Practical Approaches in Applying Process Analytical Technology (PAT) Tools to Early Active Pharmaceutical Ingredient (API) Development

Introduction

The amount of time required to successfully develop a new drug candidate from discovery to commercialization is estimated to be 12-15 years and the associated cost to be approximately $1 Billion dollars [1]. For each new medicine successfully delivered to the patient there are thousands of drug candidates that fail to make it to the market for a variety of reasons, i.e., efficacy, safety, etc. Many more drug candidates are tested in early development stage (Pre-clinical, Phase I and II) than in late stage (Phase III). On average, for every 10,000 compounds that are discovered/screened only one successfully makes it to commercialization. To cope with the workload and to contain the ever increasing development cost, early drug development has to be efficient and fit for purpose [2], while maintaining the quality of clinical trial material to ensure the safety of the clinical trial subjects.

Traditional analytical support for early API synthesis process development has mainly relied on chromatographic techniques. These analytical tools are accurate and reliable, but are off-line and need sample preparation with at least one step of sample dilution, which does not allow real-time feedback to the chemists with respect to what is occurring in a chemical reaction. Moreover, they often cannot be used in situations where there are unstable or transient intermediates. Spectroscopy based Process Analytical Technology (PAT) tools such as FTIR and Raman process analyzers are increasingly used in pharmaceutical API and drug product development. They have become valuable tools for evaluating design space and indispensible tools for Real Time Release (RTR) [3-5].

PAT tools can also be applied to early API development for reaction monitoring. However, the focus is somewhat different from that for late stage and commercial product development. The primary goal of reaction monitoring in early API development is to obtain better process understanding. As the focuses are different, so do the approaches. We here share our experience and practical approaches in applying various online/inline PAT tools to early API development in Pfizer Groton Laboratories.

Practical Approaches for Applying PAT Tools to Early API Development

Formation of a Reaction Monitoring Core Team

In 2008, a multidisciplinary reaction monitoring core team was formed with members from process chemistry, analytical development, and API scale-up facility. The core team focused on evaluating the feasibility and the implementation of PAT tools during chemistry enabling work, which is typically performed as a drug candidate transitions from Drug Discovery to the Pharmaceutical Sciences Research group. The core team promoted the use of PAT tools by serving as technical experts, championing the development and adaptation of new PAT tools, and acting as trainers to the process and analytical chemists on the use of the PAT tools. To further facilitate the utilization of the PAT tools, several analytical members of the core team collocated into the chemistry laboratories. Thus, formation of a reaction monitoring core team not only enhanced communication, but also promoted broader and closer collaboration between process and analytical chemists.

Figure 1 - Reaction monitoring implementation work flow process in early API development

Adaptation of Reaction Monitoring into the Workflow Process

The core team optimized and implemented a workflow process (See Figure 1 ) that served the needs of both the process and analytical chemists. The goals of implementing this work flow include (1). to encourage the process and analytical chemists to evaluate the use of PAT technologies as soon as they are assigned a new project, (2) to foster collaboration between the process and analytical chemists, (3). to remain in contact with them during the entire synthetic route development, (4). to gain better understanding of the chemical process in less time and using less materials during the route optimization iterations, and (5). to confidently nominate a suitable PAT-based analytical method for In-Process Control (IPC) and transfer the online method to internal or external API manufacturing scale-up facilities.

Simultaneous Use of Multiple Probes to Monitor a Single Reaction

The core team promoted simultaneous use of multiple vibrational spectroscopic tools such as both middle IR and Raman process analyzers for each chemical reaction to be studied. We found that this is especially beneficial for the early stage API development. Often in early API development, the project analytical chemist does not have much prior information about how to monitor a particular reaction, because the process chemist generally has not run the reaction before and/or just initiates a synthetic route enabling work screening. The benefits of using multiple probes simultaneously include the following: (1) to obtain multiple sets of data from a single experiment to save time, (2) to improve process understanding from richer and complementary information derived from using multiple probes with different selectivity’s, and (3) to allow for comparing and contrasting of the data in order to select the best technique for an online IPC method going forward.

It is interesting to note the similarities and the differences of Raman vs. IR spectroscopy as applied to reaction monitoring. Both are spectroscopic tools originating from molecular vibrations, and both can be used as process analyzers with immersion probes that have almost the same chemical and thermal tolerance. The Raman spectrum usually has fewer but sharper peaks, so peak overlapping is less of a problem in a Raman spectrum than in an IR spectrum. However, fluorescence is a major issue that may hinder the use of Raman spectroscopy as a reaction monitoring tool. IR spectroscopy is a universal tool, as IR absorption is ubiquitous for organic molecules. Nevertheless, it also makes IR reaction monitoring more susceptible to interference from bands of solvents and other reagents. It is generally harder to find an isolated single band for trending analysis in IR monitoring than in Raman monitoring. Multivariate analysis may have to be used. As water is a very stronger IR absorber but a weak Raman scatter, Raman has advantage for monitoring reactions in aqueous media. IR monitors changes in the solution phase only, while Raman monitors changes in the illuminated volume, whether liquid or solid.

Focus on Key Applications

To cope with large number of drug candidates that have to be developed and high attrition rate in early drug development, the core team has to be prudent in resource allocation. Therefore, we decided to focus on key chemical reactions that benefit the most from real-time reaction monitoring. The reactions of interest include those where there are reactive/unstable intermediates, critical endpoints, high energy reagents, and those that are difficult to monitor by conventional offline methods (i.e., HPLC, GC, NMR, etc.) or difficult to sample, e.g., reactions involving high pressure, high temperature or cryogenic reactions. Previously, one of the core team members has published a case study of halogen-lithium exchange reaction under cryogenic conditions, where unstable starting material and product were involved [6]. We present here another example of how reaction monitoring was used to monitor the reductive amination via imine formation, where the imine intermediate is not suitable to offline HPLC analysis.

Reaction Monitoring to Understand Imine Chemistry

Imine formation is an important class of organic reactions that process chemists frequently use to make secondary amines. A general reaction scheme is shown below. The imine is formed via equilibrium and the reaction rate is substrate dependent. Adventitious water was believed to influence the equilibrium negatively.

In one of Pfizer’s early development projects, a reductive amination reaction suffered from poor and variable yields. The chemist wondered whether this was due to incomplete imine formation, adverse effect of ‘wet’ solvents (containing water), or something else. Conventional off-line HPLC analysis is not applicable to monitor this reaction because the imine is not stable under HPLC condition and reverts back to the starting materials in the presence of excess amount of water. A bench top Raman spectrometer had been used to monitor imine formation [7]. We decided to use both Raman and IR process analyzers with immersion probes to study our reaction of imine formation and subsequent reductive amination. The Raman process analyzer was equipped with a 785 nm laser source and a back illuminated CCD detector cooled to -40 ⁰C. The probe used was a fiber optic probe attached to an alloy C-276 based probe head (dimension: 8’ long X ¼’ O.D.) with a Sapphire optical window. The IR process analyzer was equipped with an interferometer and a MCT detector, which was cooled by liquid nitrogen. The IR probe used was a 1.5 m fiberoptic attached to an alloy C-276 based probe head (dimension: 8’ long X ¼’ O.D.) with a Diamond optical window.

Figure 2 - 2-D plot of reaction monitoring via Raman process analyzer. Time represents reaction progression time, only Raman shift range of interest is shown.

The reaction sequence was as follows: (1). add the aldehyde and potassium acetate in methanol and acetic acid to the reaction flask, and start reaction monitoring using both Raman and IR process analyzers; (2). add the amine to initiate the reaction, and monitor imine formation until the reaction is complete; (3). add sodium triacetoxyborohydride (STAB) to initiate reductive amination, and monitor the conversion; (4). add water to quench the reaction, and work up the product for yield. Raman band changes are shown in Figure 2 as a 2-D plot and in Figure 3 as a 3-D plot. The band at 1700 cm-1 is due to C=O stretching mode of the aldehyde and can be used to monitoring the disappearance of the aldehyde starting material. The band at 1643 cm-1 is due to C=N stretching mode of the imine intermediate and can be used to monitor the formation and subsequent disappearance of the intermediate. The trends from these bands are shown in Figure 4 with illustration of reaction sequence.

Figure 3 - 3-D plot of reaction monitoring via Raman process analyzer. Time represents reaction progression time, only Raman shift range of interest is shown.

Figure 4 - Raman trending plots for the imine (blue) and the aldehyde starting material (red).

Raman monitoring successfully profiled the reaction progression over time. It demonstrated that the imine formation under the experiment condition was complete in less than 5 minutes. The subsequent reductive amination was complete in about 25 minutes. However, there was still about 30% intermediate imine remaining in the reaction mixture, and the remaining imine was reverted back to the reactants as soon as 50 mL water was added to quench the reaction. This and subsequent more reaction monitoring by Raman process analyzer demonstrated that the imine formation step was fast and robust. The reaction tolerated a small amount of water. It was the reductive amination that did not go to completion, causing low yield. The real-time visualization of the chemical processes was invaluable to the chemist in understanding why the reaction did not produce the expected yields, and allowed him to adjust the conditions to improve yields in a much shorter time that would had been impossible without the real-time reaction monitoring.

The IR profile was qualitatively in agreement with the Raman profile. However, the IR spectra had strong solvent bands from methanol and acetic 2011acid, which made trending analysis of both the aldehyde and the imine bands much harder. Raman monitoring for this reaction gave better quality data that was easier to analyze due to no overlapping bands.

Conclusions

A multidisciplinary team consisted of process chemists, analytical chemists, and personnel from API scale-up facility is an effective way to promote use of PAT tools in early APT process development. Reaction monitoring has to be integrated into the work flow process to allow core team members to engage process and analytical chemists as soon as a new project is assigned and to remain in contact with them during the entire synthetic route development.

Simultaneous use of multiple PAT tools for reaction monitoring provides rich and complimentary information for better process understanding and optimization in less time. It also allows for the comparing and contrasting of data in order to select the best technique for an online IPC method going forward.

Focus on key applications enables efficient use of limited resources and maximum benefits to cope with the large number of drug candidates that an early development team has to work on. To date PAT tools have been successfully applied to study over 100 reactions in early phase API synthesis route enabling work, resulting in quicker optimization, less use of materials, and improved process understanding.

Acknowledgements

The authors thank many RAPI process chemists who collaborated with us, and other current and past RM core team members, John Sagal, Zhijun Zhang, Yong Zhou, Eric Hansen, Brian Preston, Sarah Griffin, Michael Pelletier, and Peter Larkin, who participated in many reaction monitoring activities. We also thank RA and RAPI managers, Richard Irwin, Russell Linderman, Cheryl Hayward and Stephen Brune, for support. Finally we thank Steve Brune, who critically reviewed the manuscript.

References

1. Adams, C. P., Brantner, V. V. Spending on New Drug Development. Health Economics 2010, 19: 130–141.
   
2. Van Arnum, P., Strategies for Early-Stage Drug Development Bristol-Myers Squibb’s fit-for-purpose model. Pharmaceutical Technology Sept. 2010. s17-s19.
   
3. E. K. Read, R. B. Shah, B. S. Riley, J. T. Park, K. A. Brorson, A. S. Rathore, Process analytical technology (PAT) for biopharmaceutical products: Part II. Concepts and applications. Biotechnology and Bioengineering 2009, 105(2), 285-295.
   
4. A. S. Rathore, R. Bhambure, V. Ghare, Process analytical technology (PAT) for biopharmaceutical products. Analytical and bioanalytical chemistry 2010, 398(1), 137-54.  
5. W. Chew, P. Sharratt, Trends in process analytical technology. Analytical Methods 2010, 2, 1412-1438.
   
6. Li, S., Application of Online reaction Monitoring By Raman and Infrared Spectroscopy in Early Drug Development: Halogen-Lithium Exchange Chemistry. American Pharmaceutical Review 2010, 13(2), 62-67.
   
7. Lee, M., Kim, H., Rhee, H., and Choo, J., Reaction Monitoring of Imine Synthesis Using Raman Spectroscopy. Bull. Korean Chem. Soc. 2003, 24(2), 205-208.

This article was printed in the July/August 2011 issue of Pharmaceutical Outsourcing, Volume 12, Issue 4. Copyright rests with the publisher. For more information about Pharmaceutical Outsourcing and to read similar articles, visit www.pharmoutsourcing.com and subscribe for free.