About
Welcome to Potencyassay.com
I’ve been working in the analytical development field at a major pharmaceutical company for the last ten years or so. Most of that time, I’ve been involved with development of various kinds of potency assays. I have looked around the internet several times for a resource where people in the bioassay and potency assay community can share ideas and best practices. I haven’t found such a place yet. So instead, I thought I would start this blog and share my experiences and ideas about these often difficult assays.
The topics I plan on covering in this space include:
- Bioassays
- Immunoassays
- In vivo potency models
- Assay development practices
- Assay optimization
- Design of Experiments
- Data analysis and statistics for potency assays
- Laboratory Automation
- Six sigma and other process optimization methodologies as they relate to potency assays
I would also like to invite all of you to comment on my posts or write a guest post if you’re an expert on a related topic. Just leave a comment below and I will get in touch with you.
I also want to point out that the opinions that are posted on this blog are strictly my own and are in no way endorsed by or affiliated with my employer.
Thanks for visiting and I hope you come back soon.
Dan
2 Comments for About
Philip De Decker | July 9, 2010 at 8:50 am
Author comment by Dan | July 20, 2010 at 3:57 pm
Hi Philip,
Sorry for the slight delay in a response.
This is a tough question, but I would go about it this way:
First of all, the samples that failed parallelism should not have potency values associated with them. As soon as you have a lack of parallelism, a potency value has no meaning.
Then I would require at least three replicates of the assay that pass the assay acceptance criterion. The mean of those replicates need to be within the spec for the lot to pass.
Since this lot is on the border of failing the specification, you may want to consider a 95% confidence interval around the mean value and require this interval to be within specification instead of the point value. That way you can be more certain if the lot passes or fails.
I would also closely examine the response curves for the failed test to determine why the samples are failing parallelism. A failed parallelism test if often the first indicator that you have degredates in your product.
Does anyone else have an opinion?
Thanks!
Dan
Hi Dan,
We have an issue regarding potency analysis of stability samples which might be discussed on the blog. The issue deals with failing of sample suitability criteria (e.g. parallelism) of a test sample and the related interpretation and reporting of such observations.
During stability studies, it can be expected that a stability sample will fail sample suitability criteria. However, failing sample suitability criteria might also be caused by analytical errors. Therefore our default strategy is to reanalyze a sample in duplicate upon failure of initial sample suitability criteria.
What now if after this reanalysis 2 out of 3 measurements yield failing sample suitability criteria? Which relative potency value would be scientifically sound to be reported? And how should this be interpreted in view of the specifications?
An example to illustrate the situation: specs are set for 0.8-1.2.
Run 1 relative potency = 0.86 with failing parallelism
Run 2 relative potency = 0.78
Run 3 relative potency = 0.82 with failing parallelism
Does this sample is an out of spec? What if Run 2 would yield a relative potency of 0.83, would this be still an out of spec?
Moreover is it allowed to report a relative potency in case of failing sample suitability criteria?
Awaiting your response with best regards!
Philip De Decker