3 Assay Development Rules for Tech Transfer
At Kymanox, we get to see a lot of processes that are being developed for commercial approval. One of the biggest problems we come across on our projects is related to analytical assays. On fast-tracked tech transfer and scale-up projects, assay development – including qualification, transfer, and validation – are almost guaranteed to be rate-limiting for the overall project schedule.
Assays come in a lot of different flavors. Here is a short example list of some assays that we routinely deal with on our projects:
- Purity by HPLC
- ID by IR-Spectra
- ID by ELISA
- Trace Heavy Metals Analysis by Titration
- Concentration by UV
- Cleaning Verification by TOC
- (Many others!)
We often see assays that are modified from their published USP compendial methods. Compendial assays are self-validating since they use internal controls and follow validated methods. These assays only need to be minimally qualified to support GMP manufacturing. So this leads to the first rule on assay development for successful technology transfer.
RULE # 1: Follow Compendial Methods
The FDA is starting to crack down on companies who claim they are following compendial assay methods, but are actually modifying the methods to suit their needs – or scientific urges – and are not validating the modifications. For example, a modification to “USP pH” requiring a mandatory waiting or settling period outside the method parameters should be separately validated since this is a modification to the USP compendial method. In the Medical Device world, the FDA has even created a new form to ensure companies who reference compendial methods are following the methods exactly. This form is called Form 3654 and requires any method modifications to be clearly stated and justified.
Anyone doing R&D or Process Development should be trained to select and follow USP compendial assays whenever possible. This eliminates the need for stand-alone method validation and ensures assays can be outsourced and transferred with minimal time and cost.
The next problem we see related to assays is the eagerness to accept methods that only produce qualitative results (i.e. PASS/FAIL or attribute data). Ten years ago, this was not an issue; but in the 21st Century, it is imperative to utilize assays that yield quantitative (i.e. numerical) results. The main reasons behind this are Quality By Design (QbD) and ICH Q10. These universal quality principals and systems are rooted deeply in process characterization and understanding; quantitative analysis is simply more adept at characterization and understanding than qualitative analysis. Furthermore, executing QbD principles requires statistical-based science and risk management. At the end of the day, qualitative (i.e. PASS/FAIL) results provide a lot less information in comparison to their quantitative (i.e. continuous and numerical) counterparts. And once you start doing risk assessments and have to provide statistical rationale supporting those assessments, you’ll quickly come to the same conclusion that assays which produce attribute (i.e. PASS/FAIL) data are a big handicap to fully implementing QbD.
As an example, we had a process validation exercise where we only had attribute data supporting critical quality attribute. All we knew was that a contaminant level was below a specified threshold. Because we did not have continuous numerical data to statically and mathematically analyze, we required several more samples within each lot (i.e. batch) and more lots (i.e. batches) to sample. In the end, the lack of a quantitative assay cost the project hundreds of thousands of dollars in additional process validation runs and samples. In most cases where there is a need to demonstrate quality assurance at a specific confidence level, quantitative data can lower the number of samples required by ten-fold (i.e. 10x) as compared to qualitative (i.e. PASS/FAIL) data.
For processes in early development, we strongly recommend minimizing product quality and safety assays that yield only attribute (i.e. PASS/FAIL) data. Once the process is fully developed and the product is in late-stage development, there is usually very little time or resources available to make this assessment and improvement. This leads to the second rule.
RULE # 2: Select Assays That Produce Numerical Results
When developing and selecting assays for product quality and safety, as well as for in-process controls, avoid assays that yield only attribute (i.e. PASS/FAIL) data. Rather, select assays and methods that yield continuous numerical results so those results can be leveraged for statistical analysis, which in turn can support the overall QbD quality approach.
Let’s skip to the third rule and provide more detailed rationale in a future post:
RULE # 3: Minimize Assay Variability
When developing assays, do everything possible to eliminate variability in the method.
By minimizing variation, the assay becomes a much more powerful tool for making process improvements and detecting process changes. A lot of people are focused on assay accuracy; however, assay variability plays a much more important role in the realm of quality control and QbD. If you want more on RULE # 3, I suggest the following text book:
Six Sigma In The Pharmaceutical Industry: Understanding, Reducing, And Controlling Variation In Pharmaceuticals And Biologics (Paperback, June 2007) by Brian K. Nunnally (Author) and John S. Mcconnell (Author)
Abbreviations used in this post:
- GMP = FDA’s Good Manufacturing Practices
- USP = United States Pharmacopeia
- ID = Identification
- ELISA = Enzyme-Linked Immunosorbent Assay
- TOC = Total Organic Carbon
- UV = Ultra-Violet
- R&D = Research and Development
- FDA = Food and Drug Administration