Kieran O’Connor is Analytical Science and Technology senior chemist at SK biotek, Ireland, kieran.oconnor@sk.com. (formerly Bristol Myers Squibb)

Kieran O’Connor

	Submitted: Sept. 28, 2018. Accepted: Oct. 16, 2018.

The biopharmaceutical industry is responding to the need for accelerated development timelines of new medicines and identifying opportunities to accelerate development and optimize the manufacturing process. Spectroscopic-based controls have been demonstrated to be quicker, more flexible, and easier to operate and automate than traditional analytical control applications. This paper describes two separate case studies in which spectroscopic methods were introduced as equivalent alternative methods, first to a gas chromatographic method to monitor an in-process solvent exchange step and second to a potentiometric titration method to release a process input material. The spectroscopic techniques implemented were Raman and Fourier transform near infrared (FT-NIR) spectroscopy. The implementation of the spectroscopic methods generated significant increases in sample throughput as well as reductions in plant cycle time, analysis times, and incidences of analytical deviations.

The Food and Drug Administration (FDA) Safety and Innovation Act of 2012 (1) established a breakthrough designation program to expedite the development and approval of innovative drugs for serious and life-threatening conditions. As biopharmaceutical companies have identified the potential to file new drug entities as breakthrough designation candidates, they need to concurrently identify opportunities and strategies that will allow them to accelerate the development process and optimize manufacturing. FDA and other regulatory authorities are encouraging the industry to adopt a more risk-based, scientific approach to drug development and manufacture, as detailed in the FDA Guidance for Industry document 

Process Validation: General Principles and Practices 

(2) and the International Council for Harmonization (ICH) Q8, Q9, Q10, and Q11 documents (3–6). These documents detail techniques, procedures, and strategies by which increased process knowledge and understanding, and ultimately greater process control, are obtained. These advances are accomplished through comprehensive development activities, including creation of a process design space, thorough and challenging risk assessments, and use of process analytical technologies (PAT) to develop critical process parameters (CPP) for control. The knowledge obtained through these approaches can subsequently be applied to risk-based decisions to accelerate the development process. 

	Bristol-Myers Squibb (BMS) actively embraces this risk-based approach to drug development and manufacture and has established PAT-focused research groups within both drug substance and drug product development to enable robust data generation and process understanding. The Enabling Technologies Group (ETG), within the Chemical and Synthetics Development department, specializes in the development and implementation of spectroscopic-based methods for process monitoring of pharmaceutical and biopharmaceutical drug substance development and manufacture. The ETG collaborated with BMS Ireland’s small-molecule drug substance commercial manufacturing site (which changed ownership to become SK biotek Ireland in January 2018) to identify opportunities to introduce spectroscopic-based methods of analysis to generate richer batch data sets and increase the overall efficiency of the site manufacturing operations. 

	One unit operation that was quickly identified as an opportunity to increase the efficiency of both the plant cycle time and the quality control (QC) laboratory was the monitoring of solvent “swaps” or exchanges in the reactors. Solvent exchange is one of the most common unit operations in a drug substance manufacturing facility, as solvents routinely need to be removed for multiple reasons, such as for compatibility with subsequent reaction steps, to facilitate crystallization, and/or to purge impurities. The team evaluated the solvent exchanges used in the production of the high-volume products that the site manufactured for opportunities to implement spectroscopic-based methods. These opportunities would afford the greatest gains in plant and laboratory efficiency. The target identified was an in-process solvent exchange step that employed gas chromatography (GC) as an in-process control (IPC) test. The overall impact to the plant cycle time was significant, as the reactor needed to be cooled to a safe temperature before a sample could be taken and transferred to the QC laboratory where the sample analysis and data processing takes approximately two hours to complete before the result is communicated back to the manufacturing team. Thus, the following steps all contributed to the overall time for this IPC: cooling the reactor, sampling the reactor, transferring the sample to the QC lab, preparation of GC instrument and analysis of system suitability solutions, analyzing the sample, reviewing the data, reporting the results, and re-establishing the reactor temperature. 

	The ETG identified this IPC as being suitable for Raman analysis, as both of the solvents (dichloromethane [DCM] and acetone) generate significant and distinct responses in the Raman spectrum. A two-part plan was devised with the local analytical group. This plan was to initially develop and validate an at-line Raman method for the solvent exchange analysis as an equivalent alternative to the GC analysis and then subsequently introduce an equivalent, in-line Raman method. Once the team had demonstrated the feasibility of the at-line method, the second part of the plan, an in-line approach, was implemented. The rationale for the in-line analysis was to further improve process throughput, as the at-line Raman analysis still required that the reactor be cooled before sampling. An in-line analysis was developed using a Raman probe (Pilot E, Kaiser Optical Systems) that was placed directly in the reactor. The results were exported directly to a distributed control system (DCS) in the plant, realizing additional time savings through the elimination of the cooling and sampling steps. 

	In addition, the team identified an opportunity to enhance the plant efficiency by using spectroscopy for the molarity determination of hydrochloric acid in methanol (HCl in MeOH), a key reagent used in the manufacture of one of the drug substance processes run on the site. The plant had been determining the molarity using potentiometric titration, and the testing was required for both QC release testing of the reagent as well as an IPC test during batch processing. Despite the simplicity of titration testing, it was a time-consuming test that took a number of hours to complete because solutions needed to be prepared and standardized prior to sample analysis. FT-NIR was identified as an appropriate spectroscopic technique for the development of an at-line method. A method was developed and validated and then introduced as an equivalent alternative to the titration method, achieving greater than ten-fold reduction in sample turnaround time, as described in case study 2.

	Case study 1: solvent exchange monitoring using Raman spectroscopy

	As described in the previous section, GC analysis for the solvent exchange step in the manufacture of a drug substance was being used to perform the IPC test. The total analysis time in the QC laboratory typically took approximately four hours to complete, including instrument setup, standard and sample preparation, analysis of system suitability, working standard, blank and reaction completion sample solutions, processing of the relevant data using chromatography data software (Empower, Waters), and the documentation review and approval. Because both solvents (DCM and acetone) possess strong Raman responses with unique and characteristic bands, Raman was a suitable spectroscopic technique for monitoring the solvent exchange. 

Feasibility trials were performed using an analyzer (RXN2 Raman-785 nm, Kaiser Optical System) equipped with a HoloLab Analytical Sample Compartment (HLSC, Kaiser Optical System). Sample and standard solutions were analyzed in glass cuvettes. The absence of interfering fluorescence or other matrix effects was confirmed using process sample solutions. 

	A multivariate analysis model was developed that employed the ratio of the DCM peak response divided by the sum of DCM and acetone peak responses. Based on this simple model, a linear curve was established, bracketing the IPC specification for the solvent exchange, and was applied to the reaction samples. 

Validation was successfully performed following ICH Q2(R) guidelines (7) and using the GC method as the primary reference method. Validation tests performed included specificity, linearity, limit of detection, precision, and accuracy. An equivalency protocol, in which process samples were analyzed by both the GC and Raman methods and the results from both data sets were compared against each other using predefined acceptance criteria, was also successfully performed. The Raman method was then introduced into the QC laboratory as an equivalent alternative to the GC method. Appropriate training and ongoing technical support were provided to the IPC chemists to ensure the Raman analysis was effectively introduced into the laboratory. The relevant regulatory filings were updated to include the application of the Raman analysis as an equivalent alternative to the GC analysis, thereby enabling use of the Raman analysis for commercial release. 

	The introduction of the at-line Raman method into the QC laboratory was successful, with QC chemists quickly understanding and appreciating the benefits of this mode of analysis. Key factors in the successful introduction were focused and concise documentation (including test methods, work instructions, etc.), a thorough and comprehensive training program, and the straightforward execution of the analysis. The sample analysis involved the placement of the sample in the HLSC and running the method on the RXN2 analyzer. Sample analysis time for the Raman analysis was 15 minutes compared with 240 minutes for the GC analysis, thereby enabling a significant reduction in result turnaround time using the Raman analysis (see 

It is important to demonstrate the continuing validity and accuracy of the chemometric model used in the spectroscopic method at regular time periods throughout the lifecycle of the method. This maintenance is necessary to account for the following: potential changes in vendors (which may utilize alternative synthetic routes for process reagents and solvents), instrument/model drift over time, and changes in the process manufacturing route. Validity is typically achieved by applying a change-control protocol and a model maintenance program to the chemometric model. Model maintenance for this IPC method involves repeating selected validation tests (e.g., specificity, accuracy, and precision) periodically, using the GC method as the primary reference method. The activities are documented in a protocol that is included in the maintenance report. Maintenance is performed annually or following any significant changes to the manufacturing process or raw material supply that are prompted by change-control actions. 

The introduction of the at-line Raman method resulted in reduced sample analysis times, reduced consumables for analysis (solvents, glassware), and increased sample throughput in the QC laboratory. As a result, the team assessed the feasibility of introducing an in-line application for the process to achieve even further savings by eliminating the need for cooling the reactor and sampling the reaction. Based on the high commercial volumes and the need to run frequent manufacturing campaigns, it was identified as an opportunity to save significant plant cycle time with a positive return on investment. The proposed in-line solution would involve the installation of a Raman probe (Pilot E, Kaiser Optical Systems) in the reactor, which would allow the reaction to be sampled directly in the reactor and eliminate the need to cool it. In addition, the processed IPC result would be automatically transferred to the DCS as part of the batch record. A cross-functional team comprised of representatives from manufacturing, engineering, analytical, information technology, and quality was assembled to perform a failure-modes effect analysis on the process step. Areas that were analyzed included in-line monitoring infrastructure requirements, mode of result generation and results reporting into the DCS, qualification and validation requirements, and equivalency/comparability requirements for the in-line analyzer. 

	The in-line application included developing an equivalent analysis model for an ATEX-rated Raman analyzer (RXN3-785, Kaiser Optical Systems) and acquisition of Raman spectra from the reactor vessel via a Raman probe (Pilot E, Kaiser Optical Systems) connected to the analyzer using fiber optic cables.

	Equivalency between the at-line and in-line models was demonstrated using an equivalency protocol based on the data generated from the at-line and in-line models for in-process samples from the same reaction. Software (SynTQ Lite, Optimal Industrial Automation) was applied to the analyzer to provide a 21 

) Part 11-compliant operating platform. Results generated were automatically exported to the DCS, which enabled real-time decisions to be made regarding progression of the manufacturing process. This installation eliminated the requirement to cool down the reactor, remove a sample, and perform an at-line analysis in the IPC laboratory, thereby generating even greater manufacturing cycle time gains.

	Process knowledge obtained by observing real-time Raman trends during the solvent-exchange step resulted in a temperature-control step being added to the manufacturing batch record to ensure the product remained in solution, which further improved batch throughput. The knowledge was obtained through investigating irregular DCM concentration profiles during the solvent-exchange step using the in-line Raman probe. Negative concentration results were being reported, and upon review, it was noted that the raw Raman spectra were atypical for the negative concentration results. At-line analysis of the reactor supernatant displayed typical concentration results for DCM. The irregular profiles were typically generated in the early morning during the winter/spring season. The process solvent (acetone) was stored in an outside tank and was introduced into the reactor through pipework that was exposed to the outside elements. There was no temperature control of the reactor lagging jacket. Following a thorough review of the parameters of the process (e.g., time of year, external tank, and bulk process solvent being added in large volumes), it was hypothesized that the bulk acetone was being added at a sufficiently low temperature to cause the product to precipitate out of solution, which impacted the Raman scattering effect and generated the negative concentration values. A corrective measure introduced to counteract this effect was to introduce a temperature control of not less than 18 °C to the reactor lagging jacket. Following the introduction, the irregular DCM concentration profiles were no longer observed, which improved batch throughput. 

	Case study 2: HCl molarity determination using FT-NIR

	A second opportunity identified to leverage spectroscopic-based methods to enhance the efficiency of the manufacturing plant and QC laboratory was the analysis of critical reagents and solutions. All reagents that are used in the commercial manufacturing process require release testing. In addition, many solutions that are prepared from these reagents will typically require an in-process release test before they can be used on the plant floor. The team identified the reagent HCl in MeOH as an application that might be suitable for analyzing using a spectroscopy-based alternative method. Because solutions of HCl in MeOH are hygroscopic and readily degrade on exposure to air, a rapid and accurate method was required to ensure that the correct concentration of reagent was being added to the reaction for each batch. The regulatory filed process called for the concentration of solutions of this reagent to be determined using simple acid/base titration. Despite the simplicity of this analytical technique, the analysis typically required over six hours from start to finish to generate and approve a result. This length of time was due to the numerous steps involved, including gathering the necessary glassware, solvents, and reagents; preparing and standardizing the various solutions; analyzing the sample; calculating the result manually; and reviewing and approving the result. The in-process analysis of this reagent solution had been identified as a process bottleneck, and the team assessed the option of introducing an equivalent spectroscopic method of analysis that could generate a result in a more expeditious manner. Based on initial feasibility studies including Raman, FT-IR, and FT-NIR, FT-NIR spectroscopy was identified as an appropriate technique for analyzing commercially sourced solutions of HCl in MeOH. The rationale for the selection of FT-NIR included sensitivity and selectivity for the compound of interest, ability to generate quantitative models, and availability of identical instrument models at both sites. 

	Results were generated using an analyzer (Antaris II FT-NIR, Thermo Scientific) in the transmission mode. The molarity of the reagent was determined using a partial least-squares chemometric model. The resulting FT-NIR method was capable of generating sample results in only two minutes. This short analysis time was enabled by embedding the chemometric model into the method, requiring only a single sample scan be acquired to generate a result. The analytical team at the BMS Ireland commercial manufacturing site had purchased an identical instrument to the one used by the ETG in the United States as part of the collaboration between the two groups. It had been previously recognized that alignment of both hardware and software components of the instrumentation at both sites would enable more effective knowledge transfer and troubleshooting/optimization activities, as well as enhance technology transfer of analytical methods. 

Transfer of the FT-NIR method to the manufacturing site was conducted following the co-validation mode of technology transfer as defined by 

<1224> (8). Validation tests performed per ICH Q2 (R1) guidelines including specificity, linearity, repeatability, accuracy, and limit of detection/limit of quantitation (7). Acid/base titration analysis was used as the primary reference method during the validation. A comparability protocol was also successfully completed, in which 10 different lots of HCl in MeOH were analyzed using both the primary titration method and the secondary FT-NIR method and the results generated compared against predetermined acceptance criteria. Upon conclusion of the co-validation and comparability protocol testing, the results were summarized, and a proposal was submitted to the BMS Ireland site quality assurance function to introduce the FT-NIR method as an alternative equivalent to the titration method. The proposal was accepted, and the necessary activities to introduce the method, including raising a change control action, issuing of the test method report, and training the QC and IPC chemists were performed.

	The acceptance of the introduction of the FT-NIR method has been uniformly positive from the process and analytical team members. Analysis is straightforward: the analyst decants the sample solution into a 2-mL vial, places the vial in the transmission cell of the spectrometer, and selects the appropriate workflow. A result is automatically generated by the instrument within two minutes. The ease of use and rapid result generation is well suited for the in-process control environment because it enables a result to be provided back to the process chemists within minutes of sample submission. Using the previous titration method, the processing team could be waiting hours for results, resulting in significant plant idle time. 

As with the Raman model detailed in case study 1, model maintenance is performed by repeating the validation tests using the primary titration method on an annual basis. If the site change-control management system captures any changes to the reagent supply parameters, instrument operation, or drug-substance manufacturing description that could impact the accuracy and validity of the partial least-squares model, model maintenance is triggered.

	The introduction of the FT-NIR method in the commercial setting was successful. Result turnaround times were more than 10 times quicker than with the titration method, which was particularly beneficial to the in-process control work stream because it simplified analysis and facilitated improved manufacturing throughput times for this process, while at the same time freeing up analyst time in the QC laboratory to work on other high priority issues. As a result of the benefits and efficiencies gained through the introduction of the FT-NIR method for HCl in MeOH, the site identified three additional reagent concentration IPC methods that were creating bottlenecks in the plant workflows. Alternative, equivalent FT-NIR methods have subsequently been developed and introduced into the QC laboratory, resulting in faster turnaround times, simplified sample analyses, and a reduction of the resource demands in the laboratory. 

	The adoption of spectroscopic-based methods can improve the efficiency of analytical operations both within the plant environment (i.e., in IPC) and the QC laboratory. The two case studies highlight the time savings and analytical throughput increases obtained by introducing Raman spectroscopy as an alternative equivalent method to GC for solvent exchange monitoring and FT-NIR spectroscopy as an alternative equivalent to titration analysis for process reagent analysis.

	In the Raman spectroscopy case study, significant efficiencies were obtained using both in-line and at-line applications. This example highlights an effective strategy for the introduction of spectroscopic technologies to a laboratory/manufacturing site. Initially introducing the at-line application to the site enabled the staff and management to gain a fuller understanding and appreciation of the advantages of the technology with a modest investment of time and effort. The subsequent introduction of the in-line application was achieved with strong support of management due to the prior experience with the at-line application. 

	The ease of use and ruggedness of the at-line applications facilitated enthusiastic uptake and ownership of the technologies by the laboratory chemists. At-line spectroscopic applications can facilitate improved analytical performance while minimizing the need for large resource investments, infrastructure, and maintenance activities associated with in-line/PAT applications. In addition, the at-line applications generated significant analytical operation efficiencies through improved sample throughput; reduction in analysis time, laboratory errors, and laboratory waste; increased laboratory capability; versatile analysis; and the capability to automate and network. 

	The authors wish to thank the various project team members in the US and Ireland that participated in discussions and/or collaborated in the development of these applications. In addition, they would like to thank the technical specialists at Kaiser Optical Systems Ltd, Optimal Industrial Automation Ltd., and Thermo Scientific Ltd., who facilitated the introduction of the spectroscopic instrumentation and software into the BMS manufacturing site in Swords, Ireland.

	1. FDA Safety and Innovation Act (FDASIA) Section 902, 126 STAT. 993, Public Law (July 2012), pp. 112–144.
	2. FDA, 

Guidance for Industry, Process Validation: General Principles and Practices 

(Rockville, MD, 2011).            3. ICH, Q8 (R2) 

Development and Manufacture of Drug Substances (Chemical Entities and Biotechnological/Biological Entities) 

Validation of Analytical Procedures: Text and Methodology

(ICH, 1997).
	8. USP, General Chapter <1224> “Transfer of Analytical Procedures”, 


	Vol. 43, No. 2
	February 2019
	Pages: 35–39

	When referring to this article, please cite it as J. Wasylyk, M. Huang, B. Wethman, and K. O'Connor, "Spectroscopy Facilitates Lean Analysis," 

	John Wasylyk* is senior principal scientist, john.wasylyk@bms.com; Ming Huang is senior research investigator I; and Bob Wethman is senior research scientist II, all at Bristol-Myers Squibb, USA; Kieran O’Connor* is Analytical Science and Technology senior chemist at SK biotek, Ireland, kieran.oconnor@sk.com.

	* To whom correspondence should be directed.

Articles

Spectroscopic-based control methods were introduced as equivalent alternative methods, first to a gas chromatographic method to monitor an in-process solvent exchange step and second to a potentiometric titration method to release a process input material for drug substance manufacturing. 

Spectroscopy Facilitates Lean Analysis

	Submitted: May 27, 2016. Accepted: June 24, 2016.  

An important operation in the registration approval of a new drug product is the commercial-scale validation campaign at a receiving manufacturing site, which requires the technology transfer of analytical methodology from the transferring site. Traditionally, Bristol-Myers Squibb (BMS) used the comparative testing technology transfer model. In a recent launch of a product with breakthrough designation status, however, BMS used the covalidation technology transfer model, which involves simultaneous method validation and receiving site qualification. The authors describe the development and implementation of covalidation strategies to accelerate analytical method transfer using examples from a pilot study.

	In today’s challenging business environment, the speed of launch to the market for new drug products is crucial. The FDA Safety and Innovation Act of 2012 established a breakthrough designation program (1) to expedite the development and approval of innovative drugs for serious and life-threatening conditions. This program has resulted in biopharmaceutical companies prioritizing research on potential breakthrough drug entities and reviewing and assessing business practices to identify opportunities to speed up the development and registration process.
	
	Within the pharmaceutical industry, method transfer is an integral part of the lifecycle of an analytical method (2). The 

defines the transfer of analytical procedures (TAP) as “the documented process that qualifies a laboratory (the receiving unit) to use an analytical test procedure that originates in another laboratory (the transferring unit, also named the sending unit), thus ensuring that the receiving unit has the procedure knowledge and ability to perform the transferred analytical procedure as intended” (3). The comparative testing technology transfer model involves initial validation of the analytical method at the transferring site, followed by subsequent transfer of the method to the receiving site. The covalidation technology transfer model involves simultaneous method validation and receiving site qualification.
	Covalidation is based on the premise that the laboratory that completes the validation of an analytical procedure is qualified to run the procedure. According to 

 <1224>, “the transferring unit can involve the receiving unit in an interlaboratory covalidation, including them as a part of the validation team, and thereby obtaining data for the assessment of reproducibility” (3). 

		Comparative testing performed on homogeneous lots of the target material

		The complete or partial revalidation of the analytical procedure by the receiving unit

	The transfer of analytical methods from a research and development laboratory to a production site is an important step towards the commercialization of a new pharmaceutical product. While traditionally, at BMS, method transfers were performed by comparative testing, in today’s environment of accelerated pharmaceutical development, covalidation between the transferring and receiving laboratories offers great advantages and hence warrants a detailed examination. The following study of covalidation includes the evaluation of the benefits of covalidation versus its potential risks, a decision-tree driven process to assess the suitability of using covalidation, and a case study of a recently completed analytical transfer involving 50 release testing methods.

	In contrast to comparative testing (TAP Type#1) or revalidation (TAP Type#3), the covalidation approach does not require the transferring laboratory to complete method validation prior to the initiation of covalidation at the receiving unit. This unique feature of covalidation enables method validation and method transfer to be performed in parallel instead of in series, which is particularly advantageous in situations where there is limited time for the qualification of the receiving laboratory. 
	
	The nature of covalidation requires the early involvement of the receiving laboratory as part of the validation team, which is a departure from the classical validation paradigm whereby development and validation of the method are performed by the transferring laboratory with little or no input from the receiving laboratory (4). This early engagement of the receiving laboratory enables methods to be evaluated and validated in the most relevant laboratory setting for its intended use and affords the opportunity to incorporate receiving laboratory input, align with global and local practices, and build receiving-laboratory-friendly features into the method conditions.

	Documentation for covalidation is streamlined by incorporating the procedures, materials, acceptance criteria, and results of the covalidation in validation protocols and validation reports, eliminating the need for separate transfer protocols and reports used in comparative testing.

	An effective covalidation process demands and encourages close collaboration and knowledge sharing. As knowledge transfer is the most important aspect of method transfer, enhanced method knowledge at the receiving laboratory facilitates receiving laboratory ownership for the methods.
	
	 

	The very nature of covalidation however, carries some risks when compared to the other TAP approaches.
	
	

Readiness of the method for covalidation.

 Methods subjected to covalidation are yet to be fully validated, therefore, it remains uncertain whether the method can meet all of the validation acceptance criteria. The risk that the method could fail to meet the validation criteria is much higher in instances where a method is not fully optimized and/or robustness and ruggedness of the method have yet to be extensively investigated. Any covalidation time saving will be lost if the validation at the receiving laboratory has to be repeated. The scenario of having a receiving laboratory participate in covalidation for a method that might need to be reworked is a risk that needs to be addressed prior to the commencement of covalidation.
	
	

Preparedness of the receiving lab at a commercial manufacture site is required earlier than normal. 

In contrast to a method that is transferred to a long-term stability testing site, where it will usually be put into use immediately after laboratory qualification is completed, transfer of a method to a commercial site usually occurs several months prior to its routine use for product release. The timing of covalidation, however, is dictated by the timing of method validation and as such, covalidation usually takes place earlier than method transfer. As a result, the time lag between covalidation and routine execution at a manufacture site poses two additional risks as follows:

 The possibility of a long time between covalidation and actual use of the method at receiving site with no re-enforcement of method experience can lead to a loss of knowledge.

 If the receiving laboratory is not fully invested in rapid execution, the time for validation completion may lengthen.

	A decision tree was developed to address the potential risks associated with using covalidation as the approach for method transfer (see 

). The key decision points to assess the suitability of a method for covalidation are:

		Are the method robustness results from the transferring laboratory satisfactory?

		Does the method use a technique with which the receiving lab is familiar?

		Is there a significant instrument or critical material difference (e.g., filter for drug product methods) between the laboratories?

		Is the time between method validation and commercial manufacture < 12 months? (This applies to a commercial manufacture testing laboratory only.)

	Because method robustness is the most important factor to determine if a method is suitable for covalidation, it is therefore crucial for the transferring laboratory to adopt a systematic approach for method robustness evaluation during method development to ensure adequate understanding and confidence for the method.

	A project in Phase III clinical development was chosen as the pilot for the application of the covalidation strategy for the analytical transfer for the drug substance and drug product. The following case study highlights the authors’ experience with drug substance covalidation. The covalidation exercise was undertaken between scientists from analytical groups in research and development (transferring unit), both in the United Kingdom and the United States, to a quality-control laboratory in the Global Commercial Manufacturing organization (receiving unit) in Ireland.

The pilot project included the API, two isolated intermediate compounds, three regulatory starting materials (RSMs), and all associated reagents used in the process steps associated with these compounds. The traditional approach to analytical method transfer at BMS required the comparative testing on several homogeneous lots of the target material for all critical tests, such as determination of assay, impurities and degradants, residual solvents, counter ions, enantiomeric impurity, and inorganic impurities, combined with transfer waivers for compendia tests such as identity, water content determination, appearance, and color. The comparative testing model would have required comparative testing be conducted on 60% of the total number of analytical methods transferring to the commercial facility. For the pilot project, the analytical method transfer needs were reviewed against the newly mapped covalidation technology transfer model. A comparison of the two approaches is shown in 

By using the covalidation approach, the proportion of methods transferred using comparative testing was reduced to only 17%. Comparative testing was not replaced entirely due to several of the methods being fully validated prior to the decision to accelerate the project timelines.

	The covalidation approach was exclusively applied to high-performance liquid chromatography (HPLC) and gas chromatography (GC) methods for the various manufacturing steps in the case study. Validation criteria for both modes of analysis are detailed in 

, and reproducibility testing was performed at the receiving laboratory.

	To gain a greater understanding of the impact of this change, resource utilization figures were compiled for the two approaches. 

 outlines the number of hours spent on each key step for one drug substance analytical method following each of the two processes. 

 demonstrates that traditional comparative testing would take approximately 11 weeks to complete from starting the analytical method validation process to approval of the technology transfer report and subsequent qualification of the receiving site to perform analyses using the method. Adopting the new covalidation model delivers the same end result in eight weeks--a saving of over 20% less time for each covalidation as compared to comparative testing. Overall, the time for the traditional model was 13,330 hours and the time for the covalidation model was 10,760 hours (see 

 In the case study, quality by design (QbD) approaches were used to evaluate method robustness as part of the method development process. For example, for the development of the HPLC purity/impurity methods for Intermediate #1, multiple variants (e.g., binary organic modifier ratio, gradient slope, and column temperature) were evaluated in a model-robust design (see 

). Based on the identification of critical method parameters during method development, the method robustness ranges and performance-driven acceptance criteria were established.

Troubleshooting and method understanding.

 The collaborative approach used during the covalidation workflow enhanced the knowledge transfer between the transferring and receiving laboratories. Regular communication ensured that both units were fully aligned in achieving the optimum outcome. This collaboration required a cultural change by both units, as the transferring unit needed to engage with the “customer” (receiving unit) during the method validation stage of the analytical method lifecycle, and the receiving unit needed to engage in method troubleshooting and optimization of the analytical method where required. This new mode of operation modified the traditional work practices and resource allocations within both units. For example, there was a greater requirement for technical expertise amongst the staff engaged at the receiving unit than would be typically required for a transfer using the comparative testing model.

	Early input from the receiving unit identified potential roadblocks to the application of several analytical methods in their laboratory. A validation team was formed, including representation from both the transferring and receiving units who worked together to identify solutions to the issues and incorporate the changes into the validation exercise and final method. 

	One such issue arose during the covalidation of a method to determine residual solvents by GC that used headspace sampling and flame ionization detection. At the commencement of covalidation discussions between the two laboratories, an equipment list was created that indicated both laboratories had the same GC equipment. However, during familiarization trials at the receiving lab, when data from the two laboratories were compared, it was noted that early eluting peaks observed at the transferring unit were consistently larger than those obtained by the receiving unit. A thorough review of the make and models of the instrumentation confirmed that the instrumentation was the same in both laboratories, but the instrument configurations were different.

	If a comparative testing approach had been used, this issue would have halted the technology transfer while the issue was being resolved. It could have resulted in additional validation testing by the transferring laboratory and would have lengthened the time required to qualify the commercial laboratory. The collaborative nature of the covalidation approach resulted in the validation team rapidly identifying the root cause of the issue and incorporating the different instrument configurations into the covalidation exercise and final method documentation with negligible impact to the overall timeline.

Another outcome of the more collaborative and intensive communication between the two units was the identification of a planning tool. Frequent meetings, phone calls, and e-mail correspondence between both units, in addition to a significant volume of duplicate/redundant work, were identified as inefficiencies within the transfer. A review  identified a more streamlined and efficient process, which resulted in adoption of the Microsoft Project planning tool for managing the overall transfer. The planning tool was customized to reduce the number of meetings, streamline documentation and communication activities, and optimize the scheduling of transfer testing through the use of Gantt charts.

		Use of Gantt charts to link and track all project activities

		Identification of critical risk factors for the technology transfer via technical risk and associated risk assessment grading (RAG) columns

		Use of filters to facilitate efficient review of relevant tranfer activities

		Controlled write access to the tool to ensure all information is current and accurate

		Use of hyperlinks for access to controlled transfer documentation

		Centralized location for all relevant project information (e.g., test methods, validation documentation, project timelines).

	The tool has proven to be an effective mechanism to manage the multiple simultaneous activities ongoing at the transferring and receiving sites during covalidation.

	Traditionally, analytical method qualification of receiving laboratories is performed using comparative testing technology transfers, which involve validation of the method at the transferring unit, followed by technology transfer to the receiving unit. Covalidation is an alternative technology transfer model that combines method validation and receiving unit qualification in one process. To reduce the business risk of potential covalidation failure, a decision tree was established with the key requirements outlined. The decision tree highlights the importance of robustness study of any method before proceeding with covalidation.

	The primary impact of using the covalidation model is the expedited analytical method qualification of both the transferring and receiving laboratories. Covalidation of analytical methods also affords other benefits to both transferring and receiving units, namely through the earlier engagement by both units in the analytical method lifecycle, thereby ensuring the relevant test method will be optimized for operation in the receiving unit. This early engagement facilitates input from the receiving unit, supporting a more efficient subsequent covalidation and qualification of the receiving unit than the comparative testing model. This approach also results in more efficient knowledge transfer between both units through method optimization and troubleshooting activities. 

	A cultural change is required to optimize the output of the validation team, which is comprised of members from both the transferring and receiving laboratories. Additionally, resource allocation for covalidation is very different from the comparative model. The covalidation process requires the receiving laboratory to allocate resources at an earlier stage in the analytical method qualification process, and also potentially to a greater degree than the comparative transfer model. Additionally, covalidation requires efficient organization and planning at both laboratories to ensure simultaneous activities are coordinated and scheduled so that transfer deadlines can be met. Use of a tool, such as Microsoft Project, facilitates the planning of covalidation activities and scheduling of resources.

	The covalidation pilot detailed in the case study was successful in ensuring qualified analytical methods were available to support the manufacturing validation campaign at the receiving site as part of the expedited launch of the new drug product. Application of the covalidation model and transfer waivers greatly reduced the cycle time for the analytical transfer element of product development. As a result, covalidation is the preferred mode of analytical technology transfer within BMS where early identification and engagement of the receiving site is required.

	The authors wish to thank the various project team members in UK, US, and Ireland who actively participated in the pilot programs for covalidation. 

		FDA Safety and Innovation Act (FDASIA) Section 902, 126 STAT. 993, Public Law (July 2012), pp. 112-144.	

		S. Scypinski and J. Young, “Analytical Methodology Transfer” in 

Handbook of Modern Pharmaceutical Analysis

, S. Scypinski and S. Ahuja Eds. (Elsevier, Amsterdam, Netherlands, 2nd ed., Vol 10, 2011), pp. 507-526.	

 General Chapter <1224> “Transfer of Analytical Procedures”, pp.1638.	

Validation of Analytical Procedures: Text and Methodolog


	Vol. 41, No. 4
	Pages: 38–41, 44–48

:
	When referring to this article, please cite it it as K. O’Conner, N. Hulme, and Y. Shi, “Covalidation Strategies to Accelerate Analytical Method Transfer for Breakthrough Therapies,” 


	Kieran O’Connor is a quality control systems technical specialist, Bristol Myers Squibb, Ireland, kieran.oconnor@bms.com; Nicola Hulme is a senior research scientist, Bristol Myers Squibb, UK, nicola.hulme@bms.com; and Yueer Shi is group leader, Bristol Myers Squibb, USA, yueer.shi@bms.com.

The primary impact of using the covalidation model is the expedited analytical method qualification of both the transferring and receiving laboratories.