Raman/NIR Roundtable

 What are the most notable advantages of using spectroscopic techniques such as NIR and Raman in pharmaceutical analysis?

MK:  Most of the findings on incompliance during FDA inspections involve the handling of raw materials. Even large pharmaceutical companies have received warning letters from the FDA concerning cGMP for raw materials. Using molecular spectroscopy techniques like NIR and Raman provide rapid and accurate identification and statistical verification of raw materials. These instruments are now available in handheld packages and can be used in the warehouse or loading dock, to ensure authenticity and avoid costly manufacturing errors that can lead to recalls or even patient injuries.

GR/CNIRS: There are many advantages of NIR for the analysis of pharmaceuticals. Since it has been in use for many years now, pharmaceutical manufacturers can leverage abundant in-line/on-line process analytical experience from other industries (petrochemical, polymers/materials, chemicals, and food/agriculture). Analyses are highly accurate, with results that may be as reproducible, or even more reproducible than the reference method, depending on “sampling” that is governed by how much of the sample is measured by the spectroscopic method. More than a single analyte or quality attribute can be determined simultaneously. The speed of measurement (hundreds of spectra can be collected in fraction of seconds with some of the newer devices) allowing for many, many more assays, giving a better picture of the uniformity of a product. This also makes it useful for supporting high-throughput screening efforts, and is well-suited for emerging continuous manufacturing technologies. It is non-destructive and doesn’t require contact with the sample. Both NIR (and Raman) allow the dosage form to be examined in its “native” state, giving much more information as to its 1) delivery characteristics (dissolution, polymorphic form, etc.) and 2) the process parameters (for upgrade via QbD). Measurements can be done in situ, or in settings other than in a laboratory, for example on a loading dock to check incoming raw materials and the ability for real time, on-line analysis during drug substance and drug product manufacturing. Also analysis can be performed if the sample is in a transparent container or a blister-packaging. Samples require little or no preparation. NIR has limited maintenance, low cost per analysis, adaptability to most manufacturing situation and locations, and is highly customizable as is evident in the myriads of devices now available in the market place.

Because NIR is sensitive to both the sample matrix and the active ingredients, it is possible to detect changes in the matrix (particle size, contaminations, etc.) that are totally invisible to other traditional methods (HPLC, Titrimetric). This sensitivity allows NIR to be very versatile as physical states of the sample can be detected in addition to chemical signatures. NIR is green chemistry friendly, as it doesn’t require the use of solvents. In traditional pharmaceutical analyses, HPLC, wet chemistry, titrations, etc., destruction of the sample is the norm and the chemicals need to be bought and disposed of.

SP: The high speed (only seconds) and simplicity (little to no sample preparation) of the analysis are without any doubt the most compelling assets of NIR and Raman spectroscopy in pharmaceutical analysis. When using fiber-optic cables, they might be well-suited tools for the in-line monitoring of relevant quality parameters in production processes by giving real-time information. Both techniques enable non-invasive measurements which do not require contact with the sample nor the process stream, which is also an important advantage, especially for processes such as freeze-drying, where contact with the sample would disturb the process. For me, the most important advantage regarding freeze-drying (approximately 50% of the biopharmaceutical formulations are freeze-dried) is that Raman and NIR are the first tools that may allow a continuous evaluation of the product itself during the freeze-drying process. Because of the different selection rules, NIR and Raman spectroscopy can be considered complementary. NIR requires a change in dipole moment in the vibrating bond for absorption to occur, while Raman scattering occurs by a change in polarizability in the vibrating bond. Via one or both techniques, a wealth of chemical and physical information can be obtained about the sample and the process. Compared to techniques habitually used in the pharmaceutical industry, such as HLPC and MS, the obtained spectra are complex and the instrument selectivity of the NIR and Raman analysis is much lower (they are not separation techniques). Therefore, developing proper spectroscopic methods usually requires multivariate calibration. Capturing the co-variance between the measured signals and the response of interest by partial-least-squares (PLS) is the most popular way to build multivariate calibration models. The effect of the response of interest on the measured signals is preferably known and explained by sound chemical knowledge, yet the calibration itself is performed in a purely data-based, black-box approach. Therefore, it should be noted that the predictive ability of a PLS model will strongly depend on the representativeness of the calibration set, because any correlation will be used, whether it is specific or not. This should be fully kept in mind during method development. Besides developing calibration models, the multivariate nature of the data generated by these instruments allows that a number of informative parameters (scores, loadings, residuals...) can be very useful for exploratory and diagnostic purposes, such as for improving the model, flagging out-of-control samples, understanding why the model does not perform well, where samples differ from others, detecting and understanding outliers, etc. This opportunity is not available when only using standard techniques.

CH & RK: For pharmaceutical analysis, there are several very notable advantages of using NIR spectroscopy over other analytical techniques including analysis in seconds, analyzing samples in-situ, analyzing a large sample size, analysis without sample preparation, remote sampling for in-process analysis and replacement of primary wet chemistry techniques that have consumable costs and use hazardous reagents. The ability of NIR energy to travel without attenuation through fiber optics to probes or flow cells installed in pharmaceutical production processes makes NIR spectroscopy an ideal choice for process monitoring and control of pharmaceutical production processes in a Process Analytical Technology (PAT) framework. Examples of pharmaceutical processes where real-time NIR analysis has been successfully implemented include granulators, dryers, blenders, bioreactors and hot melt extruders. In-process NIR analysis allows production processes to continue to operate without interruption for thieving a sample to be taken for off-line analysis in a laboratory. This results in a significant increase in production efficiency since more material can be produced in a given time period leading to decreased production equipment capital spending. Also by eliminating laboratory analysis there are savings in consumables, elimination of instrumentation maintenance and reduced use of hazardous chemicals by laboratory personnel.

NIR’s ability to analyze samples in-situ, through plastic or glass packaging material is a significant advantage over most other analytical techniques. NIR is widely used for incoming material identification and lyophilized material analysis due to its ability to analyze samples in their original plastic or glass packaging material. For identification and adulteration testing of incoming materials, NIR brings multiple analysis advantages to the table including the ability to analyze bulk samples without sample preparation in their original packaging material making NIR the technique of choice in the pharmaceutical industry.

For quality control testing of tablets and softgels prior to product release, NIR is widely utilized for content uniformity (API%), coating thickness and identification. NIR energy penetrates opaque tablets allowing for successful bulk samples analysis for content uniformity. NIR spectroscopy has been adopted as the technique of choice for high throughput QC screening for content uniformity and other key parameters due to this ability to effectively “see” through the whole tablet.

RC: NIR and Raman have the advantage of sampling through packaging material and rapid response times. In the last ten years, these technologies have become as portable and easy to use as a bar code reader. Ironically, many of these handheld systems now incorporate bar code readers and Blue Tooth technologies. Most handheld NIR and Raman analyzers weigh from five pounds to less than one pound. Handheld innovation has trumped the former key benefit of “no sample prep” when the sample was inherently transported from the loading dock to the lab, or the instrument was wheeled to location on a cart. Today handheld NIR and Raman systems are holstered or located “at” the loading dock.

Like laboratory systems, handhelds require validatable methods and procedures. The systems are easier to employ for those not skilled in the art and method and library development software packages have rapid learning curves for novices. Simple step-by-step method development routines are commonplace and the user is no longer required to translate spectra or interpret complex chemometric curves. One of the biggest challenges in using PCA spectral decomposition models for discriminate analysis is tagging outliers and selecting the proper number of factors in the models. Chemometric software packages now have automated outlier detection and determining the proper number of factors in multivariate discrimination models.

Lastly, the systems are more robust because of the adaptation of military applications that require MilSpec testing. The most recent is MilSpec 810 G which requires users to design chamber test methods to challenge their equipment for long term stability. Standard tests include drop (shock), vibration, dunkability and other field conditions that the equipment is expected to encounter. Many of these tests can transform to NEMA and ISP67 standardization tests. In summary, the NIR and Raman handheld systems are easier to employ and are more robust than ever. They provide very rapid response times which streamline product flow, and allow customers to react quickly to problem areas in manufacturing.

GLR: Spectroscopic tools (e.g., mid infrared, near infrared, Raman, UV-Vis) are common analytical tools applied in the development, understanding, optimization and manufacture of drug substances and dosage forms.

Each spectroscopic technique has inherent advantages (and disadvantages) that can be leveraged as appropriate (just like more traditional techniques) to solve the problem at hand, or provide the information sought in the required timeframe. Advantages of spectroscopic tools include:

  • analysis speed – real time analysis
  • in-situ capabilities (inserting a probe into the process or measuring through a sealed, optically transparent material to get a measurement)
  • non-destructive analysis
  • potential to analyze reactive intermediates and the ability to monitor both chemical and physical aspects of the matrix (NIR)

With the introduction of handheld spectrometers, it became practical to take the instrument to the sample, be it in the warehouse or loading dock for identification of received materials or into the field for counterfeit detection.

During development of the drug substance, spectroscopic tools are invaluable for many tests including raw material ID (using handheld devices in the warehouse), and real time assessments for reactant loss, product formation, polymorphic form, solvent composition during distillation and solvent swaps, drying and particle size. With calibrations, quantitative measurements can be made when appropriate. Informatics are improving to the point where multiple data streams (e.g., MIR, Raman, Calorimetry, Chromatography) can be overlaid for a holistic view of the process in real time.

During development of the dosage form, spectroscopic tools are invaluable for many tests in including raw material ID (using handheld devices in the warehouse), and real time assessments for blend homogeneity, end-point determination, drying, potency, and water content. More advanced (and user friendly) chemometric and modeling tools will allow for relations to be more easily developed between raw material attributes, process parameters and end product attributes such as drug product performance for each dosage form.

Rigaku: Raman spectra are characteristic for most substances, including active pharmaceutical ingredients (API). High quality Raman spectra can be easily be acquired using portable spectrometers. Additionally, Raman spectra are measureable through transparent packaging and drug products need not be exposed to potential contamination. With software-based search and spectral identification tools, analysts build reference libraries for discriminant analysis. The portable spectrometer can be carried directly to receiving warehouse or other locations for sample identification. Bringing analysis to the point of need, these systems provide data with a high degree of chemical specificity in a variety of environments and can significantly cut the time and costs associated with lab testing, ensuring product quality and authenticity before the products reach the consumer. These tools are seeing use in the area of pharmaceutical counterfeit determination, meeting a critical need to protect health and safety of consumers.

Pharmaceutical companies worldwide lose billions of dollars in revenue due to counterfeit drugs. Consumer drugs are a common target for criminal organizations who typically sell these substitute products over the internet. Pharmaceutical industry globalization has increased the danger that counterfeit drug products or adulterated materials are contaminating the health-care supply chain (Woo, Wolfgang, and Battista, Nature 83: 494- 497 (2008)) creating a serious public health risk. Sophisticated counterfeit drug products include not only those contaminated by potentially harmful substances or those absent of any API, but also products containing diluted, ineffective amounts of API. The pharmaceutical industry has an interest in aggressively protecting consumer safety by reducing the amount of counterfeit drug products in the market. Along with manufacturing, packaging, and product tracking advances, spectro-analytical approaches including portable or handheld Raman instrumentation are important weapons against counterfeit pharmaceutical products.

A common problem with Raman spectroscopy has been interference from fluorescent molecules, often excipients. If present, fluorescence interference is typically orders of magnitude higher than the Raman signal, preventing successful chemical identification and/or analysis. Selection of the appropriate excitation wavelength is critical for successful chemical identification. An example of this is seen with Sodium Alginate, a common pharmaceutical excipient. Analysis with 785nm spectra shows significant fluorescence, reliable information about the sample is unavailable. In contrast, the 1064nm spectra are clean and informative, and can be used to produce confident chemical identification. Rigaku Raman handheld analyzers offer several different excitation options and include the first ever a dual wavelength 785nm and 1064nm combination in the Xantus-2.

What are the challenges of working with Raman/ NIR in biopharmaceutical applications?

MK: Current challenges include the ongoing acceptance of validation standards for NIR and Raman that are similar in detail to HPLC and GC-MS instruments. The new generation of instruments, like B&W Tek’s NanoRam, are designed to support stringent validation tests.

GR/CNIRS: The biggest challenge is the calibration process. The calibrations will generally be useful as long as the spectral properties of the product do not change. However, if the process changes, the new variables may invalidate the applicability of the model. This means that appropriate qualification of the sample spectrum must be established so that results for the sample are reported. It may require considerable work to produce the calibrations and the validation data required to implement a spectroscopic application. Also, for on-line measurements, there’s the possibility of the measurement point being a possible source of contamination, or of disrupting the flow of product through the process. Clearly, both are secondary (Raman can be considered a slight bit more “primary”) so they need primary methods to calibrate them for routine measurements. The Chemometrics are more complex than the high school math involved in traditional lab methods. One consequence is that more samples are needed to calibrate and generate an equation, but that is more than made up in the day-to-day use of both techniques. What has become evident over the past several years in trying to find useful ways to use NIR in biopharmaceutical processes, is that whereas it has been the approach for traditional solid or parenteral dosage forms to measure single components for chemical and physical information, the whole spectrum trajectory is measured during a fermentation process for instance, and has been shown to be useful as an indicator of the total health of the process. Critical Quality Attributes (CQA) of bio-pharm materials generally involve “low-level” compositional properties- and thus greatly “stretch” the sensitivity capabilities of “conventional” NIR and Raman methods. For instance:

  • “inferential” NIR/Raman of such properties is possible (for example, via secondary spectral effects of properties on major spectral features), but this requires good understanding of these effects and confidence that said effects are stable over time.
  • “signal enhancement” schemes (ex. SERS, pre-concentration) are possible, but not yet well-developed for commercial “validate-able” use

Even if sufficient sensitivity to a CQA is observed, it can be challenging to establish method specificity to the CQA. This is especially true for NIR, where bands are generally broad and highly overlapped. For NIR, high moisture content of many biopharmaceutical materials generally “swamps out” NIR signals for common biopharmaceutical CQAs. Raman on the other hand is less sensitive to water, making it a more viable option for aqueous systems, but still up against sensitivity limitations. In addition, Raman has laser safety concerns associated with it. In-line deployment of NIR/Raman is particularly challenging in bio-pharma. As previously stated, there are contamination concerns/precautions, and the sampling hardware must be robust to clean in place/sterilization in place (CIP/SIP) operations.

It will be exciting to see what developments occur over the next decade for NIR, Raman and other spectroscopies as well in this area, but clearly it is an area of active research and is bearing fruit at this very moment.

SP: NIR has the disadvantage of being very sensitive to water signals, hence when there is much water around, like in most biological environments, help would be limited with NIR. However, for any drying method (i.e. drying in freeze-drying, spray-drying, etc.) it might be a very accurate tool to determine or monitor the moisture content. Raman spectroscopy is much less sensitive to water, and also shows good selectivity to the folded structure of proteins (secondary and tertiary structure, and information on amino acids in side chains). This is not surprising, as a Raman spectrum gives more selective sharp bands compared to NIR spectra (constituting of combination and overtone bands). Nevertheless, it has been shown that NIR, having less selectivity for this attribute, might be an interesting tool for indirectly monitoring the expansion of the hydrogen bonds of the protein backbone when a folded protein loses its 3-D structure. Another challenge for (N)IR and Raman spectroscopy applications in protein analysis is the detection sensitivity. Compared to more sensitive techniques for analyzing protein conformation, such as fluorescence spectroscopy and circular dichroism, it is rather low (typically protein concentrations in the mg/ml range and not lower). For some case studies, it has been shown that surface enhanced Raman spectroscopy (SERS) can increase the signal-to-noise ratio of the Raman analysis significantly, but this option is only limited to solutions. Other limitations of Raman spectroscopy are that the generation of identical measurement conditions is difficult, its sensitivity to fluorescence artifacts in the spectra, and that the high energetic laser power may decompose sensitive samples. Hence, the measurement conditions need to be carefully optimized bearing in mind all these concerns in the method development.

CH & RK: In the biopharmaceutical industry there is a move toward more timely, in-line analysis in order to better monitor and control biopharmaceutical processes. There are economic advantages of using real-time NIR monitoring in order to optimize the biopharmaceutical cell growth and maximize production. Also, by using NIR for process monitoring, companies and researchers can avoid having to discard an entire batch due to sub-optimum cell growth.

One of the biggest challenges when performing NIR analysis in biopharmaceuticals is that the complex sample matrices often require higher level and more sophisticated model development strategies. It is critical to incorporate sufficient physical and chemical spectral variability from multiple batches into the calibration. The inherent batch-to-batch media differences seen in
biopharmaceutical production runs must be included with accurate wet chemistry analysis in the standards used for calibration model development. Also enough spectral variation in the components of interest must be present so that a robust model can be developed with good accuracy across the full range of components. By getting all expected component, physical and media differences built into the calibration model, accurate and robust quantification of nutrients, by-products as well as products can be achieved in a matter of seconds.

Another challenge when using NIR for bioreactor analysis is representative sampling using fiber optic probes. One of the common biopharmaceutical applications for NIR analysis is in-line monitoring of the progression of a bioproduction process, from start to completion. To ensure a robust and accurate analysis can be achieved over the entire cell growth phase, the probe must be installed in a location to guarantee representative sampling over time. Probe location issues that need to be overcome include clearance from agitator blades striking the probe or creating bubbles or turbulence which could get in the pathlength of the probe that interferes with the spectral measurement. Also solids, changing cell growth opacity and the ability for the sample to fill the probe pathlength gaps of 0.5 – 1.0 mm offer unique challenges for biopharmaceutical analysis. The changing opacity of the cell culture growth over time can be overcome with a single probe designed to perform both reflection and transflectance analysis without removal of the probe from the bioreactor during the cell growth.

RC: Biopharmaceutical samples tend to be challenging for NIR and Raman because biological samples inherently contain fewer spectral features than pure materials. The fundamental modes overlap (or overtones and combination modes in the case of NIR) because biological samples are complex mixtures. In order to bring the information out of this complex spectral data, additional samples are required for model development and it is not uncommon that a larger number of factors are required fitting multivariate models. Raman has more spectral features than NIR, but it can be more challenging because of competing fluorescence, so it is important to choose the proper laser wavelength to minimize fluorescence. Most competing fluorescence can be removed in using lasers outside of the visible regime i.e. UV and NIR. A commercial handheld UV Raman is not available and handheld NIR Raman is limited to 785 nm, 808nm, 1030 nm and 1064 nm excitations.

Biologically complex samples that are heterogeneous tend to yield the same result independent of spot size. Under conditions that the area of interrogation does not fully represent the mixture, the sample must be interrogated in several locations. NIR may contain less spectral information, but has the advantage of larger spot sizes ~mm, while handheld Raman systems have spots sizes <200 μm. Some of the Raman processing probes used in at-line and on-line applications can increase the sampling locations size to mm diameters.

In what ways do these spectroscopic tools assist in counterfeit detection?

MK: With billions of dollars being lost each year by the pharmaceutical industry to makers of counterfeit drugs, NIR and Raman spectroscopic instruments can be used for on the spot analysis of blister packs and individual tablets in retail stores, customs seizures and other locations. The instruments can analyze the main active pharmaceutical ingredients and excipients in a range of concentrations and compare the spectra with brand name reference spectra, typically within tens of seconds.

GR/CNIRS: In addition to having the advantages of the basic technology (speed, non-destructive and non-invasive), counterfeit materials can be identified as being different than the branded or generic product, even if the nature of the difference is not known. Spectroscopy physics is often sufficient to detect differentiating factors- especially if they involve moderate-to-high-level compositional effects (> 0.1%) or morphological or moisture effects. In some instances, this approach can also be applied to contaminated pharmaceutical active pharmaceutical ingredient (API) or excipient (non-active), where subtle differences arising from nefarious suppliers who skirt current good manufacturing practices (cGMP) leave tell-tale signs of their processes, material handling during storage, shipping etc. The subtle differences are “seen” by both techniques: polymorphic forms, different ingredients (excipients), and other process signatures. In traditional and spectroscopic methods, the active pharmaceutical ingredient (API) can be determined, but speed of both is obviously quite different... plus, when you destroy a dosage form, it can’t be presented as evidence in a court of law. So in the right amount of the API and the excipients, NIR (and Raman) can provide information about the presence of contaminants, and polymorphs. The matrix contribution to the spectra can also be very useful. The manufacturing process will provide a spectral signature that is unique to the manufacturing conditions. Thus, the counterfeit drugs will have a different signature and thus be differentiated from branded and generic formulations. This is an active area of research for companies and is receiving much attention from the pharmaceutical manufacturers as well as other industries where brand protection has become a serious consideration mainly due to the financial impact that counterfeiting is having on those industries. Beyond its negative financial impact, drug counterfeiting is a serious global threat to the safety of the public everywhere.

The cost-benefit ratio of NIR and Raman in thwarting drug counterfeiters is obviously still huge. However with serious attempts being made to reduce the cost of these devices (below $10k a unit and perhaps even below $1k a unit as suggested by one expert), while packing the handheld devices with more intelligent features, these features will undoubtedly make handhelds a much more attractive approach in the long run, for better reliability, lower cost, ease of use, and speed of information to prevent further infiltration of rogue products into the supply chain. The hurdle is not that it can’t be done, but can it be done in high enough volume to bring the cost per unit down to levels that organizations, manufacturers and even governments can absorb and justify. The incentive to do this is very high as counterfeiters become more savvy in their approach. Billions of dollars are being lost to the underground black market, so it may come to pass that if the spectroscopic technology can achieve better and more reliable detection, testing, and reporting of data with a lower threshold of training for ease of use, leading to more criminals being caught and prosecuted because of better and more efficient intelligence from these devices, then the cost per unit will become less and less of an issue.

SP: Compared to traditional methods for counterfeit detection, such as chromatography and MS, analyzing suspicious samples with these spectroscopic tools has the advantage of being extremely fast, nondestructive and not requiring solvents. Hence, they might be very suited tools for a rapid first screening based on fingerprint comparison. Again, the complementary capacity of Raman and NIR can be an asset here. Rather than measuring at a certain spot, transmission spectroscopy and hyper-spectral imaging techniques may provide information over a much larger sample area or even over the entire pharmaceutical form, from the core to the outside of a tablet. Differentiation between genuine and counterfeit drug products using these spectroscopic tools will often need a multivariate data analysis approach. For detecting subtle differences between spectra similarity analysis parameters (e.g. based on correlation or distance measurements) can be used, and multivariate calibration models can be built for 2 or multi-class classification.

CH & RK: Near Infrared is commonly used as a screening tool to detect differences from what is considered normal or expected. NIR analysis gives spectra representing the chemical makeup of the sample and with the use of chemometric modeling techniques, small or subtle change in the chemical makeup of a sample can detect differences between normal and counterfeit drug products. The application of NIR spectroscopy as a positive identification technique is accomplished by comparing the spectra of an unknown sample to a spectral library of known compounds. The results generated from the spectral library method can include pass/fail results, chemical compound positive identification and degree of match or similarity to the identified compound. The NIR spectral library can tell the analyst the level of similarity or difference between a counterfeit drug and the drug it is claimed to be. The speed of analysis, ability to interrogate a large sample, analyze without sample preparation including directly through plastic or glass packaging makes NIR an ideal choice for screening of counterfeit drugs. Further investigation by other analytical techniques can aid the investigator in identifying the type of adulterants present in a counterfeit drug product.

Raman spectroscopy is also an extremely versatile tool, offering the ability to detect subtle differences in oral dosage formulations. The complexity of Raman spectra provides a powerful means for uncovering slight changes in composition, analyzing structure and morphology, and even revealing the source of a specific material. Advanced multi-component search tools and comprehensive commercial libraries can readily distinguish and quantify a wide range of pharmaceutical compounds – from excipients to active pharmaceutical ingredients. When applied to detection of counterfeit drugs, the unique signature of an imitation product can be quickly compared with that of an authentic formulation. The key advantage as compared with other analytical techniques (such as NMR and liquid or gas chromatography) is the ease with which samples can be prepared for Raman analysis. A typical scan of a solid tablet can reveal detailed chemical and morphological information in a matter of minutes, greatly streamlining the process of classifying counterfeit products in batches.

RC: Handheld Raman and NIR are only most recently becoming available to thwart counterfeit drugs. There are two popular approaches to counterfeit detection with vibrational spectroscopy. The first is the detection of the API, excipient or final formula in the drug, or to detect a visible on invisible phototag on the packaging. Alternatively, phototags can be inserted into the final ingredient, but this is less popular. The limitations to handheld Raman and NIR in the last five years are the instrument costs comparable to competing handheld technologies based on portable UV/VIS absorption and fluorescence spectrometers (ca. < $5,000). Needless to say UV/VIS absorption and luminescence responses contain less spectral information and are easier for counterfeiters to fool. Raman phototags are available for packaging which have very high selectivity and sensitivity, and are difficult to reverse engineer. Once the costs for handheld Raman approach those of UV/VIS and fluorescence systems pigment/dye manufactures and pharmaceutical companies will invest further into Raman phototag technology. Today it is possible to purchase large quantities of handheld Raman systems for less than $8,000.

What are the benefits of working with handheld spectrometers?

MK: Handheld spectrometers have several key benefits, particularly for analysis of incoming raw materials. A major cost/time savings benefit is moving verification of raw materials from a central research laboratory to point of use analysis in the warehouse or loading dock. The handheld instruments support cGMP policies and procedures, as mandated by US FDA and the other regulatory bodies. Another benefit is ease of use. These instruments are small, can be operated singlehandedly and enable sampling of a range of different kinds of materials with little or no preparation. They can also be operated by non-specialists following well-defined standard operation procedures. Rather than reviewing spectra, the operator sees clear indications for Match/No Match for identification or Pass/Fail for statistical verification.

GR/CNIRS: By definition, “hand-held” means that it can be readily moved from one sample to another (e.g., samples in 55-gal drums, or other situations where it’s easier to move the instrument than the sample). Real time material identification, process monitoring, can be used by all users with limited training. The immediacy of them are the biggest benefit; the time involved in bringing a sample to a lab versus having the “lab” with you in the field is significant. Also, you can view many packages and samples on the spot. The lack of delay enhances law enforcement groups in detaining suspects.

What features of Raman and NIR instruments help to confront regulatory initiatives head-on?

MK: There are several features for NIR and Raman instruments that will assist with achieving 100% testing of raw materials to meet global regulatory initiatives. The main feature is the trend toward handheld packages with a wide range of sampling accessories for different materials. With state of the art optical, electronic and computing components, these instruments support rapid analysis and identification with the ability to do statistical verification that meet stringent cGMP regulations.

GR/CNIRS: Regulatory initiatives involve the elements of process analytical technology (PAT) and quality by design (QbD) such as design of experiments, risk management, continuous improvement and quality and life-cycle management. With respect to Quality by Design (QbD) elements:

  • Risk Assessment: High-throughput, non-invasive measurements of CQAs to support process optimization during development
  • Design Space: High-volume, and high-relevance, development data to allow “richer” description of design space
  • Control Strategy: high-speed on-line/at-line analytical to support advisory or closed loop control schemes
  • Lifecycle Management: If NIR/Raman deployed on-line: highfrequency and high-relevance data to support continued process verification.

All of these parts are designed to increase process understanding before, during and after the manufacturing process. The key benefit that both NIR and Raman bring to this endeavor is “Instantaneous” results that allow for actual real-time control. Understanding that this stands in stark contrast to the current regulatory reality known as change control; manufacturers who adopt these technologies will have to realign their quality and regulatory systems to accommodate these new approaches. This is actually a good thing and while currently viewed as a big leap, will happen when the financial impact to savings that comes with developing, filing, approving and manufacturing the new drugs of the 21st century (primarily biopharmaceuticals) is realized by those companies that have adopted these technologies. Specifically, some of the changes needed for adopting these technologies will come from the possibility to have with each spectrum taken diagnostics that tell the user if the sample is normal (variability has already been seen), if the instrument is working properly, if the sample is an outlier and the causes of the outlier (changing manufacturing process & raw materials, change in particle size, etc.) that can have an impact but not be seen because the process is still in control, and the ability to see if the process is in control by looking at real-time on-line trends in the parameter of interest) using MSPC statistics such as Hotteling’s T2 and Q residual tests to name just a few. Now, you have compendial chapters used for guidance and most manufacturers have gotten on the 21CFR210-211 bandwagon. With better knowledge and instruments (largely) being built by larger, experienced manufacturers (many small companies being bought), the adherence to ISO standards is also commonplace and reproducibility of hardware is better than ever. These are huge changes for a conservative bunch like pharmaceutical manufacturers. Big changes won’t happen all at once, but will occur in small incremental steps. Everyone agrees that while the technology can (almost) make it happen tomorrow, the will of the culture, inextricably linked to a large and diverse regulatory structure, itself undergoing continual change, will evolve at a slower pace. But it is occurring as is evident by some of the other comments by my colleagues.

The bigger picture holds more promise however. It was asked once (tongue in cheek) if the FDA one day sees itself as monitoring every drug manufacturing plant from some central office complex in Washington using these technologies. While it may not be in their foreseeable future, the technology certainly makes it possible to do remote monitoring locally within a plant now. So to extend that to the next level and have centralized up to the minute monitoring of manufacturing plants in realtime is not out of the realm of possibility. As it was once told to me by an ex FDAer, “It’s all about control.” If the industry gains process understanding through adoption of these technologies and through this also gains increase assurance of product quality, then the business and regulatory risk of non-compliance for them has been lowered or even eliminated. Then there would be no issue with sharing instantaneous access to real time data with the authorities during manufacturing. They would have the understanding and assurance that the agency wants them to have, while the agency would have the control (for real time release or RTR) that has been mandated by congress. That is, the manufactured product is of acceptable quality and can be released to the market. That is their promise to the American people; essentially pure, safe, and efficacious products and that they are available in a timely manner.

CH & RK: The ability of NIR to analyze a large sample size in-situ, without sample preparation and nondestructively, makes it the analysis technique of choice for identification and adulteration testing. NIR identification models generate conclusive results showing the level of difference between adulterated and nonadulterated samples. Also, NIR is an easy to implement technology for accurate and robust analysis by production operators. Regulatory bodies often turn to NIR as a preferred analytical technique because of these inherent benefits NIR has over other analytical techniques. These advantages are the reason why NIR is applied world-wide for identification testing of raw material and ingredients prior to their use in pharmaceutical, dietary supplements and food products. Public concern about the safety of the drugs, dietary supplements and food that they consume has received significant publicity in recent years, especially with the melamine in baby formula discovery in 2008, which have led to more regulation in these industries to protect consumers.

NIR spectroscopy has a long history of regulatory acceptance in the United States, European and Japan Pharmacopoeias. The pharmacopoeias give a baseline to users for the installation, operational and performance qualification required for a NIR spectrometer prior to its use for analysis of regulated products. The tests and specifications that are spelled out in the pharmacopoeia help ensure that an NIR spectrometer is performing at a high level and that consistent results are generated when analyzing ingredients, processes or final product in regulated industries such as pharmaceutical, biopharmaceutical, dietary supplement and certain food industries.

How will spectroscopic capabilities develop and expand in the next five years?

MK: Portable spectroscopic instruments will have laboratory quality performance and have capabilities to perform analyses of complex mixtures for a range of concentrations. The instruments will support laboratory information systems (LIMS) and quality management systems (QMS). A wide range of spectral libraries will be available that can be accessed from secure networks. We may see the development and adoption of instruments with complementary spectroscopic technologies, in portable or even handheld packages.

GR/CNIRS: The only “simple” answer is the extension of trends currently under way: smaller size & weight of instruments platforms: likely “borrowing” hardware/ firmware/software from other high-volume industries (telecom, security, gaming/entertainment), and lower power requirements. There already is a movement for on-line applications and instrumentation from all branches of spectroscopy and sensor technology. Bench work will be reduced to a minimum (but still be required during development, manufacturing and market support) and scientists will have the capacity to know what their process is doing at any time. Better hardware through experience will lead to marginally improved sensitivity and stability of conventional hardware; clever new optical/electronic/sampling schemes to enhance sensitivity and specificity, better data/calibration model visualization, better Chemometrics from competition; and a movement towards the introduction of the relevant subjects to teach the next generation in college curricula (fingers crossed) is what can be expected.

CH & RK: Near-Infrared (NIR) spectroscopy has matured into a common technique in the pharmaceutical industry and there will continue to be growth in the use of NIR as a Process Analytical Technology (PAT) tool. NIR will become more widely applied as a PAT tool across the whole pharmaceutical manufacturing life cycle from raw material identification to granulation and drying to blending and tablet production. We will see further introduction of fit-for-purpose and total solution NIR analyzers to unlock the full potential of NIR spectroscopy as a Process Analytical Technology for monitoring and controlling pharmaceutical production processes.

Additional advances in NIR analyzer design will allow for faster and smaller instruments with new accessories and probes that further optimize sampling of production processes. For many process analysis applications, representative sampling is the biggest challenge to overcome. Some of the lessons that have been learned when PAT was in its infancy will lead to improved probe designs and better sampling devices to overcome the sampling difficulties that have made certain applications challenging. Also managing the sheer volume of real-time, multivariate data that is generated by NIR analyzers, when used for real-time process analysis, has been a challenge. Further advances in data analysis and data management will lead to growth in NIR as a PAT tool. NIR process analyzer software with embedded automated archival and industry standard process communication protocols will become more common. This will allow NIR analyzers to be more easily integrated into the pharmaceutical production automation network, a critical step when implementing any PAT analyzer. NIR analyzer software packages that support multiple data exchange formats, process communication protocols and the ability to seamlessly execute external applications will further simplify the implementation process and enhance the value proposition of NIR as a process analysis tool.

As instrument technology continues to advance at an increasingly rapid pace, the market will witness less emphasis on the technique in favor of a stronger focus on obtaining answers. While hardware will play a significant role (detector sensitivity, sample handling and targeting, acquisition speed), solutions-oriented software will continue to add significant value for simplifying the workflow in the laboratory, as well as on the production line. Expect to see an expansion of commercial library offerings and even more robust search algorithms that take advantage of hardware improvements. The complete instrument package will continue to be easier to use, provide greater reliability and uptime, and ensure that perfect results can be achieved every time with minimal user training.

GLR: Analytical tools appear to be diverging into two classes: (1) cutting edge, highly capable research grade instruments with many user controlled parameters and (2) smaller, less expensive, portable, rugged and highly user friendly (plant and loading dock friendly). This divergence corresponds to the needs of the users, and the desire to place more straightforward, easier to use spectroscopic tools in the hands of project teams’ chemists.

Not all applications require the most capable or sophisticated tool. However, when the less capable tool does not meet the analysis need, it does not necessarily mean the technique is not appropriate, just the specific instrument is not capable of providing suitable data. In the R&D arena, having highly capable tools (with increased ease of use) to develop an understanding of the performance requirements is valuable for downstream use.

Developments will continue evolutionarily with spectrometer size reducing with an equivalent to better performance. Disruptive/Revolutionary technologies will continue to be developed, for example with the adoption of technologies developed for the defense and telecom industries. For instance, the use of Quantum Cascade Lasers (QCL) as intense, diffraction limited light sources for mid infrared spectroscopy will continue to evolve and become more commonplace in the future. Work on enhanced (less expensive, more rugged) couplings between MIR source/detectors and probes is an area of need due to the expense, fragility and limited length of silver halide fibers.

As much as hardware enhancements will add to capabilities, software improvements are key to enable pharmaceutical chemists to get more information and use from spectroscopic tools, and for more users to get quality information from these tools. We now live in an “Apple ecosystem” where there is an (inexpensive and well-functioning) app for everything. As such, there is an expectation for integrated informatics solutions where (multiple) hardware platforms (spectroscopic systems) are seamlessly connected and data pre and post processing intelligence (“An expert in the box”) will have facile user interfaces. As such, a non-expert can use the system to good capability (with a minimization of user produced errors).

Any additional or closing thoughts on this topic?

MK: Handheld Raman instruments are emerging as the most flexible tools for verifying most pharmaceutical materials. Raman has many advantages including non-destructive analysis, little to no sample preparation, typically sharp spectra that facilitate detailed chemometrics, and it can measure a wide range of gels, liquids, powders and solids even in aqueous solutions. In addition, sampling accessories enable Raman analysis of materials inside transparent and translucent containers.

GR/CNIRS: We still struggle to show that PAT can be useful in manufacturing and not only during development. PAT is a complement of QbD. PAT should be in place during manufacturing as a way to ensure that the process is still operating in the design space and that unforeseen raw material changes and variability in the process does not negatively affect the final product quality. It is likely that the spectroscopic techniques will be of most value when they are used in the development of new products, and can assist in learning about the manufacturing process and optimizing the steps. Then the measurements themselves can be used in controlling the process. This is distinct from using the spectroscopic techniques to replace, say, HPLC measurements in existing processes, where the pay-back for the calibration effort may be more difficult to achieve. However, for the future sustainability of NIR science, practitioners are now realizing that the idea of integrating NIR into current college curricula is perhaps the most important issue facing NIR science right now. NIR still isn’t being formally taught (although some programs are and have successfully integrated NIR and Chemometrics into pharmacy graduate studies here in the United States; Duquesne University, University of Kentucky, Purdue University and University of Maryland to name just a few), 212 years after its discovery! It may take multi-disciplinary teams, since it is only slightly spectroscopic and largely Physics, optics, sampling theory, and Math. Near Infrared education has now been formally adopted by both the Council for Near-Infrared Spectroscopy and the International Conference for Near Infrared spectroscopy (ICNIRS) as an objective for both organizations. Dr. Rodolfo Romanach along with some of his graduate students has created a series of PowerPoint slides for use by universities for introducing NIR into their curriculum. These PowerPoint slides have been developed through collaboration between the education committee of CNIRS and ICNIRS, the Engineering Research Center for Structured Organic Particulate Systems (http://ercforsops.org), and students in Dr. Rodolfo Romanach class QUIM 8995 - Special Topics in Solid State Vibrational Spectroscopy. The slides have been developed to facilitate the teaching of NIR spectroscopy. Thus, professors teaching instrumental analysis are welcome to use the slides in their courses. Feedback regarding the use of these slides and future improvements or corrections to be made is welcome, please send to [email protected]. They currently can be found in the Purdue Pharma Hub. The web site is: http://pharmahub.org/resources/teachingmaterials/ and are titled “NIR Fundamentals and “a little more”, and “NIR Spectroscopy: An Advanced Alternative.”

GLR: Spectroscopy is just another analytical tool. Each and every analytical technique a pharmaceutical chemist applies will have strengths and weaknesses. Many of the technique strengths overlap, and choice often comes down to analysis speed, familiarity/availability of the equipment (in the lab and transfer site) and status of the method as for informational purposes, fault detection or control. Increased staff awareness, to increased equipment capability/capacity, to increased quantity of technically skilled staff is required to continue to make progress on the routine use of spectroscopy in Pharma.

Spectrometers are routinely utilized in Pharma. These tools are used in the development, understanding, optimization, fault detection and ultimately in the control of processes when required. The vast majority of PAT tools are used in non-control applications (e.g., design and optimization of processes) though. Thus, when processes are transferred to a manufacturing setting, there is high confidence in each unit operation and knowledge of which (if any) step requires an in-situ control method.

In chromatography, specificity is high, as components are generally resolved by time (separation). The time and resources required to develop, validate and use a quantitative method on multiple instruments is generally low (most methods are not instrument specific and will run acceptably on multiple instruments from multiple vendors). Specificity in spectroscopy is related to unique bands (or ratios) that can be chemometrically resolved and modeled for quantitative results. The time and resources required to develop, validate, maintain and use a quantitative method on one instrument is typically high. Additionally, models normally have to be rebuilt on each instrument (method transfers instrument to instrument are challenging in practice even within the same vendor and model). The data from models must be confirmed (via a diagnostic) to ensure the data is contained within the model (or the model must be updated/maintained) prior to generating reportable results. As the matrix may contribute to the spectra, extensive variance must be added to the model to obtain a stable and widely applicable model.

In part due to the significant additional time and resources required to develop and implement quantitative spectroscopic models for control, these tools will find increasing use in internal fault detection and monitoring, and be used in combination with traditional (and off-line) methods used to generate batch results.

  • <<
  • >>

Join the Discussion