I understand how to calculate limit of blank (LOB), limit of detection (LOD), and limit of quantitation (LOQ) in the traditional way i.e., average and SD of raw analytical signal of blanks and low concentration samples. But the enzyme assay makes things unclear. I am not sure what measure to use in my calculations.
The enzyme assay with seven calibrators produces a very nice dose-response that I fit with a nonlinear regression model (4-parameter aka. Hill equation). Running blank samples gives three relevant measures that I could use to calculate LOB, LOD, LOQ:
- Raw absorbance at 405 nm
- Problem here is absorbance increases over time as the enzyme goes to town on substrate. Results across multiple experiments are not comparable. Not a good candidate.
- Normalized enzyme activity (as a % of uninhibited control)
- I like this option the best because it's not time-dependent. Blanks run on various days are comparable.
- Concentration (ng/mL)
- Concentration is interpolated on calibration curve using % activity as the input. Problem here is sometimes the nonlinear model cannot calculate results that are beyond the upper asymptote (parameter A for those familiar with the Hill equation). So roughly half of my blanks don't get a concentration.
I feel like I want to proceed using % activity as my measure for calculating LOB, LOD and LOQ but it just doesn't feel right.
For example, my 20 blanks run over 8 experiments have an average enzyme activity of 99.4% and standard deviation is 2.2%. If I calculate LOB the traditional way LOB = mean of blank + 1.645*SD of blank, I get 103% which doesn't make sense (it should be mean - 1.645*SD instead, which gives 95.7%). First red flag.
I need to report my LOB, LOD and LOQ to external stakeholders. No one is going to understand what it means when I say my LOB is 95.7% activity, LOD is 92.1% activity, and LOQ is 77.2% activity. They'll want the limits expressed in terms of analyte concentration.
But... to go from % activity to ng/mL, I have to interpolate on my calibration curve. But I got the data from 8 separate experiments. Am I supposed to average out the curves for the 8 experiments and use that to turn % activity into concentration?
I would really appreciate some help on this one. Everyone in my lab is pure analytical chemistry and they don't get the whole enzyme thing.