Supplementary Material to Physiologically Based Absorption Modelling to Explore the Impact of Food and Gastric pH Changes on the Pharmacokinetics of Alectinib
Supplementary Material 1. A brief summary of clinical studieswith alectinib used in this work.
absolute bioavailability study [1] / A study in six healthy male subjects aged between 22 and 43 years. After a standard breakfast, subjects received a single oral 600 mg dose of alectinib as capsules. 3.75 hours after oral dose administration, subjects were administered a 50 microgram intravenous microtracer dose of [14C]-alectinib as a 15 minute infusion.food effect & esomeprazole drug-drug interaction [2] / A study investigating the effect of food and the effect of esomeprazole on alectinib pharmacokinetics. 18 healthy subjects were dosed 600 mg as capsules following an overnight fast of at least 10 hours and then within 30 minutes after a high-fat, high-calorie meal.
24 healthy subjects received a single oral 600 mg dose after a standardized meal either alone or following 40 mg esomeprazole given once daily for 5 days.
Supplementary Material 2. Disposition Model Parameters
Parameters for the 3 compartment disposition model fit to plasma concentrations for a 15 minute intravenous infusion administration of 0.05 mg alectinib.
Parameter / Value / Units / CVCL / 0.452 / L/h/kg / 20.95%
Vc / 0.43 / L/kg / 32.94%
CL2 / 4.406 / L/h/kg / 3.21%
V2 / 3.322 / L/kg / 3.70%
CL3 / 0.426 / L/h/kg / 50.17%
V3 / 3.017 / L/kg / 3.85%
Fitting was performed using a 1/(Y+Yhat)2 weighting in the PKPlus module of GastroPlus™ using a body weight of 73 kg which was the mean body weight of the six subjects dosed. A three compartment model was preferred over 1 and 2 compartments based on both Akaike and Schwarz criteria. The total plasma clearance of alectinib was estimated as 33 L/hr. Assuming that all clearance is hepatic (CLh) and with a blood/plasma ratio (R) of 2.6 (Table 1) and a hepatic blood flow (Qh) of 90 L/hr the liver first pass effect was estimated as 14 % according to ((CLh/R)/Qh).Supplementary Material 3.
Simulated profile (line) with the preliminary absorption model for an oral dose of 600 mg. Observed plasma concentrations for the capsule formulation mean+/-sd are shown as symbols.
For these preliminary simulations, the fed state solubility measured in biorelevant medium (77 ug/mL at pH 5) was used as solubility input and regional changes due to pH were calculated with the Hendersson-Hasselbach relationship based on the measured pKa valueand the in silico estimated (ADMET Predictor) solubility factor (see below). (In addition to the measured pKa, 2 additional less influential pKa values predicted with ADMET Predictor were retained)
Supplementary Material 4.
Simulated profiles and regional absorption for an oral dose of 600 mg in the fed and fasted states using the preliminary model
FED STATEFASTED STATE
Supplementary Material 5.
Solubility versus pH profiles
The plot above shows the measured solubility of alectinib free base at a range of pH from 1 to 12 as filled diamonds. The dotted line is the profile fitted to these measurements which resulted in the fitted model shown below.The dashed line shows the solubility versus pH profile resulting from use of a single measured buffer solubility at pH 1.2, a measured pKa and an in silico estimated solubility factor of 7124.
The solid lines show the estimated lumenal solubility versus pH profiles obtained by using the biorelevant measurements of solubility a measured pKa and an in silico estimated solubility factor of 7124.
Supplementary Material 6.
Simulated profiles and regional absorption for an oral dose of 600 mg in the fed state considering measured solubility versus pH profile and a range of bile salt solubilization ratios derived from solubility measurements in biorelevant media.
Based on the FaSSIF measurement the solubilization ratio was 1.9*107 while using the FeSSIF solubility a 10-fold lower value of 1.9*106 was obtained. Simulations with this range of SR resulted in simulated profiles which encompassed the range of observed plasma concentrations and gave simulated bioavailability values ranging from 12% to 59%. Optimization of the SR to match the mean plasma exposure yielded a value of 7.7*106 with a simulated bioavailability of 35%./ Left is shown the regional absorption profile for the simulation with the optimized SR of 7.7*106 with a simulated bioavailability of 35%.
Supplementary Material 7.
Simulated oral profiles for a dose of 600 mg in the fed and fasted states considering measured solubility versus pH profile and a range of bile salt solubilization ratios derived from solubility measurements in biorelevant media.
Simulated AUC Fed/AUC Fasted = 8.8 / Simulated AUC Fed/AUC Fasted = 7.4 / Simulated AUC Fed/AUC Fasted = 7.1Solid lines show the simulations for the fed state while dotted lines are for the fasted state.
Supplementary Material 8.
Optimization of the solubilization ratio to match the clinical fasted and fed state profiles.
A optimization was performed for the solubilization ratio parameter, SR. This was optimized to match the fasted and fed state profiles separately. The optimization was performed in GastroPlus with an objective function of : 1/(Y+Yhat)^2 and with weights of 1 set on CMax, TMax and Concentration.
For the fasted state the default “Human - Physiological – Fasted” model was used and the optimized SR was 2.2* 107. This gave the results below:
For the fed state the default “Human - Physiological – Fed” model was used but with a lag time of 3 hrs added to match measured concentrations at early time points and Tmax (see below). The optimized SR was 5.8*106 .
Expanded plot of initial absorption phase showing lag time of 3 hrs before significant levels of drug appear in plasma. /The lag time shown above cannot be captured by the first order gastric emptying rate which is implemented in GastroPlus and we described it by using the “mixed multiple dose” feature to introduce a delay before the administration of the dose which is then handled via the usual first order gastric emptying.
Supplementary Material 9.
Modelling of the effect of the timing of alectinib dosing with respect to food.
The absorption of alectinib is sensitive to the luminal solubility and so modelling of the effect of the timing of alectinib dosing with respect to food must account for the solubility changes due to changed bile salt levels. As described by Jantratid[3] there are various time-dependent changes occurring in the GI tract after a meal which may affect drug solubility. These include changed intestinal pH, changed osmolality and changes in the composition of the luminal fluids in addition to bile salt levels. As media which capture these changes have been described [3] one possible procedure to modeling would be to first measure dissolution in these media and then incorporate these measured data into models for different post-prandial stages. An alternative, simpler procedure would be to build different ACAT models which capture the physiological changes in intestinal pH and bile salt levels and then use these to simulate the expected differences in absorption due to calculated changes in regional solubility based on the pH vs solubility profiles and bile salt solubilization ratio. This 2nd method is complicated by a discrepancy in the regional pH values used in the GastroPlus Fasted and Fed models and the values reported for the upper GI tract in the paper of Jantratid[3]. Thus in GastroPlus the Jejunum 1 pH in the fasted state 6.2 changes to 5.4 in the fed state while in Jantratid’s paper the early upper GI post-prandial pH is 6.5 which changes to 5.4 in the late post-prandial state. This change is difficult to reconcile because the direction of the change is different with a reduction in pH with food in the GastroPlus models while the pH is decreasing after a meal to a lower fasted state value in Jantratid’s paper.
The pH changes for the post-prandial time periods reported by Jantratid[3] can be translated into the associated effect on the aqueous solubility using our measured data for solubility vs pH as is done below.
Bile salts (uM) / pH / Aq. Sol (ug/mL)early / 10 / 6.5 / 0.022
middle / 7.5 / 5.8 / 0.039
late / 4.5 / 5.4 / 0.071
Looking at these changes in bile salt concentrations and at the changes in aqueous solubility due to pH, both follow a trend which is close to linear between the early, middle and late post-prandial stages. While the trend to reducing bile salt concentrations with time will affect alectinib luminal solubility negatively, the trend to reducing pH with time affects alectinib aqueous solubility positively. The overall effect as estimated from the clinical data is for a higher luminal solubility with food and overall we feel that the changing bile salts is the dominant factor and our assumption of an overall linear trend for the changing luminal solubility between the early fed and fasted states is supported.
At the time that this modelling was performed, the possibility to make additional solubility measurements was constrained and a simple approach was followed which assumed firstly that the differences in absorption would be mainly due to changes in luminal solubility and that these changes were driven by the changes inbile salt concentrations. This was achieved by modifying the regional solubility profiles using the GastroPlus .SPD (Solubility Data File) which allows a solubility to be set for each ACAT model compartment separately by linking the solubility value to the pH of the compartment as shown below (shaded cells are the values entered into the .SPD file)
ACAT model compartment / pH / Solubilitystomach / 4.9 / 0.02
duodenum / 5.4 / 0.07
jejunum1 / 5.41 / 0.048
jejunum2 / 6 / 0.03
ileum1 / 6.6 / 0.014
ileum2 / 6.9 / 0.009
ileum3 / 7.4 / 0.0025
caecum / 6.4 / 0.002
asc colon / 6.8 / 0.00056
Note that for compartments which share the same pH in the default ACAT model, such as duodenum and jejunum1, it was necessary to slightly modify the pH and save a custom ACAT model. When used in this way, and with the bile salt model turned off, the solubility is no longer estimated as a function of pH and bile salt concentrations but depends only on the value specified in the SPD file. To allow the creation of the regional solubility profiles for the post-prandial states a further optimization to the clinical data for the early and fasted states was performed. Since an automatic optimization of an SPD file is not possible in GastroPlus, this optimization was performed manually. The optimized solubility profiles and the associated simulated plasma profiles are shown below.
For the fasted with the default “Human - Physiological – Fasted” model was used and gave the results below:
For the fed state the default “Human - Physiological – Fed” model was used and gave the results below:
References
1.Morcos, P.N., L. Yu, and K. Nieforth, Absorption, distribution, metabolism, and excretion (ADME) of the ALK inhibitor alectinib: results from an absolute bioavailability/mass balance study in healthy subjects. ; . Clin Pharmacol Ther, 2016. 99: p. Abstract PI-118.
2.Morcos, P.N., et al., Effect Of Food And The Proton Pump Inhibitor (PPI) Esomeprazole On The Pharmacokinetics (PK) Of Alectinib, A Highly Selective ALK Inhibitor, In Healthy Subjects. Clin Pharmacol Ther, 2016. 99: p. Abstract PI-120.
3.Jantratid, E., et al., Dissolution Media Simulating Conditions in the Proximal Human Gastrointestinal Tract: An Update. Pharmaceutical Research, 2008. 25(7): p. 1663.