A couple of questions we are mulling over...

1.We are running biological samples at a very low concentration, therefore the run time is long to separate out impurities, followed by a meOH:H2O wash in each run (80 minutes and not possible to reduce). We have run several calibration curves and found good linearity. So in future with "real" samples we want to avoid having to inject 21 samples (7 calibration samples, 5 system suitability samples and 3 QC samples at 3 different concentrations) before any patient samples are run. Is there any way to do single or dual point calibrations??

2. In a separate method we have run accuracy and precision experiments. Precision is well within our limits, however probably due to a pipetting error, the higher QC samples are all out by 15-20% consistantly. Linearity of the method has been proven, is there any way to include this data or will the accuracy and precision experiments need to be repeated?

Thanks

James

1. So long you can show that the range you analyse falls in your linear range( I mean validated the method), I don't see why not a single (through zero point) or dual point calibration.

2. I think you can probably repeat the affected samples and disqualify those 'out' samples attributing to pipetting error. Else you have to repeat everything from scratch, not just by saying probably pipetting error.

Mike

1. Depends on your regulatory situation. In the US (FDA), any modification to a regulatory method (and changing from a 5-point calibration to a 1-point calibration is *definitely* a modification) requires revalidation. Strictly speaking, this applies to cGMP (i.e., manufacturing), but if your data will end up being submitted to FDA (e.g., bioanalytical work under GLP), then you should assume the same guideline will apply. If non-regulated (e.g., not related to clinical studies), then it's a judgement call.

2. Again, depends on your regulatory situation. Under cGMP, this situation would result in throwing out the entire run unless a laboratory investigation could pin down the source of the error. Under GLP, you have to assume the same. If non-regulated, then it's your call.

Thanks for the advice, were running to GcLP so it looks like we will have to do the full set of calibration samples as we don thave time to re-validate that one.

And we've already decided to re-do the whole lot of the accuracy + precision samples for the other - just to be on the safe side.