Drug effect calculation

NCA for drug effect data requires the following input data variables:

Dependent variable, i.e., response or effect values (continuous scale)

Independent variable, generally the time of each observation. For non-time-based data, such as concentration-effect data, take care to interpret output parameters such as “Tmin” (independent variable value corresponding to minimum dependent variable value) accordingly.

It also requires the following constants:

Dose time (relative to observation times), entered in the Dosing worksheet

Baseline response value, entered in the Therapeutic Response tab of the Model Properties or pulled from a Dosing worksheet as described below

Threshold response value (optional), entered in the Therapeutic Response tab of the NCA Diagram

If the user does not enter a baseline response value, the baseline is assumed to be zero, with the following exception. When working from a Certara Integral study that includes a Dosing worksheet, if no baseline is provided by the user, Phoenix will use the response value at dose time as baseline, and if the dataset does not include a response at dose time, the user will be required to supply the baseline value.

If there is no response value at dose time, or if dose time is not given, see “Data checking and pre-treatment” for the insertion of the point (dosetime, baseline).

Instead of Lambda Z calculations, as computed in NCA PK models, the NCA PD model can calculate the slope of the time-effect curve for specific data ranges in each profile. Unlike Lambda Z, these slopes are reported as their actual value rather than their negative. See “Lambda Z or Slope Estimation settings” for more information.

Like other NCA models, the PD model can compute partial areas under the curve, but only within the range of the data (no extrapolation is done). If a start or end time does not coincide with an observed data point, then interpolation is done to estimate the corresponding Y, following the equations and rules described under “Partial area calculation”.

For PD data, the default and recommended method for AUC calculation is the Linear Trapezoidal with Linear Interpolation (set as described under “Options tab”.) Use caution with log trapezoidal since areas are calculated both under the curve and above the curve. If the log trapezoidal rule is appropriate for the area under a curve, then it would underestimate the area over the curve. For this reason, Phoenix uses the linear trapezoidal rule for area where the curve is below baseline when computing AUC_Below_B, and similarly for threshold and AUC_Below_T.