Zake2021 - PBPK model of metformin in mice and humans
December 2021, Model of the Month by Krishna Kumar Tiwari
Original model - BIOMD0000001027,BIOMD0000001028,BIOMD0000001029,BIOMD0000001039
Diabetes mellitus, commonly known as Type 2 diabetes (T2D), is an aggregative metabolic disorder where blood glucose levels increase higher than normal levels, due to improper insulin release or an inadequate response of cells to insulin. Metformin, a biguanide derivative, is a FDA (U.S. Food and Drug Administration) approved drug, commonly prescribed to manage T2D [1–3]. Despite broader use, there is insufficient data for metformin concentration-time profiles in different human tissues which restricts the pharmacokinetic apprehension and limits the development of patient-specific precision therapies. Physiology based pharmacokinetic model (PBPK) development is widely used in pharmaceutical companies to study drug distribution at different organs/tissues [4-5]. Using such an approach, Zake et al.  has developed an ordinary differential equation (ODE) based PBPK model of metformin in mice which along with in-vitro data was extrapolated to create a human PBPK model that can be personalized by customising measurable values (such as tissue volume, blood flow) and time-course profile of metformin concentration in blood and urine after a single dose of metformin. This model can be potentially used as a decision support tool for patient-specific therapy development.
This ODE based PBPK model included 21 tissues and body fluid compartments and can simulate metformin concentration in the stomach, small intestine, liver, kidney, heart, skeletal muscle adipose, and brain depending on the body weight, dose, and administration regimen (Figure 1). Metformin doses ranging from 500 to 1500 mg given as per-oral (PO) or intravenous (IV) have been analyzed through simulation of PBPK model in a 70kg adult human. The mice PBPK model consisted of 20 compartments lacking 1 compartment i.e. red blood cells. The Mice model was simulated with a single PO dose and single IV dose of 50mg/kg of Metformin. Human models were simulated with a single PO dose and multiple PO doses with eight administration in 12 h intervals.
Figure 1. PBPK model of metformin consisting various organs/tissues compartments. Figure taken from Zake et al, 2021 .
The concentration-time profile obtained from the model simulation was compared with the experimentally determined data. The concentration-time profile of 50 mg/kg metformin in various compartments in mice can be observed in Figure 2.
Figure 2. Simulation generated metformin concentration-time profile in major compartment of Metformin action following a single PO dose of 50 mg/kg in mice.
Most of the concentration-time profile in mice matched with experimental data except few tissues like plasma, heart, liver, kidney and brain. In the portal vein compartment, Cmax and AUC24 were lower than the experimental data by 22% and 34% respectively in the model simulation. Differences were also significant in the intestinal compartment, where the AUC24 and T1/2 were 22% and 35% lower respectively. The primary reason for the difference between model and experimental data is that the parameters concerning active excretion of metformin parameters were estimated by simultaneously running simulation for the PO and IV. Similarly, in brain tissue, the difference indicates a complex transport mechanism at the blood-brain barrier that is not covered by the model.
The human PBPK model of metformin got a good fit to the experimental data and reflects the experimentally similar concentration-time profile with varying doses (500,1000,1500mg) and multiple dosing regimens (Figure 3).
Figure 3. Simulation generated metformin concentration-time profile in the major compartment of metformin hydrochloride action following multiple PO doses of 500 mg, 1000 mg and 1500 mg in humans.
Varying doses of metformin, the model simulation results were validated against separate data published in previous publications (details in ). On single PO doses of 500mg,1000mg and 1500mg of metformin hydrochloride, the simulation predicted that the intestine and kidney tissue reached the highest concentration. Interestingly, kidney accumulation of metformin (840nmol/ml for 500mg dose) was found to be greater than that of the small intestine, known to accumulate metformin. Model simulation also predicted a dose-dependent absorption of metformin. After 500mg dose, 1000 mg and 1500mg of metformin dose, 50.7% (197.1mg), 46.3% (362mg) and 44.1% (517 mg) were excreted in the urine respectively. That reflected a dose-dependent absorption with a higher fraction of the drug getting absorbed at low doses. The model also predicted from a multi-dosing regimen that in a steady-state metformin concentrations was reached after the third dose (around24 hours) for all tissues except red blood cells which showed the metformin accumulation. The overall maximal concentration in nearly all compartments was 2X higher under multiple dosing regimens. Also under multiple dosing regime, the maximal concentration in plasma were higher for a single PO dose of 500mg (plasma: 6.1nmol/mL, Cmax: 6.9nmol/mL) and 1000mg (plasma: 11.2nmol/mL, Cmax=12.8nmol/mL) of metformin hydrochloride. These results were comparable to the observed experimental data.
In this study, a simplified PBPK model for metformin in humans and mice has been developed. Model is used to study the metformin concentration-time profile on different tissues by applying some parameters from mice and other human tissue data. Using this model, one can predict the dynamics of metformin distribution in humans based on the information of the physiological properties of humans, the physicochemical properties of metformin and its absorption/excretion. Also, the model can be applied to a specific patient by entering the proportion of tissue and parameter of blood flow and using single/multiple dose experiment data to record metformin dynamics in blood and urine. The resultant concentration-time profile leads to a patient-specific parameterized model that can be potentially used as a decision support tool to develop precision therapy.
1. L. B. A. Rojas and M. B. Gomes, ‘Metformin: an old but still the best treatment for type 2 diabetes’, Diabetol. Metab. Syndr., vol. 5, no. 1, p. 6, Feb. 2013, doi: 10.1186/1758-5996-5-6.
2. H. Nasri and M. Rafieian-Kopaei, ‘Metformin: Current knowledge’, J. Res. Med. Sci. Off. J. Isfahan Univ. Med. Sci., vol. 19, no. 7, pp. 658–664, Jul. 2014.
3. C. J. Bailey, ‘Metformin: historical overview’, Diabetologia, vol. 60, no. 9, pp. 1566–1576, Sep. 2017, doi: 10.1007/s00125-017-4318-z.
4. H. M. Jones, I. B. Gardner, and K. J. Watson, ‘Modelling and PBPK Simulation in Drug Discovery’, AAPS J., vol. 11, no. 1, pp. 155–166, Mar. 2009, doi: 10.1208/s12248-009-9088-1.
5. H. Jones and K. Rowland-Yeo, ‘Basic Concepts in Physiologically Based Pharmacokinetic Modeling in Drug Discovery and Development’, CPT Pharmacomet. Syst. Pharmacol., vol. 2, no. 8, p. e63, Aug. 2013, doi: 10.1038/psp.2013.41.
6. D. M. Zake, J. Kurlovics, L. Zaharenko, V. Komasilovs, J. Klovins, and E. Stalidzans, ‘Physiologically based metformin pharmacokinetics model of mice and scale-up to humans for the estimation of concentrations in various tissues’, PloS One, vol. 16, no. 4, p. e0249594, 2021, doi: 10.1371/journal.pone.0249594.