versión impresa ISSN 0366-1644
Bol. Soc. Chil. Quím. v.45 n.2 Concepción jun. 2000
GROUP (FRAGMENT) CONTRIBUTION METHOD FOR
PREDICTING THE BRAIN / BLOOD PARTITIONING OF
Elsa Abuin (*) and Eduardo Lissi
Departamento de Ciencias Químicas. Facultad de Química y Biología.
Universidad de Santiago de Chile. Casilla 40- Correo 33 Santiago - Chile.
(Received: February 23, 2000 - Accepted: April 26, 2000)
In memoriam of Doctor Guido S. Canessa C.
Experimental values of the partition constants between brain and blood, log K(Br/Bl), of a series of compounds including simple organic molecules and some drugs were correlated with experimental values of the partitioning between sodium dodecylsulfate (SDS) micelles and water, log K(SDS/w). From a training data set of 21 compounds a good linear correlation was found and a function was derived to relate log K(Br/Bl) with log K(SDS/w). This correlation together with a simple group (fragment) contribution method that is proposed to calculate log K(SDS/w) provided the basis for a succesfull calculation of log K(Br/Bl) values for several compounds outside the training dataset.
Key Words: Drugs, partitioning, brain, blood, sodium dodecylsulfate, group contribution.
Valores experimentales de las constantes de reparto entre cerebro y sangre, log K(Br/Bl), de una serie de compuestos que incluyen moléculas simples y algunas drogas fueron correlacionados con valores experimentales de las constantes de reparto entre micelas de dodecilsulfato de sodio (SDS) y agua, log K(SDS/w). Usando un conjunto de prueba formado por 21 compuestos se ha obtenido una buena correlación lineal de la cual se derivó una ecuación que relaciona log K(Br/Bl) con log K(SDS/w). Esta correlación, junto con un método simple de contribución de grupos (fragmentos) que se propone para calcular log K(SDS/w), permiten la evaluación exitosa de valores de log K(Br/Bl) para varios compuestos no constituyentes del conjunto de prueba.
Palabras claves: Drogas, reparto, cerebro, sangre, dodecilsulfato de sodio, contribución de grupos.
To knowledge the partitioning of therapeutic compounds between brain and blood is important in drug design1). However, the experimental determination of brain/blood partition ratios, K(Br/Bl), is difficult . It involves the synthesis of the compounds , often in radiolabeled form, and the measurement of the drug concentration in the brain and blood of laboratory animals2-4). It is then desirable to develop procedures able to predict K(Br/Bl) values of the compounds of interest from their physical chemical properties or, ideally, from their molecular structures. In this sense, several approachs have been attempted. Young et al.4) examined a series of twenty histamine H2 antagonists and found a correlation between log K(Br/Bl) and log P (octanol/cyclohexane)5) (r = 0.83 ; s = 0.44). Similarly, van de Waterbeemd and Kansy 6) using the same series of compounds that Young et al., found a significant correlation between log K(Br/Bl) and log P (cyclohexane/water) when a calculated volume descriptor was introduced in the parametrization. Abraham et al.7-9) have related log K(Br/Bl) values with molecular descriptors including the excess molar refraction, solute polarizability, hydrogen bond acidity and basicity, and molecular volume. Abraham et al.7) obtained an equation that was shown to be useful for the prediction of K(Br/Bl) values of complex molecules and Abraham's descriptors for many fragments are available7-10). Recently, Lombardo et al.1) found a significant correlation between experimental log K(Br/Bl) values and computed solvation free energies in water.
Group contribution methods or additivity schemes have been proposed to predict the partitioning of solutes between micelles and water11,12), although such schemes have been usually restricted to only a few groups of atoms. Surprinsingly, no attempt to predict K(Br/Bl) values by group contribution methods have been made.
In this work, we explore on the usefulness of a group (fragment) contribution method based in correlations with the partitioning between sodium dodecylsulfate (SDS) micelles and water to predict the K(Br/Bl) values of organic compounds, including some drugs.
RESULTS AND DISCUSSION
(A) Correlation between experimental log K(Br/Bl) and log K (SDS/w) values
A series of 21 compounds including some drugs and simple organic molecules was examined and a significant correlation between experimental log K(Br/Bl) and log K(SDS/w) values12) was found [ K(Br/Bl) and K(SDS/w) defined in equations (1) and (2), respectively].
K (Br/Bl) = (solute concentration in brain) / ( solute concentration in blood) (1)
K (SDS/w) = (solute mole fraction in micelle) / (solute mole fraction in water) (2)
The compounds considered and the experimentally determined values of the partition constants are compiled in Table I.
|Compound||log K(Br/Bl)||Reference||log K(SDS/w)||Reference|
|Chloroform||0.340||7||0000002.70 (2.83)||0000018 (19)|
|Halothane||0.350||7||0000003.05 (2.80)||0000020 (18)|
Figure 1 shows the correlation between log K(Br/Bl) and log K(SDS/w) obtained. From this figure, equation (3) was derived (r = 0.977 ; sd = 0.2)
| ||Fig.1 Relationship between experimen-tal log K(Br/Bl) and log K(SDS/w)values for the compounds compiled in Table I.|
A notable feature of the line relating log K(Br/Bl) with log K(SDS/w) is the low value of the slope. If K(SDS/w) is considered as a measure of the solute hydrophobicity13), the low value of the slope in Figure 1 implies that solute hydrophobicity plays a much minor role in determining the partitioning between brain and blood than in the SDS /w partitioning.
The data set used in the correlation shown in Fig. 1 includes a limited number of drugs (6 out of 21 compounds). Nevertheless, we consider that the correlation is good enough as to permit the determination of K(SDS/w) values from group (fragment) contribution calculations , and then to translate the K(SDS/w) values to K(Br/Bl) through the use of Eq. (3). In this sense, our effort was directed to obtain K (SDS/w) values from group (fragment) additivity.
By assuming that log K (SDS/w) can be obtained by group additivity according to Eq. (4)
log K (SDS/w) = S ni ai
where, ni is the number of groups of a particular type and ai is its contribution to log K (SDS/w), ai values of several groups where derived from experimental data obtained in series of closely related compounds, including hydrocarbons, alkanols, ketones, amines, aromatic amides, etc. Some of the values obtained are summarized in Table II24).
(a) slope average of log K (SDS/w) vs the carbon number in the series of hydrocar-bons and 1-alkanols.
(b) log K (aliphatic ketones) - log K of the corresponding hydrocarbon (e.g., butane for 2-pentanone).
(c) log K(2-methylpropan-1-ol) - (2aCH3 + aCH2 + aOH)
(d) log K(propan-2-ol) - (aCH3 + aOH)
(e) log K(diphenylamine) - 2aphenyl
(f) log K(naphtylamine) - anaphthyl
(g) log K(1-alkanols) - (aCH3 + n aCH2). Average value for the series of alkanols from methanol to decanol
(h) log K(N,N - dimethylaniline) - 2 aCH3
(B) Evaluation of log K(Br/Bl) values
log K(Br/Bl) values can be evaluated from Eq. (3) (with K(SDS/w) experimentally determined or
calculated by group additivity) or directly, by group additivity by assuming that the contribution of each group to log K(Br/Bl), ai' is equal to 0.38 ai. This latter procedure is valuable to predict the effect of substituents on the partition constant in series of complex but structurally related molecules, particularly for drugs. In this case, log K(Br/Bl) values can be evaluated from Eq. (5)
log K(Br/Bl) = log K(Br/Bl)ref + S D ni ai'
where ai' = 0.38 ai and D ni is the difference in the number of i groups between the desired compound and the reference compound.
(C) Calculated versus experimental values
The goodness of the group contribution method for giving K(SDS/w) values and the predictive ability of Eqs. (3) and (5) for giving log K(Br/Bl) values was assesed for a series of 21 compounds for K(SDS/w) and 10 drugs for K(Br/Bl). The compounds considered were out of the data set used in the correlation obtained from experimental values. Figure 2 shows the relationship between the calculated and experimentally determined partition constants. A very good correlation, with r= 0.95 is obtained.
Fig.2 Calculated versus experimental values.
For a series of 21 compounds including simple organic molecules and some drugs a good linear correlation between experimental log K(Br/Bl) and log K(SDS/w) values is obtained (Eq.(3)). The correlation is better than those previously found between log K(Br/Bl) and partition coefficients between bulk phases5,6). The group (fragment) contribution approachs described in Eqs. (4) and (5) can be successfully applied to calculate the partitioning between brain and blood of simple organic molecules and complex drugs.
Financial support of this work by Dicyt (USACH) and Fondecyt (Grant # 1980211) is acknowledged.
1. F. Lombardo, J.F. Blake and W.J. Curatolo. J. Med. Chem., 39, 4650 (1996). [ Links ]
2. W.M. Pardridge and L.J. Mietis. J. Clin. Invest., 64, 145 (1979). [ Links ]
3. E.G. Chikhale, K.-Y. Ng, P.S. Burton and R.T. Borchardt. Pharm. Res.,11, 412 (1994). [ Links ]
4. R.C. Young, R.C. Mitchell, T.H.Brown, C.R.Ganellin, R.Griffith, M. Jones, K.K. Rana, D. Sauders, I.R.Smith, N.E.Sore and T.J.Wilks. J. Med. Chem. 31, 656 (1988). [ Links ]
5. P. Seiler. Eur. J. Med. Chem., 9, 473 (1974). [ Links ]
6. H. van der Waterbeemd and M. Kansy. M. Chimia, 46, 299 (1992). [ Links ]
7. M.H. Abraham, H.S.Chadha and R.C. Mitchell. J. Pharm Sci., 83,1257 (1994). [ Links ]
8. H.S. Chadha, M.H. Abraham and R. C. Mitchell. Bioorg. Med. Chem. Lett.,4, 2511 (1994). [ Links ]
9. M.H. Abraham. Chem. Soc. Rev., 22, 73 (1993). [ Links ]
10. M.H. Abraham, H.S. Chadha, J.P. Dixon, C. Rafols and C. Treiner. J.Chem.Soc. Perkin Trans. 2, 887 (1995). [ Links ]
11. G.A. Smith, S.D. Christian, E.E. Tucker and J.F. Scamerhorn. Langmuir , 3, 598 (1987). [ Links ]
12. C.D. Jafvert, P.L. Van Hoof and J.K. Heath. Water Res., 28, 1009 (1994). [ Links ]
13. F.H. Quina, E. O. Alonso and J.P.S. Farah. J. Phys. Chem., (1995). [ Links ]
14. K.T. Valsaraj and L.J. Thibodeaux. Sep.Sci.Technol., 25, 369 (1990). [ Links ]
15. P.Stilbs. J. Colloid Interface Sci., 87, 385 (1982); ibid 94, 463 (1983). [ Links ]
16. G.A. Smith, S.D. Christian, E.E. Tucker and J.F. Scamehorn. J. Colloid Interface Sci., 130, 254 (1989). [ Links ]
17. S.Causi, R. de Luisi and S.Milioto. J. Solution Chem., 19, 995 (1990). [ Links ]
18. C.Treiner and M.H. Mannebach. J. Colloid Interface Sci.,118, 243 (1987). [ Links ]
19. K.T. Valsaraj, A.Gupta, L.J.Thibodeaux and D.P. Harrison. Water Res., 22, 1173 (1988). [ Links ]
20. S. Kaneshina, H. Kamaya and I. Ueda. J. Colloid Interface Sci., 83, 589 (1981). [ Links ]
21. M.G. Casarotto and D.J. Craik. J. Phys. Chem., 95, 7093 (1991). [ Links ]
22. E. Abuin (unpublish results) [ Links ]
23. L. Sepulveda, E. Lissi and F.Quina. Adv. Colloid Interface Sci., 25,1 (1986). [ Links ]
25. M.H. Gehlen, P.B. Fo and M.G. Neumann.J. Photochem. Photobiol. A: Chem. 59, 335 (1991). [ Links ]
26. J.C. Scaiano and J.C. Selwyn. Can. J. Chem., 59, 2368 (1981). [ Links ]
27. Y. Ulloa, M.A. Rubio, E. Lissi and A. Aspée. Bol. Soc. Chil. Quim., 39, 129 (1994). [ Links ]