|
|
|
|
|
|
|
|
Home > QbD Viewpoint > Categories
|
4/29/2009
Despite a bountiful enumerated list of benefits, implementation of Quality by Design in pharmaceutical development is moving at glacial speed. Needed are cheaper, easier and faster methods for achieving QbD goals. The source of the problem and its potential solution, featuring Inference for QbD, are outlined in an article by Paul Thomas, Senior Editor of pharmaQbD, entitled:
Can Bayesians Get QbD Past Tipping Point?
The article is bases on a persistent refrain heard in pharmaceutical development comprised of the following:
- Clear business benefits of QbD need to be demonstrated before large-scale implementation. However there is a reluctance to initiate demonstration programs because current QbD approaches require a large investment of resources (people, time and materials) to execute the requisite scope of experimental work.
- There is a plethora of prior data from past development studies. However, using classical risk assessment, DOE and multivariate analysis, there is no way to integrate prior data with new studies. Huge benefits would be gained if one could make effective use of the abundance of prior data for QbD planning, execution, decision making and CMC filings.
10/16/2008In the previous blog on "Transitioning to QbD," it was noted that QbD workers will need an accessible modeling and simulation environment for QbD to achieve a central role in pharmaceutical development. Discussions with workers indicate that they are currently using a bewildering array of software, which was not designed for QbD. Some software like DOE software is highly restrictive and performs a limited subset of QbD tasks. Other software like general purpose statistical analysis software is too broad and performs many more tasks than necessary for QbD. The net result is that today QbD work processes involve a complex combinatorial array of software to get the QbD work done.
Unfortunately cobbling together software from a variety of internal and external sources, illustrated above, has led to an array of problems and shortcomings including the following:
- QbD workers get a lot more features than they need for completing QbD tasks.
- There is huge learning and retention problem for QbD workers.
- QbD workers are faced with complex data integration and management issues.
- There is no audit trail of how the results were obtained.
- Assembling QbD results require inefficient and non-reproducible cut-and-paste processes.
- There is no easy way to extend the functionality of the software as functions are hard wired in.
- And, the situation is costly to set up and maintain.
My experience in developing and implementing electronic lab notebooks in pharmaceutical R&D as replacements for the paper notebook (P. van Eikeren, "Intelligent Electronic Laboratory Notebooks for Accelerated Organic Process R&D," Organic Process Research and Development 2004, 8, 1015-1023) made it clear that pharmaceutical R&D workers expect focused software tools, devoid of extraneous functions, designed to help them get their jobs done. General purpose tools simply won't do. In response to this need, Blue Reference has initiated a project, entitled Inference for QbD project, directed at development of a software umbrella appropriate for implementation of QbD practices. The Inference for QbD project, described on the website at www.InferenceForQbD.com, encompasses a number of novel elements including the following:
- A comprehensive software solution for the implementation of QbD practices in a pharmaceutical environment;
- It is being constructed on the patent-pending Inference platform developed by Blue Reference;
- Users access it through the easy-to-use and familiar user interface of Microsoft Office;
- It is being progressed from technical feasibility to development in a project funded by the National Science Foundation (NSF) Small Business Innovation Research (SBIR) program; and
- It is being development into a commercial product within the context of a consortium of pharmaceutical companies whom are providing guidance on requirements and testing prototypes in production settings.
Pharmaceutical customers tell us that Inference for QbD should address the needs of a broad audience and enable guided decision making-that is, analyze development and manufacturing data within the context of QbD objectives with the intent to identify the best way to move pharmaceutical development forward. The resulting implementation, illustrated in the figure above, represents a greatly simplified approach to the implementation of QbD practices. Specific benefits include the following:
- Capabilities of the software are tailored to QbD requirements and can be further tailored to company-specific best practices;
- Users only see the functions that they need to get the job done without extraneous distractions;
- Users experience a shallow learning curve as a result of using the familiar Microsoft Office interface;
- Data preparation and management are tightly integrated;
- The software enables concurrent data assembly and preparation, analysis and documentation, thereby providing an audit trail of how the results were obtained;
- The software generates QbD results documents automatically with the press of a single button;
- The software is adaptable and extensible using the Inference platform SDK allowing for future extension and re-direction; and
- The software is inexpensive to set up and maintain because it is based on Microsoft Office, which is already deployed.
Demonstration of these benefits through relevant, illustrative examples will be the subject of future blog entries. 1/29/2008
The International Foundation Process Analytical Chemistry (IFPAC) completed the first day of their 22nd International Forum on Process Analytical Technology covering a full cross-section of subjects focused on the application of Process Analytical Technology to QbD. Leading the Plenary lectures was Dr. Janet Woodcock, Deputy Commissioner, FDA on the subject of the FDA’s Critical Path Initiative.
Woodcock provided new, sobering statistics on the state of today’s pharmaceutical industry including:
- Mergers and acquisitions have decreased the number of new drug candidates because post-merger, similar candidates are dropped.
- New compounds entering Phase I clinical trials have only an 8% chance of reaching the market versus a 14% chance 15 years ago.
- More worrisome is the fact, even after significant investment and “sunk costs,” the Phase III failure rate is now 50% versus 20% a decade ago.
- Costs continue to escalate and companies are less willing and able to bring new candidates forward.
- Today 65% of all prescriptions are generics and there are tremendous challenges in getting them to market and ensuring that they are truly equivalent to brand-name drugs.
- Although there has been huge private and public investment in basic research and specific product development, there has only been minor investment in tools and public standards to aid development.
Woodcock indicated that in 2004, the FDA concluded that “…basic research isn’t enough. We have to look at the critical path that a product must follow before it is introduced to market...The science required to evaluate new product safety and efficacy and to enable manufacture is different from basic discovery science.” The outgrowth of that effort was the Critical Path concept, with quality-by-design as a core feature.
Woodcock made a number of suggestions for insuring the success of the Critical Path concept including:
- We need to build a generalized knowledge base of manufacturing science.
- Industry needs to share their existing knowledge and databases. This information should not stay within the walls of one organization, but must be shared to develop generalized knowledge.
- Sharing will be required in order to develop public standards that will enable evolution of the field for everyone to use.
Woodcock explained, the 21st Century GMPs document was a prototype for the larger Critical Path Initiative. “We are accelerating the pace of introduction of new science and technology,” she said. “PAT was the poster child. It may be a small piece of the picture, but is emblematic of the problems facing the industry.” She added, “We want to move from empirically derived trial and error methods (e.g., formulation, excipient selection) to rigorous, mechanistically based and statistically run processes. We need to break down silos between R&D and production and to be less conservative.”
She noted that the challenge of linking product attributes to clinical outcomes continues. There remains “a disconnect” between clinical and manufacturing sides, Woodcock said, adding, “We still don’t know how the attributes that you measure and control for pharmaceuticals actually control the clinical performance, or the extent to which manufacturing failures adversely affect clinical outcomes.”
Today, the drug safety debate is centered on intrinsic safety problems inherent to the drug itself, rather than errors in manufacturing. However, Woodcock said that suboptimal formulations occasionally get to market and fail to improve over time. She noted that “…for science-based manufacturing, you need to have a better idea of how parameters impact clinical performance and to exercise tighter controls.” |
|
|
|
|
|
|
|
| |
|
© 2008 Blue Reference, Inc. All rights reserved. | 3052 NW Merchant Way, Suite 100, Bend, Oregon 97701 | 541-316-2343 |
Terms of Use
 |
|
 |
|
|
|
|