Figure 1.
Model structure and simulated ANC time-course.
(A) The structure of semi-mechanistic pharmacokinetic-pharmacodynamic Friberg model (from ref [26]) used to describe cytotoxic effect of drugs on proliferating neutrophils. Drug PK was linked to the semi-mechanistic PD model for neutrophil kinetics and unique PK-ANC profile for given drug at various schedules was generated. (B) Docetaxel PK-ANC simulation was carried out at an equivalent total dose to 100 mg/m2 every 21 days. The plots show the plasma PK profile (blue lines) and ANC upon treatment (red lines) for schedules of 28 on-0 off, 7 on-21 off, 4 on-18 off, 1 on-9 off, 4 on–24 off and 1 on-27off.
Figure 2.
Effect of schedule on docetaxel induced neutropenia.
Summary of population simulation of PK-ANC model for 1000 virtual individuals at two total dose levels of docetaxel shows constant dosing induces a less severe nadir than punctuated dosing. Population median ANC nadir for low (A) total dose and high (C) total dose as well as probability of grade 4 neutropenia (nadir <0.5×109/L) are shown for low (B) and high (D) dosing total dose levels. Each plotted captures schedules with the “days-on/days-off” format with number of consecutive days on in given treatment period shown on the X-axis and treatment period (sum of days on and days off) on the Y axis. Population estimation of median ANC nadir and probability of grade-4 neutropenia compared with common PK parameters across variety of schedules for low (blue) and high (red) total dose. (E) Cmax plotted against median ANC nadir for each of the schedules tested and shows weak correlation of −0.32 (R2 = 0.10, p<0.01). (F) Total cycle AUC over all schedules shows overall correlation of −0.64 (R2 = 0.41, p<0.01) with median ANC nadir, but the correlation is mainly driven by the differences in total dose as it loses ability to predict neutropenia at a fixed total dose level (R2<0.01 and p>0.75 at low and high total dose).
Table 1.
Equivalent total dose used in generation of simulated dosing schedules.
Figure 3.
Moving average of PK describes neutropenia.
The maximum of the moving average concentration over the dosing interval was examined for its ability to predict the simulated median ANC nadir. (A) Example of the 16-day moving average calculated from concentration profile from 1on-9off dosing schedule. (B) The correlation between maximal moving average concentration and ANC nadir was calculated for sliding windows of 1 day to 28 days for low dose (blue squares), high dose (red circles), and combination (black line). The maximum ability to predict neutropenia for the combined low and high total doses occurs when the moving average is calculated over 16 days. (C) The maximum of the moving average concentration, max(Cavg,16day), accounts for most of the variability in median ANC nadir across total dose and schedule (R2 = 0.86, p<0.01).
Figure 4.
Consistency of moving average of PK in describing severity of neutropenia across drugs.
The maximum of the moving average concentration over the dosing interval was examined for its consistency to predict neutropenia across different drugs. Topotecan induced neutropenia was simulated using linear drug effect model while in case of Etoposide, simulation was carried out using nonlinear drug effect (Emax) model. (A) The maximum of moving average concentration over 18 days, max(Cavg,18day), turns out to be a good predictor of Topotecan induced median ANC nadir(R2 = 0.79, p<0.01). While in case of Etoposide the maximum of moving average concentration below threshold over 16 days i.e IC50, max(Cavg,16day<IC50), shows maximum ability to predict Etoposide induced median ANC nadir across total dose and schedules(R2 = 0.84, p<0.01).
Figure 5.
Moving average PK describes rat neutrophil counts.
Rat neutrophil counts were measured response to the investigational PLK inhibitor TAK-960 and the relationship between plasma PK and neutrophil nadir was assessed. A) Absolute Neutrophil Count (ANC) normalized to control group plotted over the course of 21 days on a variety of dosing regimens. B) Normalized ANC nadir plotted versus total cycle AUC for each schedule tested. C) Normalized ANC nadir plotted versus the Cmax for each schedule tested. D) R2 plotted for Cavg,ndays for n-days from 1 to 14 days. E) The corresponding p-values of the Cavg,ndays correlation showing correlation. F) Normalized ANC nadir is highly correlated with max(Cavg,4days) with R2 = 0.70 (p<0.05), a better correlation than either Cmax or AUC.
Table 2.
Pharmacodynamic model parameters used in the simulation of ANC kinetics.