Pharmaceuticals are manufactured by one of the most regulated of all industries. And its not as simple as the vial has to match what is on the label. Every aspect of development, testing, marketing approval, promotion, manufacturing, packaging, warehousing, shipping, and post-market surveillance is subject to regulatory scrutiny.
Among the most closely examined business processes are those by which data is captured, analyzed, and reported. Drug companies often adopt a methodology called Continuous Process Verification (CPV) for the commercialization phase. All processes used in manufacturing the drug generate data, and how the data is created, captured, analyzed, shared, and reported must be validated against the Food and Drug Administration’s guidelines, which mandates that those steps must be replicated exactly from each batch to the next. If you change a step in the process, it must be validated all over again. CPV is a method for continuously assessing the business processes to keep them within the parameters documented when those processes were validated. The FDA has documented its requirements in detail.
To benefit from CPV, a pharmaceutical company must first maintain a state of control in the manufacturing process. This is typically achieved through designing an experiment and making sure the process is under control. This is done by analyzing the variability in the process and making sure that variability is associated with the drug product and not in the process or the systems involved in its development. This involves setting appropriates specifications for the process. After the developer reaches the state of having the process under control, that’s where CPV kicks in. The challenge is to make sure that the process stays in control, and that the developer can guarantee that the drug product being manufactured is exactly the product specified.
Especially within Biologics or other compounds that require complex, drawn-out processes, monitoring adherence to the process is very complicated. With Biologics in particular, the developer must be sure that in cell banking the cultures are exactly as specified. When the developer is at the stage of growing the media in lots where the protein will be generated and harvested, that process must be verified at each step and at each point in time, to ensure that the developers are adhering to those tight specifications. When the time comes to begin harvesting the product, process verification ensures that the product is up to the developer’s quality standard. Still another verification takes place when the product is packaged, making sure the materials used to package the drug will not leach contaminants into it.
Hundreds, sometimes thousands, of variables must be monitored to verify that they remain within the specifications established for this process. That’s really what CPV is—making sure to monitor from the very beginning of the process to the very end, that every step is tightly controlled and the product is exactly as described in the package insert.
A differentiating focus for TIBCO Statistica in CPV is the ability to access data in most any format, on a wide variety of platforms. For example, in Biologics production, there are numerous sensors to be monitored to make sure that the temperature, pH and other environmental factors for cell culture are correct and remain stable. The data is likely to come from several different instruments. And that data must be aligned with data from laboratory information management systems (LIMS), a quality management system like TrackWise (Sparta Systems), and other data sources. The TIBCO Statistica platform provides the means to set up the data connections once and maintain them throughout the life of the installation.
This is not a novel concept, but this type of connectivity has been difficult to achieve between disparate systems, especially behind a validated environment. Data typically is captured by people on the shop floor, and must be validated when it is committed to a database. Anytime data is pulled from that database, analyzed separately, and then the results are stored back to the database, there is yet another validation required. Statistica provides a single platform for analysis and validation of data from multiple sources. The user can enter data using a web browser, directly into a validated database, eliminating double entry and the need to validate each time data is captured. The reporting is driven from the same platform as well—one system, one code base.
The functionality to support CPV was designed into the platform based on the guidance from the Food and Drug Administration and the International Council for Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH). It is fully functional for pharmaceutical company adoption out-of-the-box without the need to re-validate the individual algorithms that are included. Statistica has over 16,000 functions, and adopters can buy validation packages for the functions they need. This option makes it easier to validate the system on the customer site. Validation is relevant not only to pharmaceuticals but to food and beverage, finance and insurance as well. Many of the analytic models generated in those industries must be validated by regulators.
This is feasible with Statistica because of the open architecture of the platform. Models can be generated in many formats, including the industry-standard predictive modeling markup language PMML, and are fully documented. There will be great platform synergies as the Statistica platform is integrated into the TIBCO infrastructure; new capabilities are expected to be announced once the merger of Statistica into TIBCO is completed. For example, the ability to leverage TIBCO’s visualization tool will enable us to display the results of our analysis in a way that is easier for end-users to digest.
For pharmaceutical companies, Statistica is in a position to offer something they have never really had before: A replicable, templatized system that can be implemented and re-implemented as new products enter development, so that common processes can be set up without requiring a new software license and with minimal service. Interested in learning more? Statistica will be at BioTech Week Boston from September 25-28. Meet the team and find out more at booth 1029.