The Digital Review
It’s time to address the bottlenecks that lie in quality assurance review and release – and digital systems can help us
As the COVID-19 pandemic grips the world with fear and uncertainty, scientists are working long hours to develop novel testing, antiviral medicines, and vaccine technologies. And every day, incremental progress is being reported. However, there are some fundamental challenges being faced by pharmaceutical manufacturers; bottlenecks that have affected development too long. In my view, it is time to address these challenges with tools that are already at our disposal.
Consider clinical trial material quality assurance. To ensure that clinical trial material is of sufficient quality, traditional release review and approval operations include the preparation of quality summaries. These summaries exist as human-readable documents (which is to say, document files presented to scientists on their computer screens and visually interpreted). Consequently, such summaries must be thoroughly examined and interpreted by expert quality assurance staff. The most fundamental decision quality assurance staff must make is, “Does this material conform to the quality specifications established for its intended use?” A comparative assessment is made, and quality test results are compared with acceptance criteria across all critical quality parameters. Moreover, QA staff must also confirm that the approved processes for generating materials were adhered to during manufacturing and downstream testing. The consequence of either a misinterpretation of a quality summary or an overlooked discrepancy is significant; releasing product batches that do not conform to quality specifications can have a significant impact on patient safety.
Digital systems (for example, LIMS, ELN, and LES) are used to capture all of the pertinent data used to prepare human readable reports to support the QA review and approval release step – and I believe these digital systems can be leveraged to provide better QA support overall. The documents used in materials release – upon generation, submission, and QA – are effectively decoupled from the systems used to generate them (electronic signature and document-traceability controls notwithstanding). Should these documents raise questions that require supplemental information to be prepared, additional human documentation efforts are conducted. The “forensic” investigations and further document preparation can create significant delays in batch release.
We can make the QA review and release step more efficient by reducing reliance on human-prepared document-driven decision making. First, by structuring specification and test data into machine-readable formats, software could then help augment QA decision making and approval. An example case is to prepare standardized, formatted chromatographic peak information from impurity profile tests. These machine-readable datasets can be presented to various business intelligence applications, which could programmatically compare peak information (for example, area percent values), and confirm whether these results conform to quality specifications – thus augmenting QA staff review oversight.
In addition, by storing and managing structured datasets (impurity profile chromatographic data, identity-confirming spectral data, material quality-confirming image data, and so on) in data management systems, the effort to prepare “portable” reports (document files that do not require specialization software applications) is greatly reduced; QA staff can simply perform queries themselves within specialized decision support applications in lieu of requiring colleagues to prepare and submit reports. The software also reduces the effort when the need to conduct forensic examinations arises. For example, when specific batch data does not conform to specification, any related batch information (for example, batch information for precursor materials) can be accessed – thus allowing for root cause assessment of non-conformance.
In addition to internal QA review innovation, machine-readable structured datasets could also accelerate external party review-and-approval steps – such as those to healthcare authorities. Machine-readable quality summaries could be submitted and reviewed by the recipients’ own decision support software applications, greatly reducing the amount of effort required by external staff. The result is a new review-and-approval paradigm where internal and external review steps can be executed within a “human review by exception” model; in other words, business intelligence software can perform automated assessments of submissions, leaving more challenging assessments to review staff. In this new world, “availability notifications” to third parties could serve as an initial contact event – as opposed to a document submission.
Though scientists on the frontline are working to develop new treatments and vaccines against COVID-19, we need to remember that all of us in the pharma industry still have a role to play. Purveyors of digital transformation and innovators within quality assurance can support frontline colleagues by considering structural changes to how QA review and release approval is undertaken. Ultimately, we can make the entire process more efficient to bring new therapies to patients faster – whether for COVID-19 or any other disease area.