Bioequivalence and Bioavailability Forum

❝ So the question is: how to calculate the AUC part from tlast to infinity in this case with Winnonlin (what kind of points to collect for lambda z calculation & hl)?

• Generally it is not a good idea to include C_max/t_max in the estimation of λ_z, since absorption is not complete. That’s the reason why Phoenix’ default “Best Fit” would not give you an estimate (only two data points >t_max). On the other hand WinNonlin ≤5.3 would happily give you an R²_adj of ~0.999 – in log-scale the last part of the profile looks nice.
  
  
  
  
• Some ideas:
  • Overrule Phoenix’ default and use the last three data points. You could justify that by the almost identical λ_z obtained from the last two.
    
  • According to the Russian GL the main PK metric for extent of absorption is AUC_t (if AUC_t > 80% AUC_∞). If you cannot demonstrate that according to the protocol (e.g., three data points after t_max required), consider excluding the subject from the com­parison of AUC_t. That should no hurt in terms of power because the critical PK metric is C_max (likely higher CV).