EP1456646A2

EP1456646A2 - Computer system and method for calculating adme properties

Info

Publication number: EP1456646A2
Application number: EP02783091A
Authority: EP
Inventors: Walter Schmitt; Jörg Keldenich; Jannis Batoulis; Michael Beck; Roger-Michael Brunne; Thorsten Bürger; Thorsten Pötter; Felix Reichel; Stefan Willmann
Original assignee: Bayer Technology Services GmbH
Current assignee: Bayer AG
Priority date: 2001-12-07
Filing date: 2002-11-25
Publication date: 2004-09-15
Also published as: CA2469388A1; US20050119832A1; US7765092B2; WO2003048720A2; JP2005512176A; CN1659435A; RU2004120787A; DE10160270A1; WO2003048720A3; AU2002346846A1

Abstract

The invention relates to a computer system and a method for calculating an ADME property of a substance, which comprises the following steps: a. input of molecular properties of said substance into a biophysical model, whereby said biophysical model describes a correlation between the molecular properties and the ADME property and b. output of the ADME property.

Description

Computer system and method for calculating ADME properties

The invention relates to a computer system and a method for calculating ADME properties of a substance, in particular for a substance with a pharmacological effect or a substance for crop protection applications.

The effectiveness of active substances is determined by their interaction with the molecular, biological target, as well as by the concentration at the target location. Both parameters are usually determined by different molecular parameters and can therefore be optimized independently of one another within certain limits. While the intrinsic biochemical effect can be determined in in-vitro tests at a very early stage of research for large numbers of substances, the concentration in the site of action can only be investigated by experiments on the whole organism (animal, plant or fungus). The latter means that due to the complexity of the tests, the information can only be carried out in late research stages and is therefore not available for the first optimization cycles.

In recent years, this has led to pharmaceutical and

The crop protection industry was looking for alternative ways to obtain information about the ADME behavior (absorption, distribution, metabolism, excretion) of active substances at an early stage. Since large parts of the ADME behavior are influenced by easily measurable physicochemical properties or quantities that can be calculated from the chemical structure, this has meanwhile

Approach established to experimentally determine or calculate such quantities with high throughput [H. van de Waterbeemd, D.A. Smith, K. Beaumont, D.K. Walker, J. Med. Chem. 44, 1-21 (2001)].

Typical properties that are usually taken into account are e.g. B.

Lipophilicity, water solubility, permeabilities over artificial membranes or Cell layers, molecular weight and number of certain structural features such as hydrogen donors and acceptors. The assessment of the substances is then usually carried out by observing certain limits, which are usually obtained from empirical values or from the statistical distribution of properties for commercial products [CA Lipinski, F. Lombardo, BW Dominy, PJ Feeney, Adv.

Drug Delivery Rev. 23, 3-25 (1997) and C. M. Tice, Pest Management Sei. 57, 3-16 (2001)].

The disadvantage of this method is that rigid limits are considered for individual properties that are only indirectly relevant. However, the actually important ADME properties generally depend on several of these variables at the same time, which means that the tolerable limits of the individual variables also depend on their value, and therefore absolute limit values can only be roughly defined.

A computer program product for at least partially automatic calculation of molecular sizes based on a substance library is known from US Pat. No. 5,901,069. However, this method does not allow the calculation of ADME properties.

From US-A-5 680 590 is a simulation system for the calculation of physiological

Data regarding pharmacokinetic and pharmacodynamic parameters are known. A disadvantage of this simulation system is that the calculation is not based directly on the molecular properties of a substance to be assessed. This model is only suitable for educational purposes and does not allow the calculation of the ADME properties of a new substance.

The invention is based on the object of providing an improved method for calculating an ADME property of a substance and a corresponding computer program and computer system. The object on which the invention is based is achieved in each case with the features of the independent patent claims.

Preferred embodiments of the invention are specified in the dependent patent claims.

The invention can advantageously be used both for already synthesized substances and for “virtual structures” of substance libraries. According to the invention, biophysical relationships between the molecular and physicochemical properties of the substances and the ADME properties under consideration are used to calculate the ADME properties The mathematical description is then made either using an analytical formula or, in the case of more complex relationships, using a numerical simulation.

A particular advantage of the invention is that the calculation of ADME

Properties are based directly on the molecular properties of the substance to be examined. For this purpose, a biophysical model is used, in which the molecular properties are entered as input variables. The biophysical model establishes a connection between these molecular properties and the ADME property or properties. In this way, one obtains an immediate statement about the ADME properties and not just about correlated surrogate sizes, as is the case in the prior art.

According to a preferred embodiment of the invention, lipophilicity, binding to proteins and molecular size are used as molecular properties. For example, this allows modeling the absorption of active substances in a human, animal or plant organism. Lipophilicity is determined by distribution coefficients between a lipoid phase (eg octanol, cooking oil, hexane, phospholipid membranes) and Described water. For example, the molecular weight or the molar volume can serve as a measure of the molecular size.

Since this is usually determined by a permeation process, and the permeability is known to depend on the lipophilicity and the size of the permeating

Depending on the molecule, conclusions can be drawn about the absorption rate from these properties. However, since the dependencies are opposite, higher molecular weights can be tolerated, for example, with increasing lipophilicity in order to achieve the same absorption rates. However, this fact cannot be adequately taken into account in the prior art by means of fixed limit values for optimal lipophilicity and molecular size.

On the other hand, the invention allows direct conclusions to be drawn from the molecular properties of the ADME properties, so that a calculation with improved accuracy is possible.

According to a preferred embodiment of the invention, the biophysical model is a physiologically based pharmacokinetic model. For the investigation of the ADME properties of a substance, for example in the human body, the model at least includes those for

Examination of major organs, such as B. the lungs, liver and kidneys and blood circulation. The various sub-models of the organs are linked to one another by mass conservation equations.

According to a further preferred embodiment of the invention, the

Mass conservation equations expressed in the form of a system of differential equations, the input variables of the system of differential equations being obtained directly from the calculated molecular properties.

According to a further preferred embodiment of the invention, the

Enter the chemical structure of a substance to be investigated into a data bank. For this purpose, the chemical structure can also be represented, for example, in the form of a descriptor or a so-called fingerprint.

The chemical structure can be entered decentrally from a client computer. The client computer is located, for example, directly at a chemist's workplace for entering new chemical structures for which the ADME properties are to be predetermined.

This database is then queried cyclically, for example by a server computer. Once the new chemical structure is entered by the

Server computer has been determined, the molecular properties of this chemical structure are automatically calculated by a corresponding program call.

Alternatively or additionally, the server computer can access a further database in which experimentally determined molecular properties of the substance are stored. After determining the molecular properties, these are incorporated into the biophysical model by another. Entered the program call so that the desired ADME properties are calculated automatically.

This process is preferably carried out repeatedly for different substances, for example the same chemical base body. The results of the ADME calculations are then output in a structured form, for example sorted according to the value of a specific ADME property or sorted according to a weighted index of ADME properties.

According to a further preferred embodiment of the invention, a statistical method for calculating the molecular properties from the chemical structure is used, such as a QSAR or HQSAR method, or a method based on a neural network. Such methods for determining molecular properties - for example from a Descriptor of the chemical structure - are known per se from the prior art.

For pharmaceutical applications, the invention in particular enables the following ADME properties to be calculated:

fraction of a dose absorbed after oral application (fraction absorbed),

- concentration-time curves in the portal vein after oral application,

- free fraction in plasma,

- organ / plasma partition coefficient,

- Blood / plasma partition coefficient

- Concentration-time curves in blood plasma and in organs after oral or intravenous administration.

- Common pharmacokinetic parameters derived from the concentration-time curves, e.g. Maximum concentration, time of max. Concentration, volume of distribution, half-lives.

For crop protection applications, the invention in particular enables the calculation of the following ADME properties:

- Key figure for the speed of uptake in the sheet after spray application,

- Indicator of speed of admission into an insect via the cuticle, - Key figure for the speed of distribution in the plant after foliar application (phloem mobility),

- Key figure for the speed of distribution in the plant after root application (xylem mobility),

- Distribution of concentration between the organs of an insect.

The calculated data are either stored directly in a database or output in a table. The data available in this way form the basis for a

Ranking of the considered substances or structures and the selection of the candidates for further optimization based on this ranking.

For rahking, the properties that are decisive for the intended application of the active ingredient to be optimized must first be selected. The

Evaluation is carried out manually using suitable data evaluation and visualization software. The substances or structures that are in the optimal distribution of properties are searched for by ranking (e.g. the 10 substances with the highest absorption after oral application). If more than one property is taken into account in the ranking, an index can be calculated that includes this (in the simplest case, the sum of the values of all sizes), and the properties can be weighted according to relevance.

As an alternative to manual evaluation, there is also the option of using output masks for the data that perform an automatic evaluation by checking for each structure when using the data whether certain values or combinations of values (indices) are optimal for the application under consideration Range (display, e.g. over traffic light colors). The project-specific. Rules are then stored in the evaluation mask (eg in a spreadsheet program). The preferred embodiments of the invention are explained in more detail below with reference to the drawings. Show it:

FIG. 1 shows a flow diagram for calculating ADME properties from the molecular properties of a substance,

Figure 2 shows an embodiment of a biophysical model for

Establishing a connection between the molecular properties and the ADME properties,

FIG. 3 shows a table for calculating organ / blood distribution coefficients for the model in FIG. 2,

Figure 4 shows an embodiment of a computer system according to the invention.

FIG. 1 shows a flow chart for the calculation of ADME properties of a substance. In step 1, the molecular property of a substance is determined. This can be done experimentally if the substance has already been synthesized. Furthermore, the molecular properties can also be determined by means of a calculation.

For this purpose, methods are known per se from the prior art which make it possible to calculate molecular properties from the chemical structure. Examples of such methods are QSAR, HQSAR and neural networks. Descriptors or serve as the input variable for such calculation methods

Fingerprints of the chemical structure of the substance to be examined.

Furthermore, in step 1 it is possible to access both experimentally determined molecular properties and molecular properties of the substance determined by calculation. This way you can Supplement experimental methods with the calculation methods for determining the molecular properties.

The molecular properties of the substance determined in step 1 are entered in a biophysical model in step 2. The biophysical

Model establishes a connection between the molecular sizes and the ADME properties of interest. This can be, for example, a physiologically based pharmaco-kinetic model. An embodiment of such a biophysical model is explained in more detail below with reference to FIGS. 2 and 3.

In step 3, the ADME properties are output from the biophysical model. It is particularly advantageous here that the ADME properties are determined directly from the molecular properties, without the interposition of surrogate variables that require interpretation. This allows a fully automatic process for the calculation of the ADME properties.

FIG. 2 shows a biophysical model 4 of a warm-blooded animal, for example a human. The biophysical model 4 contains a number of partial

Models for those organs that are most relevant for the distribution of the substance in the body. In the example in FIG. 2, these are partial model 5 for the lungs, partial model 6 for the liver, partial model 7 for the kidneys and partial models 8 for various other organs X.

The partial models 5, 1, 8 are "interconnected" by venous blood 9 and arterial blood 10. The venous blood 9 enters the partial model 5 for the lungs, where it is converted into arterial blood 10. The arterial Blood 10 then passes into the further sub-models 6, 1 and 8, from where it emerges again as venous blood 9. The various sub-models of the organs are thus rendered venous

Blood 9 and arterial blood 10 "connected in parallel". Furthermore, the biophysical model 4 contains an excretion model 11 for the sub-models 6 and 7, that is to say for the liver and the kidneys.

In the considered example of FIG. 2, the biophysical model is provided for modeling the temporal concentration distribution of a dose of the substance to be examined. The dose is supplied via the venous blood 9, that is to say, for example via the portal vein after oral administration or by injection into a vein.

A corresponding table of empirical values can be accessed for the flow rate Qu _me of venous blood 9 through the lungs 5. Also, for the flow rate of arterial blood 10 through the other body part models 6, 7, 8 corresponding empirical values for the flow rates of O - he _b, _n e and _r Que Qx be accessed.

The quantities C _x are the concentration of the substance in the organ X in question at a specific point in time. The parameter K _x denotes the distribution coefficient of the substance between the blood and the organ X in the state of equilibrium. The parameters CL e er and CLNieren denote the intrinsic excretion of the liver and kidneys.

Based on the biophysical model 4, a mass equilibrium relationship can be established for each organ X by means of a differential equation of the following form:

dC C

V - = 0 - C - O • ~ -

^Vχ dt ^{üx ar ü *} K _x

V _x = volume of the organ X Cx = concentration of the substance in the organ X Qx = flow rate of blood through the organ X f _u = proportion of the substance that is not bound in plasma

C _ar = concentration of the substance that the organ reaches via the arterial blood K _x - coefficient of distribution of the substance between the blood and organ X in the state of equilibrium

The parameter f _{u is} calculated from the reciprocal of the distribution coefficient of the substance in the equilibrium between blood plasma and water.

The corresponding differential equations for the liver and kidneys contain an additional term that describes the excretion of the substance. Such a differential equation is given below for the kidneys; the same applies to the liver:

Vj = volume of the kidneys

C i = concentration of the substance in the kidneys

Q _k i = flow rate of blood through the kidneys f _u = proportion of the substance that is not bound in plasma

C _ar ⁼ concentration of the substance that reaches the kidneys through the arterial blood

Kw = partition coefficient of the substance between blood and kidneys in the

Equilibrium

CLw ⁼ intrinsic excretion of the kidneys

From this, an equation for the venous blood can be drawn up by

Summation of all "output" concentrations of the various organs and the

CL _k . intravenously added dose of the substance. The term depends on Lipophilicity value of the substance can also be determined from a molecular property.

The differential equation for the lungs creates a connection between the venous and arterial blood. The corresponding equations are given below:

~ = ∑Q _x -Q, - _vs + can. l (t)

V ^dClu - Q. r -O. ^ i dt K _lυ

V _ve = volume of venous blood

C _ve ⁼ concentration of the substance in the venous blood supplied to the lungs

Q _x = flow rates of blood through the organ X f _u = proportion of the substance that is not bound in plasma C _x = concentration of the substance in the organ X

K _x = distribution coefficient of the substance between blood and organ X in the state of equilibrium

Qι _u = blood flow through the lungs = Σ of all blood flows

Vi _u = lung volume Ciu = concentration of the substance in the lungs

Ki _u = distribution coefficient of the substance in the balance between blood and lungs

Dose = dose of the substance that is added to the venous blood over time t according to an input function I (t).

To solve the resulting differential equation system of the biophysical model 4, knowledge of the different distribution coefficients in the equilibrium state is necessary. These can be from the molecular properties of the substance, which have previously been determined or calculated experimentally.

The distribution coefficients for a substance in the equilibrium between fat and water (K _fat ) and between protein and water (Kp _ro t _e in) can be determined by calculation. The distribution coefficients mentioned are determined either by calculation or experimentally; There are methods known per se from the prior art for the computational as well as the experimental determination.

In addition, the organ compositions of water, fat and protein are used for the calculation. These can be determined from the table in FIG. 3. The table in FIG. 3 shows the total mass of the organ in question and the absolute proportions in grams of water, fat and protein in the organ for various organs.

From this, Fwasser, Fpett and mp _ro tein determine where

F aster ⁼ total mass of water in the organ / total mass of the organ

F fat ⁼ total mass of fat in the organ / total mass of the organ

Fprotein ⁼ total mass of protein in the organ / total mass of the organ

From this, the distribution coefficient of the substance between an organ and water in equilibrium (Koi-gan / _W a _s ser) can be calculated:

Kθrgan / water ⁼ F aster + KFett ^' FFett + Kp _ro tein ^' Fp _ro tem (1)

The organ / blood or organ / plasma can be derived from this

Calculate distribution coefficients: Kθrgan Blood ^- Korgan / Water / Kfilut / Water (2)

- ^ Organ / Plasma ^- - "^ Organ / Wasser 'piasma Water ()

The coefficients Kßiut / water and Kpι _asm a / water are also calculated according to formula (1).

Using the biophysical model 4 (see FIG. 2), it is possible directly from the molecular properties of the substance KFett and Kp _ro t _e i _n the concentration of the substance in a particular organ X (Cx) as well as the

Determine the concentration of the substance in the arterial blood (C _ar ) by solving the above-mentioned systems of equations as well as the time course of these concentrations.

FIG. 4 shows a block diagram of a computer system for the calculation of ADME properties of a substance.

A file 12 contains the chemical structure of the substance, for example in the form of a so-called descriptor or a fingerprint. File 12 is manually entered by a chemist or is part of one

Substance library of chemical structures whose ADME properties are to be determined.

The file 12 is entered into a database 13. The database 13 is used to store files 12 describing chemical structures. The file 13 is queried cyclically by a program 14, specifically to determine whether a new file 12 has been entered in the period between the previous query.

For example, the file 12 can be entered from a client computer. The database 13 is located on, for example Server computer and also the program 14, which polls the database 13 cyclically. A distributed system can be implemented in this way.

If the program 14 determines that a new file 12 has been entered into the database 13 in the meantime, the program 14 starts automatically

Program 15 for calculating the molecular properties of the substance described by file 12. The molecular properties calculated by the program 15 are temporarily stored in a file or stored in a database 16.

After completion of the calculation of the molecular properties and their storage in the database 16, the program 15 automatically starts a program 17 for the calculation of one or more ADME properties of the substance. For this purpose, the program 17 accesses the database 16 in order to call up the molecular properties of the substance previously calculated by the program 15.

Alternatively or additionally, the program 17 accesses a database 18, which contains further experimentally determined molecular properties of the substance. This assumes that the substance has been previously synthesized in order to experimentally determine the molecular properties of the substance in a file 19 in the database

18 to be able to enter.

The ADME properties calculated by the program 17 based on the molecular properties stored in the database 16 and / or the database 18 are stored in a database 20. A program 21 accesses the database 20 in order to generate a structured output. This can be done in the form of a table in the form of a spread sheet. The output can also be sorted according to certain ADME properties. Reference numerals

biophysical model 4 partial model 5

Part model 6

Part model 7

Partial model 8 venous blood 9 arterial blood 10

Elimination model 11

File 12

Database 13

Program 14 Program 15

Database 16

Program 17

Database 18

File 19 Database 20

Program 21

Claims

claims

1. Procedure for calculating the ADME properties of a substance with the following steps:

Entering the molecular properties of the substance in a biophysical model, the biophysical model describing a relationship between the molecular properties and the ADME property,

Output of the ADME property.

2. The method according to claim 1 with the following further steps:

- input of the chemical structure of the substance into a database,

cyclic querying of the database for an input of a chemical structure,

- automatic calculation of molecular properties of the

Substance, based on the chemical structure, after the input to the database has been determined by the cyclical polling,

- Automatic entry of the molecular properties in the biophysical model.

3. The method according to claim 1 or 2 with the following further steps:

Access to a second database with further experimentally determined molecular properties of the substance, automatic input of the calculated and other molecular properties into the biophysical model.

4. The method according to claim 1, 2 or 3, wherein the biophysical model is a physiologically based pharmacokinetic model, the model comprising several organs and including a partial model for each of the organs, the partial models are connected to each other by a blood flow model and the relationship is described by mass balance relationships.

5. The method according to any one of the preceding claims 1 to 4, wherein the molecular properties of the substance from the chemical structure, preferably in the form of a descriptor or a fingerprint, are calculated by a statistical method, preferably QSAR or HQSAR, or by means of a neural network.

6. The method according to any one of the preceding claims 1 to 5, wherein it is a substance with a pharmacological effect on the human and / or animal body and one or more of the following ADME properties are calculated:

fraction of a dose of the substance absorbed after oral administration,

- Concentration-time curve course in the portal vein after oral

Application of the substance,

free fraction in plasma,

- organ - plasma distribution coefficients, Concentration-time curves in the blood plasma and in the organs after oral or intravenous administration of the substance.

7. The method according to any one of the preceding claims 1 to 6, wherein the substance is a substance for a crop protection application, and one or more of the following ADME properties are calculated:

a. Key figure for the speed of inclusion in a sheet after. a spray application,

b. Key figure for the speed of ingestion into an insect via the cuticle,

c. Key figure for the speed of distribution in the plant

Leaf application (phloem mobility),

d. Key figure for the speed of distribution in the plant after root application (xylem mobility),

e. Concentration distribution between the organs of an insect.

8. The method according to any one of the preceding claims 1 to 7 with the following further steps:

a. repeated calculation of the ADME property of different substances, in particular of different substances of the same basic chemical body,

b. Sorting of substances according to the value of the ADME property.

9. The method according to claim 8, wherein in each case a plurality of ADME properties are calculated for each of the substances and the sorting of the substances is carried out by forming an index from two or more of the ADME properties.

10. The method of claim 9, wherein the index is formed by a weighted sum of the at least two ADME properties.

11. The method according to claim 8, 9 or 10, wherein the output of the ADME properties of the substances in tabular form, preferably in a relational table.

12. Computer program for performing a method according to one of the preceding claims 1 to 11.

13. Computer system for performing a method according to one of the preceding claims 1 to 11.

14. Computer system according to claim 13 with

a. a first database (13) for entering the chemical structure of a substance,

b. Means (14) for cyclically querying the database for input of a chemical structure,

c. Means (15) for automatically calculating the molecular properties of the substance based on the chemical structure after the input of the chemical structure has been recognized by the cyclical query of the database, d. Means (17) for automatically calculating one or more of the ADME properties.

15. Computer system according to claim 13 or 14 with a second database (18) for storing further experimentally determined molecular

Properties of the substance, the automatic calculation of the ADME property based on the calculated and / or the further experimentally determined molecular properties.

16. Computer system according to one of the preceding claims 13, 14 or 15 with means (21) for outputting the calculated ADME properties of various substances in a structured form, preferably in tabular form, in particular in the form of a rational table.

17. Computer system according to one of the preceding claims 13 to 16 with

Means for calculating an index from the ADME properties for sorting the substances.