Dear Simon,
I came across a potential pitfall in NCA, please check. Real case: In a multiple dose, every 24h dosing study no samples were taken at time zero (predose) and at time 24h.
For the first dose Phoenix sets correctly predose C(0) = 0. BUT, for extravascular & Infusion, “…for steady state C0 conc is set to the minimum observed in the dosing interval…” In our case, a quickly cleared drug, there are high conc. at the last time point, t=8h, but in reality at t=24h (where we have no sample) conc. are very low. PHX sets, by default, if Tau = 24 is entered, for steady state C(0h) = C(8h), which is wrong. The only workaround I have found to avoid that, is not to add a value for Tau (in which case C(0) is set to 0, which reflects the low C(24) better than the high C(8).
In my opinion, in this case C(0) should be set per default to C(24) if dosing interval is 24h (i.e., set to C(tau) ). In addition, the user should have the possibility to set C(0) to another value if e.g. the C24 are not available since no sample was taken. In this case, if justified, C(24) might be predicted and set as C(0).
What do you think?
Dieter, one of the issues of NCA as an analysis approach is you only describe what you observe, so I think there is an argument for suggesting the study sample schedule was not optimally designed if this parameter should be generated.
That said we are working on extra options for Cmin (or Ctau) to be presented to the user as QC3885, since although the current implementation of Cmin;
Tmin - Time of minimum OBSERVED concentration sampled during a dosing interval.
Cmin - Minimum OBSERVED concentration between dose time and dose time + Tau (i.e. at Tmin).
Is the classic definition e.g. Encyclopedia of Biopharmaceutical Statistics (2003) etc. (imagine a formulation/drug with slow absopbtion and slow clearance: the trough conc might be one or two hours after dosing). We do see requests for Ctau to be reported and will investigate how best to present this option in NCA GUI inthe future; however your request goes a little further to request an option of a predicted conc at Ctau when no sample was actually taken.
I’ll add this to the request however i can see this too having it’s own problems as somtimes the Lz used to predict this conc will be unavailable or unreliable, especially as you say the profile is quite sparsely sampled.
Simon