By Andrew Anderson, Vice President, Innovation and Informatics Strategy
In pharma, drug substances (and the resulting formulated drug products) must conform to a variety of quality specifications in order to be approved for use by healthcare practitioners and patients. While most of us who have worked in pharma know the various regulatory statutes and advisory guidance (and can quote them chapter and verse!), my belief is that there is a challenge in the practical and efficient implementation of quality practices that support conformance. When considering the increasing ‘fracturing’ of supply chains that support demand for drug substances in both clinical and healthcare systems worldwide, this challenge only continues to grow.
As many of this blog’s readers are aware, one of ACD/Labs’ core capabilities is to manage the digital representation of analytical experiments. These experiments support process development and substance characterization efforts, among many other things. For many years, our customers across industries have not only used our tools to support efforts to make decisions on the identity, composition, and overall quality of substances, but also the comprehensive management of impurity data.
As analytical impurity control strategies are critical when producing drug substances and other formulated products, it was interesting to read about a recent drug recall for the common heart drug, Valsartan, as the reason behind the recall was exposure to a potentially genotoxic impurity, N-nitrosodimethylamine. In particular, I questioned whether the control methods employed when managing the specific impurity in question would detect its presence in any drug substance lots (within relevant ‘loss on drying’ and possibly ‘limit of quantification’). Sure, UV methods at the appropriate wavelength can detect the presence of the molecule, assuming of course that the appropriate quality by design (QbD) approach was taken to account for the potential presence of this impurity. But I’m also mindful that this could be a “Black Swan” event, where this particular molecule’s presence was not expected. While I know there is no such method, my mind then wanders into the realm of universal, infinitesimally sensitive detection methods. So are we, the purveyors of QbD and innovative analytical chemistry, relegated to accepting that in some cases, impurities will appear in substances with no chance of catching them?
While I’m unaware if the impurity was substance-related or supply chain process-related in the manufacturing of this particular drug substance nor do I know anything about the specific quality practice employed for this particular supply chain, I can say that it is difficult to share rich analytical data between contract manufacturing organizations (CMOs) and sponsor firms.
I remember my personal experience of having to review quality summary reports that were carefully and dutifully constructed by my external partners. While the human effort that went into such reports was extensive, the reports were also static, point-in-time summaries of the detailed analyses that were performed on substances. As an analytical chemist, my inquisitive nature always created a desire to look deeper into the pictorial representation of the data that a well-qualified chemist from my CMO partner prepared. Candidly, this desire would usually result in a request to my CMO partner for the digital file that I could open in our own ACD/Spectrus Processor!
Regarding this particular recall, my optimistic sense is that we can at least mitigate such risk by continuing to advocate for increasing the extent of digital data exchange. Here are some questions worth considering if this is a topic of interest for you:
- Can CMOs supply sponsor companies with the analytical data (that form the basis of certificate of analysis based quality summaries for each lot of materials?
- Can software and system providers facilitate data exchange between CMO and sponsor companies that reduce the manual preparation and submission of such quality summaries?
- Can sponsor companies implement decision support systems that incorporate live analytical data that allow decision makers (e.g., QA staff, project leaders, etc.) to look at both numerical quality metrics and the actual chromatograms and spectra acquired to form such metrics?
- Can we store, organize, and allow for rapid visual comparisons of analytical data on a lot-to-lot, process-to-process, supplier-to-supplier, and stage-to-stage basis?
As you might suspect, ACD/Labs is working with a variety of stakeholders in these areas. We have room to work with others, so if this is of interest to you, I’m all ears!