Is it considered as peak re-integration or not?

[Star Stella]

Hi,

Does anyone experience the problem with chromatographic peak recognition while using Analyst software? Quite often lowest concentrations of our samples are not recognized by Analyst software as a peak, and since we are working in the regulated environment , we are using integration method used in validation and we need to report all re-integrations explicitly. But to us, it seems that if you are not changing noise threshold or peak splitting factor (which individually affect on determination of peak start and end), instead using "Peak Selection" function, which changes only Expected Retention Time, it's not considered as re-integration.

1. I would like to hear how you are dealing with this type of problem in your labs (in regulated environment)?

2. Do you consider it as re-integration?

Thanks in advance!

[bunnahabhain]

Our QAU says: Everything that is not done fully automatic is manual. This also applies to what you describe.
If only the retention time has shifted (why is this the case for low concentration standards only? Do you see different retention times for different concentrations?) then go ahead, change the expected retention time or increase the RT window in the integration method, save the method and document that you use this method from now on. Or even better, integrate all other sequences from this study with the most recent integration method, too. That way you are waterproof, even if it is a bit more effort.

Regards
Jörg

[Steve Reimer]

The other consideration is do you treat your samples this way?
If only the standards are processed this way then you are likely to have false negatives when the concentration in your samples in similar to your low standard.

[lmh]

I assume that if you cannot detect signal in low concentration samples, one of two things has happened. Either:

(1) the samples are a lower concentration than your lowest standard, or
(2) the samples are giving a much noisier signal than your lowest standard, and as a result, the integrator cannot find the peak.

If (1), then many would argue that you are already operating outside the valid range of the method, and you shouldn't be worrying about whether what you are doing is reintegration - you should be going back and extending your standard curve to encompass lower concentrations.

If (2) then the fact your current settings can't integrate the peaks is telling you something: it is telling you that there are other compounds too close, and interferences that again may be impacting the quality of your results. It may be that the limit of quantification in these samples is a lot worse than the limit of quantification of the method as a whole given clean standards, and the samples may contain too little to quantify accurately.

Sorry to be gloomy!

On a happier note, if it's that your integrator settings were created during method development, using perfect standards, it could just be that they were always too optimistic, and there's scope to make a good method here, but it really needs someone to do some real method development finding appropriate integration parameters, and checking appropriate LOQ, in real samples?

[Star Stella]

Thank you all for your valuable feedback!

Dear Jörg,
Retention time was changed a littile bit, but the problem is that it's within retention time window, but Analyst don't recognize it as a peak. Most of the time it takes place with lowest concentrations, blanks, but sometimes with high concentrations also. The problem is why it didn't integrate the peak, while it is within RT window, despite of the fact, that RT was changed a little bit.

Thanks,

[Steve Reimer]

Some integrator routines won't perceive a peak if it doesn't find an end within the window. A long peak tail will prevent peak assignment.

[Star Stella]

Dear Steve,

Thank you so much for your help. Now it explains why big peaks are not integrated while within retention time window. And also that points out the fact that we need to be super cautious with retention time shift with small RT windows.
And another question: 1. Which retention tome window do you use generally - default from software?
2. For Analyst it is +-15 seconds, will it be considered OK from regulatory point of view if we will increase it twice?

Thanks in advance!

[James_Ball]

Dear Steve,

Thank you so much for your help. Now it explains why big peaks are not integrated while within retention time window. And also that points out the fact that we need to be super cautious with retention time shift with small RT windows.
And another question: 1. Which retention tome window do you use generally - default from software?
2. For Analyst it is +-15 seconds, will it be considered OK from regulatory point of view if we will increase it twice?

Thanks in advance!

As long as all of your determinations are done with the same settings then you should be ok in a regulatory setting. If the regulatory method dictates what the retention window must be then you are held to that definition of RT window. Some methods dictate that you must run a series of injections at a certain concentration level and do a statistical evaluation to determine window, others just set it to a specific +/- minutes.

The main thing is to always treat standards and samples the same way.

[Star Stella]

Dear Steve,

Thank you so much for your help. Now it explains why big peaks are not integrated while within retention time window. And also that points out the fact that we need to be super cautious with retention time shift with small RT windows.
And another question: 1. Which retention tome window do you use generally - default from software?
2. For Analyst it is +-15 seconds, will it be considered OK from regulatory point of view if we will increase it twice?

Thanks in advance!

As long as all of your determinations are done with the same settings then you should be ok in a regulatory setting. If the regulatory method dictates what the retention window must be then you are held to that definition of RT window. Some methods dictate that you must run a series of injections at a certain concentration level and do a statistical evaluation to determine window, others just set it to a specific +/- minutes.

The main thing is to always treat standards and samples the same way.

Thank you!

[mckrause]

I am making the following assumptions:

1. You are using an older version of Analyst, thus you probably have an older instrument
2. You are probably NOT using UHPLC coupled to capillary LC columns.

If this is the case then +/- 15 seconds is WAY too tight for HPLC windows. We run 2.1 mm columns at optimum flow rates and still get 45 second wide peaks. Analyst will not integrate correctly unless the entire peak is contained within the retention window, so you need to set the window wide enough to accommodate your peak width at maximum concentration +/- the variance you see in the peak retention times (this can be quite a bit, depending upon the polarity of your compounds). A typical window for us is 1 minute to 1.3 minutes wide; this is the unfortunate reality of conventional HPLC. Capillary GC it ain't!

[Star Stella]

I am making the following assumptions:

1. You are using an older version of Analyst, thus you probably have an older instrument
2. You are probably NOT using UHPLC coupled to capillary LC columns.

If this is the case then +/- 15 seconds is WAY too tight for HPLC windows. We run 2.1 mm columns at optimum flow rates and still get 45 second wide peaks. Analyst will not integrate correctly unless the entire peak is contained within the retention window, so you need to set the window wide enough to accommodate your peak width at maximum concentration +/- the variance you see in the peak retention times (this can be quite a bit, depending upon the polarity of your compounds). A typical window for us is 1 minute to 1.3 minutes wide; this is the unfortunate reality of conventional HPLC. Capillary GC it ain't!

Thank you so much for comprehensive answer. We are using UHPLC with 2.1 mm columns, and you are right: peak width of our max concentration is more that 30 seconds, so in the next methods we need to increase retention time window in order to not have problems with peak integration.
Thanks you again!

[Kenn]

I currently run Analyst Versions 1.4 through 1.6 on 3 to 5 different mass specs daily. Usually 6-8 different compounds with different methods. I typically sit crunching numbers each morning for the first few hrs so I deal with what you are talking about ALL the time, all strictly GLP.

If you are the analyst when you click through the chromatograms you know if a peak stands out as wrong. If it is wrong, I fix it. Never with a smooth. Usually bumping the noise threshold up or down. Analyst will label every cgram with either none, modified, or manual for integration so whatever you do it's going to be in your data. If a couple c-grams seem to be doing the same, go back a build a new quant method on your low std. Nothing wrong with building a new quant method after looking at the c-grams you generated. If you modify those parameters and everything integrates, great, if they don't then fix the ones you have to. It is sometimes difficult if not impossible to get a method to integrate a a high and low standard properly and analyst will report everything you do. The key here is if someone decides to scrutinize your data and look at that integration, is it right. If someone says why did you re-integrate that peak you can highlight the peak, re-integrate it using the original parameters and say look at it yourself, that's why, and do it comfortably. I would much rather have my peak integrated properly than have someone say "what do you call that?"

On versions of software other than Analyst on different instruments I used to initial a cgram as having had something modified but Analyst does it for you. If someone asks if a peak has been reintegrated you can say "what does the chromatogram say on it?"

[Kenn]

Our QAU says: Everything that is not done fully automatic is manual. This also applies to what you describe.
If only the retention time has shifted (why is this the case for low concentration standards only? Do you see different retention times for different concentrations?) then go ahead, change the expected retention time or increase the RT window in the integration method, save the method and document that you use this method from now on. Or even better, integrate all other sequences from this study with the most recent integration method, too. That way you are waterproof, even if it is a bit more effort.

Regards
Jörg

If you are using Analyst and have to manually integrate a peak it is recorded in the data and the cgram will say it. Every sequence stands by itself and there is no need whatsoever to have the quant method the same for each. It is perfectly acceptable to build a quant method on your low std with every sequence, actually it makes sense to. What the ideal integration parameters were 2 weeks ago may not be the same today so trying to have them match over time is only detracting from the accuracy of your data. The scientific integrity of the data is far more important than QA. If your RT is moving around you have bigger problems than integration. Again, the key is to report everything you do and be able back up those decisions scientifically.

[Kenn]

Dear Steve,

Thank you so much for your help. Now it explains why big peaks are not integrated while within retention time window. And also that points out the fact that we need to be super cautious with retention time shift with small RT windows.
And another question: 1. Which retention tome window do you use generally - default from software?
2. For Analyst it is +-15 seconds, will it be considered OK from regulatory point of view if we will increase it twice?

Thanks in advance!

Yes it will be fine from a regulatory point. As long as you report what you do and have sound reasoning behind that decision you should be fine. I think our default is 20 and I can't remember ever having to change it.