Bioequivalence and Bioavailability Forum

Dear Charl!

❝ Can I report concentrations that has 20% less values than the Lower Limit Of Quantitation in my final study report?

No, because they are below the ‘Lower Limit of Quantitation’.

❝ once we already accepts these values in our validation as a part of accuracy measures...!

That’s another story.
These values were back-calculated spiked (known) samples. Their purpose was to establish an LLOQ of accepted accuracy and precision.
Actually it’s a two step procedure:

• you assume an LLOQ:
  setting up the calibration curve, the QCs, the LLOQ-samples accordingly
  
• you run your validation: if at the assumed LLOQ you find:
  inaccuracy (bias) ≤20% and precision ≤20%, then your LLOQ is established

If you feel that your measurements at LLOQ were too accurate and/or precise you are free to repeat this procedure.
But once you have established the LLOQ, this represents a definite number you will have to follow for your unknown samples (fuzzy logic is not applicable).

To stay with your example: if we measure an unknown at exactly LLOQ, we know that it’s ‘true’ concentration may be even <80% of LLOQ due to the accepted inaccuracy/precision.
Leaving out any statistical issues, if you suggest reporting a value at 80% of LLOQ, the true value may be <0.8×0.8 = <64% of LLOQ (whereas you assume a ‘true’ value of >LLOQ, which was only brought down by chance).
And so on and so forth.

At least this is the ‘message from the frontline’ in BA/BE (in other words, the only way regulators will accept your study).

I know that there’s another point of view represented by pharmacokineticists (especially population PK modelers). They will fight for any single data point above LOD, argueing that they are able to incorporate measurement error into the model.
Besides statistical problems (negative concentrations are impossible, unproven distributional assumptions, etc.) there remains one unresolved issue:
We simply have no idea about the bias of values between LOD and LLOQ.