur through inherent vibrations during points for the coordination between water molecules blender discharge, batch transfer to the filling or and mannitol molecules, namely at 0.11 and 0.25 g/g, compression area, and even within the equipment.
respectively. The authors speculate that the transitions Since NIR techniques are able to recognize are associated with different stages of microdissolu- chemical and physical changes of particulate blends tion of the solid, thereby changing the hydrogen- whole tablets and filled capsules, noninvasive bonded network between water and mannitol, e.g. the NIR monitoring of tabletting and capsule filling, from molecular response of water and mannitol in the the very beginning to the very end of the process, spectra. This result argues for the potential of NIR in- would be valuable to increase production speed and line measurements in predicting the quantity of water improve product quality. A NIR sensor on the feed required for the successful formation of granules hopper of a capsule-filling machine or a tablet press could effectively identify the powder mixture and With the opportunity to monitor solid/water inter- detect segregation problems of particulate matter upon actions, i.e. to detect different states of water feeding the equipment. The final product could be molecules in a solid, it is not surprising that NIR further assessed for content uniformity, dissolution spectra may also provide information on pseudopoly- properties, and, in the case of tablets, for hardness (see morphic transitions during wet granulation. In two also Section 5.2). Indeed, there are some industrial subsequent papers, R7sanen et al. and Jorgensen approaches leaning in this direction, although they et al. demonstrated the efficiency of NIR have not yet been fully exploited, due to limitations in spectroscopy to study the state of water and, thus, spectra collection of tablets or capsules produced at
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 high speed. However, it might be expected that method provided predictions of applied polymer films progress in process instrumentation and chemometric with SEP values of 1.07% or less, depending on the data processing will speed up the development of NIR coating formulation. For pigment-free coating formu- process monitoring in tabletting and capsule filling in lations, the calibration model was based mainly on the near future.
distinct absorption peaks of the coating polymer. Informulations containing high concentrations of water- 5.3.6. Film coating insoluble dyes and opacifying agents, such as titanium Film coating is a process commonly employed in dioxide, baseline shifts were the primary spectral the pharmaceutical industry to either improve the taste change caused by an increase in film thickness.
or swallowing of tablets, or to control drug dissolution Subsequent papers on this topic were published by rate from the solid dosage forms. Regardless of the Andersson et al. who described an indus- intended use, the functionality of a film coat is closely trial in-line approach for film coat monitoring of related to its thickness and uniformity around the solid pharmaceutical pellets with fiber-optic probes. Cali- core. In most production settings, the endpoint of a bration models for the determination of film coat coating process is determined by in-process sample thickness were based on reference data obtained from acquisition, the weighing of a known sample size and the determination of the theoretical amount of applied Despite these interesting and excellent papers polymer. Correct film coat thickness and uniformity clearly reflecting the great value of NIR techniques are evaluated indirectly by disintegration and/or for at-line or in-line monitoring of a coating process, dissolution testing. In the PAT sense, this analytical the multivariate potential of NIR spectroscopic meth- procedure has two major disadvantages: first, deter- ods has not been fully exploited in this field. As mination of mass increase does not account for mass indicated by Reich and Frickel in a series of conference loss of core material, thus, reducing the accuracy of proceedings NIRS could be implemented as the method; and secondly, disintegration and/or a useful at-line or in-line tool to survey and determine dissolution testing are only indirect, rather time- the effect of process conditions on film coat uniformity consuming methods for the measurement of coating (and related biopharmaceutical properties (see levels and uniformity.
also Section 5.2.1). As will be discussed in section NIR techniques, on the other hand, allow for a 6.3, imaging techniques might be an additional tool to rapid, noninvasive at-line and in-line monitoring and improve product quality and the production speed of control of film coating processes prior to biopharma- film-coated dosage forms ceutical testing. Kirsch and Drennen and Wargoand Drennen were among the first to describethe use of NIR for at-line monitoring of film coatingprocesses on tablets and pellets. A Wurster columnwas retrofitted with a sample thief, allowing with-drawal of 10-tablet samples during coating. Sampleswere collected after different time intervals andmeasured on a grating-based NIR spectrometer inreflectance mode. In the case of pellets coatingsamples were classified by a bootstrap patternrecognition technique. The bootstrap standard devia-tion plot made a qualitative identification of coatingendpoints possible. In the case of tablets quantitative calibration models for the determinationof applied polymer solids, namely ethylcellulose andhydroxypropylmethyl cellulose formulations, were Fig. 6. NIR discrimination of Eudragit L film coats on tablets; effect developed based on mass increase reference data (0– of spraying temperature before ageing (20bT: 20 8C, 30bT: 30 8C) 30% w/w) corrected for core attrition. The NIR and after ageing (20aT, 30aT).
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 Packaging is the last step in the production line of a pharmaceutical product. To ensure the product 6.1. Basic principles and instrumentation safety of pharmaceuticals, a last identity check of theproduct on the packaging line would be highly NIR imaging is a combination of NIR spectroscopy desirable. Such an inspection system based on the with digital image processing. A NIR imaging system combination of a conventional high resolution camera is basically composed of an illumination source, an with an on-line diode NIR spectrometer ranging from imaging optic, a spectral encoder selecting the wave- 900 to 1700 nm at 6 nm resolution has been lengths, and a focal plane array (FPA) as indicated in developed recently. The system is supposed to NIR light from an illumination system is perform a 100% identity check at full line speed focussed upon the sample. The diffuse reflectance (i.e. 12,000 tablets per minute) before closing the image of the sample is collected by an imaging optic, blister. The potential of this type of equipment has the configuration of which depends on the sample size been evaluated in a feasibility study Using and type. For macroscopic or microscopic images a hard gelatin capsules of different shell and fill focusing lens or a microscope objective are used, compositions, the authors could demonstrate that the respectively. Data collection proceeds by recording a real-time algorithms used in this system work as series of images on the near-infrared (i.e. InSb or reliably and accurately as a PCA-based data evalua- InGaAs) FPA at each wavelength position selected by tion of spectra collected on an off-line lab spectro- a spectral encoder, such as a liquid crystal tunable meter to ensure the identification of flawed products.
filter element (LCTF) or an interferometer. The result It may, therefore, be expected that other configu- is a three-dimensional data set, known as a spectral rations based on high speed NIR spectrometer or NIR hypercube with the x and y axis representing spatial imaging techniques will be developed in the near information and the z axis representing the spectral future for identity check on packaging lines.
False Colour Image Fig. 7. Basic configuration of a near-infrared imaging system.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 Regarding instrumentation there are basically two because the spectral data are collected in parallel different approaches. The first approach is the wave- and, thus, are not hampered by a dilution effect in length scanning method, also known as the bstaring the same way as NIR bulk measurements are. This imager methodQ. Sample and camera are kept sta- is a great advantage over conventional NIRS when tionary and single images are recorded for each analyzing low dose actives or excipients in a wavelength. The spectral information is provided either by a number of discreet filters, by tuneable ! Moreover, NIR imaging enables quantitative infor- filters, or by combination with an imaging Fourier- mation to be obtained without running separate transform spectrometer. The images recorded for the calibration samples, since pure component spectra different wavelengths are combined by the software are directly available from the spectral imaging and the spectra calculated. The second approach, also data cube of heterogeneously mixed samples. This known as bpush-broom scanningQ method, requires a approach can help to save time and money when relative movement between camera and sample to building a quantitative calibration model for scan over the surface. The imaging system records the pharmaceutical applications, in particular for spatial information linewise and provides the spectral expensive peptide or protein drug formulations.
information for each pixel along the line by projectionalong the second axis of the two-dimensional camera- NIR spectroscopic imaging has only a short chip. The spectral encoding is provided by either history when compared with MIR and Raman linear variable filters, a digital micro-mirror array in imaging techniques. This is due to the fact that its combination with a grating, or dispersive optics. The advantages over Raman and MIR imaging techni- computer software combines the slices, derives the ques, such as adaption to a wide variety of fields- second axis and, thus, reconstructs the full image.
of-view (FOV) and extreme tolerance to variations Experimental setups based on the staring imager in sample geometry, have only recently been fully method are mainly used in research and quality exploited With the use of simple quartz– control laboratories with data acquisition times of tungsten halogen sources and an image filtering, typically 2 min or less. The second approach is used instead of a source filtering approach, NIR imaging for conveyor belt survey with data acquisition times techniques enable wide-field illumination for a depending on the spectral encoder. A detailed variety of magnifications and imaging modes, description of the different principles can be found ranging from around 0.2 to 125 mm. In addition, in some recent textbooks flatness of the sample is not a prerequisite as inRaman and MIR imaging. On the contrary, NIR 6.2. Analytical targets and strengths imaging systems allow experiments to be performedon very irregular samples, since NIR imaging Conventional, i.e. non-imaging NIR spectroscopy, systems perform well in the reflectance mode with analyzes the sample in bulk and determines an large depths-of-field and an excellent signal-to-noise average composition across the entire sample. NIR ratio of the arrays.
imaging, on the other hand, provides informationabout the spatial distribution of the components 6.3. Pharmaceutical applications comprising the sample. It is, therefore, a powerfulbline extensionb of conventional NIR analysis in a With the addition of spatial information and number of different ways parallel data collection, NIR imaging certainly meetsthe challenging analytical needs of pharmaceutical ! The opportunity to visualize the spatial distribution quality and process control, and may serve as a of a chemical species throughout the sample enables versatile adjunct to conventional, non-imaging NIR the degree of chemical and/or physical heteroge- spectroscopy in many fields. Despite the obvious neity within a given sample to be determined.
strengths of NIR imaging techniques, the number of ! The array-based spectral sensing of a NIR imaging scientific papers and technical notes describing their system also allows for trace sample measurements, practical use is limited and mainly in other fields, G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 e.g. plastic sorting high-throughput screening NIR imaging to identify mixing problems as being of biological material on conveyor belts, responsible for bad and good flow characteristics of remnant analysis of works of art and powder blends, as well as tablet sticking and tablet identification of atherosclerotique plaques by means fracture. The results clearly reveal that NIR imaging is of an intra-arterial catheter imaging system recently a powerful tool for matrix characterization not only in developed at the University of Kentucky. Pharma- final product control, but also in research, develop- ceutical papers, as discussed in more detail in the ment and scale-up of solid pharmaceutical dosage following paragraphs, focus on three different forms, i.e. for process and formulation optimization aspects, namely blend uniformity analysis in pow- ders and tablets, composition and morphological The same group pointed out that matrix character- features of coated tablets and multi-layer granules, ization of complex solid dosage forms requires an and spatial changes in biodegradable PLGA matrix understanding of the spatial relationship and inter- systems upon matrix hydration, degradation and action of drug formulation components. NIR imaging active release.
was, therefore, used to examine the internal structure El-Hagrasy et al. used an InSb imaging of time-release granules The chemical image of camera with discrete bandpass filters encompassing a bisected granule (0.9 mm2) was obtained at 10-nm absorption bands of the blend components, in intervals from 1000 to 1700 nm through a 10 addition to a conventional NIR fiber-optic probe in microscope objective with a total acquisition time of six sapphire windows mounted at different locations approximately 2 min. In contrast to the visible image, in a V-blender, to monitor powder blend homoge- the NIR chemical image clearly revealed that the neity of salicylic acid/lactose mixtures and compare distinct layers and boundaries were consistent with the the potential of the two techniques. Data analysis expected physical structure and composition of this indicated the necessity of using multiple sampling points for mixing endpoint determination by tradi- Another interesting application of NIR imaging is tional NIRS and clearly revealed that coupling both the chemical visualization of coating layers on techniques might provide a very robust tool for tablets. In a technical note published at the AAPS monitoring powder blending, since the volume of Annual Meeting in 2001, Lewis and co-workers powder captured by the imaging technique is much showed the chemical image of a sectioned multilayer-coated tablet. The macroscopic chemical Koehler et al. demonstrated the use of NIR image depicted the tablet core and two distinct imaging to visualize and quantify the spatial dis- coating layers of different thickness. Due to the large tribution of the active ingredient in a tablet. The field-of-view (FOV), a detailed examination of the authors used an unsupervised PCA score plot to film coat uniformity on the tablet core was not qualitatively visualise the degree of chemical hetero- feasible. Moreover, sectioning of the tablet was geneity of the formulation showing the active in necessary to achieve the multilayer chemical image unevenly distributed clumps. An alternate least of the tablet. However, considering that formulation- square regression method, based on pure component and/or process-induced microheterogeneities in film spectra isolated from the spectral data cube of the coats on tablets or pellets might have rather tablet, was used to build a quantitative concentration important implications on their biopharmaceutical distribution estimate of the active in the tablet.
properties, the necessity of spectroscopic imaging Although in this special case, the active concen- techniques for film coat uniformity analysis is tration was 20% by weight, the example clearly obvious. Interestingly, the application of microscopic demonstrates the strength of NIR imaging for the ATR-FTIR imaging rather than NIR imaging has analysis of low dose drugs.
been reported for this purpose Nondestructive Correlation of physical properties and technolog- chemical images (250 Am 250 Am) of Eudragit FS ical functionality of powder blends with their chem- 30 D film coats were obtained from different areas ical heterogeneity is the approach described by (i.e. at the center part and at the edges) of the tablets Hammond and Clarke The group has used to visualize and relate different coating levels, G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 process and/or curing conditions to film coat phase protein adsorption on the PLGA matrix uniformity. The study revealed that, due to its low certainly occurs ( penetration depth, ATR-FTIR imaging may provide In conclusion, the literature data discussed in this interesting new insights in the processes involved in section clearly reveal that spectroscopic imaging film coating and, thus, a better understanding and approaches, with NIR imaging in particular, have a control of manufacturing defects resulting in func- huge potential for gaining rapid information about the tionally important microheterogeneities. Although chemical structure and related physical or biopharma- using the mid-IR, this example again indicates the ceutical properties of all types of pharmaceutical overall great potential of spectroscopic imaging dosage forms, thus improving product quality and techniques in research, development, scale-up and enhancing production speed.
production control of pharmaceutical dosage forms.
Structurally even more complex than film-coated oral tablets or granules are biodegradable poly(d,l- 7. Concluding Remarks lactide-co-glycolide) (PLGA) matrix systems forparenteral use. As discussed in Section 5.2.4, This review has covered some of the recent hydration, degradation and drug release kinetics can methods and pharmaceutical applications of NIR be successfully monitored by classical NIR spectro- spectroscopy and imaging. As a fast and noninvasive scopy, however, without any information on the multivariate technique, conventional NIR spectro- spatial changes. In an attempt to fill this gap, NIR scopy has already gained wide industrial acceptance imaging was used (1) to investigate the time-depend- for raw material identification and/or qualification, ent spatial microenvironmental changes within bio- and nondestructive chemical analysis of intact dosage degradable PLGA films upon in vitro hydration and forms. Considering the continuing improvements in degradation in different media and (2) to hardware and software design, and the analytical chemically visualize the distribution and relative requirements of the most recent concepts of quality by abundance of a model protein, namely lysozyme, in design and real-time or parametric release, it is PLGA matrix tablets, immediately after processing anticipated that in the near future both NIR spectro- and during the release phase Within these scopy and imaging may progressively become routine studies it could be demonstrated for the first time methods for pharmaceutical process monitoring and without fluorescence-labeling that during the release process control.
Fig. 8. False-color near-infrared images of lysozyme distribution (10% initial loading) at the surface of a poly(d,l-lactide-co-glycolide) tablet(A) after 4 days in PBS pH 7.4 and (B) after 14 days in PBS pH 7.4 (T = 37 8C).
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143  T. Naes, T. Isaksson, T. Fearn, T. Davies, Multivariate Calibration and Classification, NIR Publications, Chichester,2002, pp. 27 – 54.
 P.C. Williams, K.H. Norris, W.S. Zarowski, Influence of  T. Naes, T. Isaksson, T. Fearn, T. Davies, Multivariate temperature on estimation of protein and moisture in wheat by Calibration and Classification, NIR Publications, Chichester, near-infrared reflectance, Cereal Chem. 59 (1982) 473 – 480.
2002, pp. 93 – 104.
 E.W. Ciurczak, Principles of near-infrared spectroscopy, in:  W. Plugge, C. van der Vlies, The use of near infrared D.A. Burns, E.W. Ciurczak (Eds.), Handbook of Near- spectroscopy in the quality control laboratory of the Infrared Analysis, 2nd edition, Marcel Dekker Inc., New pharmaceutical industry, J. Pharm. Biomed. Anal. 10 York/Basel, 2001, pp. 7 – 18.
(1992) 797 – 803.
 L. Bokobza, Origin of near-infrared absorption bands, in:  T. Naes, T. Isaksson, T. Fearn, T. Davies, Multivariate H.W. Siesler, Y. Ozaki, S. Kawata, H.M. Heise (Eds.), Near Calibration and Classification, NIR Publications, Chichester, Infrared Spectroscopy: Principles, Instruments, Applications, 2002, pp. 221 – 260.
Wiley-VCH Verlag GmbH, Weinheim, 2002, pp. 11 – 41.
 W. Plugge, C. van der Vlies, Near-infrared spectroscopy as  T. Burger, Radiative transfer in disperse media—new an alternative to assess compliance of ampicillin trihydrate procedures in spectroscopic infrared analysis, PhD thesis, with compendial specifications, J. Pharm. Biomed. Anal. 11 (1993) 435 – 442.
 J.M. Olinger, P.R. Griffiths, T. Burger, Theory of diffuse  Near-infrared spectrometry, Chapter 2.2.40, European Phar- reflection in the NIR region, in: D.A. Burns, E.W. Ciurczak macopoeia, 4th edition, 2001, pp. 55 – 56.
(Eds.), Handbook of Near-Infrared Analysis, 2nd edition,  Near-infrared spectrophotometry, Chapter 1119, United Marcel Dekker Inc., New York/Basel, 2001, pp. 19 – 51.
States Pharmacopoeia USP26NF21, 2003, pp. 2388 – 2391.
 S. Kawata, New techniques in near-infrared spectroscopy, in:  American Society for Testing and Materials, E1866-97: H.W. Siesler, Y. Ozaki, S. Kawata, H.M. Heise (Eds.), Near Standard Guide for Establishing Spectrophotometer Perform- Infrared Spectroscopy: Principles, Instruments, Applications, ance Tests, and E1655: Standard practices for infrared, Wiley-VCH Verlag GmbH, Weinheim, 2002, pp. 75 – 84.
multivariate, quantitative analysis, Official ASTM Publica-  J.J. Workman Jr., D.A. Burns, Commercial NIR instrumen- tions, Philadelphia, PA, 2000.
tation, in: D.A. Burns , E.W. Ciurczak (Eds.), Handbook of  Report by the analytical methods committee, evaluation of Near-Infrared Analysis, 2nd edition, Marcel Dekker Inc., analytical instrumentation, Part XIII. Instrumentation for New York/Basel, 2001, pp. 53 – 70.
UV–visible-NIR spectrometry, Analyst 125 (2000) 367 – 374.
 W.J. McCarthy, G.J. Kemeny, in: D.A. Burns, E.W. Ciurczak  Validation of compendial methods, Chapter 1225, United (Eds.), Handbook of Near-Infrared Analysis, 2nd edition, States Pharmacopoeia USP26, 2002, pp. 2439 – 2442.
Marcel Dekker Inc., New York/Basel, 2001, pp. 71 – 90.
 ICH Quality Guidelines for Method Validation, Q2A, Text on  S. Kawata, Instrumentation for near-infrared spectroscopy, in: Validation of Analytical Procedures, and Q2B, Validation of H.W. Siesler, Y. Ozaki, S. Kawata, H.M. Heise (Eds.), Near Analytical Procedures: Methodology, November 1996, FDA Infrared Spectroscopy: Principles, Instruments, Applications, Wiley-VCH Verlag GmbH, Weinheim, 2002, pp. 43 – 74.
 A.C. Moffat, A.D. Trafford, R.D. Jee, P. Graham, Meeting  A. Candolfi, D.L. Massart, S. Heuerding, Investigation of the International Conference on Harmonisation's Guidelines sources of variance which contribute to NIR-spectroscopic on Validation of Analytical Procedures: quantification as measurement of pharmaceutical formulations, Anal. Chim.
exemplified by a near-infrared reflectance assay of para- Acta 345 (1997) 185 – 196.
cetamol in intact tablets, Analyst 125 (2000) 1341 – 1351.
 S. Kawano, Sampling and sample presentation, in: H.W.
 PASG Guidelines for the Development and Validation of Near Siesler, Y. Ozaki, S. Kawata, H.M. Heise (Eds.), Near Infrared (NIR) spectroscopic methods, May 2001, pp. 1–39 Infrared Spectroscopy: Principles, Instruments, Applications, Wiley-VCH Verlag GmbH, Weinheim, 2002, pp. 115 – 124.
 USFDA Draft Guidance for Industry: PAT—a framework for  M. Blanco, J. Coello, H. Itirriaga, S. Maspoch, C. de la innovative pharmaceutical manufacturing and quality assur- Pezuela, Near-infrared spectroscopy in the pharmaceutical ance, 2003, pp. 4–21 ( industry, Analyst 123 (1998) 135R – 150R.
 H.M. Heise, R. Winzen, Fundamental chemometric meth-  P. Corti, L. Savini, E. Dreassi, G. Ceramelli, L. Montecchi, S.
ods, in: H.W. Siesler, Y. Ozaki, S. Kawata, H.M. Heise Lonardi, Application of NIRS to the control of pharmaceut- (Eds.), Near Infrared Spectroscopy: Principles, Instruments, icals identification and assay of several primary materials, Applications, Wiley-VCH Verlag GmbH, Weinheim, 2002, Pharm. Acta Helv. 67 (1992) 57 – 61.
pp. 125 – 162.
 M. Ulmschneider, G. Barth, B. Reder, A. Vfgel, D. Schilling,  Mark Howard, Data analysis: multilinear regression and Transferable basic library for the identification of active principal component analysis, in: D.A. Burns, E.W.
substances using near-infrared spectroscopy, Pharm. Ind. 62 Ciurczak (Eds.), Handbook of Near-Infrared Analysis, (2000) 301 – 304.
2nd edition, Marcel Dekker Inc., New York/Basel, 2001,  M. Ulmschneider, G. Barth, E. Trenke, Building transferable pp. 129 – 184.
cluster calibrations for the identification of different solid G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143 excipients with near-infrared spectroscopy, Pharm. Ind. 62  M. Blanco, J. Coello, H. Iturriaga, S. Maspoch, F. Gonzalez, (2000) 374 – 376.
R. Pous, Chapter 64, in: K.I. Hildrum, T. Isaksson, T. Naes,  K. Kr7mer, S. Ebel, Application of NIR reflectance spectro- A. Tandberg (Eds.), Near Infra-Red Spectroscopy (Bridging scopy for the identification of pharmaceutical excipients, the Gap Between Data Analysis and NIR Applications), Ellis Anal. Chim. Acta 420 (2000) 155 – 161.
Horwood, Chichester, 1992, pp. 401 – 406.
 K. Beckenkamp, M. Ohm, K. Molt, O. Mandal, The  A.J. O'Neil, R.D. Jee, A.C. Moffat, The application of challenge of identity testing, Eur. Pharm. Rev. (2001) 28 – 33.
multiple linear regression to the measurement of the  M. Glania, Entwicklung NIR-spektroskopischer Qualit7ts- median particle size of drugs and pharmaceutical exci- und Prozesskontrollmethoden fqr Gelatine und Phenolharze, pients by near-infrared spectroscopy, Analyst 123 (1998) Dissertation, Essen (1998).
2297 – 2302.
 S.S. Thosar, R.A. Forbess, M. Kemper, A.J. Shukla,  P. Frake, I. Gill, C.N. Luscombe, D.R. Rudd, J. Waterhouse, Determination of copolymer ratios of poly(lactide-co-glyco- U.A. Jayasooriya, Near-infrared mass median particle size lide) using near-infrared spectroscopy, J. Pharm. Biomed.
determination of lactose monohydrate evaluating several Anal. 20 (1999) 107 – 114.
chemometric approaches, Analyst 123 (1998) 2043 – 2046.
 C. Gustafsson, C. Nystrfm, H. Lennholm, M.C. Bonferoni,  A.J. O'Neil, R.D. Jee, A.C. Moffat, Measurement of C.M. Caramella, Characteristics of hydroxypropyl methyl- cumulative particle size distribution of microcrystalline cellulose influencing compactibility and prediction of particle cellulose using near infrared reflectance spectroscopy, and tablet properties by infrared spectroscopy, J. Pharm. Sci.
Analyst 124 (1999) 33 – 36.
92 (2003) 494 – 504.
 E. Storz, K.-J. Steffens, Bestimmung der spezifischen  E. Storz, Untersuchung funktioneller Parameter pharmazeu- Oberfl7che hochdisperser Siliciumdioxide mittels Nahin- tischer Hilfsstoffe mittels Nahinfrarotspektroskopie. Disser- frarot-Spektroskopie, Pharm. Ind. 64 (2002) 398 – 402.
tation, Bonn (2003).
 A.D. Patel, P.E. Luner, M.S. Kemper, Quantitative analysis of  J. Beyer, Nahinfrarotspektroskopische Untersuchungen an polymorphs in binary and multi-component powder mixtures pharmazeutischen Hilfsstoffen und festen Arzneiformen.
by near-infrared reflectance spectroscopy, Int. J. Pharm. 206 Dissertation, Bonn (2003).
(2000) 63 – 74.
 C.I. Gerh7user, K.-A. Kovar, Strategies for constructing near-  A.D. Patel, P.E. Luner, M.S. Kemper, Low-level determi- infrared spectral libraries for the identification of drug nation of polymorph composition in physical mixtures of substances, Appl. Spectrosc. 51 (1997) 1504 – 1510.
near-infrared reflectance spectroscopy, J. Pharm. Sci. 90  W.L. Yoon, R.D. Jee, A.C. Moffat, P.D. Blackler, K. Yeung, (2001) 360 – 370.
D.C. Lee, Construction and transferability of a spectral  E. R7sanen, J. Rantanen, A. Jargensen, M. Karjaleinen, T.
library for the identification of common solvents by near- Paakkari, J. Yliruusi, Novel identification of pseudopoly- infrared transflectance spectroscopy, Analyst 124 (1999) morphic changes of theophylline during wet granulation 1197 – 1203.
using near infrared spectroscopy, J. Pharm. Sci. 90 (2001)  A. Candolfi, R. De Maesschalck, D. Jonan-Rimbaud, P.A.
389 – 396.
Hailey, D.L. Massart, The influence of data processing in the  A. Jaergensen, J. Rantanen, M. Karjaleinen, L.
pattern recognition of excipients near infrared spectra, Khriachtchev, E. R7sanen, J. Yliruusi, Hydrate formation J. Pharm. Biomed. Anal. 21 (1999) 115 – 132.
during wet granulation studied by spectroscopic methods,  M.S. Kemper, L.M. Luchetta, A guide to raw material Pharm. Res. 19 (2002) 1285 – 1291.
analysis using near infrared spectroscopy, J. Near Infrared  G. Buckton, E. Yonemochi, J. Hammond, A. Moffat, The use Spectrosc. 11 (2003) 155 – 174.
of near infra-red spectroscopy to detect changes in the form  J.E. Sinsheimer, N.M. Poswalk, Pharmaceutical applications of amorphous and crystalline lactose, Int. J. Pharm. 168 of the near infrared determination of water, J. Pharm. Sci. 57 (1998) 231 – 241.
(1968) 2007 – 2010.
 H. Bauer, T. Herkert, M. Barthels, K.-A. Kovar, E. Schwarz,  X. Zhou, P. Hines, M.E. Borer, Moisture determination in P.C. Schmidt, Investigations on polymorphism of mannitol/ hygroscopic drug substances by near infrared spectroscopy, sorbitol mixtures after spray-drying using differential scan- J. Pharm. Biomed. Anal. 17 (1998) 219 – 225.
ning calorimetry, X-ray diffraction and near-infrared spectro-  W. Dziki, J.F. Bauer, J.J. Szpylman, J.E. Quick, B.C. Nichols, scopy, Pharm. Ind. 62 (2000) 231 – 235.
The use of near-infrared spectroscopy to monitor the mobility  J.J. Seyer, P.E. Luner, M.S. Kemper, Application of diffuse of water within the sarafloxacin crystal lattice, J. Pharm.
reflectance near-infrared spectroscopy for determination of Biomed. Anal. 22 (2000) 829.
crystallinity, J. Pharm. Sci. 89 (2000) 1305 – 1316.
 E.W. Ciurczak, R.P. Torlini, M.P. Demkowicz, Determination  S.E. Hogan, G. Buckton, The application of near-infrared of particle size of pharmaceutical raw materials using near- spectroscopy and dynamic vapor sorption to quantify low infrared reflectance spectroscopy, Spectroscopy 1 (1986) amorphous contents of crystalline lactose, Pharm. Res. 18 (2001) 112 – 116.
 I.L. Ikari, H. Martens, T. Isaksson, Determination of particle  S.E. Hogan, G. Buckton, Water sorption/desorption-near IR size in powders by scatter correction in diffuse near-infrared and calorimetric study of crystalline and amorphous raffi- reflectance, Appl. Spectrosc. 42 (1988) 722 – 728.
nose, Int. J. Pharm. 227 (2001) 57 – 69.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143  A. Columbano, G. Buckton, P. Wikeley, A study of the two related pharmaceutical preparations by reflectance near- crystallization of amorphous salbutamol sulphate using water infrared spectroscopy: validation of the procedure, J. Pharm.
vapour sorption and near infrared spectroscopy, Int. J. Pharm.
Biomed. Anal. 22 (2000) 139 – 148.
237 (2002) 171 – 178.
 T. Li, A.D. Donner, C.Y. Choi, G.P. Frunzi, K.R. Morris, A  M. Blanco, A. Villar, Development and validation of a method statistical support for using spectroscopic methods to validate for the polymorphic analysis of pharmaceutical preparations the content uniformity of solid dosage forms, J. Pharm. Sci.
using near infrared spectroscopy, J. Pharm. Sci. 92 (2003) 92 (2003) 1526 – 1530.
823 – 830.
 M. Blanco, J. Coello, H. Iturriaga, S. Maspoch, C. de la  T. Burger, J. Kuhn, R. Caps, J. Fricke, Quantitative Pezuela, Quantitation of the active compound and major determination of the scattering and absorption coefficients excipients in a pharmaceutical formulation by near infrared from diffuse reflectance and transmittance measurements: diffuse reflectance spectroscopy with fibre optical probe, application to pharmaceutical powders, Appl. Spectrosc. 51 Anal. Chim. Acta 333 (1996) 147 – 156.
(1997) 309 – 317.
 K. Molt, F. Zeyen, E. Podpetschnig-Fopp, Quantitative  T. Burger, J. Fricke, J. Kuhn, NIR radiative transfer near infrared spectrometry used for the determination of investigations to characterize pharmaceutical powders and the drug content of tolbutamide tablets, Pharm. Ind. 58 their mixtures, J. Near Infrared Spectrosc. 6 (1998) 33 – 40.
(1996) 847 – 852.
 J.D. Kirsch, J.K. Drennen, Near-infrared spectroscopy:  B.R. Buchanan, M.A. Baxter, T.-S. Chen, X.-Z. Quin, P.A.
applications in the analysis of tablets and solid pharma- Robinson, Use of near-infrared spectroscopy to evaluate an ceutical dosage forms, Appl. Spectrosc. Rev. 38 (1995) active in a film coated tablet, Pharm. Res. 13 (1996) 139 – 174.
616 – 621.
 Pharmaceutical assays, chapter 5, in: E.W. Ciurczak, J.K.
 P. Merckle, K.-A. Kovar, Assay of effervescent tablets by Drennen III (Eds.), Pharmaceutical and Medical Applications near-infrared spectroscopy in transmittance and reflectance of Near-Infrared Spectroscopy, Marcel Dekker, New York/ mode: acetylsalicylic acid in mono and combination for- Basel, 2002, pp. 73 – 105.
mulations, J. Pharm. Biomed. Anal. 17 (1998) 365 – 374.
 M.A. Dempster, J.A. Jones, I.R. Last, B.F. McDonald, K.A.
 A. Eustaquio, P. Graham, R.D. Jee, A.C. Moffat, A.D.
Prebbe, Near-infrared methods for the identification of tablets Trafford, Quantification of paracetamol in intact tablets using in clinical trial supplies, J. Pharm. Biomed. Anal. 11 (1992) near-infrared transmittance spectroscopy, Analyst 123 (1998) 1087 – 1092.
2303 – 2306.
 P.K. Aldridge, R.F. Mushinsky, M.M. Andino, C.L. Evans,  A.D. Trafford, R.D. Jee, A.C. Moffat, P. Graham, A rapid Identification of tablet formulations inside blister packages quantitative assay of intact paracetamol tablets by reflectance by near-infrared spectroscopy, Appl. Spectrosc. 48 (1994) near-infrared spectroscopy, Analyst 124 (1999) 163 – 167.
1272 – 1276.
 P. Corti, G. Ceramelli, E. Dreassi, S. Matti, Near infrared  N. Sondermann, K.-A. Kovar, Screening experiments of transmittance analysis for the assay of solid pharmaceutical ecstasy street samples uring near infared spectroscopy, dosage forms, Analyst 124 (1999) 755 – 758.
Forensic Sci. Int. 106 (1999) 147 – 156.
 Y. Chen, S.S. Thosar, R.A. Forbess, M.S. Kemper, R.L.
 J. Gottfries, H. Depui, M. Fransson, M. Jongeneelen, M.
Rubinowitz, A.J. Shukla, Prediction of drug content and Josefson, F.W. Langkilde, D.T. Witte, Vibrational spectro- hardness of intact tablets using artificial neural network and scopy for the assessment of active substance in metoprolol near-infrared spectroscopy, Drug Dev. Ind. Pharm. 27 (2001) tablets: a comparison between transmission and diffuse 623 – 631.
reflectance near-infrared spectrometry, J. Pharm. Biomed.
 J.-K. Drennen, R.A. Lodder, Non-destructive near-infrared Anal. 14 (1996) 1495 – 1503.
analysis of intact tablets for determination of degradation  S.S. Thosar, R.A. Forbess, N.K. Ebube, Y. Chen, R.L.
products, J. Pharm. Sci. 79 (1990) 622 – 627.
Rubinovitz, M.S. Kemper, G.E. Reier, T.A. Wheatley, A.J.
 G. Reich, Noninvasive and simultaneous monitoring of Shukla, A comparison of reflectance and transmittance near- chemical and physical properties: an analytical approach to infrared spectroscopic techniques in determinung drug lyophilized proteins, sugars and protein/sugar mixtures, content in intact tablets, Pharm. Dev. Technol. 6 (2001) Proc. 3rd World Meeting APV/APGI. Berlin, 3/6 April (2000), pp. 279 – 280.
 J.L. Ramirez, M.K. Bellamy, R.J. Romanach, A novel  J.D. Kirsch, J.K. Drennen, Determination of film-coated method for analyzing thick tablets by near infrared spectro- tablet parameters by near-infrared spectroscopy, J. Pharm.
scopy, Pharm. Sci. Tech. 2 (2001) (article 11, pp. 1-10 ( Biomed. Anal. 13 (1995) 1273 – 1281.
 K.M. Morisseau, C.T. Rhodes, Near-infrared spectroscopy as  M. Blanco, J. Coello, A. Eustaquio, H. Iturriaga, S. Maspoch, a nondestructive alternative to conventional hardness testing, Development and validation of a method for the analysis of a Pharm. Res. 14 (1997) 108 – 111.
pharmaceutical preparation by near-infrared diffuse reflec-  J.D. Kirsch, J.K. Drennen, Nondestructive tablet hardness tance spectroscopy, J. Pharm. Sci. 88 (1999) 551 – 556.
testing by near-infrared spectroscopy: a new and robust  M. Blanco, A. Eustaquio, J.M. Gonza´lez, D. Serrano, spectral best-fit algorithm, J. Pharm. Biomed. Anal. 19 Identification and quantification assays for intact tablets of (1999) 351 – 362.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143  N.K. Ebube, S.S. Thosar, R.A. Roberts, M.S. Kemper,  R.G. Buice Jr., T.B. Gold, R.A. Lodder, G.A. Digenis, R.L. Rubinowitz, D.L. Martin, G.E. Reier, T.A. Wheatley, Determination of moisture in intact gelatin capsules by near- A.J. Shukla, Application of near-infrared spectroscopy infrared spectroscopy, Pharm. Res. 12 (1995) 161 – 163.
(NIRS) for nondestructive analysis of powders and tablets:  O. Berntsson, G. Zachrisson, G. O ¨ stling, Determination of simultaneous determination of avicel type, crushing moisture in hard gelatin capsules using near-infrared spectro- strength and lubricant concentration of intact tablets by scopy: applications to at-line process control of pharmaceu- diffuse reflectance technique, Pharm. Dev. Technol. 4 tics, J. Pharm. Biomed. Anal. 15 (1997) 895 – 900.
(1999) 19 – 26.
 G. Reich, Effect of drying on structure and properties of  G. Reich, Use of NIR transmission spectroscopy for non- gelatin capsule shells: a study with model films, Pharm. Ind.
destructive determination of tablet hardness, Proc. 3rd World 57 (1995) 63 – 67.
Meeting APV/APGI, Berlin, 3/6, 2000 (April), pp. 105 – 106.
 T.B. Gold, R.G. Buice Jr., R.A. Lodder, G.A. Digenis,  Y. Chen, S.S. Thosar, R.A. Forbess, M.S. Kemper, R.L.
Determination of extent of formaldehyde-induced crosslink- Rubinowitz, A.J. Shukla, Prediction of drug content and ing in hard gelatin capsules by near infrared spectroscopy, hardness of intact tablets using artificial neural network and Pharm. Res. 14 (1997) 1046 – 1050.
near-infrared spectroscopy, Drug Dev. Ind. Pharm. 27 (2001)  G. Reich, Use of DSC, DMTA and NIRTS to determine hard 623 – 631.
gelatin capsule characteristics associated with brittleness: a  C. Nystrfm, G. Alderborn, M. Duberg, P.-G. Karehill, mechanistic study, Pharm. Sci. Suppl. 1 (1999).
Bonding surface area and bonding mechanism-two important  G. Reich, H. Frickel, A.M. Gil Bernabe, Use of near-infrared factors for the understanding of powder compactability, Drug transmission spectroscopy to detect batch-to-batch variability Dev. Ind. Pharm. 19 (1993) 2143 – 2196.
of hard gelatin capsule shells, Pharm. Sci. Suppl. 1 (1999).
 G. Reich, Nondestructive determination of tablet hardness  G. Reich, Fast and non-destructive quality control of gelatin and consolidation characteristics using NIR transmission capsule shells, Proc. 3rd World Meeting APV/APGI, Berlin, spectroscopy, AAPS Annual Meeting and Exposition, Den- 3/6, 2000 (April), pp. 493 – 494.
ver, 2001, p. M 2283.
 M. Rama, G. Reich, Monitoring structural changes in hard  P.N. Zannikos, W. Li, J.K. Drennen, R.A. Lodder, Spectro- and soft gelatin films and capsules using NIR transmission scopic prediction of the dissolution rate of carbamazepine and reflectance spectroscopy, Proc. 3rd World Meeting APV/ tablets, Pharm. Res. 8 (1991) 974 – 978.
APGI, Berlin, 3/6, 2000 (April), pp. 507 – 508.
 J.K. Drennen, R.A. Lodder, Qualitative analysis using near-  G. Reich, NIR spectroscopy—an alternative approach to infrared spectroscopy. A comparison of discriminant methods assess the mechanical performance of hard gelatin in dissolution testing, Spectroscopy 8 (1991) 34 – 39.
capsules, AAPS Annual Meeting and Exposition, Denver,  G. Reich, H. Frickel, Use of transmission spectroscopy to 2001, p. M 2233.
determine physical and functional film coat properties on  G. Reich, Mechanism and optimization of plasticizers in soft tablets, Proc. Int. Symp. Control. Release Bioact. Mater. 26 gelatin capsules, Pharm. Ind. 56 (1994) 915 – 920.
(1999) 905 – 906.
 G. Reich, Effect of sorbitol specification on structure and  G. Reich, H. Frickel, NIR spectroscopy—a rapid method to properties of soft gelatin capsule shells, Pharm. Ind. 58 evaluate gastroresistance and drug release kinetics of film- (1996) 941 – 946.
coated tablets, Proc. 3rd World Meeting APV/APGI, Berlin,  T.B. Gold, R.G. Buice Jr., R.A. Lodder, G.A. Digenis, 3/6, 2000 (April), pp. 627 – 628.
Detection of formaldehyde-induced crosslinking in soft  H. Frickel, G. Reich, NIR spectroscopy of film-coated elastic gelatin capsules using near-infrared spectrophotom- tablets—fast and nondestructive evaluation of film coat etry, Pharm. Dev. Technol. 3 (1998) 209 – 214.
uniformity and drug release kinetics, Proc. Int. Symp.
 J. Mqller, G. Reich, Importance of Wavelength Selection in Control. Release Bioact. Mater. 27 (2000) 740 – 741.
Quantitative Near-Infrared Spectroscopy of Polyols in Cap-  H. Frickel, G. Reich, Performance optimization of NIR sule Formulations, ICNIRS, Cordoba, 2003, p. 6.7.
transmission spectroscopy of coated tablets—a light-guided  J. Mqller, G. Reich, A comparison of multivariate regression journey, Proc. 3rd World Meeting APV/APGI, Berlin, 3/6, models in quantitative near-infrared spectroscopy, Poster at 2000 (April), pp. 625 – 626.
DPhG Meeting of PhD students, Frankfurt, 2002.
 H. Frickel, G. Reich, Nondestructive monitoring of  J. Mqller, G. Reich, Influence of wavelength selection in physical and biopharmaceutical film coat properties on quantitative near-infrared spectroscopy, Arch. Pharm. 335 (1) tablets, AAPS Annual Meeting and Exposition, Denver, 2001, p. M 2309.
 J. Mqller, G. Reich, Determination of plasticizer content in  H. Frickel, P. Aebi, G. Fricker, G. Reich, Real thickness of softgel capsule formulations using near-infrared spectroscopy Eudragit-based film coats on tablet cores—nondestructive and multivariate regression methods, AAPS Annual Meeting analysis with near-infrared transmission spectroscopy, and Exposition, Toronto, 2002, p. W 5343.
Pharm. Sci. Suppl. 1 (1998) 22.
 M.S. Kamat, R.A. Lodder, P.P. DeLuca, Near-infrared  R.A. Lodder, M. Selby, G.M. Hieftje, Detection of capsule spectroscopic determination of residual moisture in lyophi- tampering by near-infrared reflectance analysis, Anal. Chem.
lised sucrose through intact glass vials, Pharm. Res. 6 (1989) 59 (1987) 1921 – 1930.
961 – 965.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143  I.R. Last, K.A. Prebbe, Suitability of near-infrared methods matrices, AAPS Annual Meeting and Exposition, Toronto, for the determination of moisture in a freeze-dried injection 2002, p. T 3355.
product containing different amounts of active ingredient,  T. Schfnbrodt, G. Reich, Protein Release from Biodegradable J. Pharm. Biomed. Anal. 11 (1993) 1071 – 1076.
PLGA Tablets Monitored by Near Infrared Spectroscopy,  J.A. Jones, I.R. Last, B.F. Macdonald, K.A. Prebbe, Develop- ICNIRS, Cordoba, 2003, p. P.6.4.
ment and transferability of near-infrared methods for  H.W. Siesler, Application to industrial process control, in: determination of moisture in a freeze-dried injection product, H.W. Siesler, Y. Ozaki, S. Kawata, H.M. Heise (Eds.), Near- J. Pharm. Biomed. Anal. 11 (1993) 1227 – 1231.
Infrared Spectroscopy, Wiley-VCH Verlag GmbH, Wein-  M.W.J. Derksen, P.J.M. van de Oetelaar, F.A. Maris, The heim, Germany, 2002, pp. 247 – 268.
use of near-infrared spectroscopy in the efficient prediction  H.W. Ward II, S.S. Sekulic, M.J. Wheeler, G. Taber, F.J.
of a specification for the residual moisture content of a Urbanski, F.F. Sistare, T. Norris, P.K. Aldridge, On-line freeze-dried product, J. Pharm. Biomed. Anal. 17 (1998) determination of reaction completion in a closed-loop 473 – 480.
hydrogenator using NIR spectroscopy, Appl. Spectrosc. 52  T.P. Lin, C.C. Hsu, Determination of residual moisture in (1998) 17 – 21.
lyophilized protein pharmaceuticals using a rapid and non-  C. Coffey, B.E. Cooley, D.S. Walker, Real time quantitaion invasive method: near infrared spectroscopy, J. Pharm. Sci.
of a chemical reaction by fiber optic near-infrared spectro- Technol. 56 (2002) 196 – 205.
scopy, Anal. Chim. Acta 395 (1999) 335 – 341.
 L. Sukowski, M. Ulmschneider, Near Infrared Spectroscopy:  T. Norris, P.K. Aldridge, S.S. Sekulic, Determination of end- High Speed Noninvasive Qualification of Lyophilized Vials, points for polymorph conversions of crystalline organic ICNIRS, Cordoba, 2003, p. O.6.1.
compounds using on-line near-infrared spectroscopy, Analyst  G. Reich, Noninvasive and simultaneous monitoring of 122 (1997) 549 – 552.
chemical and physical properties: an analytical approach to  K. De Braekeleer, R. De Maesschalk, P.A. Hailey, D.C.A.
lyophilized proteins, sugars and protein/sugar mixtures, Sharp, D.L. Massart, On-line application of the orthogonal Proc. 3rd World Meeting APV/APGI, Berlin 3/6, 2000 projection approach (OPA) and the soft independent model- (April), pp. 279 – 280.
ling of class analogy approach (SIMCA) for the detection of  K. K7lkert, G. Reich, Near infrared spectroscopic evaluation the end point of a polymorph conversion reaction by near of stress-induced structural changes of proteins in solid and infrared spectroscopy (NIR), Chemom. Intell. Lab. Syst. 46 solution state, Proc. 3rd World Meeting APV/APGI, Berlin 3/ (1999) 103 – 116.
6, 2000 (April), pp. 335 – 336.
 H. Chung, M.A. Arnold, M. Rhiel, D.W. Murhammer,  K. Murayama, B. Czarnik-Matusewicz, Y. Wu, R. Tsenkova, Simultaneous measurements of glucose, glutamine, ammo- Y. Ozaki, Comparison between conventional spectral analysis nia, lactate and glutamate in aqueous solution by near- methods, chemometrics, and two-dimensional correlation infrared spectroscopy, Appl. Spectrosc. 50 (1996) 270 – 276.
spectroscopy in the analysis of near-infrared spectra of  K.S.Y. Yeung, M. Hoare, N.F. Thornhill, T. Williams, protein, Appl. Spectrosc. 54 (2000) 978 – 985.
J.D. Vaghjiani, Near-infrared spectroscopy for bioprocess  R.L. Brashear, D.R. Flanagan, P.E. Luner, J.J. Seyer, M.S.
monitoring and control, Biotechnol. Bioeng. 63 (1999) Kemper, Diffuse reflectance near-infrared spectroscopy as a 684 – 693.
nondestructive analytical technique for polymer implants and  M.R. Riley, H.M. Crider, The effect of analyte concentration microspheres, PharmSci 1 (1998) 20.
range on measurement errors obtained by NIR spectroscopy,  R.L. Brashear, D.R. Flanagan, P.E. Luner, J.J. Seyer, M.S.
Talanta 52 (2000) 473 – 484.
Kemper, Diffuse reflectance near-infrared spectroscopy as a  H.J. Venables, J.I. Wells, Powder mixing, Drug Dev. Ind.
nondestructive analytical technique for implants, J. Pharm.
Pharm. 27 (2001) 599 – 612.
Sci. 88 (1999) 1348 – 1353.
 E.W. Ciurczak, Pharmaceutical mixing studies using near-  G. Reich, Use of DSC and NIR spectroscopy to study the infrared spectroscopy, Pharm. Technol. 15 (1991) 140 – 145.
hydration, degradation and drug release behavior of PLA/  F.C. Sanchez, J. Toft, B. van den Bogaert, D.L. Massart, S.S.
PLGA microparticles and films with free and blocked Dive, P. Hailey, Monitoring powder blending by NIR carboxylic end groups, Proc. Int. Symp. Control. Release spectroscopy, Fresenius J. Anal. Chem. 352 (1995) 771 – 778.
Bioact. Mater. 27 (2000) 642 – 643.
 D.J. Wargo, J.K. Drennen, Near-infrared spectroscopic  G. Reich, T. Schfnbrodt, Fast and noninvasive near infrared characterization of pharmaceutical powder blends, J. Pharm.
analysis of PLA and PLGA microparticles, films and tablets, Biomed. Anal. 14 (1996) 1415 – 1423.
Proc. Int. Symp. Control. Release Bioact. Mater. 28 (2001)  S.S. Sekulic, H.W. Ward, D.R. Brannegan, E.D. Stanley, C.L.
387 – 388.
Evans, S.T. Sciavolino, P.A. Haily, P.K. Aldridge, Online  T. Schfnbrodt, G. Reich, Near infrared spectroscopy-a novel monitoring of powder homogeneity by near-infrared spectro- approach to monitor drug release kinetics from biodegradable scopy, Anal. Chem. 68 (1996) 509 – 513.
PLGA tablets, Proc. 4th World Meeting ADRITELF/APGI/  P.A. Hailey, P. Doherty, P. Tapsell, T. Oliver, P.K. Aldridge, APV, Florence, 8/11, 2002 (April), pp. 303 – 304.
Automated system for the on-line monitoring of powder  T. Schfnbrodt, S. Mohl, G. Winter, G. Reich, Nondestruc- blending processes using near-infrared spectroscopy, tive quantification of lyophilised protein drugs in lipid J. Pharm. Biomed. Anal. 14 (1996) 551 – 559.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143  S.S. Sekulic, J. Wakeman, P. Doherty, P.A. Hailey, Auto-  S. Watano, H. Takashima, Y. Sato, T. Yasutomo, K.
mated system for the on-line monitoring of powder blending Miyanami, Measurement of moisture content by IR sensor processes using near-infrared spectroscopy: Part II. Qualita- in fluidized bed granulation: effects of operating variables on tive approaches to blend evaluation, J. Pharm. Biomed. Anal.
the relationship between granule moisture content and 17 (1998) 1285 – 1309.
absorbance IR spectra, Chem. Pharm. Bull. 44 (1996)  R. De Maesschalck, F. Cuesta Sanchez, D.L. Massart, P.
1267 – 1269.
Doherty, P. Hailey, On-line monitoring of powder blending  P. Frake, D. Greenhalgh, S.M. Grierson, J.M. Hempemstall, with near-infrared spectroscopy, Appl. Spectrosc. 52 (1998) D.R. Rud, Process control and endpoint determination of a 731 – 752.
fluid bed granulation by application of near infra-red  C. Ufret, K. Morris, Modeling of powder blending using on- spectroscopy, Int. J. Pharm. 151 (1997) 75 – 80.
line near-infrared measurements, Drug Dev. Ind. Pharm. 27  S.G. Goebel, K.-J. Steffens, On-line measurement of mois- (2001) 719 – 729.
ture and particle size in the fluidized-bed processing with  N.-H. Duong, P. Arratia, F. Muzzio, A. Lange, J. Timmer- near-infrared spectroscopy, Pharm. Ind. 60 (1998) 889 – 895.
mans, S. Reynolds, A homogeneity study using NIR  J. Rantanen, S. Lehtola, P. R7met, J.-P. Mannerma, J.
spectroscopy: tracking magnesium stearate in Bohle bin- Yliruusi, On-line monitoring of moisture content in a blender, Drug Dev. Ind. Pharm. 29 (2003) 679 – 687.
instrumented fluidized bed granulator with a multi-cannel  A.S. El-Hagrasy, H.R. Morris, F. D'Amico, R.A. Lodder, NIR moisture sensor, Powder Technol. 99 (1998) 163 – 170.
J.K. Drennen III, Near-infrared spectroscopy and imaging for  J. Rantanen, O. Antikainen, J.-P. Mannerma, J. Yliruusi, Use the monitoring of powder blend homogeneity, J. Pharm. Sci.
of the near-infrared reflectance method for measurement of 90 (2001) 1298 – 1307.
moisture content during granulation, Pharm. Dev. Technol. 5  O. Berntsson, L.-G. Danielsson, S. Folestad, Characterization (2000) 209 – 217.
of diffuse reflectance fiber probe sampling in moving solids  J. Rantanen, E. R7sanen, J. Tenhunen, M. K7ns7koski, J.-P.
using a Fourier transform near-infrared spectrometer, Anal.
Mannerma, J. Yliruusi, Inline moisture measurement during Chim. Acta 431 (2001) 125 – 131.
granulation with a four-wavelength near infrared sensor: an  O. Berntsson, Characterization and application of near infra- evaluation of particle size and binder effects, Eur. J. Pharm.
red reflection spectroscopy for quantitative process analysis Biopharm. 50 (2000) 271 – 276.
of powder mixtures. Doctoral Thesis, Stockholm (2001).
 G. Buckton, E. Yonemochi, W.L. Yoon, A.C. Moffat, Water  O. Berntsson, L.G. Danielsson, B. Lagerholm, S. Folestad, sorption and near IR spectroscopy to study the differences Quantitative in-line monitoring of powder blending by near between microcrystalline cellulose and silicified microcrys- infrared reflection spectroscopy, Powder Technol. 123 (2002) talline cellulose before and after wet granulation, Int. J.
185 – 193.
Pharm. 181 (1999) 41 – 47.
 J.G. White, On-line moisture detection for a microwave  H.M. Derbyshire, Y. Feldman, C.R. Bland, J. Broadhead, G.
vacuum dryer, Pharm. Res. 11 (1994) 728 – 732.
Smith, A study of the molecular properties of water in  S.C. Harris, D.S. Walker, Quantitative real-time monitoring hydrated mannitol, J. Pharm. Sci. 91 (2002) 1080 – 1088.
of dryer effluent using fiber optic near infrared spectroscopy,  A. Mica, T. Yajima, S. Itai, Prediction of suitable amount of J. Pharm. Sci. 89 (2000) 1180 – 1186.
water addition for wet granulation, Int. J. Pharm. 195 (2000)  K.R. Morris, J.G. Stowell, S.R. Byrn, A.W. Placette, T.D.
Davis, G.E. Peck, Accelerated fluid-bed drying using NIR  D.J. Wargo, J.K. Drennen, AAPS Annual Meeting, Seattle, monitoring and phenomenological modeling, Drug Dev. Ind.
WA, 1996 (paper # PT6116).
Pharm. 26 (2000) 985 – 988.
 G. Radtke, K. Knop, C. Lippold, In-process control of direct  P.L. Wildfong, A.-S. Samy, J. Corfa, G.E. Peck, K.R. Morris, pelletisation in the rotary fluidized bed using NIR spectro- Accelerated fluid-bed drying using NIR monitoring and scopy, NIR News 10 (1999) 4 – 12.
phenomenological modeling: method assessment and for-  C. Roeseler, K.-A. Kovar, In-process control, Eur. Pharm.
mulation suitability, J. Pharm. Sci. 91 (2002) 631 – 639.
Rev. 2 (2003) 63 – 67.
 G.X. Zhou, Z. Ge, J. Dorwart, B. Izzo, J. Kukura, G. Bicker,  J.D. Kirsch, J.K. Drennen, Near-infrared spectroscopic J. Wyvratt, Determination and differentiation of surface and monitoring of the film coating process, Pharm. Res. 13 bound water in drug substances by near infrared spectro- (1996) 234 – 237.
scopy, J. Pharm. Sci. 92 (2003) 1058 – 1065.
 M. Andersson, M. Josefson, F. Langkilde, K.-G. Wahlund,  M. Brqlls, S. Folestad, A. Spare´n, A. Rasmuson, In-situ near- Monitoring of a film coating process for tablets using near infrared spectroscopy monitoring of the lyophilization infrared reflectance spectroscopy, J. Pharm. Biomed. Anal.
process, Pharm. Res. 20 (2003) 494 – 499.
20 (1999) 27 – 37.
 K. List, K.-J. Steffens, Supervising and controlling of  M. Andersson, S. Folestad, J. Gottfries, M.O. Johansson, M.
granulation processes by near-infrared spectroscopy, Pharm.
Josefson, K.-G. Wahlund, Anal. Chem. 72 (2000) 2099 – 2108.
Ind. 58 (1996) 347 – 353.
 M. Andersson, B. Holmquist, J. Lindquist, O. Nilsson, K.-G.
 S. Watano, K. Terashita, K. Miyanami, Posture feed-back Wahlund, Analysis of film coating thickness and surface area control and process automation in fluidized bed granulation, of pharmaceutical pellets using fluorescence microscopy and Adv. Powder Technol. 3 (1992) 255 – 265.
image analysis, J. Pharm. Biomed. Anal. 22 (2000) 325 – 339.
G. Reich / Advanced Drug Delivery Reviews 57 (2005) 1109–1143  G. Reich, Potential of attenuated total reflection infrared and infrared spectral imaging, Near Infrared Spectrosc. 11 near-infrared imaging for quality assurance/quality control of (2003) 71 – 81.
solid pharmaceutical dosage forms, Pharm. Ind. 64 (2002)  J.R. Mansfield, M.S. Sowa, C. Majzels, C. Collins, E.
870 – 874.
Clontis, H.H. Mantsch, Near infrared spectroscopic reflec-  T. Herkert, H. Prinz, K.-A. Kovar, One hundred percent tance imaging: supervised vs. unsupervised analysis using online identity check of pharmaceutical products by near- an art conservation application, Vibr. Spectrosc. 19 (1999) infrared spectroscopy on the packaging line, Eur. J. Pharm.
Biopharm. 51 (2001) 9 – 16.
 F. Clarke, NIR microscopy: utilisation from research through  F. van der Meer, S.M. De John (Eds.), Imaging Spectrometry: to full-scale manufacturing, Presentation No. 6 at the Euro- Basic Principles and Prospective Applications, Kluwer pean Conference on Near Infrared Spectroscopy, Heidelberg, Academic Publishers, 2000.
 G.H. Bearman, R.M. Levenson, D. Cabib (Eds.), Spectral  E.N. Lewis, J.E. Carroll, F. Clarke, NIR imaging: a near Imaging: Instrumentation, Applications, and Analysis, SPIE infrared view of pharmaceutical formulation analysis, NIR Publications, 2000.
News 12 (2001) 16 – 18.
 F.W. Koehler IV, E. Lee, L.H. Kidder, E.N. Lewis, Near  G. Reich, Potential of ATR-IR and NIR spectroscopic infrared spectroscopy: the practical chemical imaging sol- imaging for quality assurance of solid pharmaceutical dosage ution, Spectrosc. Eur. 14/3 (2002) 12 – 19.
forms, Proc. 4th World Meeting ADRITELF/APGI/APV,  A. Kulcke, C. Gurschler, G. Spfck, R. Leitner, M. Kraft, Florence, April 8-11, 2002, pp. 291 – 292.
On-line classification of synthetic polymers using near
Corporate Information Biographical Details of Directors and Senior Management Chairman's Statement Corporate Governance Report Management Discussion and Analysis Directors' Report Independent Auditor's Report Consolidated Financial Statements Consolidated Income Statement Consolidated Statement of Comprehensive Income Consolidated Statement of Financial Position