The standard curve reagent in our validated assay is a polyclonal antibody. It is used to measure a drug substance that is also a polyclonal antibody. This is not measuring like we would like, but it is the closest we can do. However we are having problems with our primary curve antibody which is starting to degrade upon storage. It is stored in aliquots at -80C, but still it is starting to fail. We need to replace the standard curve antibody but of course it is next to impossible to get an exact replica as it is a polyclonal. We have one polyclonal preparation that gives results with samples from 120 -130% of that obtained with the original standard curve (Both have concentrations derived from measuring the A280 to get a protein concentration for the affinity purified polyclonal preparation). But I was wondering what people think if we now 'normalize' the new preparation against the old preparation. I would do this by treating the new preparation as a sample and getting an 'ELISA' concentration for it. Then 100 ng/mL of the new preparation would give exactly the same O.D. signal as the original standard curve. To achieve the same signal I have essentially increased the amount of antibody in that well. Paralellism may be an issue, therefore I would test the range of the assay with spike recovery samples to test for accuracy and precision over the range of the assay. These can be prepared with the original preparation and curves of both run on the same plate. This would ensure that any samples tested will give the same result with both standard curve preparations? We still have available original standard curve material that has not failed and we have independent controls to verify this, but it is degrading fast!
You have a sound plan to replace your degrading standard. Replacement standards, like polyclonal antibodies in ELISA, can never be identical to the original standard in terms of activity (binding to antigen). New standard is calibrated against the current standard in multiple assays and a unitage is assigned relative to current standard. You have to monitor for the slope of the new standard, which depends upon the avidity of antibodies. If the slope is different, then there will be problems with the parallelism, as you mentioned. In the calibration of a new standard, you will evaluate the parallelism of standard curves generated by new and current standards.
You have a sound plan to replace your degrading standard. Replacement standards, like polyclonal antibodies in ELISA, can never be identical to the original standard in terms of activity (binding to antigen). New standard is calibrated against the current standard in multiple assays and a unitage is assigned relative to current standard. You have to monitor for the slope of the new standard, which depends upon the avidity of antibodies. If the slope is different, then there will be problems with the parallelism, as you mentioned. In the calibration of a new standard, you will evaluate the parallelism of standard curves generated by new and current standards.
Thanks Rajesh, it's good to know we are on the right path. We have had discussions and are changing our approach slightly. We will now during the bridging study, measure a sample with both curves and work out what correction factor will need to be applied to samples measured on the new curve, to bring them back in line with the old curve. That way we are not 'altering' the concentration of the new curve to get the equivalent answer, but keeping it the same and then applying a correction factor to the results. We think this approach will be more acceptable to regulatory auditors and is commonly applied when changing batches of standard curve material in analytical assays. We are saying we know the new standard is different and we are accounting for the difference by the correction factor. Of course the parallelism studies still apply and we will lose a part of the range of the assay. Now we just need to do the work quickly while we still have original material to assess against.
Hi Alison, The regulatory agencies don't like correction factors. Also, it may be misleading to determine correction factor from one sample. Most acceptable approach is to calibrate the new standard against the current standard in at least 10 determinations and assign the unitage to new standard based on the mean of multiple assays (at least 10). If you want to do more, you can select a few samples and test these against both standards, current standard and new standard calibrated against current standard. You should be getting results of samples within the inherent variability of the method (ELISA usually 15% CV), determined with both standards.
Dear Alison, I think you already got your question answered by Rajesh K Gupta, but I just wanted confirm you are on the right track. We have encountered the same problem in the past and solved it the same manner. We bridged to a new standard curve by first analyzing the new standard curve as a sample and determined the nominal value of the new standard curve by analyzing the sample at least 20 times in at least six independent runs. I hope you still have enough proper standard curve material to analyze your new standard. Good luck!
Hi Alison, just curious, do you perform this assay frequently? Is it for validation of a drug that has to be administered to patients? If the first answer is "yes" I would spend some money in making a pair, or at lest the capture antibody, monoclonal. If the second is also yes, you have to be sure that your assay is in compliance with Regulatory Organisms rules. If instead it's for in vitro use of the compound and no diagnostics or clinical pharmacokinetics is involved, you can proceed as you suggest.
- Badges
- Science method
- Similar topics
- Medicine
- Immunology
- Immunology Techniques
- Immunoassay
- ELISA