Development Considerations of Adapting Raman Spectroscopy for Raw Material Fingerprinting

• Amgen Inc.

Abstract

Raman spectroscopy has taken large strides in recent years as more Raman vendors have developed handheld Raman units capable of carrying out raw material identification or verification. Raman spectroscopy is used as a fingerprinting tool to capture the unique Raman spectrum of each raw material. There are many benefits of adapting this new Process Analytical Technology (PAT) tool to the biopharmaceutical industry. Critical technical considerations and unique challenges that must be overcome to implement Raman spectroscopy in the biopharmaceutical industry are discussed.

Introduction

Application of Raman spectroscopy in the biopharmaceutical industry has increased due to advances of the technology^1-7. It is in alignment with the FDA's guidance to utilize more Process Analytical Technology (PAT) tools to increase better process control and process understanding⁸ . Recent advances in Raman spectroscopy have allowed development of handheld Raman units capable of achieving acceptable resolution to perform raw material identification and verification in Good Manufacturing Practice (GMP) facilities^9-11. This has allowed the biopharmaceutical industry to adopt these new Raman PAT tools in place of traditional techniques to increase efficiency of the incoming raw material process to achieve real-time release. However, a number of critical considerations are needed before incoming raw materials Raman fingerprints can be used to accept or reject a batch of raw material.

Raman spectroscopy is applicable in the incoming raw material area^9-11. The main benefits of utilizing Raman handheld technology are high accuracy, high reproducibility and mobility as well as the non-destructive nature of the technology^1-7. Traditional identification methods such as high-performance liquid chromatography (HPLC) or other compendial methods can take up to 3 days to release a lot of material for process use. Figure 1 below shows a schematic of traditional incoming raw material identification test versus the new process with the implementation of Raman technology. The yellow path shows the traditional flow of an incoming raw material. The green path shows the new flow where Raman technology is used. In addition to increased operational efficiency, the information rich Raman data can be analyzed using multivariate (MV) statistics to gain further understanding about the raw material (i.e. identify outliers, drifts and clusters). If raw material suppliers also adopt Raman technology, the data can be shared with their clients to monitor and trend the raw materials which can ultimately lead to improved raw materials quality.

Figure 1. Schematic of traditional versus Raman implemented workflow

Method Development

There are many technical considerations when developing methods that are suitable to perform raw material identification. The type of material to be analyzed and its primary packaging often drive some of these considerations. Another primary consideration is the choice to perform material identification or verification.

Identification Versus Verification

The difference between identification and verification of a raw material must first be distinguished. Although identification and verification are sometimes used interchangeably in general discussions, they have very different meanings in quality. A clear path of whether the developed method is used for identification or verification must be distinguished first before going further in method development. Identification refers to subjecting an unknown raw material spectrum to all available raw material spectra in the spectral library and allowing the instrument software to determine the unknown material’s identity. On the other hand, verification would subject the unknown raw material spectrum to a specific spectrum in the spectral library and the instrument software would compare the two spectra and determine closeness of the match. There are advantages and disadvantages to both modes of testing. For a GMP implementation, the quality organization should be consulted to determine the best path forward.

Fluorescence

Fluorescence is a physical property of molecules that can impact the Raman spectrum. Depending on the purpose and the method being developed, the fluorescence property of the raw material can sometimes be used as a mean to identify the material. If fluorescence is an undesired property, there are hardware as well as software solutions. Sometimes, the fluorescence from an unknown impurity in raw materials can overwhelm the Raman signal when they are excited with visible laser line such as 532 nm and 785 nm. To avoid fluorescence interference, the 1064 nm laser excitation is used to reduce or completely eliminate the fluorescence. The drawback of using 1064 nm laser excitation is that it will decrease the instrument sensitivity of the Raman device as Raman scattering intensity is proportional to the fourth power of inverse wavelength.

The shorter the excitation wavelength, the stronger the Raman signal. Using 1064 nm is the hardware solution to overcome fluorescence in raw materials. Pre-processing techniques such as applying 2nd derivatives to the raw Raman spectrum is a common way of removing fluorescence background from the spectrum.

Sensitivity to Raw Material Packaging

Container contribution to the Raman spectra is something to be considered as containers do have measurable Raman peaks. Depending on several container factors (thickness, color, opaqueness) the device settings should be optimized to minimize any container contribution. Each presentation should be analyzed in order to qualify that a particular container is suitable for Raman analysis. If necessary, library entries should be created in the raw material vendor’s original packaging. If multiple presentations exist for a given raw material, then multiple library entries should be created for each container. Figure 2 shows Raman spectra of linear low-density polyethylene (LLDPE) and high-density polyethylene (HDPL) acquired using a handheld Raman analyzer at different focal lengths.

Figure 2. LLDPE and LDPE Raman spectra with different focal length

Focus length of zero denotes a Raman spectrum acquired at the surface of the container. Focal length greater than zero denotes the acquisition of the Raman spectrum at a specific distance inside the container. As expected, LLDPE and LDPE containers have different Raman spectra. The intensity of the Raman spectrum unique to the container decreases as the focal length increases. For GMP implementation, if acquisition of raw material Raman spectrum is carried out inside a container, then it is critical that the material, thickness and quality of the container be controlled as well. Otherwise false results could be obtained. If sampling through glass containers, the same considerations should be applied. Opacity of the glass is an important factor. Some amber glass will emit fluorescence that could impact the Raman spectrum as discussed earlier.

Polymorphism

Raman spectroscopy is a great way to detect polymorphism^7,8,12,13. However, when using Raman spectroscopy for incoming raw material detection, polymorphism becomes a factor that must be built into the method. If the material exhibits multiple polymorphic forms, then each specific polymorphic form’s Raman spectrum must be collected and understood. Typically each polymorphic form will have a unique Raman spectrum. Figure 3 shows four lots of 4-(2-hydroxyethyl)-1- piperazineethanesulfonic acid (HEPES) with two different polymorphic forms.

Figure 3. Four lots of HEPES Raman spectra with two different polymorphs

Understanding the various polymorphs of the raw material is key to developing a consistent Raman identification method. The various polymorphs must be incorporated into the method development to prevent false results.

Moisture Content

For powdered raw materials, moisture content can be crucial to the consistency of the Raman scan. Water can contribute to the Raman spectrum in certain raw materials at the high frequency region where hydrogen bonding of water exhibit. Intensity of a Raman spectrum decreases as the moisture content increases. This phenomenon exists for all materials, but is most obvious for a hydrophilic material. If a raw material is hydrophilic, the developed method must include a check of moisture content to ensure that all incoming material of this type is tested within a set range of moisture content; otherwise, the intensity deviation can lead to incorrect results. Similarly storage condition of raw materials is equally important to ensure invariable moisture content. Figure 4 shows Raman scans of a liquid (148 g/L) and a powered L-tyrosine and L-tyrosine dihydrate samples, respectively. The powdered L-tyrosine dehydrate sample has more pronounced peaks in the 950 and 1100 cm^-1 peak regions than the liquid sample. Please also note that 148g/L is a concentrated L-tyrosine solution and as the sample gets diluted the peak intensity will also drop. Due to Raman’s sensitivity, these differences should be considered to ensure a robust method is developed.

Figure 4. Raman spectra of 148g/L liquid L-Tyrosine (blue trace) and L-Tyrosine Disodium Dihydrate (black trace)

Instrument Settings

Depending on the system, instrument settings should be kept as consistent as possible for each developed method to ensure robustness. Some handheld instruments have autocalibration capabilities that adjust to environmental conditions to ensure consistency of each Raman scan; however, there are other instruments available which allows the user to adjust the settings including background correction, integration time, spectral range selection, number of spectra to average, focal range, etc. There are many development advantages to being able to adjust the instrument setting as it provides higher flexibility, which could lead to development of greater variety of raw materials methods. On the other hand, due to the higher flexibility, during implementation, complexity of the validation protocols and operating procedures will also increase accordingly.

Calibration and Maintenance

In order to ensure consistent Raman spectral acquisition, some Raman handheld analyzers would undergo a number of calibration and verification cycles for a 4 hour scanning session (minimally bracketing sample batches with verification scans using a commonly used standard like benzonitrile). Given the possibility of subtle wavenumber drift over relatively short periods of time it is imperative that these calibrations be performed; otherwise the accuracy of the acquired data could be at risk.

If possible, the instrument laser’s output should be trended over the lifetime of the instrument to prevent degradation of the acquired Raman signal and a corresponding loss of instrument sensitivity. Most Raman handheld instruments control this with their internal software. Nonetheless, it is recommended to monitor the laser output by trending the raw calibration spectra over time.

GMP Implementation

In addition to completing the Installation/ Operation Qualification (IOQ) from the vendors, additional internal measures must be taken to ensure proper instrument operation and method development. These include instrument validation protocols and method validations. Another important consideration is to ensure that the instrument is 21 CFR Part 11 compliant¹². Most of the commercial Raman handheld units available adhere to the standard. However, typically internal company standards are higher than the base 21 CFR Part 11 guidelines. In order to evaluate the instrument’s true compliance to the company standard, the Quality Compliant group should be consulted.

Data integrity is another major consideration for GMP implementation. For 21 CFR Part 11 compliant Raman handheld analyzers, data integrity is guaranteed from the time the Raman is acquired to its instrument data storage location. However, once the raw data file is exported to an internal database, it is no longer guaranteed. An internal data migration procedure and data archiving infrastructure must be in place to ensure the data stays integral.

Conclusion

There are many critical technical considerations and unique challenges that must be overcome to implement Raman spectroscopy in the biopharmaceutical industry. All raw materials are unique and should be considered case-by-case. Inherent understanding of the material is key in development of a good method. Furthermore, understanding the instrument and its capabilities as well as its shortcomings is also important. Lastly, familiarity with the internal standards as well as external regulations is critical to a successful on-the-floor implementation of Raman technology.

References

1. Wen Z-Q 2007. Raman spectroscopy of protein pharmaceuticals. J Pharm Sci 96(11):2861- 2878
2. Sasic S. 2008. Pharmaceutical Applications of Raman Spectroscopy. 1st ed.: John Wiley & Sons, Inc
3. Cao X 2013. Raman spectroscopy in support of biotherapeutic production: applications in protein formulation and purification. European Pharmaceutical Review. 18 (5): 34-38
4. Wen Z-Q, Cao X, Phillips J 2010. Application of Raman spectroscopy in biopharmaceutical manufacturing. Am Pharm Rev 13(4):46, 48, 50-53
5. Jayawickrama D, El Hagrasy A, Chang S-Y 2006. Raman applications in drug manufacturing processes. Am Pharm Rev 9(7):10, 12-15, 17
6. Fini G 2004. Applications of Raman spectroscopy to pharmacy. J Raman Spectrosc 35(5):335-337
7. Vankeirsbilck T, Vercauteren A, Baeyens W, Van der Weken G, Verpoort F, Vergote G, Remon JP 2002. Applications of Raman spectroscopy in pharmaceutical analysis. TrAC, Trends Anal Chem 21(12):869-877
8. US Food and Drug Administration (September, 2004). Guidance for Industry: PAT — A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance. Available at: http://www.fda.gov/downloads/Drugs/ GuidanceComplianceRegulatoryInformation/Guidances/UCM070305.pdf. Accessed September 8, 2015
9. Green RL, Brown CD 2008. Raw-material authentication using a handheld raman spectrometer. Pharm Technol. 32(3):148, 150, 152, 154, 156, 158, 160, 162
10. Green RL, Brush R, Jalenak W, Brown CD 2009. Verification Methods for 198 Common Raw Materials Using a Handheld Raman Spectrometer. Pharm Technol 33(10):72-82
11. Yang D, Thomas R 2012. The benefits of a high performance, handheld Raman spectrometer for the rapid identification of pharmaceutical raw materials. Am Pharm Rev 15(7):126738/126731-126738/126710
12. US Food and Drug Administration (Mar, 1997). Title 21 – Food and Drugs Chapter I – Food and Drug Administration Department Health and Human Services Subchapter A – General Part 11 Electronic Records; Electronic Signatures. Available at: http://www.accessdata.fda. gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?CFRPart=11. Accessed September 8, 2015

Author Biographies

Tony Wang is a senior engineer in Process Development at Amgen. He is responsible for multivariate data analysis, computer simulation and modeling including model predictive control and furthering PAT with Raman technology. Tony holds a B.Sc. in chemical engineering and a M.Sc. in biomedical engineering from University of Calgary.

Dave Meriage is a senior scientist in the Forensics group in Process Development at Amgen. He is responsible for raw material and product characterization as well as integrating new technology into various production environments. Dave holds a B.S. in biochemistry and cell biology from the University of California at San Diego.

Lise Ann Craig is a specialist in Quality Control at Amgen. She is responsible for the management of critical reagents and reference standards and led the implementation of the Raman technology within QC. Lise Ann has over 15 years of GMP experience in Quality Control and holds a degree in Chemistry from Rhode Island College.

Katie Parks is a Senior Manager in Quality at Amgen. She was part of the team that is responsible for the implementation of Raman Technology for raw material identification and verification as an alternate use to traditional quality control methods. Katie holds a B.S in cellular and molecular biology from Tulane University.

Xiaolin Cao is currently a principal scientist in the Attribute Sciences, Amgen, where he employs a number of analytical techniques to support rapid identification of products and raw materials, formulation study and pilot plant manufacturing. He has authored over 60 peer-reviewed research papers and is a member of ACS and the Society for Applied Spectroscopy.

Zai-Qing Wen is currently a principal scientist working at the Attribute Science, Process Development at Amgen. He joined Amgen in 1999 and has led an analytical group responsible for drug product NC investigation; manufacturing incident investigation, customer complaints investigation, raw material characterization, analytical method development, validation and transfer to Quality organization. His major interests are in vibrational spectroscopy, in particular in Raman spectroscopy, protein biophysics and material characterization. He has author and co-authored more than 60 papers and three of them won the best paper published in PDA Journal of Pharmaceutical Science and Technology (2009, 2010 and 2013).

Cenk Undey is an Executive Director of Process Development at Amgen, based in the headquarters in Thousand Oaks, CA, USA. He is leading the Digital Integration and Predictive Technologies group targeted to implement platforms to reduce variability, improve process development cycle time hence increasing speed to market with reduced development cost. Dr. Undey received his B.Sc., M.Sc. and Ph.D. degrees all in Chemical Engineering from Istanbul University, Turkey and his Executive MBA from the University of California, Los Angeles, Anderson School of Management.