Bioequivalence and Bioavailability Forum

Hi sree1805,

please follow the Forum’s Policy in the future. THX.

❝ I want know why the pre-dose response is represented in the concentration data even though it is less than BLOQ.

This would be rather unusual. Generally values below the LLOQ are given in the analytical report as a string of characters (“<LLOQ”, “BLQ”, …). Reporting BLQs as zero would be stupid as well because in­formation is lost. BTW, it is good practice to use other non-numeric codes for missing samples (sub­ject didn’t show up at a late scheduled sampling time, problems in blood draw, vial broken, …), non-re­portable results (bad chromatography, unreliable repeats, …), etc.

❝ Is there any use of the response in the statistical parameters calculation.

If a value <LLOQ is reported (i.e., against all guide­lines extrapolated beyond the validated range of the method), PK-software will use it. The AUC will have a positive bias and you should expect a de­fi­ci­ency letter. On the contrary most (all?) PK-software treat non-numeric codes after an extravascular dose at t≤0 as zero sampled at t=0. Example of the beginning of a profile:

  t       C   
──────────────
-0.3333   BLQ   ⇐ 20 minutes prior to dosing
 0.25    5.00 
 0.5    15.0  
 1.0    20.0  
 1.5    17.0  

In all versions of WinNonlin you will get in EV NCA:

Dose time:           0.00
Calculation method:  Linear Trapezoidal Rule for for Increasing Values,
                     Log Trapezoidal Rule for Decreasing Values

   Time       Conc.   AUC    
-----------------------------
  0.0000 @   0.0000   0.0000 
  0.2500     5.000    0.6250 
  0.5000    15.00     3.125  
  1.000     20.00    11.88   
  1.500     17.00    21.10   

@) Note - the concentration at dose time was added for extrapolation purposes.

(my emphases)

For BLQs at t>0 you should have an SOP in place describing how these values will be handled in NCA. Most people treat BQLs occurring at 0 > t < t_max as zero and ignore any BLQ after t_max (i.e., truncate the AUC at the time point of the last quantifiable concentration). Some people set the first BLOQ after t_max to LLOQ/2. IMHO, that’s not a good idea (demonstrated to give biased parameter estimates in pop­u­la­tion PK). If t_i is the time point of the last measured concentration C_i and we (arbi­tra­rily!) set C_i+1 at t_i+1 to LLOQ/2 we would imply a t_1/2 of ln(2)/{[ln(C_i)–ln(LLOQ/2)]/(t_i+1–t_i)}, which strongly depends on ∆t and is not related to the PK of the drug. That’s not justified. Example (LLOQ = 5):

 t   C      t   C   
────────────────────
24  10     24  10   
36  (2.5)  48  (2.5)
t½   6         12   

Don’t set them to zero as well (or use WinNonlin’s AUCall).
Suggestion: Do not only refer to an SOP but state in the pro­to­col what you intend to do.

Since you asked for what guidelines say about it:

• FDA
  • Guidance for Industry – Bioanalytical Method Validation (2001): “Estimation of con­cen­tra­tion in unknown samples by extrapolation of standard curves below LLOQ […] is not re­com­mended.“
    Nothing about how BLQs should be reported.
    
  • BMV – Draft (2012): Lines 416 and 710: “Concentrations below the LLOQ should be re­ported as zeros.”
    That’s a bizarre idea; there was a consensus at the Crystal City conferences that post-dose con­cen­tra­tions <LLOQ should not be reported as zero. I don’t expect this statement to make it to the final version.
• EMA
  • GL on the Investigation of Bioequivalence (2010): Nothing.
    
  • Appendix IV to the BE-GL (2011): Nothing. Only report the number of pre-dose samples >5% of C_max for each formulation.
    
  • GL on bioanalytical method validation (2011): Nothing.
• Others
  See the Forum’s Policy. The link to Guidelines / Guidances is located on top of the page.