Page 1 of 1

%RSD Assay

Posted: Thu Mar 30, 2017 2:34 pm
by etaxene
Hello guys. I'm trying a new method for isopropanol (Assay) into a cream.
Customer has not indicated me what must be %RSD of 5 standard injections.

Do you think I can use <10% or <15%?

Thank you

Re: %RSD Assay

Posted: Thu Mar 30, 2017 5:13 pm
by Peter Apps
Haven't we been here before ? search.php?author_id=67961&sr=posts

Different analysis, same question, same answer; follow official guidance if there is any, if not ask the customer, if they don't care why would you ?

Peter

Re: %RSD Assay

Posted: Fri Mar 31, 2017 6:19 am
by etaxene
No.
Question is no different. First issue of cleaning. Now issue the official batch release. However, if you dont' have the answer please do not intervene to pollute the post. thank you

Re: %RSD Assay

Posted: Fri Mar 31, 2017 8:02 am
by HPLCaddict
No.
Question is no different. First issue of cleaning. Now issue the official batch release. However, if you dont' have the answer please do not intervene to pollute the post. thank you
It IS the same question.
"I have a XY method, what should be the specification for % RSD?"

Peter's answer is appropriate!
Follow official guidance - if not available: follow customer's suggestions - if not available: follow your SOP - if not available: (re)validate the method to see what the method is capable of and be able to implement an evidence-based specification for %RSD.

Re: %RSD Assay

Posted: Fri Mar 31, 2017 8:35 am
by etaxene
I do not agree. The questions are different because an %RSD of cleaning (with very low values ​​of standard) non Could be likened to a assay.

I simply ask if there are official guidelines on GC (where areas are low)

Re: %RSD Assay

Posted: Fri Mar 31, 2017 12:53 pm
by HPLC chemist
Even a chromatographic (HPLC or GC) 'cleaning validation analytical method' must comply with USP section <621>. That is the %RSD of the reference standard must be 2.0% for 5 injections or more (example 5.0%) for 6 injections.

A %RSD of 10.0% is the statistical LOQ of the method. I would severely criticize any method that had a %RSD of 10.0%!

Re: %RSD Assay

Posted: Fri Mar 31, 2017 2:01 pm
by Consumer Products Guy
For batch release, for finished products the FDA ORA regulation (can't remember the number) allows 5% RSD for release and for validation studies; I remember fighting with our QA Director who was hung up on the 2% cGMP RSD for pharmaceutical active materials. I won that fight.

For ethanol added to finished products, we had no issues meeting 2% RSD with our own validated assay, using GC with internal standard.

Re: %RSD Assay

Posted: Fri Mar 31, 2017 3:08 pm
by HPLCaddict
Even a chromatographic (HPLC or GC) 'cleaning validation analytical method' must comply with USP section <621>. That is the %RSD of the reference standard must be 2.0% for 5 injections or more (example 5.0%) for 6 injections.
!
WHERE does USP 621 say that?
Sorry, but this is nonsense! Specifications for RSD should ENTIRELY depend on what you want to measure. It is a huge difference if you want to measure %assay of an API with specification of 98-102% (you'd need tight RSD here) or if you're doing related substances with something like <0.2% or if you're measuring trace amounts during cleaning validation!
A %RSD of 10.0% is the statistical LOQ of the method. I would severely criticize any method that had a %RSD of 10.0%!
This is nonsense, too. What if you use a reference near the LOQ to measure RSD? For trace measurements (and cleaning validation can be trace measurement) it might be perfectly OK to have %RSD of 10%.

Re: %RSD Assay

Posted: Fri Mar 31, 2017 3:58 pm
by Peter Apps
Even a chromatographic (HPLC or GC) 'cleaning validation analytical method' must comply with USP section <621>. That is the %RSD of the reference standard must be 2.0% for 5 injections or more (example 5.0%) for 6 injections.
!
WHERE does USP 621 say that?
Sorry, but this is nonsense! Specifications for RSD should ENTIRELY depend on what you want to measure. It is a huge difference if you want to measure %assay of an API with specification of 98-102% (you'd need tight RSD here) or if you're doing related substances with something like <0.2% or if you're measuring trace amounts during cleaning validation!
A %RSD of 10.0% is the statistical LOQ of the method. I would severely criticize any method that had a %RSD of 10.0%!
This is nonsense, too. What if you use a reference near the LOQ to measure RSD? For trace measurements (and cleaning validation can be trace measurement) it might be perfectly OK to have %RSD of 10%.
The RSD of replicate injections of a standard depends on the repeability of injection volume and repeatability of integration, both if which depend on the instrument and its settings - there is no sample prep involved and so the RSDs should be as tight as the instrument can deliver - which I would expect to be a lot less than 10%. The RSD of the whole method with the same size of of peak will always be larger than the repeatability of standard injections, because the variation in sample prep contributes to the variation in the final result - which could sensibly be 10%, 20% or even more depending on the nature of the samples and what has to be done to them.

Peter

Re: %RSD Assay

Posted: Mon Apr 03, 2017 6:03 am
by etaxene
Exact. That is the question. An GC analysis has very small areas (about 100 mAU), despite being a determination of Assay.
Below 2% in this case is almost impossible. That's why I asked if it is tolerated on 10% for example.

Re: %RSD Assay

Posted: Mon Apr 03, 2017 2:07 pm
by HPLC chemist
Why is your peak area of the sample SO low? Are you trying to quantitate the peak below your LOQ?

Re: %RSD Assay

Posted: Mon Apr 03, 2017 8:11 pm
by HPLCaddict
Exact. That is the question. An GC analysis has very small areas (about 100 mAU), despite being a determination of Assay.
Below 2% in this case is almost impossible. That's why I asked if it is tolerated on 10% for example.
As I tried to point out: It depends on the specifications! If the specification for the assay are something like 95-105% or even 90-110%, than 10% RSD is practically impossible to use. If it's something like 20-130%, as it's sometimes the case for preservatives, then ... you may live with it, if you use enough repetitions for averaging.

But still: If the peak area is so low that the method is practically useless, then don't use it. It's bad habit to widen acceptance criteria in order to use bad methods...this is asking for trouble. Honestly, you should ask your customer how to proceed. They provided the method, they should have knowledge about it.