After determining the regression equation, I used it to calculate the concentrations of the samples using the peak area ratio of analyte/IS and for some samples am getting negative concentrations that I cannot use to plot the plasma concentration time plot
Hello,
If the HPLC method is under validation procedure, the limit of detection but mainly the limit of quantification will show the lowest value that you can measure accurately and precisely, all the other values (positive or negative) below that limit should be considered zero. If you are using a previously published validated method again a partial re-validation will be required in any case. Did you perform back calculation of your calibration samples as well as quality control samples so to be sure that the regression equation is appropriate in terms of specificity, sensitivity, accuracy, precision, range, qantitation limit and detection limit? If all these are estimated and still getting negative concentrations even though they should be positive and inside the regression limits, consider to check the calculation method if something is missing i.e. conversions etc. I think you will find a lot of information in the links attached.
Good luck!
http://www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidances/ucm368107.pdf
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2556583/pdf/cbr29_s_pgs49.pdf
Thanks Laurent,
I do not think its a conversion problem. I have double- checked.
Thank Mario,
I did back calculation for all the calibration concentrations and they were with 15% of the nominal values using the regression equation determined (r2=0.9979)
@Markus,the regression line did not pass through zero. What do I do to salvage the situation?
Near the detection limit you should get significant variability in the readings. Negative calculated values are not uncommon, especially if you are working with a non-zero intercept. This is normal; it does not indicate a negative concentration but rather indicates that there is statistical variability in the readings near zero. You don't need to "salvage" anything; you SHOULD get this if your plasma concentrations are low enough.
You do need to investigate the non-zero intercept and make sure that it is real and not a small concentration of contaminant in your reagents.
One way to avoid this situation is to use known addition. It is much more accurate for low level concentrations than using standard calibration curves.
- Badges
- Science method