Clarifications on FDA Guidelines on Method Validation

[drhlrao]

Hi everybody,

Suppose I am performing Bioanalytical Method Validation (by LC-MS) as per USFDA guidelines (May 2001), and wish to first prepare a standard curve with 6 non-zero concentrations.

1. What would be the minimum number of injections of EACH standard concentration?

2. Are the results of single injection of each standard concentration acceptable as per USFDA guidelines?

3. Is unweighted regression analysis accepted as per USFDA guidelines?

4. Can we include the zero sample (matrix sample processed with the internal standard) in the statistical analysis?

5. If weighted regression is compulsary, what are the weighting factors to be tried in order to justify the statistical goodness of fit?

Thanks in advance.

[bert]

Do you include QC samples in your validation runs? For your calibration curve, the simplest model that adequately describes the concentration-response relationship should be used. Unweighted regression is possible, but for wider concentration ranges, a weighting factor is often used. This is discussed on this forum in several posts. Single injection is possible, but you should consider reinjection at the end of your analytical run.

regards Bert

[Alfred88]

Dear:
The purpose of validation is to demonstrate that method works as intended.
So, you need to prepare the matrix with different concentrations of interested compound (range may be 80-120% or better yet 50-150%), plus a placebo without it, and test them as you normally test samples.
Before each run, do a system suit check, by running your standard six (or five) times.
Run each sample two times, and your bracketing standard once. Loop this cycle to the end. (But if one injection takes 1 hr, then you may run 4-5 sample injections, then a bracketing)
Each concentration level must be determined at least three times, for a total of three different levels.

If you need more detailed, step-by-step instruction, you can attend one of many workshops offered by Institute of Validation Technology, www.ivthome.com. There are many catchy titles such as "SOP for Validation of Chromatographic Methods," "Developing and Validating Stability-Indicating Methods."
Note: I am not related anyway to Institute of Validation Technology.
Alfred

[guenzia]

Dear drhlrao,
I could not understand completely your questions.
If you refer to establishing accuracy and precision of the method, then validation according to FDA guideline is performed by analyzing calibrators (at least 6 non-zero) and QC samples (at least 3 different concentrations, and at limit of quantification) over n independent series (normally 5 or 6).
If you refer to establishing the calibration curve model, you can perform a single run containing a blank sample, a zero sample and six to eight non-zero samples, and then follow the criteria clearly stated in the guideline, including "excluding the standards should not change the (regression) model used".
Therefore, coming back to your questions:
1) 1 injection
2) Yes
3) Yes
4) Absolutely NO
5) See 3)

PS: don't be misled by the answer of Alfred88, who refers to a QC/QA validation problem, not to bioanalytical assays (drugs in biological fluids)

[james little]

Also need to be concerned about matrix suppression/enhancement. For example, see my website..

http://users.chartertn.net/slittle/default.htm

the first topic on page..

In my paper is an excellent reference from P Rudiwicz which in detail outlines calibration, QC, freeze thaw, etc. Get a copy and read and I think it will answer a lot of your questions.

reference 21 in:

http://littledomain.com/james/files/text.pdf