Trial Injections in a chromatographic analysis

[mataji]

When performing a HPLC or GC analysis, it is customary to make a few trial injections to ensure that peak RT, shapes etc are acceptable and the system is ready for analysis. How many trial injections are acceptable to establish system readiness?

[tkubowicz]

Hello

There is no requirements for trial/test injections. It is totaly up to you.
When you see that instrument (GC or LC) is working stable:
-detector signal is stable
-pressure is stable
-temperature is reached
You can start run.

At least it is my understanding and common practice

Regards

Tomasz Kubowicz

[tom jupille]

Depends on your working environment. In a regulated environment (pharmaceuticals, environmental, etc.) a validated method will have an associated "system suitability" test to confirm that the method is working and is, in fact, suitable for the intended purpose. That test will specify the number of replicates to be run and typically will have minimum values for things like resolution, plate count, baseline noise. See section J of this document for "typical" values: http://www.fda.gov/downloads/Drugs/Guid ... 134409.pdf

[lmh]

just to add: even if you have a requirement to pass certain system suitability tests, you can still run all your samples without doing any trial injections first. For example, if you want to run a batch of samples overnight, you can set up your calibration standards, qc samples, blanks, analytical samples and everything to run. You then look first at the qc samples and any other system suitability things that need to be passed to prove the method is working as intended. If these pass, you're allowed to look at the samples.
This is why the quant-browser in Xcalibur, for example, on opening a batch of samples, will ask first whether you want to see all the samples or just the qc samples.

[Mattias]

In the light of recent warning letters, I think you just have to be careful that you set it up correctly in the written method. That it is documented how it is done.

There have been cases where companies have been running "test sequences" and then just kept the ones with the good results and reported these...

[mattmullaney]

Hi Mattias,

I agree with your assessment. To clearly be on the side of "good", perform test injections only using standard solutions and document as such. The main pitfall is to inject sample preparations, please do not do this.

For these types of tests, I have injected singly resolution and sensitivity standards (for example) that are separate from a formal system suitability. I also ensure in the documentation that this is what I have done in as clear a way as I can muster.

[Rndirk]

One thing i can add to the posts above is, when you have to decide yourself how to set up / evaluate the system suitability test (SST), is to look for the weak or critical 'points' in your chromatogram.

For example, in our GCMS analysis of polyaromatic hydrocarbons (PAH) in waters, our SST is the injection of one standard with all the PAHs in it of a low concentration. Here we evaluate the separation between the critical pair (has to be separated more than 50%) and the absolute area of another component (has to be above a certain area) that we know by experience is more prone to loss in sensitivity compared to the other ones.

In one of these 2 tests fail, the instrument needs maintenance (can be as simple as changing the liner). If our SST is OK, we know that all the other ones will be OK because those are the weakest points.