how to evaluate robustness in gradient method

[Peter Apps]

[quote="Mattias

You would never trust that the method would work everywhere, just becuase you have succeeded to show that the method is robust (using maybe 3-4 parameters).[/quote]

No, I certainly would not trust a method to work everywhere just because I had shown that it still worked when some of its parameters had been changed by small and controlled amounts. Showing that a method will work everywhere is not what robustness is supposed to do. A method that works "everywhere" is rugged.

Peter

[rc_123]

Hi, interesting discussion .Very useful too.I just want to share few thoughts about flow rate changes in robustness study.The pumping capability and maintaining consistent flow rate across gradient runs is very crucial as the ratio and the miscibility or mixing of solutions keep on changing continuously.Every one of us might be aware of priming loses as one moves to higher organic portions of gradient runs.This is very clear evident that HPLC pumps suffer from some , if not considerable, sucking pressure.This problem is more pronounced if viscosity of mobile phase mixtures varies considerably from initial composition to final composition.Hence, flow rates may not be uniform across whole gradient run(May be with in the 95% to 105% of set flow.But it may be 100% at start of gradient,95% at end of gradient.This 5% variation between the start and the end of runs matters a lot) .More over flow rate accuracy calibration tests are typically performed with isocratic mode only.Hence flow rate variations should be part of robustness testing.

[wburt]

Definitely an interesting topic. As the definition of robustness testing was put to me, "small but deliberate changes, but not so large to explore points of failure".
I recently finished a gradient validation and below are the parameters we examined.
Initial organic % gradient value
Final organic % gradient value
column temperature
buffer ionic strength
initial wavelength
wavelength change (we're deliberately changing the wavelength during the run because of component wavelength intensity)
flow rate

These were examined using a design of experiment program, so multiple parameters were changed simultaneously. The max and min values were typically 5-10% of the nominal value. Most of these are instrument parameters. Why? Because parts are moving and they wear out. It's possible the proportional valve may not be perfectly accurate and get back to my starting point of 5 /95. There may be some variability in temperature due to the oven itself, maybe the location of the column in the oven or if there is influence of the room temperature. Maybe the pump piston seal is starting to leak so the flow rate may be a bit off, but not enough to shut down the system.

Remember, your suitability parameters should catch something if there is a problem. Robustness testing provides the ability to make notes of caution within your test method if a certain parameter is sensitive and to avoid going beyond that threshold.

[gtma]

For gradient methods, changing in organic should be a good indicator for flow rate and dwell volume changes. For area percent methods, changing the wavelength helps evaluate whether slight changes in wavelength (intentional or unintentional) affect the impurities results. When you change the critical parameters for chromatographic parameters, do you typically analyze the sample using the reference standard (assume external std approach) ran at the corresponding (non-standard) conditions or against the nominal (standard) condition? Is chromatographic robustness evaluate intentional or unintentional changes in these critical parameters.....my answer would be "both" and the way we evaluate them in the validation is critical in understand these parameters impact. I have seen too many parameters evaluated for impurities with limited solution stability, leading to significant variability (flaw) due to multiple runs.