-
- admin
- Site Admin
-
- Posts: 647
- Joined: Wed Aug 11, 2004 12:19 am
By Anonymous on Monday, May 24, 2004 - 02:41 am:
I've analysed the same working sample solution on two different HPLC instruments and observed a difference of approx. 0.10% in the level of total impurities as outlined below:
Total Impurities (%Area)
Instrument A 0.38%
Instrument B 0.29%
Given that both instuments passed the detector calibration test and has shown to detect the same number of impurities, can such a difference be attributed to variations between the two HPLC detectors ?
Are there any factors that need to be consider that could explain the difference in total impurity results observed between Instrument A and Instrument B ?
-------------------------------------------------------------------------------------------------------
By Anonymous on Monday, May 24, 2004 - 10:47 am:
Within the bounds of experimental error, I would consider those 2 numbers the same. Do you have enough data to statistically prove that they are different?
-------------------------------------------------------------------------------------------------------
By Tim on Tuesday, May 25, 2004 - 03:44 am:
Agree with first responder - this is within bounds of experimental error. In a method transfer, we would have a limit of +/-0.25% absolute difference between results for <=1% detected. Also, as first responder asks - do you have enough data to prove they are different - have you injected the same solutions multiple times on each instrument. Have you injected different preparations of the samples on both instruments?
-------------------------------------------------------------------------------------------------------
By chris allen on Sunday, May 30, 2004 - 10:13 am:
If you are calculating all impurities against a single standard component (or in this case area normalisation) a minor variation in the detection wavelenghths (e.g. 1nm) could cause this effect. I have seen impurity values reduce by half on some systems due to the detection wavelength being at the maxima of the main assay component but on the slope of the UV spectra for one of the impurities, as a result this impurity had to be quantified against its own standard. I suggest running on 1 system and varying the wavelength by 1 or 2nm each way and see what happens.
I've analysed the same working sample solution on two different HPLC instruments and observed a difference of approx. 0.10% in the level of total impurities as outlined below:
Total Impurities (%Area)
Instrument A 0.38%
Instrument B 0.29%
Given that both instuments passed the detector calibration test and has shown to detect the same number of impurities, can such a difference be attributed to variations between the two HPLC detectors ?
Are there any factors that need to be consider that could explain the difference in total impurity results observed between Instrument A and Instrument B ?
-------------------------------------------------------------------------------------------------------
By Anonymous on Monday, May 24, 2004 - 10:47 am:
Within the bounds of experimental error, I would consider those 2 numbers the same. Do you have enough data to statistically prove that they are different?
-------------------------------------------------------------------------------------------------------
By Tim on Tuesday, May 25, 2004 - 03:44 am:
Agree with first responder - this is within bounds of experimental error. In a method transfer, we would have a limit of +/-0.25% absolute difference between results for <=1% detected. Also, as first responder asks - do you have enough data to prove they are different - have you injected the same solutions multiple times on each instrument. Have you injected different preparations of the samples on both instruments?
-------------------------------------------------------------------------------------------------------
By chris allen on Sunday, May 30, 2004 - 10:13 am:
If you are calculating all impurities against a single standard component (or in this case area normalisation) a minor variation in the detection wavelenghths (e.g. 1nm) could cause this effect. I have seen impurity values reduce by half on some systems due to the detection wavelength being at the maxima of the main assay component but on the slope of the UV spectra for one of the impurities, as a result this impurity had to be quantified against its own standard. I suggest running on 1 system and varying the wavelength by 1 or 2nm each way and see what happens.
