Abstract
During active pharmaceutical ingredient (API) manufacturing process development, a balance is needed between effective management of limited budgets and resources and obtaining and implementing sufficient process knowledge in a way that demonstrates process capability and control. This balance is necessary in order to manufacture APIs safely, reliably, and economically. This is particularly challenging for companies that are mostly or completely virtual, since their prevalent bias is towards both cost minimization and rapid development. A requirement common to all strategies, at all types of organizations, is adequate documentation and justification of the ultimate process used for late-stage manufacture and commercial launch, in order to demonstrate that quality by design (QbD) has been implemented. There is an inherent tension between the need to move as rapidly as possible, and the need to accumulate sufficient knowledge along the way that can be leveraged at each successive stage of development. This article explores these competing priorities and proposes strategies for how these seemingly contradictory needs may be accommodated in a single, integrated approach.
Introduction
The impetus for this article was provided by a panel discussion held at PharmaChemOutsourcing in September 2014, during which a major focus of the discussion was how much effort to invest in developing and understanding the chemical processes used to manufacture active pharmaceutical ingredient (API) at early and middle stages of new chemical entity (NCE) drug development.1
Attrition at early stages of drug development is high. Budgets and resources for small, innovative start-up companies, from which many approved drugs now emerge, are severely constrained. These are both reasons not to indiscriminately undertake an exhaustive exploration of all aspects of the synthesis and characterization of the molecule of interest.
On the other hand, the potential longer term costs and adverse consequences, both financial and time-wise, of not accumulating sufficient process knowledge at even the earliest stages of API process development are considerable. Lack of adequate understanding and control of the chemistry being implemented can result in batch failures, delays, and time wasted in retroactively attempting (ie, back-filling) to gather and use information that should already have been in hand. In addition, the need for significant synthetic route or process changes presents the problem that a suboptimal manufacturing process used at previous stages limits the flexibility to implement changes for regulatory reasons, unless time and money is also available for bridging nonclinical development work, particularly toxicology studies.
Virtual sponsor pharmaceutical companies may also hew to this more short-sighted approach based on the plan to partner or sell a given program before there is an absolute need to address the critical shortcomings of an API process. This rationalization ignores the fact that the value realized in such business agreements depends on the value created in the asset prior to negotiation. The perceived and actual value of having API and drug product processes that can be moved forward and transferred with minimal technical and regulatory restrictions is vastly underestimated by the uninitiated. In addition, waiting to attempt to build quality into a product until it is prepared for sale or partnering, although it is commonly done, is akin to testing quality into an API or a drug product, and is not a recommended strategic approach or considered acceptable from a regulatory point of view.2 Quality is an integral aspect of a product that needs to be present and apparent from relatively early on in the development cycle.
Quality by Design and Its Relevance to Early and Mid-Stage API Development and Manufacture
Ronald D. Snee is a mathematician and statistician who is an expert in process improvement and implementation of quality by design (QbD). According to Snee,3 QbD at its essence represents “greater process understanding.” He describes 3 critical systems for the implementation of QbD:
- Knowing when you have adequate process understanding
- Strategy for getting the right data in the right amount at the right time to create the design space
- Integrated system for process control and process improvement
The system most relevant to this article is the first one. Adequate process understanding is able to identify and explain all critical sources of process variability and allows development of a process that manages that variability.4 In his description of a strategic approach for experimentation necessary to implement QbD, Snee describes 3 phases3:
- The screening phase, in which the effects of a large number of variables are explored, in order to narrow the initial set to a smaller set of variables for further study
- The characterization phase, in which individual effects of variables and how the variables interact with each other is assessed
- The optimization phase, in which a predictive model is developed, leading to a design space, in which a distinction is made between parameters which are critical and those which are not
The screening phase of experimentation has the most relevance and timeliness, regarding the manufacturing process for an API that is in a development program in transition from Phase I to Phase II development. The characterization and optimization phases of experimentation depend critically on the input obtained from the screening phase of QbD implementation.
Subsequent to (or at times, overlapped with) Phase I, a virtual company’s key objective is to be able to plan and facilitate timely Phase II supply delivery by their selected vendors. If the sponsor (virtual company) has moved as quickly as possible to file an Investigational New Drug application, there is a likelihood that little/no work was done to accumulate sufficient knowledge about the API (eg, impurity profile and identification and its implications for process control, or characterization, particularly solid state) or how it was made. This results in a technology transfer package (or a state of knowledge, if the work is kept at the same contract development and manufacturing organization as used for the Phase I work) that has numerous gaps. It compounds a situation in which time and budget allocations for additional work may not be sufficient to improve the chemistry to manage risk and ensure timely delivery of the next quantities of API. One often hears the phrase “plan for success” in the context of pharmaceutical development. The irony is that, even if it is taken to heart, a lack of understanding of what this really means often prevents this sort of planning to the extent necessary.
Most virtual pharmaceutical companies have as their Chemistry, Manufacturing, and Controls (CMC) objective just in time manufacturing of clinical supply, without allowance (budgetary and/or time-wise) for accumulation of the sort of information that increases the likelihood for success at larger scales and later stages of development and manufacturing, leading to process definition, registration, and validation. This often puts the company in the highly risky position of needing to demonstrate, scale, register, and validate a process that is not ready for any of these activities.
How can this problem be avoided, resolved, or minimized? The remainder of this article defines the problem in a way that allows identification of the most critical areas of focus for risk management and a suggested approach to addressing them in a timely and economic fashion.
Analysis of the Conflict: Competition in the Management of 2 Types of Risk
The tension between the need for “API, ASAP” and the need for incremental accumulation of process experience and knowledge at all stages of development has been aggravated in recent years. The development lifecycle for NCE drugs is now typically significantly less than 10 years (at times as little as 5 years) from candidate nomination5; as a result, drug supply is required over increasingly shorter time spans at all stages of development. CMC development and manufacturing activities therefore spend a much higher proportion of time on the critical path for overall development timelines. This is not unusual for earlier stage development, but it is also now often the case for mid- and late-stage development, in which there was previously more “breathing room” to address or mitigate limitations and risks in API processes, while clinical and nonclinical data was being evaluated.
The compression of the drug development cycle as described above has had the effect of bringing into much more stark relief the conflict described above.
Tactical vs. strategic API development boils down to competition between the attempts to manage 2 types of risks inherent to NCE drug development:
Management of the Risk of Compound Attrition is a strategy based on the high failure rate of compounds in early- to mid-stage development and is, at its most minimal, purely tactical. This approach entails little upfront investment in the development or understanding of a given production and manufacturing process. It emphasizes investment of sufficient resources and effort to provide enough material to allow forward progress of the nonclinical and clinical programs and to meet development milestones nominally. An advantage of this approach is that it can result in effective cost containment and timeline compression. If the candidate molecule fails early, less money and resources are wasted. The downside is that this approach is often to the detriment of necessary incremental understanding of the process and the molecule. A poorly defined enabling API manufacturing process is often ill-suited for or at odds with the rapid next stage development and scale-up effort which is in line with the spirit of this strategy. This often has a “ripple effect” on drug product manufacturing, since incomplete understanding of physicochemical properties leads to isolation of an API with suboptimal characteristics for its intended downstream use.
Management of Risk of Lack of Preparedness in the Event of Program Success is a truly strategic approach, more in line with a traditional approach to API process development. It takes a prospective rather than a retrospective stance on risk management and makes the identification and resolution of avoidable problems possible and likely. As a result, any problems that are encountered are more likely to be those that were truly not able to be anticipated, and this allows for increased flexibility and options for solutions to unexpected problems, based on sufficient process understanding. Judicious application of this approach results in a lean approach to process research and development (PR&D) that still reveals technical gaps and reduces risk. There are potential disadvantages. If the effort is not appropriately managed, a pursuit of knowledge for its own sake may occur rather than focusing on issues specific to prioritized needs and objectives. Poorly focused decisions or inappropriate prioritization may result in a strain on budgets and foster a further lack of appreciation of manufacturing issues and potentially less budget and resources for subsequent efforts.
Approaches to Reconciliation of Apparently Competing Objectives
The risk of attrition cannot be disregarded, nor can the realities of a “new normal,” in which the development cycle is increasingly compressed. However, proceeding solely on this basis, without an element of strategic risk management based on expectations, implications, and demands of clinical success, also works against the best use of available resources. A hybrid strategy is necessary. In this strategy, aspects of the API manufacturing process essential to the success in later stages of development are addressed and accommodated. This is done within a framework that critically evaluates all potential API-related studies for the expected magnitude of their contribution to the immediate milestone being pursued as well as to risk mitigation. Below are 3 suggestions for how to proceed:
Move Early Chemical Development Activities Upstream into Late Discovery “Front Load” the Development Effort
- Investigate chemistry that is common to a discovery scaffold from a PR&D point of view, once the potential candidates have been narrowed down to 3 or less fairly closely related molecules that have a similar synthetic route
- Expand on and capture the understanding of rudimentary physicochemical properties (solubility, crystallinity, handling properties, formation of simple salts, etc)
- If warranted by an impractical synthesis (extremely costly and/or difficult) that would be anticipated to strain allocated development budgets and projected cost of goods or hard to adapt for scale-up, perform cursory route scouting
Successful implementation of front loading results in a more rapid path to a reliable initial enabling API synthetic route, which allows focus to be on more difficult, unanticipated problems when chemical development is undertaken in earnest. This strategy has been previously suggested,5 and is currently implemented, to varying degrees. It tends to be used by organizations that have sufficient resources to accommodate it. That does not mean that smaller, virtual companies cannot find a way to implement aspects of front loading judiciously, using some suggestions provided in the sections below.
Recognize and Exploit Intervals During Development When API Manufacturing is Off the Critical Path
At points when the immediate supply needs have been met and preclinical, nonclinical, or clinical data are pending, undertake studies that are short in duration and relatively low cost, eg:
- Physicochemical characterization, salt screening, and polymorphism studies
- Storage stability of intermediates
- Improvement and optimization of problematic (safety, yield, equipment limitations, etc) process steps and operations
- Feasibility evaluation of alternate chemistry for particularly troublesome transformations or sequences
- Potential for implementation of, eg, flow chemistry for transformations that would be amenable to this approach
- Investigate the possibility of re-sequencing of process steps to improve practicality and streamline operations
- Perform feasibility/route scouting for catalytic and/or asymmetric transformations
These opportune “punctuation points” during development are becoming less frequent and are of shorter duration. Therefore, good advance planning, a keen eye on timelines as they evolve, and good communication skills and powers of persuasion are necessary to ensure that the above sorts of work have the best chance of being fit in during API’s temporary breaks from the critical path.
If it is Clear that Significant Remediation and Back-Filling is Inevitable (Due to Time and Resource Limitations) Perform Gap and Risk Analyses and Use Them to Drive Timing, Prioritization, and Selection of Development Studies
This activity should be done regardless of how much of the other 2 approaches are followed and implemented, but it is especially critical if the decision has been made to proceed to later stage manufacturing at high risk, without sufficient supporting development work.
- It provides a list of supporting studies which will ultimately be needed prior to NDA submission, regardless of when they are actually undertaken
- Prioritization of this list allows sequencing of the activities in a way that could allow a more gradual performance of the work
Identification and definition of work needed in support of mid- and late-stage development, particularly scope, cost, and duration, positions the sponsor to exploit off-the-critical-path time that arises unexpectedly, by inserting this work at these points. This opportunity is lost if the sponsor is unprepared. Continuity is important for work placed at vendors. Having the same project leader and team members on a given project placed at a vendor results in the most efficient accumulation of knowledge and experience. Work that is interrupted and set aside for prolonged periods results in a higher likelihood that familiarity with the project and the chemistry will be diminished due a lack of continuity, and this will have an adverse effect on the accumulation of knowledge and process understanding.
A combination of elements of all 3 approaches suggested above into an overall strategy is most effective. They are by no means mutually exclusive, and there is significant synergy to be realized from a plan in which the approaches are matrixed.
Summary
Many partially or completely virtual organizations engaging in drug development are limited in resources, time, and budget. Instead of taking a strategic approach, these circumstances often lead such organizations to take a reactive approach to the planning and execution of development activities, which can leave them in unenviably vulnerable and difficult situations. Ironically, this is especially the case if the clinical data from Phase I are encouraging and the pressure to move rapidly to the next stage of development/manufacture is considerable. This can lead to a more pronounced conflict between doing what is merely immediately necessary and being proactive enough to provide an adequate foundation for later stages of API manufacturing. Gap analyses and risk assessment and management are iterative processes. Given that most virtual organizations necessarily tolerate and accept higher risk than their better-resourced counterparts, it is clear that a corresponding increased assessment and management of risk is imperative. Risk management-driven decision-making and planning is required as an integral aspect of API development strategy at the outset of development, rather than as a last-minute add-on to appear to satisfy regulatory guidances related to QbD and process validation. Early use of these tools results in identification and resolution of any avoidable development and manufacturing issues. As a result, the ability to address predictable and identified risks throughout the development cycle reduces the likelihood of costly delays and failures.
Three complementary general approaches to address the conceptual uncoupling of expediency and QbD have been discussed:
- Front load API development by moving it up into late discovery
- Exploit intervals at which CMC and particularly API activities are off of the critical path
- Perform iterative gap analyses and risk assessments to assist with prioritization of activities deemed necessary, as well as tracking of progress in process understanding and control
Some combination of all 3 approaches should be able to accommodate the seemingly opposed needs of controlling costs and moving very rapidly vs. establishing an adequate database of process knowledge. The latter makes possible the establishment and maintenance of API manufacturing process control and quality throughout the development cycle, per the principles of QbD.
References
- Levy, S (Chair). Process Optimization & Scale Up: Challenges vs Human Ingenuity & Technology. PharmaChem Outsourcing. Long Branch, NJ, September 17, 2014.
- International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use. ICH Harmonized tripartite guideline: Pharmaceutical development Q8(R2). Current Step 4 Version. August 2009. Available at: http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q8_R1/Step4/Q8_R2_ Guideline.pdf. Accessed January 6, 2015.
- Snee, RD. Using Quality-by-Design to enable CMO manufacturing process development, control and improvement. Pharm Outsourcing. 2011;12(1). Available at: https://www.pharmoutsourcing.com/Featured-Articles/37652-Using-Quality-by-Design-to-Enable-CMO-Manufacturing-Process-Development-Control-and-Improvement/.
- U.S. Department of Health and Human Services; Food and Drug Administration; Center for Drug Evaluation and Research (CDER); Center for Veterinary Medicine (CVM); Office of Regulatory Affairs (ORA). Guidance for Industry PAT — A framework for innovative pharmaceutical development, manufacturing, and quality assurance. September 2004. Available at: http://www.fda.gov/downloads/Drugs/Guidances/ucm070305.pdf. Accessed January 6, 2015.
- Federsel, HJ. Chemical process research and development in the 21st century: challenges, strategies, and solutions from a pharmaceutical industry perspective. Acc Chem Res. 2009;42:671.
Stuart G. Levy, PhD, is Principal Consultant, SGL Chemistry Consulting, LLC. Dr. Levy provides expertise in synthetic organic chemistry, chemistry R&D, API manufacturing and other aspects of CMC development to emerging pharmaceutical companies. The expertise provided by SGL Chemistry Consulting is a result of employment at small, innovative biotech pharmaceutical companies (SUGEN, EPIX, Elixir) as well as at chemistry CROs/CDMOs (SERES Laboratories, Ricerca). Stuart is an expert in outsourcing, and has managed outsourced chemistry R&D and API manufacturing for 16 of his 19 years in the industry. Stuart obtained his PhD in Chemistry from the University of Illinois at Chicago, and did postdoctoral work in the School of Medicine at the University of California, San Diego.