Clinical Pharmacokinetics of Oral Semaglutide: Analyses of Data from Clinical Pharmacology Trials

Abstract

Objective: The absorption, distribution and elimination of oral semaglutide, the first oral glucagon-like peptide-1 receptor agonist for treating type 2 diabetes, was investigated using a population pharmacokinetic model based on data from clinical pharmacology trials.

Methods: A previously developed, two-compartment pharmacokinetic model, based on subcutaneous and intravenous semaglutide, was extended to include data from six oral semaglutide trials conducted in either healthy volunteers or subjects with renal or hepatic impairment. Five trials employed multiple doses of oral semaglutide (5-10 mg) and one was a single-dose (10 mg) trial. In a separate analysis, the model was re-estimated using data from a trial in subjects with type 2 diabetes.

Results: The model accurately described concentration profiles across trials. Post-dose fasting time, co-ingestion of a large water volume, and body weight were the most important covariates affecting semaglutide exposure. Bioavailability was 0.8% when oral semaglutide was dosed using the recommended dosing conditions (30 min post-dose fasting time, administered with ≤ 120 mL of water), increasing with a longer post-dose fasting time and decreasing with higher water volume. Within-subject variability in bioavailability was 137%, which with once-daily dosing and a long half-life translates into 33% within-subject variability in steady-state exposure. There was no significant difference in oral bioavailability of semaglutide in healthy subjects and subjects with type 2 diabetes.

Conclusions: The updated model provided a general characterisation of semaglutide pharmacokinetics following oral, subcutaneous and intravenous administration in healthy subjects and subjects with type 2 diabetes. Within-individual variation of oral bioavailability was relatively high, but reduced considerably at steady state. CLINICALTRIALS.

Gov identifiers: NCT01572753, NCT01619345, NCT02014259, NCT02016911, NCT02249871, NCT02172313, NCT02877355.

Conflict of interest statement

Rune V. Overgaard, Andrea Navarria, Steen H. Ingwersen, Tine A. Bækdal and Rasmus Juul Kildemoes are employees of, and shareholders in, Novo Nordisk A/S.

© 2021. The Author(s).

Figures

Fig. 1

Population pharmacokinetic model. Dashed boxes represent model components included in the previously developed subcutaneous/intravenous model. BW body weight, CL clearance, F oral bioavailability, ka oral absorption rate constant, Q inter-compartmental clearance, T2D type 2 diabetes, Vc central volume, Vp peripheral volume

Fig. 2

Observed and model-predicted mean trough concentrations (with tail) across trials (a) and mean concentration profiles following last dose (b), both on an arithmetic scale; observed and model-predicted geometric mean concentration profiles and 90% range of trough concentrations since the first dose during 10 days of dosing (c) and following the last dose (d), both on a log scale. a, b Data points are observed geometric mean concentrations with 95% confidence intervals on an arithmetic scale. Lines are population predictions from the final model. c, d Data points are observed geometric mean concentrations with 90% ranges on a semi-logarithmic scale. Shaded areas are model-predicted geometric mean and 90% prediction intervals based on approximately 10,000 simulated profiles using the final population pharmacokinetic model. Data are from the following trials: dosing conditions (NCT01572753) [15], renal (NCT02014259) [17], hepatic (NCT02016911) [18], drug–drug interaction omeprazole (NCT02249871) [19] and food effect (NCT02172313) [20]

Fig. 3

Observed and model-predicted trough concentration–time profiles by trial and key covariates in each trial: post-dose fasting time in the dosing conditions trial (a), water volume in the scintigraphy trial (b), renal impairment in the renal trial (c), hepatic impairment in the hepatic trial (d) and during repetitive dosing (a, c, d) and following a single dose (b); observed concentrations and model-predicted full semaglutide pharmacokinetic profiles by trial for the drug–drug interaction (DDI) omeprazole and food-effect trials (e, f). Data points with error bars are observed geometric mean concentrations with 95% confidence intervals. Lines are population predictions. Data are from the following trials: dosing conditions (NCT01572753) [15], scintigraphy (NCT01619345) [16], renal (NCT02014259) [17], hepatic (NCT02016911) [18], DDI omeprazole (NCT02249871) [19] and food effect (NCT02172313) [20]

Fig. 4

Absolute bioavailability vs post-dose fasting time by water volume. Data points are means with 95% confidence intervals of individual model-derived parameter estimates. Lines are model-derived relationships to the fasting time and shaded areas are 95% confidence intervals obtained by bootstrapping. The model does not discriminate between 50 and 120 mL, and therefore the modelled lines are identical for these two water volumes. Data are from the following trials: dosing conditions (NCT01572753) [15], scintigraphy (NCT01619345) [16], renal (NCT02014259) [17], hepatic (NCT02016911) [18], drug–drug interaction omeprazole (NCT02249871) [19] and food effect (NCT02172313) [20]

Fig. 5

Concentration profiles for deviations from the recommended dosing regimen and dosing conditions for once-daily dosing of 14 mg of semaglutide: one missed dose day 5 (a); two doses day 5 (b); change of post-dose fasting to 15 min from day 5 (c); change of post-dose fasting to 15 min throughout (d); change of post-dose fasting to 120 min from day 5 (e); change of post-dose fasting to 120 min throughout (f); change to administration with 240 mL of water from day 5 (g); and change to administration with 240 mL of water throughout (h). Data are population predictions with 90% prediction intervals. Illustrated are recommended dosing conditions (dark blue lines) and deviations from the dosing regimen and dosing conditions (light blue and purple lines). Recommended dosing conditions: administration in a fasting state with up to 120 mL of water and post-dose fasting time of at least 30 min

Fig. 6

Distribution of variability components for the oral bioavailability of semaglutide (a), area under the curve (AUC) values following a single oral dose (b), and average concentration (Cavg) at steady-state dosing of 10 mg of semaglutide once daily (c). Data are model-derived bioavailabilities and exposures based on simulations of a reference subject profile using the final population pharmacokinetic model. CV coefficient of variation