How Certain Should We Be About Uncertainty in E&L Analysis

Welcome once again from Maven E&L and my semi-regular Blog for issues surrounding the topic area of extractables and leachables (E&L).



This week I had the opportunity to present on the topic of “Future Trends to control risk from leachables”. If you would like to see this presentation, it’s available here.


One of the subjects I touched upon within this presentation was the subject of uncertainty factors as they relate to the topic of quantitation of extractables or leachables and the use of AETs (Analytical Evaluation Thresholds). This particular subject certainly is a issue which has been long debated by the E&L community and has many aspects in need of discussion, so I thought I would discuss some of these in this Blog entry.


So, let me start by saying, there is no agreed current position on this, and it is a area which I hope we can soon gain a greater agreement but it will not be a quick fix. Why are there no quick fixes here – well let me give you my view.


A lay person view of analytical chemistry is an interesting starting point and perhaps we can “blame” the media here. The public seen TV shows like CSI, NCIS and and other depictions of GCMS and LCMS analysis as high-tech black boxes which can detect anything and everything then tell you precisely how much is present and where its from.


Sadly, at the moment in the real world the truth is identification and quantitation are much more challenging and a lot less glamorous.



When we attempt to quantify extractables and leachables we are confronted with several issues. Including detection and quantitation limits, and most importantly for this discussion a variation in response for the substances we are attempting to detection then quantify. As every analytical chemist knows, GCMS and LCMS has its inherent limitations. In GCMS, substance must have a relatively high vapor pressure to by successfully detected and in LCMS, substances must be soluble in the mobile phase and be ionisable by the detector and well as other requirements.


But returning to uncertainty factors, the analysis for extractables and leachables is somewhat usual because the analytical chemist is being asked to detect “everything”. That is an untargeted analysis. To attempt to achieve this, analytical chemists add an internal standard to the analysis at a known amount and use this to determine when to quantify and what response to use for peaks detected.


In 2006, PQRI published its first set of recommendations covering E&L. One of the most significant part was the creation of the term AET. This allowed a hypothetical line across a chromatogram represent a limit which could be applied to the reporting of extractable or leachable data in conjunction with an internal standard. They recognised (as every good scientist should) that this was an imperfect solution and that different substances analysed will have different responses to a detector. So, they included potential solutions in their document.


  • Solution 1: Identify every substance, then obtain and use identified reference materials to demonstrate response variation and correct for it in the analysis conducted.

  • Solution 2: Apply an uncertainty factor of 50% around the AET to compensate for variation in response or an uncertainty factor value derived from a response factor database constructed from a representative set of substances, whichever is the greater.


But, and it’s a big but – Solution 1 is very difficult to achieve, you have to work long and hard to identify all the relevant substances and you may or may not be able to obtain the reference materials to derive the response factors. Solution 2 was therefore more readily adopted and with it the 50% AET solution.


Unfortunately, in many cases the second part of the sentence (the creation and use of a response factor database – where a greater variation required) was neglected.


The reasons for this I can only speculate, but what a blog entry without some speculation!


In GCMS analysis (which has been for many years the workhorse for E&L analysis), the responses the majority of substances found as extractables or leachables will not vary by more than an order of magnitude. Indeed, if you select an internal standard which itself is sits with a relative lower response then applying a final AET at half that response will be protective in almost all situations and hence a 50% value can be supported. Additionally, when the 2006 document was written the main focus was on leachables in metered dose inhalers and here the major sources of leachables were the elastomeric valve components which had extractables and leachables best detected by GCMS. So perhaps it is not that surprising that the 50% AET value took hold.


However, if we turn our attention to LCMS, then it is far from clear that a relative response factor based on 50% of an AET would be adequate to cover the range of responses potentially seen. LCMS has many useful qualities, but it is much harder to operate LCMS as a detector for everything. Each detector type and ionisation source has its unique characteristics making non-targeted universal detection very challenging. LCMS is perhaps now growing in importance and use for E&L. So, in screening for substances found as extractables or leachables, setting the limits without a consideration for response variation which might be much larger than that found in GCMS must be done very carefully.


Various approaches to this have been suggested, there have been some great contributions from both Jordi Labs and Nelson Labs. The Jordi Labs proposal is based around a multi-detector approach where the detector found to have best response is selected from a multi-detector set up and thus keep the factor low. The Nelson Lab approach uses less detectors but seeks to benefit from its large database in setting an appropriate factor to apply to GC and LC analysis but the problem remains, one thing is clear, consideration and some kind of consensus to drive this forward is becoming overdue. No agreement on the this helps no-one and at worst allows for the under reporting of leachables of concern, if labs are not thinking through their position on selection of an appropriate final AET. ICH Q3E which is being discussed now may or may not be able to take a view on this. Perhaps the individual analytical chemists need to be more thoughtful here too. I do not think it was ever acceptable just to say, “Well PQRI, has 50%” without a clear understanding of when and when it is not appropriate. Particularly with much of the analysis for extractables and leachables now conducted by 3rd party CROs, I would suggest pharmaceutical companies and other commissioning groups need to more fully appreciate what uncertainty factors truly mean for the analysis of their products for leachables.


Hope you enjoyed this latest article. The others in this series are available on the Maven E&L website here


If you have suggests for future articles or comments on this one. Please let me know


16 views0 comments