01 January 2004
New technology can aid in meeting FDA rules, add to bottom line.
By Justin O. Neway
In today's pharmaceutical and biotechnology manufacturing environments, compliance has taken on new meaning. It once implied a system of warnings that required attention. Today, the U.S. Food and Drug Administration (FDA) is demanding a new focus on compliance. Recent headlines reveal continuing industry problems and new efforts by the FDA to reduce them. And when you look beneath the regulation jargon, there are new opportunities to improve manufacturing efficiencies as well as compliance in ways that benefit the bottom line rather than cut into it.
Since the FDA first defined regulations for good manufacturing practices (GMPs) in the late 1970s, several changes have occurred. The agency's role in healthcare has grown as more drugs have hit the market, dramatically increasing the agency's resource constraints. Advances in pharmaceutical sciences and manufacturing technologies require advanced knowledge and scrutiny of processes to ensure quality, so the need to train inspectors has grown exponentially.
As the industry has become more and more global, pressure has mounted for the FDA to monitor drug imports and work more closely with the European Union. The reliance on foreign imports is increasing, posing potential threats to product quality that the FDA must address.
Although U.S. drug product quality is high overall, trends show manufacturing problems, resulting in drug recalls, disruptions in operations, and drug shortages that affect revenues and consumer confidence. In 2001, 300 prescription drugs underwent recall. Compliance creates huge challenges for manufacturing operations, and often users view this as a hindrance to efficiency rather than as a way of improving it.
As a result of its resource constraints and the new industry environment, the FDA has introduced two important new initiatives. The first is an effort to guide the industry in applying process analytical technologies (PATs) to manufacturing, and the second is a new quality initiative focused on the GMP regulation.
Interestingly, when we look at trends in FDA enforcement since its systems-based inspection approach began in February 2002, 59% of the citations issued were as a result of quality systems, production systems, and lab systems. Fully 70% of the observations in subsequent warning letters covered the same categories. The implication is that manufacturing is under FDA scrutiny, because the problems too often are not adequately addressed, especially in the areas of quality, production, and lab systems. The data demonstrates an urgent need for the industry's focus to turn to manufacturing.
In 2001, the FDA formed a subcommittee to design a philosophical and regulatory framework to guide the industry in applying PAT to measure and use quality indicators in real time during the manufacturing process. PAT allows online, nondestructive measurement, even potentially for individual tablets.
Multivariate statistical treatment of manufacturing data helps one better understand control manufacturing processes. Thus, PAT has the potential to get more information into the hands of operating staff and decision makers more quickly, empowering them to improve quality compliance and global operational efficiency in a time frame relevant to making the adjustments that can improve outcomes.
As a result of PAT, you can expect greater control of product uniformity. This will result in FDA inspection resource savings and some regulatory relief for manufacturing changes. Another potential benefit is shorter cycle and batch release times thanks to less dependence on wet chemistry techniques, a capability to conduct multivariate specification setting, and the cost savings associated with supply chain predictability, provided that the right supporting software technology is in place. PAT also has the potential to enable parametric product release, which can save costs and reduce inventory across the industry, as well improve process control and make continuous operations more practical.
The PAT framework fits into an overall quality initiative designed to help better manage the FDA's limited inspection resources. The initiative goes beyond current good manufacturing practices (cGMPs) to regulate product quality through a more scientific and risk-based approach.
The FDA, with participation from the industry, is evaluating systems for standards setting, submission, and review and inspections to determine objectives and effectiveness in achieving goals. Part of this evaluation involves assessing criteria used to prioritize FDA responsibilities, so the most resources apply to those areas with the greatest risk to public health. Combining a science-based risk management approach with quality systems inspection represents a new way of working for the FDA.
This approach will force the industry to more tightly link product development to manufacturing. For example, capturing batch records early in the discovery process can provide data to strengthen control in manufacturing. Both the FDA and manufacturers will benefit from fewer post-approval surprises.
Another result is likely to be more intensive review of injectables manufacturers, with an emphasis on biotech, because this class of medications represents a higher risk to the public than oral dosage forms. The implications of the FDA's focus on critical process control points as a greater risk area creates a need to better understand parameter interactions. And science-based review and enforcement necessitates that manufacturers be better able to show data and support conclusions on demand—not a simple challenge to address.
Pharmaceutical and biotech companies are sizing up the FDA's new approach and heeding warnings from competitors' missteps. After a $500 million fine leveled an industry powerhouse in 2002, other companies are recognizing that return on investment (ROI) in research and development and sales and marketing is only realized when you can give a very high assurance to the FDA that a drug will remain effective and safe during commercial operations. The new systems approach to inspections implies there will be less explicit guidance from multiple 483 citations, because the FDA will most likely only inspect one manufacturing site and it will be up to the manufacturer from there. If the same problem occurs at a different site, the FDA will treat this as cause for a warning letter even though the FDA may not have cited the other plants in the system with a 483 for the same deficiency. So a response in one plant should carry out across the entire manufacturing organization, with no exceptions for other locations.
As if the regulatory environment weren't difficult enough, pricing pressures impose a bottom-line mentality on dealing with compliance. As U.S. demographics shift with the aging baby boomer population, drug prices are rising (e.g., a 17% increase occurred between 1999 and 2000).
Reimbursement pressure on consumers from insurance companies favors generic drugs, and pressure from Congress favors competition.
As these external forces affect product margins, companies also must cope with shrinking drug pipelines. They are expected to get more profit out of fewer winning candidates. One solution that appeared to have potential was the promise of mergers and acquisitions to gain increased efficiency from operations. An enormous pressure exists to create centers of excellence, so that fewer manufacturing facilities are producing equal or greater revenue. This creates a burden on manufacturers in addition to that already imposed by the regulatory environment.
Technology offers answers and challenges to the tougher environment global manufacturers now face.
The first major challenge as pharmaceutical and biotech manufacturers strive to comply with FDA regulations is the belief that a company cannot achieve operational efficiency without compromising consistent quality compliance. One must not come at the cost of the other, and today's technology can help ensure that this is not the case.
Closing the gap between compliance and operational efficiency first requires understanding that although technology plays an important role, other commitments are also required. Imperatives that must exist within an organization include senior management's philosophy, commitment, and follow-through; staff-level commitment and follow-through; training and documentation; correct materials and workflows; and, finally, compliant equipment and utility systems.
A familiar example of the wrong mentality within a corporate culture was brought to life last year, when that drug manufacturer was hit with a $500 million fine by the FDA. According to a now infamous whistleblower, "They indicated that there has been in the past a continual push for increased production and decreased downtime, sometimes at the expense of quality work and [FDA] compliance."
This view of a trade-off creates the gap, but technology, in part, can close that gap. Users can leverage software to close the part of the gap that includes document generation, tracking and control, "infrastructure" data gathering systems, 21 CFR Part 11 implementation and compliance, and, most importantly, data-intensive decision making.
Data-intensive decision making is one of the most challenging aspects for the industry today. Data is everywhere, but very little of it is useful information. Most plants generate mountains of data, but without the right software capabilities to make the information content quickly available to the right people in a time frame that is relevant to the manufacturing process, the data remains essentially untapped. How can manufacturers possibly handle the mountains of data that relate to both focus areas—quality with regulatory compliance (GMP) and operational/process stability?
On the quality side, easy access to the data in a useful form is necessary for activities such as parameter review for batch release, setting defensible specifications, investigation of out-of-specification (OOS) batches, manufacturing process validation, production trend analysis, and annual product reviews (APRs).
Easy access to the data also can help shorten process start-up, scale-up, and troubleshooting and adverse trend reversal times; improve productivity, quality, and return on net assets; and finally, leverage existing technology investments, such as enterprise resource planning (ERP) systems, laboratory information management (LIMS) databases, data historians, and batch records.
In today's typical plant, there are numerous "inefficiencies" hindering the use of data for fast, effective decision making. It takes several weeks, and sometimes months, to manually retrieve and align data of multiple types from multiple sources where it is stored. In most cases, manufacturing data is not a top priority for corporate information technology. Once gathering the data, a user usually has to endure combinations of "Excel add-ins"—discrete, continuous, and replicate data create spreadsheet madness. While many software programs exist that potentially could help make this data more useful, the choices are bewildering and usually inadequate for the exact tasks at hand. Programming is most often required to use these packages, and the users—those who need data to make informed decisions—are not likely to be professional programmers or statisticians. After accomplishing the programming, a user usually communicates the results in old-fashioned, ineffective ways. Tables of highlighted numbers and traditional two-dimensional plots seem to be the limits of most programs' sophistication.
These inefficiencies cost companies millions in missed opportunities, representing the cost of not leveraging data-intensive decision making for the organization.
Decision-making inefficiencies also leave companies more vulnerable to compliance issues. GMP-related FDA warnings include failure to investigate the root causes of OOS results, failure to adequately validate the process, failure to properly justify specifications, and inadequate/incomplete APRs. A specific citation being made more frequently in recent times specifically cites "failure to establish and follow control procedures to monitor the output and to validate the performance of those manufacturing processes that may be responsible for causing variability in the characteristics of in-process material and the drug product."
So, whether a company chooses the side of the gap that favors operational efficiency or the side that strives to maximize compliance, data-intensive decision making is a must for success. And, the best news is, with today's available technology, you do not have to make a choice that sacrifices one or the other.
More than 150 technology vendors claim they can help with Part 11 compliance alone. So how can manufacturers effectively evaluate and implement solutions that will truly enhance compliance while improving the manufacturing process?
It is important to review the various types of data in the real manufacturing environment and the disparate systems that gather it. Discrete data is measured once per batch. Continuous data includes strip charts and other time series profiles stored in supervisory control and data acquisition historians, distributed control systems, and programmable logic controller systems. Replicate data includes several measurements from the same sample and/or time. Finally, paper records include operator, equipment, and quality records.
Data collection in the manufacturing process begins during the early steps—from bringing in raw materials to lab tests and utility systems. Data collects in different departments and stores in database systems ranging from ERP and LIMS to electronic and paper batch and other record systems.
In fact, AMR Research estimates that paper records can often contain as much as 80% of pharmaceutical and biotech manufacturing data. As material moves through the manufacturing process, these disparate data sources fill with more data from mixing, granulating, and milling equipment; tablet presses and capsule fillers; and final steps such as coating and packaging.
With all of this data captured either in electronic or paper systems, the challenge lies in the practical requirements of getting to the data and extracting its information content so a user can make informed decisions. The concept of relevant time is of utmost importance. To be useful, information must be available in time to affect at least the next batch—whether that is in seconds, minutes, hours, or weeks. In most plants today, the wait time for getting data to work with is typically months.
In a typical pharmaceutical or biotech manufacturing organization, a user must address simple and complex decision-making problems. Data must be easily available for both types of problems and must be readily useable by professionals who are neither programmers nor statisticians.
Different parts of the process may take place in different parts of the world, so one needs a global view of dispersed data from a single point of access, and the use of the system must become a widespread, cross-functional habit across the global organization.
If data is to be useful it must provide the right view of the entire manufacturing process in front of the right people at the right time. Fortunately, a new bridging technology exists to convert the data on paper records into electronic form efficiently and in full compliance with 21 CFR Part 11. This software allows retrieval of data from paper records and captures it in useful form without having to use a full-blown electronic batch record system that requires a much larger implementation effort because of the additional workflow requirements. Manufactur-ing professionals already have to use data from paper records today for trending, investigations, and batch release, but they are doing so using antiquated systems that require typing the data laboriously into unvalidated spreadsheets, where it tends to become lost or discarded after a short time. This creates a risk of not being compliant, making fragmented decisions, and losing the investment in previous data entry when the next data-intensive problem arises. It also requires more senior-level skills to take care of these problems. By converting the data on paper records into electronic form using an enterprise manufacturing intelligence system that includes on-screen replicas of the batch records, admin-level skills can get the data into an easily accessible form that models the process and is available companywide going forward. Moreover, double blind data can occur routinely, and the data is automatically checked for accuracy, a step that ad hoc spreadsheet approaches most often omit.
So how is this data available to any user who needs access? It can happen with a point-and-click view of the entire manufacturing process, parameter-by-parameter in an expandable hierarchy. This allows users anywhere in the world to get value from the data by selecting any view of the process based on a specific organizing principle, such as the batch, operator, equipment, or materials.
Regardless of where you store the data and whether it is discrete, continuous, replicate, or paper based, immediate access to the data from a personal terminal is available without having to type in command lines, write queries, or ask someone else to retrieve the data.
Once there is immediate data availability in a single point of immediate access, the choices can be unlimited, but they must be structured in the way that is most useful for pharmaceutical and biotech decision making. A user can see all parameters for each batch in a process-centric view filtered by selecting only the parameters needed for a particular decision or investigation.
Once data has been assembled and filtered, an analysis group is created for doing work ranging from fast batch release and root cause investigation to production trending—with customized alerts set to match individual process requirements. Specification limits can be set wide enough to ensure higher batch success based on acceptable failure rates, but within a range that will be acceptable for regulatory authorities. Before now, a manufacturer used the "rule of thumb," because the available tools were not useable enough. Today's technology puts more control in manufacturers' hands, enabling them to quickly predict outcomes using available data.
Continuous data can often be "dirty"—misrepresenting what is actually happening in a fermentation process, for example. Filtering out this "noise" and preserving fine detail is possible using new technology that incorporates a properly designed data-conditioning interface. Feature extraction allows manufacturers to select a region of continuous data they want to understand in more detail, simply by clicking on the region of interest directly in the screen. By selecting the correct values for plotting and excluding irrelevant data, manufacturers can make more meaningful interpretations more quickly.
The business case
In today's pharmaceutical and biotech corporations, gaining approval for new software technology can be a tremendous challenge. Unfulfilled expectations from past enterprise implementations haunt executives. Execution nightmares due to separate departmental systems and skilled resource limitations create a fear-of-change mentality. The most hindering factor is disbelief that new software technology can at last enable what management really wants—a complete environment for data-intensive decision making that achieves ongoing manufacturing compliance and lasting operational efficiency improvement.
Decision makers and their influencers evaluate new technology based on how well it achieves the following key manufacturing metrics: (1) sustained improvements in operational efficiency, (2) predictable yield and quality, (3) shortest time to revenue, and (4) lowest FDA risk. Technology that cannot deliver these metrics will not get past the initial demo, so it is important to identify systems that truly deliver these benefits and emphasize them in business cases that also meet the following criteria:
- Application to existing priorities
- Leveraging existing investment
- Clear project definition and scope
- Near-term return to the bottom line
- Significant positive impact on corporate goals
Simulated case study
The challenge: A major pharmaceutical manufacturer wanted to reduce lot failures and increase process predictability to improve profitability. The manufacturer needed to identify the combination of critical drivers across the process that was determining the outcome, namely the tablet dissolution rate.
The solution: Using a properly designed manufacturing software system with immediate enterprise data access, and user-centric investigational and visualization capabilities, this manufacturer was quickly able to work with data from more than 60 product batches from one plant.
Using principal component regression (PCR), the manufacturer quickly found the smallest combination of controllable process parameters that had the greatest effect on the process outcomes. There were only four such parameters, and they were in widely disparate parts of the process. The manufacturer was able to implement specific recommendations for process improvements based on the best combination of ranges of these parameters predicted to give the best outcomes.
For all the data used in this study, the process parameters had operated within their approved ranges. Therefore, the manufacturer was able to test these findings directly within the manufacturing process without the need for additional small-scale experimentation—a major cost advantage. In other words, the process improvement recommendations came from the variations that occurred in the manufacturing process when operated under approved conditions. When implemented, the recommended adjustments allowed the manufacturing staff to bias its process toward the best possible outcomes without the need to change the manufacturing technology or get it reapproved by the FDA—a tremendous savings in time and money.
The bottom-line impact of this implementation included a significant reduction in atypicals by stabilizing tablet dissolution rates, lowering failure rates to increase yield by 25%—ultimately improving the bottom line by more than $5 million per year per plant.
Although the pharmaceutical and biotech industry faces challenges to meet the demands of today's marketplace, as well as existing and forthcoming FDA requirements, new and existing technology investments can take a new approach—one that must be accepted by global organizations as the new reality. MP
Behind the byline
Justin O. Neway, Ph.D., is executive vice president and chief science officer at Aegis Analytical Corp. in Lafayette, Colo. Neway presented this paper at the World Batch Forum North American Conference in April 2003. His e-mail is firstname.lastname@example.org.
Return to Previous Page