do you report bracket standard?

[flora0975]

why bracket std is need? do we need to report them? how to report them if it is needed?

thanks!

[Consumer Products Guy]

You need what your beloved QA department says you need...

[flora0975]

Ther is no QA SOP reference.

Is there a general rule for bracket std check?

You need what your beloved QA department says you need...

[jh1]

why bracket std is need? do we need to report them? how to report them if it is needed?

thanks!

You can use bracketed standards to insure integrity over the course of the run.

One of the simplest ways is to use relative change in response and make sure that standard peak (conc./area count/peak height) does not change by over a set percentage between standard brackets and/or over the course of the entire run.

[sanS]

We need bracketing standard because it is required as per USP.

[unmgvar]

and what type of bracketing do you think you want to use? and for what purposes?
there are several kind used

do you want to report and calculate by addition mode
by group?
by bracket mode

sometimes it is used in order to get results in a work that would be otherwise deemed canceled due to standard changes like RT and area.
as a gross example,
you generally use all the standards for all the samples, but in a certain work the RT has changed and so has the area and RSD does no longer fit,
but if you brake the work to 2 parts then each part stands within the specs and you can still get the results.
be careful with that. you better know how to explain yourself if you are audited, this is not a thing to use lightly.
if you do it every other day, then you will be asked on why did you not solve the problems and issues that made the work so unstable to begin with

and you better have an SOP written as well before you start using this option for that

[Karen01]

I've always been on the R&D side and could never see the point of recalibrating DURING a run. If the instrument/method is that unstable that it's necessary to do that in a run of reasonable length, I would be looking for another method, or finding a way fixing the instability, perhaps using an internal standard.

That said, I do believe running a QC sample after every X samples to prove instruments/method stability (for RT and recovery) should be done... With X being the number of sample results you can afford to lose if the QC sample is not with-in spec, balanced with reasonable throughput.

- Karen

[HW Mueller]

The possible interferences, changes of conditions are so manyfold in chromatography that I can not see how one can trust even a robust method for all time after one has set it up. Of course my experience includes analysis of human body fluids, which are always good for surprises, especially when samples are from patients of the intensive care station. (This was discussed not too long ago). If one shuns the cost of running regular standards one might be confronted with a much more costly shock.

[Consumer Products Guy]

Our sequences for cGMP and GLP samples are set up to inject System Suitablity/Calibration samples at beginning, then samples (up to nine vials), then bracket with a Calibration Standard after that.

Bisically this shows that the system was operating fine before and after the samples, so it's pretty darn etched in stone that it didn't fail, then "heal itself".

[Karen01]

Our sequences for cGMP and GLP samples are set up to inject System Suitablity/Calibration samples at beginning, then samples (up to nine vials), then bracket with a Calibration Standard after that.

Do you use that "bracketing standard" to recalibrate the run for the next 9 samples, or just make sure the concentration/recovery you get for that sample is in spec?

If it's the latter then that is what I used QC samples for.

Typically I created a 5 pt standard curve (if I'm doing a wide range - and in R&D I almost always am) by diluting a solution made with a single weighing.

Then I created one or more QC solutions (depending on how wide a range I'm measuring and what the method is used for) from a separate weighing. I run the QC sample(s) right after the standard curve to verify that I get the expected concentration for the QC (to cross check the standard prep to check weighing/dilution errors) and then run the QCs about every 10 samples (assay specific number) and at the end of the run. I set recovery specs for the QC samples and if one fails the results for all samples since the last good one are considered suspect. I also set RSD specs for all QC samples (assuming at least 3) in the assay to catch method stability issues.

Reported as part of system suit,were the individual QC recoveries, along with the mean and RSD QC Recovery for the assay

Bisically this shows that the system was operating fine before and after the samples, so it's pretty darn etched in stone that it didn't fail, then "heal itself".

[/quote]

Basically what I did, though the details differ.

- Karen

[Consumer Products Guy]

Karen: after the HPLC or GC is equilibrated, and ready, we set up a Sequence Table. We typically use ESTD% under Specify Report, and save data to FILE. The first line is a set as a calibration standard, with the appropriate column set as "Replace", one injection. The second line is is the same vial set as a calibration standard, with the appropriate column set as "Average", with four injections. The next line would be Sample #1, entered in as a Sample, and have the weight entered into the Amount column, and the volume the sample taken to into the Dilution column. Say eight more sample vials are then injected, and then the next line would be the calibration vial again, one injection, set as calibration standard, with the appropriate column set as "Average". Then say five more samples are added, and the final line as calibration standard, with the appropriate column set as "Average" again.

With an automated sequence summary report and "Reprocessing only" selected, and check mark in the box right next to it, and partial sequence, you can choose whatever calibration injections you choose to be included in your System Suitability Report (Choice #7 under Edit options in Sequence Output).

Now, for quantitation, the software (Choice #9 under Edit options in Sequence Output, and Compund Summary, not Sample Summary), the software uses the last injection of the line that has "Replace" and averages that with all the lines that have "Average" that are BEFORE the nine sample vials that are injected to quantitate those. For the last five samples, that average is alo averaged in with the calibration data from the calibration injection after those nine samples.

If one wants to use the overall calibration average of all calibration vials, one can run a partial sequence of just those calibration injections, then run a second partial sequence of just the sample injections.

I didn't write the software, I just use it.

There is a bracket function, and I've used it twice, but remember it was way more complicated to set up.

[lmh]

Just for the record, and working in typical barely-regulated academia, I do almost exactly what Karen does. If nothing else, my first introduction to hplc and lc-ms was Agilent's Chemstation, and I almost always set up calibration curves after running the samples (rather than before), using batch processing. With Chemstation's batch processing you can't do bracketing.

The QC sample approach is also very convenient because if you're obliged to report responses (peak areas) of uncalibrated peaks, assuming they are also present in the QC sample (which is true if you've used some sort of mixed sample as a qc), you can estimate the measurement error, and check for systematic variations through a sequence. Particularly useful in metabolomics.