By Alexander Kele, Senior Global Product Manager – Digital, Cytiva
The primary role of a process development (PD) scientist is to develop robust processes that can successfully transition from laboratory scale to a safe and reliable commercial manufacturing plan. Data management plays a critical role in doing this, as the decisions process developers make about critical quality attributes, process parameters, yield, purity, and other factors are based on information collected from a wide range of sources. Traditional ways for capturing, storing, querying, and managing data are time-consuming, prone to error, and can hinder collaboration. They also make it harder to find and access data already created.
With much of the drug development process dependent on collaboration and sharing of knowledge, relying on tools that create executional bottlenecks and lower the quality of insights can lead to unnecessary repetition of experiments and incomplete transfer between internal or external teams. Additionally, there are expectations from regulators that effective strategies for managing data integrity and traceability will be implemented during drug development. These internal and external drivers put a burden on pharmaceutical companies to rapidly address data management challenges. However, doing so calls for a deeper understanding of the shortcomings of today’s methods and the factors you need to consider when adopting a solution to successfully overcome them.
Data challenges impacting the successful exchange of information in PD
Data generated during the development of biopharmaceuticals serves as an essential tool for process understanding, process control, continuous improvement, and filings. For example, measuring and understanding the critical quality attributes of your molecules as well as identifying and quantifying potential sources of variation are key to rapidly developing and delivering a safe and effective drug to the market. This information comes from several data sources and departments, putting the onus of data management on PD scientists, the technical project leaders, and their colleagues in analytical development, who already juggle multiple responsibilities.
In addition, the instruments they rely on often come from numerous vendors, all using their own proprietary data formats and user interfaces, which also results in information stored in separate systems and locations. The traditional method used today to find data is to ask your peers and colleagues for it, relying on their timely availability and access to systems. To answer the scientific questions and perform essential PD tasks, data from these systems must be combined in one location for overview, insights, and sharing.
Outside of these distributed proprietary vendor systems, data is stored in flat files in uncontrolled storage spaces, such as network drives and folders. This offers little structure or insight into the data context and/or its interrelationship with other process or analytical data. A lack of context makes it difficult for someone unrelated to the actual experiment to understand the purpose of the data or to properly hand it over to the next team in the chain.
The manual combination of process and analytical data is often a “cut and paste” approach, with the combined data stored in an untraceable Microsoft Excel spreadsheet. Not only is this method time-consuming and cumbersome, but it also makes it difficult to utilize valuable historical data that could help a drug get to market faster. Should more data become available or a different avenue need to be researched, the manual process must start from the beginning, creating even more Excel spreadsheets and broken audit trails. These current ways of working can make it challenging to evaluate the data or repurpose it to answer a follow-up question.
The ability to share information between teams cannot be overstated. Yet, the distributed storage of data, the uneven non-democratic access to it, and the issue that many of the tools today are not crafted for collaboration makes this a difficult task. Working in silos can lead to lower efficiency and an inability to capture important insights. Ensuring a constant flow of information among multidisciplinary teams can bridge the gaps and allow you to reuse collected data for future improvements and innovations. It also helps in the digital transformation of the organization, as information can be shared more easily across the different decision-making levels.
Finally, there are many different types of reports that serve as essential tools for documenting, summarizing, and relaying conclusions. A report may involve several scientists from different domains who collaborate to create, lay out, and review the validity of the conclusions and the data used. The constantly recurring process of creating these reports takes substantial time due to distributed data, manual cut-and-paste methods, and the use of general-purpose tools (i.e., Microsoft Excel, PowerPoint, and Word). Cytiva’s customers have estimated that 10% to 20% of a PD scientist’s weekly time is consumed with finding data, and it takes even more time to create presentation-quality reports. This is time that could be used for more value-creating tasks.
How to pick the best data management solution for your laboratory
When choosing the vendor for a data management solution, it is important to consider not only the features and functionalities but also a range of factors, such as their level of expertise, capability to provide technical support, and the lifetime cost of the solution. There are a few top-level options to consider depending on your company’s capabilities, workflows, and budget.
An internal solution can be constructed, but it often leads to customized solutions that are hard to maintain and costly to support. This approach can also consume internal resources that are not core to the company’s business. Another option may be to go with a broad enterprise solution. However, these solutions often require a major upfront investment of time and money and can take months or even years to roll out completely before the promised return on investment (ROI) is achieved. If such an enterprise solution is not modular enough to achieve a seamless integration and full ROI, it may require the replacement of other infrastructure components that were initially not meant to be replaced, generating even more cost and complexity. In addition, we have seen in our industry research that many end users of enterprise solutions feel disconnected from the vendor that supplies the software, hampering both rapid support and continuous improvements according to the changing needs of the company.
Often, the most flexible and lower risk option is to go with a vendor solution that is modular and easily integrates with your existing infrastructure. It should be standardized for lower and more predictable costs, yet flexible enough to not add unnecessary costs related to changes in workflows. The deployment time should be minimal, so usage can start as soon as possible and ROI can be quickly reached. A good modular solution should enable you to mature into the solution over time, while also allowing users to continuously create value from day one. The focus can then shift over to the connectivity of devices and data sources, as needed and budgeted for, at a pace decided by the company itself.
Change can be intimidating, especially in the pharmaceutical industry, with millions of dollars and, more importantly, patients’ lives at risk. For quick and persistent adoption across the organization, even if it solves the problems stated earlier, the solution must be user centric as deemed by the users themselves, from the data savvy to the novice. Any trial and testing leading to a purchase should clearly show this.
If you are interested in a flexible solution that gives access to combined process and analytical data in one place, from a vendor with deep understanding of the biopharma domain, visit Cytiva's PDCentral page or send an email to [email protected].
- FDA. (December 2018). Data Integrity and Compliance With Drug cGMP, Questions and Answers Guidance for Industry. https://www.fda.gov/media/119267/download