justification of lower recovery

[pliva]

dear friends,
how to justify the lower recovery from a biological matrix. my question is-------------if i carry out recovery using one method and it turns out to 80% not 100%and all other parameters of validation are ok. some guideline tells that lower recovery will do for validation if we justify. can antbody clear me my doubt on how to justify that the method is valid.

[Rafael Chust]

As far your 80% recovery is reproducible - if it is linear with the concentration of your analyte - them there is no problem!

I use to say as a joke that if a variable is constant, them it becomes a constant!

Just as an analogy: no machine has 100% eficiency...

[Uwe Neue]

I would not be happy with a 80% recovery. It essentially means that you are loosing some analyte somewhere (or with MS detection you may have a signal suppression problem). What if this phenomenon changes the next time you have a slightly different matrix? Would you be willing to live with a method where sometimes you get 100% and sometimes 60%, and thus have a very large error in your analysis.

Unless you understand the source of the error, I think it is a large gamble. In my opinion, it is necessary to chase down where the problem is. If you know where the problem is, you can either fix it, or control it and live with it. To rely blindly on the assumption that the recovery will always be 80% for some unknown reason is not a good idea.

(I assume in all of this that you are using an internal standard).

[Mary Carson]

If you are using an internal standard (and I would NEVER assume this), then what you are probably looking at is "accuracy" not "recovery." As to what is acceptable in either case, it really depends on your application, your customer, and what regulations, if any, apply to your analysis.

[Uwe Neue]

On second thought, I can see how a liquid-liquid extraction method can work very well and with a very high reproducibility with a recovery of 80%. However, this is not true for a protein precipitation method or a solid-phase extraction method. If the recovery is that low, there is a fundamental problem with the method that one should chase down. This is independent of what regulations apply or what the customer might be happy with.

[pliva]

On second thought, I can see how a liquid-liquid extraction method can work very well and with a very high reproducibility with a recovery of 80%. However, this is not true for a protein precipitation method or a solid-phase extraction method. If the recovery is that low, there is a fundamental problem with the method that one should chase down. This is independent of what regulations apply or what the customer might be happy with.

yes friends,
i agree ,if the sample is linear with the standard. so how can we justify that the lower recovery is due to ion suppression. is anybody go through this case?

[HW Mueller]

Mary,
this has been mentioned many times, an internal standard (IS) can or should be used in conjunction with an external standard (ES), that way your IS allows precision as well as accuracy (or recovery). If you do this, the area/hight of the IS indicates, with each injectioin (!), whether something has changed, you don´t even have to use the IS to quantify, if you don´t want to do so. All the diatribe against IS is, therefore, not very much thought through. The only argument against an IS, that I can think of, is that it represents additional effort. The lowering of precision due to an IS, which is often cited, is without basis as one does not need the IS for the quantitation. In my opinion it is licentious not to use an IS if your recovery could vary. (I go with Uwe´s first statement above)

[Uwe Neue]

If the lower recovery is always a fraction of the internal standard, it does not mean that it must or must not be a particular mechanism that is causing it. It could be loss in the sample preparation method, it could be protein binding, and even ion suppression. If it is perfectly linear, than my frist bet would be a loss in the sample preparation method.

[Uwe Neue]

Ladies and gentlemen,

May I suggest here as well the reference to Matuchevski, Constanzer and Chavez-Eng in Anal. Chem 75 (2003), 3019-3030. It covers the subject of standards, how to use them and how to look at problems in a bioanalytical method very well.

Uwe