Method changes

[wanda50]

I am in the same pickle having to re-evaluate our 20 year old methods. One that I'm working on uses uBondapak, and the peak of interest elutes over 3 minutes. In these days of concern over impurities and potential effects on the body, IMHO this method has to be investigated. Who knows what lurks under a 3 minute peak? another product of ours also had a huge tailing peak with 2 impurities hiding underneath as determined by spiking studies. Luckily that method has been redone.

Our customers are asking these questions more and more often, and as an API company, we have to get it together if we want to stay in business.

[HW Mueller]

Who knows what lurks under any peak if the main concern is to satisfy regulations? Did you sell pills for 20 years with extra poison in it?

[AdrianF]

Mr Mueller
Your aphorisms should be collected together and published!

Excessive regulation is hindering innovation. For instance when monolithic columns first arrived on the scene I wanted to change many of our methods to use monoliths. In pilot studies I showed that several methods could be greatly greatly speeded up. T tests showed they gave equivalent results - initially the idea was greeted enthusiatically but then I was told not to bother.

Why is not sufficient to run say ten samples by both methods. If they are shown to be statistically comparable, move to the new method.

[tom jupille]

Excessive regulation is hindering innovation.

Actually, I agree with that statement. The discussion centers on your definition of "excessive".

When I run a method development course, many of the questions typically center around what you have to do to validate a method. By definition, validation is demonstrating that a method does what it purports to do. How much work that involves hinges on two questions:
- who wants to know?
- what will it take to convince them.

If you deal with pharmaceuticals, the answer to the first question is "regulatory agencies". The answer to the second is spelled out in detail in the regulations (e.g., ICH, USP, BP, etc.). In some cases (e.g., USP), to run a changed method, all you need to do is to demonstrate equivalence. In AdrianF's example, that should have been sufficient. In other cases (e.g., FDA requirements), adjustments are OK but modifications require revalidation; the distinction between the two is what the appendix to ORA Lab 5.4.5 is all about.

This makes sense (to me, anyway!). If you have modified the method, it ain't the same method any more. For some applications (e.g., potency, content uniformity, or dissolution) where you have a only a few (or one!) large peaks to deal with, comparison with the results of an established (i.e. your old) method via appropriate statistical tests would be appropriate. For others (e.g., impurity profiles or stability testing), I would want to see a lot more proof that the change in stationary phase chemistry is still showing all the same peaks.

[Rande]

Excessive regulation is hindering innovation.

When dealing with an old method for an approved drug, you typically need a financial incentive to invest the time and effort needed to revalidate an analytical method. Just because there are newer tools available, does not mean that it is worth performing a method validation and equivalency study with the old method.

(This is usually a financial decision - not scientific).

If an antique instrument is ready to break down for the last time, and the new instrumentation is automated - cheaper to run - and much more accurate, then you will be happy to change. Just because you like to play with new toys, will not convince your management (or the FDA) that it is a needed change.

If the old method is a USP test - then you have a lot more explaining to show why you are not following the approved method. Just because the USP method might be "50 year old wet chemistry" that is labor intensive and has poor sensitivity - is not an excuse.

I mean no offense to the USP, but many old methods are still in use because "If it ain't broke - don't fix it"!


Is this always the best science ???

Many years ago, I know of an approved drug (Biologic) that was using an older HPLC column for release testing. The R&D guys were using a competing column that had much better resolution. The QC group decided to evaluate the new column, because it did look much better.

There was a minor problem , none of the product lots would pass release specifications with the new column.
Very simple solution - DO not change the method, no need to explain to the FDA why you want to change your specifications and allow twice the level of contaminants to be injected in a patient.

In reality, the only difference would be the reported numbers. This would not change the fact that there was over 10 years of patient safety data with the existing product.

[AdrianF]

Tom

Thank you for your illuminating reply. Can I clarify the issue of equivalent methods. If a method involving one main component is statistically equivalent to an old method would it be acceptable to the regulatory authorities? You seemed to suggest it would but then added some caveats to indicate perhaps it wouldn't.

It is this kind of uncertainty which means it is always easier to stick to the status quo.

Rande

The example you gave of a method which was 'too good' involved downright dishonesty. When lead was found in the paint of some toys from China - the reaction wasn't - never mind-even though similar toys had been sold for years. If impurities are discovered they should clearly be brought to the attention of the authorities.

I had a similar problem recently where a titrimetric method had been used. I tried an HPLC method and found the product failed. Some argued that as it passed using the registered method it was OK. But wiser counsels prevailed

Finally

I have recently been reading a book called 'Wikinomics' ( How mass collaboration changes everything) It would be great if this principle could be used in the development of regulated methods. 'WikiMethods' so methods could move on much more quickly.

chromforum does a great job of enabling collaboration it would be brilliant if there was a way of distilling the wisdom of thousands of analysts out there!

[Bryan Evans]

Wanda -

Just curious, are looking to redo the method completely or just
replace the column because the older column technology is not
giving you reproducible results?

[tom jupille]

AdrianF: My reading of the regulations (note, this is not necessarily what I would like to see!) is that USP explicitly allows modified methods so long as you demonstrate equivalence. FDA does not. Which, as you point out, leaves you in a state of uncertainty.

If it were my problem, I would put together an SOP, defining how your lab will deal with modifications/adjustments/equivalence. I think regulators generally are comfortable with good science, provided the ground rules are set ahead of time. What causes problems is "retrospective" setting of guidelines.

[wanda50]

Since the USP and EP are harmonizing on allowed changes, that is what I would work with. Tom is correct that you need an SOP stating what you will do.

next to my problem - someone asked if I was just changing the method due to the column. No, I'm not. The validation was improperly done even for the early 90's, and we basically have to revalidate the method. I really hate trying to make a silk purse out of a sow's ear.

So, here is the 3 minutes peak for the lurkers, I mean with lurkers.



for some reason I can't get the thing resized. sorry

[Uwe Neue]

Just a small comment in defense of Ye Olde Column...

This peak is clearly massively overloaded, comparing the peak width of the large one to the smaller ones next to it. If it is overloaded, it won't get any narrower on a modern column with smaller particles.

[wanda50]

Oh, I know it is overloaded. I imagine they did that so they wouldn't have to do a separate Rel Subs method. In any case, I have to fix something. Even reducing the concentration on the column will make me have to revalidate the whole thing. Again - silk purse from sow's ear?

[Bruce Hamilton]

I suppose my standards are somewhat lower, because I don't see that method as a sow's ear. The peak of interest elutes around 9.5 minutes, and although massively overloaded, is resolved from some impurities eluting before and after.

Yes, it can be improved, but the main peaks isn't eluting at 3-4 minutes, and the baseline is fairly horizontal, and retention times and areas are probably repeatable.

The method has presumably served it's purpose for 20 years, and should have recovered development costs many times over. Given the column technology and standards of that time, the method may have been considered suitable and acceptable.

It's clients who define the amount of quality applied to products, including analytical methodology used for IPC and CoA. The need to revisit and review methods and products is often defined by clients who provide more onerous purchasing specifications ( perhaps by including an updated compendial monograph ). If clients want well-characterised or compendial products, they will pay.

Bruce Hamilton