Why do some batches of tablets prepared using the same ingredients and following the same process show different dissolution profiles? This question remained largely unanswered until about five years ago, because the chemical composition measured with traditional laboratory equipment, such as HPLC (high-performance liquid chromatography), often could not explain the dissolution failure. However, with the emergence of near infrared (NIR) chemical imaging systems in troubleshooting laboratories, the situation has changed, because this new technology provides access to the structure, or the arrangement of chemical entities, within solid dosage form products.
It was quickly realized that simple descriptors, such as components distribution uniformity and aggregate size, could be correlated with dissolution data and, more importantly, often help trace back the cause of the performance differences to particular manufacturing variables. This article describes the technique of NIR chemical imaging with an emphasis on how this laboratory instrument is used in advanced troubleshooting of dissolution failures in tablets and granules.
The recent efforts in the pharmaceutical industry to improve manufacturing efficiency, combat counterfeiting, understand products and processes, and develop ever more complex drug delivery systems have emphasized the need for NIR chemical imaging as a technology that can nondestructively provide new, valuable, and detailed information about finished solid dosage forms and product intermediates. The novelty of the information can be summarized in one sentence: NIR chemical imaging combines spectroscopy with digital imaging and produces images and statistics that describe the spatial distribution of sample components in the tablet.
How does this relate to dissolution? The answer is both simple and complicated. Dissolution test results are affected by changes in dosage strength, the domain size distribution of that material, and the proximity of the other components in the formulation that may affect solubility. NIR chemical imaging can measure these parameters of “how much” and “where,” and best of all, can do it in the intact tablet. The latter is important because the same tablet can be subjected to dissolution testing, making it possible to establish links between dissolution behavior and the specific composition and structure of the same tablet used in the dissolution test. This is preferred over performing two destructive tests on different tablets selected from a batch and statistically correlating the relative results. Dissolution testing is performed routinely, and unfortunately results in the complete destruction of the sample. Recording both the chemical composition and the structure of the tablet before performing dissolution testing eliminates this problem.
Near Infrared Spectroscopy and Imaging
Near infrared spectroscopy is a well-established technique for raw material identification and the quantitative analysis of tablets and blends in the pharmaceutical industry. The NIR spectral range spans from 800 nm to 2,500 nm, and the measurement probes the absorption of radiation by different molecular functional groups. In this spectral range, most absorption bands arise from combination and overtone bands of O-H, N-H, and C-H stretching and bending vibrations. One of the important advantages of the NIR spectral range is a fairly long depth of penetration, a significant benefit to analyze intact samples or samples contained in packaging material. A typical NIR spectrometer design includes a source of NIR radiation, a means of sample presentation, a device for separating wavelengths to produce a spectrum, and an NIR detector. The apparatus for performing the spectroscopy is often a spectrograph, but it also can be an interferometer.
NIR imaging is based on the same principles implemented differently. One type of implementation has the single detector replaced with a focal plane array detector (i.e., a camera), and the wavelength selection is performed using a tunable filter, which provides the advantage of random access to any single or subset of wavelengths in the spectral range. The technology and various implementations have been discussed elsewhere and will not be detailed further.5 Because NIR spectroscopy and NIR imaging use the same spectroscopic method, the images are generated using optical absorptions arising from the intrinsic molecular structure without tagging or staining the sample.
With vibrational spectroscopy at its base, chemical imaging matches the qualitative and quantitative analytical information content provided by traditional bulk NIR spectroscopy, but the imaging approach adds significant insight into the functionality of heterogeneous materials because it “sees” where the chemical species are positioned nondestructively in the finished product or powder blend.
This approach has many key practical differences from traditional NIR spectroscopy.6 A lower detection limit is obtained with NIR imaging compared to traditional NIR spectroscopy because each detector probes a much smaller sample volume; another interesting advantage is that generally simpler multivariate analysis models suffice to gain a tremendous amount of insight into component concentration and distribution.3 Together, the increased magnification provided by the microscope optics and the chemical specificity of NIR spectroscopy combine to produce a picture of the distribution of components.
Figure 1 shows an example of a chemical image of two similar OTC pain medication tablets from two different pharmaceutical companies. This false color image highlights the distribution of chemical species; red, green, and blue each indicate agglomerations, or clusters, of high concentration of a specific ingredient, while the very dark areas contain average to low abundance of all three components. These chemical images contain a NIR spectrum at each pixel, and once processed into images with chemically relevant contrast, traditional image analysis tools can be used to gather quantitative information about the differences in the observed structures, such as cluster size and relative component distribution.
The quantity of active pharmaceutical ingredient (API) in a pharmaceutical tablet is typically measured using UV Visible spectroscopy or chromatography. More recently, traditional single-point NIR spectroscopy also has been used for this purpose because it is fast and, when calibrated properly, very accurate. The quantity of excipients often is not monitored and is expected to be relatively equal in all tablets within a batch.
Figure 1: False-color image of two OTC tablets of pain medication with similar chemical composition but different blending characteristics8, 9, 10
Figure 2: Schematic representation of varying content distribution for a common content value
When a problem occurs, hypotheses to explain the dissolution failure, such as an influence from variations in the quantity of some excipients or the structure of the product in the finished tablets, often are not evaluated because traditional wet chemistry and spectroscopic methods prove incapable, impractical, or too onerous to access that information. For example, do API agglomerations surrounded by a disintegrant and similar agglomerations surrounded by a plasticizer in an otherwise similar matrix yield different dissolution rates? Do tablets or granules containing more disintegrant on the outer edges dissolve faster than samples that have the same amount concentrated in the core? This is where NIR chemical imaging has proven an invaluable tool. Various other application examples can be found in the literature, including the assessment of coatings on granules, blend homogeneity, blister pack screening,4 and counterfeit detection and characterization.2
Novel Information Accessed With NIR Chemical Imaging
Experience has shown that the factors most often found by NIR chemical imaging to correlate with dissolution failure are API distribution and key excipient distribution.
API Distribution
Active pharmaceutical ingredient distribution in finished tablets directly affects dissolution test results in some products. Figure 2 shows a schematic that is intended to illustrate that samples A, B, C, and D contain equivalent amounts of excipients (white) and active ingredient (gray). Using a conventional assay, such as HPLC or NIR spectroscopy, all four samples would be determined to be compositionally identical. However, using an imaging system, like that shown in Figure 3, would result in different spectra being measured by different pixels of the NIR array detector, resulting in four completely different measurements for all four samples.
The most important question at this point is: When does cluster size start to matter? This is a question that many pharmaceutical formulators try to answer as part of the product and process understanding stages of their PAT (process analytical technology) implementation plan. However, it also is a question that too often gets left for later, when dissolution failure occurs and there is a need to know what went wrong.
In some products, fairly modest increases in cluster size may affect dissolution, while in others, there must be many-fold increases before the dissolution profile changes. Many formulations produce aggregates or clusters of ingredients, and changes in conditions or sources of ingredients can affect the cluster size through various processes, such as changes in the flow properties of the powders or in mixing and compaction behaviors. For example, changes in humidity conditions in a blender may result in larger clusters of an API.
Figure 3: Schematic representation of the data acquisition mechanism in diffuse reflectance NIR chemical imaging
How can one determine the relationship between structural changes and dissolution profiles? The size tolerances for API cluster in different products can be established by measuring agglomerate size distribution in hundreds of representative samples and documenting dissolution behaviors of either all tablets, or only those that exhibit size distributions of clusters that differ. The measurement of agglomerate size distribution of API is simple; the abundance of API is measured at every pixel of the chemical image of a surface of the tablet using a selected statistical method applied to the spectral image. Clusters of pixels exhibiting a higher concentration of API are then isolated and the area they cover is measured. The speed of data acquisition (about one minute per sample) is an important advantage of NIR chemical imaging using a camera, because in combination with automation of data processing, it makes it possible to analyze a large number of units and thereby obtain conclusive information rather than relying on indicative data obtained from a small sample size imposed by a time-consuming technique. The fact that the data is acquired in diffuse reflectance (Figure 3) also is a benefit because this noncontact, nondestructive approach preserves the tablets for further wet chemistry tests. This way, the dissolution profile can be correlated directly to the component distribution measured in the same tablets. Diffuse reflectance is commonly used in conventional NIR spectroscopy and results in a complex interaction of scattering and absorption.7 It is the absorption component that gives rise to the unique spectral signature identifying the chemical composition at each pixel. The two tablets illustrated in Figure 1 were purchased from a standard distribution channel and therefore are expected to have passed dissolution testing. They, however, visibly differ in terms of the distribution of their components represented by the red, green, and blue colors.
This image was created by assigning the respective colors to pixels with abundance in the three particular components greater than a set threshold. In tablet A, the blend generally is more homogeneous because it presents fewer colored spots indicative of clusters. Quantitative analysis of the clusters in the form of the measurement of their size and number in each tablet provides a statistical means of comparing these samples beyond the pretty picture (procedure described in Reference 4). This is an example of a product where there probably needs to be many-fold differences in cluster size to cause dissolution failure.
Other products are more sensitive. In 2005, the FDA’s Office of Compliance published results indicating a link between API mixing and excipients differences as measured by NIR chemical imaging and dissolution failure in pharmaceutical tablets purchased on the Internet.10 While there are not really any general rules to predict which formulations are more sensitive to cluster sizes, this measurement is often significantly more relevant for lower-dose products because greater numbers of larger cluster sizes are not only indicative of possible dissolution problems, but also of possible assay failures.
Simply put, if the API of a very low-dose product is present in large clusters, there is an increased chance that some tablets contain a few clusters too many or too few. Either case can represent significant deviation from the prescribed dosage. In this case, cluster size measurements using NIR chemical imaging would have been an invaluable tool had it been available during formulation development.
Figure 4: False-color image of pharmaceutical granules highlighting the difference in content and content distribution between granules. These granules also were measured and their size and shape correlated to a particular plasticizer content uniformity issue affecting dissolution.
Measuring Key Excipient Distribution
The NIR image cube contains information about the chemistry present at all pixels, including the spectral contributions from organic excipients in addition to the API. The approach described earlier to isolate clusters of API is equally useful to isolate agglomerations of excipients, as seen in Figure 1. Classification partial least squares (PLS) calibrations often are used for this purpose, while principal component analysis sometimes is sufficient to measure the distribution of key ingredients. In the next example, a PLS model was used to analyze granules from different batches with slow and fast dissolution patterns.
Figure 4 shows the distribution of three main components in a false color image. The increased presence of green clusters is obvious in sample A relative to sample B; however, when a large number of granules was analyzed, this difference did not exist between the means of the two groups. This observation emphasizes the importance of analyzing a larger sample size to obtain more conclusive results. Taking advantage of the fact that these were measured by an imaging approach, a comparison was made between the chemical composition of many granules from different batches in relation to their size and shape. In this experiment, larger granules with higher elongation were shown to contain an increased proportion of clusters of a plasticizer (shown in green). As a consequence, when a sample contained more of these larger granules with a heterogeneous distribution of plasticizer, the batch showed a slower dissolution profile.9
Many examples of NIR chemical imaging measurements of excipients distribution can be found in the literature. The objective always is quite simple: Once specific clustering phenomena have been measured, it becomes possible to trace their presence back to certain critical manufacturing variables and possibly correct a problem or modify the process to obtain the desired outcome. For example, compaction force was determined to affect dissolution by forming disintegrant clusters that were 40% larger in a product. Root cause analysis of a processing issue also traced the problem to increased cluster size of a polymer component of the tablets.1 A granulation issue was solved by adding a premixing step to avoid components agglomeration observed by NIR chemical imaging of granules.8
Figure 5: False-color image of sections of three tablets showing A) uniform coating, B) erosion of the coating, and C) chipping and erosion.
Similarly, coatings also can be measured for both uniformity in chemical composition and thickness. For example, Figure 5 shows images from sections of three tablets, two of them having defective coatings. The tablet on the left (A) is intact, while tablet B displays a gradual change in thickness in one area, as measured by the change in the spectral contribution from the coating material.
The tablet on the right (C) has a chipping problem. In the later case, the area where there is no coating displays spectra without any contribution from the coating. These phenomena consequently display very different coating distribution curves, and the mean, standard deviation, and skew of these curves normally are used to characterize the samples rather than the images. By looking at the data in Table 1, it is clear that while the mean coating thickness is approximately equivalent, the coating thickness distribution, as measured by the skew, is very different and is a strong indicator of significant non-uniformities in the coating distribution. Again, this phenomenon would be completely overlooked using a simple technique such as average weight gain.
Photo 1: The SyNIRgi near infrared chemical imaging system from Malvern Instruments
Table 1: Statistics of the coating layer distribution in the tablets of Figure 5
Conclusion
Understanding the impact of particular structural characteristics on the performance of pharmaceutical solid dosage forms, then establishing the critical manufacturing variables that impact these structural characteristics, is an integral part of product and process understanding and an important aspect of the FDA’s QbD (quality by design) initiative. NIR chemical imaging is a versatile and powerful tool to measure the spatial distribution of pharmaceutical ingredients and help provide a greater understanding of out-of-specifications situations.
References
1. Clarke, F. (2004) “Extracting process-related information from pharmaceutical dosage forms using near infrared microscopy,” Vibrational Spectroscopy, 34:25-35.
2. Dubois, J., Wolff, J-C, Warrack, J.K., Schoppelrei, J.S., Lewis, E.N. (2007) “NIR Chemical Imaging for Counterfeit Pharmaceutical Products Analysis,” Spectroscopy, 22(2), 40-50.
3. Gendrin, C., Roggo, Y., Collet, C. (2007) “Content uniformity of pharmaceutical solid dosage forms by near infrared hyperspectral imaging: A feasibility study,” Talanta, 73(4):733-741.
4. Lewis, E.N., Schoppelrei, J.W., Lee, E. (2004) “Near-infrared Chemical Imaging and the PAT Initiative,” Spectroscopy, 19(4):26-36.
5. Lewis, E.N., Schoppelrei, J.W., Lee, E., Kidder, L.H. (2005) Near-infrared chemical imaging as a process analytical tool, Process Analytical Technology, Chapter 7, K.A. Bakeev Ed., Blackwell Publishing, 451 p.
6. Lewis, E.N., Kidder, L.H., Lee, E. (2005) “NIR chemical imaging — near infrared spectroscopy on steroids,” NIR News, 16(5):2-4.
7. Olinger, J.M., Griffiths, P.R. (1992) Theory of diffuse reflectance in the NIR region. Handbook of Near-Infrared Analysis, Chapter 3, D.A. Burns and E.W. Ciurczak Eds., Marcel Dekker, NY, 681 p.
8. Roggo, Y., Ulmschneider, M. (2008) Chemical imaging and chemometrics: Useful tools for process analytical technology. Pharmaceutical Manufacturing Handbook: Regulations and Quality, Chapter 4.3, S. Cox Gad Ed., Wiley, NJ, 841 p.
9. Vargas, R., Dubois, J., Lewis, E.N. (2008) Simultaneous size, shape and chemical analysis of pharmaceutical granules. Poster presented at IFPAC 2008.
Posts
This is very detail and informative
ReplyDelete“…the imaging approach adds significant insight into the functionality.” – That is good news! With the incorporation of the imaging system, the quality-checking of raw materials and substances used on each tablet will have a more in-depth perspective and look.
ReplyDelete