Chromatography Forum

We have several chromatographic methods where the run time is much longer than the retention time (RT) of the principal peak. No other peaks are observed after the principal peak until the end of the chromatogram.
eg: peak comes out at 4.5 minutes and run time is specified as 30 minutes (isocratic).

We work in an FDA environment so we have to stick to methods religiously. However I see it silly to waste all that time and solvent, with no obvious benefit. I understand that one cannot shorten runs when there is a gradient chromatography, during a related substance test and any test done on stability batches (due to degradant peaks that might be large enough that they appear in assays and dissolutions)
But... when doing release testing of assays, dissolutions and content uniformities the only peak of interest is the principal peak and at release testing there is no significant degradant impurities eluting after the principal peak.

So how reasonable is it to shorten chromatographic runs where it makes sense, does any one know how FDA looks at this matter?

BTW anyone knows a way I or someone in my company can ask this directly to them to clarify the matter?

regards

From the chromatographic point of view, there should be no problem with running a steep gradient after your peak(s) of interest to wash off late eluters (there are a couple of caveats, which I'll talk about later). From the regulatory point of view, it depends on exactly how the method is written (and was validated).

If the method specifies that you run the chromatogram out until the late peaks have eluted, then adding a washout before that would very likely be seen as a modification to the method and therefore require revalidation.
If the method doesn't specify the total run time, then you should be OK without revalidating, so long as you meet system suitability.

Now for the caveats:
- you will have to study the effect of re-equilibration time (after the gradient step before you run the next sample).
- you may run into instrument-to-instrument transfer problems if you have significant differences in dwell volume (gradient delay volume)
- make sure that you also include the gradient wash out in your calibration runs (calibrators and samples should be run under identical conditions).

For a given API or formulation, we have much shorter versions of the assay in place for BU, CU, and dissolution testing. They're usually isocratic and <15' (frequently more in the 5-6' range), unless we do have an issue with a late eluting peak that may interfere with subsequent injections. In that case, we're looking at a steep post-peak-of-interest gradient and equilibration time. Even these methods typically take no more than 15' between injections.

Not sure if it's just a convention, or whether it's documented somewhere, but I usually alllow isocratic methods to run for 2x the target peak's retention eg 4.5 minutes would mean a 14 minute run time, and 10 minutes produces a 30 minute run time.

Obviously, there are exceptions for many types of materials. The only approved ways to shorten existing methods would be revalidation, or to formally review historical data, perform a risk assessment, and have your QA sign off on the changes, perhaps with ocassional long runs when new raw material suppliers, or process changes, are introduced.

I'm not sure that you would have to revalidate the methods if you have robust data to defend such a modification, but there are others here who could better advise on that.

My feeling would be that adding a gradient step at the end could be counterproductive in instrument time, and more likely to attract a negative response from QA, but judicious dispensing of boxes of chocolates might help...

Please keep having fun,

Bruce Hamilton

The key test is to run a number of samples for the longer run time and observe if any other peaks are visible. If there are none, I can see no objection to shortening the run time to one minute after the peak of interest. To revalidate for such a trivial change is what gives validation a bad name!!

If the late eluted peaks are not calculated, I always cut the run time so that the peaks eluted in the next chromatogram.

Thanks for all your contributions.
Unfortunately we have to keep to the absurdly long runtimes for the time being as the general feeling in the company is that it would require revalidation (as AdrianF said minor issues like this and sometimes wording in the pocedure combined with people who do not have enough technical knowhow give validation a bad name)

@syx---If the late eluted peaks are not calculated, I always cut the run time so that the peaks eluted in the next chromatogram.

Does not that give interference in subsequent runs.

Ideally, I would allow for the late eluting peak to come out between the end of one injection and the start of the subsequent injection or just around the solvent front of the subsequent injection. The only reason I can see for having such a long gradient method for assay is because that set of conditions provides a stability indicating method. But I think it is a waste of time for dissolution.

Hi,
If the method specifies the run time of 30 minutes, you will have to stick with it. The problem has been faced in our lab when we were doing validation of a product in which an unknown peak appeared after every 10-12 runs. The run time was extended and found that the unknown peak is highly retentive and is eluted very very slowly. You never know without doing validation that shortening the run time can really affect your method.

Kamal

Oliver reports no peaks eluting after the peaks of interest, so I'd go with what AdrianF stated: cut down the run time. What's important is that the chromatography of your sepoaration is exactly the same, so I'd say go for it. If it was me, yes, I'd shorten the run time.

@syx---If the late eluted peaks are not calculated, I always cut the run time so that the peaks eluted in the next chromatogram.

Does not that give interference in subsequent runs.

It can, unless things are very consistent and your timing is good... I've resorted to doing this in the past, if I had a column that would take forever to settle down after a gradient and the late eluter's RT was 3x or greater than that of my peak of interest. The only problem is that it can take 3-4 injections to see if injection 1's late eluter is interfering w/ injection 4 and so on... this method is for risk takers only.