Lock or Not to Lock Integration Parameter?

[Terry]

As I read the FDA warning letter (Google Key Word: 320-13-22) regarding integration, it was found that FDA expect the integration method to be locked and the same integration method is used on each analysis. Here below is the warning letter words in Italic.

a. The inspection documented that HPLC processing methods (including integration parameters) and re-integrations are executed without a pre-defined, scientifically valid procedure. Your analytical methods are not locked to ensure that the same integration parameters are used on each analysis. A QC operator interviewed during the inspection stated that integrations are performed and re-performed until the chromatographic peaks are “good”, but was unable to provide an explanation for the manner in which integration is performed. Moreover, your firm does not have a procedure for the saving of processing methods used for integration.

Your response did not include a description of the method by which chromatograph integrations are to be performed (e.g., what constitutes a chromatographic peak, how shoulder peaks are to be handled, etc.). In addition, your response did not include an audit of past chromatographic data to determine whether data used to support release and stability studies originated from appropriately integrated chromatograms.

In my lab, the analyst is assigned the privilege to change the integration parameter for processing method. And we have an SOP to define what is an good integration. The SOP also requires that the same processing method (integration parameter) should be used for all the results (blank, standard and sample) obtained in one HPLC/GC run.

If we lock the processing method, for simple chromatogram, such as isocratic HPLC assay, it will be okay. But for complicated chromatogram, such as gradient elution of stability sample related substances analysis. It is unavoidable that the existing integration parameter need to be changed to get a good integration for different impurity levels at multiple time points. Therefore administrator need to do this routine work, it will be a lot of work for one person. The lab productivity will be affected.

To summarize, my questions are:

Is it a mandatory requirement to lock CDS processing method?
Is the current practice in my lab accept acceptable from your perspective?

Regards,

Terry

[bunnahabhain]

In our lab, we also see the problem that adaptions have to be made to integration parameters over time. However, we always try to adapt it in a way that the tuned method also works with older samples. So we can be consistent at least inside one project. Of course, this will not always work. If changes appear to dramatic for this approach, we run an adequate number of standards. It is a good idea to store a sufficient amount of blank sample under stability test conditions for blank and fortified aged matrix etc.

Cheers,
Jörg