The Calculation of Physicochemical Descriptors and their
Application in Predicting Properties of Drugs and Other Compounds
A Thesis Presented to the University of London for the Degree of Doctor of
Philosophy in the Faculty of Science
By JOELLE LE
Sir Christopher Ingold Laboratories
Chemistry Department
University College London January 2001 ProQuest Number: 10010399
All rights reserved
INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest.
ProQuest 10010399
Published by ProQuest LLC(2016). Copyright of the Dissertation is held by the Author.
All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code. Microform Edition © ProQuest LLC.
ProQuest LLC 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 To my mother Abstract
The work presented may be divided into two main sections:
The first section focuses on the important aspect of compound descriptor determination. The method by which descriptors are obtained indirectly through compound solubility in organic solvents and direct water-solvent partition measurements is illustrated by example for drug compounds. This approach is extended through the derivation of gas-water and water-solvent partition equations for the n-alcohols which in the future will be available for use in descriptor determination. Importantly, the equation coefficients are also interpreted to deduce various physicochemical properties of the homologous series of alcohols. An alternative method to assign descriptors is probed through reversed-phase HPLC. Measurements are recorded for a series of solutes on several bonded phases and multivariate analysis is used to investigate the interrelationship between columns in an effort to isolate the most suitable phases.
The second section is concerned with application of the Abraham General Solvation Equation to examine processes of special interest in drug design; aqueous solubility and intestinal absorption. An algorithm to predict water solubility is obtained containing an additional Eocz^x Ep 2^ cross-term which is found to compensate at least partly for a melting point correction term. The amended equation is shown to be comparable in accuracy to commercially available packages for a test set of 268 structurally diverse compounds. Of further importance in drug delivery is the process of intestinal absorption. An extensive literature search provides evaluated absorption data for a large set of drug compounds and forms a strong basis for subsequent QSAR analysis. Intestinal absorption is found to be comparable in humans and rat, and predominantly dependent on the hydrogen-bonding capability of the drug. The mechanism of absorption is considered through transformation of the percent absorption data to an overall rate constant. Table of Contents
Page no.
Abstract 1 Table of Contents 2 List of Tables 6 Acknowledgements 10
Chapter 1 Introduction to the Abraham General Solvation Equation
1.1 History of QSAR and LFERs 11 1.2 Physicochemical Descriptors 13 1.3 Linear Solvation Energy Relationships 20 1.4 The Abraham General Solvation Equations 22
1.4.1 The Excess Molar Refraction, R 2 24
1.4.2 Solute Hydrogen-bond acidity, (% 2^ 25
1.4.3 Solute Hydrogen-bond acidity, p 2^ 26
1.4.4 Effective solute scales of S a 2^, Sp 2^, 7t2^ 27 1.4.5 McGowan’s Characteristic Volume, Vx 30 1.4.6 Estimation of Descriptors using Group Contribution Approach 31 1.5 Multiple Linear Regression Analysis (MLRA) 33 1.5.1 Difficulties with MLRA 37 1.6 References 38
Chapter 2 Aims of the Present Work
2.0 Aims of the Present Work 42
Chapter 3 The Determination of Solute Descriptors
3.1 Method for Descriptor Determination 45 3.2 Example : Descriptors for Vinclozin 49 3.3 Descriptors for Diazepam Analogues 51 3.4 Descriptors for P-blockers 65 3.5 Conclusion 71 3.6 References 72
Chapter 4 The Solvation Properties of the Aliphatic Alcohols
4.1 The Solubility of Gases and Vapours in Alkan-l-ols 73
4.2 Water-alcohol Partitions 86 4.3 References 92 4.4 Data tables 100
Chapter 5 Characterization of HPLC phases and use of HPLC in descriptor determination 5.1 Introduction to HPLC 117 5.2 Concepts of HPLC 118 5.2.1 Fundamental Relationships of Chromatography 120 123 5.2.2 Retention Mechanisms 126 5.2.3 Stationary Phases 129 5.2.4 Solvents 5.2.5 Instrumental Aspects 130 5.3 Experimental Section 132 5.4 Comparison of Stationary Phases 140 5.5 Use of Water-solvent Partition Measurements (WSPM) to obtain 147 Abraham Descriptors and Comparison with HPLC Systems
5.5.1 WSPM in Descriptor Determination 147 5.5.2 Characterization of HPLC Systems 149 5.5.2a Vector Analysis 151 5.5.2b Principal Component Analysis (PCA) 153 5.5.3 Comparison of HPLC and Water-solvent Partition Systems 154 5.5.4 Application of HPLC in Descriptor Determination 156 5.5.5 Conclusion 163 5.6 References 167 Chapter 6 The Solubility of Compounds in Water
6.1 Introduction 169 6.2 Prediction Methods 170 6.2.1 Comparison of Literature Models 175 6.3 Application of the Abraham General Solvation Equation in Prediction 178 6.3.1 Test set results : Comparison with Other Prediction Methods 184 6.3.2 Final Equations (for total dataset) 187 6.3.3 Influence of Very High and Very Low Soluble Compounds 189 6.3.4 The Factors that Influence Aqueous Solubility 191 6.4 The Solubility of Bronsted Acids and Bases 193 6.5 References 198
6.6 Data tables 202
Chapter 7 Gastrointestinal (GI) Absorption of Drug Compounds
7.1 Introduction 225 7.1.1 General Mechanisms for Transport of Substances Across 226 Biological Membranes 7.1.2 Factors Influencing Intestinal Absorption 229 7.1.2a Physicochemical Properties 230 7.1.2b Physiological Properties 235 7.1.3 Hepatic Drug Metabolism 238 7.1.4 The Prediction of Human GI Absorption 240 7.2 Human GI Absorption 244 7.2.1 Evaluation of Human Absorption Data 244 7.2.2 Relationship Between Human GI Absorption and Abraham 269 Descriptors
7.3 Rat GI Absorption 283 7.3.1 Evaluation of Rat GI Absorption Data 284 7.3.2 Relationship Between Rat GI absorption and Abraham 290 Descriptors
7.4 Comparison Between Human and Rat GI absorption 299 7.5 The Mechanism of Human GI Absorption 302 7.5.1 Bronsted Acids and Bases 309 7.5.2 Characterization of the Absorption System 313 7.5.3 Conclusion 317 7.6 Partitioning of Drug Compounds onto a C]g Disk 319 7.6.1 Experimental 319 7.6.2 Calculations 320 7.6.3 Results 321 7.7 References 328
Chapter 8 Visualisation of the Abraham General Solvation Equation
8.0 Visualisation of the Abraham General Solvation Equation 354
Chapter 9 Conclusions and Suggestions for Future Work
9.0 Conclusions and Suggestions for Future Work 357 List of Tables
Chapter 1 Introduction to the Abraham General Solvation Equation Page no. Table 1.1 Multivariate statistical techniques 20
Table 1.2 Comparison of effective solute descriptors and those based on 28 1:1 equilibrium constants
Table 1.3 Characteristic atomic volumes, Vx in cm^ mol'^ 31
Table 1.4 Results of regression using 81 modified parameters to estimate 32 Abraham descriptors
Table 1.5 Results of training and test set regressions using Group 32 Contribution approach
Chapter 3 The Determination of Solute Descriptors Table 3.1 Eqn coefficients for partition between water and solvents 47
Table 3.2 Eqn coefficients for partition between the gas phase and 48 solvents Table 3.3 Solvent solubilities of vinclozolin (S in mol dm'^), and derived 49 partition coefficients
Table 3.4 Observed and calculated values of log P and log for 50 vinclozolin
Table 3.5 Descriptors for Diazepam analogues and (3-blockers 71
Chapter 4 The Solvation Properties of the Aliphatic Alcohols Table 4.01 Calculation of log L in dry octan-l-ol at 298K 75
Table 4.02 Descriptor space for the octan-l-ol regression (eqn 14) 79
Table 4.03 Coefficients in the log L equations for gas-solvent partitions at 80 298K
Table 4.04 Some properties of bulk solvents 82
Table 4.05 A term-by-term analysis of solvation of gaseous solutes at 84 298K
Table 4.06 Values of the Kamlet-Taft solvatochromie parameters for 85 water and some alcohol solvents
Table 4.07 Coefficients in the log? equation for water-solvent partitions 88
6 Table 4.08 Solute Abraham descriptors and log values used in alkan-1- 100 ol regressions
PrOH Table 4.09 Values of log L"^"calc, log L'^, log and log 104 cale for solutes at 298K
Table 4.10 Values of log log L'^, log and log 106 calc for solutes at 298K
Table 4.11 Values of log log L'^, log and log 108 calc for solutes at 298K
Table 4.12 Values of log log L'^, log P " " ° " ^ and log 110 calc for solutes at 298K
Table 4.13 Values of log log L’^, log and log 111 pHeptOHAv 298K
Table 4.14 Values of log log L", log p°'='°H™' and log P°"°"™ 112 calc for solutes at 298K
Table 4.15 Values of log log L*, log pD“ °H^ and log pD“=°H™' 115 calc for solutes at 298K
Chapter 5 Characterization of HPLC phases and use of HPLC in Descrijptor Determination Table 5.01 Solvent properties 130
Table 5.02 Abraham descriptors for training set solutes 134
Table 5.03 HPLC columns used 136
Table 5.04 Log k’ values for compounds obtained from each stationary ^37 phase and mobile phase composition (MeCN/H 2 0 )
Table 5.05 LFERs obtained for each of the HPLC systems 139 Table 5.06 Coefficients of 60/40 (MeCN/HzO) mobile phase composition for the different stationary phases
Table 5.07 Ratios of LEER coefficients for the seven stationary phases 142
Table 5.08 Properties of selected solvents 143
Table 5.09 Equation coefficients for water-solvent systems 147
Table 5.10 Coefficients for reversed phase HPLC systems 149
Table 5.11 Matrices of cos 0 and 0 for each of the possible column pairs 152
Table 5.12 Results of descriptor back-calculation using HPLC 156
Table 5.13 Error associated with the ‘critical quartet’ equation coefficients 158 Table 5.14 Error associated with the HPLC equation coefficients 158
Table 5.15 Test set compounds and corresponding database descriptors 162
Table 5.16 Test set log k’ values and predictions of and 2 ^2^ 162
Table 5.17 Coefficients of other reversed phase HPLC systems studied 163
Table 5.18 Examples of Roche compounds for which descriptors have 164 been obtained from this work
Chapter 6 The Solubility of Compounds in Water Table 6.01 Models for the correlation and prediction of aqueous solubility, 175 as log Sw, that require additional data (ASf,mp,ô')
Table 6.02 Models for the correlation and prediction of aqueous solubility, 176 as log Sw, that do not require additional data
Table 6.03 Comparison of different compilations of solubility data 177
Table 6.04 Solubility values for compounds miscible with water 179
Table 6.05 Stepwise regression matrix for training set, n=803 182
Table 6.06 Correlation matrix for training set 182
Table 6.07 Comparison of methods using test set compounds 185
Table 6.08 Descriptor space covered by eqn 27 188
Table 6.09 Correlation equations using the amended equation with the 190 Eaz^ X term without compounds of low and high solubilities Table 6.10 Hydrogen-bond effects on solubility, as log Sw 191
Table 6.11 Observed and calculated log Sw values for strong Bronsted 196 acids, pKa<4
Table 6.12 Observed and calculated log Sw values for strong Bronsted 196 bases, pKa>10
Table 6.13 Training set data for log Sw regressions 202
Table 6.14 Test set data : log Sw predictions using Abraham method, 216 WsKow (Meylan) and TOPC(Klopman) models, and mp and log Poet data Chapter 7 Intestinal Absorption of Drug Compounds Table 7.01 Drug absorption in different pH environments 234
Table 7.02 Molecular weight, exp. and calc, solubility values, and exp. 254 and calc, octanol-water partition coefficients
Table 7.03 Human absorption, dose and percentage of excretion in urine, 259 bile and faeces of drugs from literature
Table 7.04 Regression results of different training sets by Abraham 271 descriptors (human absorption)
Table 7.05 Details of drugs used for analysis (human) 272
Table 7.06 Comparison of statistics for equations for % absorption 276
Table 7.07 Observed and predicted absorption from Model 1 278
Table 7.08 Rat absorption, percentage excretion in urine, bile and faeces 286 of drugs and drug-like compounds
Table 7.09 Observed rat absorption and calculated absorption from 291 Abraham descriptors
Table 7.10 Regression results of training and test + cross-validation sets 295 using Abraham descriptors (rat absorption)
Table 7.11 Details of drugs used in analysis (rat) 296
Table 7.12 System coefficients for water/phase transfers 308
Table 7.13 System coefficients for diffusion processes 308
Table 7.14 Analysis of absorption of Bronsted acids on eqn 25, with 311
Table 7.15 Analysis of the absorption of Bronsted acids on the mechanism 312 ofeqn29, withH=10 ‘^ ^
Table 7.16 Characterization of systems 315
Table 7.17 Rank order of difference from log k for human intestinal 318 absorption
Table 7.18 Descriptors and experimental data for diuron and drug 323 compounds
Table 7.19 Equations for log Keq, log kup and log koff 325
Chapter 8 Visualisation of the Abraham General Solvation Equation Table 8.1 Calculated change in aqueous solubility and human intestinal 354 absorption with respect to change in substituent group
9 Acknowledgements
The completion of this thesis would not have been possible without the constant guidance of Dr. Michael H. Abraham; his knowledge and wisdom throughout my postgraduate years has been invaluable. With greatest sincerity I thank Dr. Abraham for his unequivocal patience and kindness.
It must be emphasized that the work presented in Chapter 7.2 - 7.5 is the result of work that has been conducted conjointly with Dr. Yuan Zhao. I am indebted to Yuan for his unbounding enthusiasm, and for the selfless sharing of his expertise in QSAR methodology.
I am grateful to past and present members of the research group for their encouragement over the years, and for making the time spent at UCL so pleasant. Many thanks to Dr. Caroline Green, Joelle Gola, Dr. Jamie Platts, Julian Dixon, and Vikas Gupta for their friendship during my PhD quest. I wish Kei Enomoto, Andreas Zissimos and Rui Fuigera all the best in their continuing research work.
I would like to express my utmost appreciation for all the advice and support afforded by my industrial supervisors - Dr. Brad Sherborne, Dr. Ian Cooper, Dr. Gordon Beck, and Dr. Nadine Randolph. I thank Dr. Sherborne for his numerous helpful and critical comments regarding this thesis, and Dr. Cooper for organizing my work on-site. I also thank Mr. Brian Scott and Mr. John Whateley for their help regarding the HPLC practical work.
I wish to extend my gratitude to Roche Products Ltd. for making this work possible by funding a research studentship.
Finally, I acknowledge the invaluable contribution of my mother, to whom this thesis is dedicated. I thank my mother for all the sacrifices, patience, and encouragement which have made university, and ultimately the completion of this PhD possible. For the pivotal role of her unfailing love in so many aspects of my life and achievements Thank you doesn’t seem enough!
10 Chapter 1 Introduction to the Abraham General Solvation Equation
1.1 History of OSAR and LFERs
Inherent to chemistry is the concept that there is a relationship between bulk properties of compounds and the structure of the molecules of those compounds. This provides a connection between the macroscopic and the microscopic properties of matter. Quantitative structure-activity relationships (QSARs) represent an attempt to identify these relationships between molecular structure and activity/property by correlating structural or property descriptors of compounds with activities. These physicochemical descriptors, which include parameters to account for hydrophobicity, t topology, electronic properties, and steric effects, are determined empirically or, more recently, by computational methods. Activities used in QSAR include chemical measurements and biological assays. QSAR are currently being applied in many disciplines, with many pertaining to drug design and environmental risk assessment.
A general formula for a quantitative structure-activity relationship (QSAR) can be given by the following :
Activity = f (molecular structure) = f (descriptors)
QSAR date back to the 19^^ century. The first attempt to relate a physicochemical parameter to a pharmacological effect was reported by Crum Brown and Fraser (1868- 9) who noted that the paralysing properties of a series of strychnines depended on the quatemising group and hence physiological activity was some function of the constitution of the molecules\ In the 1890’s, Hans Horst Meyer^ of the University of Marburg and Charles Ernest Overton^ at the University of Zurich, working independently, noted that the toxicity of organic compounds depended on their lipophilicity. Lipophilicity represents the affinity of a molecule for a lipophilic environment and is commonly measured by its distribution behaviour in a biphasic system such as partition coefficient in octanol/water.
11 Linear Free Energy Relationships (LFERs) Little additional development of QSAR occurred until the work of Louis Hammett in 1937 who correlated electronic properties of organic acids and bases with their equilibrium constants and reactivity. From the ionization of benzoic acids, Hammett derived equations for the rate constants and equilibrium constants of reactions of the meta- and para-substituted benzoic acid derivatives.
log kx = log /:h + per ( 1)
log Kx = log Kh + pa (2)
The constants ku and Kh refer to the unsubstituted compound, while kx and K% refer to a meta- or para-substituted version. The magnitude of the substituent constant, a, reflects the polar effect a given substituent has on the rate or equilibrium of a reaction, relative to hydrogen. If the substituent is electron-withdrawing, a is positive and if the substituent is electron-donating, a is negative. This effect is independent of the reaction. The slope p (referred to as the reaction constant) measures the sensitivity of the process to electronic effects exerted by substituents and is dependent on the nature and experimental conditions of the reaction under consideration. The aqueous ionization of benzoic acid at 25°C is considered to be the reference reaction for which p=l.
Hammett provided the pioneering work on showing the usefulness of parametric procedures in describing an empirical property in terms of a parameter describing molecular structure. The Hammett equation is a prominent example of LFER; the term being derived from the use of these relationships as mathematical tools for correlating changes in free energy in different reaction series. Logarithms of rate or equilibrium constants are used instead of free energy because rate and equilibrium constants are logarithmically related to free energy (AG) through the van’t Hoff equation
AG = -2.303RT log K (3)
where R is the gas constant and T is the temperature.
I 12
! 1.2 Physicochemical Descriptors
QSAR then rapidly developed as a natural extension of the LFER approach, with a biological activity correlated against a series of parameters that described the structure of a molecule. This was necessary because difficulties were encountered when investigators attempted to apply Hammett-type relationships to biological systems, indicating that other structural descriptors were necessary.
Tafias substituent constants The most obvious limitation of the Hammett equation is that it does not hold for orr/io-substituents. Based on Ingold’s hypothesis that the ortho-effect was steric in nature, Taft ^ used the acid- and base-catalysed hydrolysis of esters as a basis for the substituent constant a*. Taft’s substituent constants are a measure of the polar effects of substituents in aliphatic compounds when the group in question does not form part of a conjugated system. They are based on the hydrolysis of esters and are calculated from eqn 4, where k represents the rate constant for the hydrolysis of the substituted compound, and ko that of the methyl derivative. a*=(l/2.48)[log(Â:/^o)B-log(log(Â:/^ojA] (4)
The bracketed term with the subscript B represents basic hydrolysis and the other, with the subscript A, acid hydrolysis. The factor 2.48 brings the constants on to the same scale as the Hammett constants. The equation depends on the fact that although both basic and acid hydrolysis are sensitive to steric effects, only basic hydrolysis is influenced by polar effects; hence, by subtracting the acid term from the basic term, only the polar effect remains.
Taft’s steric substituent constant ^ (Eg) is a corollary of eqn 4. It depends on the fact that acid hydrolysis is determined almost completely by steric factors, and is defined by the eqn :
Es = lo g (% )A (5)
13 Taft substituent constants are different from the others in that methyl, rather than hydrogen, is the standard group, for which the constant is zero. However, they can be
compared with other constants by writing the methyl group in the form, CH 2-H and identifying it as the group for H. Another substituent constant representing a group X
can then be compared by using the Taft constant for CH 2-X. Under these circumstances, Taft and inductive substituent constants are approximately related by
(7* = 2 .51(71 (6)
However, a number of corrections were made to eqn 5 when it was apparent that Eg contained resonance effects. Dubois rejected the notion that all four reactions on which the Eg values were averaged would respond identically to steric effects. Instead, a single standard was chosen which was the acid-catalysed estérification of carboxylic acids in methanol at 40°C. The resulting Taft-Dubois steric parameter, Eg has gained widespread use and is one of the standard measures of steric effects in organic chemistry ^
The list of Es values has been limited by the experimental difficulties in obtaining the physicochemical data upon which Es values are based. However, recently, measurement of steric effects has been computed by Ligand Repulsive Energies, E r , which is a step in overcoming this problem. Using the E r values obtained using the
Cr(C0 )5 fragment provided the most consistent and generally useful measure of relative steric sizes log Pact Robert Muir, a botanist at Pomona College, was studying the biological activity of compounds that resembled indoleacetic acid and phenoxyacetic acid, which function as plant growth regulators. In attempting to correlate the structures of the compounds with their activities, he consulted his colleague, Corwin Hansch. Using Hammett sigma parameters to account for the electronic effect of substituents did not lead to meaningful QSAR. However, Hansch recognized the importance of the lipophilicity, expressed as the log octanol-water partition coefficient (log Poet) on biological activity and this has now become the most well known and most used descriptor in QSAR.
14 Ferguson was the first person to rationalise the early observations regarding the depressant action of structurally non-specific drugs. Ferguson reasoned that the important parameter for the correlation of narcotic levels was their relative saturation; this hypothesis arising from simple thermodynamic principles which govern equilibrium between immiscible phases However, Hansch and Fujita assumed a steady-state model to explain the cut-off point in certain homologous series, thus suggesting that the equilibrium conditions required by Ferguson’s theory were not established. Indeed, the parabolic model of Fujita-Hansch and the bilinear model of Kubinyi describe this empirical observation that the relationship between partition coefficients and certain biological endpoints is not always linear, but displays an optimum value: log 1/C=a (log?) + b (7) log 1/C=a (logP)^ + b (logP) 4- c ( 8 ) log 1/C=a (logP) - b (log(PP-hl)) + c (9) where C is the molar concentration that produces a certain effect, P is often the octanol/water partition coefficient, and a, b,c and |3 are regression coefficients.
Furthermore, Hansch and Fujita considered that at least two conditions were required for biological activity, drug transport and distribution: 1) A random walk process in which the molecule makes its way from a very dilute solution outside the cell to its site of action inside the cell. 2) Factors that were required for drug interaction with the site of action. These include different kinds of bonding : hydrogen-bonding, ionic forces, van der Waals or hydrophobic, as well as dipole-dipole interactions. These conditions may be parameterised to some extent in a QSAR expression and the so-called Hansch equation takes into account these effects
log 1/C=a (logP)^ + b (logP) -I- cEs + da -h e (10)
Where Eg is Taft’s steric descriptor, a is the well-known Hammett constant, and P is the octanol/water partition coefficient.
15 Since the pioneering work of Hansch in the 1960’s, many other different molecular and fragmental descriptors have been used in these extrathermodynamie or linear free- energy relationships (LFER).
Hydrogen-bonding parameters Hydrogen-bonding is an important property in drug aetivity; it contributes to solubility, partitioning and reeeptor binding. It is however, difficult to quantify, and consequently its use in QSARs is sometimes as an indicator variable. However, there exist more quantitative measures of hydrogen-bonding ability. The hydrogen-bond parameters of Abraham are discussed in detail in Chapter 1.4, but there are also the hydrogen-bond parameters of Raevsky^^, Carr^^ and Famini
Raevsky et al have devised scales of hydrogen-bonding using 1:1 complexation constants in tetrachloromethane, summarized in the following eqn :
AG° = 2.43 C a . Cb + 5.70 (11) n = 936, r = 0.9840, SD = 1.11, F = 28556
where AG° is the Gibbs free energy change for complexation in kJ m ol'\ and Ca and
Cb are the solute hydrogen-bond acidity and basicity respectively. Raevsky et al. have also set out the only general equation yet developed in terms of enthalpy :
AH° = 4.96 Ea . Eb ( 12) n = 936, r = 0.9540, SD = 2.70, F = 9553 where AH° is the standard enthalpy change for 1:1 eomplexation in tetrachloromethane in kJ mol‘\ and Ea and Eb are the solute hydrogen-bond acidity and basicity
The method of Carr is very similar to that of Abraham. Retention data for various series of solutes on a number of GLC stationary phases are analysed through an LFER equation
16 log SP — C + d.d2 + S7Ï2^ + Ü.0t2^ + b.p 2^ + l.logL^^ (13) log SP is a set of retention data. The descriptors are : d2 : Kamlet’s polarizability correction parameter, taken as 1.0 for aromatic solutes, 0.5 for polychloroaliphatic solutes and zero for the rest. 712^ : Dipolarity/polarizability parameter
(%2^ : Hydrogen-bond acidity
p 2^: The 1:1 hydrogen-bond basicity of Abraham logL^^ : The logL^^ of Abraham
Data on nineteen GLC phases were analysed using a round-robin procedure until the regression coefficients and the 7t2^ and az^ descriptors had converged. The other descriptors were taken as fixed values. A scale of hydrogen-bond basicity denoted as p 2^ was then set up using two GLC phases of 4-dodecyl-a,a-bis(trifluoromethyl) benzyl alcohol and the corresponding ether. The relationship between the hydrogen- bond basicity descriptor of Carr and the equivalent basicity term of Abraham is not linear since ^2^ is much larger for the strong hydrogen-bond bases
The theoretical linear solvation energy relationship (TLSER) descriptors of Famini use the LSER philosophy (described in Chapter 1.3) and general structure, but replace the empirically derived descriptors with computationally derived descriptors. TLSER descriptors were developed to correlate closely with the LSER descriptors; to give equations with r (correlation coefficient) and SD (standard deviation) close to those with LSER and to be as widely applicable to solute-solvent interactions as the LSER set.
SSP = SSPq + 6711 4- ê £ a + y q + + C£b + dq. 4- aV m c (14)
SSP represents a solute-solvent interaction property; this is generally taken as a logarithm of a measured quantity, 71] is the polarizability index (obtained by dividing polarizability volume by the molecular volume)
£a is the covalent contribution to the hydrogen-bond acidity and is the magnitude of the difference between the lowest occupied molecular orbital of the solute and the
17 the difference between the lowest occupied molecular orbital of the solute and the highest occupied molecular orbital of water. q+ is the electrostatic contribution to hydrogen-bond acidity and is taken as the most positive formal charge on a hydrogen atom in the molecule. Es is the covalent contribution to the hydrogen-bond basicity and is calculated as the difference between the HOMO of the solute and the LUMO of water, q. is the electrostatic contribution to hydrogen-bond basicity and is taken as the absolute value of the most negative formal charge in the molecule. Vmc is the molecular van der Waals volume, in units of 100 cubic angstroms.
There is a degree of correlation between q+ and q. and the Abraham and EP 2".
Abraham Za 2 ^ as a function of Famini TLSER descriptors Til q+ c S D n -2.809 2.812 0.194 0.816 0.094 194 2.784 -0.113 0.788 0.101 194
Outliers : dichloroacetic acid, trichloroacetic acid, propan-1,3-diol, benzaldehyde, 3,5- dichlorophenol, 3-nitrophenol, 4-nitrophenol, cajfeine, estradiol.
Abraham as a function of Famini TLSER descriptors 7li q Vmc c r ' SD n 1.868 1.503 0.069 0.109 0.698 0.124 192 1.070 1.541 0.084 0.691 0.125 192 1.580 -0.043 0.688 0.125 192
Outliers : propan-1,3-diol, quinoline, pyrazine, pyridazine, uracil, thymine, cajfeine, progesterone, testosterone, pregnenolone, estradiol.
Steric Parameters Charton^^.21 developed a new set of steric measures, termed v values to bypass the limitations which accompany the Taft method.
Vx = rvx - rvH = rvx (15) where r is the van der Waals radius of the symmetrical substituent and 1.20 is the radius of hydrogen in angstroms. The effective steric value Vgff was later introduced to incorporate effects which were dependent on energetic factors.
18 Topological indices represent chemical structures in numerical form and encode both molecular size and shape of the whole molecule i.e. they encode connectivity as well as steric features. Since the first index was developed by Weiner in 1947, more than 50 topological indices have been presented in the literature. The most notable are the molecular connectivity index of Randic and the Kappa Indices (graph theoretical index) developed by Kier 25-27
Sterimol parameters were developed to overcome the criticism of steric parameters so far mentioned in that they represent only one aspect of the shape of the group. Each chemical group is allocated 5 Sterimol parameters. L, which is the distance the group protrudes from the parent molecule, and B1-B4, which give the widths of the group in four directions, 90° to each other and perpendicular to the axis along which L is measured. Cross sectional dimensions increase from Bi to B4. The Bi parameter has been shown to be highly correlated with Taft’s Es parameter as well as Charton’s veff 29
New Methods and Descriptors Although theoretical quantities were involved in QSAR models in the early 1960s, the widespread use of computational chemistry to produce quantitative descriptions of chemical structure did not happen until the 1980s Extensions to the Hansch method and the development of alternative approaches to QSAR have since been prolific, aided by faster and more powerful computers. The driving force behind the move away from the use of classical substituent constants has been prompted by the advent of combinatorial technology and the need for the application of QSAR to noncongeneric series of molecules. Alternatives to the classical multivariate statistical techniques of multiple linear regression and nonlinear regression have also gained widespread use (see table 1.1). Pattern recognition QSAR methods such as ADAPT SIMCA and CASE are now actively used to search for patterns which predict the category of compounds which have been classified as active, partially active, inactive etc. There has also been evolution in 3D techniques, examples of which include CoMFA EVA and WHIM^’ descriptors.
19 Table 1.1 Multivariate statistical techniques Pattern recognition Correlation analysis Cluster Analysis Multiple Linear Regression Principal Component Analysis Principal components Regression Non-linear Mapping Partial Least Squares Regression Neural networks Neural networks
1.2 Linear Solvation Energy Relationships
Over the years, an enormous number of descriptors have been used in both LFERs and QSARs, both empirical and computational, to develop usable regressions. Some of the difficulties with such a large base from which to choose is which descriptors will provide the best regressions. By “best”, both goodness of fit (through r^, F-statistic, standard deviation) and the chemical meaning of the regression(s) must be taken into account. The real utility originates from the ability to design compounds with desirable properties and since an excellent correlative regression with difficult to interpret parameters does not provide characterization capabilities, it is imperative that parameters be chosen that have chemical meaning. Using this argument, Kamlet and Taft developed a new generalized relationship for studying LFERs of solute-solvent interactions. Termed the linear solvation energy relationship (LSER) it has the form:
log SP = S?o + dipolarity/polarizability term + hydrogen-bonding term(s) + cavity term (16)
Each of these descriptors was empirically derived and are named solvatochromie parameters because they were originally derived from solvent effects on UV-visible spectra. The cavity term measures the endoergic process of separating the solvent molecules to provide a suitably sized cavity for the solute. The magnitude of the cavity terms depends on Y 2 and (6h^)i. V2 is the solute molar volume, taken as its
molecular weight divided by its liquid density at 20°C; V 2/IOO is used so that the scale of the cavity term is similar in magnitude to the other descriptors. The complementary
solvent parameter, ( 6h^)i, is the square of the Hildebrand solubility parameter, and is
20 often referred to as the solvent cohesive energy density. The dipolarity/polarizability term (n ) measures the exoergic effects of solute-solvent dipole-dipole and dipole- induced-dipole interactions. Exoergic hydrogen-bonding terms measure the effects of hydrogen-bonding involving the solute as hydrogen-bond base and the solvent as hydrogen-bond acid and/or the solvent as hydrogen-bond base and the solute as hydrogen-bond acid. The pm and Om solvatochromie parameters measure the ability of the solute to accept or donate a share of a proton in a solute-solvent hydrogen-bond. The subscript m indicates that for compounds that are capable of self-association by hydrogen-bonding (amphihydrogen-bonding compounds), the parameter applies to the non-self associated “monomer” solute rather than the self-associated ‘oligomer’ solvent.
Accordingly, eqn 16 with the solvatochromie parameters appropriately included becomes :
log SP = SPo + AJt*i 7i*2 + Ba,(Pm)2 + CPi(c(m)2 + D (8 h^),(V2/100) (17)
where subscripts 1 and 2 denote properties of the solvent and solute respectively.
When correlating the many diverse logSPs in terms of solute parameters, eqn 17 reduces to :
log SP = SPo + mV 2/100 -+- sn * 2 + ^(0 ^ ) 2 + /^(Pm)2 (18)
Conversely, when dealing with the effects of different solvents on properties of single solvents, logSP is correlated in terms of solvent parameters:
log SP = SPq + /i(0h^)i/100 + S7Z 1 + cicxi + b^\ (19)
21 1.4 The Abraham General Solvation Equations
Because the Kamlet-Taft solute parameters were partly based on solvent properties, Abraham and co-workers defined new solute scales of dipolarity, hydrogen-bonding, and volume. These new parameters led to the construction of the General Solvation Equations
The General Solvation Equations
log SP = c 4- r.R 2 + s.Ti'^ + 4- v.Vx (20) log SP = c -i- r.R 2 + 5.712^ 4- ût.Eot2^ + Z?.E(32^ + /.LogL^^ (21)
log SP : A biological or chemical property of a series of solutes in asystem, for example, the rate of absorption of drugs through abiological membrane, or the partition coefficients of a number of compounds in a particular water solvent system.
c : The equation constant
r, 5, a ,b ,v : The equation descriptor coefficients which are obtained by multiple linear regression.
R2 : The excess molar refraction in units of (cm^ mol'^)/10, which reflects solute polarizability. It provides a quantitative measure of the ability of a solute to interact with a solvent through n and 7i electron pairs.
712^ : The solute dipolarity/polarizability parameter. This measures the ability of the solute to stabilize a charge or dipole.
Zoc2^ : The solute hydrogen-bond acidity summation parameter. A measure of the extent of hydrogen-bonding by the solute in a basic solvent.
Zp 2^ : The solute hydrogen-bond basicity summation parameter. A measure of the extent of hydrogen-bonding by the solute in an acidic solvent.
22 Vx : McGowan’s characteristic volume in units of (cm^ mol'^yiOO. Represents the three-dimensional space occupied by the solute.
Log : The gas-hexadecane partition coefficient or the Ostwald Solubility coefficient at 298K. Accounts for cavity size and dispersion interactions.
Eqn 20 is applicable to processes within condensed phases since the change in dispersion interaction can be assumed to be negligible when a solute is transferred from one liquid to another. The Vx equation has been applied to numerous water- solvent partitions, water-plant cuticle partition water-micelle distribution blood-brain distribution water-skin permeation chromatographic systems and sorption onto soil
Eqn 21 is applicable to processes in gas to condensed phases (gas-solvent partitions). Dispersion interactions are more significant in these systems since a solute is transferred from the gas phase to the liquid phase. Whilst there is no interaction between molecules in the gaseous state, when the gaseous solute dissolves in a solvent, dispersion interactions will be set up between the solute and solvent molecules. The logL^^ equation has been applied to numerous gas-solvent partitions, gas-plant partition gas-biological tissue distribution"^^, gas chromatography data^®, nasal pungency and eye irritation in man
These two generalized linear solvation energy relationships are based on a cavity theory used by Abraham, Kamlet and Taft, conceptualized as the solvation of a gaseous solute into a liquid solvent. The process can be broken down as follows: 1) The creation of a cavity of a suitable size in the solvent in order to accommodate the solute. This involves the endoergic breaking of solvent-solvent bonds and hence the Gibbs free energy change will be positive (AG = + ve). 2) The reorganisation of the bulk solvent molecules around the cavity, for which the Gibbs free energy change is assumed to be negligible. 3) The insertion of the solute into the cavity, setting up various solute-solvent interactions all of which are exoergic and liberate energy (AG = -ve).
23 Solute-solvent complex
Solute Solute Solvent Solvent
The various elements of this solvation process can be accounted for by quantitative solute descriptors which model the various interactions involved; these are the
Abraham descriptors, R 2, Ttz", Za]", ZPz", and Vx.
1.4.1 The Excess Molar Refraction, R 2
R2 a measure of polarizable n and n electrons and is the molar refraction of a solute in excess of the molar refraction of an alkane of the same characteristic volume It is calculated from the eqn :
R2 = MRx (observed) - MRx (alkane of same Vx) (22) where MRx is the solute molar refraction in units of (cm^ mor')/10, defined by :
MRx = 10f(T|)Vx (23)
T| is the refractive index taken at 293 K with the sodium D line and
(24)
Hence knowing f(r|) and Vx for any solute, MRx and the R 2 can be calculated. This method is only applicable to liquids. However, since MR is an additive property, it is reasonable to assume that excess molar refraction is also additive and can therefore be
24 calculated by summing relevant fragments. By definition, R 2=0 for all n-alkanes, and by calculation R 2=0 for branched chain alkanes and for the rare gases.
1.4.2 Solute Hydrogen-Bond Acidity, Abraham and co-workers established a scale of solute hydrogen-bond acidity, based on log K values for the 1:1 hydrogen-bond complexation reaction :
A-H-kB ^ A-H...B (25)
Both the acid and base were present at low concentrations so that they were present in solution as monomeric, unassociated solutes.
The log K values for a series of acids against 45 reference bases in an inert solvent (tetrachloromethane) at 298K were set up
log K (series of acids against reference base B) = Lelog Ka^ + Dg (26)
Where Lg and Dg characterize the base and log Ka^ characterize the series of acids.
An example of one of the 45 equations is that for a series of acids against the reference base dimethylacetamide:
log K (acids against dimethylacetamide) = 1.1706 log Ka^ + 0.8165 (27) n = 35, r^ = 0.9964, SD = 0.0748
In order that a general hydrogen-bond acidity scale be set up, it was required that a plot of log K (against reference base x) vs. log K(against reference y) yield a straight line. By plotting the log K values for one set vs. another set, it was evident that the lines intersected at a ‘magic point’ which was found to be -1.1. This enables all compounds with zero hydrogen-bond acidity to be assigned log Ka^ = logK = - 1.1, where K is in molar concentrations. The origin was then shifted from -1.1 to zero and compressed by converting log Ka^ to in order that all compounds with zero hydrogen bond-acidity have an value of zero.
25 = log (Ka “ + l,l)/4.636 (28)
Hence the 45 equations gave rise to a general scale of hydrogen-bond acidity which was thermodynamically related to Gibbs energies of complexation.
1.4.3 Solute Hydrogen-Bond Basicity, P 2 ” In an analogous manner to that for the derivation of solute hydrogen-bond acidity, a scale for solute hydrogen-bond basicity was constructed by Abraham and co-workers. In this case, a series of 34 bases was plotted against a reference acid to yield a point of intersection which was again at -l.I. This led to 34 eqns of the form :
log K(series of bases against reference acid A) = LAlogKe^ + Da (29)
where La and Da characterize the given reference acid and log K^b characterizes the base. An example of one of the 34 equations is that for a series of bases against the reference acid 4-chlorophenol :
log K(bases against 4-chlorophenol) = 1.065 log K^b 4- 0.074 (30) n= 38, SD = 0.054
The origin was again shifted and compressed into a more convenient scale
P2“ = (logKB” +l.l)/4.636 (31)
The combination of the Kz" and Pz" scales leads to a simple equation which correlates and estimates the log K values for the 1:1 complexation in tetrachloromethane at 298K.
log K = 7.354 0C2" X P2" - 1.094 (32) n = 1312, r^ = 0.9956, SD = 0.09, F = 148535
This equation can be used to calculate further values of « 2" and p 2^ given that the logK value is known, and also the value of either a-^ or p 2^- However, these
26 equations are not completely general, and a few acid-base combinations had to be excluded. These “exceptions to the general rule” were combinations of pyridine and ether bases with weak nitrogen acids (e.g. anilines, indole) and also carbon acids (e.g. chloroform, bromoform). However, there are no problems associated with these acids in combination with other bases and so these are included in the scale.
There exists a connection between hydrogen-bond acidity or basicity and full proton transfer acidity or basicity within families. However, the relationships collapse when applied across families of solutes. An example of this is the fact that for phenol is the same as for carboxylic acids (0.60 and 0.61 respectively), yet phenol is 10'^ less as strong a proton acid in water. Also, for EtsN (0.67) is smaller than for DMSO (0.78), yet the former is a strong proton base in water, and the latter is a very weak proton base.
1.4.4 Effective Solute Scales of , Zpz" and 712 ”
Since the (X2" and P2" parameters refer to 1:1 complexation, it was not clear whether they would be applicable in situations where a solute molecule is surrounded by solvent molecules and hence undergo multiple bonding. In order to test this, Abraham and co-workers used the original 0(2" and P2" scales to set up “effective” or
“summation” scales of hydrogen-bond acidity, Zcx 2^ and hydrogen-bond basicity,
ZP2" and determined Zoc2^ and Sp 2^ for polyfunctional compounds.
In general it was found that for simple molecules, the parameters based on 1:1 equilibrium constants could be used as solute parameters even in cases where the solute was surrounded by excess molecules. Such processes include the solubility of gases and vapours in liquids and the partition of solutes between liquid phases. However, there were exceptions, and a number of these representative monofunctional solutes are shown in Table 1.2
27 Table 1.2 Comparison of effective solute descriptors and those based on 1:1 equilibrium constants
Solute b " n-heptane 0.00 0.00 0.00 0.00 Water 0.38 0.35 0.35 0.82 Ethanol 0.44 0.48 0.33 0.37 Phenol 0.22 0.30 0.60 0.60 Aniline 0.38 0.41 0.26 0.26 Acetonitrile 0.44 0.32 0.09 0.07
Polyfunctional solutes cannot be dealt with by this method and so are back-calculated using multiple linear regression equations. Abraham and co-workers used the McReynolds data set which contained Vq^ values for up to 376 solutes on up to 77 stationary phases at 393K. Since 75 of these stationary phases had no hydrogen-bond acidity, the P 2^ term could be dropped from the General Solvation Equation leaving :
log Vg^ = c -h rR2 -I- sn + aai^ + I Log (33)
This gave an equation for each of the stationary phases, where the constants c, r, s, a and / were determined by multiple linear regression analysis (MLRA) using known values of the solute parameters R 2,7i*, 0(2" and log for as many solutes as possible.
The resulting equations were then placed into a vertical matrix in which
Vn= log V^G(n) - Cn became the dependent variable and the constants rn , 5n , «n , and
/n (where n = 1-75) were the explanatory variables. This inverse matrix enabled the unknown solute coefficients R 2, n , and log to be evaluated by MLRA. Since the input data was now entirely related to properties of the solute, % was replaced by 7t2^ which is an experimentally obtained parameter.
Vn = R2rn 4- sn + n'^s^ + a 2 ^an + Log k (34)
An initial analysis was carried out by forcing the regression equation through the origin. Various solutes were studied, each giving rise to reasonable values of R 2, n ,
28 (X2^ and log R 2 is either known or can be easily calculated, and so this explanatory variable was incorporated into the independent variable, thus yielding the eqn :
log V°G(n) - Cn - rnR 2 = + log In (35)
The regression equation was again forced through the origin. The solvation parameters obtained by this inverse MLRA method are in effect averages for a given solute over 30-70 stationary phases. The 7t2^ values obtained are for solute molecules surrounded by an excess of solvent molecules, and so may be more correctly denoted as Z7t2^.
The analysis of Patte et al.^^ retention data for 240 solutes on five non-acidic stationary phases yielded five regression equations of the form :
log L’ = c + rR 2 + S7i2 + aot 2^ +1 Log L^^ (36) where log L’ = log L - log L(decane), which affects only the constant in the regression equations. The reverse MLRA could not be applied to this data set because for a given set of solutes the reverse matrix requires at least fifteen stationary phases. The determination of the unknowns by means of simultaneous equations proved unsuccessful, probably because the range of constants in the solvation equations were not markedly different. However, from the 240 x 50 data matrix. Patte et al. had derived five characteristic solute parameters. Abraham et al.^® were able to correlate these characteristic solute parameters against the Abraham solute parameters to obtain
712^. Alternatively if R 2, logL^^ and E« 2^ are known for all five phases, 712^ for the given solute can be taken as the average of the five back-calculated results. Once R 2, 712^ and logL^^ are known, measurements on very basic phases can lead to the determination of Conversely, measurements of acidic phases through the full log L^^ solvation equation can yield Zp 2^
A complication was noted by Taylor and co-workers^^ who identified a number of solutes for which Sp 2^ altered depending on the solvent system. For certain solutes such as anilines, substituted anilines, alkylpyridines and sulfoxides, a modified Zp 2^
29 descriptor denoted as is necessary for partitions from water into organic solvents that contain a high proportion of water at saturation 20
1.4.5 McGowen’s Characteristic Volume, Vx Originally, the characteristic volume used to measure the endoergic cavity term was
the bulk solute molar volume, V 2. This was calculated as the molecular weight divided by the liquid density at 298K. It was found necessary to add lOcm^mol'^ to V 2 for
aromatic and acyclic compounds, giving an adjusted molar volume V 2(adj)- However,
V2 and V2(adj) are not true solute parameters since they are measured as a bulk property. A further disadvantage is that the use of these terms is restricted to solutes that are liquid at room temperature
A preferred alternative was the computer-calculated intrinsic volume, Vi of Leahy. These intrinsic volumes could be calculated for both liquids and solids and furthermore led to better correlations when used instead of V 2 or V 2(adj)- However, the set of intrinsic solute volumes, Vx of McGowan were chosen because they are more easily calculated than Vi.
Vx is calculated by simply adding up the atom contributions (Table 1.3) for the compound and then subtracting 6.56cm^mol'' for each bond irrespective of whether it is a single, double or triple bond.
Vx = S all atom contributions - (6.56*B) (37)
The Vx values are arbitrarily divided by 100 to scale them to values similar to the other solute descriptors in the General Solvation Equation. Calculating the number of bonds present is simplified by using an algorithm derived by Abraham :
B = N -1+Rg. (38) where B = number of bonds N = the total number of atoms Rg = the total number of ring structures.
30 Table 1.3 Characteristic atomic volumes, Vx in cm^mol^ c 16.35 N 14.39 0 12.43 F 10.48 H 8.71 Si 26.83 P 24.87 S 22.91 Cl 20.95 B 18.32 Ge 31.02 As 29.42 Se 27.81 Br 26.21 Sn 39.35 Sb 37.74 Te 36.14 I 34.53
Abraham and McGowan found that Vi and Vx yielded similar solvation equations and coefficients. The two sets of volumes are very well correlated.
Vi = 0.597 + 0.6823VX (39) n = 209, r = 0.9976, SD = 1.24
1.4.6 Estimation of Descriptors using Group Contribution Approach More recently, a group contribution scheme called ABSOLV has been successfully obtained to estimate the Abraham descriptors The basis of this method were the 31 fragments of Klopmann which were defined as SMARTs strings. A C program was then written to read molecules as SMILES strings and to count the number of matches of each of the 31 fragments within a molecule. This program was then applied to all the molecules contained in the Abraham database. The resulting counts were then used as independent variables in a standard least squares regression against all available descriptors.
The model was refined by identifying classes of molecules that were poorly modelled by the Klopmann fragments, and then modifying the fragment set in order to deal with the large residuals. This approach of refining the Klopmann fragments proved to be satisfactory for five of the six descriptors. In total, 81 fragments were required in order to estimate these Abraham descriptors and resulted in an improved fit. In the case of the hydrogen-bond acidity descriptor, 2%^, the results were unsatisfactory (Table 1.4) and hence it was required that an entirely new set of fragments be developed. The most fundamental acidic atom types such as OH, NH and NH 2 were used as the cornerstone of the method, and the model was then refined as previously, with corrections to and distinctions between these fundamental acid types iteratively defined until a satisfactory regression was obtained. This iterative procedure identified
31 a separate set of 51 fragments which was required to adequately predict Separation of the data into a training set and test set showed the predictive ability of the training set was almost as accurate as the full regressions (Table 1.5).
The obvious advantage of ABSOLV is the ability to estimate descriptors rapidly with the knowledge of just compound structure. These descriptors may then be used to predict physicochemical and biological processes for which Abraham solvation equations have already been constructed.
Table 1.4 Results of regression using 81 modified parameters to estimate Abraham descriptors Descriptor r\adj) RMSE F statistic N Min. Max. R2 0.979 0.093 2574.4 3375 -1.37 4.62 0.921 0.163 570.3 2875 -0.54 4.15 ZOz"* 0.881 0.083 465.7 3692 0.00 2.10 ZPz" 0.914 0.121 417.7 2541 0.00 3.20 0.919 0.123 435.6 2568 0.00 4.52 0.989 0.242 3333.5 1947 -0.83 29.98
* Results were considered to be unsatisfactory RMSE: Root mean square error, RMSE=[S(calc-obs)^/n]°’^
Table 1.5 Results of training and test set regressions using Group Contribution approach Descriptor Set r^(adj) RMSE n Rz Training 0.978 0.093 3308 Test 0.976 0.099 67
712“ Training 0.922 0.164 2818 Test 0.954 0.122 57 Zctz" Training 0.945 0.058 3619 Test 0.943 0.055 73 ZPz" Training 0.908 0.121 2491 Test 0.918 0.108 50 Training 0.903 0.129 2517 Test 0.929 0.131 51 Training 0.990 0.240 1908 Test 0.994 0.180 38
32 1.5 Multiple Linear Regression Analysis (MLRA)
Multiple linear regression was the first and foremost statistical method applied in QS AR studies. This method has been popularized by Hansch to relate bioactivity data to the measures of lipophilic, electronic, and steric properties of a series of derivatives. Multiple linear regression is a regression model that involves more than one regressor variable.
In general, a response y may be related to k regressor variables :
y = (3o + Pi Xi + ^ 2^2 + ...... + Pk^k + E (40) where x j, j = 0 , 1, ...... , k represent k different independent variables,
Po is the intercept (value when all the independent variables are 0 ),
p j, j = 1, 2 ,...... , k represent the corresponding k regression coefficients, e is a random error with mean zero and variance
Multiple linear regression models are often used as approximating functions. That is, the true functional relationship between y and xi, X 2, Xk is unknown, but over certain ranges of the regressor variables the linear model is an adequate model.
In the case of the General Solvation Equation, we have :
log SP = c + r.R 2 + sK'2 ^ + fl.Eoc2^ + Z?.EP2^ + v.Vx (20)
This is a multiple linear regression model with five regressor variables. The equation is a linear function of the unknown parameters r, 5, a ,b and v. The model describes a plane in five-dimensional space of the regressor variables R 2, 7t2^, Zp 2^ and Vx. The parameter c is the intercept of the regression plane. The parameter r indicates the expected change in log SP per unit change in R 2 when 712^, E a2^, Zp 2^ and Vx are held constant. Similarly, 5 measures the expected change in log SP per unit change in 712^ when the other regressor variables are held constant and so forth. The regression coefficients r, s, a, b, and v are estimated by the method of least squares which is used to fit a straight line of best fit to the scattered plot.
33 The least squares estimate of the line of best fit is that which leads to aminimum value for the sum of squares for the deviation of the observations of y from the predicted line. It is an important assumption of regression analysis that the independent variables have minimal error. In the case of the General Solvation Equation there may be error in the explanatory variables which are either experimentally obtained or are calculated. However, the results will not be greatly affected because the descriptors will be consistent within whichever method they are obtained. Furthermore, in structure-biological activity correlations, the error in determining structural descriptors can generally be expected to be low as compared to the error of biological data.
A number of statistics are derived in conjunction with multiple linear regression calculations that allow the statistical significance of the resulting correlation to be assessed; the correlation coefficient (r), the standard deviation (SD), the root mean square error (RMSE), the average absolute error (AAE), the average error (AE), the student’s t-test (t), and the Fisher E-statistic (F).
The correlation coefficient, r : The Pearson correlation coefficient quantifies the variation in the data and measures how closely the observed data tracks the fitted regression line. Errors in either the model or in the data will lead to a bad fit.
1/2
^ A I Yi-y, r = 1 ^ (41) 2 I y , - y ^
where yi and yi are the measured and predicted values of the dependent variable respectively, and ÿ is the mean (Ey/n). The square of the Pearson correlation coefficient, r^ is the coefficient of determination or percentage of data variance accounted for by the regression equation. This varies from 0 if the points are completely random, to a correlation of 1.0 for perfect fit of the line to the data. It is convenient to convert the value to a percentage, thus when r^ = 0.90 the regression equation explains 90% of the variance. It is very important to consider the correlation
34 coefficient in relation to the number of compounds in the data set and also to the nature of the data itself. It may be reasonable to consider correlation coefficients above 0.90 to be satisfactory (accounting for above 81% of data set), and those above 0.95 as very satisfactory (accounting for above 90% of data set) for physical data such as HPLC retention measurements. However, such an assertion would not be rational for comparison to a biological assay in which the correlation quality would be expected to be poorer due to the intrinsic error associated with the experimental technique. Furthermore, as the size of the data set is decreased by removing outlying data points, the correlation coefficient is invariably improved.
The standard deviation, SD : Residual standard deviation is an expression of the modelling error and is dependent upon the number of samples in the data set(n) and also the number of descriptors(p) used in the model.
/A \ 2 -, 1/2 E y i - y i (42) SD = . - n - 1 - p
Standard deviation has the same units as the property being measured and becomes a more reliable expression of precision as n gets larger. It is a means of assessing the quality of the regression, and is also used in assessing the significance of deviant points and highlighting possible erroneous data. As the SD decreases, the correlation improves by producing tighter fits of the data. Standard deviation is relative to the numerical range of the data set and to the magnitude of the y dependent variable.
Root Mean Square Error, RMSE : The RMSE is the square root of the average of the squared differences between predicted and measured responses and is a direct measure of the prediction error of the model given in the same units.
1 / 2 A Z yi- y, RMSE = n (43)
35 Average Absolute Error, AAE : The average absolute error is sometimes used, and is a measure of the average absolute difference between predicted and measured responses. It is always smaller than the corresponding SD value, and is defined as
f A
AAE = Y i- Yi (44) n V y
Average error, AE : Average error gives an indication of whether an equation is biased i.e. whether it systematically over- or under- predicts the desired response variable. The nearer the average error is to zero, the more unbiased the model.
Yi- Yi (45) AE =
Studentt-test, t : In multiple linear regression analysis, the student’s t-test is performed on each independent variable as a significance test and to assess the relative merits of correlation coefficients. It is used to assess the likelihood of the correlation arising purely by chance.
n - 2 1/2 t = r (46) 1 -
The T-test assumes that the distribution of errors in the data is normally distributed. Generally it is accepted that a correlation is good enough if on average it could only have arisen by chance one time in 20 or better (5% significance level). Hence, the t- test is usually set at a confidence limit of 95% but can be raised depending on the accuracy of the test required. The level of significance of each variable indicates its individual importance in the correlation, and as such the t-test also aids in the removal of insignificant variables. A common rule of thumb is to drop from the equation all variables not significant at the 0.05 level or better.
36 Fis her *3 F-statisticj F : Given the assumption that the data has a Gaussian distribution, the F-statistic or the Fisher statistic is used to assess the statistical significance of association of the regressors with the dependent variable. The F-statistic gives an indication of the quality of the regression, the higher value of F, the more significant it will be for given degrees of freedom and the better is the regression. By definition it is the ratio of the variation in y explained by the model to the unexplained residual variation in y:
where r is the correlation coefficient, n is the number of data points and v is the number of independent variables.
1.5.1 Difficulties with MLRA In a multiple regression if two or more of the explanatory variables are nearly linear combinations of each other, the variables are multicollinear. The existence of such dependence implies that it is almost impossible to vary one of these variables while holding the others constant. Multicollinearity can make the calculations required for the regression unstable. Although it may be possible to find a good multiple linear fit for the Y variable, the values of the individual coefficients may be highly variable in that unexpectedly large estimated standard errors for the coefficients of the independent variables may be produced. Thus, Y may be predicted with reasonable accuracy, but it would not be possible to draw any reliable conclusions about the coefficients. Furthermore, fitted coefficients could vary widely from sample to sample of the data, or if a single independent variable is added or deleted from the equation.
In terms of the data used for multiple linear regression analysis, both the quality and quantity needs to be taken into account. To obtain meaningful and statistically significant coefficients, a wide spread of explanatory variables is required. Furthermore, as a rule of thumb, there should be no more than one descriptor per 5 compounds in the regression. If more descriptors are used, the predictive capability of the model is liable to be significantly affected since the extra descriptors are more likely to be describing random error in the Y variable.
37 1.6 References
1. Brown, A.C.; Fraser, T.R. On the connection between chemical constitution and physiological action. Part.l.- On the physiological action of the salts of the ammonium bases, derived from
strycnia, Brucia, Thebaia, Codea, Morphia and Nicotia. Trans. R. Soc. Edinburgh 1868, 25, 151- 203.
2. Meyer, H. Arch.Exp.Pathol.Pharmakol. (Naunyn-schmiedebergs). 1899, 42, 109 and 1901, 46, 338. 3. Overton, E. Studies uber die Narkose, Fischer, Jena, Germany, 1901. 4. Hammett, L.P. The effect of structure upon the reactions of organic compounds. Benzene
derivatives. J. Amer.Chem.Soc. 1937, 50, 96-103. 5. Hammett, L.P. Physical Organic Chemistry, chapter 11, McGraw-hill, New York, 2°^ edn. 1970. 6. Taft, R.W. Steric effects in Organic Chemistry. Newman, M.S. ed. Wiley : New York, 1956, 556-675. 7. Isaacs, N. Physical Organic Chemistry, 2“^ ed.; Longman:Essex, England 1995. 8. White, D.P.; Anthony, J.C.; Oyefeso, A.O. Computational measurement of steric effects: The size of organic substituents computed by Ligand Repulsive Energies. J. Org. Chem. 1999, 64, 7707- 7716.
9. Hansch, C. A quantitative approach to biochemical structure. Ace. Chem. Res. 1969, 232-239. 10. Hansch, C. Quantitative structure-activity relationships and the unnamed science. Acc. Chem. Res. 1993, 26, 147-153. 11. Ferguson, J. Relations between thermodynamic indices of narcotic potency and the molecular structure of narcotics. In Mechanisme de la Narcose, 1951, 25-39, Paris (Colloques Internationaux du Centre National de la Recherche scientifique. 12. King, L.A. Ferguson’s Principle and the prediction of fatal drug levels in blood. Human Toxicol. 1985,4, 273-287. 13. Hansch, C.; Maloney, P.P.; Fujita, T.; Muir, R.M. Correlation of biological activity of
phenoxyacetic acids with Hammett substituent constants and partition coefficients. Nature 1962, 178-180 14. Kubinyi, H. QSAR: Hansch Analysis and Related Approaches. Methods and principles in Medicinal Chemistry, vol.l, R. Mannhold, P.Krogsgaard-Larsen, H.Timmerman, eds., VCH, Weinheim, 1993. 15. Raevsky, O.A.; Grigor’ev, V.Y.; Kireev, D.B.; Zefirov, N.S. Complete thermodynamic
description of H-bonding in the framework of multiplicative work. Quant. Struct.-Act. Relat. 1992, 11,49-63. 16. Li, J.; Zhang, Y.; Dallas, A.J.; Carr, P.W. Measurement of solute dipolarity, polarizability and hydrogen-bond acidity by inverse gas chromatography. J. Chromatogr. 1991, 550, 101-134. 17. Famini, G.R.; Kassels, R.J.; King, J.W.; Wilson, L.Y. Using theoretical descriptors in quantitative
structure-activity relationships - comparison with the molecular transform. Quant. Struct.-Act. Relat. 1991, 10, 344-349.
38 18. Famini, G.R.; Ashman, W.P.; Mickiewicz, Wilson, L.Y. Using theoretical descriptors in quantitative structure-activity relationships-opiate receptor activity by fentanyl-like compounds. Quant. Struct.-Act. Relat. 1992, 11, 162-170. 19. Famini, G.R.; Wilson, L.Y.; Penski, C.A. Using theoretical descriptors in quantitative-structure-
activity relationships- some physicochemical properties. J.Phys. Org. Chem. 1992, 5, 395-408. 20. Abraham, M.H. New solute descriptors for linear free energy relationships and quantitative
structure-activity relationships. In Quantitative treatments of solute/solvent interactions. Theoretical and computational chemistry, vol 1. Ed. Politzer, P.; Murray, J.S. 1994, Elsevier Science. 21. Charton, M. Steric effects 1. Estérification and acid-catalyzed hydrolysis of esters. J. Am. Chem.
Soc. 1975,97, 1552-1556. 22. Charton, M. Steric effects 7. Additional v constants. J.Org.Chem. 1976, 41, 2217- 2220. 23. Wiener, H. Structural determination of paraffin boiling points. J. Am. Chem. Soc. 1947, 69, 17-22. 24. Randic, M. On characterization of molecular branching. J. Am. Chem. Soc. 1975, 97, 6609-6615. 25. Kier, L.B. Quant. Struct.-Act. Relat. A shape index from molecular graphs. 1985, 4, 109-116. 26. Kier, L.B. Shape indexes of orders one and three from molecular graphs. Quant. Struct.-Act. Relat. 1986, 5, 1-7. 27. Kier, L.B. Quant. Struct.-Act. Relat. Distinguishing atom differences in a molecular graph shape index. 1986, 5,7-12. 28. Verloop, A. The STERIMOL Approach to Drug Design, Marcel Dekker, New York, 1987. 29. Hansch, C.; Leo, A.J. Substituent constants for Correlation Analysis in Chemistry and Biology, John Wiley & Sons, New York, 1979 30. Livingstone, D.J. The characterization of chemical structures using molecular properties. A survey. J. Chem. Inf. Comput. Sci. 2000, 40, 195-209. 31. Stuper, A.J.; Brugger, W.E.; Jurs, P C. Chemometrics : Theory and Application. Ed. Kowalski, B.R. American Chemical Society, Washington, D C. 1977.
32. Wold, Svante. Pattern Recognition by means of Disjoint Principal Components Models, Pattern Recognition, 1976, 8, 127. 33. Klopman, G.; Dimayuga, M.L. Computer automated structure evaluation (CASE)of the
teratogenicity of retinoids with the aid of a novel geometry index. J. Comput.-Aided Mol. Design. 1990,4, 117-130. 34. Kubinyi, H. ed. 3D QSAR in Drug design. Theory, methods and applications, Escom, Leiden, 1993. 35. Ferguson, A.M.; Heritage, T.; Jonathon, P.; Pack, S.E.; Phillips, L.; Rogan, J.; Snaith, P.J. EVA
:A new theoretically based molecular descriptor for use in QSAR/QSPR analysis. J. Comput.- Aided Mol. Des. 1997, 11, 143-152. 36. Turner, D.B.; Willet, P.; Ferguson, A.M.; Heritage, T. Development and validation of the EVA descriptor for QSAR studies. Abstr. Pap. Am. Chem. Soc. 1997, 214, 158-COMP, part 1.
39 37. Todeschini, R.; Gramatica, P. 3D-modelling and prediction by WHIM descriptors 5. Theory, development and chemical meaning of WHIM descriptors. Quant. Struct.-Act. Relat. 1997, 16, I13-II9. 38. Taft, R.W.; Abraham, M.H.; Doherty, R.M.; Kamlet, M.J. The molecular properties governing
solubilities of organic nonelectrolytes in water. Nature, 1985, 313, 348-346. 39. Taft, R.W.; Abraham, M.H.; Famini, G.R.; Doherty, R.M.; Abboud, J.-L.M., Kamlet, M.J. Solubility properties in polymers and biological media 5. An analysis of the physicochemical properties which influence octanol water partition- coefficients of aliphatic and aromatic solutes. J. Pharm. Sci. 1985, 74, 807-814. 40. Abraham, M.H. Scales of hydrogen-bonding - Their construction and application to physicochemical and biochemical processes. Chem. Soc. Rev. 1993, 22, 73-83. 41. Platts, J.A.; Abraham, M.H. Partition of volatile organic compounds from air and from water into
plant cuticular matrix : An LFER analysis. Environ. Sci. Technol. 2000, 34, 318-323. 42. Abraham, M.H.; Chadha, H.S.; Dixon, J.P.; Rafols, C.; Treiner, C. Hydrogen-bonding 40. Factors
that influence the distribution of solutes between water and sodium dodecyl-sulfate micelles. J. Chem. Soc., Perkin Trans. 2. 1995, 887-894. 43. Abraham, M.H.; Chadha, H.S.; Dixon, J.P.; Rafols, C.; Treiner, C. Hydrogen-bonding 41. Factors that influence the distribution of solutes between water and hexadecylpyridinium chloride micelles. J. Chem. Soc. Perkin Trans. 2. 1997, 19-24. 44. Abraham, M.H.; Chadha, H.S.; Martins, F.; Mitchell, R.C.; Bradbury, M.W.; Gratton, J.A. Hydrogen bonding -Part 46: A review of the correlation and prediction of transport properties by an LFER method iphysicochemical properties, brain penetration and skin permeability. Pesticide Sci. 1999, 55,78-88. 45. Abraham, M.H.; Martins, F.; Mitchell, R.C. Algorithms for skin permeability using hydrogen bond descriptors : The problem of steroids. J. Pharm. Pharmacol. 1997, 49, 858-865. 46. Abraham, M.H.; Roses, M. Hydrogen-bonding 38. Effect of solute structure and mobile-phase composition on reversed-phase high performance liquid chromatographic capacity factors. J. Phys. Org. Chem. 1994, 7, 672-684. 47. Abraham, M.H.; Roses, M.; Poole, C F.; Poole, S.K. Hydrogen-bonding 42. Characterization of
reversed-phase high-performance liquid chromatographic C-18 stationary phases. J. Phys. Org. Chem. 1997, 10, 358-368. 48. Poole, S.K.; Poole, C.F. Chromatographic models for the sorption of neutral organic compounds by soil from water and air. J. Chromatog. A. 1999, 845, 381-400. 49. Abraham, M.H.; Weathersby. Hydrogen-bonding 30. Solubility of gases and vapours in biological liquids and tissues. J.Pharm. Sci. 1994, 83, 1450-1456. 50. Abraham, M.H.; Poole, C F.; Poole, S.K. Classification of stationary phases and other materials by gas chromatography. J. Chromatogr. A. 1999, 842, 79-114. 51. Abraham, M.H.; Kumarsingh, R.; Cometto-Muniz, J.E.; Cain, W.S. An algorithm for nasal pungency thresholds in man. Arch. Toxicol. 1998, 72, 227-232.
40 52. Abraham, M.H,; Kumarsingh, R.; Cometto-Muniz, J.E.; Cain, W.S.; Roses, M.; Bosch, E.; Diaz, M.L. The determination of solvation descriptors for terpenes, and the prediction of nasal
pungency thresholds. J. Chem. Soc,, Perkin Trans. 2, 1998, 2405-2411. 53. Abraham, M.H.; Kumarsingh, R.; Cometto-Muniz, Cain, W.S. A Quantitative Structure-Activity Relationship for a Draize eye irritation database. Tax. In Vitro 1998, 12,201-207 54. Abraham, M.H.; Kumarsingh, R.; Cometto-Muniz, Cain, W.S. Draize eye scores and eye
irritation thresholds in man can be combined into one QSAR. Annuls of the New York Academy of
Sciences. 1998, 855, 652-656. 55. Uhlig, H.H. The solubilities of gases and surface tension. J.Phys. Chem. 1937, 41, 1215-1225. 56. Pierotti, R.A. Scaled particle theory of aqueous and non-aqueous solutions. Chem. Rev. 1976, 76, 717-726.
57. Abraham, M.H.; Liszi, J. J.Chem.Soc.,Faraday Trans. CalculationsI. on ionic solvation.l.Free- energies of solvation of gaseous univalent ions using a one-layer continuum model. 1978, 74, 1604-1614 58. Abraham, M.; Whiting, G.S. Hydrogen Bonding. Part 13. A new method for the
characterization of GLC stationary phases - The Laffort data set. J. Chem. Soc. Perkin Trans.2, 1990,2, 1451-1460. 59. Abraham, M.H. Scales of solute hydrogen-bonding: Their construction and application to physicochemical and biochemical processes. Chem. Soc. Rev., 1993, 22, 73-83. 60. Abraham, M.H.; Whiting, G.S.; Doherty, R.M.; Shuely, W.J. XVI. A new solute solvation
parameter, Ttz", from gas chromatographic data. J. Chromatography, 1991, 587, 213-228. 61. Patte, P.; Etcheto, M.; Laffort, P. Solubility factors for 240 solutes and 207 stationary phases in gas-liquid chromatography. Anal. Chem. 1982, 54, 2239-2247. 62. Abraham, M.H. Hydrogen-bond descriptors for solute molecules. Computational approaches in supramolecular chemistry, 63-78. Ed. Wipff, G. 1994 Kluwer academic publishers. Netherlands. 63. Leahy, D.E.; Morris, J.J.; Taylor, P.J.; Wait, A.R. Model solvent systems for QSAR 3. An LSER
analysis of the critical quartet - new light on hydrogen-bond strength and directionality. J.Chem. Soc. Perkin Trans. 2. 1992, 705-722. 64. Abraham, M.H.; McGowan, J.C. The use of characteristic volumes to measure cavity terms in reversed phase liquid chromatography. Chromatographia, 1987, 4, 243-246. 65. Platts, J.A.; Butina, D.; Abraham, M.H.; Hersey, A. Estimation of molecular linear free energy
relation descriptors using a group contribution approach. J. Chem. Inf. Comput. Sci. 1999, 39, 835-845. 66. Klopman, G.; Li, J.Y.; Wang, S.; Dimayuga, M. Computer automated log P calculations based on an extended group contribution approach. J. Chem. Inf. Comput. Sci. 1994, 34, 752-781. 67. Klopman, G.; Wang, S.; Balthasar, D.M. Estimation of aqueous solubility of organic molecules by the Group Contribution Approach. Application to the study of biodégradation. J. Chem. Inf. Comput. Sci. 1992, 32, 474-482.
41 Chapter 2 Aims of the Present Work
The present work aims to provide algorithms which will aid in drug design by isolating or suggesting new chemical entities which will have improved physicochemical properties. The Abraham General Solvation Equation will be used to achieve this aim; the obvious advantage being that the equation coefficients are interpretable and enable characterization of the system under investigation.
log SP = C + r.R2 + 5.712^ + 6f.Z(%2^ + Z?.Sp2^ + v.Vx
Here, SP is a biological or chemical property of a series of solutes in a system, R 2 is the excess molar refraction, 712^ is the solute dipolarity/polarizability parameter, Za 2^ and ZP2" are the solute hydrogen-bond acidity and hydrogen-bond basicity summation parameters respectively, and Vx is McGowan’s characteristic volume.
Although ABSOLV (Chapter 1.4.6, page 31) is available to predict descriptors, the method is only reliable for fragments upon which the group contribution scheme is based. A different technique will be required to deal with novel fragments/compounds that may be encountered; one aspect of the work will be to evaluate potential methods. Two possible routes are : i. Descriptor assignment using gas-solvent and water-solvent partition measurements in conjunction with a least-squares procedure. ii. Utilisation of reversed-phase HPLC as an alternative to conventional water- solvent partition measurements. Any descriptors that are obtained from these methods may then be incorporated into the Abraham UCL Database to enhance the solute physicochemical parameters available. Once compound descriptors are available, the Abraham General Solvation Equation may be applied to predict the solute property of interest.
The majority of drugs are taken in the form of tablets and capsules. Thus, as a first step, the solid must dissolve so that its concentration in solution is sufficient to exert
42 the required pharmacological effect. In this respect, the aqueous solubility of the drug is of fundamental importance. Since traditional methods of solubility measurement are time consuming, a reliable computational method to predict this property would be invaluable to deal with the increasing numbers of drug candidates available. To this end, it is hoped that literature data will form the basis of a solvation equation which will be able to predict water solubility with a reasonable degree of accuracy. Furthermore, the equation should not contain a melting point term, thus enabling prediction based simply on compound structure.
For an orally bioavailable drug, absorption must take place before the compound may be transported to its site of action. Intestinal absorption therefore plays an integral role in determining the amount of drug that is made available to the target tissues. As such, an extensive number of in vitro and in situ techniques have aimed to provide a suitable preliminary surrogate as a marker for in vivo data. However, these experimental methods are again subject to time constraints and require a reasonable large quantity of compound. Thus, the concept of a QSAR method to offer a solution to this problem has rapidly gained momentum. The Abraham General Solvation Equation will be applied to a comprehensive dataset of drug compounds in order to isolate the physicochemical properties which are most influential to intestinal absorption in a quantitative manner.
A general scheme of the thesis is given in Fig. I
43 Fig. I General Scheme of thesis :
Chapter 8
Chapter 6 Chapter 7
The Abraham Descriptors Chapter 1 (physicochemical parameters)
LFER equations
Chapter 3 Chapter 4 Chapter 5
Determination Solvation Use of HPLC of Solute Properties of in Descriptor Descriptors Alkan-l-ols Determination
New Fragments
ABSOLV
44 Chapter 3 The Determination of Solute Descriptors
The first step in any QSAR analysis is to obtain molecular descriptors which will form the basis of solute predictions. In the case of the Abraham method, it is necessary for six descriptors to be determined for each solute if both eqns 1 and 2 are to be utilised. The calculation of Vx presents no problem because this can easily be obtained from atomic contributions. In the case of liquids, R 2 may be calculated from the liquid refractive index, whilst for solids R 2 may be obtained from addition of relevant fragments. The remaining descriptors 712^, and Ep 2^ can be determined by a method based on water-solvent and gas-solvent partitions which will be described in this chapter.
3.1 Method for Descriptor Determination
When considering partitioning of a solute between two liquid phases, the Abraham General Solvation Equation takes the following form. log P = c + r.R 2 + s.TC'^ + (3.2(X2^ + è.Sp 2^ + v.Vx (I) where log P is the partition coefficient of a series of solutes between water and the given solvent. For partition of solutes between the gas phase and solvents, the alternative, eqn 2 is used
log = c + r.R 2 + s.Ti'^ + fl.2 (X2^ + 6 .2 ^2^ + /.L (2) where is the gas-solvent partition coefficient, or Ostwald Solubility coefficient for a given solvent. The independent variables in eqns 1 and 2 have been set up and explained in Chapter 1.4, pages 22-31.
In addition to direct measurements, the partition coefficient of a solid between water and a solvent phase can be obtained from the solubility of the solid in water, Sw (mol dm'^\ and the solvent, S (mol dm'^), provided that the same solid phase is in equilibrium with the saturated solutions in water and the given solvent^'^.
45 log p = log s - log Sw (3)
The solubility of the compound together with available solubilities in a variety of nonaqueous solvents can therefore be used to deduce log P values for the partition of the compound from water to the respective solvents.
The values of log for the given solute in various solvents can then be obtained from the gas phase concentration of the compound at 298K, Cq (mol dm'^) and log Sw through eqns 4-6,
log Cg = log VP - log RT (4) log = log Sw - log Co (5)
Log = log + log P ( 6 ) where R is the gas constant (dm^atm K'^), T is the temperature (K), and VP is the solid saturated vapour pressure (atm).
Numerous water-solvent and gas-solvent partition equations have previously been obtained; a selection of which are shown in Tables 3.1 and 3.2. These equations form an integral part of an in-house Excel spreadsheet called Lpcalc which contains the coefficients for each of the equations. The descriptors R 2 and Vx which have been calculated are added into the spreadsheet along with the partition measurements of the compound in as many solvents as are available. The three unknown descriptors are then calculated using ‘Solver’ which is an add-in available with the Microsoft Office Excel software; the descriptor values being those that reproduce the observed partition measurements with the minimum standard deviation possible. However, Solver merely obtains descriptor values which ensure the best answer statistically, but certainly the user needs to use some intuition in deciding whether the descriptors are reasonable given compound structure and functional groups present. Certain data values may be highlighted as erroneous in nature and these will need to be removed before the standard deviation minimisation step is repeated. This type of data refinement should lead to descriptors which are valid in terms of chemistry. Thus, using the predetermined equations in Tables 3.1 and 3,2, it is possible to obtain
46 descriptors for a particular compound if either partition measurements or solubilities are available in a sufficient number of solvents.
Table 3.1 Coefficients in eqn 1 for partition between water and solvents
Solvent c r s a b V 1,2-Dichloroethane 0.227 0.278 -0.167 -2.816 -4.324 4.205 Acetonitrile 0.262 0.471 0.107 -1.377 -4.358 3.526 Benzene 0.142 0.464 -0.588 -3.099 -4.625 4.491 Chlorobenzene 0.040 0.246 -0.462 -3.038 -4.769 4.640 Chloroform 0.205 0.194 -0.412 -3.319 -3.455 4.403 Cyclohexane 0.159 0.784 -1.678 -3.740 -4.929 4.577 Dibutyl ether 0.184 0.817 -1.495 -0.830 -5.090 4.694 Diehl oromethane 0.314 0.001 0.022 -3.238 -4.137 4.259 Diethyl ether 0.255 0.605 -1.096 -0.097 -5.000 4.381 Diisopropylether 0.197 0.695 -1.220 -0.238 -4.912 4.388 Ethyl acetate 0.253 1.157 -1.397 -0.054 -3.755 3.726 Ethylene glycol 0.336 -0.075 - 1.201 -3.786 - 2.201 2.085 Hexadecane 0.087 0.667 -1.617 -3.587 -4.869 4.433 n-Butyl acetate -0.468 0.712 -0.397 0.010 -3.743 3.865 Nitrobenzene -0.181 0.576 0.003 -2.356 -4.420 4.263 Octanol 0.088 0.562 -1.054 0.034 -3.460 3.814 Oleyl alcohol -0.359 -0.270 -0.528 -0.035 -4.042 4.204 Olive Oil -0.011 0.577 -0.800 -1.470 -4.921 4.173 PGDP 0.139 0.376 -0.715 -1.034 -4.852 4.245 Tetrachloromethane 0.260 0.573 -1.254 -3.558 -4.588 4.589 Toluene 0.143 0.527 -0.720 -3.010 -4.824 4.545 Tributylphosphate 0.015 0.804 -0.862 1.389 -4.647 4.129 Trichloromethane 0.327 0.157 -0.391 -3.191 -3.437 4.191 Pentane 0.369 0.386 -1.568 -3.535 -5.215 4.514 Hexane 0.361 0.579 -1.723 -3.599 -4.764 4.344 Heptane 0.325 0.670 -2.061 -3.317 -4.733 4.543 Octane 0.223 0.642 -1.647 -3.480 -5.067 4.526 Butanone, dry 0.354 0.003 -0.164 -0.979 -4.706 4.160 Dimethylformamide, dry 0.105 0.317 0.462 1.154 -4.843 3.757 Dimethylsulfoxide, dry -0.231 0.520 0.757 1.799 -4.652 3.428 Ethanol, dry 0.208 0.409 -0.959 0.186 -3.645 3.928 Methanol, dry 0.329 0.299 -0.671 0.080 -3.389 3.512 Gasphase -0.994 0.577 2.549 3.813 4.841 -0.869
47 Table 3.2 Coefficients in eqn 2 for partition between the gas phase and solvents
Solvent c r s a b I
1,2-Dichloroe thane 0.011 -0.150 1.436 0.649 0.736 0.936 Benzene 0.107 -0.313 1.053 0.457 0.169 1.020 Chlorobenzene 0.053 -0.553 1.254 0.364 0.000 1.041 Chloroform 0.116 -0.467 1.203 0.138 1.432 0.994 Cyclohexane 0.163 - 0.110 0.000 0.000 0.000 1.013 Dibutyl ether 0.145 0.074 0.250 2.621 0.000 1.019 Dichloromethane 0.121 -0.450 1.677 0.404 0.786 0.940 Diethyl ether 0.245 -0.360 1.079 3.316 0.000 0.889 Heptane 0.281 -0.211 0.000 0.000 0.000 0.990 Hexadecane 0.000 0.000 0.000 0.000 0.000 1.000 Hexane 0.307 -0.227 0.000 0.000 0.000 0.976 Isopropyl ether 0.131 0.024 0.396 3.402 0.000 0.963 Nitrobenzene -0.273 0.039 1.803 1.231 0.000 0.929 Octanol -0.198 0.002 0.709 3.519 1.429 0.858 Olive oil -0.230 0.009 0.795 1.353 0.000 0.888 Tetrachloromethane 0.282 -0.303 0.460 0.000 0.000 1.047 Toluene 0.121 -0.222 0.938 0.467 0.099 1.012 Water -1.271 0.822 2.743 3.904 4.814 -0.213 Nonane 0.200 -0.145 0.000 0.000 0.000 0.980 Pentane 0.335 -0.276 0.000 0.000 0.000 0.968 Hexane 0.292 -0.169 0.000 0.000 0.000 0.979 heptane 0.275 -0.162 0.000 0.000 0.000 0.983 Octane 0.215 -0.049 0.000 0.000 0.000 0.967 Decane 0.156 -0.143 0.000 0.000 0.000 0.989 Acetonitrile, dry -0.042 -0.122 2.138 2.383 0.452 0.745 Butanone, dry 0.171 -0.453 1.694 2.699 0.000 0.891 DMF, dry -0.161 -0.189 2.327 4.756 0.000 0.808 DMSO, dry -0.619 0.131 2.811 5.474 0.000 0.734 EGLY,dry -0.937 0.215 1.511 4.651 2.591 0.571 Ethanol, dry 0.012 -0.221 0.819 3.636 1.249 0.854 Methanol, dry -0.004 -0.215 1.173 3.701 1.432 0.769 TEE, dry -0.133 -0.611 1.457 1.899 4.461 0.633
48 3.2 Example ; Descriptors for Vinclozolin
Solubility data in water and organic solvents will be employed to obtain descriptors for the pesticide vinclozolin to provide an example of how descriptors may be determined for a compound which is solid at room temperature. R 2 was obtained from the hypothetical refractive index which was calculated using ACD Chemsketch, version 2. Vinclozolin does not contain any hydrogen-bond donor groups and as such, the situation is somewhat simplified since E a 2^ is constrained to be zero. The McGowan’s characteristic volume is calculated from atomic constants and the number of bonds. This leaves 7t2^, Zp 2^ and logL^^ to be calculated if processes correlated through both eqn 1 and 2 are to be considered. The vapour pressure of 0.016mPa was obtained from the ARS pesticide database"^ and the water solubility of 2.6 mg dm'^ taken from the SRC Physprop database^. Application of eqns 3-6 resulted in 9 indirectly determined log P values and 9 logL^ values from the original solubility data. A directly determined water-octanol partition value was also available for this compound^, although this was highlighted as an approximate value. Furthermore, an additional two LFERs are available where the dependent variable is log (specified as Gas phase and Water in Tables 3.1 and 3.2 respectively). This resulted in a total of 22 eqns, see Table 3.3.
log VP -9.80 log Sw -4.93 log Co" -11.19 > lo g L ^ 6.27 O
Table 3.3 Solvent solubilities of vinclozolin, (S in mol dm'^), and derived partition coefficients
Solvent log Sobs logP: logL^^ Benzene -0.236" 4.689 10.954 Cyclohexane -1.394" 3.531 9.796 Dichloromethane 0 .220^ 5.262 11.527 Ethanol -1.207" 3.718 9.983 Heptane -1.803^ 3.238 9.503 Methanol -1.269^ 3.773 10.038
49 Toluene -0.419^ 4.622 10.888 Gas phase 6.265 6.265 Chloroform (not used) -0.124" 4.801 11.066 Diethyl ether (not used) -0.511" 4.415 10.680 Octanol (not used) 3.100 9.365 ^Calculated as log VP- logRT (eqn 4) Calculated as log S w - log Cg (eqn 5) Solubility taken from ref 4; data converted from units of ppm using density of respective solvent ^ Solubility taken from ref 5 ® Calculated as log S - logSw (eqn 3) '^Calculated as logL^ + log? (eqn 6)
Through Lpcalc and Solver, the unknown descriptors 712^, Zpi" and logL^^ are assigned by a least-squares procedure that minimises the difference between observed and calculated log S, log P and log values. The log? and log values for chloroform, diethyl ether and octanol were anomolous and were omitted. The subsequent best-fit solution of the remaining 16 equations in log? and log yielded the following values: 712^(1.553), ZP 2" (0.7814), logL^^ (9.630). How well these descriptors, together with R 2 = 1.948 and Vx = 1.8449, reproduce the log? and log and values is shown in Table 3.4. For reference, if the value of S a 2^ had been left to float. Solver would have assigned very similar descriptors, viz 712^(1.569), (-0.022) ZPz" (0.791), logL’^9.667).
Table 3.4 Observed and calculated values of log P and log for vinclozolin Solvent log Fobs logPcalc calc- log log calc- obs L^obs Locale obs Benzene 4.689 4.804 0.115 10.954 11.087 0.133 Cyclohexane 3.531 3.673 0.142 9.796 9.704 -0.092 Dichloromethane 5.262 4.985 -0.277 11.527 11.515 -0.012 Ethanol 3.718 3.914 0.196 9.983 10.053 0.070 Heptane 3.238 3.112 -0.126 9.503 9.426 -0.077 Methanol 3.773 3.700 -0.073 10.038 9.923 -0.115 Toluene 4.622 4.667 0.045 10.888 10.968 0.080 Gas to water 6.265 6^:68 0.000 6.265 6.301 0.036
50 3.3 Descriptors for Diazepam Analogues
Under circumstances in which only one or two partition measurements are available for a compound, the method previously detailed is somewhat limited. However, if a parent compound exists for which there is enough data available, descriptors for analogues may then be assigned through deduction. Whilst this may lead to slight errors, the method ensures consistency in descriptors within the group of compounds. This method has been applied to obtain descriptors for two series of drug compounds - diazepam analogues and p-blockers.
For diazepam, R 2 and Vx were calculated in the same way as previously described for vinclozolin. Again, = 0. Water-solvent partition data was obtained from the Medchem 2000 database^; data was available in five different solvents. Since a value for the vapour pressure could not be found, only 712^ and ZP2" may be determined from the available data, i.e. log cannot be derived. The descriptors are able to reproduce the logP values well, and since five partition measurements have been used to obtain two descriptors there is confidence that the values are appropriate.
Diazepam log P values
R2 = 2.078 Solvent obs calc Octanol 3.12 3.11
CHCI3 4.45 4.30
Benzene 3.69 3.54
Heptane 1.96 1.94
PGDP 2.36 236
ClogP 3.29
R2 712“ Z ai" ZPz" Vx
Final descriptors : 2.078 1.48 0 1.30 2.0739
* refers to the hypothetical liquid refractive index calculated using Chemsketch version!
51 From the descriptors determined for diazepam, it is possible to assign descriptors for analogues by addition/subtraction of appropriate fragments. These descriptors are used to calculate log Poet, and if required, refinement is then carried out so that the descriptors best reproduce the available measured log Poet or ClogP value (calculated log Poet using Clog P for Windows, version 2.0.0b, Biobyte Corp). This method has been applied to the compounds that follow.
Prazepam
obs calc
log Poet 4.14
ClogP 4.30
Rz Zotz" Vx
Diazepam 2.078 1.57 0 1.30
+ Cyclopropane 0.408 0.23 0 0
Final descriptors : 2.486 1.71 0 1.30 2.3880
52 Iclazepam
obs calc
log Poet 4.01
ClogP 4.21
Rz Saz” Zpz" Vx Prazepam 2.486 1.71 0 1.30
+ Ether linkage 0 0.22 0 0.45
2.486 1.99 0 1.75
Final descriptors : 2.486 1.91 0 1.69 2.7285
Pinazepam
HC obs calc
log Poet 3.29
ClogP 3.26
Rz Ttz” Saz” ZPz" Vx Diazepam 2.078 1.48 0 1.30
+ CH2-C=CH 0.183 0.25 0.12 0.12
Final descriptors : 2.261 1.73 0.12 1.42 2.2697
53 Tetrazepam
obs calc
log Poet 3.89
ClogP 1.88
R2 Vx
Diazepam 2.078 1.48 0 1.30
+ (1-Me-cyclohexene -0.210 -0.34 0 -0.05 - toluene)
Final descriptors : 1.868 1.14 0 1.25 2.1599
Menitrazepam
obs calc
log Poet 3.06
ClogP 2.92
R2 %2" Zocz" ZP2* Vx
Tetrazepam 1.868 1.14 0 1.25
NO] for Cl 0.153 0.46 0 0.21
Final descriptors : 2.021 1.60 0 1.46 2.2117
54 Nimetazepam
obs calc
log Poet* 2.16 2.18
ClogP 2.34
R2 Zaz" Z b " Vx Diazepam 2.078 1.48 0 1.30
NOz for Cl 0.153 0.46 0 0.21
Final descriptors : 2.230 1.94 0 1.51 2.1257
Flunitrazepam
obs calc
log Poet* 2.06 2.26
ClogP 2.13
R2 %2** Saz” ZPz" Vx Nimetazepam 2.230 1.94 0 1.51
+ (PhF - benzene) -0.133 0.05 0 -0.04
Final descriptors : 2.097 1.99 0 1.47 2.1434
55 Fludiazepam
obs calc
log Poet 2.75 3.00
ClogP 2.96
Rz Ttz" Zaz" Vx Flunitrazepam 2.097 2.06 0 1.50
ClforNOz -0.153 -0.46 0 -0.21
Final descriptors : 1.947 1.60 0 1.29 2.0916
Flutoprazepam
obs calc
log Poet 4.19
ClogP 3.93
Rz Zotz" ZPz" Vx Fludiazepam 1.947 1.60 0 1.29
4- Cyclopropane 0.408 0.23 0 0
Final descriptors : 2.348 1.83 0 1.29 2.4057
56 Flurazepam C21H23CIFN3O
obs calc
log Poet* 3.20 3.14
ClogP 4.16
R2 712“ Za2^ Z k " Vx Fludiazepam 1.950 1.60 0 1.29
+ Triethylamine 0.101 0.15 0 0.79
Final descriptors : 2.051 1.75 0 2.08 2.8959
2-Oxoquazepam
obs calc
log Poet * 3.36 338
ClogP 4.24
R2 712“ Z(%2** ZP2” Vx Fludiazepam 1.950 1.60 0 1.29
+ C F 3 -CH 2 -0.280 0.20 0.10 0
Final descriptors : 1.670 1.80 0.10 1.29 2.2856
57 Proflazepam
HO obs calc
MO' log Poet * 1.51 1.52
ClogP 1.58
R2 Z(%2* ZP2* Vx
Fludiazepam 1.940 1.60 0 1.29
-k CH3-CHOH-CH2OH 0.373 0.90 0.58 0.80
2.310 2.50 0.58 2.09
Final descriptors : 2.310 2.50 0.47 1.95 2.4908
The values of Z 0C2" and obtained from the simple addition of fragments are too high to reproduce the available log Poet value for proflazepam. This discrepancy is due to the potential for hydrogen-bonding between the two hydroxyl groups and also between the hydroxyl group and the carbonyl group (see Fig 3.1 - Graphics created using PC Model version 7, Serena software, 1999). Due to this intramolecular hydrogen-bonding, the ability of proflazepam to participate in intermolecular hydrogen-bond interaction with the solvent will therefore be reduced; this should be reflected in both S a 2^ and 1 ^2^ which theoretically should be lower-tlian-t)ie.valn£S prWlcW. from the addition of relevant fragments which assume no intramolecular bonding.
58 Fig 3.1 (A) H-bonding between two hydroxyl groups
(B) H-bonding between hydroxyl group and carbonyl group
(A)
(B)
O Oxygen atom O Chlorine atom
® Nitrogen atom O Fluorine atom
[ ] Hydrogen-bond
59 Six directly determined water-partition measurements were also available for nitrazepam from the Medchem 2000 database^. These are very useful and required because nitrazepam differs from diazepam in that there exists a protonated nitrogen atom in the seven-membered ring; a value for Z a 2^ must therefore be assigned. The experimental data in conjunction with Lpcalc and Solver assigned descriptors which reproduced the water-partition values with an SD of 0.17 log units. This error is perfectly acceptable considering that the measurements were undertaken by different researchers using varying protocols. The value of Za2^=0.29 is certainly reasonable if it is compared with that for N-methylaniline for which 2a2^=0.17 (taken from the Abraham UCL Database). The increased hydrogen-bond acidity in nitrazepam may be ascribed to the nitro-group in the para-position of the benzene ring which has an enhanced electron-withdrawing effect.
Nitrazepam log P values
Solvent obs calc
Octanol 2.12 2.25
HN CHCI3 2.42 2.56
Benzene 1.59 1.65
Hexane -0.10 -0.32
O Et20 1.57 1.32
CHCI2 1.91 1.97
ClogP 2.31
R2 712H Za2H Vx
Final descriptors : 2.230 1.60 0.29 1.44 1.9848
60 Clonazepam
HN obs calc
log Poet * 2.41 2.57
ClogP 2.37
R2 Ttz" Zaz" z k " Vx Nitrazepam 2.230 1.60 0.29 1.44
+ (2 -chlorotoluene 0.160 0.13 0 -0.07 - toluene)
2.390 1.73 0.29 1.37 2.1072
Final descriptors : 2.390 1.80 0.29 1.44 2.1072
Meclonazepam
obs calc HN
CH log Poet* 2.72 3.00 ClogP 2.89
R2 712^ Zctz" ZPz" Vx Clonazepam 2.390 1.80 0.29 1.44
+ CH] 0 0 0 0.03
Final descriptors ; 2.390 1.80 0.29 1.47 2.2481
61 Nordazepam
HN obs calc
log Poet * 2.93 TÔT
ClogP 3.01
R2 Zotz" Vx Nitrazepam 2.230 1.60 0.29 1.44
Cl for NO 2 -0.150 -0.46 0 -0.21
2.080 1.13 0.29 1.23
Final descriptors : 2.080 1.21 0.29 1.28 1.9330
Nortetrazepam
calc HN obs log Poet 3.68
clogP 3.73
R2 712 “ Vx
Nordazepam 2.08 1.21 0.29 1.28
+ (1-Me-cyclohexene -0.21 -0.34 0 -0.05 - toluene
Final descriptors ; 1.87 0.87 0.29 1.23 2.019
62 Delorazepam
HN obs calc
log Poet* 3.15 3.34
ClogP 3.07
R2 Ttz" Zk" Vx Clonazepam 2.390 1.80 0.29 1.44
Cl for NO 2 -0.150 -0.31 0 -0.14
2.240 1.49 0.29 1.30
Nordazepam 2.080 1.21 0.29 1.28
+ C1 0.160 0.13 0 -0.07
2.240 1.34 0.29 1.21
Average 2.240 1.42 0.29 1.26 2.0554
63 Bromazepam
HN obs calc log Poet* 1.69 1.78
ClogP 1.69
R2 712^ Zotz" Vx Nitrazepam 2.230 1.60 0.29 1.44
Br for NO 2 0.011 -0.38 0 -0.19
+ (2-Me-pyridine -0.003 0.23 0 0.34 - toluene)
Final descriptors : 2.31 1.45 0.29 1.59 1.9445
Overall, the descriptors are as would be expected from the change in substituent. The introduction of more electronegative substituents increases the dipolarity/ polarizability of the compound and also the hydrogen-bond acidity. Conversely, hydrogen-bond basicity will be reduced depending on the strength of the electron- withdrawing effect. In the few cases where the descriptors do not produce the log P value with an acceptable degree of accuracy, the effect of substituents is less than would be expected from fragment addition. This may be attributable to cancelling effects, and for example it is difficult to assign the reduction in dipolarity/polarizability if two electron-withdrawing substituents are in close proximity; certainly the term will not be additive, because each substituent will influence the other.
64 3.4 Descriptors for B-blockers
A second set of descriptors was obtained in much the same way for a series of P- blockers. These compounds have the same chemical backbone and five partition measurements were available for each of two analogues - atenolol and alprenolol. As such, it was possible to obtain descriptors for these two compounds and then subtract fragments in order to obtain an average value for the backbone which was then used to calculate descriptors for subsequent p-blockers.
nACD^ 1 540 log P values Atenolol R2 = 1.330 Solvent obs calc Octanol 0.16 0.37
H3C CHCI3 -0.13 -0.17 CH. Benzene -1.85 -1.90 NH HgN- Toluene -2.16 -2.28 \ O y OH Ethylacetate -0.07 -0.20
ClogP -0.11
H R2 712 Zocz" Zpz" Vx
Final descriptors : 1.330 2.04 0.76 1.92 2.1763
log P values Alprenolol n^^°= 1.518 R2 = 1.089 Solvent obs calc Octanol 2.89 3.18
CHCI3 3.73 3.98 CH Benzene 2.63 2.73
HN CH Cyclohexane 1.47 1.40 HO Et2 0 2.74 2.52
ClogP 2.65
H 712 ZPz"" Vx
Final descriptors : 1.089 1.57 0.29 1.27 2.1587
65 Chemical backbone of analogues (*)
CH
HN CH HO
Rz 712® Eaz" SPz” Vx Atenolol 1.330 2.04 0.76 1.92 2.1763
+ (toluene -0.350 -0.75 -0.44 -0.75 - PhCHzCONHz) 0.980 1.29 0.32 1.17
Alprenolol 1.089 1.57 0.29 1.27 2.1587
+ (benzene -0.107 -0.08 0 -0.08 - allylbenzene) 0.980 1.49 0.29 1.19
Average - Final 0.980 1.39 0.31 1.18 descriptors for (*)
66 Metoprolol Water/solvent partition Solvent obs calc
Octanol * 1.88 1.86 H, CH, Dichloroethane 2.23 1.78 HN CH HO ClogP
R2 712^ ZPz" Vx Fragment (*) 0.982 1.39 0.31 1.18
+ (l-phenyl- 2- 0.020 0.30 0 0.45 methoxyethane - benzene)
Final descriptors : 1 . 0 0 0 1.69 0.31 1.63 2.2604
Practolol
H3C obs calc CH. log Poet* 0.79 0.78 CH. NH ClogP 0.76 HN (! ^ ---- 0 OH
R2 712 ” ZPz" Vx Fragment (*) 0.982 1.39 0.31 1.18
+ (acetanilide 0.260 0.84 0.46 0.55 - benzene
Final descriptors : 1.242 2.23 0.78 1.73 2.1763
67 Betaxolol
H3C obs calc CH, NH log Poet * 2.81 2.87
ClogP 2.17 OH
y
R2 712*^ ZPz"" Vx Fragment (*) 0.982 1.39 0.31 1.18
+ (l-phenyl- 2 -methoxyethane 0.020 0.30 0 0.45 - benzene)
+ Cyclopropane-benzene 0.408 0.23 0 0
1.410 1.92 0.31 1.63
Final descriptors : 1.410 1.96 0.31 1.67 2.5745
Moprolol
obs calc H3C log Poet* 1.69 1.76
ClogP 1.12 CHNH
OH
R2 712^ Zaz" ZP2'' Vx Fragment (*) 0.982 1.39 0.31 1.18
+ (Anisole - benzene) 0.098 0.23 0 0.15
1.080 1.62 0.31 1.33
Final descriptors : 1.080 1 . 6 6 0.31 1.37 1.9786
68 Pindolol
obs calc
log Poet * 1.75 1.98
ClogP 1.67
R2 712^ Zotz" Zk" Vx Fragment (*) 0.982 1.39 0.31 1.18
+ (Indole - benzene) 0.590 0.60 0.44 0.08
1.570 1.99 0.76 1.26
Final descriptors : 1.570 2.04 0.76 1.31 2.0090
Procinolol
obs calc
log Poet * 3.01 3.03
NH CH ClogP 2.52 OH
Rz 712^ Zotz" z k " Vx Fragment (*) 0.982 1.39 0.31 1.18
+ (Cyclopropane propylbenzene 0.196 0.13 0 0.01 - benzene)
Final descriptors : 1.178 1.52 0.31 1.19 2.0931
69 Propranolol
H3C obs calc CH, log Poet* 2.98 3.08 NH ClogP 2.75 OH
R2 Ttz" z k " Vx
Fragment (*) 0.982 1.20 0.31 1.20
+ (Naphthalene 0.730 0.40 0 0.06 - benzene) Final descriptors : 1.712 1.79 0.31 1.24 2.1480
Carazolol
HO obs calc HN. log Poet * 3.59 3.47
ClogP 3.06 HN
R2 2%^ Vx
Fragment (*) 0.982 1.20 0.31 1.20
+ (Carbazole -0.61 -0.52 0 -0.14 - benzene)
Final descriptors ; 2.159 2.37 0.67 1.28 2.1480
70 3.5 Conclusion
This work demonstrates that once descriptors have been calculated for a parent compound, substitution of functional groups is generally simply a matter of addition or subtraction of relevant fragments. However, this is only applicable to compounds in which intramolecular hydrogen-bonding is considered to be negligible. For compounds in which there is hydrogen-bonding between functional groups, a correction will need to be applied which will appropriately reduce both S a 2^ and
EP2^. The final descriptors for the analogues are shown in Table 3.5.
Table 3.5 Descriptors for diazepam analogues and P-blockers
Rz ZPz" Vx Diazepam 2.078 1.48 0 1.30 2.0739 Prazepam 2.486 1.71 0 1.30 2.3880 Iclazepam 2.486 1.91 0 1.69 2.7285 Pinazepam 2.261 1.73 0.12 1.42 2.2697 Tetrazepam 1.868 1.14 0 1.25 2.1599 Menitrazepam 2.021 1.60 0 1.46 2.2117 Nimetazepam 2.230 1.94 0 1.51 2.1257 Flunitrazepam 2.097 1.99 0 1.47 2.1434 Fludiazepam 1.947 1.60 0 1.29 2.0916 Flutoprazepam 2.348 1.83 0 1.29 2.4057 Flurazepam 2.051 1.75 0 2.08 2.8959 2-Oxoquazepam 1.670 1.80 0.10 1.29 2.2856 Proflazepam 2.310 2.56 0.47 1.98 2.4908 Nitrazepam 2.230 1.60 0.29 1.44 1.9848 Clonazepam 2.390 1.80 0.27 1.44 2.1072 Meclonazepam 2.390 1.80 0.27 1.47 2.2481 Nordazepam 2.080 1.21 0.27 1.28 1.9330 Delorazepam 2.240 1.42 0.27 1.26 2.0554 Nortetrazepam 1.870 0.87 0.27 1.23 2.019 Bromazepam 2.310 1.45 0.27 1.59 1.9445 Atenolol 1.330 2.04 0.76 1.92 2.1763 Alprenolol 1.089 1.57 0.29 1.27 2.1587 Metoprolol 1.000 1.69 0.31 1.63 2.2604 Practolol 1.242 2.23 0.78 1.73 2.1763 Betaxolol 1.410 1.96 0.31 1.67 2.5745 Moprolol 1.080 1.66 0.31 1.37 1.9786 Pindolol 1.570 2.04 0.76 1.31 2.0090 Procinolol 1.178 1.52 0.31 1.19 2.0931 Propranolol 1.712 1.79 0.31 • 1.24 2.1480 Carazolol 2.159 2.37 0.67 1.28 2.1480
71 3.6 References
1. Abraham, M.H.; Le, J.; Acree, Jr, W.E.; Carr, P.W.; Dallas, A J. The solubility of gases and vapours in dry octan-l-ol at 298k. Submitted to Chemosphere. 2. Abraham, M.H.; Le, J.; Acree, Jr, W.E.; Carr, P.W. Solubility of gases and vapours in propan-l-ol at 298K. J. Phys. Org. Chem. 1999, 12, 675-680. 3. Abraham, M.H.; Le, J.; Acree, Jr, W.E. The solvation properties of the aliphatic alcohols. Collect. Czech. Chem. Commun.1999, 64, 1748-1760. 4. Agricultural Research Service Pesticide Properties Database (ARS PPD) 5. SRC The Physical Properties (Physprop) Database, available via chemfinder.com.
6 . The Pesticide Manual, Eleventh Edition, Ed. Tomlin, C D S, The Crop Protection Council, 1997. 7. Leo, A.J. The Medchem Database 2000, Biobyte corporation and Pomona College in co-operation with Daylight.
72 Chapter 4 The Solvation Properties of the Aliphatic Alcohols
Although a number of methods are available for the correlation and prediction of the solubility of gases and vapours in water, there has been little attention paid to solubilities in other associated solvents such as alcohols \ This chapter aims to rectify this inadequacy and provides chemically interpretative algorithms to estimate vapour solubilities in a number of alcohols. The term ‘vapour’ is used to cover both permanent gases and vapours of compounds that are liquid or solid at 298K. In addition, an indirect method for obtaining water-to-dry alcohol partition coefficients is highlighted and utilised to yield a further set of predictive algorithms in several alcohols. These two sets of algorithms can, in the future, be used to determine compound descriptors by the method outlined in Chapter 3.
4.1 The Solubility of Gases and Vapours
The solubilities of gases and vapours in water and alcohols can be correlated through the General Solvation Equation, detailed in Chapter 1.4, pages 22-31.
log L = C + rR 2 + 5712^ + (ÆéQL'^ + bZP2^ 4- 1 log (1)
In eqn 1, L is the Ostwald solubility coefficient defined through eqn 2. If concentrations in the solvent and the gas phase are in the same units, for example mol dm'^, then L is a dimensionless quantity.
L = [conc. of solute in solution]/ [conc. of solute in the gas phase] (2)
There are three main methods of obtaining values of L in alcoholic solvents, i. Traditionally, the most common method is through vapour-liquid equilibrium measurements in which it is possible to obtain the activity coefficient of a solute in an alcohol. Extrapolation to zero solute concentration will yield the infinite dilution activity coefficient, y°. However extrapolation to zero solute concentration is not easy in the case of asymmetric mixtures, and the method is
73 difficult to apply to involatile solutes in volatile solvents\ Henry’s Law constant is given by eqn 3
K" = Y”.P° (3)
Here, P° is the solute saturated vapour pressure (taken as equivalent to fugacity). L may be calculated from eqn 3 since it is simply the inverse of K^, although due regard must be given to units. The conversion of units from atm to molar gas concentration is given by R.T where R is the gas constant and T the temperature. The conversion from mole fraction in solution to concentration in solution is given by lOOOp/MW, where p is the density and MW the molecular weight of the solvent respectively. Thus, the final equation relating L to K" is
R.T. 1000p K“.MW
where R is in units of dm^ atm. K '\ m ol'\ T is in K, p is in g.cm*^, is in atm and MW in g.mol'\ ii. L may also be obtained by a direct determination of the concentration of the solute in the headspace above a dilute solution of the solute in a given alcohol. Again, the most difficult systems to study by this method are those of a relatively involatile solute in a volatile solvent. This difficulty also arises in the case of the gas chromatographic method in which the solvent is the stationary phase iii. Due to the intrinsic difficulties associated with the previous two methods for volatile solvents such as water and methanol, some other method of obtaining L- values in alcohols, would be very useful. In this respect, an indirect method can be utilised for sparingly soluble compounds since can be obtained as the ratio of the molar concentration of solute in the saturated
solution, Salc , to the vapour concentration of the gaseous solute, Cq (mol dm'^) at 298K.
L^‘^ = Salc /C o or log = log Salc - log Cg (5)
The gas phase concentration, Cg is obtained from eqn 6
74 log Cg = logVP-logRT (6)
where VP is the pure liquid or solid vapour pressure, R is the gas constant and T is the temperature. Eqn 5 holds for both liquid and solid solutes, but there are two important provisos, (a) the solute must not be too soluble, because then the secondary medium activity coefficient of the solute in the saturated solution may be far away from unity, and (b) the solid or liquid in equilibrium with the saturated solution must be the unsolvated compound.
The values of log L in alcohols used for the correlation analyses in this chapter were derived from all three available methods and were mainly obtained from literature. However, the data derived for octan-l-ol were mainly obtained from experimental GC work by Dallas^ using the headspace analysis procedure.
In Table 4.01 are examples of calculated log L values for a number of solids in octan- l-ol. Data for these solids are important because the range of descriptor values is considerably increased. In a number of cases in which it is possible to compare these results of indirectly determined values of log from method (iii) to those which have been directly determined, the agreement is very good.
Table 4.01 calculation of log L in dry Octan-l-ol at 298K
Solute logSoct LogCo logL‘^'““ logL®''®" (indirect) (direct) Naphthalene -0.15 -5.34 5.19 Anthracene -1.91 -9.46 '' 7.55 Phenanthrene -0.45 -7.97 ” 7.52 7.57" Acenaphthene -0.59 -6.90 ” 6.31 Fluorene -0.62 -7.45 6.83 6.79" Fluoranthene -0.76 -9.37 ' 8.61 Pyrene -0.90 -9.65 '' 8.75 8.80" Biphenyl -0.13 -6.28 ' 6.15 1,4-Dichlorobenzene 0.20 -4.26 4.46 1,3,5-Trichlorobenzene -0.16'' -5.01 ' 4.85 1,2,3,5-Tetrachlorobenzene 0.15 -5.40 5.55 1,2,4,5-Tetrachlorobenzene -0.92 -6.54 « . 5.62 5.63®
75 Pentachlorobenzene -0.56 -7.05 ^ 6.49 6.27® Hexachlorobenzene - 1.86 -9.03 ® 7.17 6.90^ Hexamethylbenzene -0.89 -7.20 ' 6.31 1,4-Dibromobenzene -0.30 -5.51 ' 5.21 Hexachloroethane -0.28 -4.75 4.47 4-Ethoxyacetanilide -0.84 -10.43 " 9.59 Diphenylamine 0.03 -7.61 ' 7.64 trans-Stilbene - 1.1 0 ' -8.58 * 7.48 4-Hydroxybenzoic acid -0.17 -8.25 8.08 2-Hydroxybenzoic acid 0.17* -7.27 7.44 Methyl 4-hydroxybenzoate -O.lOj’*^ -8.67 8.57 Hexadecan-l-ol -0.03 -9.93 ° 9.90 Octadecan-l-ol -0.45 -11.38“ 10.93 Eicosan-l-ol -0.71 -12.77 “ 12.06 All data from ref 111 unless otherwise stated, Ref 4, Directly determined values from Ref 119, Ref 120, 'R ef 121, Ref 122, ® Directly determined values from Ref 106, Ref 123, 'R ef 59, jRef 86, •‘Ref 86, 'Ref 78, Ref 112, "Ref 113
Differing numbers of log L values were found for each of the alcohols investigated. The coefficients in the regression equations will be discussed and compared for all the alcohols together. Values of the solute descriptors, logL^ and the logL"^^ values are given at the end of the chapter in Tables 4.08 to 4.15 (pages 100-116).
For propan-l-ol, a total of 79 logL values could be obtained, covering a wide range of solutes. The required descriptors were available for all except fluromethane, and 3- methylpyridine was a pronounced outlier. Data for the remaining 77 solutes led to the correlation equation
logL"^" = -0.028 - 0.185 R 2 + 0.648 + 4.022 Z ai" + 1.043 ZPz" + 0.869 logL'* n = 77, = 0.9976, SD = 0.120, F = 6073 (7) where n is the number of data points (solutes), r is the correlation coefficient, SD is the standard deviation and F is the F-statistic. The statistical fit of eqn 7 is good, and suggests that the equation could be used to predict further values of log .
For butan-l-ol, 92 values of logL were assembled. Application of eqn 1 to the 92 logL values, leads to the correlation equation:
76 log L®“°“ = -0.039 - 0.276 R; + 0.539%" + 3.781 + 0.995 + 0.934 log L‘® n = 92, = 0.9966, SD = 0.158, F = 5099 ( 8 )
In the case of pentan-l-ol, 61 logL*"'®" values led to the equation: log L"'®" = -0.042 - 0.277 Rz + 0.526%" + 3.779 + 0.983 Zk" + 0.932 log L'* n = 61, = 0.9994, SD = 0.076, F =19143 (9)
For hexan-l-ol, heptan-l-ol and decan-l-ol, values could only be obtained for a relatively small number of 38-46 solutes. However, the statistical fit of all the equations was shown to be good, and as a result, eqns 7-12 could be used to predict further log L values for the respective alcohols. Even with the various methods for obtaining logL values, there was insufficient data available for nonan-l-ol and so a correlation for this solvent was not possible. log L"“ °" = -0.035 - 0.298 Rz + 0.626%" + 3.726 Z«z" + 0.729 Zgz" + 0.936 log L 16 n = 46, r" = 0.9996, SD = 0.089, F =18181 (10) log = -0.062 - 0.168 Rz + 0.429712" + 3.541 Zttz" +1.181 ZPz" + 0.927 log L 16 n = 38, r =0.9998, SD = 0.067, F = 23045 (11) log L°“ °" = -0.136 - 0.068 Rz + 0.325%" + 3.674Zaz" + 0.767 ZPz" + 0.947 log L 16 n = 45, r =0.9996, SD = 0.090, F = 15984 (12)
In the case of octan-lol, log values for 161 solutes were assembled covering a reasonably wide range of compound type. Since a large number of compounds was available, there was scope to investigate the possible prediction of logL°^^°" which would obviously enhance the value of the data compilation. A preliminary analysis revealed that five compounds were outliers, and so these were removed from the dataset. Four of the five outliers were the aliphatic aldehydes and since it is known that these compounds form aldols with aliphatic alcohols, their exclusion is warranted on chemical grounds. The other outlier was dimethylacetamide, but the log value for this compound is so large that there is certainly considerable experimental
77 uncertainty involved. The remaining 156 solutes were separated into a training set of 124 compounds and a randomly chosen test set of 32 compounds. Application of the solvation eqn 1 to the training set yielded the correlation equation;
log = -0.108 - 0.177 R 2 + 0.567712“ + 3.565 Z a 2“ + 0.703 + 0.932 log L'^ n= 124, = 0.9970, SD = 0.125, F = 7731 (13)
To assess the predictive capability of eqn 13, this algorithm was applied to an external test set of compounds. For the 32 data test set, values of log were predicted with a standard deviation between predicted and observed log values of only 0.131. The average absolute error is 0.085, and the average error is 0.009 log units. It is therefore clear that log values can be predicted through eqn 13 to around 0.13 log units, for any solute for which descriptors are available.
The training set and the test set are then combined to give the final and ‘best’ equation for the correlation and prediction of log values: log = -0.120 - 0.203R2 + 0.560%" + 3.576 2% " + 0.702 Zgz" + 0.939 log n = 156, r^ = 0.9972, SD = 0.125, F = 10573 (14)
Although the coefficients and statistics for eqn 14 are very similar to those for eqn 13, it is more advisable to use the former equation for further prediction of log values because it contains more data points and as such is more robust. Considering the thousands of solutes which are available in the Abraham UCL Database, it is clearly evident that this equation could be used to predict values for a large number of compounds. However, it must be expressed that the expected accuracy of the prediction will only be around 0.13 log units for compounds with descriptors that are within the descriptor space of the solutes used to set up the correlation equation, see Table 4.02.
78 Table 4.02 Descriptor space for the octan-l-ol regression, eqn 14
Descriptor Min Max R2 -0.06 2.81 Ttz" -0.26 1.72 Zcxz" 0.00 0.81 0.00 0.97 logL'" -1.74 10.69
The coefficients in eqns 7-12 and 14 can be compared with those previously obtained for the solution of gases and vapours in other alcohols, other non aqueous solvents and water (Table 4.03). The coefficients refer to solute transfer from the gas phase where there are no solute-solvent interactions at all to bulk solvents, and as such they can be regarded as “absolute” measurements of the various solvent properties Thus the coefficients may be compared quantitatively and also interpreted to deduce various physicochemical properties of the homologous series of alcohols.
79 Table 4.03 Coefficients in the log L equation for gas-solvent partitions at 298 K ^
Solvent c r s a b I n" r' SD F Water ^ -1.271 0.822 2.743 3.904 4.814 -0.213 Methanol ^ -0.004 -0.215 1.173 3.701 1.432 0.769 93 0.9952 0.130 3681 Ethanol ^ 0.012 -0.206 0,789 3.635 1.311 0.853 68 0.9966 0.140 3534 Propan-l-ol ^ -0.028 -0.185 0.648 4.022 1.043 0.869 77 0.9976 0.120 6073 Butan-l-ol ® -0.039 -0.276 0.539 3.781 0.995 0.934 92 0.9966 0.158 5099 Pentan-l-ol ® -0.042 -0.277 0.526 3.779 0.983 0.932 61 0.9994 0.076 19143 Hexan-l-ol ® -0.035 -0.298 0.626 3.726 0.729 0.936 46 0.9996 0.089 18181 Heptan-l-ol ^ -0.062 -0.168 0.429 3.541 1.181 0.927 38 0.9998 0.067 23045 Octan-l-ol ® - 0 .120 -0.204 0.564 3.582 0.694 0.939 156 0.9972 0.125 10571 Decan-l-ol ® -0.136 -0.068 0.325 3.674 0.767 0.947 45 0.9996 0.090 15984
Chloroform ^ 0.168 -0.595 1.256 0.280 1.370 0.981 150 0.9850 0.230 1919 DMF® -0.161 -0.189 2.327 4.756 0.000 0.808
Obs vs calculated log L values for solutes in each of the alkan-l-ols are presented in Tables 4.09 - 4.15
‘Reference 5, ^ Reference 1, Reference 5, This work and ref 4, ® This work and ref 6, Reference 7, ® Reference 8
80 Little change is evident in either the constant term or the r-coefficient (Fig.4.1) as the number of carbons is increased moving from methan-l-ol to decan-loi. The alkan-l- ols all display moderate-to-weak dipolarity/polarizability, as characterized by the s- coefficient which decreases slightly as the carbon number increases, although the value for octan-l-ol is rather out-of-line. This small decrease is as expected from the alcohol dipole moments that gradually decline in the same way, see Table 4.04. The a- coefficient, a measure of the solvent hydrogen-bond basicity, is almost constant amongst all the alcohols, at 3.715 ± 0.142; including water the average value is 3.734 ± 0.146. This is quite remarkable, considering that the fraction of OH groups decreases very considerably from methanol to decan-l-ol, and decreases even more from water to decan-l-ol. There is a gradual decline in the 6 -coefficient from methanol to decan-l-ol, but the value for heptan-l-ol is out-of-line. The magnitude of the coefficients indicate that the alkan-l-ols have very considerable hydrogen-bond basicity and moderate hydrogen-bond acidity.
Fig. 4.1 Coefficients of log L regressions for aliphatic alcohols
0-. O
0) 3 - 3 1 C 2 - .S>0
O1
23456789 10 Carbon number of alcohol
81 Table 4.04 Some properties of bulk solvents
Solvent Î1 ' MR ^ Pol " CED P' Water 1.333 3.71 1.45 549.0 1.85 Methanol 1.329 8.23 3.26 205.2 1.70 Ethanol 1.361 12.92 5.07 162.8 1.69 Propan-l-ol 1.384 17.49 6.77 143.2 1.68 f Butan-l-ol 1.399 22.15 8.79 129.5 1.66 f Pentan-l-ol 1.4101 26.82 10.63 119.8 1.70 f Hexan-l-ol 1.4178 31.55 12.48 113.1 1.65 f Heptan-l-ol 1.4249 36.28 14.32 108.4 1.67 f Octan-l-ol 1.4295 40.64 16.17 103.3 1.65 f Nonan-l-ol 1.4333 45.50 18.01 101.4 1.61 Decan-l-ol 1.4372 50.21 19.86 f 98.6 1.62 ecu 1.4604 26.43 10.47 73.8 0.00 f DMF 1.4305 19.92 7.48 138.9 3.82 Butanone 1.3788 20.68 8.19 86.0 2.76 Ethyl acetate 1.3722 22.25 8.89 f 79.2 1.78
^ Refractive index at 293K. Molar refraction in cm^mol V ^ Molecular polari ability from ref 9. ^ Hildebrand cohesive energy density ' Dipole moment. Molecular polari ability calculated by method used in
Comparison of the /-coefficients in Table 4.03 indicate that the alcohols behave no differently from standard aprotic organic solvents such as chloroform^ and N,N- dimethylformamide^ (DMF) towards nonpolar solutes. The /-coefficient increases with increasing alkyl chain length and hence hydrophobicity so that decan-l-ol is almost as hydrophobic as hexadecane for which l=\ at 298K. There is a reasonable plot (not given) of the /-coefficient vs the Hildebrand cohesive energy density, CED, again as expected. The most distinct observation from the coefficients in Table 4.03 is the extraordinary /-coefficient of water as compared to the values for all the non- aqueous solvents. The /-coefficient is due to (1) an unfavourable creation of a cavity to accommodate the solute in the solvent phase and ( 2 ) a favourable general dispersion term in which solute-solvent bonds are set up. The negative /-coefficient for water is because there is a greater increase in process ( 1) and a smaller increase in process (2) with increase in solute size. As a result, for solutions of gases and vapours in water, an increase in logL’^ (or size) invariably leads to a slight decrease in solubility. For all the non-aqueous solvents, the reverse applies in that an increase in
82 solute size will lead to a very large increase in solubility. Since the /-coefficient for methanol is large and positive, the unique result for water cannot be explained by the effect of self-association because both solvents are highly self-associated \
Thus overall, although one or two values are out-of-line, the coefficients in eqn 1 for solubility of gases and vapours in the alkan-l-ols fall into a reasonably coherent pattern. The small discrepancies are most likely due to different sets of solutes being used in the various equations.
The factors that influence the overall solubility of the gaseous solute can be shown through a term-by-term analysis of the various solvation equations. In Table 4.05 are the contributions to log L made by each term in the equation for solution in water, ethanol, decan-l-ol, and DMF. As already mentioned, water as a solvent is unique because the /-coefficient is negative. However, there is an additional pecularity of water in that the c constant is also very much more negative than for any non-aqueous solvent. Both highly negative values are probably a consequence of the cavity effects in water mentioned previously, and result in very low solubilities for any hydrophobic solute. Thus the hydrophobic solutes ethane and octane are poorly soluble in water, but much more soluble in the three nonaqueous solvents.
The alcohols behave quite similarly to aprotic nonaqueous solvents as regards solvation of hydrophobic solutes. In the case of acetic acid, which is polar, and both a hydrogen-bond acid and a hydrogen-bond base solute, the three polar terms aZot]" and 6ZP2") are all very positive and so considerably aid solution in water. However, opposing these polar terms are the /logL*^ and constant terms which are both highly negative and hence disfavour solubiltiy in water, with the result that the total logL value in water is hardly greater than that in ethanol, and only 0.56 log units greater than in decan-l-ol. For the basic solvent, DMF, both the aLaj^ and sU2 ^ terms are very large, leading to a very large calculated logL value indeed. The very large term makes the singular most considerable contribution in the solution of trimethylamine in water and is due to the interaction between the solute strong hydrogen-bond base and the solvent strong hydrogen-bond acid. However, because the unfavourable /logL^^ term and the constant term again play a part, the total logL
83 value for trimethylamine is smaller than that in ethanol, where the term is much less. All-in-all, the solvation of solutes in the alcohols more resembles solvation in nonaqueous polar solvents than solvation in water.
Table 4.05 A term-by-term analysis of solvation of gaseous solutes at 298K
rR2 5712^ 6ZP2" L log Total Total calc “ obs Solvent water Ethane 0.00 0.00 0.00 0.00 -0 .10 -1.37 -1.34 Octane 0.00 0.00 0.00 0.00 -0.78 -2.05 - 2.11 CH3CO2H 0.22 1.78 2.38 2.11 -0.37 4.85 4.91 MesN 0.11 0.55 0.00 3.22 -0.35 2.26 2.35
Solvent ethanol Ethane 0.00 0.00 0.00 0.00 0.42 0.43 0.44 Octane 0.00 0.00 0.00 0.00 3.14 3.15 3.17 CH3CO2H -0.06 0.52 2.22 0.55 1.49 4.74 ---- Me3N -0.03 0.16 0.00 0.87 1.38 2.37 2.67
Solvent decanol Ethane 0.00 0.00 0.00 0 .0 0 0.47 0.33 ---- Octane 0.00 0.00 0.00 0 .0 0 3.48 3.35 3.30 CH3CO2H -0.02 0.21 2.24 0.34 1.66 4.29 ---- Me3N -0.01 0.07 0.00 0.51 1.53 1.96 ----
Solvent DMF Ethane 0.00 0.00 0.00 0.00 0.40 0.24 0.22 Octane 0.00 0.00 0.00 0.00 2.97 2.81 2.81 CH3CO2H -0.05 1.51 2.90 0 .0 0 1.41 5.61 ---- Me3N -0.03 0.46 0.00 0 .0 0 1.31 1.58 1.77
^ Includes the intercept term o f -1.27 (water), 0.01 (ethanol), -0.14 (decan-l-ol) and -0.16 (DMF).
84 The present method yields very consistent results for the alkan-l-ols from methanol to decan-l-ol. Nevertheless, there is an oddity in the results of Table 4.03; the inference that water and alcohols have almost identical hydrogen-bond basicity (as shown by the ^-coefficient in Table 4.03). This finding is quite contrary to other measures of solvent hydrogen-bond basicity such as the Kamlet-Taft solvatochromie parametersA number of solvatochromie hydrogen-bond basicity values of water have been recorded, but in all cases the basicity of water is shown to be always less than that of the alcohols, see Table 4.06. Furthermore, from solvatochromie studies, the Kamlet-Taft hydrogen-bond acidity of water is similar to that of methanol but slightly greater than that of the higher alcohols. From this work, the ^-coefficients indicate that the hydrogen-bond acidity of water is very much larger than that of all the alcohols. There is little explanation for the discrepancy in results except to highlight the difference in the two measures of basicity; the solvatochromie parameters are spectroscopic in nature whilst the Abraham hydrogen-bond basicity parameter is Gibbs-energy related.
Table 4.06 Values of the Kamlet-Taft solvatochromie parameters for water and some alcohol solvents.
Solvent ai Pi water 1.02 0.14 0.18 0.31 0.42 1.16 0.43 Methanol 0.99 0.62 1.07 0.79 1.09 0.79 Ethanol 0.85 0.89
0.88 0.77 0.92 0.90
Octanol 0.70 0.86
Values in table taken from ref
85 4.2 Water-Alcohol Partitions
Having assembled the log values for solutes in the alkan-l-ol solvents, the corresponding water/alcohol partition coefficients can be calculated. The L values for the gases and vapours are combined with the respective values of the Ostwald solubility coefficient in water at 298K, to yield values for the transfer of solutes from pure water to pure alkan-l-ol, through eqn 15
pALCAV ^ L ^ lc / l W oj . log = log - log (15)
Although these logP values are for the hypothetical partition from pure water to the pure alkanol, they are useful in two ways, i they provide another measure of the solubility properties of the alkanols, this time with reference to water, and ii they can be compared to practical partitions from water to water-saturated alkanols.
These water-solvent logP values are correlated through the alternative solvation equation, eqn 16:
log P = C + rR2 + ■S'712^ + ûe2CC2^ + 62^2^ 4" V Vx (16)
where the final descriptor is the McGowan characteristic volume, Vx.
A summary of the correlation equations with logP as the dependent variable is given in Table 4.07. The statistical fits of the logP equations are always worse than those of the corresponding logL equations (Table 4.03). This is as expected, because most of the logP values have been obtained through eqn 15 and will be subject to errors in and an additional experimental error in log values. However, the correlation equations are reasonably self-consistent.
86 Again, a large number of data was available for octan-l-ol and it was possible to construct a training and test set for logPoct, before the data was combined to give the final equation shown in Table 4.07. The same training set of 124 compounds, and test set of 32 compounds was used as in the logL regression and the same five outliers were excluded.
log PoctoH = -0.192 - 0.373 Rz - 0.853 + 0.100 - 4.339 zPz" + 4.343 Vx n = 124, = 0.9949, SD = 0.162, F = 4387 (17)
When eqn 17 was applied to the test set compounds, the log PoctOH values were predicted with a standard deviation between predicted and observed log values of 0.162, average absolute error of 0 .112, and average error of 0.028 log units.
The coefficients can be interpreted in the same way as the coefficients in the logL equations, but now refer to the difference in the complementary water and the alcohol phases. The r-coefficient is positive, indicating that dispersive interactions involving the alcohols are more important than those involving water. The negative s- coefficients indicates that the alcohols are less dipolar/polarizable than water and since they are more polarizable, they must all be less dipolar than water. The r and s- coefficients are roughly constant along the series and this is due to the combination of dipolarity (greater for water and the lower alcohols) and polarizability (greater for the higher alcohols). The «-coefficient, which is a measure of the difference in hydrogen- bond basicity of the alkanol and water is essentially constant along the series and very close to zero. This is evidence to confirm the deduction that the alkanols and water have the same basicity (in the hydrogen bond sense). Considering that a completely nonacidic organic solvent such as hexadecane has a 6 -coefficient of -4.9, the magnitude of the negative 6 -coefficients for the alkanols indicates that they are very much weaker hydrogen-bond acids than water. The v-coefficient, rather like the /- coefficient in eqn 1 is a measure of the solvent hydrophobicity, but now relative to water as zero. The large negative value of the v-coefficient is in line with that observed for numerous water-solvent partitions'^'*^ and shows that the alcohols are somewhat hydrophobic. As expected, the larger the alcohol and the less water there is in the alcohol phase, the larger is the hydrophobicity.
87 Table 4.07 Coefficients in the logP equation for water- solvent partitions ^
Solvent c r s a b V n .....F " - - SD F [U20r
Dry solvents Methanol" 0.329 0.299 -0.671 0,080 -3.389 3.512 93 0.9880 0.160 1440 Ethanol^ 0.208 0.409 -0.959 0.186 -3.645 3.928 64 0.9809 0.170 1205 Propan-1-0^ 0.152 0.442 -1.093 0.386 -3.882 4.019 77 0.9950 0.136 2808 Butan-l-ol^ 0.076 0.374 -1.170 0.149 -3,920 4.258 92 0.9947 0.159 3259 Pentan-l-or 0.080 0.521 -1.294 0.208 -3.908 4.208 59 0.9960 0.112 2597 Hexan-l-of 0.044 0.470 -1.153 0.083 -4.057 4.249 46 0.9978 0.114 3775 Heptan-l-ol'^ -0.026 0.491 -1.258 0.035 -4.155 4.415 38 0.9972 0.081 2333 Octan-l-or -0.066 0.463 -1.043 0.020 -4.259 4.275 156 0.9949 0.144 5835 Decan-1-of -0.062 0.754 -1.461 0.063 -4.053 4.293 45 0.9980 0.123 3843
Wet solvents® Pentan-l-ol 0.175 0.575 -0.787 0.020 -2.837 3.249 40 0.9899 0.154 333 3.36 Hexan-l-ol 0.143 0.718 -0.980 0.145 -3.214 3.403 49 0.9854 0.167 289 3.32 Octan-l-ol 0.088 0.562 -1.054 0.034 -3.460 3.814 613 0.9974 0.116 23162 2.36 Decan-l-ol 0.008 0.485 -0.974 0.015 -3.798 3.945 51 0.9929 0.124 630 1.65 ^ Obs vs calculated log P values for solutes in each of the alkan-l-ols are presented in Tables 4.09 - 4.15
" Coefficients in eqn 16 for partition from water to dry methanol from Ref 1, * Ref. 5, This work Molar concentration of water in the water-saturated alcohol from Ref 13, ^ Ref. 3.
88 It is of some consequence to ascertain the effect of water on the solubility properties of the alkan-l-ols. Correlation equations for logP values in the practical partitions between water and a number of the higher alkan-l-ols'^ have previously been obtained. These practical partitions refer to partition between water saturated with the alkanol and the alkanol saturated with water. The solubility of the higher alkanols in water is comparatively small, and so differences in partition between water and dry alkanols, and practical partition will very largely be due to the water in the water- saturated alkanol.
There is no great change in the r- and «-coefficients on going from the dry to wet alcohols. Thus the presence of water in the wet alcohols has no effect on the hydrogen- bond basicity of the solvent which is probably attributable to the similarity in hydrogen-bond basicity of water and alcohols. There are larger effects on the other three coefficients which are shown in Fig.4.2. These coefficients for the water/dry alkanol partitions and those for the practical partitions previously studied (shown in Table 4.07) are plotted against the carbon number, N, of the alkan-l-ols. For convenience water is assigned zero coefficients (0,0) which is by definition corrrect since transfer will be from water to water itself. Although the position of N = 0 for water on the x-axis is arbitrary, the general conclusions as regards wet and dry alkanols are not affected.
Addition of water to the alcohols makes the 5-coefficient slightly less negative; that is, the wet alcohols are a little more dipolar than the dry alcohols. But similar addition of water makes the 6-coefficient, that reflects the difference in hydrogen-bond acidity of water and the alcohols, very much less negative, so that the wet alcohols are much stronger hydrogen-bond acids than the dry alcohols. An almost mirror-image effect is shown by the change in the v-coefficient; the wet alcohols have lower coefficients than expected from consideration of the dry alcohols, and so are less hydrophobic/lipophilic than expected. It is very clear that as the length of the alcohol chain increases, the solubility of water in the alcohol decreases, and the effect of water levels off; therefore the coefficients for the water/dry alkanol partitions approach those for the water/wet alkanol partitions. Indeed for partitions into decan-l-ol, the two sets of coefficients are statistically indistinguishable. Thus for the various solutes
89 used to construct the solvation equations, the solvation properties of dry decan-l-ol and wet decan-l-ol are practically the same.
Fig. 4.2 Plot of the coefficients in eqn 16 for partitions between water and alcohols
5
4
3
2 5 1 I 0 1 1 Ü 2
3
4
5 0 2 4 6 8 10 Carbon number
Coefficients for partition between water and dry alcohols N=l-10, and for partitions between water and wet alcohols, N=5-10, against the carbon number of the alcohol : ■ ^-coefficient, # 6-coefficient, ▲ v- coefficient. Closed symbols are for the dry alcohols, open symbols are for the wet alcohols.
There have been several extensive studies on the solvation properties of wet and dry octan-l-ol, although the results are not consistent. Cabani et al. found that there were differences up to 0.79 log units for partition into wet and dry octan-l-ol, but in the work of Dallas and Carr^^, the maximum difference was found to be only 0.13 log units. More recently, Kristi and Vesnave/^ have shown that the solubility ratio of a number of drug molecules between water and dry octan-l-ol and between octanol- saturated water and water-saturated octanol could be as large as 0.99 log units. However, the solubility ratio method will yield consistent results only if the same solid phase is in equilibrium with the various solvents under study. The conclusion seems to be that solvation in wet and dry octan-l-ol is generally quite similar, but that there may be specific instances where this is not so. Indeed, it might be expected that
90 solutes with very hydrophilic functional groups might complex with the water in water-saturated octan-l-ol and bring about a decrease in the standard Gibbs energy. From the work presented in this chapter, it is suggested that water and octan-l-ol have similar hydrogen-bond basicities, so it is unlikely that a solute with an acidic functional group will preferentially complex with the water in water-saturated octan- l-ol. But the hydrogen-bond acidity of water is very much greater than that of octan- l-ol (see the ^-coefficients in Table 4.07), so that functional groups that are strong hydrogen-bond bases might complex preferentially with water in the water-saturated octan-l-ol.
Kristialso found very different solubility ratios for drug molecules in the other dry and wet alkanols, heptan-l-ol and nonan-l-ol. Whether these ratios are due to complexation with the water in the water-saturated alkan-l-ols, or to solvate (or hydrate) formation, or to both, is not really known. If they are due to complexation, then this work of Kristi will illustrate again that there is always a possibility that certain compounds will be solvated differently in wet alkan-l-ols than in dry alkan-l- ols.
Although the two General Solvation Equations contain descriptors that refer to solute size (log and Vx), neither of the equations contains any descriptor that refers to the solute shape. The compounds that have been included in the correlations are molecules of quite different shape, for example spherical molecules such as helium and sulfur hexafluoride, long chain alkanes and alkan-l-ols, and a variety of other shaped compounds including diisopropyl ether, benzene, pyrene, diuron, and diphenylsulfone. Thus, for the simple transport processes considered, any effects due to solute shape appear to be relatively small. This has also been found for the solvation of conformational isomers by organic solventsand for the solvation of structural isomers by hexadecane and olive oil^^. In the latter work, the solvation of the three dimethoxybenzenes or the three dimethylbenzenes or the cis/trans isomers of 1,2-dichloroethene are essentially the same. Hence, it is reasonable to conclude that predictions of gas-solvent and water-solvent partitions through the General Solvation Equations should be valid for any compound that is within the descriptor space used to set up the correlations provided, possibly with the proviso that the compound is not of a very unusual shape.
91 4.3 References
1. Abraham, M.H.; Whiting, G.S.; Carr, P.W.; Ouyang, H. Hydrogen bonding. Part 45. The
solubility of gases and vapours in methanol at 298K; an LFER analysis. J.Chem. Soc., Perkin
Trans. 2,1998, 1385-1390. 2. Abraham, M.H.; Grellier, P.L.; Mcgill, R.A. Determination of olive oil-gas and hexadecane-gas partition coefficients, and calculation of the corresponding olive oil-water and hexadecane-water
partition coefficients. J. Chem. Soc., Perkin. Trans. 2, 1987, 797-803. 3. Abraham, M.H.; Le, J.; Acree, Jr, W.E.; Carr, P.W.; Dallas, A.J. The solubility of gases and
vapors in dry octan-l-ol at 298k. Submitted to Chemosphere. 4. Abraham, M.H.; Le, J.; Acree, Jr, W.E.; Carr, P.W. Solubility of gases and vapours in propan-l-ol at 298K. J. Phys. Org. Chem. 1999, 12, 675-680. 5. Abraham, M.H.; Whiting, G.S.; Shuely, W.J.; Doherty, R.M. The solubility of gases and vapours
in ethanol - The connection between gaseous solubility and water- solvent partition. Can. J.
Chem. 1998, 76,703-709. 6. Abraham, M.H.; Le, J.; Acree, Jr, W.E. The solvation properties of the aliphatic alcohols. Collect.Czech.Chem.Commun.1999, 64, 1748-1760. 7. Abraham, M.H.; Platts, J.A.; Hersey, A.; Leo, A.J.; Taft, R.W. Correlation and estimation of gas- chloroform and water-chloroform partition coefficients by a linear free energy relationship method. J. Pharm. Sci. 1999, 88 , 670-679. 8 . Abraham, M.H. Unpublished results. 9. Miller, K.J.; J.A.Savchik, J.A. New empirical method to calculate average molecular
polarisabilities. J.Am.Chem.Soc., 1979, 101,7206-7213. 10. Kamlet, M.J.; Abboud, J.-L.M.; Taft, R.W. An examination of linear solvation energy
relationships. Prog. Phys. Org. Chem. 1981, 13,485- 630. 11. Taft, R.W.; Abboud, J.-L.M.; Taft, Kamlet, M.J.; Abraham, M.H. Linear solvation energy
relations. Solution Chem. 1985, 153-186. 12. Abraham, M.H.; Chadha, H.S.; Whiting, G.S.; Mitchell, R.C. Hydrogen-bonding.32. An analysis
of water-octanol and water-alkane partitioning and the AlogP parameter of Seiler. J. Pharm.Sci. 1994, 83,1085-1100. 13. Abraham, M.H.; Chadha, H.S.; Dixon, J.P. Hydrogen bonding.39. The partition of solutes between
water and various alcohols. J.Phys.Org.chem. 1994,7,712-716. 14a. Berti, P.; Cabani, S.; Conti, G.; Mollica, V. Thermodynamic study of organic compounds in
octan-l-ol - Processes of transfer from gas and from dilute aqueous solution. J.Chem.Soc.,
Faraday Trans. 1. 1986, 82, 2547-2556. 14b. Berti, P.; Cabani, S.; Conti, G.; Mollica, V. Thermodynamic study of organic compounds in octan -1 -ol. 2.Enthalpy chenges in the transfer of primary, secondary, and tertiary amines from gas
and the dilute aqueous phase. Thermochim. Acta 1987, 122, 1-8. 14c. Bemazzani, L.; Cabani, S.; Conti, G.; Mollica, V. Thermodynamic study of the partitioning of
organic compounds between water and octan-l-ol. J.Chem. Soc.,Faraday Trans. 1995, 91-649.
92 15. Dallas, A.J.; Carr, P.W, A thermodynamic and solvatochromie investigation of the effect of water on the phase transfer properties of octan-l-ol. J.ChemSoc.,Perkin Trans. 2. 1992, 2155-2161. 16. Kristi, A.; Vesnaver, G. Thermodynamic investigation of the effect of octanol water mutual
miscibility on the partitioning and solubility of some guanine derivatives. J.Chem.Soc., Faraday
Trans. 1995,995-998. 17. Kristi, A. Thermodynamic investigation of the effect of the mutual miscibility of some higher
alkanols and water on the partitioning and solubility of some guanine derivatives. J.Chem.Soc.,
Faraday Trans. 1996, 1721-1724. 18. Abraham, M.H.; Xodo, L.E.; Cook, M.J.; Cruz, R. Solvent and gas phase effects on the equilibrium
between confurational isomers of some 4-r-butylcyclohexanes. J.Chem.Soc.,Perkin Trans.2. 1982, 1503-1509. 19. Abraham, M.H.; Grellier, P.L.; McGill, R.A. Determination of olive-oil-gas and hexadecane-gas partition coefficients, and calculation of the corresponding olive oil-water and hexadecane-water
partition coefficients. J.Chem.Soc.,Perkin Trans.2. 1987, 797-803. 20. Abraham, M.H.; Andonian-Haftvan, J.; Whiting, G.S.; Leo, A, Taft, R.W. Hydrogen bonding.34. The factors that influence the solubility of gases and vapors in water at 298K, and a method for its determination. J.ChenuSoc.,Perkin Trans.2. 1994, 1777- 1791. 21. Cabani, S.; Gianni, P.; Mollica, V.; Lepori, L. Group contributions to the thermodynamic properties of non-ionic organic solutes in dilute aqueous solution. J.Soln. Chem. 1981, 10, 563- 595. 22. Revised value 23. Abraham, M.H.; Grellier, P.L.; Nasehzadah, A.; Walker, R.A.C. Substitution at saturated carbon. 26. A complete analysis of solvent effects on initial states and transition states for the solvolysis of the tert-butyl halides in terms of G, H, and S using the unified method. J.Chem.Soc., Perkin. Trans.2. 1988, 1717-1724. 24. Benson, G.C., Int. DATA Ser., Sel. Data Mix.1985, 92-94. 25. Bo, S.; Battino, R.; Wilhelm, E. Solubility of gases in liquids. 19. Solubility of He, Ne, Ar, Kr, Xe, N2, 02 , CH4, CF4, and SF 6 in normal 1-alkanols at 98.15K. J.Chem. Eng. Data 1993, 38, 611- 616. 26. From the solubility data project. 27. Tokunaga, J. Solubilities of oxygen, nitrogen, and carbon-dioxide in aqueous alcohol solutions. J.Chem.Eng.Data 1975, 20, 41-46. 28. Mijyano, Y.; Hayduk, W. Solubility of butane in several polar and nonpolar solvents and in an
acetone butanol solvent solution J.Chem.Eng.Data, 1986, 31, 77- 80. 29. Dohnal, V.;Vrba, P. Limiting activity coefficients in the 1-alkanol plus n-alkane systems: Survey, critical evaluation and recommended values, interpretation in terms of association models. Fluid
Phase Eq. 1997, 133, 73-87. 30. Gmehling,!.; Menke, J.; Schiller, M. Activity Coefficients at Infinite Dilution; Chemical Data Series, Volume IX, Part 3, DECHEMA, Frankfurt/Main, Germany, 1994.
93 31. Vrbka, P.; Dohnal, V.; Trejo, L.M.; Costas, M. Molecular shape effects on limiting activity
coefficients: normal, branched and cyclic alkanes in 1-propanol or 2-propanol. Fluid Phase Eq. 1997, 137, 133-140. 32. Ashraf, S.M.; Ramakrishna, M.; Prasad, D.H.L. A simple and accurate method for isothermal
phase equilibrium studies. Phys.Chem.Liq.l991, 34, 89-102. 33. Heintz, H.; Dolch, E.; Lichtenthaler, R.N. New experimental VLE-data for alkanol alkane
mixtures and their description by an extended real association (ERAS) model. Fluid Phase Eq. 1986, 27,61-79. 34. Loeffler, G.E.; McKetta, Jr., J.J. Solubility of cyclopropane in some lower alcohols.
J.Chem.Eng.Data, 1966, 11, 431-433. 35. Letcher, T.M.; Mercer-Chalmers, J.; Prasad, A.K. Excess molar enthalpies of mixtures of a
cycloalkane and an alkanol. Thermochim.Acta 1991, 188,157-162. 36. Nagata, I.; Kazuma, K. Heats of mixing for ternary-system ethanol-1 -propanol-cyclohexane at
25°C. J.Chem.Eng.Data 1977, 22, 79-84. 37. Boyer, F.L.; Bircher, L.J. The solubility of nitrogen, argon, methane, ethylene and ethane in
normal primary alcohols. J.Phys.Chem. 1960, 64, 1330-1331.
38. Silva.; C.S.O, Fonseca, I.M.A.; Lobo, L.Q. Solubility of methyl fluoride in some alcohols. Fluid
Phase Eq. 1977, 135,137-144. 39. Matteoli, E.; Lepori, L. Excess gibbs free-energies of tetrachloromethane and an alcohol or a cyclic ether at 298.15K. J.ChenuThermodynam. 1986, 18, 1065-1076. 40. Gmehling, J.; Onken, U.; Arlt, W. Vapor-Liquid Equilibrium Data Collection; Chemical Data Series, Volume I, parts 2b-d, DECHEMA, Frankfurt/Main, Germany, 1978-1984. 41. Abraham, M.H.; Grellier, P.L. Substitution at saturated carbon. Part XX. The effect of 39 solvents on the Free Energy of EtgN, EtI, and the EtsN-Etl transition state. Comparison with solvent effects on the equilibria : EtgN 4- EtI Et4KT' and EtgN + EtI Et 4N^ + I . Chem.Soc., Perkin Trans.2. 1976, 1735-1741. 42. Dohnal, V.; Novotna, M. Infinite-dilution activity coefficients by comparative ebulliometry - the
mixtures of FREON-112 with oxygenated solvents and hydrocarbons. Fluid Phase Eq. 1985, 23, 303-313. 43. Salas, J.A.; Arancibia, E.L.; Katz, M. Excess molar volumes and isothermal vapor-liquid equilibria
in the tetrahydrofuran with propan-l-ol and propan-2-ol systems at 298.15K. Can.J.Chem. 1997, 75, 207-211. 44. Abraham, M.H.; Grellier, P.L. Substitution at saturated carbon. Part XIX. The effect of alcohols and
water on the Free Energy of solutes and on the Free Energy of transition states in Sn and Se reactions.
J.Chem.Soc., Perkin Trans.2. 1975,1856-1863. 45. Park, J.H.; Hussan, P.; Couasnon, P.; Fritz, D.J.; Carr, P.W. Experimental re-examination of selected partition coefficients from Rohrschneider data set. Anal. Chem. 1987,59,1970-1976. 46. Comelli, F.; Francesconi, R. Excess molar enthalpies and excess molar volumes of dimethyl
carbonate plus seven alkyl acetates at 298.15K. J.ChemEng.Data. 1997, 42,705- 709.
94 47. Fernandez, J.; Berro, C.; Andrade, M.I.P. Excess thermodynamic functions of 1-propanol and methyl propanoate and 1-propanol and methyl butanoate systems. Fluid Phase Eq. 1985, 20, 145- 153. 48. Abraham, M.H. Substitution at saturated carbon. 8 .Sol vent effects on free-energy of trimethylamine, nitrobenzyl chlorides, and trimethylamine-nitrobenzyl chloride transition-states.
J.Chem.Soc.B. 1971, 299-308. 49. Oracz, P. Int. DATA Ser., Sel.Data Mix.1986,75-77; 87-89; 96-98 50. Benson, G.C. Int. DATA Ser., Sel.Data Mix. 1973, 77; 85 51. Activity coefficient taken as unity. 52. Polak, J.; Murakami, S.; Benson, G.C.; Pflug, H.D. Excess enthalpy, volume and Gibbs free-
energy of n-propanol-isopropanol mixtures at 25°C. Can.J.Chem. 1970, 48, 3782- 3785. 53. Sun, H.H.; Christensen, J.J.; Izatt, R.M.; Hanks, R.W. Excess enthalpies of 6 normal-pentanol alcohol mixtures at 298.15K. J.ChenuThermodyn. 1980, 12,95- 99. 54. Benson, G.C. Int. DATA Ser., Sel.Data Mix. 1985, 92-94. 55. Abraham, M.H. Substitution at saturated carbon. 14. Solvent effects on free-energies of ions, ion-
pairs, non-electrolytes, and transition-states in some Sn and Sg reactions. J.Chem.Soc.,Perkin Trans.2. 1972, 1343-1357. 56. Endler, I.; Hradetzky, G.; Bittrich, H.-J. Limiting activity coefficients in binary mixtures benzene normal alcohol. J.Prakt.Chem. 1985, 327, 693-697. 57. Abraham, M.H.; Green, C.; Acree,Jr., W.E.; Hernandez, C.E.; Roy, L.E. Descriptors for solutes
from the solubility of solids: trans-stilbene as an example. J.Chem.Soc.,Perkin Trans.2. 1998, 2677-2681. 58. Warycha, S. Int. DATA Ser., Sel.Data Mix. 1994, 130-132. 59. Orszagh, A.; Kasprzycka-Guttman, T. Thermodynamic properties of some heterocyclic base aliphatic n-alcohol systems.3. Bull.Acad.Pol.ScL, Ser.Chim. 1974, 22, 625-629. 60. Triday, J O. Vapour liquid equilibria in binary-systems formed by thiophene and light alcohols.
J.Chem.Eng.Data 1983, 28, 307-310. 61. Pardo, J.; Lopez, M.C.; Santafe, J.; Royo, P.M.; Utieta, J.S. Solubility of gases in butanols. 1. Solubilities of nonpolar gases in 1-butanol from 263.15 to 303.15K at 101.33Kpa partial pressure of gas. Fluid Phase Eq. 1995, 109, 29-37. 62. Bo, S.; Battino, R.; Wilhelm, E. Solubility of gases in liquids. 19.Solubility of He, Ne, Ar, Kr, Xe, N2, O2, CH4, and SFg in normal 1-alkanols n-(CH 2+ 10H) J.Chem.Eng.Data 1993, 38, 611-616. 63. Sahgal, A.; Hayduk, W. Ethylene solubility and diffusivity in hexane-dodecane and ethylene
glycol butanol solutions. J.Chem.Eng.Data 1979,24,222-227. 64. Vernier, P.; Raimbaul, C.; Renon, H. Thermodynamic properties of infinitely diluted hydrocarbon
solutions in mixtures of polar solvents. J.Chim.Phys. 1969,66,960-969. 65. Thomas, E.R.; Newman, B.A.; Long, T.C.; Wood, D.A.; Eckert, C.A. Limiting activity coefficients of non-polar and polar solutes in both volatile and non-volatile solvents by gas- chromatography. J.ChemEng.Data 1982,27,399-405.
95 66. Garriga, R.; Sanchez, F.; Perez, P.; Valero, J.; Gracia, M. Vapour pressures at several temperatures and excess functions at 298.15K of n-butanol plus 2- methylpentane and plus 3-
methylpentane. Fluid Phase Eq. 1994, 101,227-236. 67. Garriga, R.; Sanchez, P.; Perez, P.; Valero, J.; Gracia, M. Vapour pressures at several temperatures and excess functions of n-butanol with 2,2-dimethylbutane and with 2,3-Dimethylbutane at 298.15K.
Fluid Phase Eq. 1994,102, 85-95. 68 . Hayduk, W.; Asatani, H.; Miyano, Y. Solubilities of propene, butane, isobutane and isobutene
gases in normal-octane, chlorobenzene and normal-butanol solvents. Can.J.Chem.Eng. 1988,66,466- 473. 69. Miyano, Y.; Hayduk, W. Solubility of acetylene in several polar and nonpolar solvents and
solvent mixtures. Can.J.Chem.Eng. 1981, 59, 746-751. 70. Christensen, J.J.; Rowley, R.L.; Izatt, R.M. Handbook of Heats of Mixing: Supplementary Volume, Wiley-Interscience, New York, 1988. 71. Landau, I.; Belfer, A.J.; Locke, D C. Measurement of limiting activity-coefficients using non steady state gas chromatography. Ind.Eng.Chem.Res.1991, 30, 1900-1906. 72. Garriga, R.; Perez, P.; Gracia, M. Vapour pressures at eight temperatures of mixtures of Di-n- propylether plus ethanol, or plus 1-butanol. Thermodynamic description of mixtures of Di-n- propylether plus alkanol according to the ERAS model. Ber.Bunsenges. Phys.Chem.1997, 101, 1466-1473. 73. Park, J.H.; Hussam, A.; Couasnon, P.; Fritz, D.; Carr, P.W. Experimental re-examination of selected partition coefficients from Rohrschneider data set. Anal.Chem. 1987, 59, 1970-1976. 74. Garriga, R.; Putze, I.; Perez, P.; Gracia, M. Vapour pressures at several temperatures-T and excess functions at T=298.15K of (butanenitrile + ethanol or butan-l-ol). J.Chem.Thermodyn. 1995, 27, 481-491. 75. Balyatinskaya, L.N.; Filimonov, V.N.; Garkavaya, N.N. Headspace gas chromatographic
separation of toxic organic impurities from air. Chromatographia, 1997, 45, 343-346 Suppl. S. 76. Acree, Jr., W.E., unpublished results 77. Geiseler, G.; Suhnel, K.; Quitzch, K. Excess properties of binary mixed systems of butanol
isomers. Z.Phys.Chem.(Leipzig) 1973, 254,261-270. 78. Murakami, S.; Benson, G.C. Thermodynamic properties of some isomeric butyl alcohol mixtures. Bull.Chem.Soc.Japan 1973, 46, 74-79. 79. Oracz, P. Int.DATA Ser., Sel.Data Mix. 1986, pp.99-101. 80. Monfort, J.-P.; Vidal, J.; Renon, H. Activity-coefficient of an infinite dilution of hydrocarbons in
two families of solvents - effects of solvent structure. J.Chim.Phys. 1970, 67, 748. 81. Fletcher, K.A.; McHale, M.E.R.; Coym, K.S.; Acree, Jr., W.E. Solubility of trans-stilbene in organic non-electrolyte solvents. Comparison of observed versus predicted values based upon
mobile order theory. Can.J.Chem. 1997,75,258-261; from the ratio of solubilities. 82. Dickhut, R.M.; Andren, A.W.; Armstrong, D.E. Naphthalene solubility in selected organic
solvent-water mixtures. J.Chem.Eng.Data 1989, 34,438-443; from the ratio of solubilities. 83. From ratio of solubilities, this work.
96 84. Martin, A.; Wu, P.L.; Beerbower, A. Expanded solubility parameter approach.2. Para- hydro xybenzoic acid and methyl para-hydroxybenzoate in individual solvents. J. Pharm. Sci. 1984, 73, 188-194. 85. Ruelle, P.; Kesselring, U.W. The hydrophobic effect.2. Relative importance of the hydrophobic
effect on the solubility of hydrophobes and pharmaceuticals in H- bonded solvents. J.Pharm.Sci. 1998,998-1014. 86 . Abraham, M.H. Thermodynamics of solution of homologous series of solutes in water. J.Chem.Soc.Faraday Trans.1 1984, 80, 153-184 Part 1. 87. Knoop, C.; Tiegs, D.; Gmehling, J. Measurement of y-infmity using gas-liquid chromatography. 3. Results for the stationary phases 10-nonadecanone, n-formylmorpholine, 1-pentanol, m-xylene and toluene. J.Chem.Eng.Data. 1989, 34,240-247. 88 . Sayegh, S.G.; Radcliff, G.A. Excess Gibbs energies of binary systems of isopentanol and normal pentanol with hexane isomers at 25 °C - measurement and prediction by analytical group solution
model. J.Chem.Eng.Data. 1976, 21, 71-74. 89. Garriga, R.; Ilarraza, J.; Perez, P.; M.Gracia, M. Excess functions at T=298.15K of (butanenitrile + methanol, + pentan-l-ol, + heptan-l-ol, + nonan-l-ol, or + decan-l-ol) J.Chem.Thermodyn. 1996,28,233-234. 90. Abraham, M.H.; Grellier, P.L.; Mana, J. Limiting activity coefficients of triethylamine in 30 solvents by a simple gas-liquid chromatographic method. J.Chem.Thermodyn. 1974, 6, 1175- 1179. 91. Mara, K.; Bhethanabotla, V.R.; Campbell, S.W. Total pressure measurements for 1-propanol + 1- pentanol, 1-propanol + 2-pentanol, 2-propanol + 1-pentanol, and 2-propanol + 2-pentanol at 313.15K. Fluid Phase Eqilib. 1997, 127, 147-153. 92. McKelgue, K.; Gularl, E. Vapour-liquid equilibria for binary mixtures of 1-alcohols in carbon
disulfide. J. Chem. Eng. Data 1986, 31,278-283. 93. Ashraf, S.M.; Prasad, D.H.L. Isothermal phase equilibrium studies on the binary mixtures of n- butanol and n-pentanol with alkylbenzenes. Phys.Chem.Liq. 1998, 36, 149-158. 94. Prasad, A.R.; Reddy, B.S.; Rao, K.V. Vapour liquid equilibria of the 1,1,1 -trichloroethane-1 -
propanol system at 101.3-Kpa pressure. J.Chem.Eng.data 1991, 36, 311-313. 95. Brunner, E. Solubility of hydrogen in alcohols. Ber.Bunsenges Phys.Chem. 1979, 83,715-721. 96. Yaacobi, M.; Ben-Naim, A. Solvophobic interaction. J.Phys.Chem. 1974, 78, 175-179. 97. Asprion, N.; Hasse, H.; Maurer, G. Limiting activity coefficients in alcohol-containing organic
solutions from headspace chromatography. J.Chem.Eng.Data 1998,43,74-80. 98. Garriga, R.; Sanchez, F.; Perez, P.; Gracia, M. Excess Gibbs free energies at several temperatures of butanone with 1-hexanol or 1-octanol. J.Chem.Eng.Data 1996,41, 1091-1096. 99. Garriga, R.; Perez, P.; Gracia, M. Vapour pressures at several temperatures, T and excess
functions at T=298.15K of (butanenitrile + hexan-l-ol or octan-l-ol) J.Chem.Thermodynam. 1995,27,1057-1066. 100. Acree,Jr., W.E. Polycyclic Aromatic Hydrocarbons; Binary Nonaqueous systems. Part 2 (Solutes F-Z), Volume 59 in lUPAC Solubility Data Series, Oxford University Press, Oxford, UK, 1995
97 101. De Fina, K.M.; Sharp, T.L.; Acree,Jr., W.E. CanJ.Chem. submitted. 102. Hernandez, C.E.; De Fina, K.M.; Roy, L.E.; Sharp, T.L.; Acree,Jr., W.E. CanJ.Chem. in the press. 103. Hernandez, C.E.; Acree,Jr., W.E. Solubility of fluoranthene in organic nonelectrolyte solvents.
Comparison of observed versus predicted values based upon Mobile Order Theory. Can.J.Chem. 1998,76,1312-1316. 104. Beerbower, A.; Wu, P.L.; Martin, A. Expanded solubility parameter approach. 1 .Naphthalene and
benzoic acid in individual solvents. J.Pharm.Sci. 1985, 73, 179. 105. Tse, G.; Sandler, S.I. Determination of infinite dilution activity coefficients and 1-octanol water partition coefficients of volatile organic pollutants. J.Chem. Eng.Data 1994, 39, 354-357. 106. Hamer, T.; Bidleman, T.F. Measurement of octanol-air partition coefficients for polycyclic
aromatic hydrocarbons and polychlorinated naphthalenes. J.Chem.Eng.Data 1998,43,40-46. 107. Mackay, D.; Shiu, W.Y. A critical review of Henrys Law constants for chemicals of environmental interest. J.Phys. Chem.Ref.Data 1981, 10, 1175-1199 108. Stephenson, S.M.; Malanowsky, S. Handbook of the thermodynamics of organic compounds; Elsevier, New York, 1987. 109. Rordorf, B E. Thermodynamic properties of polychlorinated compounds - The vapour pressures and enthalpies of sublimation of 10 dibenzo-p-dioxines. Thermochim. Acta 1985, 85,435-438. 110. Harner, T.; Mackay, D. Measurement of octanol-air partition coefficients for chlorobenzenes, PCBs, and DDT. Environ. Sci.Technol. 1995, 29, 1599-1606 111. Yalkowsky, S.H.; Myrdal, P.; Dannenfelser,R.-M.; Simamora,P. Upper.2. Calculation of physical properties of the chlorobenzenes. Chemosphere 1994, 28, 1675-1688. 112. Alexander,K.S.; Mauger,J.W.; Petersen,Jr.H.; Paruta, A.N. Solubility profiles and thermodynamics of parabens in aliphatic alcohols. J.Pharm.Sci. 1977, 66, 42-48. 113. Domanska,U.; Gonzalez,J.A. Solid-liquid equilibria for systems containing long chain 1- alkanols.3.Experimental data for 1-tetradecanol, 1-hexadecanol, 1- octadecanol or eicosanol + 1- butan-l-ol, 1-hexanol, 1-octanol or 1-decanol mixtures, characterization in terms of DISQUAC. Fluid Phase Eq. 1997, 129, 139- 163. 114. Wilcock,R.L.; Battino,R.; Danforth,W.F.; Wilhelm, E. Solubilities of gases in liquids.2.Solubilities of He, Ne, Ar, Kr, O 2, N%, CO, CO2, CH4, CF4 and SFg in normal octane, 1- octanol, normal decane and 1-decanol. J.Chem.Thermodyn. 1978, 10, 817-822, 115. Bhatia, S.R.; Sandler, S.L. Temperature dependence of infinite dilution activity coefficients in octanol and octanol water partition coefficients of some volatile halogenated organic compounds.
J.Chem.Eng.Data 1995, 40, 1196-1198. 116. Gruber, D.; Langenheim, D.; Gmehling, J. Measurement of activity coefficients at infinite dilution using gas-liquid chromatography. 6.Results for systems exhibiting gas-liquid interface adsorption with 1-octanol. J.Chem.Eng.Data 1997, 42, 882-885 117. Cabani, S.; Conti, G.; Mollica, V.; Bemazzani. Free-energy and enthalpy changes for the process of transfer from gas and from dilute aqueous solutions of some alkanes and monofunctional saturated organic compounds. J. Faraday Trans. 1991, 87, 2433-2442.
98 118. Tewari, Y.B.; Miller, M.M.; Wasik, J.P. Calculation of aqueous solubility of organic compounds. J.Res.Nat. Bur. Stand. 1982, 87,155-158. 119. Gerrard, W. Solubility of hydrogen sulfide, dimethyl ether, methyl chloride and sulfur-dioxide in
liquids - prediction of solubility of all gases. J.Appl.Chem. Biotech. 1972, 22, 62. 120. Pollack, G.L.; Himm, j.F.; Enyeart, J.J. Solubility of xenon in liquid normal alkanols -
thermodynamic functions in simple polar liquids. J.Chem.Phys. 1984, 81, 3239-3246. 121. Pettersson, A.; Saris, P.; Rosenholm, J.B. Thermodynamic properties of binary alcohol
hydrocarbon systems. J.Chem.Soc., Faraday Trans 1986,I. 82,2435-2457. 122. Fukuchi, K.; Miyoshi, K.; Watanabe, T.; Yonezawa, S.; Arai, Y. Measurement and correlation of
infinite dilution activity coefficients of ethers in alkanols. Fluid Phase Eq. 1999, 156, 197-206. 123. Benson, G.C. Int.DATA Ser., Sel.Data Mix. 1985, 86 -88 , 89-91, 92-94,233-234. 124. Ignat, A.; Molder, L.; Eesti, N.S.V. Tead.Akad.Toim.Keem. 1989, 38, 11-16. 125. G.C.Benson, Int.DATA Ser., Sel.Data Mix. 1974, 101,102.
99 4.4 Data Tables
Table 4.08 Solute Abraham descriptors and logL^ values
Solute R2 Ttz" Zotz" Zpz" Vx logL'^ Helium 0 0 0 0 0.0680 -1.741 - 2.02 Neon 0 0 0 0 0.0850 -1.575 -1.96 Argon 0 0 0 0 0.1900 -0.688 -1.47 Krypton 0 0 0 0 0.2460 -0.211 - 1.21 Xenon 0 0 0 0 0.3290 0.378 -0.97 Hydrogen 0 0 0 0 0.1086 -1.200 -1.72 Deuterium 0 0 0 0 0.1100 - 1.200 -1.73 Oxygen 0 0 0 0 0.1830 -0.723 -1.51 Nitrogen 0 0 0 0 0.2222 -0.978 -1.80 Nitrous oxide 0.068 0.35 0 0.10 0.2810 0.164 -0.23 Carbon monoxide 0 0 0 0.04 0.2220 -0.836 -1.62 Carbon dioxide 0 0.28 0.05 0.10 0.2809 0.058 -0.08 Methane 0 0 0 0 0.2495 -0.323 -1.46 Ethane 0 0 0 0 0.3904 0.492 -1.34 Propane 0 0 0 0 0.5313 1.050 -1.44 Butane 0 0 0 0 0.6722 1.615 -1.52 Isobutane 0 0 0 0 0.6722 1.409 -1.70 Pentane 0 0 0 0 0.8131 2.162 -1.70 Hexane 0 0 0 0 0.9540 2.668 -1.82 2-MethyIpentane 0 0 0 0 0.9540 2.503 -1.84 3-Methylpentane 0 0 0 0 0.9540 2.581 -1.84 2,2-Dimethylbutane 0 0 0 0 0.9540 2.352 -1.84 2,3-Dimethylbutane 0 0 0 0 0.9540 2.495 -1.72 Heptane 0 0 0 0 1.0949 3.173 -1.96 2,4-Dimethylpentane 0 0 0 0 1.0949 2.809 -2.08 Octane 0 0 0 0 1.2358 3.677 - 2.11 2,2,4-Trimethylpentane 0 0 0 0 1.2358 3.106 -2.12 2,3,4-T rimethylpentane 0 0 0 0 1.2358 3.481 - 1.88 Nonane 0 0 0 0 1.3767 4.182 -2.30 2,5-Dimethylheptane 0 0 0 0 1.3767 3.822 -2.19 Cyclopropane 0.180 0.15 0 0 0.4227 1.314 -0.55 Cyclopentane 0.263 0.10 0 0 0.7045 2.477 - 0.88 Methylcyclopentane 0.225 0.10 0 0 0.8454 2.907 -1.17 Cyclohexane 0.305 0.10 0 0 0.8454 2.964 -0.90 Methylcyclohexane 0.244 0.06 0 0 0.9863 3.319 -1.21 Ethylcyclohexane 0.263 0.10 0 0 1.1272 3.877 -1.59 Cycloheptane 0.350 0.10 0 0 0.9863 3.704 -1.39 Ethene 0.107 0.10 0 0.07 0.3474 0.289 -0.94 Propene 0.103 0.08 0 0.07 0.4883 0.946 -0.97 Isobutene 0.100 0.08 0 0.07 0.6292 1.560 -0.85 Pent-l-ene 0.093 0.08 0 0.07 0.7701 2.047 -1.23 Isopentene 0.063 0.08 0 0.07 0.7701 1.910 -1.34 Hex-l-ene 0.078 0.08 0 0.07 0.9110 2.572 -1.16 Hept-l-ene 0.092 0.08 0 0.07 1.0519 3.063 - 1.22 Oct-l-ene 0.094 0.08 0 0.07 1.1928 3.568 -1.41
100 Solute R2 Ttl” ZPz" Vx logL'^ Non-l-ene 0.090 0.08 0 0.07 1.3337 4.073 -1.51 2-Methylbuta-l,3-diene 0.313 0.23 0 0.10 0.7271 2.101 -0.50 2-Methylbuta-l,3-diene 0.313 0.23 0 0.10 0.7271 2.101 -0.50 Cyclohexene 0.395 0.20 0 0.10 0.8024 3.021 -0.27 Ethyne 0.190 0.60 0.06 0.04 0.3044 0.140 0.00 Fluoromethane 0.066 0.35 0 0.10 0.2672 0.056 0.16 T etrafluomethane 0.570 -0.26 0 0 0.3203 -0.826 -2.29 Chloromethane 0.249 0.43 0 0.08 0.3719 1.163 0.40 Dichloromethane 0.387 0.57 0.10 0.05 0.4943 2.019 0.96 T richloromethane 0.425 0.49 0.15 0.02 0.6167 2.480 0.79 T etrachloromethane 0.458 0.38 0 0 0.7391 2.823 -0.06 1,1 -Dichloroethane 0.322 0.49 0.10 0.10 0.6352 2.316 0.62 1,2-Dichloroethane 0.416 0.64 0.10 0.11 0.6352 2.573 1.31 1,1,1-T richloroethane 0.369 0.41 0 0.09 0.7576 2.733 0.14 1,1,2-Trichloroethane 0.499 0.68 0.13 0.13 0.7576 3.290 1.46 Hexachloroethane 0.680 0.68 0 0 1.1248 4.718 0.15 1-Chloropropane 0.216 0.40 0 0.10 0.6537 2.202 0.24 1,2-DichIoropropane 0.371 0.63 0 0.17 0.7761 2.890 0.93 1-Chlorobutane 0.210 0.40 0 0.10 0.7946 2.722 0.12 2-chloro- 2-methylpropane 0.142 0.25 0 0.12 0.7946 2.217 -0.80 cis-1,2-Dichloroethene 0.436 0.61 0.11 0.05 0.5922 2.439 0.86 Trichloroethene 0.524 0.37 0.08 0.03 0.7146 2.997 0.32 T etrachloroethene 0.639 0.44 0 0 0.8370 3.584 -0.07 Dibromomethane 0.714 0.69 0.11 0.07 0.5995 2.886 1.44 Bromoethane 0.366 0.40 0 0.12 0.5654 2.120 0.54 2-B romo-2-methy Ipropane 0.305 0.25 0 0.12 0.8472 2.616 -0.59 lodomethane 0.676 0.43 0 0.13 0.5077 2.106 0.65 lodoethane 0.640 0.40 0 0.15 0.6486 2.573 0.54 Dibromochloromethane 0.775 0.71 0.07 0.08 0.7219 3.304 1.28 CFC12.CFC12 0.227 0.33 0 0.02 0.9154 3.034 -0.64 Dimethylether 0 0.27 0 0.41 0.4491 1.285 1.40 Diethylether 0.041 0.25 0 0.45 0.7309 2.015 1.17 Dipropylether 0.008 0.25 0 0.45 1.0127 2.954 0.85 Diisopropylether -0.060 0.16 0 0.58 1.0127 2.482 1.17 Dibutylether 0 0.25 0 0.45 1.2945 3.924 0.61 Diisobutylether 0 0.19 0 0.45 1.2945 3.485 0.54 Dipentylether 0 0.25 0 0.45 1.5763 4.875 0.50 t-Butyl methyl ether 0.024 0.11 0 0.63 0.8720 2.372 1.62 Tetrahydrofuran 0.289 0.52 0 0.48 0.6223 2.636 2.55 Tetrahydropyran 0.275 0.47 0 0.55 0.8288 3.057 2.29 Dioxane 0.329 0.75 0 0.64 0.6810 2.892 3.71 Propanal 0.196 0.65 0 0.45 0.5470 1.815 -- Hexanal 0.187 0.65 0 0.45 0.6879 2.270 -- Octanal 0.072 0.60 0 0.51 1.1106 3.403 -- Butanal 0.187 0.65 0 0.45 0.6879 2.270 2.33 Acrolein 0.324 0.61 0 0.46 0.5040 1.656 -- Propanone 0.179 0.70 0.04 0.49 0.5470 1.696 2.79 Butanone 0.166 0.70 0 0.51 0.6879 2.287 2.72 Pentan-2-one 0.143 0.68 0 0.51 0.8288 2.755 2.58 Pentan-3-one 0.154 0.66 0 0.51 ■ 0.8288 2.811 2.50
101 Solute Rz Ttz” Sttz” Zgi" Vx logL'^ C 3 -Methy lbutan-2-one 0.134 0.65 0 0.51 0.8290 2.692 2.38 Hexan-2-one 0.136 0.68 0 0.51 0.9697 3.262 2.41 Heptan-2-one 0.123 0.68 0 0.51 1.1106 3.760 2.23 Cyclopentanone 0.373 0.86 0 0.52 0.7202 3.221 3.45 2,4-Pentanedione 0.412 0.56 0 0.79 0.8445 2.918 -- Dimethyl carbonate 0.142 0.61 0 0.55 0.6644 2.447 2.73 Methyl formate 0.192 0.68 0 0.38 0.4648 1.285 2.04 Ethyl formate 0.146 0.66 0 0.38 0.6057 1.845 1.95 Propyl formate 0.132 0.63 0 0.38 0.7466 2.433 1.82 Methyl acetate 0.142 0.64 0 0.45 0.6057 1.911 2.30 Ethyl acetate 0.106 0.62 0 0.45 0.7466 2.314 2.16 Propyl acetate 0.092 0.60 0 0.45 0.8875 2.819 2.05 Isopropyl acetate 0.055 0.57 0 0.47 0.8875 2.546 1.94 Butyl acetate 0.071 0.60 0 0.45 1.0284 3.353 1.94 Isobutyl acetate 0.052 0.57 0 0.47 1.0284 3.161 1.73 Pentyl acetate 0.067 0.60 0 0.45 1.1693 3.844 1.84 Isopentyl acetate 0.051 0.57 0 0.47 1.1693 3.740 1.62 Hexyl acetate 0.056 0.60 0 0.45 1.3102 4.351 1.66 Methyl propanoate 0.128 0.60 0 0.45 0.7466 2.431 2.15 Ethyl propanoate 0.087 0.58 0 0.45 0.8875 2.807 1.97 Methyl butanoate 0.106 0.60 0 0.45 0.8875 2.893 2.08 Ethyl butanoate 0.068 0.58 0 0.45 1.0284 3.271 1.83 Acetonitrile 0.237 0.90 0.07 0.32 0.4042 1.739 2.85 Proprionitrile 0.162 0.90 0.02 0.36 0.5450 2.082 2.82 Butanonitrile 0.188 0.90 0 0.36 0.6860 2.548 2.67 2-Cyanopropane 0.142 0.87 0 0.36 0.6860 2.452 2.50 Pentanonitrile 0.177 0.90 0 0.36 0.8269 3.108 2.58 Hexanonitrile 0.166 0.90 0 0.36 0.9680 3.608 2.45 Ammonia 0.139 0.39 0.16 0.56 0.2084 0.319 3.15 Methylamine 0.250 0.35 0.16 0.58 0.3493 1.300 3.34 Butylamine 0.224 0.35 0.16 0.61 0.7720 2.618 3.11 Dimethylamine 0.189 0.30 0.08 0.66 0.4902 1.600 3.15 Dipropylamine 0.124 0.30 0.08 0.69 1.0538 3.351 2.68 Trimethylamine 0.140 0.20 0 0.67 0.6311 1.620 2.35 Triethylamine 0.101 0.15 0 0.79 1.0538 3.040 2.36 Nitromethane 0.313 0.95 0.06 0.31 0.4237 1.892 2.95 Nitroethane 0.270 0.95 0.02 0.33 0.5646 2.414 2.72 Nitropropane 0.242 0.95 0 0.31 0.7055 2.894 2.45 Dimethylformamide 0.367 1.31 0 0.74 0.6468 3.173 5.73 Dimethylacetamide 0.363 1.33 0 0.78 0.7877 3.717 -- Acetic acid 0.265 0.65 0.61 0.44 0.4648 1.750 4.91 Methanol 0.278 0.44 0.43 0.47 0.3082 0.970 3.74 Ethanol 0.246 0.42 0.37 0.48 0.4491 1.485 3.67 Propan-l-ol 0.236 0.42 0.37 0.48 0.5900 2.031 3.56 Propan-2-ol 0.212 0.36 0.33 0.56 0.5900 1.764 3.48 Butan-l-ol 0.224 0.42 0.37 0.48 0.7309 2.601 3.46 2-Methy Ipropan-1 -ol 0.217 0.39 0.37 0.48 0.7309 2.413 3.30 Butan-2-ol 0.217 0.36 0.33 0.56 0.7309 2.338 3.39 2-Methylpropan-2-ol 0.180 0.30 0.31 0.60 0.7309 1.963 3.28 Pentan-l-ol 0.219 0.42 0.37 0.48 0.8718 3.106 3.35
102 Solute R2 Ea," Vx logL'^ r 3-Methylbutan-l-oI 0.192 0.39 0.37 0.48 0.8718 3.011 3.24 Hexan-l-ol 0.210 0.42 0.37 0.48 1.0127 3.610 3.23 Heptane] 0.211 0.42 0.37 0.48 1.1536 4.115 3.09 Octan-l-ol 0.199 0.42 0.37 0.48 1.2950 4.619 3.00 Nonan-l-ol 0.193 0.42 0.37 0.48 1.4354 5.124 2.85 Decan-l-ol 0.191 0.42 0,37 0.48 1.5763 5.628 2.67 Undecan-l-ol 0.181 0.42 0.37 0.48 1.7172 6.139 2.50 Tetradecan-l-ol 0.163 0.42 0.37 0.48 2.1400 7.656 2.26 Hexadecan-l-ol 0.151 0.42 0.37 0.48 2.4220 8.671 2.02 Octadecan-l-ol 0.145 0.42 0.37 0.48 2.7035 9.680 1.78 Eicosan-l-ol 0.140 0.42 0.37 0.48 2.9853 10.690 1.54 Cyclohexanol 0.460 0.54 0.32 0.57 0.9040 3.758 4.01 2-Chloroethanol 0.419 0.59 0.47 0.57 0.5715 2.497 4.78 Dimethylsulfoxide 0.522 1.72 0 0.97 0.6126 3.459 7.85 Sulfur hexafluoride 0.600 -0.20 0 0 0.4643 -0.120 -2.23 Carbon disulfide 0.876 0.26 0 0.03 0.4905 2.370 -0.15 Tetramethyltin 0.100 0.08 0 0.05 1.0431 2.745 -1.53 Tetraethyltin -0.010 0.12 0 0.10 1.6067 4.943 -1.62 Tetraethyllead 0.080 0.10 0 0.12 1.6476 5.180 -1.49 Benzene 0.610 0.52 0 0.14 0.7164 2.786 0.63 Toluene 0.601 0.52 0 0.14 0.8573 3.325 0.65 Ethylbenzene 0.613 0.51 0 0.15 0.9982 3.778 0.58 o-Xylene 0.663 0.56 0 0.16 0.9982 3.939 0.66 m-Xylene 0.623 0.52 0 0.16 0.9982 3.839 0.61 p-Xylene 0.613 0.52 0 0.16 0.9982 3.839 0.59 Propylbenzene 0.604 0.50 0 0.15 1.1391 4.230 0.39 Isopropylbenzene 0.602 0.49 0 0.16 1.1391 4.084 0.22 Hexamethylbenzene 0.950 0.72 0 0.28 1.5618 6.557 1.39 trans-Stilbene 1.450 1.05 0 0.34 1.5630 7.520 2.78 Biphenyl 1.360 0.99 0 0.26 1.3242 6.014 1.95 Naphthalene 1.340 0.92 0 0.20 1.0854 5.161 1.73 Acenaphthene 1.604 1.05 0 0.22 1.2586 6.469 2.28 Fluorene 1.588 1.06 0 0.25 1.3565 6.922 2.56 Anthracene 2.290 1.34 0 0.28 1.4544 7.568 3.03 Phenanthrene 2.055 1.29 0 0.29 1.4544 7.632 2.80 Fluoranthene 2.377 1.55 0 0.24 1.5846 8.827 3.45 Pyrene 2.808 1.71 0 0.28 1.5846 8.833 3.50 Hexafluorobenzene 0.088 0.66 0 0 0.8226 2.515 -0.07 Chlorobenzene 0.718 0.65 0 0.07 0.8388 3.657 0.82 1,2-Dichlorobenzene 0.872 0.78 0 0.04 0.9612 4.518 1.00 1,4-Dichlorobenzene 0.825 0.75 0 0.02 0.9612 4.435 0.93 1,2,3-Trichlorobenzene 1.030 0.86 0 0 1.0836 5.419 0.91 1,3,5-T richlorobenzene 0.980 0.73 0 0 1.0836 5.045 0.57 1,2,3,4-T etrachlorobenzene 1.180 0.92 0 0 1.2060 6.171 0.98 1,2,3,5 -T etrachlorobenzene 1.160 0.85 0 0 1.2060 5.922 0.67 1,2,4,5-Tetrachlorobenzene 1.160 0.86 0 0 1.2060 5.926 0.98 Pentachlorobenzene 1.330 0.96 0 0 1.3284 6.716 1.56 Hexachlorobenzene 1.490 0.99 0 0 1.4508 7.624 1.34 Benzyl chloride 0.821 0.86 0 0.14 0.9797 4.384 1.38
103 l W« Solute R2 Z k" Vx logL'" 1,4-Dibromobenzene 1.150 0.86 0 0.04 1.0664 5.324 1.44 Benzyl bromide 1.014 0.94 0 0.20 1.0323 4.672 1.90 Anisole 0.708 0.75 0 0.29 0.9160 3.890 1.80 Benzonitrile 0.742 1.11 0 0.33 0.8711 4.039 3.09 Methyl 4-aminobenzoate 1.028 1.52 0.32 0.59 1.1724 6.085 6.56^ Diphenylamine 1.585 0.88 0.10 0.57 1.4240 7.086 4.10 4-Nitrobenzyl chloride 1.270 1.42 0 0.36 1.2065 6.290 3.77* 4-Ethoxyacetanilide 0.940 1.51 0.45 0.86 1.4542 6.917 8.04 Benzoic acid 0.730 0.90 0.59 0.40 0.9317 4.395 5.10 2-Hydroxybenzoic acid 0.890 0.84 0.71 0.38 0.9904 4.721 5.35/ 4-Hydroxybenzoic acid 0.930 0.90 0.81 0.56 0.9904 4.867 6.78 / Methyl 4-hydroxybenzoate 0.900 1.37 0.69 0.45 1.1313 5.665 6.84^ Diphenylsulfone 1.570 2.15 0 0.70 1.6051 8.902 7.39 Pyridine 0.631 0.84 0 0.52 0.6753 3.022 3.44 3-Methylpyridine 0.631 0.81 0 0.54 0.8162 3.631 3.50 Piperidine 0.422 0.46 0.13 0.68 0.8043 3.304 3.75 Pyrrolidine 0.406 0.96 0.23 0.42 0.6634 2.893 4.02 N-Methylpyrrolidine 0.303 0.98 0 0.40 0.8043 2.808 2.92 Thiophene 0.687 0.57 0 0.15 0.6411 2.819 1.04 Diuron 1.280 1.60 0.57 0.70 1.5992 8.060 7.97 “ log values taken from ref 20 and 21 unless shown otherwise ^Ref 20 calculated value using eqn 1 "Ref 22 Ref 23 "Ref 24, ^ Ref 4 calculated using values in Table 3
Table 3.09 Values of log log L^^^calc, log log and log pPrOH/''^calc for solutes at 298 K
Solute logL ref logL lo g P “ logP calc calc Helium -1.56 15 -1.54 0.46 0.42 Neon -1.41 25 -1.40 0.55 0.49 Argon -0.61 25 -0.63 0.86 0.92 Krypton -0.16 25 -0.21 1.05 1.14 Xenon 0.38 25 0.30 1.35 1.47 Hydrogen -1.12 25 -1.07 0.60 0.59 Oxygen -0.66 25 -0.66 0.85 0.89 Nitrogen -0.88 25 -0.88 0.92 1.04 Nitrous oxide 0.44 26 0.43 0.67 0.54 Carbon monoxide -0.75 26 -0.71 0.87 0.89 Carbon dioxide 0.35 27 0.51 0.43 0.61 Methane -0.29 25 -0.31 1.17 1.15 Ethane 0.45 26 0.40 1.79 1.72 Butane 1.45 28 1.38 2.97 2.85 Pentane 1.81 29 1.85 3.51 3.42 Hexane 2.28 29 2.29 4.10 3.99 2-Methylpentane 2.15 30 2.15 3.99 3.99 Heptane 2.74 32 2.73 4.70 4.55 2,4-Dimethylpentane 2.43 30 2.41 4.51 4.55
104 Solute logL ref logL logP" logP calc calc Octane 3.17 29 3.17 5.28 5.12 2,2,4-Trimethylpentane 2.69 31,32 2.67 4.81 5.12 2,3,4-T rimethylpentane 2.95 30 3.00 4.83 5.12 Nonane 3.55 33 3.61 5.85 5.68 2,5-Dimethylheptane 3.40 30 3.29 5.59 5.68 Cyclopropane 1.15 34 1.18 1.70 1.77 Cyclopentane 2.12 31,35 2.14 3.00 2.99 Cyclohexane 2.55 31,36 2.56 3.45 3.57 Ethylcyclohexane 3.28 30 3.36 4.87 4.69 Cycloheptane 3.13 31,35 3.19 4.52 4.16 Ethene 0.38 37 0.34 1.32 1.47 T etrafluoromethane -0.76 25 -0.81 1.53 1.47 T etrachloromethane 2.75 39 2.59 2.81 2.91 1-Chlorobutane 2.63 40 2.66 2.51 2.61 2-chloro- 2-methylpropane 2.06 22 2.16 2.86 2.67 2-Bromo-2-methylpropane 2.34 22 2.48 2.93 2.95 lodoethane 2.56 41 2.51 2.02 2.02 CFC12.CFC12 2.67 42 2.80 3.31 3.49 Tetrahydrofuran 2.90 43 3.05 0.35 0.35 1,4-Dioxane 3.27 44 3.58 -0.44 -0.27 Butanone 3.07 45 2.91 0.35 0.24 2,4-Pentanedione 3.69 40 3.62 --- Dimethyl carbonate 2.73 46 3.04 0.00 0.08 Methyl propanoate 2.85 47 2.92 0.70 0.81 Methyl butanoate 3.28 47 3.32 1.20 1.36 Ammonia 1.76 26 1.70 -1.39 -1.49 Dimethylamine 2.19 26 2.53 -0.96 -0.65 Trimethylamine 2.56 48 2.18 0.21 -0.07 Triethylamine 3.50 41 3.52 1.14 1.20 Nitromethane 2.85 44 2.74 -0.10 -0.23 Methanol 3.15 49,50 3.27 -0.59 -0.63 Ethanol 3.51 45 3.48 -0.16 -0.11 Propan-l-ol 3.98 51 3.95 0.42 0.45 Propan-2-ol 3.67 52 3.61 0.19 0.18 Butan-l-ol 4.47 49,50 4.45 1.01 1.01 Pentan-l-ol 4.97 49,53 4.89 1.62 1.57 Decan-l-ol 7.13 54 7.09 4.46 4.39 Sulfur hexafluoride -0.13 25 -0.15 2.10 1.97 Tetramethyltin 2.52 55 2.48 4.05 4.02 Tetraethyltin 4.47 55 4.45 6.09 6.08 Tetraethyllead 4.69 55 4.65 6.18 6.23 Benzene 2.78 56 2.76 2.15 2.19 Toluene 3.31 45 3.23 2.66 2.75 t-Stilbene 7.31 57 7.27 4.53 4.61 Napthalene 5.19 24 5.01 3.46 3.32 Acenapthene 6.25 24 6.21 3.89 3.92 Anthracene 7.36 24 7.29 4.33 4.46 Phenanthrene 7.30 24 7.36 4.50 4.37 Pyrene 8.40 24 8.53 4.90 4.80
105 ref Solute logL logL logP" logP calc calc Benzyl chloride 4.41 44 4.33 3.03 2.97 Benzyl bromide 4.70 44 4.66 2.80 2.94 Methyl 4-aminobenzoate 8.14 24 7.96 1.58 1.49 4-Nitrobenzyl chloride 6.26 48 6.50 2.49 2.61 4-Aminobenzoic acid -- 1.04^' 0.88 2-Hydroxybenzoic acid 7.59 24 7.71 2.24 2.41 4-Hydroxybenzoic acid 8.40 24 8.45 1.62 1.70 Methyl 4-Hydroxybenzoate 8.93 24 8.86 2.09 2.12 Pyridine 3.99 44 3.57 0.55 0.21 Thiophene 2.72 58 2.82 1.68 1.83 Fluoromethane -0.02 38 -0.18 omitted 3-Methylpyridine 4.65 59,60 1.15 omitted “ Values calculated from eqn 15 ^ From ref 4, Table 3
Table 3.10 Values of log log L®"°"calc, log L'^, and log P®uO“™'calc for solutes at 298 K
Solute logL ref logL logP * logP calc calc Helium -1.58 61,62 -1.67 0.44 0.37 Neon -1.45 61,62 -1.51 0.51 0.44 Argon -0.62 61,62 -0.68 0.85 0.89 Krypton -0.18 61,62 -0.24 1.03 1.12 Xenon 0.40 61,62 0.31 1.37 1.48 Hydrogen -1.15 61 -1.16 0.57 0.54 Deuterium - 1.12 61 -1.16 0.61 0.54 Oxygen -0.68 61,62 -0.71 0.83 0.86 Nitrogen -0.91 61,62 -0.95 0.89 1.02 Nitrous oxide 0.40 26 0.38 0.63 0.50 Carbon dioxide 0.29 61 0.45 0.37 0.56 Methane -0.31 62 -0.34 1.15 1.14 Ethane 0.46 61,26 0.42 1.80 1.74 Propane 0.99 26 0.94 2.43 2.34 Butane 1.37 26 1.47 2.89 2.94 Isobutane 1.40 63 1.28 3.10 2.94 Pentane 1.98 64,65 1.98 3.68 3.54 Hexane 2.44 65 2.45 4.26 4.14 2-Methylpentane 2.29 66 2.30 4.13 4.14 3-Methylpentane 2.34 66 2.37 4.18 4.14 2,2-Dimethylbutane 2.12 67 2.16 3.96 4.14 2,3-Dimethylbutane 2.25 67 2.29 3.97 4.14 Heptane 2.88 29 2.93 4.84 4.74 Octane 3.26 44 3.40 5.37 5.34 Cyclopropane 1.26 34 1.22 1.81 1.77 Cyclohexane 2.73 64,65 2.70 3.63 3.67 Ethene 0.34 61 0.32 1.28 1.20
106 ref Solute logL logL logP " logP calc calc Propene 1.03 6S 0.93 2.00 1.83 Isobutene 1.59 6S 1.50 2.44 2.42 Pent-l-ene 1.94 65.30 1.96 3.17 3.02 Isopentene 1.83 65 1.84 3.17 3.01 Hex-l-ene 2.36 30 2.46 3.52 3.62 Isoprene 2.12 65 2.06 2.62 2.63 Ethyne 0.60 69 0.63 0.60 0.59 CF4 -0.81 61.62 -0.79 1.48 1.53 Dichloromethane 2.41 65 2.48 1.45 1.48 Trichloromethane 2.95 65 3.01 2.16 2.23 T etrachloromethane 2.84 65,70 2.68 2.90 2.95 1,2-Dichloroethane 2.92 65.70 3.08 1.61 1.77 1-Chloropropane 2.32 65 2.27 2.08 2.08 t-Butyl chloride 2.17 22 2.25 2.97 3.04 Bromoethane 2.22 65 2.18 1.68 1.68 t-Butyl bromide 2.42 22 2.58 3.02 3.03 lodomethane 2.23 65 2.10 1.58 1.48 lodoethane 2.65 65 2.55 2.11 2.02 CFC12.CFC12 2.71 42 2.93 3.35 3.59 Diethyl ether 2.27 71 2.41 1.10 1.15 Di-n-propyl ether 3.03 72 3.30 2.18 2.33 Dioxane 3.35 73 3.61 -0.36 -0.29 Acrolein 2.35 40 2.21 -0.16 -0.17 Propanone 2.57 65 2.51 -0.22 -0.26 Butanone 3.02 44.73 2.94 0.30 0.25 Ethyl acetate 2.91 71 2.88 0.75 0.81 Acetonitrile 2.60 65 2.59 -0.25 -0.41 Proprionitrile 2.94 65 2.78 0.12 0.00 PrCN 3.39 74 3.13 0.72 0.60 Ammonia 1.76 26 1.59 -1.39 -1.61 Methylamine 2.12 26 2.48 - 1.22 -1.00 Trimethylamine 2.13 26 2.21 -0.22 -0.04 Triethylamine 3.56 44 3.64 1.20 1.33 Nitromethane 2.82 65 2.69 -0.13 -0.32 Acetic acid 4.35 40 4.62 -0.56 -0.24 Methanol 3.02 75 3.12 -0.72 -0.80 Ethanol 3.56 73 3.38 -0.11 -0.24 Propanol 3.83 75 3.90 0.27 0.36 Propan-2-ol 3.68 76 3.55 0.20 0.10 Butan-l-ol 4.47 51 4.43 1.01 0.95 2-Methylpropan-1 -ol 4.21 77,78 4.24 0.91 0.99 Butan-2-ol 4.03 77,78 4.08 0.64 0.70 2-Methylpropan-2-ol 3.76 77,78 3.68 0.48 0.60 Pentan-l-ol 4.96 79,53 4.91 1.61 1.55 SF6 -0.19 61 -0.09 2.04 2.06 Carbon disulfide 2.26 65 2.10 2.41 2.07 Tetramethyltin 2.57 44 2.65 4.10 4.19 Tetraethyltin 4.53 44 4.75 6.15 6.38
107 Solute logL ref logL logP * logP calc calc Benzene 2.87 64,80 2.82 2.24 2.20 Toluene 3.31 44,73 3.32 2.66 2.79 Stilbene 7.34 81 7.44 4.56 4.71 Naphthalene 5.20 82 5.11 3.47 3.34 Anthracene 7.42 76 7.40 4.39 4.46 Hexafluorobenzene 2.54 40 2.64 2.61 2.84 Benzyl chloride 4.47 44 4.43 3.09 3.00 Benzyl bromide 4.67 44 4.75 2.77 2.97 4-Nitrobenzyl chloride 6.22 48 6.61 2.45 2.62 2-Hydroxybenzoic acid 7.52 83 7.64 2.17 2.26 4-Hydroxybenzoic acid 8.36 84 8.36 1.58 1.51 Methyl 4-hydroxybenzoate 8.86 84 8.80 2.02 1.97 Pyridine 3.95 44 3.58 0.51 0.17 Tetradecan-l-ol 8.99 85,86 9.17 6.73 6.93 Hexadecan-l-ol 9.98 85,86 10.12 7.96 8.13 Octadecan-l-ol 11.92 85,86 11.07 10.14 9.32 Eicosan-l-ol 12.01 85,86 12.01 10.47 10.52 “Values calculated from eqn 15
Table 3.11 Values of log log L*’'°"calc, log L'^, log and log pPeOH/W^gjj. 298 K
Solute logL ref logL lo g P “ logP calc calc Helium -1.63 '62 -1.63 0.39 0.40 Neon -1.49 62 -1.47 0.47 0.47 Argon -0.64 62 -0.65 0.83 0.91 Krypton -0.19 62 -0.21 1.02 1.14 Xenon 0.39 62 0.33 1.36 1.49 Hydrogen -1.22 26 -1.13 0.50 0.57 Oxygen -0.70 62 -0.69 0.81 0.88 Nitrogen -0.94 62 -0.92 0.86 1.04 Nitrous oxide 0.37 26 0.39 0.60 0.46 Carbon monoxide -0.82 26 -0.75 0.80 0.89 Methane -0.33 62 -0.32 1.13 1.16 Ethane 0.44 26 0.44 1.78 1.74 Pentane 1.94 87 1.98 3.64 3.50 Hexane 2.44 87 2.44 4.26 4.09 2-Methylpentane 2.42 88 2.29 4.26 4.09 3-Methylpentane 2.44 88 2.36 4.28 4.09 2,2-Dimethylbutane 2.15 88 2.15 3.99 4.09 2,3-Dimethylbutane 2.31 88 2.28 4.03 4.09 Octane 3.39 71 3.37 5.50 5.27 Isooctane 2.85 87 2.85 4.97 5.27 Methylcyclopentane 2.57 87 2.66 3.74 3.64 Cyclohexane 2.72 87 2.70 3.62 3.68
108 Solute logL ref logL logP" logP calc calc " ~ w ...... Ethene 0.35 0.34 1.29 1.22 Hex-l-ene 2.42 87 2.45 3.58 3.58 Hept-l-ene 2.88 87 2.90 4.10 4.17 CF4 -0.85 62 -0.79 1.44 1.47 Dichloromethane 2.34 87 2.46 1.38 1.45 Diethyl ether 2.29 87 2.40 1.12 1.10 Butanal 2.71 87 2.80 0.38 0.46 Acrolein 2.36 40 2.28 -0.15 -0.21 Propanone 2.49 87,75 2.48 -0.30 -0.35 Butanone 2.85 87.75 2.90 0.13 0.15 Ethyl acetate 2.84 87 2.84 0.68 0.70 Propyl acetate 3.16 87 3.30 1.11 1.30 PrCN 3.28 89 3.08 0.61 0.46 Triethylamine 3.64 90 3.62 1.28 1.30 Methanol 3.07 87 3.12 -0.67 -0.77 Ethanol 3.41 87 3.38 -0.26 -0.22 Propan-l-ol 3.98 91,53 3.88 0.42 0.36 Propan-2-ol 3.66 77 3.54 0.18 0.11 Butan-l-ol 4.45 81,53 4.41 0.99 0.94 Pentan-l-ol 4.95 51 4.88 1.60 1.52 SF6 -0.21 62 -0.10 2.02 1.97 CS2 2.43 92 2.14 2.58 2.21 Benzene 2.83 87 2.81 2.20 2.21 Toluene 3.29 87 3.31 2.64 2.79 o-Xylene 3.80 93,70 3.90 3.14 3.28 Chlorobenzene 3.82 94 3.58 3.00 2.88 trans-Stilbene 7.38 57 7.45 4.60 4.73 Naphthalene 5.05 82 5.10 3.32 3.40 Acenaphthene 6.36 b 6.33 4.08 4.02 Anthracene 7.46 7.40 4.43 4.61 Phenanthrene 7.33 d 7.49 4.53 4.50 Pyrene 8.58 8.63 5.08 4.95 4-Nitrobenzyl chloride 6.16 48 6.10 2.39 2.36 Methyl 4-hydroxybenzoate 8.81 84 8.73 1.97 1.89 4-Hydroxybenzoic acid 8.29 84 8.33 1.51 1.56 2-Hydroxybenzoic acid 7.45 f 7.61 2.10 2.29 “ Values calculated from eqn 15 * This work, from log S = -0.54, log Co = -6.90, log Sw = -4.62 ^ This work, from log S = -2.00, log Cq = -9.46, log Sw = -6.43 ‘^This work, from log S = -0.64, log Cg= -7.97, log Sw = -5.17 * This work, from log S = -1.07, log Cq = -9.65, log Sw = -6.15 ^This work, from log S = 0.18, log Cq = -7.27, log Sw = -1.92
109 Table 3.12 Values of log log L^^^^’^calc, log L^, log pH^^OHAV j^g pHexOH^calc foj. solutes at 298 K
Solute logL ref logL logP" logP calc calc .. Helium -1.67 -1.67 0.35 0.33 Neon -1.52 25 -1.51 0.44 0.41 Argon - 0.66 25 -0.68 0.81 0.85 Krypton - 0.20 25 -0.23 1.01 1.09 Xenon 0.38 25 0.32 1.27 1.44 Hydrogen -1.18 95 -1.16 0.54 0.51 Oxygen -0.72 25 -0.71 0.79 0.82 Nitrogen -0.95 25 -0.95 0.85 0.99 Nitrous oxide 0.36 26 0.39 0.59 0.46 Methane -0.35 25 -0.34 1.11 1.10 Ethane 0.43 96 0.43 1.77 1.70 Pentane 1.97 29 1.99 3.67 3.50 Hexane 2.50 97 2.46 4.32 4.10 Heptane 2.93 76 2.94 4.89 4.70 2,2,4-Trimethylpentane 2.88 76 2.87 5.00 5.30 Cyclohexane 2.75 97 2.71 3.65 3.66 Ethene 0.32 37 0.32 1.26 1.17 CF4 -0.89 25 -0.80 1.40 1.44 T etrachloromethane 2.83 97 2.71 2.89 2.96 Dipropylether 3.02 76 3.21 2.17 2.24 Butanone 2.84 98 2.87 0.12 0.17 Butanonitrile (3.16) 3.24 99 3.12 0.57 0.55 Pentanol 4.90 b 4.80 1.55 1.45 Hexanol 5.35 b 5.27 2.02 2.04 Heptanol 5.92 b 5.75 2.83 2.64 Octanol 6.29 b 6.22 3.29 3.24 Tetradecan-l-ol 8.94 76 9.08 6.68 6.81 Hexadecan-l-ol 9.94 76 10.03 7.92 8.01 Octadecan-l-ol 10.94 76 10.98 9.16 9.20 Eicosan-l-ol 12.08 76 11.92 10.54 10.39 SF6 -0.25 25 -0.09 1.98 1.97 Toluene 3.32 97 3.33 2.67 2.80 t-Stilbene 7.41 57 7.48 4.63 4.78 Biphenyl 6.11 76 6.00 4.16 4.11 Naphthalene 5.22 100 5.12 3.49 3.41 Acenaphthene 6.39 101 6.36 4.03 4.04 Anthracene 7.53 76 7.41 4.50 4.62 Phenanthrene 7.35 102 7.52 4.55 4.53 Fluoranthene 8.57 103 8.67 5.12 5.13 Pyrene 8.66 102 8.67 5.16 4.99 Benzoic acid 6.89 104 6.92 1.79 1.73 o-Hydroxybenzoic acid 7.43 76 7.57 2.08 2.22 p-Hydroxybenzoic acid 8.21 84 8.23 1.43 1.45 Methyl 4-hydroxybenzoate 8.75 84 8.76 1.91 1.93
110 Solute logL ref logL lo g P “ logP calc calc Diphenylsulfone 9.65 76 9.69 2.26 2.28 Diuron 10.80 76 10.77 2.83 2.80 “ Values calculated from eqn 15
Table 3.13 Values of log log L"'>’‘°"°''°“calc, log L"', log and lo pHeptoHcaic for solutes at 298 K
Solute logL ref logL lo g P “ logP calc calc Helium -1.71 "15...... ■■■ - 1.68 0.31 0.27 Neon -1.54 25 -1.52 0.42 0.35 Argon -0.68 25 -0.70 0.79 0.81 Krypton -0.22 25 -0.26 0.99 1.06 Xenon 0.38 25 0.29 1.35 1.43 Oxygen -0.75 25 -0.73 0.76 0.78 Nitrogen -0.98 25 -0.97 0.82 0.96 Methane -0.35 25 -0.36 1.11 1.08 Ethane 0.44 37 0.39 1.78 1.70 Propane 0.87 76 0.91 2.31 2.32 Butane 1.43 76 1.43 2.95 2.94 Isobutane 1.22 76 1.24 2.92 2.94 Pentane 1.94 76 1.94 3.64 3.56 Hexane 2.45 76 2.41 4.27 4.19 Cyclopentane 2.27 76 2.23 3.15 3.09 Methylcyclopentane 2.57 76 2.64 3.74 3.69 Cyclohexane 2.73 76 2.68 3.63 3.73 Ethene 0.31 26 0.31 1.25 1.14 CF4 -0.91 25 -0.84 1.38 1.44 Butanonitrile 3.16 99 3.08 0.49 0.47 Hexanol 5.30 51 5.31 2.07 2.04 Heptanol 5.86 51 5.77 2.77 2.66 Octanol 6.24 51 6.24 3.24 3.28 SF6 -0.27 25 -0.16 1.96 1.98 Benzene 2.81 76 2.81 2.18 2.20 t-Stilbene 7.46 57 7.52 4.68 4.85 Biphenyl 6.13 76 6.01 4.18 4.16 Acenaphthene 6.41 76 6.37 4.05 4.08 Anthracene 7.58 76 7.47 4.55 4.67 Phenanthrene 7.41 76 7.56 4.61 4.58 Fluoranthene 8.62 76 8.67 5.17 5.19 Pyrene 8.73 76 8.72 5.23 5.03 Benzoic 6.80 76 6.84 1.70 1.67 o-Hydroxybenzoic 7.40 76 7.49 2.05 2.17 p-Hydroxybenzoic 8.16 76 8.21 1.38 1.37 Methyl 4-hydroxybenzoate 8.70 84 8.60 1.86 1.84 Diphenylsulfone 9.58 76 9.67 2.19 2.22 Diuron 10.76 76 10.72 2.79 2.76 “Values calculated from eqn 15
111 Table 3.14 Values of log log L”“""calc,OctOH, log L", logW P .OctOHAV and log pOctOHAv at 298 K
Solute logL ref logL logP" logP Set calc calc Helium -1.72 110,114 -1.76 0.30 0.23 test Neon -1.57 25,26 -1.60 0.39 0.30 Argon -0.71 25,26 -0.77 0.76 0.75 Krypton -0.24 25,26 -0.32 0.97 0.99 Xenon 0.38 26 0.23 1.35 1.34 Hydrogen -1.29 26 -1.25 0.43 0.40 test Oxygen -0.77 25 -0.80 0.74 0.72 Nitrogen - 1.02 25 -1.04 0.78 0.88 Nitrous oxide 0.33 26 0.29 0.56 0.38 Carbon monoxide - 0.88 114 -0.88 0.74 0.71 Carbon dioxide 0.16 114 0.34 0.24 0.42 test Methane -0.38 25 -0.42 1.08 1.00 Ethane 0.42 26 0.34 1.76 1.60 Propane 0.97 26 0.87 2.41 2.21 Butane 1.53 26 1.40 3.05 2.81 Pentane 1.95 105 1.91 3.65 3.41 test Hexane 2.44 115 2.39 4.26 4.01 Heptane 2.95 116 2.86 4.91 4.62 Octane 3.30 117 3.33 5.41 5.22 Cyclohexane 2.71 105 2.66 3.61 3.59 Methylcyclohexane 3.05 116 2.98 4.26 4.20 test Ethene 0.28 37 0.23 1.22 1.07 Pent-l-ene 1.93 116 1.88 3.16 2.89 Hex-l-ene 2.41 118,116 2.37 3.57 3.48 Oct-l-ene 3.35 118 3.31 4.76 4.70 Non-l-ene 3.83 118 3.78 5.34 5.30 test 2-Methylbuta-1,3-diene 2.06 65 1.99 2.56 2.52 Cyclohexene 2.83 116 2.82 3.10 2.91 Tetrafluomethane -0.95 25 -1.16 1.34 1.84 Chloromethane 1.39 119 1.22 0.99 0.85 Dichloromethane 2.27 b 2.41 1.31 1.42 test T richloromethane 2.80 b 2.95 2.01 2.17 T etrachloromethane 2.79 b 2.65 2.85 2.91 1,1 -Dichloroethane 2.41 105 2.69 1.79 1.86 1,2-Dichloroethane 2.78 b 3.01 1.47 1.71 1,1,1-Trichloroethane 2.70 b 2.67 2.56 2.53 test 1,1,2-Trichloroethane 3.40 105 3.81 1.94 2.14 Hexachloroethane 4.47 4.56 4.32 4.35 1-Chloropropane 2.24 b 2.20 2.00 1.99 1,2-Dichloropropane 2.96 115 2.99 2.03 2.04 1-Chlorobutane 2.72 b 2.69 2.60 2.59 test cis-1,2-Dichloroethene 2.56 105 2.85 1.70 1.82 Trichloroethene 2.99 105 3.10 2.67 2.72 T etrachloroethene 3.48 b 3.36 3.55 3.35 Dibromomethane 3.07 b 3.28 1.63 1.81
112 Solute logL ref logL logP" logP Set calc calc Bromoethane 2.11 65 2.11 1.57 1.59 test lodomethane 2.16 65 2.05 1.51 1.42 lodoethane 2.59 65 2.50 2.05 1.95 Dibromochloromethane 3.59 105 3.53 2.31 2.30 Dimethylether 1.37 119 1.52 -0.03 -0.17 Diethylether 2.19 b 2.22 1.02 0.90 test Dipropylether 2.97 b 3.11 2.12 2.09 Diisopropylether 2.66 b 2.72 1.49 1.60 Dibutylether 3.89 b 4.02 3.28 3.29 Diisobutylether 3.40 b 3.57 2.86 3.35 Dipentylether 4.80 b 4.91 4.30 4.50 test t-Butyl methyl ether 2.58 b 2.60 0.96 0.88 Tetrahydrofuran 2.86 b 2.92 0.31 0.14 Tetrahydropyran 3.22 b 3.34 0.93 0.77 Dioxane 3.17 b 3.40 -0.54 -0.51 Propanone 2.31 b 2.31 -0.48 -0.46 test Butanone 2.77 b 2.74 0.05 0.05 Pentan-2-one 3.19 b 3.18 0.61 0.66 Pentan-3-one 3.20 b 3.22 0.70 0.69 3-Methylbutan-2-one 3.04 b 3.10 0.66 0.69 Hexan-2-one 3.68 b 3.65 1.27 1.26 test Heptan-2-one 4.15 b 4.12 1.92 1.86 Cyclopentanone 3.67 b 3.68 0.22 0.07 Methyl formate 1.75 b 1.69 -0.29 -0.32 Ethyl formate 2.19 b 2.22 0.24 0.28 Propyl formate 2.66 b 2.76 0.84 0.91 test Methyl acetate 2.31 b 2.32 0.01 0.01 Ethyl acetate 2.70 b 2.69 0.54 0.61 Propyl acetate 3.17 b 3.16 1.12 1.23 Butyl acetate 3.65 b 3.67 1.71 1.82 Pentyl acetate 4.12 b 4.13 2.28 2.42 test Isopropyl acetate 2.93 b 2.91 0.99 1.16 Isobutyl acetate 3.45 b 3.49 1.72 1.76 Isopentyl acetate 3.94 b 4.03 2.32 2.36 Hexyl acetate 4.58 b 4.61 2.92 3.02 Ethyl propanoate 3.15 b 3.14 1.18 1.25 test Ethyl butanoate 3.56 b 3.58 1.73 1.84 Acetonitrile 2.31 b 2.45 -0.54 -0.53 Proprionitrile 2.69 b 2.63 -0.13 -0.13 Butanonitrile 3.12 b 2.99 0.45 0.48 2-Cyanopropane 2.87 b 2.89 0.37 0.49 test Pentanonitrile 3.60 b 3.52 1.02 1.08 Hexanonitrile 4.08 b 3.99 1.63 1.68 Methylamine 1.90 26 2.22 -1.44 -1.29 Butylamine 3.61 <1 3.49 0.50 0.38 Dimethylamine 2.00 26 2.26 -1.15 - 1.00 test Dipropylamine 3.59 d 3.94 0.91 1.25 Nitromethane 2.52 b 2.56 -0.43 -0.42 Nitroethane 2.88 b 2.93 0.16 0.08
113 Solute logL ref logL logP" logP Set calc calc Nitropropane 3.25 b 3.30 0.80 0.75 Dimethylformamide 4.38 b 4.04 -1.35 -1.65 test Methanol 2.84 b 2.85 -0.90 -1.07 Ethanol 3.20 b 3.12 -0.47 -0.51 Propan-l-ol 3.68 b 3.63 0.12 0.09 Propan-2-ol 3.38 b 3.27 -0.10 -0.20 Butan-l-ol 4.19 117.116 4.17 0.73 0.69 test 2-Methylpropan-1 -ol 3.93 b 3.98 0.63 0.72 Butan-2-ol 3.80 b 3.81 0.41 0.41 2-Methylpropan-2-ol 3.50 b 3.38 0.22 0.28 Pentan-l-ol 4.69 b 4.65 1.34 1.29 3 -Methy Ibutan- l-ol 4.52 b 4.55 1.28 1.31 test Hexan-l-ol 5.18 b 5.12 1.95 1.89 Octan-l-ol 6.03 117 6.07 3.03 3.09 Hexadecan-l-ol 9.90 9.89 7.88 7.88 Octadecan-l-ol 10.93 10.84 9.15 9.09 Eicosan-l-ol 12.06 11.79 10.52 10.29 test 2-Chloroethanol 4.30 4.55 -0.48 -0.46 Cyclohexanol 5.18 b 5.16 1.17 1.03 Acetic acid 4.31 b 4.33 -0.60 -0.50 Dimethylsulfoxide 4.96 b 4.67 -2.89 -3.13 Sulfur hexafluoride -0.30 114,25 -0.47 1.93 2.41 test Carbon disulfide 2.28 65 2.09 2.43 2.04 Tetramethyltin 2.62 55 2.52 4.15 4.14 Benzene 2.80 b 2.76 2.17 2.14 Toluene 3.31 b 3.27 2.66 2.74 Ethylbenzene 3.72 118,116 3.70 3.14 3.32 test o-Xylene 3.90 118,116 3.87 3.24 3.24 m-Xylene 3.79 118,116 3.76 3.18 3.27 p-Xylene 3.79 116 3.77 3.20 3.26 Propylbenzene 4.09 118 4.12 3.70 3.92 Isopropylbenzene 3.98 118 3.98 3.76 3.89 test Hexamethylbenzene 6.31 6.45 4.92 5.11 trans-Stilbene 7.48 7.48 4.70 4.75 Biphenyl 6.15 5.99 4.20 4.09 Naphthalene 5.19 5.11 3.46 3.38 Acenaphthene 6.31 6.37 4.03 4.03 test Fluorene 6.83 6.83 4.27 4.30 Anthracene 7.55 7.47 4.52 4.62 Phenanthrene 7.52 7.56 4.72 4.52 Fluoranthene 8.61 8.73 5.16 5.17 Pyrene 8.75 8.76 5.25 5.03 test 1,2-Dichlorobenzene 4.36 4.41 3.36 3.46 1,4-Dichlorobenzene 4.46 4.31 3.53 3.56 1,2,3-T richlorobenzene 5.19 5.25 4.28 4.15 1,3,5-T richlorobenzene 4.85 4.83 4.28 4.26 1,2,3,4-T etrachlorobenzene 5.64 5.95 4.66 4.68 test 1,2,3,5-T etrachlorobenzene 5.55 5.69 4.88 4.74 1,2,4,5-Tetrachlorobenzene 5.62 5.69 4.64 4.73
114 Solute logL ref logL logP" logP Set ’ calc calc Pentachlorobenzene 6.49 6.46 4.93 5.23 Hexachlorobenzene 7.17 7.30 5.83 5.80 1,4-Dibromobenzene 5.21 5.16 3.77 3.96 test Anisole 4.01 4.01 2.21 2.16 Benzonitrile 4.46 4.38 1.37 1.44 Diphenylamine 7.64 7.46 3.54 3.41 4-Ethoxyacetanilide 9.59 9.25 1.55 1.36 2-Hydroxybenzoic acid 7.44 7.41 2.09 2.10 test 4-Hydroxybenzoic acid 8.08 8.06 1.30 1.29 Methyl 4-hydroxybenzoate 8.57 8.57 1.73 1.86 Piperidine 4.04 117 4.09 0.29 0.20 N-Methylpyrrolidine 3.64 117 3.29 0.72 0.79 Pyrrolidine 4.07 117 4.17 0.05 0.17 test Dimethylacetamide 5.33 b omitted Propanal 3.02 b omitted Butanal 3.39 b omitted Hexanal 4.41 b omitted Octanal 5.36 b omitted “ Values calculated from eqn 15 * This work (Andrew Dallas Data) ^ This work. Table 4.01 Personal communication from Professor Cabani
Table 3.15. Values of log L**"®", log L ®“®“calc, log LT log and log P®“° “ ™calc for solutes at 298 K
Solute logL ref logL logP" logP calc calc Helium -1.71 114 -1.79 0.31 0.23 Neon -1.60 114 -1.63 0.36 0.30 Argon -0.73 114 -0.79 0.74 0.75 Krypton -0.26 114 -0.34 0.95 0.99 Xenon 0.36 120 0.22 1.33 1.35 Hydrogen -1.34 26 -1.27 0.38 0.40 Oxygen -0.80 114 -0.82 0.91 0.72 Nitrogen -1.07 114 -1.06 0.73 0.89 Nitrous oxide 0.40 26 0.21 0.63 0.28 Carbon monoxide -0.91 114 -0.90 0.71 0.73 Carbon dioxide 0.10 114 0.27 0.18 0.33 Methane -0.40 114 -0.44 1.06 1.01 Octane 3.30 121 3.35 5.41 5.24 2,2,4-T rimethylpentane 2.86 33 2.81 4.98 5.24 Cyclohexane 2.51 121 2.68 3.41 3.65 CF4 -1.00 114 -0.96 1.29 1.26 Diethyl ether 2.16 122 2.20 0.99 0.92 Diisopropyl ether 2.74 122 2.76 1.57 1.67 Methyl t-butyl ether 2.51 122 2.63 0.89 0.99 Butanonitrile 2.87 89 2.83 0.20 0.25
115 Solute logL ref logL logP " logP calc calc Methanol 2.71 m 2.85 -1.03 -1.05 Ethanol 3.11 123 3.12 -0.56 -0.48 Propan-l-ol 3.60 123 3.64 0.04 0.11 Propan-2-ol 3.29 123 3.28 -0.19 -0.14 Butan-l-ol 4.23 124,125 4.18 0.77 0.71 2-Methylpropan-1 -ol 3.98 124 3.99 0.68 0.75 Butan-2-ol 3.94 124 3.82 0.55 0.47 Pentan-l-ol 4.80 124 4.66 1.45 1.31 3-Methylbutan-l-ol 4.53 124 4.56 1.29 1.33 Hexan-l-ol 5.20 124,125 5.13 1.97 1.91 Nonan-l-ol 6.57 51 6.57 3.72 3.71 Decan-l-ol 7.03 51 7.05 4.36 4.31 Undecan-l-ol 7.53 51 7.53 5.03 4.91 Tetradecan-l-ol 8.88 76 8.97 6.62 6.71 Hexadecan-l-ol 9.86 76 9.93 7.84 7.92 Octadecan-l-ol 10.90 76 10.89 9.12 9.12 Eicosan-l-ol 12.01 76 11.85 10.47 10.33 2-Chloroethanol 4.20 124 4.13 -0.41 -0.46 SF6 -0.36 114 -0.27 1.87 1.77 Benzene 2.62 121 2.74 1.99 2.15 Pyrene 8.83 76 8.81 5.33 5.23 Benzoic acid 6.75 76 6.74 1.65 1.59 Methyl 4-hydroxybenzoate 8.35 112,85 8.49 1.51 1.69 Diphenylsulfone 9.42 76 9.42 2.03 2.04 Diuron 10.60 76 10.56 2.63 2.63 "Values calculated from eqn 15
116 Chapter 5 High Performance Liquid Chromatography (HPLC)
The bonded phase of a column and mobile phase are the two components in a reversed-phase high performance chromatographic system that will most influence retention. This chapter will illustrate the application of the Abraham General Solvation Equation in providing a better understanding and quantification of the various properties involved in retention. The algorithms obtained from the retention of a series of solutes in acetonitrile/water mobile phases will enable the characterization of several bonded phases in terms of fundamental physicochemical parameters. In addition, this work will investigate the suitability of a HPLC method as an alternative to water-partition measurements in obtaining Abraham descriptors.
5.1 Introduction
Chromatography was discovered around the turn of the century by a Russian botanist named Mikhail S. Tswett ' who separated green leaf pigments, namely the chlorophylls, into a series of coloured bands using powdered calcium carbonate tamped firmly into a glass tube. The coloured bands shown on the column of the adsorbent bed evoked the term chromatography for this kind of separation. Although colour has little to do with modem chromatography the name has persisted and despite its irrelevance is still used for all separation techniques that employ a mobile and a stationary phase.
In the 1940s Martin and Synge developed the theory of partition chromatography and used mathematics to describe the separation process resulting from the use of a liquid- coated solid phase and a moving liquid phase. As a result, Martin was awarded the Nobel Prize in chemistry\ Although the historical start of liquid chromatography(LC) was earlier than gas chromatography(GC), GC grew more rapidly due to Martin publishing the first application of gas chromatography with James. “Gas-liquid” chromatography provided much better efficiencies than “liquid-liquid” or “liquid- solid” chromatography, supplying impetus for application of this new separation tool. As a result, many fundamental studies working toward the optimization of high-speed
117 gas separations took place during the 1950s. However, in the 1960s, the speed of LC was increased as a result of new developments such as surface-coated stationary phase packings and smaller diameter spherical packings. All of these advances contributed to the phrase, high performance liquid chromatography (HPLC); a term which today is the most common name for modem LC instrumentation.
There are three types of chromatographic development^ L Elution development. This is best described as a series of absorption-extraction processes which are continuous from the time a small volume of the sample is injected into the chromatographic system, until the various sample components exit the column in the form of concentration bands separated in time. it. Frontal analysis. In frontal analysis the sample is fed continuously into the column, either as a pure sample or as a solution in the mobile phase. Only part of the first compound is eluted in a relatively pure state, each subsequent component being mixed with those previously eluted. Hi. Displacement development. This is a version of elution chromatography which depends on the competition between solutes for the active sites of the absorbent and is applicable to strongly retained solutes which cannot be eluted by the mobile phase alone. It is assumed that the sample components distribute themselves on the column in zones according to their adsorption strength. To develop the separation, a substance (the displacer) that is even more strongly held on the stationary phase than any of the solutes is introduced into the mobile phase.
5.2 Concepts of Reversed-Phase HPLC
Liquid chromatography is a separation method in which a mixture of components is resolved into its constituent parts by passage through a chromatographic column. It is carried out by passing the mobile phase, containing the mixture of the components, through the stationary phase, which consists of a column packed with solid particles. Physical and chemical forces acting between the solutes and the two phases are responsible for the retention of solutes on the chromatographic column. It is the differences in the magnitude of these forces that determine the resolution and hence separation of the individual solutes. The term reversed-phase liquid chromatography
118 derived from the fact that the mobile phase is more polar than the stationary phase, which is the opposite of normal-phase chromatography. The reversed-phase method is popular because of numerous advantages it has for the potential chromatographer. The inertness of the stationary phase allows the exploitation of a wide range of solvent effects through variation of the mobile phase composition. Analytes with a wide range of polarity can be separated by this technique. Also, the fact that it requires aqueous mobile phases means that reversed-phase liquid chromatography is generally compatible with most aqueous samples which can be directly injected onto the column without the need for pre-treatment.
When a sample is injected onto a column, the compounds that do not interact with the stationary phase will be eluted at time to in the void volume Vq . The retention time of the sample (tr) is the time from injection to the time of maximum concentration in the eluted peak. Quite simply, in order for any two solutes to be separated, they must be retained to a different extent during their passage through the column so that different volumes of solvent are required to elute each solute (Fig. 5.1).
Fig. 5.1 Chromatographic separation of solutes
Detector response
Injection Time or volume
Here components A and B are separated, whilst a non-retained compound would be eluted in the void volume.
119 5.2.1 Fundamental Relationships of Chromatography The retention volume Vr is the volume of solvent required to elute the solute as measured from the centre of the chromatographic band. Vr and tr can be related by the following equation:
Vr = trX / (1) where/represents the flow rate.
The capacity factor k’ is a common measure of the degree of retention and providing that the flow rate of the mobile phase is kept constant, can be calculated from the following equation :
k’ = (tr — to)/to = (Vr — Vo)/V 0 (2) where k is the number of column volumes required to elute a particular solute, tr is the time taken for a specific solute to reach the detector(retention time) and to is the time taken for a non-retained species (e.g. uracil) to reach the detector(holdup time). The same value of k’ is obtained if volumes are used instead of times. The capacity factor of a column is mostly a function of the packing material but can be manipulated to a degree by varying the solvent strength.
The retention time of any solute can be calculated from:
tr = tm(l+k’) = (L/u)(l4-k’) (3) where L is the column length and u the average mobile phase velocity.
It is generally assumed that, initially, injection volumes will spread to give a Poisson distribution and then to a Gaussian distribution. A typical Gaussian peak is shown in Fig. 5.2.
120 Fig. 5.2 Characteristic properties of a Gaussian peak^
Tangent* drawn to ttie inflexion points
1.000
0.882 £ E Inflexion points
E 0.607 tvi ■ * 2 o 0.324 3 0.134 4«r 0.044 5 The efficiency of a chromatographic column is measured by the number of theoretical plates (N) to which the column is equivalent. The term was originally used to describe the process of distillation and can be visualised as a series of hypothetical layers in which the solute concentrations in the relevant phases are assumed to be in equilibrium. The number of theoretical plates is also a measure of the amount of band broadening caused by a column. It can be calculated from the formula: N = (tr /a ) (4) where a is the mathematical term representing the standard deviation of a Gaussian peak. Assuming that peak shapes are Gaussian, various peak width measurements can be employed to calculate N. A commonly used method is the tangent method where tangents to the curve at the point of inflexion intercept the baseline at a distance of 4a apart. Eqn 4 can then be written as: N = 16x(tr/wby (5) 121 Alternatively, the peak width can be measured at the inflexion point(wi) or at half height (Wh), giving rise to eqns 6 and 7 respectively. N = 4 X (tr / Wi)^ (6) N = 5.54 X (tr/ Wh) (7) N is often quoted as a measure of the column performance and the larger the number, the better the column. In general, N increases for smaller stationary phase particle diameters, low mobile phase flow rates, higher separation temperatures, less viscous solvents, and smaller solute molecules. The value of N is independent of the retention time of a solute but is proportional to column length. To enable a direct comparison to be made between different columns, it is more useful to use the height equivalent to a theoretical plate, H (also referred to as the plate height), which is given by the ratio of the column length (L) to the column plate number, eqn 8. H = L/N (8) The selectivity of the chromatographic system is a measure of the difference in retention times(or volumes) between two given peaks and describes how effectively a chromatographic system can separate two compounds. Selectivity is defined in terms of cc, which is simply the ratio of capacity factors: (9) The value for a can range from unity when the retention times of the two components are identical, to infinity if the first component of interest is eluted in the void volume. Resolution is defined as the degree of separation between two peaks and is related to the chromatographic variables of selectivity, efficiency and time. Rs = [N“/2][(a-l)(/ a +l)[k’Av/ (l+kVv)] (10) where k’av = (k’ i + k’2)/2 ; ki and k% are the capacity factors of the two peaks. 122 5.2.2 Retention Mechanisms Partitioning This is the most simple possible model and assumes that the transfer process is dominated by partitioning. The principle driving force for the transfer of solute is its relative chemical affinity for mobile- and stationary-phase molecules. On the molecular level, eluite retention involves the creation of an eluite-sized cavity in the organic stationary phase, the transfer of the eluite into the cavity and the subsequent closing of the eluite-sized cavity in the mobile phase Several stationary phase models have been proposed to describe retention by partition mechanism in reversed phase HPLC^ : i. Liquid hydrocarbon partition : The bonded chains form a liquid hydrocarbon layer over the siliceous surface. The retention mechanism is ordinary bulk phase partitioning between the organic (stationary) phase and hydroorganic mobile phase"^ (Fig. 5.3). ii. Liquid crystalline hydrocarbon partition^’. Model in which the alkyl-silica bonded phase is considered as a liquid-crystalline hydrocarbon layer. This is a refinement of the bulk liquid hydrocarbon layer because the model takes into account the organization of the stationary phase chains in terms of the bonded chain length, the intrinsic chain stiffness, surface coverage, and also the configuration of the chains in various mobile phases. Retention is believed to occur via the partition mechanism when the eluite fully penetrates the liquid- crystalline hydrocarbon layer"^ (Fig. 5.4). iii. Amorphous-crystalline hydrocarbon partition model : According to this "interphase" model, bonded chains of stationary phases will have greater orientational order near their anchored ends than their free ends. The nature of variation of stationary phase properties with distance from the interface depends on surface density and length of the chains and contrasts bulk liquid phases where properties are considered to be invariant^’^ (Fig. 5.4). 123 Fig.5.3 Lattice model of liquid hydrocarbon partition ■> Eluent molecule Mobile phase # e Eluite molecule • # e i e 0 i o o -► Hydrocarbon layer -0 -i O Stationary phase o Cubic lattice ► Siliceous surface Fig. 5.4 Lattice model of liquid-crystalline/amorphous crystalline hydrocarbon partition -►Eluent m olecule Mobile phase e I e • • I • -► Bonded ligate -► Eluite molecule Stationary phase Cubic lattice Silceous surface 124 Adsorption This is an alternative view that solute transfer is not a process of partitioning into the stationary phase but instead involves adsorption of the solute to the hydrocarbon surface of the stationary phase Retention is believed to be governed solely by the adsorption mechanism when the density of bonded non-polar functions is high enough for the chains to interact laterally among themselves and to disallow penetration of eluite molecules into the hydrocarbon layer at the chromatographic surface. The solute molecules or adsorbates migrate from the liquid phase to the interface (the adsorptive monolayer) and displace the physically adsorbed molecules of the solvent. Solvophobic theory The solvophobic theory developed by Melander and Hovarth ^ assumes that aqueous mobile phases are highly structured due to the tendency of water molecules to self associate by hydrogen-bonding and that this structuring is perturbed by the presence of non-polar solute molecules. As a consequence of the high cohesive energy of the solvent, the less polar solutes are literally ‘squeezed out’ of the mobile phase and are bound to the hydrocarbon portion of the stationary phase. This model emphasizes interactions in the mobile phase alone and suggests that interactions with the stationary phase are unimportant. If the solute contains polar functional groups then the dipolar or hydrogen bonding interactions of these groups with the mobile phase will oppose the solute transfer mechanism. Two stationary phase configurations are possible when considering the solvophobic theory with the isolated solvated hydrocarbon chains model.(Fig. 5.5) Fig 5.5 Models of stationary-phase chains^ (a) fur’ : Bonded chains are extended in organic (b) ‘stack’ : In water-rich mobile phases, modified rich eluents. the structure collapses so that the bonded chains are in close contact with each other. Eluite molecule ». Bonded chain Si -► Siliceous surface 125 5.2.3 Stationary Phases Macroporous silica gel is by far the most important adsorbent for liquid-solid chromatography\ and is also the material used to prepare most bonded phase packings. Different silica gels are characterized by their shape and mean particle size, specific surface area, mean pore diameter and specific pore volume. Their chromatographic properties are also influenced by the type and number of surface functional groups, the presence of trace metal impurities, the surface pH, and solubility in mobile phases of different pH. For analytical applications, microparticles of 5 to 10 micrometers in diameter are the most widely used. These sizes provide a reasonable compromise between column performance, stability, operating pressure and separation time. Particle shape may become more important as the particle size is reduced, and spherical microparticles are considered superior for particle diameters less than 5 micrometers. The particle type of the stationary phase is also of importance since the stability of the column bed is greatly enhanced by improved silica smoothness and spherical microparticles. The likelihood of particle shearing and breakage during the bonding or packing process and in everyday use is significantly reduced. Furthermore, backpressures are lower, leading to longer performance. There are several types of surface silanols (Fig. 5.6) that have their own unique properties that affect both chemical derivatization reactions and adsorptive interactions with solutes. The relative distribution of these different types of silanols may affect the characteristics of silica-based stationary phases more than the absolute number of surface silanols Fig. 5.6 Types of surface silanols present on chromatographic silica gel S i \ O Siloxane Si — OH Free silanol Si OH Si— 0> Si Geminal silanols H Vicinal (or associated) OH Si— O silanols H 126 The silanol groups on the surface of the silica gel may be chemically modified to give stationary phases with specific properties. The performance of the bonded phase is determined by the base silica and its pre-treatment, the choice of functional group, the carbon load and also whether end-capping has taken place. The functional group affects selectivity and efficiency; the nature of the functional group controls selectivity, while chain length controls column efficiency. Although shorter chains result in more efficient columns, the sample capacity decreases with decreasing chain length. Derivatization may be carried out in three main ways 1,9,10 1) By reaction of the silanol with an alcohol, producing an alkoxy silane. These are relatively simple to prepare but are hydrolysed easily and therefore cannot be used with aqueous or alcoholic mobile phases. I I — Si— OH + ROH ------► Si— OR + H2O I I 2) By the production of a chloride using thionyl chloride followed by reaction with an amine to give an alkylaminosilane. These products are more stable to hydrolysis than the alkoxysilanes; however, their preparation is more complex (two stage synthesis) and considerable difficulties are frequently encountered in removing unwanted reaction products. I I — Si— OH + SOCI2------► Si— Cl + SO2 + HCi + H2N—R Si— NH— R + HCI 127 3) By reaction with organosilanes to give a siloxane bond. This is the method most often used in the preparation of chemically bonded silica gel packing materials as it gives a stable product using a straightforward reaction scheme, and has largely superseded methods 1 and 2. By modifying the side chain (R) on the organosilane a wide variety of bonded stationary phases may be prepared. II II — Si— OH + Cl— Si— (CH2)„R ------► Si— O — Si— (CH2)„R All the above reactions illustrate a one-to-one relationship between the silanol group and the derivatisation reagent, resulting in one R group for each silanol. This type of phase is referred to as monomeric. By using a trichloroalkylsilane rather than a monochloroalkylsilane as the reagent in method 3, the product resulting from a one- to-one reaction can further react with a nearby silanol or another chloroalkylsilane leading to the build up of several layers. This type of phase is termed polymeric and while the process is more difficult to control than one to give a monomeric coverage, acceptable reproducibility can be obtained. Due to their extra carbon loading, retention is generally higher than on monomeric phases. Because of steric hinderance, only about 25-50% of the silanol groups present on the silica surface react with organosilanes^\ Silanol groups are weakly acidic (pKa typically 5-7) and they are thus able to undergo hydrogen-bond and dipole-dipole interactions with polar compounds. Therefore, the unreacted, or residual silanols remaining after surface modification of the silica cause peak tailing and loss of chromatographic resolution, especially for basic solutes^^'^'^. Thus, subsequent treatment involves reducing the number of accessible silanol groups in a bonded phase; a process known as end-capping. TMS (Trimethylsilane) groups are one of the smallest groups that have been bonded to silica for this chromatographic purposes, and thus the highest bonding densities achieved to date (e.g. 5.43 pmol m'^) have used TMS groups because of their reduced steric constraints^^. In addition to silanols, metal impurities, typically present at about 0.1-0.3% for chromatographic grade silica, are another source of peak tailing and loss of chromatographic resolution by the metals acting as adsorption sites themselves, or by enhancing the activity of adjacent silanols’^''"^. These quantities and types of metal impurities also affect the pH of silica 128 particles. Theoretically, pure silica should have a pKa of 7.1 ±0.5, but pKas varying from 1.5 to 10 have been reported*^. A more certain method of ensuring that silanol based interactions do not spoil a chromatographic separation is to use a column which is not silica based. Polystyrene- divinylbenzene (PS-DVB) is probably the most commonly employed polymeric stationary phase and is built up from cross-linked poly(divinylbenzyl)styrene, prepared by polymerization of mixtures of styrene and divinylbenzene^^’^^. These macroporous polymer packings are totally organic and have a more homogeneous surface devoid of strong hydrogen bonding sites, and can frequently be used at high and low pH (1-13). This allows a much greater choice of mobile phases than would be advisable with silica and, in the case of base permits the ionization to be suppressed by operating at high pH values. Chromatographic behaviour generally resembles that of octadecylsilyl (ODS) silica based stationary phases. However, polymeric columns tend to be less efficient than silica based ones and show increased retentitivity of aromatic solutes, presumably due to n-n interactions between these solutes and the benzene rings of the polymer. 5.2.4 Solvents A suitable solvent for reversed-phase HPLC will preferably have a low viscosity, be compatible with the detection system, have suitable miscibility with water, be commercially available as HPLC-grade solvent, and if possible, have low flammability and toxicity*’^. The solvent must also be able to dissolve the sample without reacting with it chemically. The low viscosity enhances column performance and minimizes the column pressure drop for a given column length and/or particle size. The solvent must be high quality and pure since solvent impurities cause a drift in the detector baseline and diminished sensitivity under isocratic conditions, and large baseline fluctuations and spurious interfering peaks when gradient elution is used. The criteria of high optical transparency at low UV wavelength and solvent properties (Table 5.01) mean that acetonitrile, along with methanol, are the most important organic solvents for use in reversed-phase HPLC. The eluting strength of a solvent for non-polar eluites is inversely related to its polarity. 129 Table 5.01 Solvent properties p ’d Solvent Viscosity^ UV^ RL Boiling (mPa s) (nm) Point (°C) n-Hexane 0.33 190 1.3749 69 0.1 Carbon tetrachloride 0.97 265 1.4652 77 1.6 Toluene 0.59 285 1.4969 111 2.4 Dichloromethane 0.44 230 1.4242 40 3.1 THF 0.46 220 1.4072 66 4.0 Ethyl Acetate 0.45 260 1.3724 77 4.4 Methanol 0.60 205 1.3284 65 5.1 Acetonitrile 0.37 190 1.3441 82 5.8 Ethylene glycol 19.9 210 1.4318 197 6.9 Dimethyl sulphoxide 2.24 270 1.4783 189 7.2 Water 1.00 <190 1.3330 100 10.2 ^Viscosity at 20°C Wavelength above which the solvent can be used i.e. UV cut-off point Refractive index, no^° ** Solvent polarity parameter of Snyder 5.2.5 Instrumental Aspects Columns Columns for analytical HPLC are typically 10-25cm long and 2.1-4.6mm i.d. They are constructed of stainless steel to cope with the high back pressure and are often glass lined to prevent corrosion which may occur if high concentrations of chloride ion or citrates are used'^. Optimum packing size is <5pm and columns with such packing would yield>10000 theoretical plates m '\ Pumps As a consequence of the large back pressures encountered due to the small particle size of packing used in HPLC columns, pumps must be employed to achieve acceptable eluent flow rates. The pump should be able to provide a wide range of flow rates while maintaining adequate levels of accuracy and precision. The flow rate should be practically free of pulsations since fluctuations in flow rate can result in loss of sensitivity due to a ‘noisy’ chromatogram. Pumps may be classified as either those which provide constant inlet pressure or those which provide constant outflow and 130 should be capable of delivering up to TOOOpsi (48.3Mpa). In the vast majority of current analytical HPLC work, it is the latter type of pump that is used. They are constructed from materials which are resistant to the various organic solvents, buffer salt solutions and solutes commonly used as eluents. Reciprocating piston pumps are the most commonly used and force the mobile phase through the chromatographic system by a piston which is driven by a powerful electric motor. Detector The most important of the spectrophotometric detectors is the UV/visible absorption detector which is the most widely used detector in liquid chromatography. The UV spectrophotometer which may be of fixed wavelength or variable wavelength design functions by monitoring the change in absorbance as the solute passes through the detector flow cell i.e. it utilises the specific property of the solute to absorb ultraviolet radiation. Since most organic compounds have some useful absorption in the UV region of the electromagnetic spectrum, these detectors are useful in application, although sensitivity depends on how strongly the sample absorbs light at a particular wavelength which is usually above 200nm. The mobile phase is selected for optical transparency at the detector wavelength such that absorbance should be zero or at least be electronically adjustable to zero in order that a steady baseline be produced. Injector In theory, the sample to be analysed should be introduced onto the head of the column as an extremely narrow band. As a result, sample injection is an important aspect of HPLC. Valve type injectors are the most common injectors used and have excellent precision, compatibility with the pressures encountered in HPLC and good facility for automatic operation. Although several designs exist (notably Valeo and Rhoedyne), all contain common features. A loop which is either external or internal to the valve controls the sample prior to introduction and also the volume injection. A valve allows the loop to be in the ‘load’ position whereby the loop is isolated from the stream of eluent from the pump, but is filled with the sample. When the valve is turned to the ‘inject’ position, the sample is flushed from the loop by the mobile phase coming from the pump and taken onto the column for separation. 131 Chapter 5.3 Experimental Section All the liquid chromatographic measurements carried out in this work were isocratic and made at ambient temperature using acetonitrile-water mobile phases. For the aqueous component of the mobile phase, 50mM potassium phosphate buffer was used, adjusted to pH7 with phosphoric acid. The peak produced by uracil was taken as the void-volume of the system. The mobile phase was filtered under vacuum and then either sonicated or sparged with helium for 15mins to remove excess oxygen. Milli-Q-ultra pure water and HPLC-grade acetonitrile purchased from BDH were used in the mobile phase. All test solutes were obtained commercially and were of HPLC- grade standard. Most samples were dissolved in the mobile phase under study. For a few which were not sufficiently soluble, acetonitrile was first used in order to dissolve the solutes and the required mobile phase composition then made up with water. The mobile phase flow rate was 1.00 ml m in'\ Retention times for each solute were measured at least in duplicate. Measurements were made with two different Kontron HPLC systems; the second being used when the first became unavailable^. The first set of apparatus consisted of a Kontron HPLC Autosampler 465, UV Dual wavelength detector 430, a model 420 pump, a DEC 103 degasser and Software : 450-MT2 V3.90. UV detection was at 210 and 237nm. The second set of apparatus consisted of a Kontron HPLC Autosampler 465, Kontron HPLC detector 535, Kontron Pump System 32X. Data acquisition and processing were performed using Waters ExpertEase Chromatography Remote Acquisition Software Version 3.2. Later Millennium^^ Version 3 software was used. UV detection was at a single wavelength of 237nm. ^ Capacity factors obtained from system 1 were adequately reproduced on system 2 132 Fig. 5.7 General scheme of apparatus setup Mobile Pumping Autosampler/ Phase Degasser system injector Reservoir Column Waste mobile phase reservoir Detector PC based data handler The reversed-phase HPLC systems may be characterized by the linear free energy relationship, eqn 11 (detailed in Chapter 1.4, pages 22-31) log k’ — c + r.R 2 + 5'. 712^ + ût.2ct2^ + 6.2^2^ + v.Vx (11) where k’ is the capacity factor defined as R o) k ’ = (r “ r (12) t tR is the time for a retained compound i.e. a test compound to pass through the column, to is the time taken for a non-retained compound (uracil) to pass through the column. Capacity factors were recorded for the training set of solutes in Table 5,02 and were averages of at least duplicate determinations. The solutes were chosen in such a way as to provide a wide spanning distribution of individual structural descriptor values. 133 Table 5.02 Abraham descriptors for training set solutes Compound Name R2 Zaz" Z k " Vx Nitromethane 0.31 0.95 0.06 0.31 0.4237 Benzene 0.61 0.52 0 0.14 0.7164 Propylbenzene 0.60 0.50 0 0.15 1.1391 Naphthalene 1.34 0.92 0 0.20 1.0854 Anthracene 2.29 1.34 0 0.28 1.4544 Fluorobenzene 0.48 0.57 0 0.10 0.7341 4-Chlorotoluene 0.71 0.74 0 0.05 0.9797 1,4-Dibromobenzene 1.15 0.86 0 0.04 1.0664 lodobenzene 1.19 0.82 0 0.12 0.9746 1 -Bromo-2-Ruorobenzene 0.78 0.78 0 0 0.9091 Anisole 0.71 0.75 0 0.29 0.9160 Benzaldehyde 0.82 1.00 0 0.39 0.8730 4-Huoroacetophenone 0.70 1.02 0 0.47 1.0316 4-Methoxyacetophenone 0.92 1.58 0 0.53 1.2135 Anthraquinone 1.41 1.70 0 0.46 1.5288 Butyl benzoate 0.67 0.80 0 0.46 1.4953 Dimethylphthalate 0.78 1.40 0 0.84 1.4288 Dibutyl phthalate 0.67 1.40 0 0.88 2.2742 3 -R uorobenzoni tri le 0.64 1.09 0 0.35 0.8888 1,4-Dicyanobenzene 0.87 1.63 0 0.53 1.0258 Aniline 0.96 0.96 0.26 0.41 0.8162 Nitrobenzene 0.87 1.11 0 0.28 0.8906 1,2-Dinitrobenzene 1.17 1.70 0 0.38 1.0648 Benzamide 0.99 1.50 0.49 0.67 0.9728 N,N-Dimethylbenzamide 0.95 1.40 0 0.98 1.2546 N,N-Diethylbenzamide 0.95 1.40 0 1.10 1.5360 2-Nitrobenzamide 1.29 2.25 0.40 0.86 1.1470 Acetanilide 0.87 1.36 0.46 0.69 1.1137 4-Ruoroacetanilide 0.74 1.39 0.62 0.56 1.1310 4-Methoxyacetanilide 0.97 1.63 0.48 0.86 1.3133 Phenacetin 0.94 1.51 0.45 0.86 1.4542 134 Phthalimide 1.18 2.09 0.40 0.42 1.0208 Phenol 0.81 0.89 0.60 0.30 0.7751 p-Cresol 0.82 0.87 0.57 0.31 0.9160 4-Chlorophenol 0.92 1.08 0.67 0.20 0.8975 3,5 -Dichlorophenol 1.02 1.00 0.91 0 1.0199 4-Methoxyphenol 0.90 1.17 0.57 0.48 0.9747 3-Hydroxyacetophenone 0.98 1.35 0.72 0.55 1.0730 4-CN-phenol 0.94 1.63 0.80 0.29 0.9298 Catechol 0.97 1.10 0.88 0.47 0.8338 Hydroquinone 1.06 1.27 1.06 0.57 0.8338 4-Hydroxybenzophenone 1.64 1.88 0.79 0.57 1.5400 3-Acetamidophenol 1.05 1.70 1.09 0.78 1.1724 4-Phenylbutan-1 -ol 0.81 0.90 0.33 0.70 1.3387 Diphenyldisulfide 1.92 1.10 0 0.28 1.6512 Benzenesulfonamide 1.13 1.55 0.55 0.80 1.0970 Testosterone 1.54 2.59 0.32 1.19 2.3830 Hydrocortisone 2.03 3.49 0.71 1.90 2.7980 Phenylurea 1.11 1.40 0.77 0.77 1.0730 Phénobarbital 1.63 1.80 0.73 1.15 1.6999 To develop and achieve the separation, it is necessary to manipulate the experimental variables that have the greatest influence on the equilibrium distribution - the composition of the mobile phase and the nature of the stationary phase. For maximum flexibility in developing a separation, the mobile and stationary phases are usually chosen to have contrasting polarities. Although the most commonly used phases are alkyl-bonded silicas, there also exist functional alkyl-bonded phases whose main advantage is a selectivity which is different from those of alkyl-bonded silicas. The synthesis and use of diverse bonded phases containing moieties such as phenyl groups, cyano groups and alkylamide groups have been developed to improve separations of solutes with similar structures. In this study, the properties of different stationary phases (Table 5.03) were compared by application of the Abraham Solvation Equation to capacity factors derived from the series of solutes in Table 5.02 135 Table 5.03 HPLC columns used Column Stationary phase Manufacturer Dim/mm YMC C18 -Polymethacrylate CIS Highcrom 250 X 4.6 PLRPS (polymer) -Polystyrene/divinylbenzene Polymer Labs 50 X 4.6 Luna 5u Phenylhexyl -CeHsCCôHn) Phenomenex 150x4.6 Luna 5u CIS -CigHsy Phenomenex 150x4.6 Amide -(CH2)3C0NHCH3 Shandon HPLC 150x4.6 Lichrospher Diol-5 -(CH2)30CH2CH(0 H)CH2(0 H) Hichrom 125 X 4.6 FluoroSep -RP Octyl (FO) -(CF2)?CF3 ES Industries 150x4.6 Hypersil CN -(CH2)3CN Shandon HPLC 100 X 4.6 Both the diol and amide columns are of interest for samples which can form hydrogen-bonds with the polar groups of the packing. The cyano column has a lower polarity than the diol and amide phases but still maintains retention for polar solutes and is particularly selective towards components with double bonds. These three columns are thus principally used to offer alternative selectivity for highly polar compounds which have little retention on the conventional highly non-polar C l8 phase. The perfluorooctyl column should be the least polar phase and have similar selectivity to a standard C8 phase, but enhanced selectivity for halogenated compounds. A disadvantage of these bonded phases is that the stability is generally considered to be less than that of n-alkyl phases, because the Si-O-Si-C bonds are less effectively shielded against nucleophilic attacks. Conversely, the C l8, phenylhexyl and polymer phases are particularly useful for the analysis of aromatic compounds; the selectivity being derived from interaction with the TC-electrons found in the bonded phase. These Ti-electrons can enhance the selectivity of aromatic compounds through an induced polarization of electrons. The polymer column contains a very high surface area of inherently hydrophobic divinylbenzene particles which should give unique separation properties for non-polar compounds. The average capacity factors obtained for each of the solutes in the studied stationary phase and acetonitrile/water mobile phase compositions are given in Table 5.04. The resulting LFER equations from analysis of this data with the Abraham General Solvation Equation are given in Table 5.05 136 Table 5.04 log k’ values for compounds obtained from each stationary phase and mobile phase composition (MeCN/HzO) in vol % terms Compound name PolyClS Polymer Phhexyl C18 Amide Diol Perfluoro Cyano 70/30 60/40 65/35 60/40 60/40 65/35 60/40 50/50 30/70 60/40 60/40 30/70 Nitromethane -0.438 -0.209 -0.398 -0.310 -0.243 -0.237 -0.551 -0.593 -0.634 -0.373 -0.651 -0.561 Benzene 0.155 0.394 0.508 0.621 0.272 0.376 0.000 -0.198 0.030 0.005 -0.372 0.147 Propylbenzene 0.449 0.767 1.000 1.171 0.640 0.897 0.425 -0.010 0.431 0.313 -0.200 0.627 Naphthalene 0.451 0.747 1.042 1.183 0.520 0.646 0.277 -0.005 0.461 0.097 -0.232 0.399 Anthracene 0.778 1.091 1.601 1.763 0.782 0.954 0.622 0.166 0.851 0.169 -0.115 0.903 Fluorobenzene 0.146 0.386 0.451 0.571 0.267 0.383 0.004 -0.184 0.078 0.046 -0.350 0.203 4-Chlorotoluene 0.420 0.695 0.914 1.058 0.533 0.701 0.306 -0.046 0.375 0.190 -0.250 0.504 1,4-Dibromobenzene 0.625 0.896 1.239 1.396 0.659 0.818 0.438 0.041 0.530 0.154 -0.198 0.622 lodobenzene 0.470 0.730 1.027 1.164 0.535 0.688 0.295 -0.020 0.409 0.082 -0.250 0.477 l-Bromo-2-Fluorobenzene 0.338 0.616 0.761 0.887 0.438 0.572 0.197 -0.068 0.334 -0.095 -0.309 0.223 Anisole 0.111 0.344 0.489 0.600 0.253 0.331 -0.024 -0.218 0.030 -0.029 -0.381 0.182 Benzaldehyde -0.051 0.181 0.207 0.301 0.058 0.110 -0.228 -0.342 -0.161 -0.161 -0.469 -0.042 4-Fl uoroacetophenone -0.057 0.171 0.150 0.260 0.091 0.161 -0.174 -0.321 -0.112 -0.083 -0.447 0.058 4-Methoxyacetophenone -0.121 0.111 0.121 0.216 0.040 0.125 -0.219 -0.342 -0.106 -0.396 -0.481 -0.048 Anthraquinone 0.479 0.758 0.920 1.056 0.477 0.568 0.262 -0.068 0.371 -0.005 -0.264 0.342 Butyl, benzoate 0.375 0.664 0.886 1.038 0.579 0.789 0.383 -0.046 0.404 0.259 -0.240 0.647 Dimethylphthalate -0.195 0.080 0.075 0.198 0.100 0.124 -0.207 -0.331 -0.100 -0.151 -0.469 -0.018 Dibutyl phthalate 0.410 0.783 0.952 1.137 0.783 1.024 0.566 0.046 0.609 0.406 -0.146 0.962 3-Fluorobenzonitrile -0.006 0.238 0.204 0.314 0.150 0.198 -0.121 -0.260 -0.058 -0.032 -0.406 0.111 1,4-Dicyanobenzene -0.120 0.130 0.017 0.134 0.060 0.036 -0.244 -0.336 -0.209 -0.186 -0.447 0.011 Aniline -0.086 0.123 -0.060 0.017 -0.086 -0.044 -0.353 -0.375 -0.254 -0.306 -0.531 -0.210 Nitrobenzene 0.108 0.344 0.370 0.494 0.201 0.259 -0.066 -0.206 0.034 -0.047 -0.377 0.151 1,2-Dinitrobenzene 0.053 0.313 0.173 0.310 0.200 0.155 -0.116 -0.169 0.147 -0.077 -0.350 0.317 Benzamide -0.478 -0.320 -0.802 -0.649 -0.522 -0.506 -0.708 -0.611 -0.456 -0.613 -0.707 -0.532 137 Compound name PolyClB Polymer Phhexyl C18 Amide Diol Perfluoro Cyano 70/30 60/40 65/35 60/40 60/40 65/35 60/40 50/50 30/70 60/40 60/40 30/70 N,N-Dimethylbenzamide -0.492 -0.289 -0.357 -0.288 -0.288 -0.234 -0.536 -0.593 -0.456 -0.292 -0.669 -0.287 N,N-Diethylbenzamide -0.267 -0.044 -0.125 -0.001 -0.016 0.053 -0.274 -0.423 -0.216 -0.112 -0.518 -0.097 2-Nitrobenzamide -0.520 -0.313 -0.802 -0.649 -0.449 -0.471 -0.712 -0.593 -0.510 -0.654 -0.727 -0.532 Acetanilide -0.341 -0.160 -0.420 -0.357 -0.288 -0.264 -0.428 -0.463 -0.297 -0.443 -0.622 -0.206 4-FluoroacetaniIide -0.329 -0.131 -0.486 -0.337 -0.261 -0.243 -0.400 -0.436 -0.231 -0.377 -0.602 -0.140 4-Methoxyacetanilide -0.444 -0.234 -0.551 -0.412 -0.323 -0.318 -0.513 -0.515 -0.284 -0.489 -0.618 -0.296 Phenacetin -0.354 -0.125 -0.426 -0.289 -0.198 -0.170 -0.380 -0.436 -0.216 -0.375 -0.558 -0.238 Phthalimide -0.221 -0.025 -0.426 -0.316 -0.253 -0.243 -0.457 -0.449 -0.288 -0.487 -0.687 -0.370 Phenol -0.135 0.068 -0.264 -0.155 -0.137 -0.084 -0.255 -0.342 -0.167 -0.386 -0.572 -0.182 p-Cresol -0.063 0.176 -0.142 -0.030 -0.016 0.022 -0.142 -0.282 -0.047 -0.294 -0.458 -0.143 4-Chlorophenol 0.081 0.305 -0.051 0.064 0.031 0.066 -0.011 -0.191 0.127 -0.271 -0.469 0.132 ,5-Dichlorophenol 0.368 0.638 0.211 0.359 0.265 0.345 0.397 -0.046 0.392 -0.105 -0.368 0.268 4-Methoxyphenol -0.232 -0.032 -0.342 -0.237 -0.205 -0.172 -0.360 -0.398 -0.231 -0.477 -0.587 -0.216 3-Hydroxyacetophenone -0.291 -0.068 -0.462 -0.346 -0.252 -0.233 -0.390 -0.436 -0.247 -0.506 -0.618 -0.319 4-CN-phenol -0.212 0.018 -0.462 -0.316 -0.206 -0.214 -0.297 -0.386 -0.181 -0.477 -0.634 -0.290 Catechol -0.288 -0.118 -0.620 -0.523 -0.368 -0.363 -0.482 -0.449 -0.334 -0.646 -0.669 -0.376 Hydroquinone -0.492 -0.320 -0.925 -0.854 -0.592 -0.609 -0.671 -0.593 -0.496 -0.947 -0.873 -0.632 4-Hydroxybenzophenone 0.062 0.322 -0.028 0.110 0.040 0.050 0.012 -0.184 0.201 -0.340 -0.481 0.144 3-Acetamidophenol -0.601 -0.410 -0.901 -0.713 -0.566 -0.584 -0.712 -0.611 -0.443 -0.854 -0.748 -0.564 4-Pheny Ibutan-1 -ol -0.099 0.121 -0.009 0.093 0.022 0.118 -0.156 -0.291 0.026 -0.228 -0.481 0.141 Diphenyldisulfide 0.797 1.137 1.608 1.783 0.932 1.108 0.711 0.189 0.921 0.230 -0.076 1.083 B enzenesulfonamide -0.368 -0.139 -0.600 -0.451 -0.285 -0.318 -0.566 -0.463 -0.344 -0.588 -0.634 -0.428 Testosterone 0.076 0.295 0.086 0.198 0.083 0.219 0.040 -0.282 0.134 -0.162 -0.436 0.249 Hydrocortisone -0.393 -0.209 -0.600 -0.493 -0.386 -0.330 -0.469 -0.593 -0.279 -0.739 -0.531 -0.203 Phenylurea -0.444 -0.266 -0.802 -0.713 -0.522 -0.506 -0.671 -0.556 -0.364 -0.689 -0.748 -0.477 Phénobarbital -0.286 -0.062 -0.600 -0.451 -0.255 -0.284 -0.482 -0.449 -0.239 -0.689 -0.618 -0.343 138 Table 5.05 LFERs obtained for each of the HPLC systems Column Mobile Phase r s a b V c r" SD n MeCN/HzO PolyClS 70/30 0.246 -0.191 -0.232 -1.023 0.75S -0.322 0.9S1 0.053 50 60/40 0.247 -0.210 -0.267 -1.124 0.S61 -0.119 0.9S2 0.057 50 Polymer 65/35 0.445 -0.437 -0.S32 -1.390 1.173 -0.211 0.9S6 0.0S4 50 60/40 0.445 -0.437 -0.S19 -1.449 1.229 -0.125 0.9S4 0.091 50 Phenylhexyl 60/40 0.077 -0.1 S2 -0.460 -1.009 0.S95 -O.lSl 0.9S5 0.04S 50 CIS 65/35 O.OSS -0.2S9 -0.502 -1.114 l.OSl -0.142 0.9SS 0.050 50 Amide 60/40 0.097 -0.225 -0.177 -1.099 0.99S -0.501 0.979 0.05S 50 Diol ^60/40 0.043 -0.076 -0.097 -0.42S 0.274 -0.447 0.933 0.042 50 50/50 0.109 -0.129 -0.107 -0.622 0.4SS -0.456 0.970 0.040 50 30/70 0.244 -0.240 -0.124 -1.003 0.913 -0.471 0.9S1 0.053 50 Perfluorooctyl 60/40 -0.0S7 -0.170 -0.439 -0.612 0.62S -0.2 IS 0.94S 0.077 50 Cyano 60/40 0.059 -0.09S -0.202 -0.426 0.411 -0.603 0.953 0.042 50 30/70 0.129 -0.269 -0.321 -0.9S4 1.056 -0.3S5 0.956 0.091 50 ^Calculated coefficients obtained from eqn 13 139 Chapter 5.4 Comparison of Stationary Phases The LFER coefficients for the different stationary phases in a fixed mobile phase may be compared (Table 5.06, Fig. 5.08 and Table 5.07, Fig. 5.09). Since the mobile phase is fixed at 60%MeCN, 40%H2O by volume, the variables corresponding to the mobile-phase properties are constant and hence the differences in the LFER coefficients reflect different properties of the stationary phases as they exist in equilibrium with the mobile phase^^. In the case of the diol stationary phase, the derived equation for the required mobile phase composition of 60%MeCN and 40%H20 was not obtained on a completely experimental basis, but was obtained as follows : Considering the well-known relationship between log k’ and organic mobile phase composition^^, ({) (see eqn 13), once experimental capacity factors for a series of solutes has been obtained in two different mobile phase compositions, it is possible to calculate further capacity factors for the required mobile phase composition. log k’ = log k’w - S({) (13) So, in the case of the diol column : log k] = log k’w -0 .5 8 (14) log k 2 = logk’w - 0.3S (15) Hence logk 2 _ ^ (16) — 0.5 + 0.3 log k’w = log kl 4-0.58 (17) Thus for the required mobile phase composition of 60%MeCN, 40%H2O, log kcaic = log k’w - 0.68 (18) Having calculated these new log k’ values, multiple linear regression analysis with the Abraham descriptors is carried out as normal to obtain an equation for a mobile phase containing 60%MeCN. 140 Table 5.06 Coefficients of 60/40 MeCN/H20 mobile phase composition for the different stationary phases Column r s a b V c PolyClS 0.247 -0.210 -0.267 -1.124 0.861 -0.119 Polymer 0.445 -0.437 -0.819 -1.449 1.229 -0.125 Phenylhexyl 0.077 -0.182 -0.460 -1.009 0.895 -0.181 Amide 0.097 -0.225 -0.177 -1.099 0.998 -0.501 Diol 0.043 -0.076 -0.097 -0.428 0.274 -0.447 Perfluorooctyl -0.087 -0.170 -0.439 -0.612 0.628 -0.218 Cyano 0.059 -0.098 -0.202 -0.426 0.411 -0.603 Fig. 5.08 Comparison of r, s, a, b and v for 7 stationary phases □ r □ -S c 0.6 □ -a □ - b Stationary phase 141 Table 5.07 Ratios of LFER coefficients for the seven stationary phases Column r/v s/v aiv hiv PolyClS 0.287 -0.244 -0.310 -1.305 Polymer 0.362 -0.356 -0.666 -1.179 Phenylhexyl 0.086 -0.203 -0.514 -1.127 Amide 0.097 -0.225 -0.177 -1.101 Diol 0.157 -0.277 -0.354 -1.562 Perfluorooctyl -0.139 -0.271 -0.699 -0.974 Cyano 0.144 -0.238 -0.491 -1.036 Fig 5.09 Comparison of LFER coefficient ratios Ratio 0.2 PolyClS Polymer Phenylhexyl Amide Perfluoro Cyano Diol 0.0 - 0.2 -0.4 □ -r/v > -0.6 □ s/v □ a/v £ -0.8 Ib/v o - 1.0 - 1.2 -1.4 - 1.6 142 Goodness of fit of all the equations can be examined from several perspectives. The r^ values are good; all greater than 0.90. Also, the SD values are less than or equal to 0.10. The plots of experimental log k’ against calculated log k’ showed no serious outliers. In the following interpretation of LFER coefficients, it will be necessary to refer to several parameters in Table 5.08. Table 5.08 Properties of selected solvents nD^^^ Til*" Pi" CXi" Water 549.0 1.333 1.09 0.47 1.17 Acetonitrile^ 147.0 1.341 0.75 0.37 0.19 Methanol 205.2 1.327 0.61 0.66 0.93 Benzene 83.8 1.501 0.59 0.10 0.00 Decane 59.7 1.411 0.10 0.00 0.00 Perfluorooctane <33.7 <1.30 -0.41 -0.08 0.00 Ethylene glycol 274 1.432 0.92 0.52 0.90 All data taken from ref. 19 unless otherwise stated "Hildebrand solubility parameter values (kcal dm'^) taken from ref 20. Value for perfluorooctane taken from ref 19 ^Refractive index ‘^Kamlet-Taft solvent dipolarity parameter ‘^Kamlet-Taft solvent hydrogen-bond base parameter ^ Kamlet -Taft solvent hydrogen-bond acid parameter ^Data for acetonitrile taken from ref 21 The r-coefficient The r-coefficient is a correction factor to the dipolarity/polarizability term {s- coefficient) and reflects the tendency of the system to interact with the solute through 71- and n-electron pairs. In contrast to the negative ^-coefficients, the r-coefficients are either nearly zero or positive. A positive r-coefficient implies that the solute undergoes stronger n-n interactions with the stationary phase than the mobile phase. The higher the r-value, the more polarizable the phase. Notably there is a higher r- value for the polymer column than any other phase. Results in Table 5.05 indicate that the r-coefficient tends to zero as (|)MeCN is increased, implying that the stationary and mobile phases are becoming similar in terms of polarizability. The negative value of r for perfluorooctyl implies that the solute undergoes stronger 7i-7i interactions with the mobile phase than the stationary phase. This is expected since fluorinated compounds are less polarizable than their hydrocarbon analogues due to the inductive effect of 143 fluorine atoms. This is indicated by the low refractive indexes of the corresponding liquid fluorocarbons. Furthermore, the fluorocarbon solvent that is most closely related to the fluorinated bonded phase is very non-polar as reflected in the Hildebrand solubility parameter which is again smaller than those of the analogous hydrocarbon liquids. In addition fluorocarbon solvents have very low Kamlet-Taft dipolarity/polarizability parameters (rci*), compared to much higher values for the corresponding hydrocarbons. The s-coefpcient The 5-coefficient is negative for all phases and is related to the difference in dipolarity/polarizability of the stationary and mobile phases. The closer this coefficient is to zero, the closer the dipolar interactions between the solute and the stationary phase are to the dipolar interactions in the mobile phase. The negative sign indicates that the dipolarity/polarizability of the mobile phase is higher than that of the stationary phases. This is attributed to the highly dipolar nature of both components of the mobile phase; Tiwater* = 1.17 and TCacetomtriie* = 0.75. In the stationary phase, the dipolar interactions are attributed mainly to the sorbed mobile-phase components. The small magnitude of coefficients indicates that dipolar solutes only slightly favour the mobile to the stationary phase. The polymer column has the largest 5-coefficient. This is probably caused by the same process that leads to the larger r-coefficient; the surface of the polymer column is not as well solvated as those of the other phases. The a-coefficient The small and negative coefficients of a for most of the stationary phases indicate that based solely on its hydrogen-bond acidity, a solute would only have slight preference for the mobile phase to the stationary phase. The solvent hydrogen-bond basicity (Pi) is the complimentary property to the solute hydrogen-bond acidity. The acetonitrile/water (60/40) mobile phase is only moderately basic since pure water (Pi = 0.47) and pure acetonitrile (Pi = 0.37) are both only modestly basic. The a coefficient seems to be small and independent of the nature of the stationary phase; the exceptions being the polymer and perfluorooctyl columns in which the coefficient is significantly more negative when the phases are compared in terms of the relative LFER coefficient. Since alkyl chains and phenyl groups have low basicities, the 144 basicity of the polymer phase must arise mainly from the sorbed mobile phase. The large negative a indicates that the basicity of the polymer column is small due to the minimal amount of sorbed mobile phase on the polymer surface. In the case of the perfluorooctyl column, it is likely that the hydrogen-bond basicity arises from both sorbed mobile phase and from the residual hydroxyl groups on the surface of the silica. In both cases, the mobile phase is much more basic than the stationary phase. The b-coefficient The ^-coefficient represents the difference in hydrogen-bond acidity of the mobile and stationary phases and is larger than a for all phases. Thus solutes with stronger hydrogen-bond acceptor ability will be significantly less retained than non-basic compounds. This also indicates that the mobile phase is a much stronger hydrogen- bond acid than the stationary phase. This is expected since the aqueous mobile phase is highly acidic due to water being a very strong hydrogen-bond acid (oci = 1.17). Amongst the phases, the polymer column has the most negative ^-coefficient. This is again evidence to suggest that the polymer phase is a poor adsorbent and repels water. The perfluorooctyl, cyano and diol columns have the least negative 6-coefficients indicating that these phases adsorb far more water onto their surfaces, thus increasing the surface acidity. In the case of phases for which more than one volume fraction of acetonitrile has been used (Table 5.05), the 6-coefficient which is negative becomes more positive as the volume fraction of acetonitrile in the eluent is increased. This results because the highly acidic water is replaced with the much less acidic acetonitrile («i = 0.19). The acidity of the stationary phase decreases as the volume fraction of acetonitrile in the mobile phase is increased because less water is sorbed to the stationary phase. The v-coefficient At a fixed mobile phase composition, variations in the v-coefficient is due to the differences in the cohesiveness and dispersion interactions of the aromatic and aliphatic phases. The v-coefficient becomes increasingly more positive as (])MeCN is decreased. This observation is consistent with the higher cohesivity of water (0H^=549cal ml'*) as compared to acetonitrile (0 h^= 147cal ml'^) which affects the dependence of retention on solute volume. As the fraction of water is increased, the 145 cohesive density of the mobile phase increases substantially. Hydrocarbon phases are significantly more dispersive than highly fluorinated phases (refractive index for alkanes higher). Hence the lower coefficient of the perfluorinated phase can be explained by weaker dispersive interactions of the solute with the fluorocarbon phase. However, the diol and cyano phases are highly dispersive and hence the low v- coefficients are more likely due to the high sorption of water on to the mobile phase. This results in these solvated stationary phases being much more similar to the mobile phase in terms of cohesivity. Thus at the fixed mobile phase composition, this increase in the stationary phase cohesive energy decreases both solute retention and also the v term. In terms of selectivity, the stationary phases are compared in terms of the sh, a/v, 6/v, and r/v ratios instead of in terms of their absolute v, 5, a, b and r values. The r/v and siv ratios are similar for most of the phases, but distinguish the polymer and perfluorooctyl phases. The fact that the r/v ratio is so different for these two phases strongly indicates that they have highly contrasting polarizabilities. The polymer column is most polarizable, whereas the perfluorooctyl is the least. Again, the a/v ratio distinguishes the polymer and perfluorooctyl columns in that they are larger than for the other stationary phases. In terms of the blv ratio, the diol column is shown to be distinct which is expected since this phase should be the most acidic due to the hydroxyl groups of the bonded phase and also the large amount of sorbed water. 146 5.5 Use of Water-solvent Partition Measurements to Obtain Abraham Descriptors and Comparison with HPLC Systems 5.5.1 Water-solvent partition measurements in descriptor determination Although the group contribution method of Platts^^ is now available to estimate Abraham descriptors, the problem of missing fragments is an important issue which needs to be addressed. Traditionally, water-solvent partition measurements have been used extensively in the past to obtain compound descriptors, see Chapter 3. Common water-solvent systems employed are shown in Table 5.09. The table includes the critical quartet of systems as suggested by Leahy^^ which are considered to be four highly contrasted solvent-water partitioning systems; octanol is amphiprotic, chloroform is a proton donor, hexane is inert, whilst propylene glycol diperlargonate (PGDP) is a proton acceptor. Table 5.09 Equation coefficients for water-solvent systems r s a b V c Octanol ^ 0.562 -1.054 0.034 -3.460 3.814 0.088 (0.014) (0.021) (0.021) (0.026) (0.015) (0.015) CHC13 ^ 0.157 -0.391 -3.191 -3.437 4.191 0.327 (0.046) (0.042) (0.043) (0.065) (0.048) (0.037) Hexane ^ 0.579 -1.723 -3.599 -4.764 4.344 0.361 (0.033) (0.038) (0.034) (0.045) (0.037) (0.028) PGDP"* 0.501 -0.828 -1.022 -4.640 4.033 0.256 (0.103) (0.159) (0.133) (0.161) (0.153) (0.129) Diethyl ether ^ 0.605 -1.096 -0.097 -5.000 4.381 0.25 (0.115) (0.098) (0.090) (0.122) (0.093) (0.071) Ethyl acetate ^ 1.157 -1.397 -0.054 -3.755 3.726 0.253 (0.110) (0.125) (0.102) (0.117) (0.100) (0.082) Benzene ^ 0.464 -0.588 -3.099 -4.625 4.491 0.142 (0.047) (0.055) (0.048) (0.074) (0.063) (0.036) Oleyl alcohol ® -0.270 -0.528 -0.035 -4.042 4.204 -0.359 (0.091) (0.087) (0.085) (0.140) (0.065) (0.089) The coefficient SD values are given in the parentheses "Ref24 '’Ref25 "Ref26 ‘‘Ref27 "Ref28 147 Principal component analysis (PCA) using Minitab for Windows 9.2, 1993 may be carried out on the water-solvent partition equations shown in Table 5.09. PCA is a data reduction technique used to identify a small set of variables that account for a large proportion of the total variance in the original variables. Analysis of the coefficients in Table 5.09 using this method indicates that three independent parameters may be derived from the original variables, although the third parameter is poorly represented because the eigenvalue is less than 1. However, the general guideline in PCA applications is to select those principle components (PCs) which account, cumulatively for at least 80% of the data variation, and so PCS is still significant because it explains 17.9% of the data variation, and brings the cumulative total above the 80% threshold. The PCA loadings indicate that R 2, Ttz", and contribute most to PCI, whilst Sa 2^ and Vx have more in common with PC2. Eigenanalysis of the Correlation Matrix for water-solvent partition equations Variable PCI* PC2* PCS* PC4* PC5* R2 0.547 -0.209 0.384 0.708 -0.095 -0.544 0.327 -0.227 0.679 0.291 -0.104 -0.615 -0.630 0.184 -0.424 ZP2" -0.564 -0.070 0.568 0.027 -0.594 Vx 0.277 0.683 -0.283 0.060 -0.611 Eigenvalue^ 2.119 1.576 0.896 0.330 0.080 Proportion‘d 0.424 0.315 0.179 0.066 0.016 Cumulative*^ 0.424 0.739 0.918 0.984 1.000 “ Linear transformation of original variables such that the derived variables are uncorrelated; all mutually orthogonal ^ The variance of the original data set described by that principal component The proportion of data variation accounted for by each principal component ** The cumulative proportion of data variation accounted for by the principal components. 148 5.5.2 Characterization of HPLC Systems Although water-solvent partitions have been used successfully to obtain Abraham descriptors, the method suffers from several impracticalities; the long timescale involved, the difficulty in experimental procedure and reproducibility in experimental results. To overcome these incessant difficulties, it was suggested that reversed phase HPLC (RP-HPLC) could be utilised to supersede the partition measurements. RP- HPLC is a widely used separation technique and has several key advantages over partition measurements. It is fast, highly reproducible, applicable to volatile substances, suitable for substances containing impurities and furthermore, pH can be controlled precisely. Chapter 5.4 dealt with stationary phases in which the mobile phase is fixed. Whilst this is useful in the context of comparing retention mechanisms and also the chemical properties of the bonded phase, it is experimentally not feasible to use the same mobile phase-composition for back-calculation of descriptors. The aim in this respect is to choose a mobile phase in which the retention time is most reasonable and selective for a wide range of compounds. With this in mind, the following systems were considered to be most appropriate. Table 5.10 Coefficients for reversed-phase HPLC systems Column^ Mobile phase r s a b V c MeCN/HzO Poly Cl 8 60/40 0.247 -0.210 -0.267 -1.124 0.861 -0.119 (0.031) (0.030) (0.027) (0.046) (0.035) (0.029) Polymer 60/40 0.445 -0.437 -0.819 -1.449 1.229 -0.125 (0.049) (0.048) (0.043) (0.073) (0.055) (0.046) Phenylhexyl 60/40 0.077 -0.182 -0.460 -1.009 0.895 -0.181 (0.033) (0.034) (0.031) (0.050) (0.035) (0.036) CIS 65/35 0.088 -0.289 -0.502 -1.114 1.081 -0.142 (0.034) (0.035) (0.033) (0.052) (0.036) (0.037) Amide 60/40 0.097 -0.225 -0.177 -1.099 0.998 -0.501 (0.032) (0.031) (0.028) (0.047) (0.036) (0.029) Diol 30/70 0.244 -0.240 -0.124 -1.003 0.913 -0.471 (0.029) (0.028) (0.025) (0.043) (0.032) (0.027) Perfluorooctyl 60/40 -0.087 -0.170 -0.439 -0.612 0.628 -0.218 (0.042) (0.041) (0.036) (0.062) (0.047) (0.039) Cyano 30/70 0.129 -0.269 -0.321 -0.984 1.056 -0.385 (0.049) (0.047) (0.042) (0.072) (0.055) (0.045) The SD values of the coefficients are given in the parentheses This work 149 Since R2 may be obtained from the refractive index of a compound or from addition of relevant fragments, and Vx is trivial to calculate from atomic contributions, only three descriptors remain to be found. This implies that a minimum of three HPLC systems are required to solve simultaneous equations with three unknowns. However, the consequence of this would be that exact values of 712”, Zo^" and ZP 2” would be calculated, which leaves no scope for experimental error. This is highly unrealistic, and as such at least a fourth system should be incorporated to highlight erroneous measurements. Theoretically, distinct HPLC systems are required for back-calculating to obtain descriptors; different methods of analysis will be employed to illustrate how it is possible to isolate the most distinct equations and hence HPLC systems. A first step in deciding which equations are the most different is to plot the respective coefficients graphically. Fig 5.10 Plot of coefficients for each HPLC column system ■polyC18 ■polymer V ■ phenylhexyl ■C18 ■amide ■did •perfluorooctyl ■cyano The radar plot suggests that in terms of individual absolute coefficients, the polymer column and perfluorooctyl column are the most different, and hence should be most useful in back-calculating to obtain descriptors. However, a further and more quantitative method is required in order to ascertain which of the other systems should be used. 150 5.5.2a Vector approach A quantitative method for distinguishing between two LFERs has recently been proposed by Ishihama and Asakawa^^. In this approach, each of the HPLC equations may be represented as a vector ; the magnitude of the angle between two such vectors is a measure of their similarity. The scalar product of the vectors a and b is defined as : a • b = a b cos0 If a and b are given in Cartesian form, a = aii + a 2j + a^k + ...... b = bji + b2j + b^k +. then a • b = aibi + azbz + agbg +..... If the angle between the vectors a and b is 0 then ci^bf - \ - +. cosa=— — ^ ^ ^ \a Where | a | | b | is the magnitude or modulus of a and b respectively. Example : a = Polymer 60/40 = 0.445R: -0.437712“ -O.SlOZaz" - 1.4492p2“ + 1.229 Vx b = Amide 60/40 = 0.097R; -O. 2 2 5 7 1 2 " -0.177Za2“ - I.O9 9 EP 2 " + 0.998 Vx a • b = (0.445.0.097)+(-0.437.-0.225)+(-0.8I9«-0.177)+(-1.449»-1.099) +(1.229*0.998) = 3.10545 |a| = 7(0.445)" + (-0.437)" + (-0.819)" + (-1.449)^ + (1.229)^ = 2.16097 |6| = 7(0.097)" + (-0.225)" + (-0.177)" + (-1.099)" + (0.998)" = 1.51498 151 3.10545 COS# = = 0.9484 .••# = 18.46 (2.16097 jcl.51498) Thus this method enables the pairwise comparison of equations. The higher the linear correlation between two equations, the closer cosO is to 1, or the nearer 0 is to zero. Table 5.11 Matrices of cos 0 and 0 for each of the possible column pairs. cos 0 matrix No. Column 1 2 3 4 5 6 7 8 1 Poly C l8 1.000 0.974 0.981 0.979 0.989 0.992 0.922 0.983 2 Polymer 0.974 1.000 0.982 0.982 0.949 0.952 0.951 0.970 3 Phenylhexyl 0.981 0.982 1.000 0.998 0.978 0.965 0.978 0.988 4 C18 0.979 0.982 0.998 1.000 0.981 0.969 0.979 0.994 5 Amide 0.989 0.949 0.978 0.981 1.000 0.993 0.929 0.991 6 D iol 0.992 0.952 0.965 0.969 0.993 1.000 0.899 0.985 7 Perfluorooctyl 0.922 0.951 0.978 0.979 0.929 0.899 1.000 0.956 8 Cyano 0.983 0.970 0.988 0.994 0.991 0.985 0.956 1.000 0 m atrix No. Column 1 2 3 4 5 6 7 8 1 Poly C 18 0.00 13.19 11.13 11.71 8.47 7.28 22.82 10.66 2 Polymer 13.19 0.00 10.93 11.02 18.46 17.74 17.97 14.10 3 Phenylhexyl 11.13 10.93 0.00 3.77 12.00 15.31 12.04 8 ^ 2 4 C18 11.71 11.02 3.77 0.00 11.22 14.34 11.82 6.24 5 Amide 8.47 18.46 12.00 11.22 0.00 6.67 21.65 7.61 6 D iol 7 J # 17.74 15.31 14.34 6.67 0.00 25.98 9 j # 7 Perfluorooctyl 22.82 17.97 12.04 11.82 21.65 25.98 0.00 17.05 8 Cyano 10.66 14.10 8 ^ 2 6.24 7.61 9 j # 17.05 0.00 Total 0 85.26 103.40 73.98 70.12 86.07 97.20 129.32 74.36 Vector analysis identifies the perfluorooctyl, polymer, and diol columns as having the least similarity with the other columns. 152 5.5.2b Princip Component Analysis (PCA) of HPLC Systems As previously carried out for the water-solvent partition systems, PCA can be used to decompose the multivariate HPLC equations into a series of orthogonal eigenvectors. The resulting correlation matrix differs substantially in that the loadings of the variables onto PCI are all very similar. This is contrary to the desired result which would be to have each PC being significantly influenced by as few original variables as possible. However, the scores plot. Fig. 5.11 is in agreement with previous methods and identifies the polymer, diol and perfluorooctyl columns as most diverse; these columns reside on the periphery of the plot. Eigenanalysis of the Correlation Matrix for HPLC equations Variable PCI PC2 PC3 PC4 PC5 Rz 0.448 -0.359 0.681 0.157 0.427 Ttz" -0.492 -0.159 0.077 -0.684 0.509 Zctz" -0.320 -0.870 -0.180 0.259 -0.205 -0.484 0.259 -0.016 0.664 0.507 Vx 0.470 -0.148 -0.706 -0.004 0.509 Eigenvalue 0.376 0.789 0.295 0.136 0.018 Proportion 0.752 0.158 0.059 0.027 0.004 Cumulative 0.752 0.910 0.969 0.996 1.000 Fig. 5.11 Scores plot for HPLC systems described by Abraham Descriptors 1.5 □ perfluorooctyl 1.0 - polymer □ 0.5 □ OC18 phenylhexyl □ cyano -0.5 - - 1.0 - □ diol - 1.5 -3 -2 0 1 PCI 153 5.5.3 Comparison of HPLC and water-solvent partition systems Since the aim of this work is to investigate the prospect of replacing water-solvent partition measurements with HPLC measurements to obtain Abraham descriptors, it is useful to compare both sets of values in terms of the equation coefficients. The methods of comparing HPLC systems in terms of coefficient ratios, magnitude of coefficients, vector analysis and PCA concluded that the perfluorooctyl column, polymer and diol column are the most distinct. However, whether the HPLC systems are able to adequately compensate for the diverse water-solvent partition systems is an important point which still remains to be answered. To assess the interrelationship between the variables as a set and to obtain a true estimate of dimensionality, PCA on the combined data is invaluable. PCA indicates that the HPLC and water-solvent equations give two clear independent factors which account for 86% of data variance. Although PC3 has a low eigenvector value, it may be considered to be useful because it encodes 11% of variation not included in PCI and PC2. Eigenanalysis of the Correlation Matrix for HPLC and water-solvent equations Variable PCI PC2 PC3 PC4 PC5 R2 0.382 0.590 -0.497 0.509 -0.017 7Z2^ -0.494 -0.297 0.095 0.810 0.043 Zctz" -0.289 0.714 0.638 0.012 -0.022 ZPz" -0.517 0.132 -0.403 -0.182 -0.721 Vx 0.509 -0.193 0.418 0.227 -0.691 Eigenvalue 3.291 0.987 0.547 0.158 0.018 Proportion 0.658 0.197 0.109 0.031 0.004 Cumulative 0.658 0.856 0.965 0.996 1.000 However, the most useful aspect of the PCA method in the context of this work is the ability of the method to provide a platform to compare the HPLC and water-solvent partition equations in a simple two dimensional plot, see Fig. 5.12. The analysis emphasizes the difference between HPLC and log P measurements and it is apparent that the HPLC equations are far more clustered, thus indicating greater similarity in 154 chemical characteristics. Conversely, the water-solvent systems lie in their own PC space and as such contain some unique information not associated with the other systems. It would therefore be expected that the HPLC systems would be far more limiting in terms of back-calculating to obtain descriptors because the original variables are representing essentially the same process and thus the same chemical information. Fig. 5.12 Plot of PC scores for water-solvent and HPLC equations 2.6 1 • ethyl acetate 2.0 - 1.5 - % octanol 1.0 - • diethylether diol polymer 0.5 - B p o l y C i f CM am id e m cyano Ü 0.0 - Q. phenylhexyl □ • PGDP -0.5 - perfluorooctyl hexane • • oleyl alcohol -1.0 - -1.5 - • benzene -2.0 - trichloromethane • -2.5 - -2 0 PCI log P (water-solvent) □ log k (HPLC) 155 5.5.4 Application of HPLC in Descriptor Determination Using the vector approach described in Chapter 5.5.2a, a total value of 0 for a given group of systems (each system consists of a stationary phase and mobile phase composition) can be obtained by adding together the 0 values for each pairwise comparison of columns. The greater the total 0 value, the greater the difference between the HPLC systems in the group. Each group was then utilised in an effort to obtain compound descriptors. The coefficients for each of the HPLC equations shown in Table 5.10 were entered into an in-house Microsoft Office Excel spreadsheet called Lpcalc. The corresponding group experimental capacity factors for a single compound at a time were are also entered into Lpcalc. Since R% and Vx had already been obtained, Zaz" and Zp 2^ remained to be calculated. Application of Lpcalc in conjunction with the Excel "Solver add-in" then assigned descriptor values for the compound as those for which the capacity factors were reproduced with an overall minimum standard deviation. From the log k’ values for several system groups, 712”, Zot 2^ and Zp 2^ have been back- calculated for the "internal" 50 compounds which comprise the training set. The results are given in Table 5.12. Table 5.12 Results of descriptor back-calculation using HPLC ... A A E - Columns Total 0 Zoz" ZPz" I Polymer, polyClS, amide, perfluorooctyl, diol 160.22 0.438 0.100 0.095 2 Polymer, phenylhexyl, amide, perfluorooctyl, diol 158.73 0.341 0.068 0.090 3 Polymer, CIS, amide, perfluorooctyl, diol 156.86 0.860 0.144 0.187 4 Polymer, poly CIS, perfluorooctyl, diol, cyano 156.68 0.459 0.126 0.091 5 Polymer, polyClS, amide, perfluorooctyl, cyano 151.98 0.612 0.154 0.103 6 Polymer, polyClS, CIS, amide, diol 120.10 0.451 0.133 0.095 7 Polymer, polyClS, CIS, phenylhexyl, diol 116.41 0.291 0.098 0.080 8 Polymer, polyClS, CIS, phenylhexyl, amide 111.89 0.316 0.089 0.080 AAE: average absolute error, AAE=E|caIc-obs|/n 156 As suggested by PCA, the results show that even for the compounds in the training set it is not possible to obtain all three descriptors with a sufficient degree of accuracy. Upon using the equations to back-calculate descriptors for a number external of test set compounds, it was even more evident that not all three descriptors could be obtained. The last two system combinations (7 and 8) appear to give the best results, especially for the calculation of 712^. This somewhat contradicts theory in that systems 7 and 8 have the smallest 0 totals. However, vector analysis is simply a tool for obtaining a quantitative measure of the linear correlation between two equations and does not give an indication of the level of error associated with each equation coefficient. It has not been possible to obtain all three descriptors from HPLC log k’ values and this may be reasoned as follows. Theoretically, considering that there are three unknown values to be found, the aim would be to have a series of systems for which a rotation matrix can be applied to transform log ki = 5 .7x2“ +a.Sct2^ + log k 2 = 5 .7x2“ + a X a ^ + log ks = 5.7x2“ + a X a ^ + 6.2^2“ to log k ’i = 5 .7 x 2 “ log k ’2 = a.Za2“ log k ’3 = 6 .2 ^ 2 “ In practice, this has been achieved to a certain degree for water-solvent partitions, but less so for the HPLC systems due to lack of sufficient coefficient diversity. An additional problem associated with the HPLC systems is the fact that the equation coefficients are very small in magnitude compared with those in the water-solvent systems. Tables 5.13 and 5.14 show the results when the critical quartet and the most distinct HPLC columns are considered. The log k’ values span -1.5 log units whilst logP values have a far larger range of -6 log units. The relative error associated with each HPLC coefficient is therefore much higher, especially in the case of 7X2“, see Table 5.14. Consequently, these inaccuracies will be propagated when back- 157 calculating descriptors, resulting in higher errors in descriptor prediction even though errors in the actual HPLC experimental measurements will be lower than that of water-solvent partition experiments. Table 5.13 Error associated with the ‘critical quartet’ equation coefficients Coefficient (SD of coefficient) logP s a b OctanoP -1.054 (0.021) 0.034 (0.021) -3.460 (0.026) Chloroform*’ -0.391 (0.042) -3.191 (0.043) -3.437 (0.065) Hexane^ -1.723 (0.038) -3.599 (0.034) -4.764 (0.045) PGDP^ -0.715(0.159) -1.034 (0.133) -4.852 (0.161) Average coefficient® 0.971 1.965 4.128 Average SD*^ (0.065) (0.053) (0.073) Relative error^ 0.067 0.027 0.018 “a is negligible - equation useful in predicting Ti'^ or ZPz" ^jis small - useful in predicting S a 2^ or ZPz" s and a are largest coefficients - most useful in predicting 7I2” or Ea 2^ ^6 is largest coefficient - useful in predicting ZP 2" ® Average coefficient = Scœfficients/4 ^ Average SD = Zsd/4 ® Average SD/average coefficient Table 5.14 Error associated with the HPLC equation coefficients Coefficient (SD of coefficient) log k’ a b Polymer -0.437 (0.048) -0.819(0.043) -1.449 (0.073) Perfluorooctyl -0.170 (0.041) -0.439 (0.036) -0.612 (0.062) Diol -0.240(0.028) -0.124 (0.025) -1.003 (0.043) Amide -0.225 (0.031) -0.177 (0.028) -1.099 (0.047) Average coefficient® 0JK8 0.390 1.041 Average SD*^ (0.037) (0.033) (0.056) Relative error® 0.138 0.085 0.058 e. f. S 0 _ _ j ' . _ . _ . f r r 1 1 f Although several columns with differing functionalities have been used, the HPLC equations are not significantly different enough because the mobile phase contains water. As a result, the hydrophobic effect is a major component in determining the retention time of the solutes. Thus, the coefficients in each of the phases follow the 158 same pattern of small s- and a- coefficients and a larger ^^-coefficient. The similarity between phases was revealed in the scores plot of the PCA, Fig.5.12 where the HPLC systems were clustered together. The water-solvent partition equations were shown to cover a larger co-ordinate space, and furthermore, the magnitude of coefficients in these equations are far larger than those obtained through HPLC. As a simple illustration of the error involved, the following diagram shows the difference between the observed logk’ and corresponding predicted log k’ values for 15 of the training set compounds on the C l8 column. o log k’ obs A log k’pred 0.6 -0.4 - 0.2 0.0 0.2 0.4 0.6 0.8 1.0 log k’ It is evident from the CIS log k’ plot (above) that the equation does not predict the correct eluting order of every compound since several lines intersect. The result of the same type of plot for octanol-water solvent partitioning for the same set of compounds is far more acceptable since there are no ‘cross-overs’ evident. o log Poet obs A log Poct’pred 0.0 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.50.5 log Poet 159 It is apparent that the most difficult descriptor to predict is 7T2". The inaccuracies in prediction for 712^ are more pronounced because in all of the HPLC systems, the s- coefficient is small (-0.170 to -0.437), and hence the associated relative error in calculating 7:2" is high. Conversely, the superior predictions of ZP 2" are due to the large ^-coefficients obtained. Of the three descriptors, Z a 2^ can be most easily calculated since it is generally an additive value for compounds in which the acidic functional groups do not interact with each other or any other substituent groups. Also, the Abraham UCL database contains E a 2^ values for many compounds including heterocycles which are of general interest. Therefore, it was decided that this descriptor should be calculated by either using ABSOLV or by addition of fragments, thus leaving only 712^ and Sp 2^ to be determined by the HPLC method. Since only two descriptors can be calculated, it was considered tedious to use Solver. Instead, a multiple linear regression equation can be devised for each of the two descriptors using the capacity factors obtained from the relevant HPLC systems as the variables. A different combination of systems can be used to predict 7ü2^ and ZP 2" because the equations to derive each descriptor are independent. However, experimentally, it is obviously advantageous if the same group of systems can be used to predict both 712^ and Zp 2^- It is reasonable to use either 3 or 4 systems for prediction since there are only two unknowns. Stepwise regression indicated that both the polyClS and CIS columns are most useful in the prediction of 7t2^, whilst polyClB, diol and phenylhexyl columns are required for prediction of ZP 2" (given that Z0C2" is calculated separately beforehand). However, the LFERs for the CIS and phenylhexyl columns have been shown to be chemically similar by both principle component and vector analysis. Although it is experimentally feasible, the use of both the CIS and phenylhexyl columns is theoretically questionable because of the high degree of redundancy that is evident. This was proved correct through numerous MLRA in which it was found that 3-system groups were as accurate as the 4-system groups in predicting 7:2" and ZP2". Furthermore, incorporation of the CIS equation rather than the phenylhexyl equation resulted in more accurate predictions for the training set compounds ^ Overall, the perfluorooctyl and cyano phases were considered to be of limited use due to low column stability 160 Best MLRA equations used to obtain and z p ” 7I2" =-0.080 + 0.043«2-0.7102'a2''+ 1164yx:-3.051/ogÆ’‘^^*'"''^^' + 2.571/og - 0.339kg r^ = 0.890, SD = 0.192, n=50 ZP2" = -0.254 + 0.176% - 0.059% " + 0.616V>: + 0.447kg . 0.829kg _ 0 359/og ^..diousono) = 0.978, SD = 0.058, n=50 These equations were used to predict the descriptors for ten test set compounds (Table 5.16) and also a series of Roche compounds (Table 5.18). Several test set compounds were relatively strong acids and had to be run at a low pH to ensure that these solutes were unionized. Theoretically, it is expected that weak acids, weak bases and neutral compounds should have the same capacity factors irrespective of whether experimental pH is acidic, basic or neutral. However, to ensure that this assumption applies in practice, 25 compounds were run at pH 2.5 (0.1%TFA (aq).) for comparison with capacity factors obtained at pH7. The results are as expected (Fig. 5.13) and so it is assured that the equations derived at pH7 are applicable to any future acidic compounds which will be run at low pH. Fig. 5.13 Plot of log k’ for compounds at both pH 2.5 and pH 7 y= 1.0151X-0.0445 = 0.9992 If) - 1.0 -0.5 0.5 -0.5 - pH7 161 Table 5.15 Test set compounds and corresponding database descriptors R2 ZŒ2" ZP2" Vx Dexamethasone 2.040 3.51 0.71 1.92 2.9132 Coumarin 1.060 1.79 0.00 0.46 1.0619 Ibuprofen 0.700 0.92 0.60 0.60 1.7771 Anisole 0.708 0.75 0.00 0.29 0.9160 4-Nitroaniline 1.220 1.83 0.45 0.38 0.9904 3,4-Dichlorophenol 1.020 1.14 0.85 0.03 1.0199 3-Nitrophenol 1.050 1.57 0.79 0.23 0.9493 Resorcinol 0.980 1.10 1.09 0.52 0.8338 4-Me-p-hydroxybenzoate 0.900 1.37 0.69 0.45 1.1313 Caffeine 1.500 1.60 0.00 1.33 1.3632 The equations were able to predict 712 ^ and ZP2" for the external test set compound: with a reasonable degree of accuracy if caffeine is excluded (Table 5.16). Table 5.16 Test set log k’ values and predictions of 712 ^ and log k’ pred - ZP2" pred polyC18 C18 diol pred obs pred obs Dexamethasone -0.100 -0.198 -0.091 3.27 -0.24 1.89 -0.14 Coumarin 0.150 0.007 -0.131 1.61 -0.18 0.51 0.05 Ibuprofen 0.465 0.495 0.280 1.18 0.26 0.67 0.07 Anisole 0.344 0.331 0.030 0.88 0.13 0.29 0.00 4-Nitroaniline 0.303 -0.073 0.000 1.81 -0.02 0.26 -0.12 3,4-Dichlorophenol 0.553 0.257 0.357 1.07 -0.08 0.03 0.00 3-Nitrophenol 0.204 -0.034 -0.005 1.14 -0.43 0.29 0.06 Resorcinol -0.209 -0.488 -0.374 1.24 0.14 0.46 -0.06 4-Me-p- 0.043 -0.154 -0.289 1.46 0.09 0.56 0.11 hydroxybenzoate RMSE = 0.21 0.07 AAE = 0.17 0.06 AE = -0.04 0.01 [ Caffeine -0.732 -0.727 -0.874 2.20 -0.60 1.45 -0.12] RMSE: Root mean square error, RMSE=[Z(calc-obs) /n] AAE: average absolute error, AAE=Z|caIc-obs|/n., AE: average error, AE=E(caIc-obs)/n. The prediction of 712^ for caffeine was far higher than the actual database value. This result was consistent with whichever combination of HPLC systems was used and as 162 such it is unlikely that an erroneous experimental value is the cause of the prediction error. The error is more likely attributed to the fact that caffeine is the fastest eluting compound and as such may be beyond the scope of a HPLC method for predicting descriptors. 5.5.5 Conclusion Unfortunately, this work has shown that it is not possible to obtain all three descriptors due to the lack of diversity and low coefficients obtained from the HPLC systems studied in this work. It appears unlikely that a different stationary phase or even a complete change in organic modifier will provide sufficient improvement, as indicated in Table 5.17, where the problem of small s- and «-coefficients are evident despite a change of solvent and use of more ‘exotic’ stationary phases. Table 5.17 Coefficients of other reversed-phase HPLC systems studied. Column Mobile phase r s « b V lAM^' 10%MeCN 0.81 -0.42 0.69 -2.00 1.87 DPC*’ 10%MeCN 0.49 -0.44 0.03 -2.76 2.68 PRP r 80%MeOH 0.45 -0.05 -1.09 -1.70 1.46 C18'' 70%MeOH 0.28 -0.58 -0.44 -1.23 1.35 C18' 60%THF -0.09 -0.20 -0.26 -0.98 0.69 " Immobilised artificial membrane phase, ref 30 ^Dipalmitoyl phosphatidylcholine coated silica phase, ref 30 Poly(styrene-divinylbenzene) column, ref 30 Spherisorb ODS-2 stationary phase, tetrahydrofuran/water mobile phase ref 31 Spherisorb ODS-2 stationary phase, methanol/water mobile phase ref 31 Whilst HPLC cannot be used in isolation to assign Ttz", Zotz", and EP 2", there exists the possibility of using HPLC in combination with water-solvent partitions. The PCA plot in Fig. 5.12 indicates that the HPLC systems are markedly different with respect to the water-solvent systems. An appropriate compromise in terms of preferable equation coefficients and experimental efficiency may therefore be the best solution to descriptor determination. The water-octanol partition is the most widely measured system, and therefore the most accessible. However, in terms of equation coefficients, it may not be the most appropriate choice due to the negligible «-coefficient. In this respect, a water-alkane system such as water-hexane would .be preferable because the «-coefficient is very large (-3.599) which is required if the aim is to derive 'La^. 163 Table 5.18 Examples of Roche compounds for which descriptors have been obtained from this work log k’ Ro number R2" 7:2"' V x" C18 polyC18 diol 65/35" 60/40" 30/70" RO-31-9790/000 0.999 1.70 1.34 2.45 2.6273 -0.910 -1.332 -0.804 RO-31-3948/000 1.323 2.11 0.60 1.12 2.2039 0.201 0.258 0.174 RO-64-0796/001 0.793 1.44 0.56 2.18 2.5598 -0.424 -0.975 -0.355 RO-31-8959/008 4.285 4.33 1.71 3.36 5.2961 0.424 0.308 0.667 RO-32-3346/000 2.614 4.58 1.09 3.26 4.8988 -0.111 -0.258 0.179 RO-32-5592/000 4.116 4.25 1.38 3.56 6.7208 1.990 1.532 1.643 RO-31-6930/000 1.709 2.51 0.00 1.50 2.1081 -0.169 -0.199 -0.171 RO-32-7066/000 3.059 3.54 1.11 2.32 3.6424 -0.048 0.002 0.346 RO-32-7315/000 2.196 3.39 0.94 2.21 3.5982 0.012 -0.009 0.265 RO-32-5287/000 2.589 3.09 0.41 1.72 3.0061 0.267 0.317 0.461 “ Calculated from addition of relevant fragments Obtained from HPLC using the C l8, polyClS and Diol phases ^ Calculated from atomic contributions ^ Mobile phase composition [(MeCN/HiO) in vol %] RO-31-9790/000 RO-31-3948/000 HO Cl O o 164 RO-64-0796/001 RO-31-8959/008 NH, 0 , ^ N H OH H RO-32-3346/000 RO-32-5592/000 NHz OH N N H 165 RO-31-6930/000 RO-32-7066/000 RO-32-7315/000 RO-32-5287/000 HO, 166 5.6 References 1. Poole, C.; Poole, S.K. Chromatography Today, Elsevier Science Publishers B.V. 1991 2. Melander, W.; Hovarth, Cs. High performance liquid chromatography (Advances and perspectives, vol. 2. Academic Press, New York, 1980. 3. Dorsey, J.G.; Dill, K.A. The molecular mechanism of retention in reversed-phase liquid chromatography. Chem. Rev. 1989, 89, 331-346. 4. Vailaya, A.; Hovarth, Cs. Retention in reversed-phase chromatography : partition or adsorption. J.Chromatogr. A. 1998, 829, 1-27. 5. Dill, K.A. The mechanism of solute retention in reversed-phase liquid chromatography. J. Phys. Chem. 1987,91,1980-1988 6. Boudreau, S.P.; Cooper, W.T. Analysis of thermally and chemically modified silica-gels by heterogenous gas solid chromatography and infrared-spectroscopy. Anal. Chem 1989, 61,41-47. 7. Sindorf, D.W.; Maciel, G.E. SI-29 NMR-study of dehydrated rehydrated silica-gel using cross polarization and magic angle spinning. J. Am. Chem. Soc. 1983, 105, 1487-1493. 8. Mauss, M.; Engelhardt, H. Thermal treatment of silica and its influence on chromatographic selectivity. J. Chromatogr 1986, 371, 235-242. 9. Meyer, V.R. Practical high-performance liquid chromatography. John wiley & Sons, Surrey, 1988. 10. Engelhardt, H.; Ahr, G. Properties of chemically bonded phases. Chromatographia 1981, 14, 227- 233. 11. Snyder, L.R.; Poppe, H. Mechanism of solute retention in liquid-solid chromatography and the role of the mobile phase in affecting the separation. J. Chromatogr. 1980, 184, 363-413. 12. Kohler, J.; Chase, D.B.; Farlee, R.D.; Vega, A.J, Kirkland. Comprehensive characterization of some silica-based stationary phases for high-performance liquid chromatography. J. Chromatogr. 1986, 352, 275-305. 13. Nawrocki, J. Silica surface controversies, strong adsorption sites, their blockage and removal 1. Chromatographia 1991, 31, 177-192 14. Nawrocki, J. Silica surface controversies, strong adsorption sites, their blockage and removal 2. Chromatographia 1991, 31, 193 15. Benson, J.R.; Woo, D.J. Polymeric columns for liquid chromatography. J.chromatogr. Sci. 1984, 22, 386-399. 16. Stuurman, H.W.; Kohler, J.; Jannson, S O.; Litzau, A. Characterization of some commercial poly (styrene-divinylbenzene) co-polymers. Chromatographia 1987, 23, 341-349. 17. Lindsay, S. High performance liquid chromatography. Chichester : Wiley on behalf of ACOL, 1987. 18. Zhao, J.; Carr, P.W. Comparison of the retention characeteristics of aromatic and aliphatic reversed phases for HPC using Linear Solvation Energy Relationships. Anal. Chem. 1998,70, 3619-3628. 19. Reta M, Carr PW, Sadek PC, Rutan, S.C. Comparative study of hydrocarbon, and aromatic bonded RP-HPLC stationary phases by linear solvation energy relationships. Anal.Chem. 1999, 71, 3484- 3496. 167 20. Abraham, M.H.; Grellier, P.L.; Abboud, J-L.M.; Doherty, R.M.; Taft, R.W. Can. J.Chem. 1988, 66, 2673-2686 21. Abraham, M.H.; Roses, M. Hydrogen-bonding 38. Effect of solute structure and mobile-phase composition on reversed-phase high-performance liquid-chromatographic capacity factors. J.Phys. Org. Chem. 1994,672-684. 22. Platts, J.A.; Butina, D.; Abraham, M.H.; Hersey, A. Estimation of molecular linear free energy relation descriptors using a group contribution approach. J. Chem. Inf.Comput. Sci. 1999, 39, 835- 845. 23. Leahy, D.E.; Morris, J.J.; Taylor, P.J.; Wait, A.R. Model solvent systems for QSAR.2. Fragment values (F-values) for the critical quartet. J. Chem. Soc- Perk.Trans.2. 1992,4, 723-731. 25. Abraham, M.H.; Chadha, H.S. Applications of a solvation equation to drug transport properties. In : Lipophilicity in Drug action. VCH Verlagsgesellschaft mbH, Weinheim, Germany, 1996. 25. Abraham, M.H.; Platts, J.A.; Hersey, A.; Leo, A.; Taft, R.W. Correlation and estimation of gas- chloroform and water-chloroform partition coefficients by a linear free energy relationship method. J. Pharm. Sci. 1999, 88, 670-679. 26. Abraham, M.H.; Green, G.E.; Platts, J.A. Unpublished results 27. Abraham, M.H. Unpublished results 28. Abraham, M.H.; Chadha, H.S.; Dixon, J.; Leo, A.J. Hydrogen-bonding.39. The partition of solutes between water and various alcohols. J. Phys. Org. Chem. 1994,7, 712-716. 29. Ishihama, Y.; Asakawa, N. Characterization of lipophilicity scales using vectors from solvation energy descriptors. J. Pharm. Sci. 1999, 88, 1305-1312. 30. Abraham, M.H.; Chadha, H.S.; Leitao, A.R.E.; Mitchell, R.C.; Lambert,W.J.; Kaliszan, R.; Nasal, A.; Haber, P. Determination of solute lipophilicity, as log P(octanol) and log P(alkane) using poly(styrene-divinylbenzene) and immobilised artificial membrane stationary phases in reversed- phase high-performance liquid chromatography. J. Chromatog. A. 1997,766, 35-47. 31. Abraham, M.H.; Roses, M.; Poole, C.F.; Poole, S.K. Hydrogen bonding .42. Characterization of reversed-phase high-performance liquid chromatographic C-18 stationary phases. J.Phys. Org. Chem. 1997, 10, 358-368. 168 Chapter 6 The Solubility of Compounds in water 6.1 Introduction Aqueous solubility is a very important molecular property which plays an integral role in many different biological and physical processes and represents an equilibrium distribution of a solute between water and the solute phase. The aqueous solubility of a drug influences its release, transport, and also the rate and extent of its absorption through biological membranes. Ferguson, in 1939, first recognized situations where poor solubility prevents a solute from reaching its full toxic potential \ For example, although the toxicity of chlorobenzenes to Daphnia magna increases with increasing chlorine substitution, hexachlorobenzene is non-toxic owing to its limited solubility^. In the environmental sciences, the aqueous solubility of agrochemicals and pollutants is a key determinant of its environmental fate. It is thus not surprising that many quantitative structure activity relationships are at least partially related to solubility. Liquid water forms molecular networks which are in essence clusters of water molecules bound tightly to one another via hydrogen-bonding. The process of solubility in water consists of essentially three energy dependent steps. (1) The endoergic formation of a cavity in the water network of equal volume to that of the incoming solute molecule. (2) A single solute molecule must then be separated from the bulk liquid solute or from the crystalline solid and deposited in the cavity. The solvent molecules reorganize around the cavity (3) Finally, the initialisation of attractive forces between the water and solute. Formation of van der Waals interactions between the solute and neighbouring solvent molecules (on account of the dipolarity and polarizability of the respective molecules), and formation of intermolecular hydrogen bonds where possible. Aqueous solubility is traditionally measured by thermodynamic solubility assay (THESA) in which a saturated solution is prepared, afterwhich the excess undissolved solute is separated. The composition of the resulting solution is then determined by a suitable analytical procedure. Although the solubility of a compound may seem a rather 169 easy quantity to determine, all too often, the procedure is beset with difficulties. Low solubilities, for example 10'^ or 10'^ mol L'* are difficult to determine. Larger errors are associated with their measurement because the measured values are closer to the lower limit of instrument sensitivity and the effect of even relatively small impurities in the compound sample are more pronounced. As a result, many compilations of solubility have differences of up to one log unit for the same compound. More recently, experimental methods such as turbidimetric solubility^ and laser nephelometry"^ have been employed in the early discovery phase of pharmaceutical research. Measurements from these methods are useful as a benchmark to highlight compounds that have poor solubility and that are thus likely to have absorption and permeability rate problems. However, the data obtained is not equivalent to a normal thermodynamic assay; instead, the apparent solubility is largely kinetically driven^ because the compounds are usually pre-dissolved in a water-miscible solvent such as DMSO, Although these methods provide a higher throughput of compounds, the numbers are dwarfed in comparison to the vast number of possible drug-like molecules within combinatorial libraries. 6.2 Prediction Methods The ability to predict aqueous solubility would therefore greatly assist the drug and agrochemical development process. A viable method for predicting this parameter would reduce time, effort and cost in synthesis and product design because an estimate of the aqueous solubility for a compound could potentially eliminate some of the large number of unsuitable candidates before they are even synthesized. Not surprisingly, numerous methods for the prediction of aqueous solubility have been proposed; these may be separated into four main techniques^. (1) Methods based directly upon group contributions. (2) Methods based on theoretically calculated properties. (3) Techniques based upon experimental physicochemical properties (4) Techniques based upon combinations of two or more parameters that may be calculated, derived empirically, or experimentally measured. Discussion and comparison of methods will be restricted primarily to methods that include solid solutes, because methods that predict only liquid solubilities are of limited use. 170 (1 ) Group contributions One of the first predictive methods for aqueous solubilities was that of Irmann,^ who set up a group contribution scheme for liquid hydrocarbons and halocompounds. For solids, Irmann used an additional term, ASm(Tm -T)/1364, where ASm is the entropy of fusion (melting) at the melting point, Tm. A value of 13 cal deg mol was taken for ASm, leading to the simplified correction term, - 0.0095 (mp - 25) (1) In eqn 1, mp is the melting point in °C for solids; liquids are assigned a mp of 25 °C, thus zeroing out the (mp - 25) term. Irmann^ gave no statistical analysis, but it was possible to use Irmann’s original data, excluding compounds for which the observed solubility was given as approximate, and to obtain the details given in Table 6.01, page 175. Several other group contribution schemes have been constructed,^'*® some of which^ do not require any mp correction term. The UNIFAC (UNIQUAC functional group activity coefficients) and UNIQUAC methods are also group contribution schemes, and have been used to estimate aqueous solubilities.**’*^ Since the reference state for solutes in these methods is the pure liquid, they require a knowledge of the solute enthalpy of fusion, or an approximate mp correction term for solids. Another type of group contribution scheme is used in the AQUAFAC program,*^'*^ applied to 970 compounds.*^ In this method, the aqueous activity coefficient is calculated by logYw = Zn^qi (2) where n, is the frequence of occurrence of group i and q, is the contribution of that group. Individual q values are obtained by fitting experimental solubility and melting point data by linear regression. Again, the applicability of AQUAFAC is limited by the need for a melting point or the entropy of fusion for solid solutes. Kuhne et al*^ made a comparative analysis of five Group Contribution methods using a test set of 351 organic liquids and 343 solids. With all methods, solubility prediction was clearly better for liquids than for solids. Therefore, the conclusion is that an explicit account of the entropy of fusion by a melting point term will generally lead to an improved performance of Group Contribution schemes. 171 (2) Theoretically calculated descriptors A number of correlations are based on theoretically calculated descriptors.None of these require any mp correction term for solids, and so are capable of predicting aqueous solubilities from structure. Computational methods are employed to create a large pool of descriptors for each compound in the dataset. These descriptors are fitted to the experimental solubility data through application of either multiple linear regression analysis (MLRA) or artificial neural networks (ANN). ANN attempt to model the way in which the human brain processes information and consist of a large number of ‘units’ or ‘neurones’ represented by a computer program which are interconnected. The signals between these units are weighted according to chosen initial conditions and the application of a learning algorithm iteratively minimizes the difference between the input (descriptor values) and output values (log Sw) by altering the weights of these signals. This leads to a small subset of descriptors being chosen that are optimally related to solubility. The number of descriptors varies from model to model and may include quantum chemical parameters, geometric parameters and connectivity indices. Neural networks have the advantage that non-linear relationships can be modelled. (3 )Techniques based upon experimental physicochemical properties Quite different types of calculation were initiated by Hansch and co-workers,^^ who showed there was a relationship, eqn 3 between log Sw and the water-octanol partition coefficient, as log Poet, for a training set of 156 liquids; Swis the solubility in mol L \ log Sw = -1.339 log Poet + 0.978 (3) n=156, r^ = 0.874, SD = 0.472 This empirical correlation may be rationalised thermodynamically as follows^®. The partition coefficient is defined as an equilibrium constant which relates the activity of the solute in two miscible phases at equlibrium dZo Poet = ----- (A\ where Oq and ttw are the activities of the solutes in octanol and water respectively. Alternatively, the activity can be replaced by a product of the activity coefficient (T) and the concentration (C) of the solute; T is the Henrys Law activity coefficient. 172 (5) = ^w w The activity coefficients can be approximated by unity in dilute solutions, and so the equation reduces to ; Poet = (6) The concentrations at saturation for most liquids are equal to the solubilities in the octanol and water phases respectively : Pact - ° 3 . a ) Taking logs : log Sw = -log P oet + logS q (8) The log So term will be nearly constant if the solubilities of liquid solutes in octanol is comparable. In fact octanol has a solubility parameter of 10.3 and as such was completely miscible at room temperature for all the liquid solutes considered. Hence if complete miscibility is approximated by So = 1, then log Sw = -log Poet. Yalkowsky and Valvani^^ extended the applicability of this relationship by incorporation of similar terms to those used by Irmann^ for solids. They showed that the entropy of fusion could be estimated, and that the entropy of fusion term could be replaced by a mp correction term, as in eqn 9. Several related equations were put forward,*^’ log Sw = -1.05 log Poet - 0.012 (mp - 25) + 0.87 (9) n= 155, r^ = 0.979, SD = 0.308 log Sw = -1.00 log Poet - ASm (mp - 25)/1364 + 0.87 (10) n = 873 ASm = 13.5-4.6 (log a) (11) In eqn 9, and elsewhere, n is the number of data points, SD the standard deviation, r the correlation coefficient and F the F-statistic. Values of log Pœt in eqn 10 were not 173 experimental ones, but were calculated by the ClogP program. The entropies of fusion were a combination of experimental values, and calculated ones using eqn 11 where a is the rotational symmetry number. However, the compound mp is still needed in order to apply eqn 10, and so log Sw values cannot be calculated from structure. Meylan and Howard^^ have further extended this empirical correlation of log Poet with solubility and propose two equations which form the basis of the Syracuse Research Corporation (SRC) Wskow program for predicting aqueous solubility. log S = 0.796 - 0.854 log Poet - 0.0728 MW + Efi (12) n = 1450, ? = 0.934, SD = 0.585 log S = 0.693 - 0.960 log Poet - 0.092(mp-25) - 0.0034 MW + Sfi (13) n = 1450, = 0.970, SD = 0.409 where MW is the molecular weight of the solute and Sfi is the summation of all correction factors applicable to a given compound; 15 correction factors were specified. Eqn 12 is useful in that it allows solubility to be calculated from compound structure since log Poet may be estimated from programs such as ClogP or WsKow itself. However, prediction accuracy is improved when experimental values are used, and the melting point equation is superior to the equation without this term. (4) Techniques based upon combinations of two or more parameters Mobile Order Theory^®'^^ has recently been applied to the estimation of aqueous solubility in volume fraction.^^ Mobile order thermodynamics leads to a ‘universal solubility equation’ which contains as many as six terms incorporating parameters of fluidization (A), entropy (B and F), non-specific forces (D) and hydrogen-bonding (O and OH). ln0B = A + B4-D4-F4-O + OH (14) The number of terms in the solubility equation is determined by the types of functional groups present in the molecules. For example, if only hydrophobes are considered, the simplified eqn 15 is adequate to predict solubility. 174 In 0B = -A - 1.948 - 0.083 Vb + 0.51nB (15) Hence for hydrophobes (nonpolar or slightly polar compounds with no hydrogen- bonding capacity), solubility is dependent on solute mp and molar volume. Although this method provides accurate prediction of solubility, not only the entropy of fusion of solid solutes (or a melting point correction term where A=0.0278(Tm-298.15)) is required, but also a modified nonspecific solute cohesion parameter. The latter is obtained either from experimental solubilities in hydrocarbon solvents or is ‘...deduced by analogy to similar compounds.’^^ 6.2.1 Comparison of Literature Models Tables 6.01 and 6.02 summarize the methods that have been applied more generally, that is to large sets of structurally diverse compounds. Some workers list outliers but other workers do not. Since the number of outliers can be very large (42 out of 258 for the general case in ref.21), care has to be taken in judging one model against another. Wherever possible in Tables 6.01 and 6.02, the standard deviation has been calculated in order to provide a uniform basis of comparison. Table 6.01 - Models for the correlation and prediction of aqueous solubility, as log Swj that require additional data (ASf, mp, 6’) ----- Training set — ------Test set ------ref n SD outliers n SD outliers 168 0.31" AAE (0.17) none 6 694 AAE (0.38) none 9 167 0.24 none 27 150 0.48" none none 31 873 0.56"^ none 97 0.56‘‘ AAE (0.41) none 15 873 0.80" none 97 0.80' AAE (0.61) none 15 1450 0.41 7 817 0.62 none 29 “Calculated in this work. The AAE is given by AAE = Z(|log Swcalc - log Swobs|)/n ^ Calculated in this work from data in Tables 2 and 3 in ref. 12. Calculated in this work. Note that the value in ref. 31 at the foot of Table 4 is incorrect. AQUAFAC method. The SD value for the test set has been calculated in this work, and the SD value for the training set has been taken to be the same. 'Eqn 10. The SD value for the test set has been calculated in this work and the SD value for the training set has been taken to be the same. 175 Table 6.02 - Models for the correlation and prediction of aqueous solubility, as log Sw 9 that do not require additional data - Training set - — Test set — ref n SD outliers n SD outliers 469" 0.46 none 25 0.50 none 8 12 0.37 1 483*" 0.53 none 25 0.55 none 8 19 0.86 2 123" 0.22 4 13 0.23 none 22 123"* 0.28 4 13 0.28 none 22 258 0.37 42 21 411 0.57 none 23 331 0.30 none 17 0.34 none 20 884" 0.47 none 413 0.60 none 24 884^ 0.67 none 413 0.71 none 24 1450 0.59 7 817 0.72 none 29 ^ Model I, that is not too general. A later analysis^ gives AAE = 0.50 for 694 compounds. Model II, that is very general. A later analysis’ gives AAE = 0.56 for 614 compounds. ^ Using a neural network with nine descriptors. Using a linear model with nine descriptors. ® Using a neural network with 30 topological descriptors. ^ Using a linear model with 30 topological descriptors. Both the quality and size of the dataset play a crucial role in the development of a prediction model. If data is erroneous, the descriptors found to be important may contain information which is of little use in predicting the solubility, but may contain information that is able to predict the erroneous values effectively. If the number of compounds on which the model is based is too small, the structures may not be sufficiently diverse and the model will have little ability to generalize. Myrdal and co-workers*^ have pointed out that experimental solubilities are a source of considerable error and that inter-laboratory differences are significant even when researchers are following the exact same protocol. This is illustrated by the results of a Japanese ring test^^ in 1984 in which 19 different laboratories were asked to determine the solubility of both anthracene and fluoranthene (Table 6.03). The recorded log Sw values for anthracene differ by 1.85 log units, and for fluoranthene by 1.15 log units. 176 Table 6.03 Comparison of different compilations of solubility data 15 Compound Average Range of Number of solubility SD solubilities different (log(Molar)) (log(Molar)) solubilities Anthracene (20°C)* -6.68 0.19 -7.08 to -6.22 17 Anthracene (20-30°C)'* -6.51 0.38 -6.79 to -5.23 22 Fluoranthene (20°C)* -6.09 0.10 -6.38 to -5.94 17 Fluoranthene (25°C)^ -5.96 0.07 -6.28 to —5.88 4 Fluoranthene (25°C)^ -5.73 - -6.01 to -5.28 7 Fluoranthene (20-30°C/ -5.96 0.19 -6.23 to -5.35 13 ....Tin...... 2";^; ...... ï " Furthermore, the several methods use only a small test set of compounds with which to test the predictive capability of the model. As such it is not possible to properly assess the applicability of exploiting these methods to accurately predict solubility for novel compounds. The comparisons of the various models is thus difficult. Bodor and Huang^® obtain a very low SD value of 0.30 log units for a 331 compound data set, using 18 theoretically calculated descriptors, and Sutter and Jurs^^ find even lower SD values of 0.27 and 0.22 for a 123 compound data set. Meylan and Howard find a slightly larger SD of 0.41 for a set of 1450 compounds using log Poet, MW, mp, and correction terms, but Myrdal and co-workers,*^ find a much larger SD of 0.56 for an 873 compound training set using the AQUAFAC model. However, the 331 training set^** includes very few complicated molecules, and the 123 compound data set^^ no complicated molecules at all, whereas the 1450 and 873 training sets*^ are much more diverse. What is also evident from Tables 6.0.1 and 6.0.2, is that there is no clear advantage, as regards SD values, of methods that require additional solute properties. In view of the importance of high throughput screening, the methods that do not require additional solute properties are certainly preferable, because they allow the calculation of log Sw from structure alone. It is reasonable to conclude that for training sets that do not contain compounds of complicated structure, SD values as low as 0.30 log units may be obtained, but that for training sets that contain a reasonable proportion of complicated structures, for many of which only one solubility determination has been made, experimental error probably precludes SD values less than around -0.50 log units. 177 6.3 Application of the Abraham General Solvation Equation in Predicting Aqueous Solubility This method starts with the General Solvation Equation, which is set up and explained in Chapter 1.4, pages 22-31. log Sw = c + rR 2 + 5712^ + + 62^2^ + vVx (16) where Sw is the solute water solubility in units of mol L \ As has been shown in Chapter 6.2, there are numerous methods for predicting aqueous solubility. Although accuracy of prediction is certainly satisfactory, most methods are simply numerical in nature and offer little physicochemical interpretation of the solubility process. The aim of the present work is to obtain an equation for the correlation of log Sw values, without the need for a melting point correction, using a large training set and also a reasonably large test set of compounds. Furthermore, the derived equation should yield quantitative information on the physicochemical factors that most influence solubility since the Abraham descriptors encompass fundamental bonding, interaction and size parameters. In constructing an equation for log Sw, a number of data bases were used in order to set out values of log Sw for 1082 solids and liquids at atmospheric pressure and temperature of 20-25°C. To satisfy the temperature criteria, a number of compounds which were included in the analyses of Huuskonen^"^ were disregarded in this work because they are known to have been measured outside this temperature range. These compounds were ethinylestradiol (log Sw = -4.30, 27°C), ethisterone (log Sw = -5.66, 27°C), Menadione (log Sw = -3.03, 30°C), sulfaethidole (log Sw = -1.94, 37°C), glutaric acid (log Sw = 1.00, 28°C), guaiacol (log Sw = -1.96,15°C), and tetroxoprim (log Sw = -2.10, 30°C). Solubility data for compounds miscible in water, such as methanol, acetonitrile were taken from the references quoted. Such values are usually obtained from the equation log L = log Sw - logCo (17) 178 where L is the Ostwald solubility coefficient in water and Cq is the molar concentration of the compound in the gas phase, corresponding to the saturated vapour pressure at 298K. Several examples of values obtained by this method are shown in Table 6.04, along with the log Sw value given from the relevant references; there is good agreement between values. Table 6.04 Solubility values for compounds miscible with water Solute logL"^ log Cg log Sw ^ log Swobs ^ Methanol 3.74 -2.17 1.57 1.57 Ethanol 3.67 -2.50 1.17 1.10 Acetonitrile 2.85 -2.32 0.53 0.26 Acetic acid 4.91 -3.08 1.83 2.00 THF 2.55 -2.06 0.49 1.15" Ethylamine 3.30 -1.25 2.05 2.06 Diethylamine 2.99 -1.90 1.09 1.03 Trimethylamine 2.35 -1.05 1.30 1.32 “ Water solubility calculated from log Lw and log Cg values ^ Water solubility taken directly from reference quoted Value for THF is cited from ref 9 as being 1.15. This value is probably a typographical error and should have been 0.15. The descriptors for each of the compounds in the data set were taken from the Abraham UCL Database, obtained from referenced partition measurements (Medchem 2000 Database, Biobyte corporation and Pomona College in co-operation with Daylight), or calculated using ABSOLV (details in Chapter 1.4.6 page 31). It was evident from preliminary regression analysis that there were problems with solubility prediction for several compounds with a high Vx descriptor value. From the histogram in Fig. 6.1, it is obvious that compounds which have vastly larger values than the optimum category of Vx from 1 to 1.5 are not well represented in the data set. Although there were only three compounds with 3.5 179 Fig. 6.1 Histogram showing distribution of dataset Vx values. 4 0 0 j 3 5 0 -- 3 0 0 -- ^ 2 5 0 -- oc 3 200 - - 1 5 0 - 100 - 50 - r —1 >0 to 0.5 > 0 .5 to 1 >1 to 1.5 >1.5 to 2 >2 to 2.5 > 2.5 to 3 > 3 to 3.5 > 3 .5 to 4 >4 Vx A further six compounds were excluded from the 1077 data set, viz: uracil, adenine, 4- methyloctane, decane, undecane and dodecane because a preliminary analysis showed that these six were outliers to all the equations constructed, i.e (pred-obs) log Sw >2 log units. This left a total of 1071 compounds for the final analysis with solubility data spanning 11 orders of magnitude. Every fourth compound in a random order was selected to form a test set, to give 803 compounds as a training set and 268 compounds as a test set. The total set of 1071 compounds is given at the end of the chapter in Tables 6.13 (pages 202-215) and 6.14 (pages 216-224), together with references. Application of eqn 16 to the 804 training set yielded eqn 18 and eqn 19, with SD values of 0.68 and 0.76 log units respectively; note that fewer compounds were used in eqn 18, because of lack of mps. log Sw = 0.579 - 0.866 R 2 +0.552%" + 0.524 Zaz" + 2.919 Z k " -3.445 V, - 0.004 (mp - 25) (IS) n = 707, r^ = 0.889, SD = 0.681, F =929, AAE = 0.529, =0.889 log Sw = 0.636 - 1.105 Rz +0.561%" + 0.046 Saz" + 2.674 XPz" - 3.261 V, (19) n = 803, = 0.859, SD = 0.758, F = 969, AAE = 0.588, r%,,p= 0.85) It is somewhat surprising that eqn 16 has led to the reasonable eqn 18 and eqn 19, because eqn 16 was not set up at all to correlate quantities such as log Sw. There is a 180 fundamental difference between processes such as water-solvent partitions, to which eqn 16 has previously been applied, and solubility in water. In the former processes, the thermodynamic standard states are those of unit molar concentration and unit activity in each phase, but for solubility in water the standard states are unit molar concentration and unit activity in water, but the pure liquid or solid in the other phase. As pointed out b e fo re ,th e standard state of pure liquid or pure solid is equivalent to a different standard state for each compound. Eqn 16 is constructed for processes in which different solutes have the same standard state in each phase. In chemical terms, this means that a solute in a given phase is surrounded by the phase molecules, whereas for the standard state of pure liquid or solid, the solute is surrounded by itself. Difficulties in application to aqueous solubility of similar equations to eqn 16 have previously been encountered. For example, Kamlet, Taft and co-workers^^ correlated the solvatochromie parameters with water solubility and found that the equations derived for liquid aromatic and aliphatic solutes were quite different. The problem lay in the uncertainty in the contribution of the n term representing solute dipolarity/polarizability. Eqn 16 can be amended in order to incorporate terms that reflect interactions in the pure liquid or solid. A term in S a 2^ x ZP2" will deal with hydrogen bond interactions between acid and basic sites in the solid or liquid, and a term in 712^ x 712^ with dipole/dipole interactions, see Tables 6.05 and 6.06. The best equations constructed on these lines are, log Sw = 0.358 - 0.776 R; +0.456%" + 1.730% " + 3.397 XPa" - 1.033 Zaz" X ZPz" - 3.469 V, - 0.005 (mp - 25) (20) n = 707, r^ = 0.910, SD = 0.614, F =1009, AAE = 0.465, 0.910 log Sw = 0.412 -1.015 Rz +0.370% "+ 1.188 Z«z" + 3.296 ZPz" - 0.827 X t t z " X zPz" - 3.343 V. (21) n = 803, 1^ = 0.889, SD = 0.672, F =1061, AAE = 0.530, 0M9 181 T able 6.05 Stepw ise regression matrix for training set, n = 803 Step Vx Z k " R] Z«2" X I P2 ” 2 :0 (2 " 7 1 2 " (7 :2 ")' r' SD F 1 -1.857 0.390 1.570 512 2 -3.740 2.809 0.797 0.906 1571 3 -3.167 2.977 -0.789 0.848 0.783 1494 4 -3.335 3.491 -0.770 -0.422 0.864 0.744 1262 5 -3.264 3.475 -0.820 -0.877 1.350 0.885 0.684 1226 6 -3.341 3.292 -1.016 -0.823 1.180 0.372 0.888 0.672 1061 7 -3.313 3.239 -1.080 -0.735 1.090 0.696 -0.088 0.890 0.669 918 Table 6.06 Correlation matrix for training set 712" (712")' Zotz" ZP 2" Zœ " X Zpz" Vx R: 0.833 0.737 0.311 0.587 0.374 0.711 712" 0.948 0.472 0.780 0.518 0.772 (7[2")' 0.496 0.779 0.605 0.739 Za2" 0.551 0.798 0.252 ZP2" 0.711 0.708 Zai" X ZPz" 0.368 The stepwise regression matrix indicates that (%")" is a redundant descriptor because of the negligible coefficient (-0.08) for this descriptor and also the minimal improvement in SD and r“ when it is introduced into the equation. This is due to an extremely high cornelation between 7T2" and (7X2”)“ which can be seen in the correlation matrix. 182 Overall, the correlations are good considering the solubility data were taken from the work of many investigators whose results were obtained by different techniques on compounds of various degrees of purity. Based on the regression analyses, there is little to be gained by inclusion of the mp correction term. From a thermodynamic point of view, the mp term is required to properly describe the solubility of crystalline compounds. However, elimination of this term may be construed as an attempt to predict the compound mp from structural information, thereby replacing the (mp-25) term with a linear combination of molecular descriptors. Thus the equations with the cross-term are significantly better than those without this term, and eqn 20 is generally more accurate in prediction than eqn 21. However, the practical advantages of eqn 21 quite outweigh the better fit of eqn 20; in any case, eqn 21 is reasonable considering the complexity of compound type in the training set. Since the Yalkowsky equation is considered to be a reliable method for solubility prediction, it is of interest to extend the equation and incorporate more experimental data to investigate whether this equation may be improved. The result is eqn 22, log Sw = 0.455 - 1.020 log Poet - 0.009(mp-25) (22) n = 540, = 0.868, SD = 0.709, F = 1760 It is also possible to obtain eqn 23, with log Poet as the only term. log Sw = -0.184 - 0.996 log Poet (23) n = 570, = 0.725, SD = 1.014, F =1538 Isnard and Lambert suggested that the log Poet parameter alone would give adequately accurate solubility predictions because according to their correlations, only a minor improvement in SD was observed when a mp term was introduced, see eqns 24 and 25. log Sw = 1.17 - 1.38 log Poet (24) n = 300, r^ = 0.931, SD = 0.665 log Sw = 1.00 - 1.26 log Poet - 0.0054(mp-25) (25) n = 300, SD = 0.582 However, eqns 22 and 23 obtained from this work contradict this suggestion. 183 6.3.1 Test Set Results : Comparison with Other Methods The predictive capability of eqn 21 can be probed through the test set of 268 (91 liquids, 177 solid) compounds. The SD value for the 268 compound test set is 0.667 log units, AAE = 0.530, and AE = 0.083, which can be taken as an estimate of the predictive power of eqn 21. The Abraham method can be compared with llitYalkowsky equation, eqns 22, 23, 24 and 25, and also commercially available packages for estimating water solubility. Two software packages were used; WsKow for Windows, version 1.26, (Syracuse Research Corporation, USA) is based on the atom/fragment contribution method of Meylan and Howard^^ and the Toolkit for estimating Physicochemical Properties of Organic Compounds (TPOC), version 1.0 (JohnWiley and Sons, Inc) uses the group contribution method of Klopman^ to predict aqueous solubility. The observed and calculated log Sw values for eqn 21, the WsKow program (without melting point term), and both models of the Klopman method are given in Table 6.14 (pages 216-224). Prediction statistics for each of the models studied is shown in Table 6.07. The results for the test set of 268 compounds indicate that the Abraham General Solvation Equation is comparable to the Meylan method in terms of solubility prediction accuracy. These two methods are significantly more accurate in prediction than the other models in Table 6.07. Although both Klopman models were reliable for the simple compounds, problems arose when trying to deal with more complicated drug-like compounds and pesticides. Predictions were specifically poor for 21 molecules containing several polar functional groups, indicating that the model is deficient in being able to account for interactions between groups and that the groups had not previously been well-defined. There were also a large number of compounds for which prediction could not be made due to missing fragments. Model I of Klopman was not able to predict solubility for 81 compounds due to missing fragments, but in the case of Model n, this number was reduced to 8 compounds. The results confirm that Model n has a much wider applicability than Model 1. Nevertheless, the parameter set needs to be increased yet further if the prediction results for multifunctional groups is to be improved. Although the Yalkowsky equation was derived from only 155 compounds, it is shown to be robust due to the similarity in coefficients with eqn 22 obtained from 540 compounds. The Yalkowsky equation performed only slightly worse than eqn 22 in 184 terms of prediction SD, but displayed greater bias in that prediction was on average 0.33 log units lower than the experimental value. The Yalkowsky equation performed reasonably well if the log Poet data were experimentally derived. If calculated log Poet values are used, there is a decline in model accuracy which is of little surprise considering that log Poet can be calculated to ~0.41og units; this will invariably contribute to the overall error in solubility prediction. For this set of compounds, the WsKow program performed better than the ClogP program in calculating log Poet. In addition to being slightly more accurate, WsKow did not suffer from the problem of missing fragments which prevented ClogP from calculating values for three compounds. The experimental log Poet and calculated log Poet values by both the ClogP and WsKow programs are also included in Table 6.14. Table 6.07 Comparison of methods using test set compounds Prediction Method n SD AAE AE No melting point Abraham" 268 0.667 0.534 0.083 Meylan* 268 0.692 0.503 -0.009 Klopman (Model 1/ 187 0.897 0.602 -0.006 Klopman (Model lY 260 0.806 0.707 -0.144 log Poet, eqn 23"" 209 1.037 0.791 0.168 Log Poet - ClogP^ 265 1.170 0.890 0.192 Log Poet - WsKow^ 268 1.093 0.848 0.189 Isnard and Lambert, eqn 24* 209 1.325 0.986 0.497 With melting point Abraham' 237 0.621 0.484 0.080 Meylan^ 237 0.641 0.448 -0.043 Modified Yalkowsky, eqn 22* 194 0.729 0.550 0.103 Yalkowsky' 194 0.745 0.659 0.333 Yalkowsky (ClogP)'” 237 0.866 0.725 0.387 Yalkowsky (WsKow)" 237 0.803 0.705 0.372 Isnard and Lambert, eqn 25" 194 0.859 0.663 0.211 AAE: average absolute error, AAE=Z|ca!c-obs|/n., AE: average error, AE=Z(caIc-obs)/n. " Eqn 21 * WsKow ^ TPOC, 61 parameter model ‘^TPOC, 33 parameter model *Eqn 23 ^Eqn 23, log Poet calculated using ClogP for Windows, version 2.0.0b, Biobyte, CA. * Eqn 23, log Poet calculated using WsKow ^ Eqn 9 ' Eqn 20 ^ WsKow * Eqn 22 ^ Eqn 3 Eqn 3, log Poet calculated using ClogP "Eqn 3, log Poet calculated using WsKow ° Eqn 25. 185 There is no agreement with Isnard and Lambert who suggested that a single parameter of log Poet can sufficiently be employed to predict solubility. It is quite conclusive from this work that mp vastly improves the statistics of prediction when used in conjunction with log Poet. Although the Isnard and Lambert dataset used to derive eqns 24 and 25 was reasonably large, it was not representative of a diverse set of compounds. The choice of compounds was in fact biased towards non-polar derivatives with low melting points which was the reason for the mp term being somewhat insignificant in eqn 25. In this test set, 26% of compounds have a mp>I50°C and it would be expected that large deviations would occur if eqn 24 is applied to highly polar compounds, and eqn 25 is applied to high mp and/or polar compounds. This is observed and the statistics of prediction for both eqn 24 and 25 are poor in comparsion to the other methods. This is an example of the importance in using a training set which is representative of a diverse set of compounds in order that the method be ‘universally applicable’. 186 6.3.2 Final equations The training set and test set can be combined to obtain eqn 26 for 1071 compounds and eqn 27 which contains an additional mp term for 944 compounds (419 liquids and 516 solids). log Sw = 0,394 - 0.954 Ra +0.318 712“ + 1.157 Z ai" + 3.255 Z k " -0,786 Z ai" X ZP 2" - 3.329 V* (26) n=1071, = 0.888, SD = 0.671, F =1401, AAE = 0.526 log Sw = 0.368 -0.711 R2 +0.407 712" + 1.730 Z a 2* +3.383 ZP 2" -1.036 Zctz" X ZP2” - 3.493 V, - 0.005 (mp - 25) (27) n = 944, = 0.908, SD = 0.615, F =1324, AAE = 0.469 Eqn 26 is considered to be the best equation constructed from the Abraham General Solvation descriptors, and it may be concluded that an amended version of eqn 16, containing the extra x Zpz" term, can correlate and predict log Sw values simply from compound structure to around 0.67 log units. For eqn 26, the t-statistics have been calculated as follows : R 2 (-20.2), Ttz" (5.3), (11.7), Zgz" (46.5), x ZP2” (-15.5) and Vx (-61.7). The standard errors of the coefficients are on average, 0.06. The result indicates that the solute size and hydrogen-bonding terms are the dominant factors influencing compound solubility in water. This is consistent with the stepwise regression carried out for the training set in which 712” terms were found to be the least statistically significant. The availability of an experimental mp will generally improve prediction through the use of eqn 27. However, it should be stressed that the levels of prediction accuracy mentioned will only extend to compounds which are within the scope of the descriptor space covered by the total data set in this work; extrapolation will result in increasingly poorer predictions. Table 6.08. shows the minimum and maximum descriptor values used, and also compound structure as a simple indication of the type of multifunctional compounds which give rise to the maximum values. 187 Table 6.08 Descriptor space covered by eqn 27 Parameter Min value Max value R2 -0.06 4.77 ( diisopropylether) (coronene) 712 0 4.20 (alkanes) (sparsomycin) HOv ,OH 0 2.72 OH (compounds with no (morin) H -bond acidity) OH HO -OH 0 5.05 (compounds with no ( raffinose) HO' OH H-bond basicity) OH OH HO Zaz"xZp2 (compounds for which 11.11 [Zaz" = 2.20, zPz = 5.05] = 0 or =0) ( raffinose) 0.2495 (methane) (etoposide) Cl ^ Cl Q, Mp 25°C 485°C (all liquids) (mirex) - 188 6.3.3 Influence of High and Low Solubility Compounds There are particular experimental difficulties with regard to compounds that have very low solubilities. In order to ascertain if such compounds were exerting any undue influence on the regression, the correlation was re-run leaving out the very insoluble compounds. The very soluble compounds were also left out separately, and finally both the very insoluble and very soluble compounds were ommited. A summary of the resulting equations is given in Table 6.09, where the coefficient of the product term X is denoted as ‘k’. By comparison wi-thecj^n21 and eqn 26, changes in the regression coefficients are not very pronounced, and so there is little disrupting effect of compounds with very low or very high solubilities. Since the intrinsic solubility of a substance depends on the particular solid phase, more important effects probably arise when the solid in equilibrium with the saturated solution is a hydrate, because the solubility of the hydrate will not be the same as the unhydrated solid, to which all the correlation equations refer. The physical form of the solid can also have a profound effect and solubility will vary depending on whether the compound is amorphous or crystalline, and will also differ between polymorphs of the same compound. Since the Abraham descriptors will be identical for polymorphs of the same compound, it is essential that eqn 20 is used when predicting solubility for different crystal lattices. Polymorphs have different activities which will be reflected in their different enthalpies of fusion and different mps. At a given temperature, the polymorph with higher activity has lower stability and higher solubility. In addition, racemic compounds are typically more soluble than their pure enantiomers. For example, S-ketoprofen and racemic ketoprofen were found to have aqueous solubility of 2.32g and I.42g L'^ respectively The reason for the difference being attributed to the difference in intercrystalline forces which resulted in a lower melting point (by 2I.7K) of S-ketoprofen in comparison wdhits racemate. 189 Table 6.09 Correlation equations using the amended equation with the Z 0C2" x term, without compounds of low and high solubilities^ Coefficients r s a b k V c r' SD n AAE condition*’ -0.922 0.295 1.210 3.129 -0.757 -3.202 0.217 0.872 0.643 993 0.516 (1) -0.866 0.266 1.105 3.180 -0.760 -3.277 0.373 0.849 0.654 1003 0.516 (2) -0.803 0.224 1.129 2.985 -0.711 -3.088 0.147 0.818 0.618 925 0.495 (3) -1.015 0.370 1.188 3.296 -0.827 -3.343 0.412 0.889 0.672 803 0.530 (4) -0.954 0.318 1.157 3.255 -0.786 -3.329 0.394 0.888 0.671 1071 0.526 (5) ^ All equations are without any mp correction term. (1) Omit very soluble compounds (log Sw>0) (2) Omit very insoluble compounds (log Sw<-6) (3) Omit both very soluble and very insoluble compounds. (4) Eqn 21 (5) Eqn 26 190 6.3.4 The Factors that Influence Aqueous Solubility Unlike most regression equations for log Sw, eqn 21 and eqn 26 can be interpreted to show the compound physicochemical properties that influence aqueous solubility. Most studies of aqueous solubility in which correlations are constructed without any correction term for solids fail to mention why a correction term is not required. Neither eqn 19 nor 26 include a solid correction term, and this may be reasoned as follows. The two main properties that lead to an increase in solubility are hydrogen-bond acidity and hydrogen- bond basicity; these no doubt reflect the strong hydrogen-bond basicity and strong hydrogen-bond acidity of water as a bulk solvent.^^’^^ However, if the compound is itself both a hydrogen-bond acid and a hydrogen-bond base, then the intermolecular hydrogen- bond interactions will lead to an increase in mp. Since the increase in intermolecular interactions in the crystalline solid will be stronger than the increase in solute-water interactions in solution, this will lead to a diminution in aqueous solubility. Furthermore, the descriptors will also account for intramolecular hydrogen-bonding that may occur between neighbouring groups within the same molecule, and which will reduce both the extent of intermolecular hydrogen-bonding and hence mp. Thus the product term, Eotz^xEpz" takes the place (at least partly) of a solid correction term. Now all hydrogen-bond acids, with the exception of carbon acids, are also hydrogen-bond bases, so that the effect of hydrogen-bonding on solubility will be a resultant of the two single terms and the product term, as shown in Table 6.10 for some representative acids. Table 6.10 Hydrogen-bond Effects on Solubility, as log Sw Compound 1 .157% " 3 .2 5 5 SP2" -0.786%" xZPz" Resultant Acetic acid 0.71 1.43 -0.21 1.93 Trichloroacetic acid 1.10 0.91 -0.21 1.80 Benzoic acid 0.68 1.30 -0.19 1.80 Phenol 0.69 0.98 -0.14 1.53 4-Nitrophenol 0.95 0.85 -0.17 1.63 Ethanol 0.43 1.56 -0.14 1.85 TEE 0.66 0.81 -0.11 1.36 Estratriol 1.62 3.97 . -1.34 4.25 191 Aniline 0.30 1.33 -0.08 1:55 Benzamide 0.57 2.18 -0.26 2.49 Pyrazole 0.62 1.46 -0.19 1.90 Morpholine 0.07 2.96 -0.04 2.99 Progesterone 0.00 3.71 0.00 3.71 Trichloromethane 0.17 0.07 0.00 0.24 It is quite clear that the net result of the presence of hydrogen-bond acid and hydrogen-bond base groups will increase solubility. Although the acid-base interaction in a solid or liquid, given by the S a 2^xZP2^ term, will reduce the hydrogen-bond effect, a positive resultant always remains. For the large number of compounds that are hydrogen-bond bases, but not acids, there is a straightforward effect of increased solubility, as also shown in Table 6.10. As mentioned, there are but few compounds that are hydrogen-bond acids and yet have no or very little hydrogen- bond basicity. Again, there will be virtually no cross-term, and all the effect of hydrogen-bond acidity will be towards an increase in solubility, as shown for trichloromethane. The single terms in the descriptors Z(X 2^ and ZP2", both lead to an increase in solubility, and are slightly smaller in magnitude than those for the solvation equation for the solubility of gases and vapours in water. The other ‘polar’ term in eqn 26 is s.tc 2^ that leads also to an increase in solubility. The product term as previously mentioned is not significant, because it is very well correlated with 712^ itself (r^ = 0.90). However, the coefficient of 712^ in eqn 26 is very much less than for the solubility of gases and vapours, so that dipolar effects within the solid or liquid counteract to some extent the solute/water effects that lead to increased solubility. Two other terms in eqn 26, rR 2 and vVx, both result in a decrease in solubility, and both the r- and v-coefficients in eqn 26 are markedly more negative than in the solvation equation for gaseous solubility. The R 2 descriptor refers to the propensity of a solute to interact with surrounding a and n electrons, the negative r-coefficient suggesting that such interaction within the solid or liquid is much larger than the corresponding interaction between the solute and bulk water. Although the Vx descriptor refers to the size of the solute, the vVx term for the solubility of gaseous 192 solutes will be the resultant of two opposing effects, (i) a cavity effect that arises from the disruption of solvent-solvent interactions and will lead to a negative coefficient, and (ii) a general solute-solvent dispersion interaction that will lead to a positive coefficient. For the solubility of gaseous solutes in water, the v-coefficient is negative, -0.869, so that the unfavorable cavity effect dominates^^. Now in solids and liquids, part of the cohesive forces will be general dispersion interactions that help to hold the solid or liquid together. These interactions within the solid or liquid oppose solubility, and will lead to an even more negative v-coefficient in eqn 26, as observed (-3.329). Thus the sign and magnitude of the coefficients in eqn 26 can be interpreted in terms of known chemical interactions. Such interpretation, in turn, leads to information about the physicochemical factors that influence the aqueous solubility of solids and liquids. 6.4 The Solubility of Bronsted Acids and Bases Grant and Higuchi"^® have noted the effect of pH on the solubility of Bronsted acids and bases, and have given equations for the variation of solubility with Most studies on the correlation and prediction of solubility ignore this pH dependency; none of the studies in Tables 6.01 and 6.02 mention this problem at all. If a Bronsted acid, such as a carboxylic acid, is dissolved in water, the pH of the resulting solution will depend both on the acid pKa and on the total concentration of the acid in solution. For a given acid, the greater the concentration, the lower will be the pH, and the larger will be the proportion of the neutral species. Hence for acids with the same pKa, the pH of the saturated solution will decrease as the solubility increases, so that for acids that are quite soluble the proportion of neutral species will be larger than for acids that are sparingly soluble. Eqn 26, and, indeed, all the other correlation equations in Tables 6.01 and 6.02 refer to the solubility of the neutral species, N, so that the predicted (neutral) solubility will be less than the observed solubility, T or Sw, the difference depending on the acid pKa value, and the actual solubility. In Fig. 6.1 the calculated values of N/T, the fraction of the neutral species, for a series 193 of acids of pKa 3, 4.37, and 5 as a function of the observed total solubility, as log Sw (log T) are given. For an acid with a pKa of 4.37 or 5, N/T is larger than about 0.5 even down to log Sw values of -4. Now an error of a factor of 0.5 (or 2.0) corresponds to an error of 0.3 log units, and is not very important in the context of SD values of 0.5 log units. However, for very insoluble acids, with log Sw of -5 or -6, errors of one or two log units will arise if no consideration is given to ionization of Bronsted acids. For stronger acids with pKa = 3, large errors will arise at log Sw values less than around - 3.5 units. The pKa value at 4.37 was chosen because this is the pKa of p-toluic acid, studied in considerable detail by Strong and cow orkers.Their determined N/T value for p-toluic acid in the saturated solution at 25°C is shown in Fig. 6.1; the calculated value from this work is in excellent agreement. Fig. 6.1 Values of N/T for Bronsted acids as a function of the total solubility, log Sw- acids pKa values are 3 (□), 4.37 (•) and 5 (0), calculated value (a ) for p-toluic acid. 1.0 0.8 0.6 0.4 0.2 0.0 •7 ■5 •3 1 log Sw A similar ionization phenomenon occurs in the solubility of Bronsted bases; many drug molecules, of course, are strong Bronsted bases, with pKa of the conjugate acid from 8-10. In Fig. 6.2 the calculated N/T values for three series of bases with pKa = 8, 194 9, and 10 as a function of the observed solubility, log Sw are plotted. If it is considered that substantial errors in predicted values arise when N/T is less than about 0.5, this will be the case for bases with log Sw < -4 (pKa 10), < -5 (pKa 9) and < -6 (pKa 8). Fig. 6.2 Values of N/T for Bronsted bases as a function of the total solubility, log Sw; pKa values are 8 (□), 9 (•) and 10 (0) 0.8 - 0.6 - I 0.4 - 0.2 - 0.0 •8■6 -4 02 log Sw Observed solubilities may be compared with those calculated on eqn 26 for the neutral species via eqn 26, for Bronsted acids and bases in the data set. In Table 6.11 are given values for strong Bronsted acids, that is with pKa values less than 4. In none of the cases is the value of N/T less than 0.5 for the saturated solution. The lowest value of N/T is for p-bromobenzoic acid which has N/T of 0.549; this would make the observed solubility some 0.26 log units more than calculated. For all the other acids in Table 6.11, the difference will be even less. Inspection of Table 6.11 shows that for two acids the calculated - observed log Sw values are -1.31 and -1.56 units, so that other interfering factors are far more important than ionization, at least as regards the acids presented in Table 6.11. 195 Table 6.11 Observed and calculated log Sw values for strong Bronsted acids, pKa < 4 Acid obs calc calc-obs pKa Trichloroacetic acid 0.600 -0.713 -1.313 0.65 o-Aminobenzoic acid -1.520 -1.474 0.046 2.11 Chloroacetic acid 1.810 0.246 -1.564 2.82 m-Bromobenzoic acid -2.276 -1.776 0.500 2.85 o-Chlorobenzoic acid -1.890 -1.641 0.249 2.94 Salicylic acid -1.820 -1.594 0.226 2.98 p-Nitrobenzoic acid -2.800 -1.680 1.120 3.42 m-Nitrobenzoic acid -1.680 -1.627 0.053 3.49 m-Chlorobenzoic acid -2.590 -2.039 0.551 3.87 o-Toluic acid -2.060 -1.946 0.114 3.95 p-Bromobenzoic acid -3.539 -2.458 1.081 3.97 p-Chlorobenzoic acid -3.310 -2.155 1.155 3.98 A similar table can be constructed for the strong bases, those with pKa >10, see Table 6.12. There is a general trend, with (calculated - observed) log Sw values always negative by 0.84 log units, on average. However, this cannot be accounted for by ionization since even the value of 0.80 for N/T for dibutylamine would make a difference of only 0.1 log units. Table 6.12 Observed and calculated log Sw values for strong Bronsted bases, pKa > 10 Base obs calc calc-obs pKa Octylamine -1.460 -2.025 -0.565 10.57 Hexylamine -0.250 -1.097 -0.847 10.64 Butylamine 0.960 -0.184 -1.144 10.66 Pentylamine 0.270 -0.641 -0.911 10.64 Heptylamine -0.900 -1.566 -0.666 10.66 Propylamine 1.520 0.284 -1.236 10.69 Ethylamine 2.060 0.742 -1.318 10.70 Triethyl amine -0.140 -0.591 -0.451 10.85 Diethylamine 1.030 0.068 -0.962 11.04 Dibutylamine -1.440 -1.764 -0.324 11.25 196 It is apparent, therefore, that only for very insoluble strong Bronsted acids and Bronsted bases will ionization lead to significant errors in calculation. However, it is worth pointing out that solubilities calculated through eqn 26, or by the methods summarized in Tables 6.01 and 6.02 (pages 175 and 176 respectively), refer to the solubility of the neutral species. For Bronsted acids and bases this will be the solubility in solutions of pH near to the compound pKa, see Figs. 6.1 and 6.2. The observed solubility is that at whatever pH the saturated solution is at. This observed solubility does not refer to any specific pH, but to a pH that has to be calculated from the observed (total) solubility and the compound pKa. If the solubility of a Bronsted acid or base is required at a given pH of 7, or 7.4, for example, then Figs. 6.1 and 6.2 can be used to obtain the correction factor N/T, at least if the difference in pH between the saturated solution and the given pH is not too large. 197 6.5 References 1. Ferguson, J. The use of chemical potentials as indices of toxicity. Proc.Roy.Soy.,London,Ser.B. 1939, 127,387-404. 2. Bobra, A.; Shiu, W.Y.; Mackay, D. Quantitative structure activity relationships for the acute toxicity of chlorobenzenes to Daphnia magna. Environ. Toxicol. Chem. 1985,4, 297-305. 3. Lipinski, C.A.; Lombardo, P.; Dominy, B.W.; Feeney, P.J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 1997,23, 3-25. 4. Bevan, C.D.; Lloyd, R.S. A high-throughput screening method for the determination of aqueous drug solubility using laser nephelometry in microtiter plates. Anal. Chem. 2000,72, 1781-1787. 5. Yalkowsky, S.H.; Banerjee, S. Aqueous Solubility. Methods of estimation for organic compounds. Dekker, New York, 1992. 6. Irmann, F. Eine einfache korrelation zwischen wasserloslichkeit und strukture von kohlenwasserstoffen und halogenkohlenwasserstoffen. Chem.Ing.Tech., 1965,37, 789-798. 7. Wakita, K.; Yoshimoto, M.; Miyamoto, S.; Watanabe, H. A method for calculation of aqueous solubility of organic compounds using new fragment solubility constants. Chem.Pharm.Bull., 1986, 34,4663-4681. 8. Klopman, G.; Wang, S.; Balthasar, D.M. Estimation of aqueous solubility of organic molecules by the Group Contribution approach. J.Chem.Inf.Comput.Sci., 1992, 32,474-482. 9. Kuhne, R.; Ebert, R.-U.; Kleint, F.; Schmidt, G.; Schuurmann, G. Group Contribution methods to estimate water solubility of organic compounds. Chemosphere 1995,3 0 ,2061-2077. 10. Nirmalakhanden, N.N.; Speece, R.E. Prediction of aqueous solubility of organic compounds based on molecular structure. 2. Application to PNAs, PCBs, PCDDs, etc. Environ.Sci.Tech., 1989,23,708- 713. 11. Banerjee, S. Calculation of water solubility of organic compounds with UNIFAC-derived parameters. Environ.ScLTechnoL, 1985,19,369-370. 12. Kan, A.T.; Tomson, M.B. UNIF AC prediction of aqueous solubilities of chemicals of environmental interest. Environ.Sci.Technol., 1996,30,1369-1376. 13. Myrdal, P.B.; Ward, G.H.; Dannenfelser, R.-M.; Mishra, D.; Yalkowsky, S.H. Aquafac 1: aqueous functional group activity coefficients; application to hydrocarbons. Chemosphere, 1992, 24, 1047- 1061. 14. Myrdal, P.B.; Ward, G.H.; Simamora, P.; Yalkowsky, S.H. Aquafac: aqueous functional group activity coefficients. SAR and QSAR in Environ. Res., 1993,1,53-61. 15. Myrdal, P.B.; Manka, A.M.; Yalkowsky, S.H. Aquafac 3: Aqueous functional group activity coefficients; application to the estimation of aqueous solubility. Chemosphere, 1995, 30,1619-1637. 16. Lee, Y. C.; Myrdal, P.B.; Yalkowsky, S.M. Aqueous functional group activity coefficients (Aquafac) 4: applications to complex organic compounds. Chemosphere, 1996, 33,2129-2144. 17. Lee,Y.-C.; Pinsuwan, S.; Yalkowsky, S.M. A comparison of AQUAFAC group q-values to the corresponding CLOGP f-values. Chemosphere, 1997, 35,775-782. 18. Pinsuwan, S.; Myrdal, P.B.; Lee, Y.-C.; Yalkowsky, S.H. AQUAFAC 5: aqueous functional group 198 activity coefficients; application to alcohols and acids.Chemosphere, 1997,35,2503-2513. 19. Bodor, N.; Harget, A.; Huang, M.-J. Neural network studies. 1. Estimation of the aqueous solubility of organic compounds. J.Am.Chem.Soc., 1991,113,9480-9483. 20. Bodor, N.; Huang, M.-J. A new method for the estimation of aqueous solubility of organic compounds. J.PhamuScL, 1992,81,954-960. 21. Nelson, T.M.; JursJP.C. Prediction of aqueous solubility of organic compounds. J. Chem. Inf. Comput. Sci., 1994, 34,601-609. 22. Sutter, J.M.; Jurs, P.C. Prediction of aqueous solubility for a diverse set of hereroatom-containing compounds using a quantitative structure-property relationship. Chem Inf. Comput.Sci., 1996, 36,100- 107. 23. Katritzky, A.R.; Wang, Y.; Sild, S.; Tamm, T.; Karelson, M. QSPR studies on vapour pressure, aqueous solubility and the prediction of water-air partition coefficients. J.Chem. Inf.Comput.Sci., 1998, 38,720-725 24. Huuskonen, J. Estimation of aqueous solubility for a diverse set of organic compounds based on molecular topology. J. Chem. Inf. Comput. Sci. 2000,40,773-777. 25. Hansch, C.; Quinlan, J.E.; Lawrence, G.L. The linear free energy relationship between partition coefficients and the aqueous solubility of organic liquids. J.Org.Chem. 1968, 33, 347-350. 26. Valvani, S.C.; Yalkowsky, S.H.; Roseman, T.J. Solubility and partitioning IV: solubility and octanol- water partition coefficients of liquid nonelectrolytes. J.Pharm.ScL, 1981,70,502- 507. 27. Yalkowsky, S.H.; Valvani, S.C. Solubility and partitioning. I: solubility of nonelectrolytes in water. J.Pharm.ScL, 1980, 69,912-922. 28. Yalkowsky, S.H.; Valvani, S.C.; Roseman, T.J. Solubility and partitioning VI: octanol solubility and octanol-water partition coefficients. J.Pharm.Sci., 1983,72, 866-870. 29. Meylan, W.M.; Howard, P.H.; Boethling, R.S. Improved method for estimating water solubility from octanol/water partition coefficient. Environ. Tox. Chem. 1996, 15, 100-106. 30. Ruelle, P; Kesselring, U.W. The hydrophobic effect. 1. A consequence of the mobile order in H- bonded liquids. J.Pharm.Sci., 1998, 87,987-997. 31. Ruelle, P; Kesselring, U.W. The hydrophobic effect. 2. Relative importance of the hydrophobic effect on the solubility of hydrophobes and pharmaceuticals in H-bonded solvents. J.PharmScL, 1998, 87,998-1014. 32. Ruelle, P; Kesselring, U.W. The hydrophobic effect. 3. A key ingredient in predicting n-octanol- water partition coefficients. J.Pharm.ScL, 1998, 87, 1015-1024. 33. Kishi, H.; Hashimoto, Y. Evaluation of the procedures for the measurement of water solubility and n- octanol/water partition coefficients of chemicals. Results of a ring test in Japan. Chemosphere 1989, 18,1749-1759. 34. Abraham, M.H. Thermodynamics of solution of homologous series of solutes in water. J.Chem.Soc., Faraday Trans. 1984,1, 80, 153-181. 35. Kamlet, M.J.; Doherty, R.M.; Abraham, M.H.; Carr, P.W.; Doherty, R.F.; Taft, R.W. Linear solvation energy relationships. 41. Important differences between aqueous solubility relationships for aliphatic and aromatic solutes. J.Phys.Chern., 1987, 91, 1996-2004. 199 36. Isnard, P.; Lambert, S. Aqueous solubility and n-octanol/water parttion coefficient correlations. Chemosphere 1989, 18, 1837-1853. 37. Kommuru, T.R.; Khan, M.A.; Reddy, I.K. Racemate and enantiomers of Ketoprofen: Phase diagram, thermodynamic studies, skin permeability, and use of chiral permeation enhancers. J. Pharm. Sci. 1998, 87, 833-840. 38. Abraham, M.H.; Chadha, H.S.; Whiting, G.S.; Mitchell, R.C. Hydrogen bonding. 32. An analysis of water-octanol and water-alkane partitioning and the AlogP parameter of Seiler. J.Fharm.Sci., 1994, 8 3 ,1085-1100. 39. Abraham, M.H.; Andonian-Haftvan, J.; Whiting, G.S.; Leo, A.; Taft, R.W. Hydrogen bonding. Part 34: the factors that influence the solubility of gases and vapours in water at 298 K, and a new method for its determination. J.Chem.Soc., Perkin Trans.2, 1994, 2,1777-1791. 40. Grant, D.G.W.; Higuchi, T. Solubility Behaviour of Organic Compounds, John Wiley & Sons, New York, 1990, pp 486-488. 41. Strong, L.E.; Neff, R.M.; Whitesel, I. Thermodynamics of dissolving and solvation for benzoic acid and the toluic acids in aqueous solution. J.Soln.Chem., 1989,18, 101-114. 42. Brunas, J.P. The aqueous solubility of drugs and other compounds. University College London. Third Year Project 1995. 43. Calculated from the gas-water partition coefficient and the saturated vapor concentration, this work. 44. Ruelle, P.; Kesselring, W. The hydrophobic propensity of water toward amphiprotic solutes: Prediction and molecular origin of the aqueous solubility of aliphatic alcohols. J.Pharm.Sci. 1997, 86, 179-186. 45. Schwarzenbach, R.P.; Gschwend, P.M.; Imboden, D.M. Environmental Organic Chemistry. John Wiley-Interscience, New York, 1993, p.618-625. 46. Pinsuwan, S.; Li, A.; Yalkowsky, S. Correlation of octanol/water solubility ratios and partition coefficients. J.Chem.Eng.Data. 1995,4 0 , 623-626. 47. Huuskonen, J; Salo, M; Taskinen, J. Neural network modelling for estimation of the aqueous solubility of structurally related drugs. J.Pharm.Sci. 1997, 8 6 ,450-454. 48. Siedell, A. Solubilities of organic compounds : A compilation of quantitative solubility data from the periodical literature. Third Edition (fourth printing). Volume 2. D. Van Nostrand Company, inc. 1941, p.777. 49. The Merck Index : An encyclopedia of chemicals, drugs, and biologicals. Eleventh edition. Merck & Co., inc. 1989. 50. Prankerd, R.J.; McKeown, R.H. Physico-chemical properties of barbituric acid derivatives : IV. Solubilities of 5,5-disubstituted barbituric acids in water. Int.J.Pharm. 1994,112,1-15. 51. Vaution, C; Treiner, C; Puisieux, F; Cartensen, J.T. Solubility behaviour of barbituric acids in aqueous solution of sodium alkyl sulfonate as a function of concentration and temperature. J.Pharm-Sci. 1981, 70, 1238-1242. 52. The Pesticide Manual. Eleventh edition. Editor : Tomlin, C.D.S. British Crop Protection Council 1997. 53. Ruelle, P; Kesselring, W. Aqueous solubility prediction of environmentally important chemicals from 200 the mobile order thermodynamics. Chemosphere 1997,3 4 , 275-298. 54. Stella, V.J.; Martodihardjo, S.; Terada, K.; Rao, V.M Some relationships between the physical properties of various 3-acyloxymethyl prodrugs of phenytoin to structure: Potential in vivo performance implications. J.PharmSci. 1998, 87, 1235-1241 55. Miller, D.J.; Hawthorne, S.B. Solubility of liquid organic flavour and fragrance compounds in subcritical (hot/liquid) water from 298K to 473K. J. Chem. Eng. Data 2000,45, 315-318. 56. Y-C. Lee, Myrdal, P.B.; Yalkowsky SH. Aqueous functional group activity coefficients (AQUAFAC 4): Applications to complex organic compounds. Chemosphere 1996, 33, 2129- 2144. 57. Nicolaide, E.; Galia, E.; Efthymiopoulos, C.; Dressman, J.B.; Reppas, C. Forecasting the in vivo performance of four low solubility drugs from their in vitro dissolution data. Pharm. Res. 1999, 16, 1876-1882. 58. Florey, K (Ed) Analytical profiles of drug substances. Academic Press, New York. 59. SRC The Physical Properties Database (PHYSPROP) available via chemfinder.com 60. Huuskonen, J.; Salo, M.; Taskinen J. Aqueous solubility prediction of drugs based on molecular topology and neural network modelling. J. Chem. Inf. Comput. Sci. 1998, 38,450-456. 61. Vandenbroek, L.; Lazaro, E.; Zylich, Z.; Fennis, P.J.; Missler, F.; Lelieveld, P.; Garzotto, M.; Wagener, D.J.T.; Ballesta, J.P.G.; Ottenheim, H.C.J. J.Med.Chem. 1989, 32, 2002-20015 62. Agricultural Research Service, Pesticide Properties database, remote Sensing and Modelling Laboratory, Beltville, Maryland, USA. 63. Shah, J.C.; Chen, JR.; Chow, D. Preformulation study of etoposide - identification of physicochemical characteristics responsible for the low an erratic oral bioavailability of etoposide. Pharm. Res. 1989, 6, 408-412 64. Mitchell, B.E.; Jurs, P.C. Prediction of aqueous solubility of organic compounds from molecular structure. J. Chem. Inf. Comput. Sci. 1998, 38, 489-496. 65. Aldrich catalogue : Handbook of fine chemicals. Sigma-Aldrich Co. USA. 1998 201 6.6 Data Tables Table 6.13 Training set data for log Sw regressions Compound log Sw ref log Sw calc- obs Mp (°C) ref log exp calc® Poet*’ Methane -0.90 34 -0.42 0.48 -184.3 5f) 1.09 Ethane -1.36 34 -0.89 0.47 -172.0 65 1.81 Propane -1.94 34 -1.36 0.58 -187.0 59 2.36 2-Methylpropane -2.55 42 -1.84 0.71 -160.0 65 2.76 Pentane -3.18 9 -2.31 0.87 -130.0 65 3.39 2-Methylbutane -3.18 S3 -2.31 0.87 -159.9 59 2-Methylpentane -3.74 9 -2.78 0.96 -154.0 65 3.73 3-Methylpentane -3.68 9 -2.78 0.90 -118.0 59 2,2-Dimethylbutane -3.55 9 -2.78 0.77 -100.0 65 Heptane -4.53 9 -3.25 1.28 -91.0 65 4.50 2,2-Dimethylpentane -4.36 53 -3.25 1.11 -124.0 59 2,3-Dimethylpentane -4.28 53 -3.25 1.03 <25 59 3,3-Dimethylpentane -4.23 53 -3.25 0.98 -134.0 59 2,2,3-TrimethyIbutane -4.36 53 -3.25 1.11 -25.0 59 Octane -5.24 9 -3.72 1.52 -57.0 65 5.18 3-Methylheptane -5.16 53 -3.72 1.44 -121.0 59 2,2,4-Trimethylpentane -4.74 9 -3.72 1.02 -107.0 59 2,3,4-Trimethylpentane -4.80 53 -3.72 1.08 -109.0 59 2,2,5-Trimethylhexane -5.05 53 -4.19 0.86 -106.0 59 Tetradecane -7.96 53 -6.55 1.41 5.5 65 Hexadecane -8.40 53 -7.49 0.91 18.2 59 Methylcyclopentane -3.30 9 -2.61 0.69 -142.0 65 3.37 Propylcyclopentane -4.74 53 -3.55 1.19 -117.0 59 - Pentylcyclopentane -6.08 53 -4.48 1.60 -83.0 59 Methylcyclohexane -3.85 9 -3.11 0.74 -126.0 65 3.88 cis-l,2-Dimethylcyclohexane -4.30 53 -3.60 0.70 -50.0 59 trans-l,4-Dimethylcyclohexane -4.47 53 -3.52 0.95 -37.0 59 Cycloheptane -3.51 53 -3.20 0.31 -12.0 65 4.00 Cyclooctane -4.15 53 -3.74 0.41 11.5 65 Decal in -5.19 53 -4.35 0.84 -31.0 59 Propylene -1.08 34 -1.06 0.02 -185.0 65 1.77 1-Butene -1.94 34 -1.53 0.41 -185.0 65 2.40 2-Methylpropene -2.33 20 -1.53 0.80 -140.0 65 2.34 cis-2-Pentene -2.54 42 -2.05 0.49 -180.0 65 2.78 trans-2-Pentene -2.54 9 -2.03 0.51 -140.0 59 2-Methyl-1-Butene -2.73 53 -2.02 0.71 -138.0 59 2-Methy-2-Butene -2.56 53 -2.06 0.50 -134.0 65 2.67 1-Hexene -3.23 9 -2.45 0.78 -139.8 65 3.39 2-Methyl-1-Pentene -3.03 53 -2.46 0.57 -136.0 65 trans-2-Heptene -3.82 9 -2.96 0.86 -109.5 65 3.84 1-Octene -4.44 9 -3.41 1.03 -101.0 65 4.57 1-Nonene -5.05 9 -3.88 1.17 -81.0 65 5.15 1,3-Butadiene -1.87 53 -1.46 0.41 -109.0 65 1.99 2-Methyl-l,3-Butadiene -2.03 9 -1.92 0.11 -146.0 65 2,3-Dimethyl-1,3-Butadiene -2.40 53 -2.30 0.10 -76.0 59 1,5-Hexadiene -2.68 9 -2.28 0.40 -141.0 65 2.87 Cyclopentene -2.10 9 -1.73 0.37 -94.0 65 Cyclohexene -2.59 53 -2.27 0.32 -104.0 65 2.86 Cycloheptene -3.18 53 -2.75 0.43 -56.0 59 1,4-Cyclohexadiene -2.06 53 -1.94 0.12 -49.2 59 2.30 Ethyne 0.29 21 -0.37 -0.66 -80.8 59 0.37 1-Butyne -1.24 53 -1.17 0.07 -125.7 59 1 -Pentyne -1.64 9 -1.58 0.06 -105.5 65 1.98 1-Hexyne -2.36 9 -2.07 0.29 -132.0 65 2.73 202 Compound log Sw ref log Sw calc- obs Mp C O ref log exp calc® Poet" 1-Heptyne -3.01 53 -2.61 0.40 -81.0 65...... 3.45 1-Octyne -3.66 9 -3.08 0.58 -80.0 65 1-Nonyne -4.24 9 -3.54 0.70 -50.0 59 Trichloromelhane -1.17 9 -1.66 -0.49 -63.7 59 1.97 Tetrachloromethane -2.31 9 -2.38 -0.07 -23.0 65 2.83 Chloroe thane -1.06 53 -1.05 0.01 -139.0 65 1.43 1,2-Dichloroethane -1.06 9 -1.42 -0.36 -35.4 65 1.48 1,1,1 -Trichloroethane -2.00 9 -2.05 -0.05 -50.0 65 2.49 1,1,2-Trichloroethane -1.48 9 -1.81 -0.33 -37.0 65 2.07 1,1,1,2-Tetrachloroe thane -2.18 9 -2.47 -0.29 -70.2 65 2.66 Pentachloroethane -2.60 53 -2.96 -0.36 -29.0 59 3.22 Hexachloroethane -3.67 9 -3.76 -0.09 4.14 2-Chloropropane -1.41 9 -1.43 -0.02 -118.0 65 1.90 1,2-Dichloropropane -1.60 53 -1.81 -0.21 -100.0 65 1.99 1,3-Dichloropropane -1.62 9 -1.90 -0.28 -99.0 65 2.00 1 -Chloro-2-methylpropane -2.00 53 -1.91 0.09 -131.0 65 2.39 2-Chlorobutane -1.96 9 -1.91 0.05 -140.0 65 2.33 1-Chloropentane -2.73 9 -2.45 0.28 -99.0 65 1 -Chiorohexane -3.12 53 -2.91 0.21 -94.0 59 3.66 1-Chloroheptane -4.00 42 -3.38 0.62 -69.0 65 4.15 Chloroethylene -1.75 19 -1.11 0.64 -153.7 59 CIS 1,2-Dichloroethylene -1.30 9 -1.49 -0.19 -80.0 65 1.86 Trichloroethylene -1.96 9 -2.18 -0.22 -86.0 59 2.61 T etrach loroethy lene -2.54 9 -2.87 -0.33 -84.8 65 3.40 Bromomethane -0.79 53 -0.92 -0.13 -94.0 65 1.19 Dibromomethane -1.17 9 -1.71 -0.54 -52.0 65 1.88 T ribromomethane -1.91 9 -2.54 -0.63 8.3 59 2.67 Bromoethane -1.09 9 -1.31 -0.22 -119.0 65 1.61 1,2-Dibromoethane -1.68 9 -1.87 -0.19 9.0 65 1.96 1 -Bromopropane -1.73 9 -1.78 -0.05 -110.0 65 2.10 1-Bromobutane -2.37 9 -2.24 0.13 -112.0 65 2.75 1 -Bromo-2-methylpropane -2.43 9 -2.23 0.20 0.0 65 1-Bromopentane -3.08 42 -2.71 0.37 -95.0 59 3.37 1 -Bromoheptane -4.43 42 -3.64 0.79 -85.0 65 4.36 1-Bromooctane -5.06 42 -4.10 0.96 -55.0 65 4.89 lodomethane -1.00 9 -1.38 -0.38 -64.0 65 1.51 lodoethane -1.60 53 -1.76 -0.16 -108.0 65 2.00 1-Iodopropane -2.29 9 -2.23 0.06 -101.0 65 2-lodopropane -2.09 9 -2.17 -0.08 -90.0 59 2.89 1-Iodoheptane -4.81 42 -4.09 0.72 -48.0 65 4.70 Bromochloromethane -0.89 9 -1.46 -0.57 -88.0 65 1.41 Bromodichloromethane -1.54 9 -1.92 -0.38 -57.1 59 1 -Chioro-2-bromoethane -1.32 9 -1.80 -0.48 -16.6 59 1,1,2-Tri ch lorotri fl uoroethan e -3.04 9 -2.26 0.78 -35.0 65 3.16 1,2-Dichlorotetrafluoroethane -2.74 53 -2.10 0.64 -94.0 65 2.82 Dipropyl ether -1.62 9 -1.41 0.21 -123.0 59 2.03 Diisopropyl ether -1.10 9 -0.94 0.16 -85.5 59 1.52 Dibutyl ether -1.85 9 -2.34 -0.49 -95.0 59 3.21 Methyl butyl ether -0.99 9 -1.01 -0.02 -115.0 59 1.66 Methyl t-butyl ether -0.24 9 -0.41 -0.17 -109.0 59 0.94 Ethyl propyl ether -0.66 9 -0.93 -0.27 -127.0 59 Ethyl vinyl ether -0.85 9 -0.57 0.28 -116.0 59 1.04 Dimethoxymethane 0.48 9 0.32 -0.16 -105.0 59 0.18 1,1-Diethoxyethane -0.43 9 0.11 0.54 -100.0 59 0.84 1,2-Propylene oxide -0.59 9 0.25 0.84 -112.0 59 0.03 1,8-Cineole -1.74 55 -2.87 -1.13 35.6 59 2.50 Tetrahydrofurane 0.49 9 -0.19 -0.68 -108.4 59 0.47 Tetrahydropyran -0.03 9 -0.65 -0.62 -45.0 59 0.95 Propionaldéhyde 0.58 9 0.11 -0.47 -81.0 59 0.59 203 Compound log Sw ref log Sw calc- obs Mp (°C) ref log calc* Poet" Butyraldéhyde -0.01 '”T ~ ...... -0.35 -0.34 -96.0 0.88 Caproaldéhyde -1.30 9 -1.25 0.05 -56.0 59 1.78 2-Ethylbutanal -1.52 19 -1.26 0.26 liquid 59 2-Ethylhexanal -2.13 19 -2.21 -0.08 liquid 59 2-Ethyl-2-hexanal -2.46 19 -2.13 0.33 liquid 59 2-Butanone 0.52 9 -0.12 -0.64 -87.0 65 0.29 2-Pentanone -0.19 9 -0.57 -0.38 -78.0 65 0.91 3-Methyl-2-butanone -0.12 20 -0.57 -0.45 -92.0 59 0.84 2-Hexanone -0.80 9 -1.03 -0.23 -57.0 65 1.38 3-Hexanone -0.83 20 -1.04 -0.21 -55.5 59 4-MethyI-2-pentanone -0.74 9 -1.02 -0.28 -80.0 65 1.31 3,3-Dimethyl-2-butanone -0.72 20 -1.03 -0.31 -52.5 59 1.20 2-Heptanone -1.45 42 -1.49 -0.04 -31.0 59 1.98 2,4-Dimethyl-3 -pen tanone -1.30 20 -1.47 -0.17 -33.0 65 1.97 2-Octanone -2.05 42 -1.95 0.10 -16.0 65 2.37 2-Nonanone -2.58 9 -2.43 0.15 -21.0 65 3.14 2-Decanone -3.30 42 -2.89 0.41 14.0 65 3.77 Cyclohexanone -0.60 9 -0.71 -0.11 -47.0 65 0.81 Carvone -2.06 31 -2.79 -0.73 Menthone -2.35 31 -2.60 -0.25 -6.0 59 Methyl formate 0.58 9 0.17 -0.41 -100.0 65 0.03 Ethyl formate 0.15 9 -0.26 -0.41 -80.0 65 Isopropyl formate -0.63 20 -0.64 -0.01 <25 59 Butyl acetate -1.37 20 -1.19 0.18 -106.2 59 Isobutyl formate -1.01 9 -1.11 -0.10 -95.0 59 Methyl acetate 0.46 20 -0.04 -0.50 -98.0 65 0.18 Ethyl acetate -0.04 9 -0.48 -0.44 -84.0 65 0.73 Propyl acetate -0.72 9 -0.94 -0.22 -95.0 65 L24 Isobutyl acetate ■1.21 9 -1.32 -0.11 -99.0 65 1.78 Pentyl acetate -1.89 9 -1.86 0.03 -70.8 59 2.29 Isopentyl acetate -1.92 9 -1.79 0.13 -78.0 65 2.25 Ethyl propionate -0.66 9 -0.95 -0.29 -99.4 59 1.21 Methyl butyrate -0.82 9 -0.96 -0.14 -95.0 59 Ethyl butyrate -1.28 9 -1.40 -0.12 -135.4 59 Methyl pentanoate -1.36 9 -1.43 -0.07 -93.0 65 Ethyl pentanoate -1.75 9 -1.85 -0.10 liquid 59 Propyl propanoate -1.34 34 -1.39 -0.05 -75.9 59 1.96 Methyl hexanoate -1.87 34 -1.87 0.00 -71.0 59 Ethyl hexanoate -2.35 34 -2.31 0.04 liquid 59 Ethyl heptanoate -2.74 34 -2.77 -0.03 -86.8 59 Ethyl octanoate -3.39 34 -3.24 0.15 -47.0 59 Methyl nonanoate -3.38 34 -3.26 0.12 Ethyl nonanoate -3.80 34 -3.71 0.09 <25 59 Ethyl decan oate -4.10 34 -4.17 -0.07 liquid 59 Methyl acrylate -0.22 9 -0.57 -0.35 -75.0 65 0.80 Glyceryl triacetate -0.60 9 -0.14 0.46 3.0 59 0.25 Acetonitrile 0.26 9 0.27 0.01 -48.0 59 -0.34 Propionitrile 0.28 9 -0.04 -0.32 -93.0 65 0.16 Acrylonitrile 0.15 9 -0.02 -0.17 -83.0 65 0.25 Propylamine 1.52 19 0.32 -1.20 -83.0 65 0.47 Butylamine 0.96 19 -0.15 -1.11 -49.0 65 0.97 Penty lamine 0.27 19 -0.60 -0.87 -50.0 59 1.49 Heptylamine -0.90 19 -1.53 -0.63 -23.0 59 2.57 Octylamine -1.46 19 -1.99 -0.53 -3.0 65 Diethylamine 1.03 19 0.11 -0.92 -50.0 59 0.58 Dibutylamine -1.44 19 -1.73 -0.29 -61.0 65 2.83 Trimethylamine 1.32 19 0.44 -0.88 -117.0 65 0.16 Triethylamine -0.14 9 -0.55 -0.41 -115.0 65 1.45 Nitromethane 0.26 9 0.11 -0.15 -29.0 65 -0.35 204 Compound log Sw ref log Sw calc- obs Mp C O ref log exp calc' Poet" Nitroethane -0.22 9 -0.29 -0.07 -90.0 65 0.18 1-Nitropropane -0.80 9 -0.82 -0.02 -108.0 59 0.87 Chloropicrin -2.00 9 -2.09 -0.09 -64.0 65 2.09 Acetamide 1.58 9 1.31 -0.27 81.0 59 -1.26 N,N-Dimethylacetamide 1.11 9 0.47 -0.64 -20.0 65 -0.77 0-Ethyl carbamate 0.85 9 0.41 -0.44 49.0 9 -0.15 Acetic acid 2.00 9 0.78 -1.22 16.6 65 -0.17 Hexanoic acid -1.06 9 -1.01 0.05 -3.0 59 1.92 Methacrylic acid 0.00 9 -0.20 -0.20 16.0 65 0.93 Chloroacetic acid 1.81 9 0.32 -1.49 56.0 65 0.22 Trichloroacetic acid 0.60 9 -0.64 -1.24 57.5 65 1.33 Fu marie acid -1.28 42 0.10 1.38 300.0 65 . -0.34 Methanol 1.57 20 1.15 -0.42 -98.0 65 -0.77 Ethanol 1.10 20 0.69 -0.41 -130.0 65 -0.31 2-Propanol 0.43 9 0.44 0.01 -88.5 59 0.05 1-Butanol 0.00 9 -0.23 -0.23 -90.0 65 0.88 2-Methylpropan-1 -ol 0.10 42 -0.23 -0.33 -108.0 65 0.76 1-Pentanol -0.60 9 -0.69 -0.09 -78.0 65 1.56 2-Pentanol -0.29 9 -0.48 -0.19 -50.0 59 1.19 3-Pentanol -0.24 9 -0.51 -0.27 -69.0 59 1.21 3-Methylbutan-l-ol -0.51 21 -0.68 -0.17 -117.0 59 1.16 2-Methylbutan-2-ol 0.15 21 -0.40 -0.55 -12.0 59 0.89 3-Methyl-2-butanol -0.18 44 -0.49 -0.31 1.28 1-Hexanol -1.24 9 -1.16 0.08 -52.0 65 2.03 2-Hexanol -0.89 9 -0.95 -0.06 1.76 3-Hexanol -0.80 9 -0.96 -0.16 1.65 3-Methyl-2-pentanol -0.72 9 -1.17 -0.45 liquid 59 4-MethylpentanoI -1.14 44 -1.15 -0.01 2-Methyl-2-pentanol .0.49 44 -0.84 -0.35 -108.0 65 4-Methyl-2-pentanoI -0.80 9 -0.94 -0.14 -90.0 65 2-Methyl-3-pentanol -0.70 9 -0.98 -0.28 liquid 59 3-Methyl-3-pentanol -0.36 44 -0.88 -0.52 -24.0 59 2,2-Dimethyl-1 -butanol -1.04 44 -1.19 -0.15 <25 59 3,3-Dimethyl-1-butanol -0.50 9 -1.16 -0.66 3,3-Dimethyl-2-butanol -0.62 9 -0.97 -0.35 4.8 59 1.47 2-Heptanol -1.55 9 -1.42 0.13 2.31 3-Heptanol -1.47 9 -1.41 0.06 -70.0 59 2.24 4-HeptanoI -1.40 9 -1.41 -0.01 -42.0 59 2.22 3-Methyl-3-hexanol -0.98 44 -1.34 -0.36 3-Ethyl-3-pentanol -0.85 9 -1.38 -0.53 2,2-Dimethylpentanol -1.52 44 -1.66 -0.14 2,4-Dimethyl-3-pentanol -1.22 9 -1.46 -0.24 -70.0 59 1-Octanol -2.39 9 -2.09 0.30 -15.0 65 3.00 2-Octanol -2.09 9 -1.86 0.23 -38.6 59 2.90 2-Methyl-2-heptanol -1.72 9 -1.79 -0.07 -50.0 59 3-Methyl-3-heptanol -1.60 9 -1.81 -0.21 -83.0 59 2-Ethyl-l-hexanol -2.11 9 -2.11 0.00 -76.0 59 2-Nonanol -2.74 44 -2.34 0.40 -35.0 59 1 -Decanol -3.63 44 -3.02 0.61 6.0 59 4.57 2-Undecanol -2.94 44 -3.27 -0.33 2-3 65 I-Tetradecanol -5.84 44 -4.88 0.96 40.0 59 I-Pentadecanol -6.35 44 -5.34 1.01 46.0 59 1-Hexadecanol -7.00 44 -5.81 1.19 54-55 59 Cyclohexanol -0.44 9 -0.77 -0.33 21.0 65 1.23 Cycloheptanol -0.88 44 -1.26 -0.38 Cyclooctanol -1.29 44 -1.79 -0.50 14.5 65 I-Hexene-3-ol -0.59 44 -0.91 -0.32 Nerol -2.46 55 -2.80 -0.34 liquid 59 2-Butoxyethanol -0.42 9 -0.30 0.12 -75.0 65 0.83 Butanethiol -2.18 9 -1.85 0.33 -116.0 59 2.28 205 ref Compound log Sw ref log Sw calc- obs Mp (°C) - log exp calc" Poet” Dimethyl sulfide -0.45 45 -0.75 -0.30 -98.0 65 Diethyl sulfide -1.34 45 -1.56 -0.22 -103.9 31 1.95 Diisopropylsulfide -2.24 45 -2.32 -0.08 -106.0 59 2.84 Dimethyldisulfide -1.44 45 -1.61 -0.17 -85.0 65 1.77 Diethyldisulfide -2.42 45 -2.51 -0.09 Triethyl phosphate 0.43 9 -0.38 -0.81 -56.4 59 0.80 Benzene -1.64 9 -1.95 -0.31 5.5 65 2.13 Toluene -2.21 9 -2.41 -0.20 -93.0 65 2.73 o-Xylene -2.80 9 -2.86 -0.06 -25.2 65 3.12 m-Xylene -2.82 9 -2.84 -0.02 -47.9 65 3.20 p-Xylene -2.77 9 -2.83 -0.06 13.3 65 3.15 Isopropylbenzene -3.27 9 -3.30 -0.03 -96.0 59 3.66 1,2,3-Trimethylbenzene -3.20 9 -3.28 -0.08 -25.0 65 3.59 1,2,4-Trimethylbenzene -3.31 53 -3.25 0.06 -44.0 65 3.63 2-Ethyltoluene -3.21 53 -3.29 -0.08 -17.0 65 3.53 4-Ethyltoluene -3.11 53 -3.25 -0.14 -62.0 59 Butylbenzene -4.06 9 -3.79 0.27 -88.0 65 4.38 t-Butylbenzene -3.66 9 -3.72 -0.06 -58.0 65 4.11 1,2-Diethylbenzene -3.28 53 -3.77 -0.49 -31.0 65 3.72 1,4-Diethylbenzene -3.75 9 -3.74 0.01 -43.0 59 2-lsopropyltoluene -3.76 53 -3.72 0.04 -71.0 59 4-Isopropyltoluene -3.77 53 -3.67 0.10 -67.0 59 4.10 Pentylbenzene -4.64 9 -4.26 0.38 -75.0 59 4.90 Pentamethylbenzene -4.00 9 -4.26 -0.26 50.8 9 4.56 Hexylbenzene -5.21 9 -4.73 0.48 -61.0 59 5.52 Hexamethylbenzene -5.23 53 -4.58 0.65 167.0 65 4.61 Diphenylmethane -4.08 53 -4.25 -0.17 25.0 59 4.14 Bibenzyl -4.62 9 -4.72 -0.10 52.0 9 4.79 Biphenyl -4.35 42 -4.17 0.18 70.5 9 4.01 Napthalene -3.60 9 -3.57 0.03 80.3 9 3.30 1 -Methylnaphthalene -3.70 9 -4.05 -0.35 -22.0 65 3.87 2-MethylnapthaIene -3.77 53 -4.01 -0.24 34.6 9 3.86 1,4-DimethylnaphthaIene -4.14 9 -4.58 -0.44 -18.0 65 4.37 1,5-Dimethlnapthalene -4.68 42 -4.57 0.11 81.0 9 4.38 2,3-Dimethylnaphthalene -4.72 9 -4.60 0.12 102-104 59 4.40 1 -Ethylnaphthalene -4.17 9 -4.56 -0.39 -14.5 65 4.39 2-EthylnaphthaIene -4.29 9 -4.52 -0.23 -70.0 65 4.38 1,2,3,4-Tetrahydronapthalene -4.37 53 -3.61 0.76 -35.0 65 3.49 Acenapthene -4.63 9 -4.31 0.32 95.0 9 3.92 Acenapthylene -3.96 9 -4.15 -0.19 90.0 9 Fluorene -5.00 9 -4.52 0.48 116.0 9 4.18 Anthracene -6.35 9 -5.35 1.00 217.5 59 4.45 2-Methylanthracene -6.96 9 -5.74 1.22 204.0 9 9-Methylanthracene -5.89 9 -5.77 0.12 79.0 9 5.07 Phenanthrene -5.26 9 -5.10 0.16 100.0 9 4.47 1 -Methylphenanthrene -5.85 53 -5.59 0.26 123.0 53 5.08 2-MethyIphenanthrene -5.84 53 -5.59 0.25 58.0 9 5.15 Benzo(a)fluorene -6.68 9 -6.67 0.01 187.0 9 5.68 Benzo(b)fluorene -8.04 9 -6.64 1.40 209.0 9 5.77 Pyrene -6.18 42 -6.18 0.00 156.0 59 4.88 Napthacene -8.60 9 -6.98 1.62 341.0 65 5.90 Chrysene -8.06 42 -6.93 1.13 255.0 9 5.73 5-Methylchrysene -6.59 53 -7.40 -0.81 117.1 59 5,6-Di methylchrysene -7.01 53 -7.87 -0.86 Triphenylene -6.73 42 -6.71 0.02 199.0 9 5.49 Perylene -8.80 42 -7.38 1.42 277.0 9 5.82 Benzo(j)nuoranthene -8.00 9 -7.49 0.51 165.0 59 Benzo(k)fluoranthene -8.49 9 -7.56 0.93 217.0 9 Cholanthrene -7.85 53 -7.40 0.45 206 ref re/ Compound log Sw log Sw calc- obs Mp (°C) ■ log exp calc® Poet” Benzo(a)pyrene -8.70 46 -7.85 0.85 181.2 S3 5.97 Benzo(e)pyrene -7.80 9 -7.91 -0.11 181.0 53 Benzo[ghi]perylene -9.02 42 -8.50 0.52 277.0 9 6.63 Fluorobenzene -1.80 9 -1.98 -0.18 -42.0 65 2.27 1,3-Difluorobenzene -2.00 53 -2.07 -0.07 -59.0 59 1,4-Difluorobenzene -1.97 53 -1.94 0.03 -13.0 59 Chlorobenzene -2.38 9 -2.65 -0.27 -45.0 65 2.89 1,2-Dichlorobenzene -3.05 9 -3.26 -0.21 -17.0 65 3.43 1,3-Dichlorobenzene -3.04 9 -3.32 -0.28 -24.0 65 3.53 1,2,3-Trichlorobenzene -4.00 9 -3.94 0.06 52.6 9 4.14 1,2,4-Trichlorobenzene -3.59 53 -3.90 -0.31 16.0 9 4.05 1,3.5-Trichlorobenzene -4.48 9 -3.93 0.55 63.5 9 4.19 1,2,3,5-Tetrachlorobenzene -4.63 9 -4.48 0.15 54.5 59 4.66 1,2,4,5-Tetrachlorobenzene -5.56 9 -4.48 1.08 139.5 9 4.60 Pentachlorobenzene -5.65 9 -5.02 0.63 84.5 9 5.18 4-Chlorotoluene -3.08 53 -3.14 -0.06 7.0 65 3.33 Benzylchloride -2.39 53 -2.30 0.09 -43.0 65 1 -Chloronapthalene -3.93 9 -4.24 -0.31 -2.3 65 4.10 2-Chlorobiphenyl -4.54 44 -4.82 -0.28 32.1 59 4.53 3-Chlorobiphenyl -4.88 44 -4.82 0.06 18.0 59 4.71 Bromobenzene -2.55 9 -2.89 -0.34 -31.0 65 2.99 1,3-Dibromobenzene -3.54 53 -3.88 -0.34 -7.0 59 3.75 1,4-Dibromobenzene -4.07 9 -3.87 0.20 87.3 9 3.79 1,3,5-Tribromobenzene -5.60 9 -4.84 0,76 119.6 9 4.51 2-Bromotoluene -2.23 53 -3.41 -1.18 -27.0 65 4-BromotoIuene -3.19 53 -3.36 -0.17 28.5 59 1 -Bromonapthalene -4.35 53 -4.57 -0.22 -1.0 59 lodobenzene -3.01 42 -3.35 -0.34 -29.0 59 3.25 1-Iodonapthalene -4.55 53 -5.05 -0.50 164.0 9 o-Fluorobromobenzene -2.70 19 -3.13 -0.43 o-Chlorobromobenzene -3.19 27 -3.71 -0.52 m-Chlorobromobenzene -3.21 27 -3.71 -0.50 -21.5 59 p-Chlorobromobenzene -3.63 27 -3.68 -0.05 m-Chloroi odoben zene -3.55 27 -4.20 -0.65 p-Chloroiodobenzene -4.03 27 -4.21 -0.18 53-54 59 p-Bromoiodobenzene -4.56 19 -4.45 0.11 90-92 59 2-Chloroanisole -2.46 53 -2.92 -0.46 2.68 3-ChloroanisoIe -2.78 47 -2.88 -0.10 <25 59 2.98 4-Chloroanisole -2.78 47 -2.80 -0.02 -18.0 59 2.78 Benzaldehyde -1.19 9 -1.68 -0.49 -26.0 65 1.47 p-Methoxybenzaldehyde -1.49 9 -1.92 -0.43 -1.0 65 1.76 Acetophenone -1.28 9 -1.85 -0.57 19.6 59 1.58 Anthraquinone -5.19 9 -3.98 1.21 286.0 59 3.39 Methyl benzoate -1.85 9 -2.09 -0.24 -12.0 65 2.12 Ethyl benzoate -2.32 9 -2.51 -0.19 -34.0 65 2.64 Diethyl phthalate -2.35 9 -2.69 -0.34 -3.0 65 2.47 Di(2-ethylhexyl)-phthalate -6.96 9 -8.23 -1.27 -50.0 59 7.45 Benzonitrile -1.00 9 -1.75 -0.75 -13.0 65 1.56 Aniline -0.41 9 -1.36 -0.95 -6.0 65 0.90 o-Toluidine -2.21 9 -1.76 0.45 130.0 9 1.32 m-Methylaniline -0.85 9 -1.72 -0.87 -30.0 59 1.40 o-Chloroaniline -1.52 9 -2.18 -0.66 -1.5 65 1.90 m-Chloroaniline -1.37 9 -2.12 -0.75 -10.0 65 1.88 p-Chloroaniline -1.66 9 -2.08 -0.42 72.5 9 1.88 m-Nitroaniline -2.19 9 -1.97 0.22 112.5 9 1.37 p-Nitroaniline -2.37 9 -1.81 0.56 147.8 9 1.39 Ethyl-p-aminobenzoate -2.10 9 -2.30 -0.20 92.0 9 1.86 Butamben -3.08 31 -3.24 -0.16 58.0 49 2.87 N-Methylaniline -1.28 9 -1.86 -0.58 -57.0 65 1.66 207 ref rè/ Compound log Sw log Sw calc- obs Mp CC) ■ log exp calc" Poet” N-Ethylaniline -1.70 ------2.34 -0.64 -64.0 59 2.16 N,N-Diethylaniline -3.03 9 -3.52 -0.49 -38.0 65 3.31 1-Napthylamine -1.92 9 -2.75 -0.83 49.2 9 2.25 p.p’-Biphenyldiamine -2.70 9 -2.86 -0.16 128.0 9 1.34 Procaine -1.78 31 -2.23 -0.45 238.0 49 2.14 Diphenylamine -3.50 46 -3.68 -0.18 52.0 59 3.50 Azobenzene -2.75 9 -4.35 -1.60 68.5 9 3.82 o-Nitrotoluene -2.33 9 -2.58 -0.25 -3.5 65 2.30 m-Nitrotoluene -2.44 9 -2.69 -0.25 15.5 59 2.42 p-Nitrotoluene -2.49 9 -2.58 -0.09 51.8 9 2.37 m-Chloronitrobenzene -2.77 9 -2.74 0.03 46.0 9 2.47 p-Chloronitrobenzene -2.92 9 -2.87 0.05 83.3 9 2.39 o-Nitroanisole -1.96 9 -2.46 -0.50 9.5 9 1.73 1,2-Dinitrobenzene -3.10 9 -2.45 0.65 118.5 9 1.69 1,3-Dinitrobenzene -2.29 9 -2.17 0.12 89.9 9 1.49 1,4-Dinitrobenzene -3.39 9 -2.17 1.22 174.0 9 1.47 2,6-Dinitrotoluene -3.00 9 -2.71 0.29 66.0 9 2.10 2,4,6-Trinitrotoluene -3.22 9 -2.81 0.41 80.1 9 1.60 1,3,5-Trinitrobenzene -2.89 9 -2.34 0.55 121.5 59 1.18 2,3-Di chloron i trobenzene -3.48 9 -3.46 0.02 61.5 9 3.05 3,4-Di ch loron i trobenzene -3.20 9 -3.49 -0.29 41.2 9 3.12 Benzamide -0.96 9 -0.77 0.19 128.0 9 0.64 p-Fluoroacetani !i de -1.78 31 -1.31 0.47 154.0 49 1.47 p-Chloroacetanilide -2.84 31 -2.00 0.85 178.0 49 2.12 p-Bromoacetanilide -3.08 31 -2.33 0.75 168.0 49 2.29 Phenacetin -2.35 46 -1.79 0.56 155.0 49 1.58 Lidocaine -1.71 31 -2.74 -1.03 66-69 59 2.26 Benzoic acid -1.55 9 -1.29 0.26 122.4 65 1.87 m-ToIuic acid -2.14 9 -1.81 0.33 112.0 65 2.37 p-Toluic acid -2.60 9 -1.75 0.85 180.0 65 2.27 o-Chlorobenzoic acid -1.89 9 -1.62 0.27 142.0 9 2.05 p-Chlorobenzoic acid -3.31 9 -2.13 1.18 243.0 65 2.65 m-Bromobenzoic acid -2.28 42 -1.76 0.51 150.0 9 2.20 p-Bromobenzoic acid -3.54 42 -2.44 1.10 254.5 65 2.86 p-Nitrobenzoic acid -2.80 9 -1.65 1.15 242.4 65 1.89 o-Aminobenzoic acid -1.52 9 -1.42 0.10 146.5 59 1.21 Aspirin -1.72 9 -0.92 0.80 -38.3 49 1.19 Ibuprofen -3.76 46 -3.51 0.25 129.0 49 3.50 o-phthalic acid -2.11 9 -0.75 1.36 191.0 9 0.73 Naproxen -4.20 46 -3.50 0.70 153.0 49 3.34 Phenol 0.00 9 -1.11 -1.11 40.9 9 1.47 3-Methylphenol -0.68 9 -1.52 -0.84 9.0 65 1.96 p-Cresol -0.73 42 -1.61 -0.88 34.8 65 1.94 2,4-Dimethylpheno! -1.19 9 -1.94 -0.75 22.0 9 2.30 3,4-Dimethylphenol -1.38 42 -1.87 -0.49 67.0 65 2.23 3,5-Dimethylphenol -1.40 9 -1.95 -0.55 64.0 59 2.35 2,4,6-Trimethylphenol -2.05 42 -2.42 -0.37 72.0 65 Thymol -2.22 9 -2.73 -0.51 51.5 65 3.30 p-Phenyiphenol -3.48 9 -3.31 0.17 164.5 65 3.20 2-Chiorophenol -1.06 9 -1.81 -0.75 8.0 65 2.15 4-Chlorophenol -0.70 9 -1.77 -1.07 43.2 65 2.39 4-Bromophenol -1.09 9 -2.08 -0.99 63.5 65 2.59 2,3-Dichlorophenoi -1.30 42 -2.47 -1.17 59.0 9 2.84 2,6-Dichloropheno! -1.79 42 -2.41 -0.62 67.0 9 2.75 3,4-DichlorophenoI -1.25 42 -2.52 -1.27 67.0 9 3.33 3,5-Dichloropheno! -1.34 42 -2.58 -1.24 68.0 9 3.52 2,3,5-Trichlorophenol -2.67 42 -2.90 -0.23 62.0 9 2,3,6-Trichlorophenol -2.64 42 -2.95 -0.31 58.0 9 3.77 2,4,5-Trichlorophenol -2.21 9 -3.02 -0.81 68.0 9 3.72 208 ref ref Compound log Sw log Sw calc- obs Mp (°C) ■ log exp calc” Poet” 2,3,4,5-Tetrachlorophenol -3.15 4i -3.49 -0.34 116.0 9 4.21 2,3,4,6-Tetrachlorophenol -3.10 42 -3.58 -0.48 70.0 59 4.12 2,3,5,6-Tetrach lorophen ol -3.37 42 -3.44 -0.07 115.0 9 3.88 o-Methoxyphenol -1.96 9 -1.48 0.48 28.0 65 1.32 p-Hydroxybenzaldehyde -0.96 9 -1.19 -0.23 116.0 65 1.35 o-Aminophenol -0.72 9 -0.67 0.05 172.0 9 0.62 o-Nitrophenol -1.74 9 -2.14 -0.40 44.0 9 1.79 m-Nitrophenol -1.01 9 -1.70 -0.69 97.0 9 2.00 p-NitrophenoI -0.74 9 -1.55 -0.81 113.0 9 1.91 p-Hydroxybenzoic acid -1.41 9 -1.10 0.31 217.0 59 1.58 1,2-Benzenediol 0.62 9 -0.70 -1.32 105.0 65 0.88 1,3-Benzenediol 0.81 9 -0.42 -1.23 110.0 65 0.80 Methylparaben -1.83 31 -1.73 0.10 128.0 59 1.96 Ethyi-p-hydroxybenzoate -2.35 9 -2.17 0.18 116-118 59 2.47 o-Hydroxybenzamlde -1.82 9 -1.45 0.37 142.0 9 1.28 1-Napthol -2.22 9 -2.82 -0.60 96.0 9 2.84 2-Napthol -2.28 9 -2.71 -0.43 121.0 9 2.70 Phenylmethanol -0.40 9 -1.01 -0.61 -15.3 59 1.10 2-Phenoxyethanol -0.70 9 -1.09 -0.39 12.0 65 1.16 Ephedrine -0.47 31 -0.97 -0.50 37-39 59 0.93 Thiopheno! -2.12 9 -2.62 -0.50 -15.0 59 2.52 p-Toiuenesulfonamlde -1.74 9 -1.14 0.60 137.1 65 0.82 Furane -0.82 9 -1.13 -0.31 -85.6 59 1.34 Furfural -0.10 9 -0.71 -0.61 -36.5 59 0.41 Pyridine 0.76 9 -0.46 -1.22 -42.0 65 0.65 2,3-Dimethylpyridine 0.38 9 -1.13 -1.51 -15.0 65 2,4-Dimethylpyridine 0.38 9 -1.07 -1.45 -60.0 59 3,4-Dimethylpyridine 0.36 9 -1.11 -1.47 -12.0 65 3,5-Dimethylpyridine 0.38 9 -1.19 -1.57 -9.0 65 1.78 2-Ethyl pyridine 0.51 9 -1.20 -1.71 -63.0 59 1.69 Atropine -2.12 49 -2.36 -0.24 181.0 49 1.83 Quinoline -1.30 35 -2.23 -0.93 -15.5 65 2.03 Isoquinoline -1.45 35 -2.16 -0.71 26-28 59 2.08 Anti pyrene 0.72 9 -0.68 -1.40 114.0 9 0.38 Morpholine 1.97 42 0.88 -1.09 -4.9 59 -0.86 Theophylline -1.39 9 -0.14 1.25 222.0 49 -0.02 Morphine -3.28 9 -1.86 1.42 167.0 49 0.76 Codeine -1.52 9 -2.27 -0.75 155.0 49 1.14 Thiophene -1.33 45 -1.72 -0.39 134.5 49 1.89 Progesterone -4.42 9 -4.84 -0.42 135.0 49 3.87 Testosterone -4.02 9 -4.17 -0.15 76.0 49 3.32 Deoxycorticosterone -3.45 47 -4.68 -1.23 158.0 49 2.88 Corticosterone -3.24 47 -4.05 -0.81 98.0 49 1.94 Cortisone -3.11 9 -3.45 -0.34 115.0 49 1.47 Hydrocortisone -3.09 9 -3.72 -0.63 150.0 49 1.61 Prednisolone -3.18 47 -3.85 -0.67 1.62 Hydrocortisone 21-acetate -4.88 31 -3.94 0.94 2.19 Estrone -3.96 49 -4.16 -0.20 252.5 49 3.13 Dexamethasone -3.59 47 -4.05 -0.46 263.0 65 2.01 Fenuron -1.60 9 -1.39 0.21 133.5 52 0.98 Fluorometuron -3.32 60 -2.16 1.16 2.42 Diuron -3.80 9 -2.99 0.81 158.5 52 2.68 Chlorotoluron -3.46 60 -2.65 0.81 148.1 59 2.41 Metoxuron -2.56 16 -2.17 0.40 126.5 52 1.64 5-Methyl-5-ethylbarbituric acid -1.23 50 -0.34 0.89 216.0 16 0.08 Barbital -2.40 51 -0.82 1.58 186.0 51 0.65 5-Ethyl-5-isopropylbarbituric acid -2.15 50 -1.24 0.91 204.0 50 1.10 Pentobarbital -2.39 51 -2.09 0.30 125.0 51 2.10 5-Ethyl-5-(3-methylbutyI)barbital -2.66 50 -2.09 0.56 2.07 209 Compound log Sw ref log Sw calc- obs Mp (°C) ref ■ log calc“ Poet” 5,5-Di isopropylbarbital -2.77 50 -1.63 1.13 227.5 50 1.56 5-Allyl-5-ethylbarbital -1.61 50 -1.01 0.61 162.0 50 0.87 5-AllyI-5-isopropylbarbital -1.71 50 -1.34 0.36 145.0 51 1.37 Secobarbital -2.36 51 -2.29 0.07 95.0 51 1.97 5-(3-MethyI-2-butenyI)-5-ethylbarbital -2.25 50 -1.90 0.35 158.3 51 1.73 5-(3-Methy!-2-butenyl)-5-isoPrbarbitai -2.59 50 -2.33 0.27 131.3 51 2.23 5-EthyI-5-phenylbarbitaI -2.32 50 -2.29 0.03 175.0 51 1.47 Phenytoin -4.10 54 -3.16 0.94 296.9 59 2.47 3-Ethanoyloxymethylphenytoin -4.47 54 -3.62 0.85 158.1 59 3-Propanoyloxymethylphenytoin -4.91 54 -4.09 0.82 170.6 59 3-PentanoyloxymethyIphenytoin -4.68 54 -4.99 -0.31 91.0 59 3 -Hexanoy loxy methylphenyltoi n -5.89 54 -5.46 0.43 105.6 59 3-Heptanoyloxymethylphenytoin -6.30 54 -5.92 0.39 86.4 59 d-Limonene -4.26 55 -3.71 0.55 95.0 59 2,4-DB -3.73 52 -3.46 0.27 118.0 52 3.53 2-butenal 0.32 24 -0.41 -0.73 -76.5 59 2-Hydroxypyridine 1.02 24 0.21 -0.81 105-107 59 Abate -6.24 59 -5.69 0.55 30 59 5.96 Acephate 0.54 9 0.39 -0.15 85.5 9 -0.85 Acrolein 0.57 24 0.05 -0.52 -87.7 59 -0.01 Adi pic acid -0.82 9 -0.18 0.64 152.0 9 0.08 Aldrin -6.31 59 -6.81 -0.50 104 59 6.50 Altretamine -3.36 56 -3.71 -0.34 172-174 59 2.73 Ametryn -3.04 9 -3.16 -0.12 88.0 9 2.98 Amigdalin -0.77 24 -1.66 -0.89 Amitraz -5.47 24 -5.26 0.21 86.0 59 5.50 Amitrole 0.52 52 0.91 0.39 158.0 52 -0.87 Amobarbital -2.47 58 -2.19 0.28 157.0 59 2.07 Androstenedione -3.69 24 -4.31 -0.62 158.0 59 2.75 Androsterone -4.40 18 -4.54 -0.14 185.0 59 3.69 59 Anethole -3.13 24 -3.33 -0.20 21.0 Ansiomycin -1.61 24 -2.00 -0.39 Antazoline -2.60 24 -2.98 -0.38 Arecoline 0.81 24 -0.44 -1.25 <25 59 0.35 Atovaquone -5.93 57 -7.11 -1.18 Atratone -2.08 56 -2.35 -0.27 94.5 16 2.69 Atrazine -3.85 9 -2.94 0.91 176.0 9 2.61 Azintamide -1.72 58 -1.79 -0.07 Azodrin 0.65 59 0.76 0.11 55 59 -0.20 Barban -4.37 59 -4.18 0.19 75 59 Bendroflumethiazide -3.59 24 -3.81 -0.22 221-223 59 1.19 Benfluralin -5.53 24 -5.46 0.07 66.0 59 5.29 Benfuracarb -4.71 59 -3.23 1.48 <25 59 4.30 Benperidol -4.28 58 -4.94 -0.66 170- 59 Bensulide -4.20 24 ■ -4.87 -0.67 34.4 59 4.22 Bentazone -2.68 59 -2.49 0.19 138 59 2.80 Benzocaine -2.62 58 -2.36 0.25 89.0 59 1.86 Benzoin -2.85 18 -3.25 -0.40 137.0 59 2.13 Benzotriazole -0.78 24 -0.87 -0.09 100.0 59 1.44 Betamethasone-17-valerate -4.71 58 -5.81 -1.10 183-184 59 3.60 Bomeol -2.32 24 -2.68 -0.36 207.0 59 2.72 Bromacil -2.52 60 -2.46 0.06 158.0 59 2.11 Bumetanide -3.56 24 -4.22 -0.66 Butethal -1.66 60 -1.75 -0.09 128.5 59 1.73 Buturon -3.90 58 -3.25 0.65 3.00 Butylate -3.68 24 -4.07 -0.39 <25 59 4.15 Captopril -0.13 58 -1.18 -1.05 106.0 59 Carbaryl -3.22 52 -2.96 0.26 142.0 52 2.36 Carbetamide -1.83 24 -2.49 -0.66 210 re/ Compound logSw re/ log Sw calc- obs Mp (°C) - log calc* Poct^ Carbofuran -2.80 .... -2.10 0.70 151.0 SO 1.63 Carboxin -3.14 58 -2.72 0.42 94.0 59 2.14 Carbromal -2.68 24 -1.59 1.09 118.0 59 1.54 Carvacrol -2.08 24 -2.78 -0.70 Chloramphenicol -2.11 24 -2.25 -0.14 150.5 59 1.14 Chlorazine -4.41 59 -5.16 -0.75 Chlorbromuron -3.92 59 -2.80 1.12 96 59 3.09 Chlordane -6.86 9 -7.05 -0.19 105.0 9 6.00 Chlorfenac -3.08 18 -3.23 -0.15 161.0 59 3.20 Chlorimuron-ethyl -4.58 59 -2.77 1.80 181 59 Chloropham -3.38 9 -3.56 -0.18 41.4 9 3.51 Chlorothalonil -5.64 9 -4.93 0.71 250.5 9 2.90 Chloroxuron -4.89 56 -4.30 0.59 151.0 9 3.20 Chlorpropamide -3.03 58 -2.88 0.15 128.0 59 2.27 Chlortetracycline -2.88 24 -4.29 -1.41 168.5 59 Chlorthalidone -3.45 58 -2.24 1.21 Chlorzoxazone -2.83 24 -2.00 0.83 191-192 59 Cimetidine -1.61 58 -0.19 1.42 141-143 59 0.40 Cinchonidin -3.07 24 -3.65 -0.58 210.5 59 2.82 Cinnamic acid -2.48 9 -2.06 0.42 133.0 9 2.13 Citric acid 0.51 9 0.67 0.16 153.0 9 -1.72 Clomazone -2.34 59 -2.14 0.20 25 59 2.54 Coniine -1.50 58 -1.66 -0.16 Coronene -9.33 59 -9.51 -0.18 437.3 59 Coumachlor -5.84 59 -5.06 0.77 168-170 59 Coumatetralyl -2.84 52 -3.88 -1.05 174.0 52 Cyanazine -3.15 9 -2.33 0.82 166.5 9 2.22 Cyclobutyl-5-spirobarbituric acid -1.66 50 -0.85 0.81 257.0 50 -0.27 Cyclohexyl-5-spirobarbituric acid -3.06 50 -1.88 1.18 288.0 50 0.91 Cyclooctyl-5-spirobarbituric acid -2.98 50 -2.93 0.06 228.0 50 1.79 Cyclopentyl-5-spirobarbituric acid -2.35 50 -1.36 0.99 271.5 50 0.24 Cycluron -2.22 59 -2.42 -0.20 138 59 Cyfluthrin -7.34 59 -7.02 0.32 60 59 Cyhalothrin -8.18 59 -7.36 0.82 49.2 59 Danazol -5.51 60 -5.75 -0.24 Dapsone -3.09 58 -2.58 0.52 175.5 59 0.97 DDE -6.90 9 -7.45 -0.55 89.0 9 6.96 DDT -7.15 9 -7.16 -0.01 108.5 59 6.91 DEF -5.14 9 -5.66 -0.52 liquid 62 3.23 Desipramine -3.66 60 -5.12 -1.47 172-174 59 4.90 Desmedipham -4.63 52 -4.55 0.08 120.0 52 3.39 D-fenchone -1.85 24 -2.66 -0.81 liquid 59 Dialifos -6.34 24 -5.40 0.94 68.0 59 Diallate -4.29 59 -4.51 -0.23 25 59 Diazepam -3.75 60 -4.36 -0.61 132.0 59 2.99 Dibenzothiophene -4.38 24 -5.05 -0.67 97.0 59 4.38 Dibucaine -3.70 24 -3.85 -0.15 64.0 16 4.40 Dicamba -1.70 9 -2.52 -0.82 115.0 9 2.21 Dichlorophen -3.95 52 -4.37 -0.42 177.5 52 4.26 Dichlorprop -2.83 52 -3.24 -0.41 117.0 52 3.42 Diclofenac -5.10 60 -4.43 0.67 4.40 Dicofol -5.67 59 -5.90 -0.23 77.5 59 4.28 Dieldrin -6.29 9 -5.70 0.59 175.0 9 5.20 Dienestrol -4.95 24 -5.09 -0.14 227.0 59 Difenoxuron -4.16 24 -3.78 0.38 138.5 16 Dihexyl phthalate -6.14 59 -7.35 -1.21 -58 59 Dimecron 0.52 59 -0.68 -1.20 -45.0 59 Dinitramine -5.47 24 -4.77 0.70 98-99 66 Dinoseb -3.38 9 -4.49 -l.l'l 40.0 9 3.56 211 Compound log Sw ref log Sw calc- obs Mp CC) ref - log calc” Poet*’ Dionine -2.08 59 -3.12 -1.04 Dioxacarb -1.57 24 -1.23 0.34 114.5 59 Disulfiram -4.86 58 -4.33 0.53 71.5 59 3.88 Disulfoton -4.23 24 -3.95 0.28 -25.0 59 4.02 DNOC -1.46 52 -2.16 -0.71 89.0 52 2.13 Doxepin -3.40 56 -4.80 -1.40 <25 16 Doxycycline -2.87 24 -3.66 -0.79 Dyphylline -0.17 24 -0.30 -0.13 161-162 59 Eicosane -8.17 59 -9.17 -1.00 36.8 59 Endothall -0.27 24 -0.08 0.19 144.0 59 Epiandrosterone -4.16 24 -4.54 -0.38 172-174 59 3.69 Epitostanol -5.41 24 -5.54 -0.13 Equilenin -5.24 24 -4.11 1.13 258-259 59 Eriodictyol -3.62 24 -3.06 0.56 Erythritol 0.70 24 1.60 0.90 121.5 59 -2.29 Estragole -2.92 24 -3.35 -0.43 Ethambutol 0.55 60 -0.32 -0.87 Ethion -5.54 24 -4.35 1.19 -13.0 59 5.07 Ethirimol -3.02 24 -2.37 0.65 2.20 Ethoxyzolamide -3.81 24 -2.27 1.54 189.0 59 2.01 Etofenprox -8.60 52 -7.71 0.89 37.0 52 7.05 Etomidate -4.73 58 -3.24 1.50 67.0 59 3.05 Etryptamine -2.57 24 -2.58 -0.01 Eugenol -1.56 24 -2.62 -1.06 -7.5 59 Fenarimol -4.38 24 -4.00 0.38 118.0 59 3.60 Fenclofenac -3.85 60 -4.95 -1.10 109-115 59 4.80 Fenfuram -3.30 24 -3.04 0.26 109-110 59 Fenitrothion -4.04 24 -3.22 0.82 3.4 59 3.30 Fenoxycarb -4.70 52 -4.79 -0.09 53-54 59 4.30 Fenpropathrin -6.02 59 -6.06 -0.03 47 59 5.70 Fentanyl -3.23 59 -4.90 -1.67 87.5 59 3.89 Fluey thrinate -6.88 59 -7.55 -0.67 <25 59 6.20 Flucytosine -0.96 58 0.37 1.33 296.0 59 Fludrocortisone -3.43 24 -3.67 -0.24 1.67 Flumethasone -5.61 59 -4.03 1.59 1.94 Flumetralin -6.78 59 -7.05 -0.27 101-103 59 Fluometuron -3.43 9 -2.11 1.32 163.0 9 2.42 Fluoromethalone -4.10 59 -4.28 -0.18 297 59 2.00 Flurbiprofen -3.74 60 -4.53 -0.79 110-111 59 4.16 Fluridone -4.44 9 -5.59 -1.15 152.0 9 3.16 Flutriafol -3.37 24 -3.32 0.05 130.0 59 2.30 Fluvalinate -8.00 59 -7.84 0.16 Formetanate -2.34 24 -1.32 1.02 Fructose 0.64 24 1.89 1.25 Glafenine -4.57 58 -4.27 0.30 Glucose 0.74 9 1.87 1.13 83.0 9 Glutethimide -2.34 58 -3.06 -0.72 84.0 59 1.90 Glybuthiazole -3.74 24 -4.28 -0.54 Glycerol 1.12 24 1.46 0.34 18.2 59 -2.42 Glycocholic acid -5.15 59 -6.28 -1.13 166.5 59 Griseofulvin -3.25 58 -3.46 -0.21 352.8 59 2.18 Haloperidol -4.43 24 -4.86 -0.43 151.5 59 3.23 Heroin -2.79 59 -3.25 -0.46 173 59 1.14 Hexestrol -4.43 24 -5.15 -0.72 Hydroxyurea 1.12 24 1.89 0.77 141.0 59 -1.80 Hyoscyamine -1.91 58 -3.18 -1.27 1.83 Ibuproxam -3.04 24 -3.01 0.03 Indapamide -3.59 58 -3.91 -0.32 Indole -1.52 59 -2.38 -0.86 52.5 59 2.14 212 Compound log Sw ref log Sw calc- obs Mp C O ref log calc' Poet” Indomethacin -4.84 58 -5.42 -0.58 158.0 3$ 4.27 Inosine -1.23 59 0.63 1.86 lodofenphos -6.62 24 -5.49 1.13 72-73 59 5.51 loxynil -3.61 24 -4.64 -1.03 200.0 59 3.43 Isazofos -3.66 59 -2.66 1.00 <25 59 Isocarboxazid -2.46 58 -1.12 1.34 105-106 59 1.49 Isofenphos -4.19 59 -4.76 -0.57 -12 59 4.12 Isoprocarb -2.86 52 -2.37 0.49 94.5 52 2.31 Isopropalin -6.49 56 -6.01 0.48 <25 16 Isoproturon -3.54 16 -2.89 0.65 155.5 52 2.87 kebuzone -3.27 24 -4.82 -1.55 Kepone -5.26 59 -5.74 -0.48 350 59 5.41 ketoprofen -3.16 60 -4.08 -0.92 94.0 59 3.12 Lactose -0.24 59 0.52 0.76 Lenacil -4.59 60 -3.16 1.43 316.3 52 Lindane -4.64 24 -4.77 -0.13 112.5 59 3.72 Lithocholic acid -6.00 59 -6.87 -0.87 184-186 59 Lorazepam -3.60 60 -3.56 0.05 166-168 59 2.51 Lovastatin -6.00 58 -6.67 -0.67 4.26 Maltose 0.36 59 0.52 0.16 Mebendazole -3.88 60 -4.07 -0.19 288.5 59 2.83 Mecarbam -2.52 59 -2.90 -0.38 Medrogestone -5.27 24 -5.95 -0.68 Mefenacet -4.87 59 -5.18 -0.31 134.8 59 3.23 Mefenacet -4.87 59 -5.18 -0.31 134.8 59 3.23 Mefluidide -3.24 24 -2.83 0.41 183-185 59 Mepazine -4.74 24 -5.10 -0.36 Meperidine -1.89 24 -3.19 -1.30 270.0 59 2.45 Methaqualone -2.92 58 -4.23 -1.30 2.50 Methazole -2.82 24 -3.32 -0.50 208.0 59 59 Methocarbamol -0.98 58 -1.73 -0.74 93.0 Methoproptryne -2.93 59 -3.55 -0.62 2.37 Methotrimeprazine -4.22 24 -5.01 -0.79 4.68 Methoxsalen -3.66 24 -2.87 0.79 143.0 59 1.93 Methyl hydrazine 1.34 24 2.41 1.07 -52.4 59 -1.05 Methyidymron -3.35 52 -3.74 -0.39 72.0 52 3.01 Metoclopramide -3.18 24 -2.83 0.35 147.3 59 2.62 Metolcarb -1.80 52 -1.84 -0.04 76.5 52 1.24 Metranidazole -1.26 24 -0.62 0.64 -0.02 Metribuzin -2.24 59 -1.17 1.07 126 59 1.70 Minoxidil -1.98 58 -0.81 1.17 248.0 59 1.24 Mi rex -6.80 9 -7.60 -0.80 485 9 5.28 Monolinuron -2.57 9 -2.55 0.02 81.5 9 2.30 Morin -3.08 24 -2.59 0.49 303.5 59 1.54 Naepaine -3.27 24 -3.48 -0.21 66.0 16 Naled -2.28 59 -2.09 0.19 26.5-27.5 59 2.19 Naprosyn -4.16 24 -3.86 0.30 153.0 59 3.34 Neburon -4.77 24 -4.23 0.54 101.5 9 3.80 Niclosamide -4.70 24 -5.82 -1.12 Niridazole -3.22 24 -1.78 1.44 260-262 59 0.95 Nitramine -3.56 56 -4.77 -1.21 129.0 59 Nitrapyrin -3.76 24 -3.33 0.43 63.0 59 3.41 Nitrofen -5.46 24 -5.58 -0.12 70.0 59 4.64 Norea -3.17 59 -2.79 0.38 177 59 Norethisterone -4.57 24 -3.97 0.60 203.5 59 Noscapine -3.14 24 -1.78 1.36 O-Ethyl carbamate 0.85 24 -0.06 -0.91 49.0 59 -0.15 Oxadiazon -5.69 59 -5.91 -0.22 90 59 4.80 Oxalic acid 0.38 9 0.98 0.60 189.5 9 213 Compound log Sw re/ log Sw calc- obs Mp (°C) re/ log exp calc" Poet" Oxamyl 0.11 SO 0.15 0.04 100-102 SO ...... -0.47 Oxycarboxin -2.28 59 -1.40 0.88 120 59 0.74 Oxytetracycline -3.14 24 -3.65 -0.51 184.5 59 Palmitic acid -5.49 59 -5.92 -0.43 61.8 59 Parethoxycaine -2.71 24 -3.42 -0.71 Pebulate -3.53 59 -3.73 -0.20 <25 59 3.84 Pecazine -4.75 60 -5.10 -0.35 Pencycuron -5.91 59 -6.16 -0.24 130 59 4.82 Pentazocine -3.80 50 -4.90 -1.10 3.31 Perfluidone -3.80 24 -3.90 -0.10 142-144 59 Perphenazine -4.16 60 -4.57 -0.42 97.0 59 4.20 Phenbutamide -3.05 24 -2.69 0.36 Phenetole -2.33 24 -2.48 -0.15 -29.5 59 2.51 Phenylhydrazine 0.07 24 0.12 0.05 19.6 59 1.25 Phorate -4.11 24 -3.48 0.63 < 15 59 3.56 Phosalone -5.23 59 -4.53 0.71 46 59 4.38 Pindone -4.11 24 -2.91 1.20 110.0 59 Piperazine 1.07 24 1.49 0.42 106.0 59 -1.50 Pipeline -3.46 24 -2.77 0.69 130-132 59 Pirimicarb -1.95 24 -1.81 0.14 90.5 59 1.70 Prasteron e -4.12 24 -4.51 -0.39 140-141 59 3.23 Primidone -2.64 58 -1.34 1.30 281.5 59 0.91 Procymidone -4.80 59 -5.17 -0.37 166 59 3.00 Promazine -4.30 60 -4.36 -0.06 <25 59 4.55 Promethazine -4.26 60 -4.33 -0.07 <25 59 4.81 Prometryn -4.10 24 -3.56 0.54 72.5 9 3.51 Propanil -3.00 24 -3.20 -0.20 91.5 9 3.07 Propazine -4.43 9 -3.35 1.08 211.0 9 2.93 Propiconazole •3.49 59 -4.06 -0.56 <25 59 3.50 Propoxur -2.05 9 -1.86 0.19 91.0 59 1.52 Pteridine 0.02 24 -0.66 -0.68 139.5 59 -0.58 Pyrazon -2.87 24 -1.81 1.06 205.0 59 2.20 Pyridazine 1.10 24 -0.10 -1.20 -8.0 59 -0.72 Pyrimidine 1.10 24 0.11 -0.99 22.0 59 -0.40 Pyrrolidine 1.15 24 0.16 -0.99 -57.8 59 0.47 Quinethazone -3.29 24 -1.53 1.76 250-252 59 Quinidine -2.81 60 -3.64 -0.83 174.0 59 2.64 Quintozene -5.82 24 -5.73 0.09 144.0 59 4.64 Raffinose -0.41 60 -1.89 -1.48 Reposal -2.70 60 -3.13 -0.44 213.0 59 Ronnel -5.72 24 -4.56 1.16 41.0 59 5.07 Rotenone -4.42 24 -4.79 -0.37 176.0 59 4.10 Rovral -4.38 59 -4.53 -0.16 136 59 3.10 Salbutamol -1.22 24 -1.24 -0.02 Salicin -0.85 9 -0.44 0.41 199.0 9 -1.22 Salicylamide -1.84 58 -1.51 0.33 142.0 59 1.28 Scopolamine -0.48 24 -2.04 -1.56 Siduron -4.11 24 -3.88 0.23 135.5 59 Silvex -3.33 18 -3.87 -0.54 181.6 59 3.80 Sorbic acid -1.77 9 -1.02 0.75 134.5 9 1.33 Sorbitol 1.09 24 1.48 0.39 11.0 59 -3.10 Sparsomycin -1.98 60 -0.88 1.09 -1.71 Stanolone -4.74 60 -4.54 0.20 181.0 59 3.66 Stearic acid -5.68 24 -6.86 -1.18 68.8 59 Stirofos -4.52 59 -4.11 0.41 97-98 59 3.53 Succinimide 0.30 24 0.56 0.26 123-125 59 Sucrose 0.79 59 -0.63 -1.42 185.5 59 -3.70 Sulfadiazine -3.40 58 -1.97 1.43 -0.09 Sulfallate -3.39 24 -3.90 -0.5i Sulfamerazine -2.85 58 -2.45 0.40 236.0 59 0.14 214 Compound log Sw ref log Sw calc- obs Mp (°C) ref log exp calc® Poet" Sulfamethazine -2.27 3$ -2.94 -0.67 198.5 59 0.89 Sulfamethoxazole -2.62 58 -2.34 0.28 167.0 59 0.89 Sulfamoxole -2.44 24 -3.03 -0.59 Sulfanilamide -1.34 9 -0.72 0.62 165.0 9 -0.62 Sulfapyridine -2.70 24 -2.47 0.23 192.0 59 0.00 Sulfathlazole -2.81 58 -2.66 0.15 189.0 59 0.05 Sulfisomidine -2.24 24 -2.69 -0.45 243.0 59 -0.33 Sulfometuron -4.56 59 -3.58 0.99 204 59 Sulindac -5.00 60 -5.06 -0.06 182-185 59 3.05 Sulpiride -2.88 58 -2.02 0.86 178.0 59 Terbacil -2.48 9 -1.99 0.49 176.0 9 1.89 Terbufos -4.76 59 -4.26 0.50 -29.2 59 4.47 Terbumeton -3.24 56 -2.70 0.54 123.5 59 3.04 Tetracycline -3.12 60 -3.41 -0.29 -1.47 Tetrafluthrin -7.32 59 -6.97 0.35 44.6 59 Thalidomide -2.68 59 -2.24 0.43 269.5 59 0.33 Thiometon -3.09 59 -3.01 0.08 Thiopropazate -4.70 60 -5.04 -0.34 Thioridazine -5.82 60 -6.61 -0.79 73.0 59 5.90 Triadimefon -3.61 24 -3.49 0.12 82.0 59 2.77 Triallate -4.88 24 -5.12 -0.24 29-30 59 Triamcinolone -3.68 60 -3.14 0.54 270.0 59 1.16 Triazolam -4.09 60 -4.92 -0.83 2.42 Trichlomethiazide -2.68 24 -2.45 0.23 0.56 Trichlorfon -0.22 9 0.45 0.67 83.5 9 0.51 Trichloronate -5.75 59 -4.78 0.97 5.23 Triclosan -4.46 24 -5.19 -0.73 54-57.3 59 Tricresyl phosphate -6.01 59 -6.12 -0.11 -33 59 Trietazine -4.06 9 -3.83 0.23 100.5 9 3.34 Tri fen morph -7.22 59 -6.07 1.15 Trifluorperazine -4.52 59 -4.63 -0.10 5.03 Trifluralin -5.68 24 -5.46 0.22 49.0 59 5.07 Trimazosin -3.64 60 -3.68 -0.04 Trimethoprim -2.86 58 -2.73 0.13 199-203 59 0.91 Tyramine -1.12 24 -0.88 0.24 161-163 59 Ursodeoxycholic acid -4.35 59 -3.66 0.69 Valproic acid -2.06 58 -2.10 -0.05 2.75 Vinclozolin -4.93 59 -5.01 -0.09 108 59 3.10 Warfarin -4.26 52 -3.91 0.35 161.5 52 2.70 Xipamide -3.79 24 -5.19 -1.40 ‘Calculated log Sw using eqn 21, this work (no mp term). ^Experimental octanol-water partition measurements taken from the Abraham UCL Database or from ClogP for Windows, version 2.0.0b, Biobyte Corp., USA. 215 Table 6.14 Test set data : log Sw predictions using Abraham, WsKow (Meylan) and TOPC (Klopman) models, and mp and log Poet data. ref ref Compound log Sw ------Prediction method —...... Mpt (°C) log Poet® ClogP^ WsKow exp Abraham® Meylan*’ Klopman r Klopman 2** log Poet® Butane -2.57 34 -1.83 -2.63 -2.79 -3.00 -138.0 66 2.89 2.81 2.89 Hexane -3.84 9 -2.78 -3.70 -3.94 -4.04 -95.0 66 3.90 3.87 3.29 2,3-Dimethylbutane -3.65 S3 -2.78 -3.29 -3.53 -3.41 -129.0 66 3.85 3.61 3.14 2,4-Dimethylpentane -4.26 53 -3.25 -3.57 -4.10 -3.93 -123.0 66 4.26 4.14 3.63 2-Methylheptane -5.08 53 -3.72 -4.16 -4.88 -4.77 -109.0 60 4.80 4.20 Nonane -5.88 53 -4.19 -4.74 -5.66 -5.60 -53.0 66 5.65 5.45 4.76 Cyclopentane -2.64 9 -2.17 -2.81 -3.26 -3.73 -94.0 60 2.79 2.68 Cyclohexane -3.10 9 -2.69 -3.29 -3.71 -4.25 6.5 66 3.44 3.35 3.18 Ethylcyclohexane -4.25 53 -3.59 -4.45 -4.57 -4.97 -111.0 60 4.40 4.08 34 Ethylene -0.40 -0.59 -0.91 -1.07 -1.05 -169.0 66 1.13 1.27 1.27 I-Pentene -2.68 9 -2.00 -2.52 -2.19 -2.11 -165.0 60 2.86 2.66 3-Methyl-1-Butene -2.73 53 -1.97 -2.46 -1.98 -1.80 -168.0 66 2.77 2.73 2.59 I-Heptene -3.73 9 -2.94 -3.86 -3.33 -3.15 -119.0 66 3.99 3.91 3.64 1-Decene -5.51 53 -4.35 -5.13 -5.05 -4.71 -66.3 60 5.50 5.12 1,4-Pentadiene -2.09 9 -1.80 -2.36 -2.29 -2.34 -148.0 66 2.48 2.37 2.52 1 -Methylcyclohexene -3.27 9 -2.74 -3.44 -2.72 -2.88 -120.0 60 3.39 3.51 Propyne -0.41 21 -0.58 -0.84 -0.85 -103.0 60 0.92 1.04 53 3-Hexyne -1.99 -2.11 -2.54 -2.31 -103.0 60 2.51 2.57 Dichloromethane -0.63 53 -1.14 -0.89 -0.87 -1.05 -97.0 66 1.25 1.25 1.34 1,1-Dichloroethane -1.29 9 -1.42 -1.45 -1.24 -1.26 -97.0 66 1.79 1.78 1.76 9 1,1,2,2-Tetrachloroethane -1.74 -2.28 -2.47 -2.36 -2.40 -43.0 66 2.39 2.64 2.19 1 -Chloropropane -1.47 9 -1.51 -1.52 -1.35 -1.36 -123.0 66 2.04 1.99 2.07 9 1 -Chlorobutane -2.03 -1.98 -2.13 -1.92 -1.88 -123.0 66 2.64 2.52 2.56 9 2-Chloro-2-methylbutane -2.51 -2.53 -2.13 -2.23 -1.84 -74.0 60 2.92 2.94 9 1,1 -Dichloroethylene -1.64 -1.64 -1.73 -1.72 -1.82 -122.0 66 2.13 2.37 2.12 9 Hexachloro-1,3-butadiene -4.92 -4.72 -5.18 -4.56 -5.08 -20.5 66 4.78 4.90 4.72 9 Tetrabromomethane -3.14 -3.62 -4.54 -2.75 -3.11 88-90 60 3.42 3.43 2.80 9 2-Bromopropane -1.59 -1.69 -1.93 -1.35 -1.33 -89.0 66 2.10 2.13 2.08 42 1-Bromohexane -3.81 -3.17 -3.65 -3.27 -3.21 -95.0 66 3.80 3.72 3.63 42 Diiodomethane -2.34 -2.56 -3.12 -2.49 -2.73 6.1 66 2.30 2.31 2.35 216 ref ref Compound log Sw ------Prediction method ------Mpt (°C) log Poct^ ClogP' WsKow exp Abraham“ Meylan'’ Klopman r Klopman 2*' log Poet® 0 ' 66...... 1-lodobutane -2.96 -2.69 -3.16 -2.73 -2.72 -103.0 3.08 3.05 3.06 Chlorodibromethane -1.90 9 -2.18 -2.57 -1.73 -2.05 -22.0 66 2.24 2.23 1.70 Diethyl ether -0.09 9 -0.50 -0.50 -0.40 -0.70 -116.3 60 0.89 0.87 1.05 Methyl propyl ether -0.39 9 -0.58 -0.78 -0.40 -0.70 <25 1.21 0.87 1.05 Propylisopropylether -1.34 9 -1.41 -1.62 -1.34 -1.43 <25 1.71 1.96 1,2-Diethoxyethane -0.77 9 -0.30 -0.63 -0.70 -1.30 -74.0 66 0.66 0.93 0.77 2-Methyltctrahydrofurane 0.11 9 -0.46 -0.99 -1.85 -111 -136.0 60 0.97 1.35 Valeraldehyde -0.85 9 -0.80 -0.95 -1.37 -0.68 -91.5 60 1.36 1.31 t-Crotonaldehyde 0.32 9 -0.19 -0.23 -0.29 0.11 -76.5 60 0.85 0.60 3-Pentanone -0.28 42 -0.59 -0.68 -0.28 -0.70 -40.0 66 0.82 1.16 0.75 3-Methyl-2-pentanone -0.67 20 -1.02 -0.93 -0.65 -0.91 <25 1.91 1.16 4-Heptanone -1.30 9 -1.49 -1.51 -1.43 -1.74 -33.0 60 2.97 1.73 20 5-Nonanone -2.58 -2.42 -2.56 -2.57 -2.78 -50.0 66 2.88 2.97 2.71 Camphor -1.96 31 -2.28 -2.31 -1.80 -2.27 179.8 49 2.18 2.34 Propyl formate -0.49 9 -0.73 -0.55 0.06 -0.17 -93.0 60 0.83 0.79 0.81 Isopentyl formate -1.52 20 -1.58 -1.52 -0.88 -0.89 -93.0 60 1.72 1.72 Isopropyl acetate -0.55 9 -0.85 -1.04 -0.51 -0.39 -73.4 60 1.02 1.28 9 Methyl propionate -0.14 -0.51 -0.56 -0.14 -0.19 -88.0 66 0.76 0.71 0.86 Propyl butyrate -1.92 9 -1.86 -1.05 -0.72 -0.71 <25 60 1.24 1.36 34 Pentyl propanoate -2.25 -2.33 -2.15 -1.86 -1.75 <25 2.30 2.34 34 Methyl octanoate -3.17 -2.80 -3.19 -3.01 -2.79 <25 60 3.36 3.32 Methyl decanoate -4.69 34 -3.73 -4.33 -4.15 -3.83 -18.0 60 4.41 4.30 Malonic acid diethylester -0.82 9 -1.01 -1.19 -0.75 -0.79 1.13 -0.15 Ethylamine 2.06 19 0.78 1.35 0.35 0.41 -81.0 66 -0.13 -0.13 1.82 Hexylamine -0.25 19 -1.06 -0.69 -1.94 -1.67 -23.0 66 2.06 1.98 1.79 Dipropylamine -0.46 9 -0.80 -0.36 -1.44 -1.41 -63.0 60 1.67 1.60 2.99 Tripropylamine -2.28 16 -1.84 -1.62 -0.98 -2.57 -93.0 66 2.79 2.98 0.87 9 2-Nitropropane -0.62 -0.74 -1.04 -0.42 -93.0 66 0.75 0.55 -1.56 Urea 0.96 9 1.91 0.85 2.92 1.30 132.7 9 -2.11 -2.11 4.02 Decanoic acid -3.44 9 -2.84 -3.56 -3.52 -3.56 31.4 66 4.09 4.04 -0.75 9 Succinic acid -0.20 0.62 0.835 1.09 0.27 185.0 66 -0.53 0.35 I-Propanol 0.62 20 0.23 0.66 0.55 0.36 -127.0 66 0.25 0.29 0.77 217 ref ref Compound log Sw ------Prediction method - ...... Mpt (°C) log Poet* Clogpf WsKow exp Abraham® Meylan** Klopman 1*^ Klopman 2** log Poet® Butan-2-ol 0.47 21 -0.23 0.25 0.28 0.16 -115.0 66 0.65 0.60 1.26 2-Methylbutanol -0.47 9 -0.71 -0.44 -0.39 -0.36 <25 49 1.22 1.22 2,2-Dimethylpropanol -0.40 9 -0.57 -0.45 -0.34 -0.12 53.0 9 1.31 1.09 1.75 2-Methylpentanol -1.11 44 -1.17 -0.93 -0.97 -0.88 <25 1.75 1.68 3-Methyl-2-pentanol -0.71 44 -0.94 -0.87 -0.67 -0.57 <25 1.53 1.75 2-Ethyl-1-butanol -1.17 9 -1.19 -0.93 -0.97 -0.88 -15.0 60 1.75 2.31 1-Heptanol -1.81 9 -1.63 -1.78 -1.74 -1.72 -36.0 66 2.62 2.41 2.20 2-Methyl-2-hexanol -1.08 44 -1.31 -1.42 -1.86 -1.16 <25 2.06 2.13 2,4-Dimethyl-2-pentanoI -0.92 9 -1.32 -1.36 -1.66 -0.84 <-20 60 1.93 2.73 3-Octanol -1.98 44 -1.88 -1.98 -2.01 -1.92 -45.0 60 2.72 3.30 1-Nonanol -3.01 42 -2.55 -2.96 -2.89 -2.76 -7.0 60 4.26 3.47 4.77 1 -Dodecanol -4.80 44 -3.95 -4.43 -4.61 -4.32 22-26 60 5.13 5.06 7.72 1-Octadecanol -8.40 44 -6.74 -7.25 -8.05 -7.44 91.0 60 8.23 1.27 4-Pentene-l-oI -0.15 44 -0.42 -0.34 -0.70 -0.91 0.87 2.88 Ethanethiol -0.60 9 -0.92 -0.74 -0.86 -0.83 -148.0 60 1.38 -1.31 Di-n-propylsulfide -2.58 45 -2.49 -2.53 -2.81 -102.5 60 2.96 3.03 Thiourea 0.32 9 1.19 0.86 0.30 1.09 176.0 9 -1.02 -1.02 3.52 Ethylbenzene -2.77 9 -2.86 -2.67 -2.94 -2.86 -95.0 66 3.15 3.17 3.63 Propylbenzene -3.37 9 -3.33 -3.23 -3.52 -3.38 -99.0 66 3.72 3.70 3.94 1,3,5-Trimethylbenzene -3.40 9 -3.24 -3.00 -3.29 -3.42 -45.0 66 3.42 3.64 4.18 Isobutylbenzene -4.12 9 -3.79 -4.18 -3.89 -3.58 -51.0 66 4.28 4.10 2.89 1,2,4;5-TetramethyIbenzene -4.59 53 -3.77 -3.60 -3.76 -3.96 80.0 9 4.10 4.04 4.30 t-Pentylbenzene -4.15 9 -4.26 -4.04 -4.40 -3.85 4.50 4.26 Styrene -2.82 9 -2.87 -2.48 -3.04 -3.08 -31.0 66 2.95 2.87 4.26 4-MethylbiphenyI -4.62 53 -4.65 -4.38 -4.84 -4.89 45.5 9 4.63 4.53 3.47 1,3-DimethyInaphthalene -4.29 9 -4.57 -4.12 -4.56 -4.66 -6.0 60 4.42 4.31 4.56 2,6-Dimethylnaphthalene -4.89 9 -4.51 -4.02 -4.56 -4.66 109.0 9 4.31 4.31 5.44 Indan -3.04 9 -3.15 -2.78 -3.53 -3.60 -51.0 66 3.18 3.15 4.93 1-Methylfluorene -5.22 9 -4.99 -5.87 -5.54 -5.66 87.0 9 4.97 4.57 6.62 9,10-Dimethylanthracene -6.57 9 -6.18 -6.68 -6.29 -6.45 182.0 9 5.69 5.49 6.07 Fluoranthene -6.00 9 -5.93 -6.19 -6.35 -6.38 111.0 9 5.16 4.95 6.11 9 7,12-Dimethylbenz(a)anthracene -7.02 -7.90 -7.13 -8.01 -8.23 122-123 60 5.80 6.66 7.05 218 ref Compound log Sw ref ------Prediction method - ...... Mpt (°C) log Poet® ClogP' WsKow exp Abraham® Meylan'’ Klopman r Klopman 2^ log Poet® 6-Methylchrysene -6.57 SJ -7.40 -7.26 -7.55 -7.69 160.3 60 6.16 6.70 Benzo(b)fluoranthene -8.23 9 -7.37 -7.09 -8.08 -8.16 167.0 60 6.12 3.28 3-Methylcholanthrene -7.92 9 -7.88 -8.29 -8.72 -8.98 179.0 9 6.75 6.62 4.57 Picene -7.87 53 -8.77 -8.41 -8.82 -8.93 368.0 60 6.84 3.18 Benzyltrifluoride -2.51 53 -2.32 -2.84 -3.90 -3.62 3.03 3.81 1,4-Dichlorobenzene -3.27 9 -3.29 -3.21 -3.16 -3.09 53.1 9 3.39 3.57 3.77 1,2,3,4-Tetrachlorobenzene -4.57 9 -4.48 -4.70 -4.68 -4.59 46.8 66 4.64 4.75 5.55 2-Chlorotoluene -3.52 53 -3.16 -3.05 -2.87 -2.87 -36.0 66 3.42 3.35 4.06 2-Chloronapthalene -4.14 9 -4.25 -3.79 -4.13 -4.12 59.5 9 4.14 4.03 3.80 1,2-Dibromobenzene -3.50 53 -3.87 -4.03 -3.83 -3.67 5.0 60 3.64 3.67 2.07 1,2,4,5-Tctrabromobcnzene -6.98 9 -5.61 -6.45 -6.01 -5.75 173.5 9 5.13 5.19 4.05 2-BromonapthaIcnc -4.40 9 -4.52 -4.18 -4.47 -4.41 54.0 60 4.09 4.18 3.15 m-Fluorobromobenzene -2.67 19 -3.08 -2.97 -3.04 -3.33 3.15 1.66 o-Chloroiodobenzene -3.54 27 -4.23 -4.19 -3.79 -3.92 1.0 60 3.98 1.09 Anisole -1.85 9 -2.13 -1.79 -1.26 -1.68 -37.0 60 2.11 2.06 1.62 Diphenyl ether -3.96 9 -4.42 -4.04 -3.26 -3.69 28.0 66 4.21 4.24 2.02 Benzophenone -3.12 9 -3.80 -3.25 -3.14 -3.68 48.5 9 3.18 3.18 2.29 Dimethyl phthalate -1.66 9 -1.86 -1.95 -1.50 -1.79 2.0 66 1.56 1.48 2.17 9 Phthalonitrile -2.38 -1.70 -1.25 -0.65 -0.76 140.0 9 0.99 1.01 1.07 p-Methylaniline -1.21 9 -1.70 -1.17 -1.54 -1.63 44.5 9 1.39 1.41 1.81 o-Nitroaniline -1.96 9 -2.14 -1.79 -1.47 -1.50 71.5 9 1.85 1.92 2.46 3J Risocaine -2.45 -2.77 -2.58 -2.15 -2.23 75.0 60 2.43 2.55 1.89 9 N,N-Dimethylaniline -1.92 -2.58 -2.06 -1.46 2.0 66 2.31 2.34 2.18 9 Benzylamine -1.54 -0.87 0.09 -1.65 -1.61 10.0 60 1.09 1.09 2.99 9 Nitrobenzene -1.80 -2.11 -2.07 -2.04 -1.99 5.7 60 1.85 1.88 1.10 9 o-Chloronitrobenzene -2.55 -2.79 -2.65 -2.80 -2.75 32.0 9 2.24 2.40 1.64 9 p-Nitroanisole -2.41 -2.42 -2.44 -1.65 -2.09 54.0 9 2.03 2.10 2.08 9 2,4-Dinitrotoluene -2.82 -2.64 -2.61 -2.89 -2.95 69.0 60 1.98 2.05 2.52 9 1-Nitronapthalene -3.54 -3.91 -3.58 -3.77 -3.78 59.5 9 3.19 3.06 1.69 Acetanilide -1.33 9 -1.13 -1.18 -1.30 -1.45 114.0 9 1.16 1.16 1.43 4-Nitroacetanilide -2.69 16 -2.22 -2.32 -1.69 -1.86 216.0 66 1.66 1.46 2.06 9 o-Toluic acid -2.06 -1.92 -2.30 -2.31 -1.95 105.0 60 2.46 2.46 2.61 219 ref ref Compound log Sw ------Prediction method — ...... Mpt (°C) log Poet' ClogP' WsKow exp Abraham“ Meylan** Klopman 1*^ Klopman 2** log Poet® 0 9 m-Chlorobenzoic acid -2.59 -2.01 -2.63 -2.60 -2.16 158.0 2.68 2.70 3.42 m-Nitrobenzoic acid -1.68 9 -1.60 -2.37 -2.23 -1.82 142.0 9 1.83 1.84 2.16 9 9 Phenylacetic acid -0.89 -1.13 -1.00 -1.51 -1.93 76.5 1.41 1.41 2.80 9 66 2-Methylphenol -0.62 -1.71 -1.08 -1.04 -1.15 30.9 1.98 1.97 3.45 2,6-Dimethylphenol -1.29 42 -2.08 -1.53 -1.51 -1.69 49.0 66 2.36 2.47 3.45 p-t-Butylphenol -2.41 9 -2.73 -2.54 -2.50 -2.15 101.0 66 3.31 3.30 4.74 3-Chlorophenol -0.70 9 -1.89 -1.70 -1.34 -1.37 32.8 66 2.50 2.48 0.24 2,4-Dichlorophenol -1.55 9 -2.49 -2.42 -2.10 -2.12 45.0 60 3.06 2.96 2.24 2,3,4-Trichlorophenol -2.67 42 -2.90 -3.31 -2.85 -2.87 80.0 9 3.60 3.58 1.03 9 9 2,4,6-Trichlorophenol -2.34 -2.92 -3.21 -2.85 -2.87 69.0 3.69 3.37 0.27 Pentachlorophenol -4.28 9 -3.97 -4.94 -4.37 -4.38 174.0 9 5.12 4.80 1.49 9 9 p-Aminophenol -0.80 -0.23 -0.03 -0.01 -0.12 190.0 0.04 0.25 0.95 Salicylic acid -1.82 9 -1.58 -1.56 -0.78 -0.44 155.0 49 2.26 2.19 3.71 1,4-Benzenediol -0.17 9 -0.24 0.07 0.48 0.36 172.0 66 0.59 0.81 1.90 p-Hydroxyacetanilide -1.03 9 -0.69 -0.70 -0.23 -0.48 167.0 9 0.51 0.49 2.17 9 1-Phenylethanol -0.92 -1.24 -0.80 -1.15 -1.34 20.0 60 1.42 1.41 3.23 Phenylthiourea -1.77 9 -1.32 0.09 -1.78 -1.18 149.0 9 0.73 0.75 0.16 Dibenzofiirane -4.60 9 -4.34 -5.06 -4.66 -3.93 83.0 66 4.12 4.09 5.01 2,6-Dimethylpyridine 0.45 9 -1.07 -0.12 -2.58 -2.65 -6.0 66 1.68 1.64 3.08 Cocaine -2.25 48 -2.99 -2.37 -4.06 222.5 49 2.30 2.72 3.72 Carbazole -5.27 9 -4.11 -4.71 -3.58 245.0 9 3.72 3.52 3.94 Caffeine -0.88 42 -0.69 -1.87 238.0 49 -0.07 -0.06 2.03 Imipramine -4.19 31 -4.55 -5.45 -6.35 174-175 60 0.23 4.41 -0.38 Hydroxyprogesterone-17a -3.82 18 -4.63 -3.81 -4.01 -5.00 132.5 49 2.74 3.15 1.51 17 a-Methyltestosterone -4.00 18 -4.43 -3.77 -4.98 -5.44 162-167 60 3.36 3.74 0.47 Estradiol -5.03 31 -4.06 -3.52 -4.29 -4.69 176.0 49 4.01 3.78 1.31 Monuron -2.89 61 -2.30 -2.31 -2.10 170.5 60 1.94 1.99 3.44 5,5-Dimethylbarbituric acid -1.74 50 0.14 0.04 0.45 278.0 50 -0.47 -0.40 5.41 Butabarbital -2.39 51 -1.62 -2.19 -1.31 168.0 51 1.57 1.58 3.32 5-Allyl-5-methylbarbital -1.16 50 -0.53 -0.84 -0.82 167.0 50 0.71 0.17 0.00 50 5,5-Diallylbarbital -2.08 -1.43 -1.70 -2.08 175.0 51 1.05 0.75 1.90 5-Allyl-5-phenylbarbital -2.37 50 -2.50 -2.43 -2.83 1.69 1.41 1.99 220 ref ref Compound log Sw ------Prediction method ------Mpt (°C) log Poet' ClogP' WsKow exp Abraham® Meylan’’ Klopman T Klopman 2^ log Poet® 3-Butanoyloxymethylphenytoin -5.07 54 -4.66 -4.71 -5.41 131.1 60 2.27 3.45 3-Octanoyloxymethylphenytoin -6.52 54 -6.38 -6.80 -7.49 65.2 60 4.38 0.05 Acridine -3.67 60 -4.30 -4.52 -5.11 -5.13 108.0 60 3.40 3.41 1.22 Alizarin -2.78 56 -3.26 -2.07 -0.65 287-289 60 -0.79 1.72 Aminocarb -2.36 56 -2.87 -2.34 -1.88 93-94 60 1.90 1.45 5.11 Ancymidol -2.60 18 -2.18 -2.19 -4.75 -4.61 110-111 60 1.91 0.76 2.97 Anilofos -4.43 60 -5.40 -5.14 -4.99 51 60 3.81 4.50 5.19 Asulam -1.66 9 -1.72 -1.96 -1.03 144.0 9 -0.27 -0.26 1.90 Aureomycin -2.88 60 -4.29 -2.89 0.18 -0.62 -1.04 3.02 Benazolin -2.61 56 -2.44 -2.36 1.34 0.11 Benomyl -4.88 60 -3.85 -3.13 -3.92 2.12 1.79 5.20 Benznidazole -2.81 56 -2.44 -2.27 -3.61 188.5-190 60 0.91 0.48 2.58 Benzoxazole -1.16 56 -2.14 -1.43 -2.94 -2.13 182.5 60 1.59 1.41 2.53 Bromophos -6.09 9 -4.91 -6.32 -4.91 -5.03 53.5 9 5.21 4.95 4.47 Buthidazolc -1.88 60 -2.55 -2.67 -3.09 1.49 1.43 1.47 Carbanilide -3.15 56 -3.55 -3.31 -2.35 -3.23 239.0 60 3.00 3.01 -0.49 Carbophenthion -5.74 60 -5.58 -6.25 -6.47 <25 60 5.33 5.39 6.38 Chloramben -2.47 56 -1.96 -2.33 -2.79 -2.42 194-197 60 1.94 5.87 Chlorbufam -2.62 60 -3.71 -3.415 -3.46 45.5 60 3.44 6.18 Chloroacetonitrile -0.09 57 -0.17 -0.40 0.19 0.11 <25 60 0.45 0.22 4.14 Chlorpromazine -5.10 61 -5.20 -6.06 -6.68 <25 60 5.41 5.20 3.86 Chlortoluron -3.48 57 -2.71 -2.81 -2.64 148.1 60 2.41 2.49 3.39 Cinmetacin -5.54 56 -5.74 -4.58 -6.49 missing 4.25 Clonazepam -3.50 59 -4.08 -3.95 -6.13 237.5 60 2.41 2.37 8.39 Coumaphos -5.38 56 -4.50 -5.67 93.0 60 4.13 4.33 2.57 Cycloheptyl-5-spirobarbituric acid -3.17 50 -2.39 -1.90 -1.84 266.0 50 1.36 1.11 5.45 50 Cyclopropyl-5-spirobarbituric acid -1.89 -0.31 0.13 0.25 325.0 50 -0.53 -1.13 3.35 60 Cypermethrin -8.02 -7.12 -7.68 -6.55 -6.91 80.5 60 6.60 6.35 5.23 9 DDD -7.20 -7.30 -6.68 -7.70 -7.51 109.5 9 6.22 6.06 2.89 60 Deltamethrin -8.40 -7.83 -8.44 -7.21 -7.49 99 60 6.20 6.53 0.04 Dialifor -6.34 60 -5.40 -6.08 -5.98 68 60 4.69 4.06 3.18 9 Diazinon -3.64 -3.06 -4.67 -5.32 -5.29 120.0 9 3.81 3.50 3.28 221 ref ref Compound log Sw ------Prediction method - ...... - Mpt (°C) log Poet® Clogpf WsKow exp Abraham® Meylan** Klopman 1** Klopman 2“* log Poet® Dicapthon -4.31 9 -3.41 -4.82 -3.45 -3.65 52.0 i) 3.58 3.38 4.08 Diclofop -5.04 60 -5.15 -5.10 -4.16 -5.30 120 60 2.81 5.30 5.15 Dienochlor -7.28 60 -8.58 -9.83 -9.93 -9.95 122-123 63 8.14 3.45 Dimefuron -4.33 60 -4.10 -3.81 193 60 2.51 2.52 -0.48 Diosgenin -7.32 59 -7.78 -5.51 -9.44 -7.99 5.07 -0.26 Ditalimfos -3.35 56 -4.74 -4.36 -3.91 3.48 3.48 0.36 Dulcin -2.17 56 -1.73 -1.61 -1.58 1.48 6.21 Endrin -6.18 56 -5.70 -6.42 -7.54 -6.46 226-230 60 5.20 3.63 2.32 Equilin -5.28 61 -3.99 -3.44 -4.40 -5.10 236-240 60 2.90 2.91 57 Ethalfluralin -6.12 -5.39 -5.99 -6.67 56.0 60 5.11 4.47 2.29 Ethofumesate -3.42 56 -2.73 -3.59 -3.34 71.0 60 2.70 2.16 2.94 61 Etoposide -3.57 -2.47 -4.00 -4.32 -3.48 236-251 60 0.60 -1.89 0.98 Fenbufen -5.06 61 -4.14 -3.39 -3.58 -4.56 186.0 60 3.2 3.14 6.34 60 Fenothiocarb -3.93 -4.11 -3.85 -3.82 40-41 60 3.28 2.99 5.67 Fenthion -4.57 56 -3.81 -4.72 -4.18 7.5 60 4.09 3.89 0.00 59 60 Flufenamic acid -4.36 -3.71 -5.74 -5.61 -5.23 133.5 5.25 5.51 3.33 61 Fluopromazine -5.30 -5.20 -6.60 -7.97 5.54 5.53 2.27 60 60 Flurochloridone -4.05 -5.14 -4.346 -6.37 42-47 3.36 3.38 2.45 60 60 Foimothion -2.00 -0.50 -0.664 -1.07 25-26 1.47 1.51 0.85 9 9 glutaric acid 1.00 0.29 0.48 0.52 -0.25 96.5 -0.29 -0.55 2.73 56 60 Glyceryl triacetate -0.60 -0.95 -1.01 -0.58 -0.56 3.0 0.25 0.45 2.32 9 Heptachlor -6.32 -6.56 -7.13 -6.73 -6.87 95.5 9 6.10 4.92 3.73 60 Hydroquinine -3.05 -3.76 -3.50 -6.33 3.42 7.43 59 Imipenem -1.48 -0.58 -1.69 -2.54 missing 3.27 61 Indoprofen -4.82 -4.73 -3.62 -5.04 213-214 60 2.77 2.74 4.39 Ipazine -3.78 60 -4.24 -4.343 -3.69 2.87 4.23 59 Isonazid 0.01 1.38 -0.92 -0.27 -0.40 -0.70 -0.71 4.79 56 karbutilate -2.93 -3.01 -2.88 -2.22 1.66 1.47 3.57 59 Khellin -3.02 -3.49 -3.66 2.35 -0.53 52 Linuron -3.59 -3.32 -3.75 -2.40 94.0 52 3.20 3.00 1.96 56 60 Malathion -3.37 -2.59 -3.62 -2.94 2.8 2.36 2.31 0.45 56 60 Mecoprop -2.55 -2.93 -3.05 -2.25 -2.98 94.5 3.13 3.01 3.01 222 ref ref Compound log Sw ------Prediction method ------Mpt (°C) log Poet® Clogpf WsKow exp Abraham® Meylan** Klopman r Klopman 2"* log Poet® Mefenamic acid -3.78 ~ 1 S ~ -4.02 -5.33 -4.28 -4.26 4.78 2.88 60 Meprobamate -1.81 59 -1.96 -1.39 -0.72 -1.35 105.0 0.7 0.28 1.85 Methoprene -5.19 56 -6.66 -6.16 -5.99 -5.83 <25 60 5.50 5.54 1.17 Methoxychlor -6.89 9 -6.70 -6.06 -6.01 -6.68 89.0 9 5.08 5.17 0.75 Metolazone -3.78 56 -4.54 -3.44 -4.77 2.42 0.21 Metronidazole -1.22 61 -0.62 -0.82 -1.31 160.5 60 -0.02 -0.70 1.03 Monotropitoside -0.74 60 -1.25 -0.56 4.09 2.73 -2.75 1.87 Napropamide -3.57 56 -4.41 -4.05 -4.67 75.0 60 3.36 3.79 3.77 Nimetazepam -3.80 16 -3.26 -3.59 -6.20 156.5-7.5 60 2.16 2.34 -0.33 Nitrazepam -3.80 16 -3.15 -3.56 -5.38 225.0 60 2.25 2.31 1.70 norflurazon -4.04 56 -2.62 -2.15 184.0 60 2.30 2.78 0.80 oryzalin -5.16 9 -4.21 -4.053 -3.95 137 9 0.51 2.71 1.21 Oxazepam -3.95 61 -3.04 -3.20 -4.41 205-206 60 2.24 2.29 5.49 Parathion -4.66 56 -3.68 -4.99 -3.84 -3.94 6.1 60 3.83 3.47 5.52 Peiietierine -0.45 56 -0.68 0.32 -1.35 0.65 2.27 Permethrin -6.29 60 -7.68 -7.60 -7.14 -7.65 34 60 6.50 7.12 -7.60 Phenmedipham -4.81 60 -4.92 -4.46 -2.05 -3.88 143 60 3.59 3.37 -4.46 Phoxim -4.86 60 -3.55 -5.13 -4.30 6.1 60 4.38 -5.13 Piperophos -4.15 60 -4.29 -5.23 -6.23 <25 60 4.04 4.83 -5.23 61 Prochlorperazine -4.40 -4.62 -6.20 -8.63 228.0 60 4.60 6.15 -6.20 Prometon -2.48 60 -2.77 -3.398 -2.17 -2.60 91.5 60 2.99 2.63 -3.40 Propctamphos -3.41 60 -2.81 -3.39 -2.16 -2.25 2.18 -3.39 Pyrazinamidc -0.67 59 0.75 -0.90 -1.16 192.0 60 -0.60 -0.71 -0.90 Pyrolan -2.09 56 -2.19 -2.66 50.0 60 2.05 -2.66 Quinonamid -5.03 56 -3.48 -3.65 -2.65 -3.99 2.96 -3.65 Riboflavin -3.68 61 -3.25 -3.16 -1.46 missing -3.16 Saccharin -1.64 9 -0.83 -2.366 -3.55 228.8 9 0.91 0.52 -2.37 Santonin -3.09 56 -3.26 -2.52 -6.20 1.38 -2.52 Simetryn -2.68 60 -3.48 -3.327 -5.38 82-83 60 1.80 -3.33 Spironolactone -4.17 59 -5.10 -4.17 134.5 60 2.78 2.25 -4.17 Strychnine -3.32 59 -4.11 -3.29 -3.95 287.0 60 1.93 0.20 -3.29 Sulfadimethoxine -2.96 56 -1.96 -2.86 -4.41 203.5 60 1.63 1.17 -2.86 223 ref ref Compound log Sw -.... Prediction method - ...... M pt C O log Poet' Clogpf W sKow exp Abraham^ Meylan'’ Klopman r Klopman 2‘* log Poct^ S6 Sulfamethomidine -2.54 -2.39 -1.87 -3.84 -3.94 146.0 60 0.61 1.16 -1.87 56 Sulfapcrinc -2.82 -2.45 -1.42 -1.35 262-263 60 0.34 0.64 -1.42 59 Suifisoxazole -2.91 -2.57 -2.01 -7.14 -7.65 191.0 60 1.01 0.32 -2.01 61 Talbutal -2.02 -2.07 -2.09 -2.05 -3.88 109.0 60 1.47 1.63 -2.09 9 9 Terbutryn -4.00 -3.54 -4.16 -4.30 104.0 3.74 3.35 -4.16 59 49 Thiamphenicol -2.15 -1.59 -1.57 -6.23 164.3- -0.27 -0.70 -1.57 56 63 Thiram -3.90 -2.45 -1.40 -8.63 155-156 1.76 -1.40 59 60 Triamterene -2.40 -3.03 -1.89 -2.17 -2.60 316.0 0.98 2.06 -1.89 57 60 Trichloroacetonitrile -2.17 -1.34 -2.31 -2.16 -2.25 -42.0 2.09 2.04 -2.31 60 66 Tridecanoic acid -3.81 -4.51 -5.06 -1.16 41-42 5.63 -5.06 56 Triflupromazine -5.30 -5.20 -6.60 <25 49 5.54 5.53 -6.60 Trogiitazone -5.35 58 -6.83 -6.57 -2.65 -3.99 5.05 -6.57 60 Vibramycin -2.85 -3.66 -3.15 -0.02 -2.31 -3.15 52 49 XMC -2.58 -2.25 -2.41 -3.55 99.0 2.23 2.21 -2.41 “ Calculated log Sw using eqn 21, this work (excludes mp term) *’ Calculated log Sw using WsKow for Windows, version 1.26, Syracuse Research Corporation, USA ^ Calculated log Sw using Klopman Model 1 (33 parameter model that is not too general) ** Calculated log Sw using Klopman Model 2 (67 parameter model that is very general) ® Experimental log Poet measurements taken from the Abraham UCL Database or from Clog? for Windows, version 2.0.0b, Biobyte Corp., USA ^ Calculated log Poet using ClogP for Windows, version 2.0.0b ® Calculated log Poet using WsKow for Windows, version 1.26. 224 Chapter 7 Gastrointestinal Absorption 7.1 Introduction In order to produce a pharmacological action, it is required that a drug be of an adequate concentration in the fluid bathing the target tissue. Medicines may be taken orally or by other routes of administration including sublingual dosing, rectal, inhalation, application to epithelial surfaces (skin patches), and injection either intravenously, intramuscularly or subcutaneously. With the exception of the intravenous route, in which the drug is administered directly into the bloodstream, a drug must be initially absorbed from its site of administration before it can enter the bloodstream and be distributed to its various sites of action. The process of absorption is thus of fundamental importance in determining the pharmacodynamic and hence the therapeutic activity of a medicine \ The oral route is the most convenient and hence most popular route of administration, although the many anatomic and physiological factors affecting both the rate and extent of drug absorption from the gastrointestinal tract present a source of variability in systemic response. Drugs administered orally must dissolve in the aqueous digestive fluids and diffuse across the intestinal wall into the hepatic portal circulation for transport through the liver before reaching the systemic circulation which transports the drug to its site of action. The small intestine which is 5-6 metres in length has the highest capacity for nutrient and drug absorption within the gastrointestinal tract and is divided into three regions, the duodenum, jejunum, and ileum. The duodenum comprises the first 20-30cm, whilst the remainder of the small intestine is divided into the jejunum and ileum in the proportion 2:3. The mucosal surface of the small intestine is superbly designed to facilitate efficient passive absorption. In relation to a plain tube this surface is increased 600-fold due to numerous folds, villi and microvilli see Fig.7.01. These finger-like projections are richly supplied with capillary blood vessels, nerves and lymph vessels, which provide an enormous area for contact with, and subsequent absorption of drug formulations designed for oral absorption. The 225 microvillus border of the intestinal epithelial cells looks something like a brush and as such is sometimes refened to as the ‘brush border’. Absorption is not equally effective in the various parts of the GI tract and in extreme cases, only a small region sometimes labelled as an “absoiption window” seems to be responsible for intestinal absorption of a particular drug. Figure 7.01 Villi, epithelial cells that cover the villi and the microvilli of the epithelial cells Epithelial cells Microvilli When a eompound is absorbed from the small intestine it is transported across the intestinal mucosa which consists of a series of separate barriers. The outermost layer is the unstirred water layer (including the mucous layer), followed by the membrane of the epithelial cells, the lamina propria and the endothelium of the capillaries. The principle permeability bamer is represented by the luminal surface of the brush border. 7.1.1 General Mechansims for Drug Transport Across Biological Membranes The principles governing the absorption of drugs from the gastrointestinal lumen are the same as for the passage of drugs across biological membranes elsewhere. In general, there are four different routes (Fig. 7.02) by which molecules may be transported across the intestinal brush border membrane. Two types are distinguished; the paracellular and transcellular routes. 226 Fig 7.02 Four main ways by which molecules can cross cell membranes: AAAAA AA/\AA /. Transport through tight junctions. 4. Carrier-mediated transport 2. Passive transcellular diffusion 3. Endocytosis Paracellular route 1. Transport through tight junctions. Small hydrophilic compounds of very small molecular weight^ (<200) such as urea and sugars are able to rapidly pass into the bloodstream via the junctions between the epithelial cells. Transcellular route 2. Passive transcellular dijfusion. Prediction of intestinal absorption concentrates on this pathway and it is the most common and important mechanism by which drugs traverse the intestinal cell membranes. Passive transcellular diffusion is dependent on the lipophilicity of the compound; the solute requires a certain affinity to lipid structures in order to enter the cell This transmembrane process is simply due to diffusion from a region of high drug concentration to one of low drug concentration and is described by Pick’s first law of diffusion ^ : D.A.(Ch-Cl) (1) d t 227 D = diffusion coefficient, A = surface area of membrane, H = membrane thickness, Ch-Cl = concentration difference (Ch refers to drug concentration in GI tract, and Cl refers to drug concentration in the portal vein). Since the drug concentration in blood or plasma will be low compared with the concentration in the GI tract, i.e. Cl « Ch, eqn 1 may be simplified to eqn 2. For solutes of approximately the same size (i.e. the same diffusion coefficient), permeabilities are directly proportional to the solute partition coefficient (P) between the aqueous solution and the membrane lipid. Thus the absorption of many drugs from the gastrointestinal tract can often appear to be first-order ^ implying that the rate at which a dose is absorbed is proportional to the concentration at the site of absorption. The MW limit for paracellular passage seems to be approx. 400-500 3. Endocytosis. There is evidence that very high MW compounds * such as oral vaccines and peptides and proteins with specific receptors on the cell surface can be absorbed with low efficiency due to endocytosis. This vesicular transport can be divided into pinocytosis and phagocytosis which differ by the type of material ingested. Pinocytosis is a process in which the cell membrane invaginates to surround the material, and then engulfs the material into the cell. Subsequently, the cell membrane containing the material is detached as an intracellular vesicle. The membrane of this vesicle is removed and the contents of the vesicle emptied into the cytosol. Pinocytosis is differentiated from phagocytosis in that pinocytosis involves the capture of extracellular solutes or fluid and is a property of most nucleated animal cells, whereas phagocytosis is refers to the engulfment of macromolecules and is a phenomenon associated with a few specialized cell types such as macrophages and neutrophils. 4. Carrier-mediated transport. Active transport is a carrier mediated process in which the membrane plays an active role, transporting solute molecules against a 228 j concentration gradient. This is an energy-consuming system driven by energy derived from cellular metabolism. There are two general categories of active transport, differentiated on the basis of the immediate source of energy each uses to transport substrate. Primary active transport processes make direct use of energy (usually derived through ATP hydrolysis) at the membrane protein itself to cause a conformational change that results in the transport of the molecule through the protein. A prominent example is the efflux transporter p-glycoprotein Secondary active transport does not use ATP directly. In this case, the transport of the solute is driven by the energy stored in ion concentrations across a membrane. The transmembrane protein couples the spontaneous transport of an ion down its electrochemical gradient to the "uphill" transport of another molecule. For example, the sodium concentration difference produced by primary active transport (Na^-K"^-ATPase) is known to power the secondary active transport mechanism of glucose and amino acids. Active transport is limited by the number of protein transporters present. If the drug concentration becomes very high, all the adsorption sites on the carriers will be occupied and the system will become saturated. Although the transfer system is relatively structure specific, it is subject to competition between similar chemical structures. Facilitated diffusion is also a carrier-mediated transport system, although it does not require an energy input. It differs from active transport in that the drug only moves in the direction of the concentration gradient, but at a much faster rate than would be anticipated based on the polarity and molecular size of the solute. 7.12 Factors Influencing Intestinal Absorption Drugs are absorbed after oral administration as a consequence of a complex array of interactions between the drug, its formulation and the gastrointestinal tract^\ The process of absorption involves the passage of a drug across one or more cell membranes and a basic knowledge of the physical and chemical principles governing this transfer of drugs is therefore necessary. In addition to chemical properties of the drug such as molecular size and shape, solubility of the ionized and non-ionized forms and particle size, factors such as gastric emptying, blood flow, and co-administration of food will play an essential role in determining the overall time taken to reach the active site. These factors that may contribute to the variability in rate and extent of 229 absorption of a drug along the intestine may be categorized as physicochemical or physiological properties, see Fig 7.03. Fig 7.03 Drug, meal and formulation interactions influencing oral drug absorption 12 Formulated drug GI transit time Bile and enzyme secretion Active or passive transport Dissolution Drug in solution Region dependent absorption Degradation Intestinal elimination Diffusion Splanchnic blood flow Physicochemical Physiological First pass effect effect metabolism Food Concomitant drug Drug in the Formulation systemic circulation Dose, volume 7.1.2a Physicochemical Properties Disintegration, water solubility, rate of dissolution and particle size Whilst a solid dosage form must hold together while dry, it must disintegrate (break into granules) rapidly in the intestinal fluids of the intestine. Once the tablet has disintegrated or the capsule coating has dissolved, particles of drug are exposed to the solution, and the release of drug molecules into solution from the solid dosage form is termed dissolution. Dissolution is considered the rate-limiting step in the absorption of most drugs from solid dosage forms'^. For these compounds, dissolution proceeds relatively slowly and although dissolved drug is readily transported across the gastrointestinal epithelium, absorption cannot proceed any faster than the rate at which the drug dissolves. Changes in dissolution will therefore profoundly effect the rate, and sometimes the extent of drug absorption. An efficient means of improving absorption 230 is to hasten the rate of dissolution. This is usually accomplished by physically reducing the particle size of the drug and thereby increasing the surface area exposed to the solvent. For many drugs that are poorly absorbed, the absorption rate is often related to the limited solubility of the drug in the dissolution medium which in effect drives the dissolution process. Even if dissolution were not the rate-limiting step, drugs such as spironolactone would still exhibit a low rate of absorption due to low aqueous solubility. An improvement in absorption may be achieved by converting the drug to a more soluble form. Salts of weak acids and weak bases generally have much higher aqueous solubility than the free acid or base and therefore if the drug can be given as a salt the solubility will be chemically promoted and dissolution should be improved. In addition, dissolution rate is also affected by whether the drug is in anhydrous or hydrate form since the solubility of the solvate may markedly differ from the non solvated form. For example, anhydrous theophylline, gluthemide and caffeine have a greater rate of dissolution than that of the corresponding hydrates whereas the anhydrous form of acyclovir dissolves almost twice as slowly as its hydrated form A general relationship describing the dissolution process was first observed by Noyes and Whitney’^: dc DA (C.y-C) (3) dt hV where dc/dt = rate of drug dissolution, D = diffusion rate constant, A = surface area of the particle, Cs = concentration of drug in the stagnant layer, C = concentration of drug in the bulk solvent, h = thickness of the stagnant layer, and V = volume of the dissolution medium. The Noyes-Whitney equation shows that dissolution kinetics may be influenced by the physicochemical characteristics of the drug, the formulation and the solvent. In addition to these factors, the temperature of the medium and the agitation rate also affect the rate of drug dissolution. Although this dissolution step can be bypassed by administration of a solution, the rate of absorption of a drug from solution may still be limited by perfusion at the site of absorption. 231 Formulation Formulation is an important aspect of design in which a drug intended for oral administration is presented in a dosage form that has the desired characteristics in terms of storage and drug release. Numerous excipients (inert ingredients) are commonly employed to stabilize the drug, facilitate manufacture of the dosage form, maintain drug integrity during handling and storage and facilitate release of drug following administration of the dosage form. For convenience, fine-particles of drugs are usually compressed or compacted into tablets and capsules. However, this manufacturing process must be reversed if the surface area is to be enlarged sufficiently to ensure adequate dissolution To achieve this, capsules and tablets can be designed to swell upon contact with water and disintegrate into granules that finally deaggregate into the original fine particles either in the acid conditions of the stomach or the alkaline conditions of the duodenum. Sometimes part of a dose is coated so that absorption of the parts occurs at different rates producing a sustained release of the drug. These excipients and also the manufacturing process can produce large variations in the rate of dissolution of the drug and more importantly, the extent of absorption the result is a large potential for variability in absorption of a drug between generic products Polymorphism Some drugs are polymorphic existing in amorphous or various crystal forms. Physical properties such as melting point, solubility and thus dissolution rate may vary substantially from one polymorph to another. The metastable polymorph is a higher energy form of the drug and usually has a lower melting point, greater solubility, and greater dissolution rate than the stable crystal form*^. Accordingly, the absorption rate and clinical efficacy of a drug may depend on which crystal is administered. The amorphous form of a drug is always more soluble than the corresponding crystalline forms because the energy required for a drug molecule to transfer from the lattice of a crystalline solid to a solvated state is much greater than that required from an amorphous solid. Chloramphenicol palmitate is one example of a drug which exists in at least two polymorphs, but for which only the B form sufficiently dissolves and is absorbed to be clinically effective^^. 232 Lipid solubility, pH and pK To get from the lumen of the gut to the blood at an appreciable rate, the drug must be sufficiently lipid-soluble to move across the epithelium. The drug’s partition coefficient between the lipid cell membrane and the aqueous cell environment, dissociation constant, as well as the pH at the absorption site often dictate the absorption characteristics of a drug from solution. In 1957, Brodie et al.^^ proposed the pH partition theory to explain the interrelationship among these parameters. Most drugs are weak acids or weak bases, existing in aqueous solution as an equilibrium mixture of non-ionized and ionized species. The ratio of non-ionized to ionized drug when in aqueous solution is pH-dependent and can be calculated from the general form of the Henderson-Hasselbach equation. For a weak base : BH+< 5_>b + H+ [BH^ ] pKa = pH + Iog,o (4) For a weak acid : ^ 0 - 4- AH< +H + pKa = pH-Hlogio^—^ (5) [A"] where pKa is the dissociation constant. Brodie reasoned that the gastrointestinal tract was a simple lipid barrier and accordingly, an ionized species, BH^ or A' has very low lipid solubility and is virtually unable to permeate membranes except, rarely, where a specific transport mechanism exists. Conversely, the lipid solubility of the non ionized form of a drug, B or AH, will depend on the chemical nature of the drug; for the majority of drugs the uncharged species is sufficiently lipid soluble to permit rapid membrane permeation. A solution of the weak acid aspirin (pKa 3.5) in the stomach at pH 1-2 will have greater than 99% of the drug in the non-ionized form and consequently it is lipid soluble and according to the pH- partition theory will be rapidly absorbed in the stomach because it exists largely in a non-ionized form at low pH values. In contrast, most basic drugs are highly ionized in the acid content of the stomach so that 233 absorption is negligible whilst in the near neutral fluids of the small intestine the absorption of weak bases such as codeine (pKa 8) is more favourable\ However, in practice, the area of the absorptive surface is of far more importance. The absorption of all orally administered drugs, whether they be weak acids or weak bases, takes place more rapidly in the small intestine than in the stomach. This is because the small intestine has been calculated to have a total absorptive area of about 200m^ and an estimated 1 litre of blood passes through the intestinal capillaries each min. The stomach has a comparatively small surface area estimated at only Im^ and blood flow of 150ml/min Furthermore, the thick protective mucus layer covering the gastric mucosa provides a poor site for absorption compared with the greater drug permeability provided within the small intestine. Despite its general appeal, the pH partition is somewhat limited in predicting the effect of pH on the rate of absorption because it sees pH as the sole determinant of drug absorption and does not account for such factors as blood flow, absorption surface area and the permeability of the biological membrane; thus, there exist many exceptions to the hypothesis However, the primary limitation of the pH-partition hypothesis is the assumption that only non-ionized drugs are absorbed, when in fact, the ionized species of some compounds can be absorbed (Table 7.01) although at a slower rate Table 7.01 Drug absorption in different pH environments^^ % Absorbed at pKa pH 4 pH 5 pH 7 pH 8 Acids 5-Nitrosalicylic 2.3 40 27 0 0 Salicylic 3.0 64 35 30 10 Acetylsalicylic 3.5 41 27 - - Benzoic 4.2 62 36 35 5 Bases Aniline 4.6 40 48 58 61 Amidopyrine 5.0 21 35 48 52 p-Toluidine 5.3 30 42 65 64 Quinine 8.4 9 11 41 54 234 The hypothesis therefore fails to explain the observation that a variety of quaternary ammonium compounds (eg propantheline bromide) which are always ionized elicit systemic effects when given orally, although movement of these compounds through the gastrointestinal membrane occurs at a slow and erratic rate 7.1.2b Physiological Properties Gastric emptying and intestinal transit In theory, weakly acidic drugs should be better absorbed from the stomach than from the intestine, because a larger fraction of the dose would be in a nonionized lipid- soluble form. However, the limited residence of the drug in the stomach and the relatively small surface area of the stomach more than balance the influence of pH in determining the optimal site of absorption. The time taken for the drug to reach the small intestine is governed in part by the gastric emptying time, which influences the rate and may indirectly affect the extent of absorption. Gastric emptying is highly variable depending on pathological, pharmacological and physiological factors^^. Fasting, diseases such as hyperthyroidism and the anti-emetic agent, metoclopramide. promote gastric emptying and thus tend to increase the absorption rate of all drugs. Conversely, disorders such as migraine, gastric ulcers, and mental depression will slow gastric emtying. In addition, fatty food in the diet and many drugs including propantheline and imipramine will produce the same effect’’^’*^. Physiological properties of the GI tract 12 Region pH Residence time(hours) Stomach 1.5-2 0-3 Duodenum 4.9 - 6.4 3 - 4 Jejunum 4.4 - 6.4 3 - 4 Ileum 6 .5 -7 .4 3 - 4 Colon 7.4 Up to 18 Intestinal transit time is only 3-4 hours and is also an important factor in drug absorption, although transit through the intestine is less variable than gastric emptying. Intestinal transit time is most influential when absorption membranes are poorly permeable to the drug, and the extent of absorption is thus limited by the residence 235 time at the uptake sites. As such, the absorption of drugs that are too polar and hence poorly lipid soluble such as acyclovir, drugs for which dissolution rate is slow such as griseofulvin, and relatively large compounds as neomycin will be affected In addition, intestinal transit time is influential when the drug has a narrow “absorption window”, and when an extended-release system is administered since any release of drug beyond the uptake sites is useless. Blood flow The blood perfusing the gastrointestinal tract plays a critical role in drug absorption by ensuring that any absorbed drug is rapidly transported into the bloodstream as soon as it passes through the intestinal membrane; this continuously maintains the concentration gradient across the epithelial membrane. The dependence of intestinal absorption of drugs on blood flow rate changes from blood flow independent to blood flow limited as the absorbability of the substances increase. Polar molecules that are slowly absorbed show no dependence on blood flow; the absorption of lipid soluble molecules and molecules that are small enough to penetrate the aqueous pores is rapid and highly dependent on the blood flow'^. However, in general, the rate of drug absorption will be unaffected by normal variability in mesenteric blood flow because blood flow is rarely the rate-limiting step in the absorption process. Changes in mesenteric blood flow that result from disease or drug effects must be substantial and sustained to significantly influence drug absorption. Complicating factors : drug-food interactions The presence of food within the GI tract modulates the oral bioavailability of drugs via changes in the rate and/or extent of absorption Historically, the presence of food was regarded as a barrier to absorption and in particular, fatty food was shown to slow gastric emptying which led to suggestions that drugs should be taken on an empty stomach^^ when a rapid onset of action is desired. However, in experiments with rats, prolonged fasting was shown to diminish the absorption of several drugs, possibly by deleterious effects upon the epithelium of intestinal villi It is currently accepted that the interaction between food and drugs should be examined on an individual basis^^’^', and that food can delay, reduce or increase gastrointestinal 236 absorption. It is well accepted that foods with a high concentration of polyvalent metals can directly interfere with drug by binding it and reducing availability, eg. calcium and tetracycline. Furthermore, recent studies indicate that the presence of food can result in an elevation of the viscosity of the contents in the upper GI tract and reduce drug absorption The effect of viscosity varies with the drug, the type and amount of viscosity-inducing agent administered and whether the drug and the viscosity-inducing agent are already dissolved or not. The effect of elevated luminal viscosity on GI absorption appears to be greater for more soluble drugs and results from a decrease in dissolution rate and gastric emptying rate. Drugs that have an intestinal absorption that increases when they are administered with food include those that have an incomplete absorption as a consequence of their poor solubility in the GI fluids. The increased intestinal uptake results from increased secretion of bile salts which may increase the dissolution rate via enhanced wetting or an increase in solubility via micellular solubilization^^. It has been suggested that the reduced absorption of hydrophilic p-blockers is due to their tendency to become tightly associated with bile acid micelles^^. Drugs whose absorption is favoured by bile, salts include griseofulvin^"^, danazol"^^, diazepam^^ and digoxin^^. The effect of food on drug absorption can also be dependent on the type of dosage form used, the excipients and the form of the drug e.g. erythromycin stearate in film coated tablets demonstrated reduced absorption with food, erythromycin estolate in suspension was unaffected by food but absorption of erythromycin ethylsuccinate in suspension and erythromycin estolate in capsules was increased by the presence of food^^ Metabolism in the intestine The enzymes responsible for gut wall metabolism are from two sources : mammalian (including enzymes from gastric and pancreatic secretions and intestinal cells) and bacteria, the latter of which are concentrated in the ileum and colon. The mucosal epithelial cells contain a variety of enzymes involved in the biotransformation of drugs. The highest concentration of mucosal enzymes is in the upper intestine, associated with the villous tips^^. Biotransformation occurs predominantly via 237 cytochromes P450; the major enzyme being CYP3A. It is estimated that between 50% and 70% of currently administered drugs may be substrates for CYP3A and are extensively metabolised by the intestine thus resulting in poor oral bioavailability and subtheraputic effects In addition, it is reported that acid/base hydrolysis can, depending on the solubility of a drug, impact upon its dissolution rate and hence its absorption rate Thus, the propensity of drugs towards certain enzymatic reactions can limit the extent of absorption. 7.1.3 Hepatic Drug Metabolism Although intestinal tissues, lung, kidney, and skin contain appreciable amounts of biotransformation enzymes, the principle site of drug metabolism is assumed to be the liver. Once absorbed from the gastrointestinal tract, an orally administered drug will enter the portal circulation. Drugs are removed from the portal circulation very efficiently by the liver and may be partially or completely metabolised by hepatic microsomal enzymes to less active metabolites; a process referred to as ‘first-pass metabolism’. The metabolic alteration of drug molecules may be divided into two major groups of reactions, phase I and phase II reactions. Phase I reactions are those which produce or introduce a new chemical group onto a molecule and usually consist of oxidation, reduction, or hydrolysis. Phase I reactions occur primarily via a complex enzyme system known as the mixed-function oxygenase system, which resides on the smooth endoplasmic reticulum. Several enzymes are involved, but the most dominant is cytochrome P-450; a haem protein which binds both oxygen and the substrate molecule and is the terminal component of the electron transfer system'^. Metabolism is dependent on the intrinsic activity of these biotransformation enzymes which themselves are under genetic control and also highly sensitive to induction or inhibition by many factors. Thus, hepatic drug metabolism varies widely even among healthy persons. There are now at least eight families of cytochrome isozymes known in human and animals and cytochrome P-450 CYP 1-3 are best known for metabolising clinically useful drugs in humans Once a suitable chemical group (e.g. hydroxyl, amino group) has been revealed or placed onto the drug molecule, it is susceptible to Phase II or conjugation reactions. 238 Phase n reactions involve coupling the drug or metabolite of a drug with an endogenous substrate (eg. glucuronide, sulphate). Examples include the conjugation of salicylic acid with glycine to form salicyluric acid or glucuronic acid to form salicylglucuronide'^. The most common Phase II reaction is glucuronide formation, involving the high energy form of the conjugating agent uridine diphosphoglucuronic acid (UDPGA) which in the presence of UDP glucuronyl transferase combines with the drug to form the conjugate. This reaction occurs with a wide variety of drugs because UDP glucuronyl transferase has a very broad substrate specificity. Other conjugation reactions include acétylation and méthylation reactions which occur via acetyl-CoA and S-adenosyl methionine (and the appropriate transferase enzymes) respectively. Phase I and Phase II reactions generally produce metabolites which are more water soluble than the parent, and thus more readily excreted in the urine or bile. Drugs that are highly metabolised by the liver demonstrate poor oral bioavailability because the amount of unchanged (active) drug that reaches the systemic circulation and is available to exert a pharmacological effect is much less than the amount absorbed into the portal vein. In some instances, the first-pass effect is so great that oral administration of the drug is rendered ineffectual. 239 7.1.4 The Prediction of Human Gastrointestinal Absorption The prediction of human intestinal absorption is a major goal in the design, optimization, and selection of candidates for the development of oral drugs. Modem drug discovery is now focused towards compound selection, based not only on pharmacological activity, but also on seeking favourable ADME (absorption, distribution, metabolism, and excretion) properties, and identifying the candidates with the greatest potential to succeed in the clinicThe growth in drug discovery of combinatorial chemistry methods, where large numbers of candidate compounds are synthesized has led to drug selection becoming a significant bottleneck in the discovery and development of new chemical entities"^^. In order to reduce time spent in the drug selection phase and hence overall project time, there is a necessity for efficient models to select compounds with optimum pharmacokinetic properties. The discovery and development of new drug entities can be facilitated by the use of appropriate in vitro, in situ and in vivo experimental models which are valuable tools in studying the rate, extent and mechanism of intestinal absorption. In vitro methods include everted sacs, brush border membrane vesicles (BBMV’s) and Ussing chambers. However, recently attention has widely been diverted to the human adenocarcinoma cell lines to estimate the barrier function of the gastrointestinal tract. In particular, Caco-2 cell monolayers have been recommended for studies on the prediction of oral drug absorption because these cells display a number of properties characteristic of differentiated intestinal cells and as such express various biological membrane properties including enzymatic and transporter systems"^^. Permeability values are calculated from the change in concentration of the serosal (receiver) solution over time"^^. The advantage of cell models is that they measure both passive and active transport of the drug across a cell membrane, although they cannot quantitatively predict the level of active transport in vivo^®. A number of laboratories have correlated permeability with fraction of drug absorbed^^ and this validates the cell model as a useful method in ranking compounds (particularly within a chemical series) for purposes of drug targeting. In situ methods involve isolated intestinal segments of the rat^^’^^ or dog^"^ and have been used to study the mechanism and rate of drug absorption when such information 240 cannot be readily derived from blood concentration data following oral administration in conscious animals. These perfusion systems base permeability calculations on steady-state disappearance of the compound from the intestinal lumen^\ The advantage is that although the animal has been surgically manipulated and is anaesthetised, the blood flow, neural and endocrine input are maintained intact and viability is not as much a shortcoming as for in vitro systems^^. Thus, non- physiological diffusion of drug through the submucosal and muscularis layers of the intestine does not occur. Furthermore, measurement of metabolites in the intestinal lumen has been made in situ when metabolites could not be observed due to enzyme inactivation in vitro^^. An important disadvantage of this system is the number of animals required to establish perfusion conditions that minimize within-treatment variability^^. In vivo techniques are generally less accessible due to experimental difficulties. However, direct measurements of human in vivo drug permeability is available from the Lennemas group using the LOC-I-GUT^^'^^ technique. This system makes use of an instrument consisting of a multichannel tube with two inflatable balloons separated by a 10cm long segment and enables single-pass perfusion of a well-defined region of jejunum. A solution of the compound of interest and a non-absorbable marker is infused into the intestinal segment. Samples of the perfusate are collected and the fraction disappearing from the perfusate when it has passed through the intestinal segment is assumed to be absorbed. The method allows accurate and direct estimates of the local absorption rate and there exists a good correlation between these measured human effective permeability values and the extent of absorption of drugs in humans determined by pharmacokinetic studies^^. The mentioned cell membrane methods^*'^^, in situ animal studies^' and in vivo human perfusion technique provide a valuable guidance in the selection of lead compounds for further development. The widespread use of cell membrane models has been fuelled by the need to find viable alternatives to animal testing. A further advantage of cell-based models is that they are more amenable to higher-throughput screening and are therefore more able to support drug discovery efforts. However, a drawback shared by all experimental methods is that they require synthesis of sufficient amounts of the compound, and as such are still costly, labour intensive, and time-consuming. 241 To overcome these limitations during the early discovery phase, computational methods have been sought to provide information concerning the vast number of compounds generated by combinatorial chemistry approaches and to provide broad synthetic direction. A plethora of quantitative structure-activity relationships (QSARs) relating the biological activity of a compound to its physicochemical properties exist; although those concerning human intestinal absorption as the dependent variable are comparatively few; notable examples include Lipinski’s Rule of 5, polar surface area and neural networks. Lipinski’s “rule-of 5” was the consequence of an objective at Pfizer to set up an absorption-permeability alert procedure that was pragmatic and could guide medicinal chemists^^. This was achieved by isolating 2245 drug compounds from the World Drug Index (WDI) computerized database. These compounds were chosen because they were likely to have entered the Phase II process and were thus considered to have superior physicochemical properties. Parameters that were likely related to absorption or permeability were calculated for each of the compounds and the distribution of these calculated properties was then analysed. Since the critical threshold values were all close to 5 or multiples of 5, the mnemonic “rule-of 5” was coined. The “rule-of 5” has since proved very popular as a rapid screen^^’^^ and identifies compounds likely to exhibit poor intestinal absorption as those that satisfy any of the following two rules (1) Molecular weight > 500, (2) Number of hydrogen bond donors >5 (a donor being any O-H or N-H group), (3) Number of hydrogen bond acceptors >10 (an acceptor being any O or N including those in donor groups) and (4) Clog P >5.0 or MlogP > 4.15. However, these rules are by no means stringent and some compounds will lie outside the parameter cuttoffs. Orally active therapeutic classes outside the rule are those that contain members which have structural features allowing the drug to act as substrates for naturally occurring transporters, namely antibiotics, antifungals, vitamins and cardiac glycosides^^. Palm^’^^ has developed a theoretical method based on the determination of dynamic surface properties of drug molecules to predict human intestinal absorption. The polar surface area (PSA) is defined as the van der Waals’ surface area contributed by nitrogen and oxygen atoms, plus the area of the hydrogen atoms attached to these heteroatoms^"^. The dynamic polar surface area (PSAd) which is the statistical average 242 polar surface area of the low energy conformations of a molecule was calculated for a set of 20 structurally diverse drug compounds. An excellent inverse sigmoidal relationship between the fraction of drug (%FA) absorbed in humans and PSAd was obtained, and analysis of the data revealed that drugs with almost complete absorption (>90%) had a PSA < 60Â^ while drugs exhibiting < 10% absorbed had a PSA >104 Â^. However, the disadvantage of this model is the long calculation times for the low energy conformations of large and flexible compounds. This limitation was overcome by Clark^^ who argued that use of calculations from a single global minimum conformation could be just as effective but far less computationally expensive; the result of using PSA instead of PSAd for the 20 compounds studied by Palm was shown to be minimal. When the dataset was extended to 74 compounds, the sigmoidal relationship between PSA and (%FA) was not so pronounced and the model predicted seven compounds to be poorly absorbed when in fact they were well absorbed. However, these compounds were actively transported and as such are not true false negatives within the scope of the PSA predictions which are applicable only to passively transported molecules; a caveat which is pertinent to all computational methods. A third computational method worthy of note is that of Wessel et al.^^ who used an artificial neural network (ANN) to model human intestinal drug absorption. In this approach, 127 descriptors were calculated for each of the 67 compounds in the training set and fed to a genetic algorithm (GA) neural network feature selection routine with the aim of developing a nonlinear model. This GA algorithm reduced the large pool of descriptors to a subset of 6 independent descriptors which were considered to best describe human absorption : (1) Number of single bonds, (2) Normalized 2D projection of molecule on YZ plane, (3) Charge on donatable hydrogen atoms, (4) Surface area of hydrogen-bond acceptors atoms/number of hydrogen-bond acceptor atoms, (5) Surface area x charge of hydrogen-bond acceptor atoms/number of hydrogen bond acceptor atoms, and (6)Cube root of gravitational index. For the combined training and cross-validation set of 76 compounds, absorption was calculated with a RMSE (root mean square error) of 9%. For the external test set of 10 compounds, which gives the best measure of the quality of the model, the RMSE of prediction was quite significantly higher at 16%. 243 The work presented in sub-chapters 7.2 - 7.5 has been conducted conjointly with Dr. Yuan Hui Zhao. I wish to express my thanks to Yuan for undertaking the major proportion (60%) of absorption data evaluation. I also gratefully acknowledge his contribution (50%) to the data analysis and subsequent derivation of Abraham models. 7.2 Human Intestinal Absorption As indicated in Chapter 7.1, drug absorption is a complex process that is dependent upon numerous biochemical, physiological and physicochemical factors. In a thorough review of the subject, Sietsema^^ accurately points out that the terms absorption and bioavailability are often incorrectly and interchangeably used. Sietsema defined absorption as “the drug passing from the lumen of the gastrointestinal (GI) tract into the tissue of the GI tract. Once in the tissue, the drug is considered absorbed”. 7.2.1 Evaluation of Human Absorption Data The largest data set of compounds yet analysed is that of Wessel et al.^^ who constructed a total set of 86 compounds. However, there was potentially much more data available, and the first aim of the present work was to collect data from the literature and, unlike previous work, to assess the data in order to obtain a much larger data set that could be analysed with some confidence. The second aim was to construct a QSAR for human intestinal absorption that could be used as a rapid screening method for candidate drugs. The names of drug and drug-like compounds and their related data are listed in Table 7.02 (pages 254-258), and Table 7.03 (pages 259-268). The absorption data was collected and evaluated from 244 and the following information concerning human drug absorption was recorded from these literature sources: • Absorption data given in the literature; • Oral bioavailability or absolute bioavailability; • Fraction of cumulative urinary excretion of unchanged drug and its metabolites following oral and intravenous administration; 244 • Fraction of metabolites in urine or first pass effect following oral and intravenous administration; • Fraction of unchanged drug in urine following oral and intravenous administration; • Fraction of excretion of drug in bile following oral and intravenous administration; • Fraction of cumulative excretion of drug in faeces following oral and intravenous administration; • Total recovery of drug in urine and faeces following oral and intravenous administration; • Single dose level in mg or mg/kg and daily oral dose in mg. Surveying the numerous papers, it was found that the absorption data were obtained by different methods. However, most of the absorption data from the literature were based on one of the three main methods outlined in the following text. This data forms the basis by which all methods of predicting absorption are judged or devised from. Method 1: Bioavailability Bioavailability measurements are one method of obtaining absorption data (Drugs 1-2, 10, 39-40, 64, 76-77 and 96-98). If bioavailability is high (>80%), it can be assumed that the bioavailability of the drug reflects absorption because the effect of first-pass metabolism is minimal and almost all the absorbed drug can reach the systemic circulation. However, this method may underestimate absorption if the bioavailability of the drug is low (Drug 241) because a fraction of the absorbed drug may not reach the systemic circulation. This is because following absorption from the gastrointestinal tract, the drug passes directly to the liver via the hepatic portal vein where it may be extensively metabolised before reaching the systemic circulation. Method!: Excretion in urine and faeces following oral administration The percentage of cumulative excretion of drug in urine or faeces following oral administration was another common method used to derive absorption data. For example, absorption data was derived from the percentage of cumulative urinary excretion of unchanged drug and its metabolites (Drugs 3, 12, 15, 45-47, 69-71 and 84-89), percentage of cumulative faecal excretion of unchanged drug (Drug 146) or drug-related material (Drugs 44, 72 and 224), and the range between the percentage of 245 parent drug in urine and 100-percentage of metabolites excreted in the faeces (Drug 234). There were fewer excretion data in faeces available in comparison to urinary excretion data. If the drug could be completely recovered from the urine and faeces, and urinary excretion was the main elimination route for the absorbed drug, use of this method is correct. However, this method would result in significant mis-estimation if one of the following two cases occurred. (1) The drug could not be completely recovered in urine and faeces. Deficit fraction of the drug may still be in the human intestinal tract or the absorbed drug may not have fully been excreted in urine because of the time limit (Drugs 147 and 224).^°^ This can be seen from the total recovery of drugs 28, 44, 60, 74, 79, 109 and 122 following oral administration, and drugs 74, 79, 109, 138, 139 and 203 following intravenous administration. (2) If the urinary data is used when the absorbed drug is also excreted by a route other than the urine. For example, following intravenous administration, less than 50% of certain drugs (Drugs 2, 151, and 207) were shown to be excreted in the urine. Thus a fraction of the absorbed drug could also be excreted in the faeces. In this respect, if it cannot be proved that urinary excretion is the main route of excretion for absorbed drug (i.e. by intravenous administration), it is difficult to state whether absorption obtained by this method is reliable (Drugs 236-240) In order to investigate the excretion route of absorbed drug, regression analysis was carried out using the percentage of excretion in urine and faeces (Table 7.03) following intravenous administration. The results indicate that the percentage of urinary excretion decreased, or faecal excretion increased with an increase in the octanol-water partition coefficient, especially for drugs with ClogP>0 (Fig. 7.04). The logarithm of the octanol-water partition coefficient (ClogP) was calculated using ClogP for Windows software (Biobyte version 2.0.0b, Claremont, CA). The result suggests that the more hydrophobic a drug, the more likely it is to be excreted in the faeces. Fig. 7.04 may suggest that if ClogP is larger than zero, the absorbed drug would not be completely excreted in urine, but would also be excreted in the faeces. Therefore, the use of the urinary excretion method would likely result in mis-estimation of absorption for highly hydrophobic drugs. 246 Fig. 7.04 Dependence of urinary excretion of drug-related material following intravenous administration on ClogP Method 3: The ratio of cumulative urinary excretion of drug-related material following oral and intravenous administration The ratio of cumulative urinary excretion of drug-related material (parent drug and its metabolites) following oral and intravenous administration was also used to evaluate the absorption of drug, e.g. drugs 22, 29, 32-36, 38, 50-51, 66 and 79-80. This method is preferable to the urinary excretion method because the absorption data can be estimated more accurately even if the urinary excretion of absorbed drug is not the main route of elimination, or if the drug is not completely recovered from urine and faeces. The greater the extent of urinary excretion of drug from intravenous administration, the greater the validity of the absorption data. Greater error in obtaining an absorption value would be associated with drugs in which both urinary excretion fractions from oral and intravenous administrations were very low. Although this method is better than the urinary recovery method, the intravenous administration of some drugs has not been determined in For some drugs, intravenous administration may not be possible due to low aqueous solubility.^^^ If the drug undergoes extensive hepatic metabolism, the absorption cannot be accurately evaluated by the ratio of urinary excretion of parent drug since absorption will be under-estimated. For example, the absorption evaluated by the ratio of cumulative urinary excretion of mercaptoethane and fenoterol is 75% and 60%, respectively. However the absorption evaluated by the ratio of urinary excretion of the 247 parent drug is 25% and 7%, respectively. The estimation error between the two approaches is therefore 25% and 53% for these two drugs. As a result, absorption evaluated by the ratio of cumulative urinary excretion of parent drug (Drugs 229-233) cannot be considered to be reliable. The following cases may result in mis-evaluation of absorption: (1 ) Low solubility and dose-limited absorption Water solubility is an important factor in drug absorption. Phenytoin is well known for its poor solubility in water; its absorption varies considerably among different preparations and dosages.^^^ For dose-limited drugs with poor solubility, incomplete dissolution and hence incomplete absorption may occur. Absorption is highly variable for these drugs (Drugs 202-219).^*°’^^*’^^^ For instance, the excretion in faeces of fosinopril varies from 63-81% and total cumulative excretion in urine and faeces of spironolactone varies from 40-95% following oral administration. One reason for this variability may be the hydrophobicity of the drug which would prevent complete dissolution in the intestinal fluid; certainly, the lack of drug dispersion is one cause of incomplete absorption^^’. For instance, the volume of water required to dissolve 20 mg of lovastatin is >15L (The solubility is 0.0013g L'^ at pH 5.0 and 23 Some of the drugs were completely insoluble in the intestinal fluid (Drugs 181, 207, 209 and 210). Oily and alcoholic solution were used to dissolve drugs 202 and 209. To correct for low solubility, Dressman^^^ introduced the absorption potential (AP). With this approach, log P is corrected for molar fraction of nonionized species at pH 6.5 (Fnon), the solubility of the nonionized species in water (Sw), the volume of the luminal contents (V J and the dose administered (X q ). AP = log ( PxFnonXSwxViyXo) (6) Doses received by subjects listed in Table 7.03 showed that 20 drugs were dosed singly over Ig (Drugs 20, 27, 30, 37, 38, 76, 101, 111, 121, 130, 144, 162, 163, 179, 188, 189, 212, 214, 218 and 228); the highest single dosage being lOg (Drug 38). However, when a large dose was given orally, urinary excretion and bioavailability decreased greatly.^^^’^^^’^^^ This could be understood on the basis of an "absorption 248 window" effect/^^ The question then arises as to whether these drugs were completely dissolved in the intestinal fluid. Generally, the average weight of subjects was taken to be 70kg for pharmacokinetic studies. The subjects either received single dosed drugs with 200ml water (usually 100~250ml), or subjects received the drugs three times a day. Because small intestine volume is assumed to be 250ml the percentage of un-dissolved drug for a single dose in 250ml water [100x(l-0.25xSw/Dose)] was calculated and listed in Table 7.03. Since there were few experimental values available, the water solubilities (Sw) at 25°C listed in Table 7.02 were estimated by use of both the WsKow for Windows program, version 1.26, SRC (William Meylan 1994-1996)^^^ and the Abraham method (eqn 26 derived in Chapter 6.3.2, page 187). However several compounds in the dataset have Vx>4 which is beyond the descriptor space and hence validated prediction accuracy of the Abraham method for solubility prediction. As such, the Meylan method was chosen as the best model for subsequent absorption correction calculations to maintain consistency in solubility prediction (where experimental values were unavailable). The results showed (Table 7.03) that there were 37 drugs for which the insoluble percentage was larger than 90% in 250ml water. However, absorption is not a partition process and water is not the same as intestinal fluid. Furthermore, large estimation errors in the Meylan solubility calculations may arise for the larger and more complex drug molecules. Table 7.03 also lists the ratio between the amount of insoluble drug and dose in 250ml water after absorption {100x[Dosex(l- fraction absorbed) - 0.25xSw]/Dose}. The results showed that there were 22 drugs for which the ratio was still over 20% after absorption. These drugs were dose-limited because of their low solubility and absorption. However, for a drug which has low solubility but a high absorption of 80%, the absorption of drug will still be reliably estimated. This is because, after absorption, the insoluble drug would represent less than 20% of the administered dose (Drugs 1, 2, 7, 9, 13, 18, 40, 44, 46, 60 and 63). (2) Formulation-dependent and salt dependent Many drugs were not administered orally as the free base or acid because of their poor water solubility, stability, hygroscopicity, crystallinity or purity. These drugs were 249 usually combined with acids or bases to form a salt, or formulated with a lipophilic solvent, a hydrophilic solvent, and a surfactant that interact to aid in dispersion and émulsification. Therefore, the absorption of some drugs was formulation-dependent or salt dependent.^^^ (3) Dose-dependent Dose-dependent absorption of the drug is based on the observation that the percentage of the oral dose absorbed, bioavailability and excretion in urine declined with increasing dose (Drugs 220-226). Variable urinary recovery or bioavailability with dose may reflect a variation in absorption. (4) The drugs were metabolised in intestinal tract Certain drugs may be metabolised by enzymes or microflora that reside in the gut or gut wall and it is difficult to quantify this intestinal metabolism. Biotransformation in the intestine occurs predominantly via cytochromes P450; the major enzyme being CYP3A. More than 50% of drugs may be substrates for CYP3A, thus resulting in poor oral bioavailability due to extensive metabolism in the intestine,^^^'^^^ e.g. drugs 35 and 228. Absorption data can therefore be unreliable if metabolism occurs in the intestine. Because the absorption data obtained from the literature was from different methods, it seems unrealistic to expect to find a single model that will accurately predict all classes of compounds if the absorption data is not classified carefully"^’^^. Therefore, it was imperative that the data be properly evaluated from the original papers before embarking on QSAR studies. The absorption data chosen for QSAR modelling in Table 7.03 (%Abs."’) was based on one of the following methods: 1. BIO: Absorption was obtained from bioavailability values after oral administration. If the bioavailability was low, the absorption should be equal to or higher than the value of bioavailability (Drug 241). 2. RA: Absorption was evaluated from the ratio of urinary excretion of drug-related material following oral and intravenous administration. 250 3. RAP: Absorption was evaluated from the ratio of urinary excretion of parent drug following oral and intravenous administration 4. EU: Absorption was obtained from cumulative urinary excretion of drug-related material following oral administration. If the urinary excretion was low (<80%) and it could not be proved that urinary excretion of absorbed drug was the main route of elimination, or nearly all the drug was recovered in urine and faeces, the absorption should be equal to or higher than the percentage of urinary excretion of the drug (Drugs 236-240). 5. EF: Absorption was obtained from the excretion in faeces (100-% excreted in faeces). 6. EUB: Absorption was obtained from the cumulative excretion of drug in urine and bile. 7. REV: Absorption was obtained from review papers. 8. EU~EF: Intravenous administration showed that nearly all of the drug was excreted in urine or that excretion in bile was low. However the drug was not completely recovered in urine and faeces. Thus the absorption value should lie between the percentage of excretion in urine and faeces (100 - % excreted in faeces). The following key was used as an indication of the quality of the data based on the above analysis: Good: 1. The absorption data is evaluated based on the ratio of cumulative urinary excretion of drug-related material following oral and intravenous administration. The result from intravenous administration showed that the percentage of urinary excretion was greater than 20%, or 2. Drugs with bioavailability >90% (Absorption would be 90-100. Estimation error would be less than 10%), or 3. Cumulative urinary excretion following oral administration is higher than 90% (Absorption would be 90-100. Estimation error would be less than 10%) or cumulative faecal excretion is less than 10% or 4. Absorption value taken from review papers, provided that the quoted absorption was greater than 80%. 251 OK: 1. The bioavailability is 80-90% (Absorption would be 80-100. Estimation error would be less than 20%), or 2. The percentage of cumulative drug-related material in urine is 80-90% (Absorption would be 80-100. Estimation error would be less than 20%) or the cumulative faecal excretion is 10-20%, or 3. The absorption data is evaluated based on the ratio of urinary excretion of parent drug following oral and intravenous administration and percentage of urinary excretion of parent drug following intravenous administration is higher than 70%. The absorption may be under-estimated by this approach. For example, the absorption evaluated by ratio of cumulative urinary excretion of sorivudine related material and parent drug is 82 and 66, respectively. There is 16% estimation error between the two approaches. This approach may (partly) reflect the true absorption if the drug has a higher intravenous administration. OK? (Uncertain): 1. Based on the analysis of the excretion in urine following intravenous administration (Fig. 7.04), the Clog P of these drugs is lower than 0. Therefore, the urinary excretion may be the main route for the absorbed drug. If the drug could be completely recovered in urine and faeces, the percentage of excretion in urine would reflect the absorption of the drug, or 2. Bioavailability, or the urinary excretion of drug-related material is between 70- 80% (Estimation error would be less than 30%). DP: The absorption of the drugs is dose-dependent based on the literature. DL: The drugs are dose-limited and more than 20% of the drug is still insoluble in 250ml water after absorption. Absorption is highly variable and incomplete for these drugs. DL?: The drugs may not be dose-limited after the correction of solubility by pKa at pH=6.5. 252 FD; The absorption is variable depending on the formulation of the drug (formulation- dependent). M: The drugs were metabolised in the intestine before passing through the membrane. IVL: The excretion in urine is so low following intravenous administration (<20%) that the absorption data may not be reliable based on the method of the ratio of urinary excretion following oral and intravenous administration. In summary, it is difficult to be absolutely certain whether some of this data refers to true absorption figures. If metabolism occured during passage across the gastrointestinal tract, the absorption data would not be reliable if derived from the ratio of urinary excretion of drug-related material following oral and intravenous administration.Although the cut-off point for good bioavailability and urinary excretion data has been specified as 80%, data down to a level of 70% may also reflect the true absorption (Drugs 152-154). Whilst there is little doubt that some drugs are dose-limited, it is difficult to give a definition of a dose-limited drug because many factors can affect the solubility in the intestinal fluid and absorption in humans, such as the absorption mechanism"^, drug formulation, food composition, chemical composition, pH of the intestinal secretions, gastric emptying time, intestinal motility and blood flow.^°^ The pharmacokinetics of some drugs were extremely complex and knowledge of their in vivo behaviour is still far from complete.^^^ Nadolol seems to be a dose-dependent drug.^°^ Some absorption data have not been checked because the original papers have not as yet been found (Drugs 164-172 and 241). Nevertheless, a great deal of absorption information contained in each of the original references was extracted and recorded in the hope that this would be of value for subsequent analysis using the Abraham General Solvation Equation. 253 Table 7.02 Molecular weight (MW), experimental and calculated water solubility using the Meylan method (Sw) and Abraham Method (SwAb) and octanol-water partition Coefficients (MlogP and ClogP) No Names MW Sw(Calc)“ SwAb(Calc)" MlogP'' ClogP' (mgU') (m gU') Training set 1® 1 Cisapride 466 2.71 1.19 3.43 2 Valproic acid 144 12547895 1052 2.75 2.76 3 Salicylic acid 138 208973808 2361 2.26 2.19 4 Diazepam 285 51759 12.26 2.99 3.29 5 Sudoxicam 337 3015 1277 1.64 2.60 6 Glyburide 358 35 0.04 4.08 7 Gallopamil 485 0.52 0.25 3.14 8 Mexiletine 179 8248 295 2.15 2.57 9 Nefazodone 470 0.060 0.07 5.00 ^ 10 Naproxen 230 14.517145 31.54 3.34 2.82 11 Lamotrigine 256 3127 180 3.24 12 Tolmesoxide 214 10850 11660 0.89 13 Disulfiram 296 4.09764 14.37 3.88 3.88 14 Torasemide 348 137 109 3.34 15 Metoprolol 267 4777 192 1.88 1.20 16 Naloxone 327 1415 1460 2.09 -0.04 17 Terazosin 387 205 1085 2.71 18 Sulindac 356 3.56717 2.81 3.05 2.81 19 Sultopride 354 724 1703 1.93 20 Topiramate 339 13640 4907926 -0.07 21 Tolbutamide 270 1097183 158 2.34 2.50 22 Propiverine 367 8.85 0.44 4.06 23 Digoxin 781 6573.78 1.26 1.32 24 Mercaptoethanesulfonic acid 142 1000000 227855 -0.52 25 Cimetidine 252 6186710460 141307 0.40 0.35 26 Furosemide 330 149 357 2.03 1.87 27 Metformin 129 1000000 129051040 -2.64 ^ 28 Rimiterol 223 399500 16085 0.36 29 Cymarin 548 99 -0.15 30 Ascorbic acid 176 1000000 5324998 -1.64 -2.21 31 Fosfomycin 138 960700 79848468 -0.48 32 Fosmidomycin 183 1000000 87688692 -3.11 33 k-strophanthoside 873 20510 -5.42 34 Adefovir 273 42380 965514 -2.08 35 Acarbose 646 1000000 -10.62 36 Ouabain 584 10340 -1.70 -4.58 37 Kanamycin 484 1000000 8772477 -7.77 38 Lactulose 342 1000000 1616483 -5.56 Test set 39 Camazepam 372 24 23.32 4.27 3.64 40 Indomethacin 358 5.1773.11 1.30 4.18 41 Levonorgestrel 312 36 2.08 3.31 42 Tenoxicam 337 442 3992 2.42 43 Theophylline 180 733372800 70212 -0.02 -0.06 44 Oxatomide 426 0.60 ■ 0.29 5.41 254 No Names MW Sw(Calc)" SwAb(Calc)' MlogP'^ ClogP' (mgU') (mg L‘‘) 45 Desipramine 266 58^0.99 2.04 4.90 4.09 46 Fenclofenac 297 41.95^2.52 3.30 4.80 4.96 47 Imipramine 280 18^1 1.61 4.80 4.41 48 Lormetazepam 335 94 13.50 2.60 49 Diclofenac 296 2.35V5.6I 11.05 4.40 3.03 50 Granisetron 312 28 779 1.79 51 Testosterone 288 27.5^68 7.86 3.32 3.22 52 Caffeine 194 25574^2632 45017 -0.07 -0.06 53 Corticosterone 346 199^143 26.14 1.94 2.32 54 Ethinylestradiol 296 116 4.78 3.67 3.66 55 Isoxicam 335 1046 2349 2.83 2.40 56 Lomoxicam 372 1572 1070 3.15 57 Nicotine 162 1000000 9941 1.17 1.32 58 Ondansetron 293 5.70 52.04 2.64 59 Piroxicam 331 23V52I 1993 1.98 2.70 60 Verapamil 455 4.47 0.23 3.79 3.71 61 Progesterone 314 l l . 94V5.OO 1.65 3.87 3.78 62 Stavudine 224 6595 105005 -0.81 -0.48 63 Toremifene 406 0.072 0.00 6.35 64 Cyproterone acetate 417 0.65 0.75 3.39 65 Praziquantel 312 88 70 3.43 66 Cicaprost 374 21 0.79 2.01 67 Aminopyrine 231 4227 745 1.00 68 Nordiazepam 270 57 91 2.93 3.01 69 Carfecillin 454 2.91 22 2.96 3.12 70 Prednisolone 360 238V221 87 1.62 1.64 71 Propranolol 259 609 34 2.98 2.75 72 Viloxazine 237 29210 4823 1.34 73 Warfarin 308 I7V17 9.37 2.70 2.44 74 Atropine 289 2192V3944 175 1.83 1.32 75 Minoxidil 209 2189V3423 29199 1.24 1.09 76 Clofibrate 243 21 30 3.68 77 Trimethoprim 290 4OOV2334 501 0.91 0.95 78 Venlafaxine 277 267 42 2.11 79 Antipyrine 188 986638^/23760 1536 0.38 0.41 80 Bumetanide 364 IOOV32 22 3.90 81 Trapidil 205 1696 228 1.94 82 Fluconazole 306 1086 3564 -0.11 83 Sotalol 272 136800 3182 -0.44 0.23 84 Codeine 299 9030Vl3400 639 1.14 0.82 85 Flumazenil 303 128 4912 1.06 86 Ibuprofen 206 36V2440 21 3.50 3.68 87 Labetalol 328 73 7.46 2.50 88 Oxprenolol 265 3182 287 2.10 1.69 89 Practolol 266 4472 1181 0.79 0.75 90 Timolol 316 2741 5264 1.83 1.61 91 Alprenolol 249 547 67 2.89 2.65 92 Amrinone 189 8067 15798 -0.59 93 Ketoprofen 254 176V120 20 3.12 2.76 94 Hydrocortisone 362 294V219 69 1.61 1.70 95 Betaxolol 307 451 14 2.81 2.17 96 Ketorolac 255 572 - 522 1.62 97 Meloxicam 351 3.60 432 3.01 3.10 255 No Names MW Sw(Calc)" SwAb(Calc)' MlogP“ ClogP' (mgU') (mg L'*) 98 Phenytoin 252 20.02^1267 109 2.47 2.08 99 Amphetamine 135 28000 828 1.76 1.59 100 Chloramphenicol 323 2507*’/389 1611 1.14 0.69 101 Felbamate 238 6116 656 -0.29 102 Nizatidine 331 77690 27617 0.50 103 Alprazolam 309 13 12 2.12 2.30 104 Tramadol 263 1151 304 2.63 2.31 105 Nisoldipine 388 25 2.56 4.53 4.24 106 Oxazepam 287 32*’/179 259 2.24 2.29 107 Tenidap 321 2676 7.52 0.63 ^ 108 Dihydrocodeine 301 6866 508 1.30 109 Felodipine 384 20 2.99 4.80 4.96 110 Nitrendipine 360 77 19 4.15 3.39 111 Saccharin 183 4192^789 23951 0.91 0.52 112 Moxonidine 242 16450 3439605 1.02 113 Bupropion 240 140 143 3.21 114 Pindolol 248 7883 1179 1.75 1.67 115 Lamivudine 229 1000000 179170 -0.93 -1.54 116 Morphine 285 150Vl3810 1289 0.76 0.24 117 Lansoprazole 369 3.43 1035 3.07 118 Oxyfedrine 313 3.43 428 2.84 119 Captopril 217 160864^6857 12513 1.19 120 Bromazepam 316 1394 683 1.69 1.69 121 Acetylsalicylic acid 180 4600 6600 1.19 1.02 122 Sorivudine 349 1207 67500 ■ -1.66 123 Methylprednisolone 374 123 35 1.96 124 Mifobate 359 3298 342402 0.69 125 Flecainide 414 1.48 59 4.43 126 Quinidine 324 502^104 72 2.64 2.93 127 Piroximone 217 11890 7671 0.96 128 Acebutolol 336 259 148 1.71 1.63 129 Ethambutol 204 723819^948800 86388 0.12 130 Acetaminophen 151 30350 19929 0.51 0.49 131 Dexamethasone 392 IOIV93 34 2.01 2.01 132 Guanabenz 231 1055 17088 2.96 133 Isoniazid 137 16700 2767987 -0.70 -0.71 134 Omeprazole 345 82 1592 2.23 2.53 135 Methadone 309 48 0.81 3.93 3.13 136 Famciclovir 321 2609 934 -0.36 137 Metolazone 366 6IV133 10.39 2.42 138 Fenoterol 303 41370 271 0.83 139 Nadolol 309 22400 291 0.71 0.23 140 Atenolol 266 685 2332 0.16 -0.11 141 Sulpiride 341 450V2275 2737 1.11 142 Metaproterenol 211 973500 32287 0.08 143 Famotidine 337 1271 8083 -0.57 -0.56 144 Foscamet 126 1000000 394605147 -1.93 ^ 145 Cidofovir 279 1000000 736098 -3.56 146 Isradipine 371 49 23 4.18 3.57 147 Terbutaline 225 212800 13262 0.08 0.48 148 Reproterol 389 1424 814 -0.98 149 Lincomycin 406 927 • 6874 0.20 -0.12 256 No Names MW Sw(Calc)“ SwAb(Calc)' MlogP" ClogP' (mg U') (mg U') 150 Streptomycin 582 1000000 -7.17 151 Fluvastatin 411 0.47 0.15 3.19 152 Urapidil 387 157 953 2.56 153 Propylthiouracil 170 514 21426 2.80 154 Recainam 263 2158 71 1.13 155 Cycloserine 102 166000 42692233 -1.72^ 156 Hydrochlorothiazide 298 722^1292 27585 -0.07 -0.40 157 Pirbuterol 240 521000 124447 -0.93 158 Sumatriptan 295 21360 1296 0.93 0.58 159 Amiloride 230 1256 1075484 -0.26 160 Mannitol 182 216300Vl000000 5105132 -3.10 -4.67 161 Ganciclovir 255 28340 120275 -2.07 -2.99 162 Neomycin 615 1000000 -9.03 163 Raffmose 504 196079^1000000 10065 -7.96 164 Phenglutarimide 288 325 130 1.54 165 Bomaprine 329 2.16 0.56 4.30 166 Scopolamine 303 100333^17400 2463 0.26 167 Ziprasidone 413 2.13 0.56 4.42 168 Guanoxan 207 16120 379348 0.33 169 Netivudine 282 5918 241465 -2.03 170 DphenylalanineLproline 171 Noloxone 172 Gentamicin-Cl 484 1000000 92579954 -3.77 Zwitterionic drugs 173 Cefadroxil 363 1110 9602 -2.06 -2.57 174 Ofloxacin 361 28260 36002 -0.24 175 Pefloxacin 333 11390 8552 0.27 0.08 176 Cephalexin 347 1789 9087 -1.74 -1.90 177 Loracarbef 349 2785 5959 -0.47 178 Glycine 75 625600^249000 1834751 -3.21 -3.21 179 Amoxicillin 365 3433 12233 -1.99 -1.92 180 Tiagabine 376 0.66 0.13 2.79 181 Telmisartan 514 0.0000029 0.00 7.26 182 Trovoflaxicin (CP99219) 416 285 443 -1.19 183 Acrivastine 348 262 1.25 1.13 184 Nicotinic acid 123 48000 81555 0.80 185 Levodopa 197 320100 78380 -2.74 -2.82 186 Cefatrizine 463 2505 225 -2.96 187 Ampicillin 349 10100”/3574 12688 -1.13 -1.25 188 Vigabatrin 129 55140 90071 -2.94 189 Tranexamic acid 157 25000 17068 -1.80 190 Eflomithine 182 256000 487936 -3.00 191 Methyldopa 211 41810 34612 -2.11 192 Ceftriaxone 554 958 445 -2.09 ^ Drugs with missing fragments from ABSOLV program 193 Distigminebromide 578 194 Zidovudine 267 311 0.05 -0.20 195 Ximoprofen 261 453 2.18 196 Clonidine 230 13580 1.57 1.37 197 Viomycin 685 1000000 • -8.03 198 Ceftizoxime 382 910000 -4.30 ^ 257 No Names MW Sw(Calc)“ SwAb(Calc)= MlogP" ClogP' (mg L ') (mg L ’) 199 Capreomycin 653 1000000 -7.25 200 AAFC 243 1000000 -3.91 ^ 201 Bretyliumtosylate 244 4280 -1.25 Dose-limited, dose-dependent and formulation-dependent drugs 202 Spironolactone 417 28^28 2.61 2.26 2.25 203 Etoposide 588 158”/59 1644 0.60 -1.89 204 Cefetamet pivoxil 511 19 1039 2.33 205 Cefuroximeaxetii 510 29 7276 0.89 0.25 206 Azithromycin 749 7.09 1.83 207 Fosinopril 564 0.000033 7.74 208 Pravastatin 425 12 0.41 0.57 209 Cyclosporin 1202 22^/0.0000076 3.80 210 Bromocriptine 654 0.0020 6.69 211 Doxorubicin 543 93 0.80 0.10 -1.45 212 Cefuroxime 424 145 9543 -0.16 -0.17 213 lothalamatesodium 613 7.24 1.43 1.42 214 Sulfasalazine 398 2.44 0.61 3.83 215 Benazepril 425 2.23 0.38 1.82 216 Lisinopril 405 13 94 -1.71 217 Enalaprilat 348 11 432 0.86 218 Amphotericin B 924 1.16 -2.46 ^ 219 Aztreonam 435 810 24721 -3.46 ^ 220 Mibefradil 516 0.041 0.01 4.41 221 Ranitidine 314 24660 25298 0.27 . 1.33 222 Chlorothiazide 296 283^1854 123818 -0.24 -0.31 223 Acyclovir 225 33990 133154 -1.56 -2.07 224 Norfloxacin 319 177900 6508 -1.03 1.57 225 Methotrexate 454 2600 1.93 -0.30 226 Gabapentin 171 4491 6977 -1.18 227 Prazosin 383 310 245 2.45 228 Olsalazine 302 1.92 8.02 4.50 Drugs expected to have higher absorption 229 Ciprofloxacin 331 11480 3764 -1.08 1.40 230 Ribavirin 244 67180 6469995 -1.85 -3.23 231 Pafenolol 337 172 87 1.67 232 Azosemide 371 201 21 1.35 233 Xamoterol 339 11810 24179 0.61 0.39 234 Enalapril 376 35 135 0.79 235 Phenoxymethylpenicillin 350 101 5228 2.09 1.90 236 Gliclazide 323 138 423 1.09 237 Benzylpenicillin 334 210 1455 1.83 1.70 238 Thiacetazone 236 3302 19162 0.88 239 Lovastatin 405 0.41^2.14 0.08 4.26 4.08 240 Cromolynsodium 468 210 9.51 1.92 1.85 241 Erythromycin 734 1.43 2.54 0.65 “Calculated water solubility (Sw) values using WsKow for Windows, v.l.26,SRC (Meylan method)^°^. ^Experimental Sw values taken from Chapter 6. ‘^Abraham eqn 26 Chapter 6.3.2, page 187. Experimental logP (MlogP) from ClogP program. ® Calculated logP (ClogP). *^Calculated logP using WsKow for Windows, v.1.26, SRC (Meylan m.ethod)^°^. ®Trainingset2: Drugs 7-8, 11, 15, 18-19,21-22,24-38, 136-141 and 143-145. 258 Table 7.03 Human absorption, dose and percentage of excretion in urine, bile and faeces of drugs from literature Ratio* Excretion'* Metab.* Parent* Excretion* Excretion*' Excretion* Oral Dose* %insoluble*‘ Between Method" Quality” No %Abs.* % Abs.** % Bio/ in urine in urine drug in bile in faeces In urine drug (IS) insoluble %Abs."" for Of the Ref. in urine & faeces in 250ml drug and obtaining %Abs."’ (% dose) (% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs."" data absorption Training set 1 I 100 100 53 <1 5 100 5-20 96 0 100 BIO Good 43,69-71 2 100 -100 90(68-100) 26/17 0 600 63 0 100 BIO+RA Good 72 3 100 100 100 0 100 EU Good 73-74 4 97-100 100 71 small /5.4 10-20 17 0 100 REV Good 75 5 100 100 REV Good 76 6 -50 -50 1.25-5 0 0 100 REV+EUB Good 46 7 -100 15 25 99 0 100 REV Good 77 8 100 88 <10 100-400 0 0 100 REV Good 78 9 100 15-23 100 100 0 100 REV Good 79 10 94-99 100 99 250 86 0 99 BIO Good 44,80 11 70 98 70 63/0 7-30 15-240 0 0 98 BIO Good 45,81 12 100 85 98 200-400 0 0 98 EU Good 82 13 91 97 -91 0 97 250 100 3 97 EU Good 83 14 96 10 0 0 96 BIO Good 77 15 95-100 >90 50 95(50F) 3 >95 300mg/d 0 0 95 EU Good 84,85 16 59/65 0.1 0 0 91 RA Good 86 17 91 -100 90 38.8 >10.4 55.6 94.4 7.5 0 0 90 BIO Good 87 18 90 40-60 25 20-30 200 98 8 90 REV Good 88 19 100 -100 89 50-100 0 0 89 EU OK 89 20 81-95 >80 59 100-1200 0 0 86 BIO OK 90 21 85 85 EU OK 91 22 84 54/75 15 85 1 84 RA Good 92 23 67 66/81 52/76 32/17 1.2 21 0 81 RA Good 93 24 49/65 17/32 800 0 0 77 RA Good 94 25 62-98 60 48/75 <2 200 0 0 64 RAP OK 43 26 61 61 61 30-50 >2/6-9 32-52 40 7 0 61 BIO+RA Good 95 /50-95 /59-100 27 50-60A 35-50/80 27 500-1500 0 0 53 RAP+BIO OK 96-97 259 Ratio* Excretion'* Metab.* Parent' Excretion® Excretion** Excretion* Oral Dose* %insoluble*‘ Between Method" Quality" No %Abs.' % Abs." % Bio/ in urine in urine drug in bile in faeces In urine drug (IS) insoluble %Abs."* for Of the Ref. in urine & faeces in 250 ml drug and obtaining %Abs."* (% dose) {% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs."* data absorption 28 44/92 40/1.4 76/94 10 0 0 48 RA Good 98 29 47 21/46 3 0 0 47 RA Good 99 30 30/85 1000 0 0 35 RA Good 100 31 25/80 20-40 0 0 31 RAP OK 101-102 mg/kg 32 30 26/86 500 0 0 30 RAP OK 103 33 16 11.3/73 4.77 0 0 16 RA Good 99 34 12 12 16/98 3mg/kg 0 0 16 RAP OK 104 35 1-2 35 1.7/94 5I/<1 200 0 0 2 RA Good 105 36 1.4 0.5/33 8 0 0 1.4 RA Good 99 37 0.7/94 /94 4000 0 0 1 RA Good 106-107 38 0.6 0.6 0.53/93 10000 0 0 0.6 RA Good 108-110 Test set 39 99 100 A 20 70 0 100 BIO Good 111 40 100 100 A 50 100 0 100 BIO Good 112 41 100 A 40-68 16-48 56-100 0 100 BIO Good 46 42 100 66 traces 33 99 10-100 0 0 100 BIO Good 76 43 96 100 A 0 100 BIO Good 113 44 100 >80 0.1 0.5 >80 60 100 0 100 EF Good 114 45 95-100 >95 40 >95 <5 50 71 0 100 EU Good 115 46 100 99.5 200-600 100 0 100 EU Good 116-117 47 95-100 >95 22-77 >95 <5 40-60 91 0 100 EU Good 115 48 100 100 80 ns (20F) 1-3 0 0 100 Good 118 49 100 90A 60-70/61 /30 /91 50 97 0 100 RA Good 119 50 100 100 60/60 5-25 36/36 97/95 0.1 mg/kg 0 0 100 RA Good 120 51 100 100 100/100 >30 small 20 15 0 100 RA Good 121 52 100 100 1 1-300 0 0 100 REV Good 122 53 100 100 100 REV Good 123 54 100 -100 59 30 3 0 100 REV Good 46,124 55 100 2 200 0 0 100 REV Good 76 56 100 4 0 0 100 REV Good 76 260 Ratio' Excretion'* Metab.* Parent* Excretion® Excretion" Excretion' Oral Dose* %insoluble‘‘ Between Method" Quality” No %Abs.* % Abs." % Bio.* in urine in urine drug in bile in faeces In urine drug (IS) insoluble %Abs."" for Of the Ref. in urine & faeces in 250 ml drug and obtaining %Abs."’ (% dose) (% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs."' data absorption 57 100 100 17-50 0 100 REV Good 122 58 100 100 60 44-53 IS 8 82 0 100 REV Good 125 59 100 100 5-10 20 71 0 100 REV Good 76 60 100 >90 10-52 70 3 15 85 80-160 99 0 100 REV Good 126 61 91-100 91 1-2.5 38 0 100 Good 127 62 100 40 40 0 0 100 BIO Good 77 63 100 120 100 0 100 BIO Good 77 64 100 100 25 99 0 100 BIO Good 128 65 100 80-90 6-50mg/kg 89 0 100 REV Good 129 66 100 60/60 35/35 95/95 0.008 0 0 100 RA Good 130 67 100 UK) 0.2 250 0 0 100 EU Good 13! 68 99 99A 10 0 0 99 BIO Good 132 69 100 99 <1 100 650 100 1 99 EU Good 133 70 99 70-100 99 60 98 10-50 0 0 99 EU Good 134-136 71 90-100 >90 30 99(60F) <1 300 49 0 99 EU Good 85,137 72 100 -100 61-98A <2 200 0 0 98 EF Good 138-139 73 98 -100 93-98 -100 traces 100 5 15 0 98 EU Good 140 74 90 65/66 5/1.5 69-86/67 2 0 0 98 RA Good 141 75 95 98 REV Good 142 76 96 95-99 78-122 43-73 11 <1 1000-2000 100 3 97 BIO Good 143-144 77 97 92-102 2 0 0 97 BIO Good 145 78 92 97 50 0 0 97 EU Good 146 79 100 -100 97A 67/69.5 2.5/3.7 0.5-1 68/70 lOmg/kg 0 0 97 RA Good 147 80 100 -100 78/81 33 16/8.7 94/90 0.5 0 0 96 RA Good 148 81 96 96 BIO Good 149 82 95-100 >90 11 64-90 50-150 0 0 95 BIO Good 150 83 95 -100 90-1OOA <20(0F) 75 240mg/d 0 0 95 BIO Good 85,151-152 84 95 91 0.18 1 mg/kg 0 0 95 EU Good 153-154 85 95 >95 16 90-95 0.2 5-10 95-100 200 84 0 95 EU Good 155 86 100 95 80 400 0 0 95 EU Good 156 87 90-95 >90 33 95(-60F) <4 600mg/d 91 0 95 EU Good 85 88 97 90 50 95(40F) 3 160mg/d 0 0 95 EU Good 85 261 Ratio' Excretion'* Metab.* Parent' Excretion* Excretion'' Excretion' Oral Dose* %insoluble'‘ Between Method" Quality” No %Abs.* % Abs.'’ % Bio.' in urine in urine drug in bile in faeces In urine drug (IS) insoluble %Abs.'" for Oftbe Ref. in urine & faeces in 250 mi drug and obtaining %Abs."’ (% dose) (% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs."’ data absorption 89 95 -100 95 0 95 25-600 0 0 95 EU Good 157 90 72 >90 75 (25F) 5 6 81 30mg/d 0 0 95 REV Good 85,158 91 93-96 >93 93 0.4 100 0 0 93 EU Good 159-160 92 93 75-150 0 0 93 REV Good 149 93 100 -100 >92A 80 80 0-50 1 25-200 73 0 92 BIO Good 161 94 89-95 84-95 84/90 -4 /4 200 60 0 91 RA Good 162-163 95 90 90 80-89 >80 15 >80 20mg/d 0 0 90 BIO Good 84 96 100 well 80-1OOA 92 75 5-10 10 0 0 90 BIO Good 164 97 90 90A 43 >43 0.25 47 90 30 97 7 90 BIO Good 165 98 90 90 90A 400 21 0 90 BIO Good 136.166 99 90 30 15-25 0 0 90 EU Good 122 100 90 80 90 5-15 2.7 250 0 0 90 EU Good 167-168 101 90-95 90 43-63 100-1200 0 0 90 EU Good 45.169 102 99 >90 >90 60 <6 -100 150mg/d 0 0 90 EU Good 43 103 80-100 1 0 0 90 BIO Good 77 104 65-75 90 -15 10 90 EU Good 170 105 74/82 -0 12.0/14 10-20 58 0 90 RA Good 171 106 97 -100 92.8 71/80 71.9 15 0 0 89 RA Good 172 107 90 89A 120 0 0 89 BIO OK 173 108 20 >89 >59 -30 60 0 0 89 EU OK 170 109 100 100 16 62/70 0 9.8/11 72/81 27.5 82 0 88 RA Good 174 110 23 88(80-113) 0.5/0.5 20 3 0 88 EU OK 175-176 111 97 88 89/101 2000 90 2 88 RAP OK 177 112 88 88 BIO OK 149 113 87 87 <1 50-200 72 0 87 EU OK 46.178 114 92-95 >90 87 60(13F) 35 15 mg/kg 0 0 87 EU OK 85 115 86-88A 100 0 0 87 BIO OK 179 116 100 -100 20-30 85 2 20 0 0 85 EU OK 180 117 85 30 97 12 85 BIO OK 77 118 85 85 BIO OK 149 119 67 71 62A 67-76/93 45-50 38/40 16/0.3 91 100 0 0 84 RA+EF Good 181-182 120 84 84 /63-76 6 0 0 84 BIO OK 183-184 262 Ratio' Excretion"' Metab.* Parent^ Excretion* Excretion*" Excretion" Oral Dose* %insoluble'" Between Method"" Quality"" %Abs.* % Abs.’’ % Bio.* in urine in urine drug in bile in faeces In urine drug (IS) insoluble %Abs."" for Oftbe Ref. No in urine & faeces in 250 ml drug and obtaining %Abs."" (% dose) (% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs."" data absorption 121 84 1200 4 0 84 EU OK 73,185 122 82 82 61 70/84 47/71 7.0/2 77/86 20 0 0 82 RA Good 186 123 82 82 0.6mg/kg 0 0 82 BIO OK 44,187 124 74-88 400-600 0 0 82 EU OK 144 125 81 - 100 100 19 81 BIO OK 188 126 80 81 81 330 89 8 81 BIO OK 189-190 127 81 0.6-1 mg/kg 0 0 81 BIO OK 149 128 90 90 50 65(30F) 18 >90 300mg/d 35 0 80 REV Good 84-85,191 129 75-80 40-80 79-94/82 15-25 mg/kg 0 0 80 REV Good 106,192 130 80-100 80 68-95 85-95 4 500-1000 0 0 80 BIO OK 193-194 131 92-100 80A 1.5 80 BIO OK 195 132 75 77-80 1.4 8-32 87-98 16-32 0 0 80 EU OK 196 133 75-95 75-95 300 0 0 80 EU OK 106 134 75-80 -0.1 20-25 95-100 80 EU OK 43 135 80 80 BIO OK 170 136 77 50-60/94 125-500 0 0 77 BIO+RAP OK 197 137 64 62-64 56/90 2.5mg/d 0 0 64 RA Good 198 138 60 35/60 2.0/27 40/15 75/75 5-20 0 0 60 RA Good 199-200 139 20-35 34 34 0 34/60 /15 /75 80mg/d 0 0 57 RA DP? 85,201, 202,310 140 50-54 50 50 10(<10F) 40/90-95 >95 200mg/d 0 0 50 BIO+RAP OK 85,201 141 36 30 40/91 200 0 0 44 RAP OK 203 142 44 10 36/82 1.6 0 0 44 RA Good 204 143 37-45 25-30 38 BIO+RAP OK 43 /65-80 144 17 17(12-22) /73-94 /73-94 16000mg/d 0 0 17 RA Good 205 145 <5 2.7/90 lOmg/kg 0 0 3 RAP OK 104 146 92 90-95 17 60-67 0 26-32 86-99 5-20 7 0 92 EF 206 147 60-73 50-73 16 (78F) 8.7/56 5 0 0 62 207 148 60 60 REV 208 149 20-35 28 REV 209 150 poor small 1 REV 106 263 Ratio' Excretion'' Metab.* Parent^ Excretion® Excretion" Excretion' Oral Dose* %insoluble'‘ Between Method" Quality" No %Abs.* % Abs." % Bio.* in urine in urine drug in bUe in faeces In urine drug (IS) insoluble %Abs."" for Oftbe Ref. in urine & faeces in 250 ml drug and obtaining %Abs.*" (% dose) (% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs."’ data absorption 151 100 >90 19-29 4.9/3.4 <1 92/89 100 2-10 98 0 100 RA iVL 210-211 152 78 78 BIO OK? 149 153 75 76(53-88)A 400 68 0 76 BIO OK? 212-213 154 71 60 18 500 98 27 71 EU OK? 214 155 64-81 250 0 0 73 EU OK? 106 156 67-90 65-72 65-72 12.5-75 0 0 69(65-72) EU OK? 215 157 60 10 10 0 0 60 EU OK? 216 158 55-75 >57 14 57 36/<15 94 200 0 0 57 EU OK? 217-219 150 50 88 20 0 0 50 EU OK? 220 100 10-20 15.6 500 0 0 16 EU OK? 108-109 161 3-3.8 3 3 0 3 50-100 0 0 3 EU OK? 221 mg/d 162 0.6 2800 0 0 1 EU OK? 107 163 0.3 0.26 8000 0 0 0.3 EU OK? 222 164 100 100 Check 61 165 100 100 Check 61 166 90-100 95 Check 63,66,68 167 60 60 Check 58,63,66 168 50 50 Check 223 169 28 28 Check 60 170 100 100 Check 58 171 91 91 Check 58 172 0 Poor Poor Check 63,209 Zwitterionic drugs 173 100 70 500 45 0 100 BIO Good 224 174 100 100 BIO Good 225 175 100 400 0 0 100 BIO Good 77 176 98 100 85-100 500 11 0 100 EU Good 224 177 100 100 100 100-500 0 0 100 EU Good 226 178 100 100 Check 58 179 94 93 71/82 375-1000 0 0 93 BIO+RAP Good 227-231 264 Ratio' Excretion** Metab.* Parent* Excretion® Excretion" Excretion* Oral Dose* %insolubIe'* Between Method" Quality" N o %Abs.* % Abs." % Bio.* in urine in urine drug in bile in faeces In urine drug (IS) insoluble %Abs.*" for Oftbe Ref. in urine & faeces in 250 ml drug and obtaining %Ahs.*" (% dose) (% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs.*" data absorption 180 90A 90 BIO Good 77 181 90 Rapid 43 <2 >98 >98/>90 40 100 10 90 Check 58 182 88 88A 23 6 63 86 200 64 0 88 BIO OK 46,232 183 88 88 59 12 100 4-8 0 0 88 EU OK 233 184 88 88 EU OK 144 185 100 80-90 86A 250 0 0 86 BIO OK 234 186 75A 42-60/80 250 0 0 75 BIO+RAP OK 235 187 46/73 500 0 0 62 RAP OK 236 188 50-65 1000-3000 0 0 58 EU OK? 81 189 55 53/>95 <10 500-2000 0 0 55 RA Good 237 190 44/80 10-20 0 0 55 RAP OK 238 mg/kg 191 41 /57 250 0 0 41 Good 239-240 192 1 1 Check 241 Drugs with missing fragments from the ABSOLV program 193 4.7 6.5/85 88/4 5 0 0 8 RA Good 242 194 100 100 63 60-75 14-20/19 lOmg/kg 89 0 100 RA Good 243-244 /60-75 195 100 98 30 0 0 98 BIO Good 245 196 95-100 100 75-95 50 30-33 0.3 0 0 95 BIO Good 246-248 197 85 85 EU OK 101 198 58/81 500 0 0 72 RA Good 249 199 50 lOOOmg/d 0 0 50 EU OK? 106 200 32 24/74 19-35 2-20mg/kg 0 0 32 RA Good 250 201 23 23(12-37) A /lOO 13/71 5-200 0 0 23 BIO+RAP OK 251 Dose-limited, dose-dependent and formulation-dependent drugs 202 >73 53 0 20 37 40-95 50-200 96 23 73 EUB DL 252 203 50 50(25-75) 6-25/30-50 /0-16 /30-66 100-600 96 46 50(25-75) BIO DL 253-254 204 47 38/80 - 500 99 52 47 RAP DL 255 205 36 36-52 36 500 99 63 44(36-52) RA DL 256 265 Ratio* Excretion'* Metab.' Parent* Excretion* Excretion** Excretion* Oral Dose* %insoluble*‘ Between Method" Quality” No %Abs.‘ % Abs." % Bio.' in urine in urine drug in bile in faeces In urine drug (IS) insoluble %Abs.'" for Oftbe Ref. in urine & faeces in 250 ml drug and obtaining %Abs."’ (% dose) (% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs."’ data absorption 206 35-37 37 4.5/12 500 100 63 37 RAP+BIO DL 257 207 36 25-29 8.6/29 63-81/42-6 72-90 10 100 64 36 RA DL 258-259 208 34 34 18 20/60 71/34 91/94 20 85 51 34 RA DL? 210,260 209 35 10-60 8 mg/kg 100 65 28(10-65) BIO DL 261-262 210 28 28 6 2.5-5.5 <5 85 0.5-2.5 100 72 28 REV DL 46,263 mg/d 211 5 trace 5 5(0.7-23) 25-45 50-60 58 53 12(0.7-23) BIO+EU DL 264-265 212 1.0/95 1000 96 95 1 RAP DL? 266 213 1.9 1.9 1.4/75 800 100 98 1.9 RAP DL 267 214 12-13 56-61 2000 100 35 59(56-61) EU DL? 268 215 37 >37 37 0.4 97 20 97 60 >37 EU DL? 269-270 216 25 25 25-50A /100 29 56 97 10-20 78 40 28(25-50) BIO+EF DL? 271-272 217 9-10 10-40 29 69 98 10 72 43 25(10-40) EU DL? 273-275 218 5 poor 2-5 0 2000-5000 100 95 3(2-5) EU DL 276 219 Drugs expected to have higher absorption 229 69-100 69 42/60 200 0 0 >69 RAP 289-290 230 33 4.4/17 6.2/29 800 0 0 >33 BIO+RAP 291 231 28 1.9/4.8 16/56 74/28 40 0 0 >29 RAP '292 232 10 2.0/20 20-80 0 0 >10 RAP 293 233 5 /90 3.1/62 50-200 0 0 >5 RAP 294 266 Ratio* Excretion** Metab.* Parent^ Excretion* Excretion" Excretion* Oral Dose* %insoluble*‘ Between Method" Quality" No %Abs.* % Abs." % Bio.' in urine in urine drug in bile in faeces In urine drug (IS) insoluble %Abs.*" for Of the Ref. in urine & faeces in 250 ml drug and obtaining %Abs.'" (% dose) (% dose) (% dose) (% dose) (% dose) (% dose) (mg) water dose after chosen %Abs.*" data absorption 234 66 60-70 29-50 61 29 90 10 13 0 66(61-71) EU+EF 274-275, 295-297 235 45 45(31-60) 49 30 0 32(22-52) 71-100 22 0 0 59(49-68) EU-EF 298 236 60-70 10-20 70-90 >65 EU 91 237 30 15-30 15-30 500 90 60 >30 EU DL? 299 238 20 150 0 0 >20 EU 106 239 30 31 9.6 83 93 20-100 99 68 >10 EU DL 210,300 240 0 0.4 82 20 0 0 >0.4 EU 301 241 35 35 >4.5 >35 BIG 44,302 “The data used for QSAR studies was taken from Clark^^ (1999) and Wessel^^ (1998), Palm ^^(1997), Yazdanian ^\l998), Yee ^^(1997) and Chiou ^'(1998). '’Absorption data was obtained from the original and review literature. Bioavailability or absolute bioavailability of oral administration. ‘'Percentage of cumulative drug and its metabolites in urine following oral/intravenous administration. ® Percentage of metabolites in urine by oral/intravenous administration or first pass effect (F). ^ Percentage of unchanged drug in urine by oral/intravenous administration. ® Percentage of excretion in bile by oral/intravenous administration. Percentage of excretion in faeces by oral/intravenous administration. ' Percentage of cumulative recovery in urine and faeces by oral/intravenous administration. ■' Single dose (mg or mg/kg) and daily dose (mg/d). '^Percentage of insoluble oral dosed drug in 250ml water (IS). ' Ratio between insoluble drug and dose administered after absorption 100x[Dosex(l-fraction absorbed)-0.25xSw]/Dose. Absorption data (or averaged values) chosen based on the analysis of literature. "Method for obtaining absorption data (%Abs."'). Quality of the data based on the analysis of literature. Notes for some of the drugs from literature 1. Drug 35: Additionally, up to 35% of radioactivity of acarbose was absorbed after degradation by digestive enzymes and/or intestinal microorganisms. 2. Drug 59: 100% oral absorption in humans was obtained from studies in rabbits. 3. Drug 61: Absorption data was obtained from the small intestinal zone 100-200 cm. 267 4. Drugs 144, 191, and 205: Absorption was evaluated from the urinary excretion ratio of oral and intravenous administration based on literature although the authors did not give the urinary excretion in detail. 5. Drug 146: Assuming all intact PN200-100 found in faeces to represent unabsorbed drug, the extent of absorption could be as high as 90-95 of the dose. Although the possibility of drug metabolism by intestinal epithelium or microbial flora cannot be predicted, rapid urinary excretion of the administered dose also supported efficient oral absorption. 6. Drug 147: The absorbed amount was calculated as the sum of unchanged drug, systemically and presystemically conjugated terbutaline plus deficit. 7. Drug 202: The extent of absorption must be at least approximately 73%, since 53% and 20% of an orally administered radioactive dose of spironolactone in an alcoholic solution were excreted in urine and bile respectively. 8 . Drug 206: Azithromycin gains entry into cells by both passive and active transport. 9. Drug 215: If biliary excretion occurred, the extent of absorption may well have been greater. In animal studies, significant biliary excretion has been found (>50% of intravenous dose or of absorbed fraction of oral dose). 10. Drug 224: Recovery of the drug in faeces from 12 healthy volunteers given single oral 400mg doses averaged 28% over the ensuring 48 hours. As the drug is excreted in the bile to only a small extent, these results imply an oral bioavailability of approximately 70%. 11. Drug 227: the data of 77-95 is the percentage of excretion in urine and bile. 12. Drug 239: Estimate for absorption in humans based upon hydroxy-acid form as intravenous reference. 268 7.2.2 Relationship Between Human Intestinal Absorption and Abraham Descriptors The QSAR used in this work is of the form of the General Solvation Equation, the origin of which is described in Chapter 1.4, pages 22-31. %Abs. = c + r.R% + + v.Vx (7) where %Abs. is the percent of intestinal drug absorption in vivo. Although this is contrary to the idea of eqn 7 as a linear free energy relationship, it is the only practical way of including all the relevant data. It is possible to convert % absorption into a rate constant, or the logarithm of a rate constant, but only by omitting all drugs with 0 % absorption and 100 % absorption. Thus all the correlation equations are couched in terms of eqn 7, with % absorption as the dependent variable. The ABSOLV program (details in Chapter 1.4.6, page 31) which was written to read molecular structures as SMILES strings has been used to calculate all the descriptors used in this work. After calculation of the solvation descriptors, an error code was given by the program for each drug as an indication of the quality of the parameter calculations. To model human intestinal absorption using the Abraham descriptors, two training sets were selected from drugs 1-145 (Tables 7.02 and 7.03) since this absorption data is considered to be comparatively reliable (Good and OK). Training set 1 is chosen by use of an alternative space filling design technique developed by Kennard and Stone.^®^ The principle of the method is based on the distribution of chosen descriptors. The descriptors of a training set should cover the whole descriptor space of the total set, and the histogram of the training set should relate to that of the total set, as shown in Fig.7.05 for training set 1. However the histogram of the total set, and of course training set 1, is completely biased towards drugs with 100 % absorption, and it might be suggested that any mathematical analysis will also be biased in this way. Therefore, another training set was chosen, based on histogram analysis, that is not biased towards 100 % absorption. Although this second training set enables a test for bias in the dependent variable, it is not statistically as sound as training set 1, because it does not take into account the distribution of descriptors. Therefore training 269 set 2 is used only as a test for the above bias. Fig. 7.05 shows the histograms for the two separate training sets, the total data set and also the histogram of the data set from Wessel et al. (1998)“ . Fig.7.05 Histograms of training set and total set Training Set 1 Total Set 20 80 15 60 10 40 5 20 0 n rn r-i r-i r~1f~1r~1 0 20 40 60 80 100 20 40 60 80 100 Training Set 2 Total set 80 c 30) 60 IT 0 ri r-l r-i r-i n n ri 0 20 40 60 80 100 20 40 60 80 100 W essel et al. training set W essel etal. total set 40 40 30 30 20 20 10 10 n 0 - a ,------,-P 0 O , 0,0, , 0 20 40 60 80 100 0 20 40 60 80 100 % Absorption Results of the two training set multiple linear regression analyses using the Abraham descriptors in eqn 7 are given in Table 7.04 (Models 1 and 3). 270 Table 7.04 Regression results of different training sets using Abraham descriptors — Training set — Test set ----- No Data set Models r^ (adj) n SD RMSE F n RMSE AAE AE 1 Training 1 %Abs. = 90 4- 2.IIR 2+ 1.70ti2^ — 20.72ot2^ 0.83 0.80 38 16 14 31 131 14 11 -1 - 22.3% "+ 15.0VX 2 Training 1 %Abs. = 92 - 2 0 .0 2 0 2 " -21.9% " + 17.2VX 0.82 0.81 38 15 14 53 131 14 11 -1 3 Training 2 %Abs. = 82 + 3.9 7 R2- 3.587 i2" - 1 9 .5 % " 0.85 0.82 31 14 12 28 139 19 16 -13 - I8 .OZP2" + 14.0VX 4 Training 2 %Abs. = 83 - 19.9%" - I 8 .6ZP2" + 13.6VX 0.85 0.73 31 13 12 50 139 19 17 -14 5 Total data %Abs. =92 + 2.94 R2 + 4 .IO7I2" - 2 1 .7 % " 0.74 0.73 169 14 14 93 - - - - - 2 I.ISP 2" + 10.6VX 6 Total data %Abs. = 96 - 2 0 .0 % " - 19.8%" + 13.9Vx 0.72 0.72 169 15 14 14 --- - 7 Training 1 %Abs. = 89 - 2 4 .0 % " - I8 .9ZP2” + 16.8Vx 0.80 0.78 49 17 16 43 - -- - -f* DL - 0.421 IS 8 Total data %Abs. = 93 - 2 1 .3 1 0 2 " - I9.OZP2" + 14.6VX 0.74 0.73 180 15 15 15 - -- - + DL - 0.386 IS Training set 1: see Table 7.02. Training set 2; drugs 7-8, 11, 15, 18-19,21-22, 24-38, 136-141 and 143-145. Total data: drugs 1-169. r^ (adj) : Adjusted r-square, r^ (adj) = l-((l-r^)(n-l/n-p)), where p is the number of descriptors used in the model. RMSE: Root mean square error, RMSE=[S(calc-obs)^/n]®’^, AAE: average absolute error, AAE=E|calc-obs|/n., AE: average error, AE=S(calc-obs)/n. IS: % insoluble oral dosed drug in 250ml water. 271 Step-wise regression was carried out to find the significant descriptors. The result showed that the significant descriptors were and Vx (Models 2 and 4 in Table 7.04) and the two dominant descriptors were Zaz" and This is in agreement with previous work that suggests hydrogen bond donors and hydrogen bond acceptors or polar molecular surface are good descriptors to model human intestinal absorption.^^’^^’^^ The coefficient standard errors of the variables are around 5 for the two training sets (Models 1-4) and 2.5 for the whole data set (Models 5-6). The results indicate that models 1, 2, 3 and 4 are relatively similar. The two different ways of obtaining the training set give very similar absorption models. However, Models 1 and 2 are statistically superior to Models 3 and 4. The use of the Kennard and Stone selection method results in training and test sets which have less Bias (Bias is defined as the average difference between the predicted and actual values in the test set). Models 1-4 show that increasing the volume (hydrophobic part) and decreasing the polarity of a compound can increase human intestinal absorption. The details of the number of drug compounds used in the analyses are presented in Table 7.05. Table 7.05 Details of drugs used for analysis Drug no. in Tables Number of drugs 7.02, 7.03, and 7.07 Total number investigated 241 1-241 Training set 1 38 1-38 Test set 131 39-169 Dose-limited drugs added for analysis 11 202-212 in Models 7 and 8 For the full equation (Model 5), the t-statistic has been calculated as follows: R 2 (1.05), 712^ (1.71), Z(%2" (-8.20), Zp 2^ (-12.6) and Vx (5.11). These show, as already indicated by the step-wise regression, that the R 2 and 7t2^ descriptors are statistically not very significant. The inter-correlation of descriptors for the full data set of 169 compounds used in Models 5 and 6 in r^ are as follows: 272 Zp2 Vx 0.617 0.053 0.312 0.373 0.197 0.426 0.460 0.272 0.037 0.404 The only cross-correlation that is rather high is that between R 2 and 712". Fortunately, neither R2 nor 7:2" is very significant and so the R 2/712" cross-correlation presents no real problem. Thus for the truncated equation (Model 6) the largest cross-correlation in r^ is only 0.404, between ZP 2" and Vx. Fig.7.06 shows the prediction result for the test set using Model 1 (Drugs 39-169). The predicted absorption is in agreement with the observed absorption for 131 compounds. Only drug 161 (0 in Fig. 7.06) is an obvious outlier by use of the Abraham model. A possible reason for this outlier is that the drug may not be completely recovered in the urine and faeces even if urinary excretion is the main route. As such, absorption may be underestimated by use of the percentage of excretion in urine. Fig.7.06 Relationship between the observed and predicted absorption Training set Æ T est s e t ■oo ^ Dose-limited 20 40 60 80 100 Observed 273 Absorption correction for dose-limited drugs was applied both to Model 2 and to the total data set by using additional descriptors, viz. solubility, octanol-water partition coefficient, molecular weight and administered dose. The result (Models 7 and 8 and Fig. 7.06) shows that the best additional descriptors were the solubility and dose. The latter was expressed by the percentage of insoluble drug (IS) administered in 250ml water (Table 7.03) for dose-limited drugs, with the percentage taken as zero for drugs 1-169. The regression result suggests that the more insoluble a dose-limited drug, the lower the absorption will be. Table 7.07 (pages 278-282) lists the predicted absorption and also the residuals (predicted-observed) for dose-dependent, dose-limited drugs before the dose and solubility correction, and drugs 229-241 obtained using Model 1 in Table 7.04. The results plotted in Fig.7.07 show that nearly all the absorption predicted for dose- dependent drugs (Drugs 220-226) is in the range of observed absorption. Absorption prediction for some of the dose-limited drugs (Drugs 203, 205, 206 and 209) is in agreement with or is in the range of the observed absorption, and for some of the dose-limited drugs (202, 204, 207, 208 and 210-212) the predicted values are higher than the observed absorption. This is in agreement with the point that absorption is highly variable and incomplete for dose-limited drugs.^'"^’^^^'^^"^ Fig.7.07 Plot of observed and predicted absorption by Model 1 100 80 0 Training set 60 □ O 1 □ Drugs 229-241 ■5 40 20 O DP 0 -20 0 20 40 60 80 100 Observed 274 The ABSOLV program cannot accurately calculate the descriptors for drugs 193-201 owing to missing fragments. The absorption prediction for some of these drugs does not agree well with the observed absorption, possibly because of the inaccurate calculation of descriptors. The prediction of absorption for drugs 229-241 is higher than the observed absorption; this is in agreement with what would be expected for these drugs based on the method from which the absorption data was derived. For example, absorption evaluated by the percentage of urinary excretion of drugs 236- 240 is lower than the predicted absorption because excretion in urine for the absorbed drug may not be the only route for excretion. These drugs have Clog P values above zero; faecal excretion may be another route for the absorbed drug. Hence, absorption would be underestimated by the urinary excretion method. All the descriptors used in the analyses refer to the neutral form of the respective drug compounds, and there has been no correction included to account for ionization of strong Bronsted acids and bases. Inspection of the calculated and observed % absorption in Fig.7.08 reveals no particular trend of ionizable drugs. Fig. 7.08 Distribution of neutral and ionized drugs in relation to absorption prediction using Model 5 100 80 oA H O o Neutral 60 I ♦ w eak acids + u w eak bases I 40 ■ strong acids II ▲ strong bases -20 0 20 40 60 80 100 Observed 275 However, in order to assess more rigorously any effects, an analysis of the 169 drugs used in Model 5 has been carried out with inclusion of an indicator variable (I) for strong acids with pKa<4.5 and bases with pKa>8.5; I is taken as unity for the strong acids and strong bases, and zero for all other compounds. The resulting equation is %Abs. = 94 + 2.90R2 + 2.71%" - 2 0 .7 % " - 2 O.9 SP2" + 11.2Vx - 3.141 (8) n=169, = 0.74, SD=14%, F=78 It can be seen that the additional indicator variable (I) is hardly significant (p=0.25, t= -1.16), and so it may be concluded that any ionization of Bronsted acids and bases has a very small effect indeed (3%) on the % absorption. It is noteworthy that previous w orkers^'have made no ionization correction, either. It is useful to compare the statistics of the various equations put forward for the correlation of % absorption. Clark carried out only a qualitative analysis, but the statistics obtained from the results of Wessel et al. and Palm et al.^^ are shown in Table 7.06. In their analysis, Palm et al. gave standard deviations in the % absorption observed for 16 drugs. The average of these standard deviations is 9%, so that it is unreasonable to expect any equation to correlate % absorption to less than this value. The RMSE values of Palm et al. and Wessel et al. in Table 7.06 thus represent about the limit of correlative equations. Of course, predictions of % absorption for a test set cannot be expected to be better than correlations, as shown by the RMSE of 16 % for 10 drugs in the test set of Wessel et al.^^ By comparison, the RMSE of 14 % obtained for this work for a test set of no less than 131 drugs represents a very good prediction, bearing in mind the experimental error in the data. Table 7.06 Comparison of statistics for equations for % absorption — Training set - — 1 est set — r' n RMSE n RMSE Palm et al. 0.94 20 9% — Wessel et al.^° — 67 9% 10 16% No 1, Table 3 0.83 38 14% 131 14% No 5, Table 3 0.74 169 14% — 276 It is therefore evident that the Abraham descriptors are able to successfully predict absorption for a diverse set of drugs. The significant descriptors are the summation of solute hydrogen-bond acidity (Sa 2^) and the summation of solute hydrogen-bond basicity (Sp 2^), but the volume term also contributes to absorption. The predicted absorption values also correlated well with the observed absorption for large drugs (Molecular weight > 500) with the proviso that they were not dose-limited (Drugs 150, 163 and 197). Most of the dose-limited drugs satisfied the “rule of 5”; molecular weight > 500 and number of hydrogen bond donors > 5 or number of hydrogen bond acceptors >10. These compounds usually have low solubility and their absorption therefore varies considerably among different preparations.T he absorption of these drugs may not only be controlled by the passive diffusion rate, but also by the in vivo dissolution rate in the small intestinal fluid. Therefore, the solubility and dose are very important pieces of information that should be known before QSAR analysis is attempted on human absorption since the drug may be dose-limited and have lower absorption. 277 Table 7.07 Observed and predicted absorption from Model 1 Method for Quality of No obtaining the Observed Predicted Pred. - R2 Vx Error %Abs. %Abs. %Abs. %Abs. obs. Code Data Training set 1 1 BIO Good 100 97 -3 2.30 3.40 0.46 2.04 3.40 -OP 2 BIO+RA Good 100 89 -11 0.24 0.47 0.59 0.44 1.31 -OP 3 EU Good 100 85 -14 1.05 0.89 0.72 0.38 0.99 -OP 4 REV Good 100 104 4 2.38 2.11 0.00 1.15 2.07 -OP 5 REV Good 100 80 -19 2.87 3.60 0.58 1.91 2.17 -OP 6 REV+EUB Good 100 87 -13 2.60 3.89 1.30 1.88 3.56 -OP 7 REV Good 100 107 7 1.72 2.56 0.00 2.28 3.99 -OP 8 REV Good 100 98 -2 0.97 0.81 0.03 0.84 1.58 -OP 9 REV Good 100 109 9 3.07 2.83 0.00 2.08 3.59 -OP 10 BIO Good 99 94 -5 1.62 1.40 0.59 0.75 1.78 -OP 11 BIO Good 98 91 -7 2.79 2.81 0.50 1.09 1.65 -OP 12 EU Good 98 92 -6 1.19 2.21 0.00 1.28 1.62 -OP 13 EU Good 97 100 3 2.12 1.25 0.00 1.39 2.29 -OP 14 BIO Good 96 73 -23 2.14 2.95 1.12 1.90 2.58 -OP 15 EU Good 95 94 -1 1.13 1.18 0.10 1.44 2.26 -OP 16 RA Good 91 75 -15 2.24 2.09 0.55 2.11 2.36 -OP 17 BIO Good 90 82 -8 3.25 3.78 0.25 2.64 2.83 -OP 18 REV Good 90 98 8 2.28 3.09 0.59 1.28 2.57 -OP 19 EU OK 89 87 -2 1.77 3.25 0.22 2.18 2.71 -OP 20 BIO OK 86 57 -29 1.20 2.40 0.54 2.78 2.21 -OP 21 EU OK 85 84 -1 1.35 2.15 0.93 1.09 2.06 -OP 22 RA Good 84 109 26 1.73 1.88 0.00 1.45 3.00 -OP 23 RA Good 81 53 -27 3.20 5.34 1.72 4.62 5.75 -OP 24 RA Good 77 79 3 1.13 1.60 0.35 0.99 0.89 -OP 25 RAP OK 64 66 3 1.53 2.11 0.59 2.14 1.96 -OP 26 BIO+RA Good 61 69 9 2.05 2.55 1.36 1.47 2.10 -OP 27 RAP+BIO OK 53 57 5 1.18 1.35 0.27 2.17 1.09 -OP 28 RA Good 48 55 7 1.55 1.45 1.41 1.68 1.73 -OP 29 RA Good 47 78 31 2.50 4.41 0.86 3.08 4.08 -OP 30 RA Good 35 59 24 1.39 2.18 0.78 1.71 1.11 -OP 31 RAP OK 31 36 5 0.67 1.42 1.52 1.78 0.86 -OP 32 RAP OK 30 23 -7 0.83 2.44 1.88 2.37 1.23 -OP 33 RA Good 16 21 5 3.94 6.54 2.13 6.12 6.14 -OP 34 RAP OK 16 29 14 2.08 3.00 1.76 2.72 1.79 -OP 35 RA Good 2 -20 -22 3.31 4.47 2.53 6.19 4.38 -OP 36 RA Good 1.4 28 28 3.13 5.12 2.32 4.09 4.16 -OP 37 RA Good 1 6 5 2.80 2.71 1.20 5.40 3.36 -OP 38 RA Good 0.6 18 18 1.95 2.57 1.70 3.53 2.23 -OP Test set 39 BIO Good 100 99 -1 2.63 2.56 0.00 1.84 2.67 -OP 40 BIO Good 100 99 -1 2.39 2.72 0.59 1.19 2.53 -OP 41 BIO Good 100 101 1 1.79 2.46 0.43 1.18 2.58 -OP 42 BIO Good 100 76 -24 2.82 3.51 0.58 2.08 2.17 -OP 43 BIO Good 100 76 -24 1.93 1.84 0.42 1.38 1.22 -OP 44 EF Good 100 96 -4 3.43 2.83 0.33 2.25 3.40 -OP 45 EU Good 100 106 6 1.99 1.57 0.09 1.04 2.26 -OP 46 EU Good 100 101 1 1.80 1.76 0.59 0.62 1.98 -OP 47 EU Good 100 107 7 1.97 1.56 0.00 1.15 2.40 -OP 48 Good 100 100 0 2.69 2.37 . 0.10 1.39 2.26 -OP 49 RA Good 100 92 -8 1.97 1.88 0.78 0.87 2.03 -OP 278 Method for Quality of No obtaining the Observed Predicted Pred.- Zaz" Vx Error %Abs. %Abs. %Abs. %Abs. obs. Code Data 50 RA Good 100 83 -17 2.18 2.53 0.37 2.03 2.45 -OP 51 RA Good 100 101 1 1.61 2.32 0.35 1.13 2.38 -OP 52 REV Good 100 85 -15 1.94 1.81 0.00 1.47 1.36 -OP 53 REV Good 100 91 -9 1.90 2.98 0.53 1.71 2.74 -OP 54 REV Good 100 89 -11 2.12 2.50 0.97 1.16 2.39 -OP 55 REV Good 100 80 -20 2.47 3.53 0.58 1.92 2.21 -OP 56 REV Good 100 79 -21 2.95 3.60 0.58 2.04 2.30 -OP 57 REV Good 100 90 -10 1.05 1.09 0.00 1.11 1.37 -OP 58 REV Good 100 100 0 2.13 2.15 0.00 1.46 2.27 -OP 59 REV Good 100 78 -22 2.84 3.61 0.58 2.06 2.25 -OP 60 REV Good 100 108 8 1.70 2.48 0.00 2.07 3.79 -OP 61 Good 100 111 11 1.58 2.47 0.00 1.16 2.62 -OP 62 BIO Good 100 71 -29 1.91 2.06 0.49 1.77 1.56 -OP 63 BIO Good 100 123 23 2.43 2.03 0.02 1.11 3.30 -OP 64 BIO Good 100 111 11 2.07 3.17 0.00 1.57 3.09 -OP 65 REV Good 100 99 -1 1.94 2.42 0.00 1.60 2.45 -OP 66 RA Good 100 84 -16 1.26 1.55 1.19 1.44 3.03 -OP 67 EU Good 100 94 -6 1.78 1.78 0.00 1.37 1.87 -OP 68 BIO Good 99 94 -5 2.33 2.21 0.28 1.24 1.93 -OP 69 EU Good 99 83 -16 2.83 3.31 0.56 2.47 3.20 -OP 70 EU Good 99 82 -17 2.19 3.26 0.72 2.00 2.75 -OP 71 EU Good 99 98 -1 1.85 1.36 0.10 1.29 2.15 -OP 72 EF Good 98 83 -15 1.15 1.42 0.32 1.47 1.87 -OP 73 EU Good 98 94 -4 2.30 2.43 0.55 1.26 2.31 -OP 74 RA Good 98 90 -8 1.44 1.71 0.35 1.48 2.28 -OP 75 REV Good 98 75 -23 2.46 2.87 0.50 1.71 1.59 -OP 76 BIO Good 97 106 9 0.93 1.23 0.00 0.69 1.82 -OP 77 BIO Good 97 83 -14 2.52 2.81 0.50 1.76 2.18 -OP 78 EU Good 97 93 -4 1.24 1.32 0.35 1.36 2.37 -OP 79 RA Good 97 95 -2 1.53 1.58 0.00 1.05 1.48 -OP 80 RA Good 96 70 -26 2.20 2.73 1.41 1.76 2.64 -OP 81 BIO Good 96 99 3 1.68 1.52 0.00 0.98 1.63 -OP 82 BIO Good 95 84 -11 1.69 2.30 0.35 1.62 2.01 -OP 83 BIO Good 95 74 -21 1.54 1.98 0.74 1.74 2.10 -OP 84 EU Good 95 86 -9 2.02 1.78 0.26 1.75 2.21 -OP 85 EU Good 95 90 -5 1.80 2.38 0.00 1.76 2.09 -OP 86 EU Good 95 97 2 0.86 0.84 0.59 0.50 1.78 -OP 87 EU Good 95 86 -9 2.20 2.13 0.77 1.62 2.64 -OP 88 EU Good 95 93 -2 1.26 1.18 0.10 1.49 2.22 -OP 89 EU Good 95 80 -15 1.45 1.95 0.58 1.64 2.18 -OP 90 REV Good 95 84 -11 1.47 1.81 0.10 2.03 2.38 -OP 91 EU Good 93 97 4 1.25 1.03 0.10 1.25 2.16 -OP 92 REV Good 93 79 -14 1.84 2.11 0.53 1.29 1.40 -OP 93 BIO Good 92 95 3 1.63 1.78 0.59 0.86 1.98 -OP 94 RA Good 91 83 -8 2.06 3.16 0.72 1.98 2.80 -OP 95 BIO Good 90 99 9 1.33 1.29 0.10 1.44 2.57 -OP 96 BIO Good 90 85 -5 1.69 2.02 0.59 1.23 1.87 -OP 97 BIO Good 90 82 -8 2.88 3.57 0.58 1.91 2.32 -OP 98 BIO Good 90 89 -1 2.21 1.68 0.48 1.21 1.87 -OP 99 EU Good 90 94 4 0.94 0.77 0.18 0.63 1.24 -OP 100 EU Good 90 79 -11 1.86 2.46 0.66 1.62 2.07 -OP 101 EU Good 90 83 -7 1.44 1.48 0.70 1.12 1.77 -OP 102 EU Good 90 77 -13 1.87 2.55 0.20 2.41 2.46 -OP 103 BIO Good 90 103 13 2.58 2.22 ' 0.00 1.32 2.20 -OP 279 Method for Quality of No obtaining the Observed Predicted Pred. - Rz Kz" Zaz" ZPz" Vx Error %Abs. %Abs. %Abs. %Abs. obs. Code Data 104 EU Good 90 88 -2 1.24 1.30 0.35 1.50 2.23 -OP 105 RA Good 90 101 11 1.71 2.43 0.32 1.54 2.92 -OP 106 RA Good 89 88 -1 2.51 2.33 0.38 1.49 1.99 -OP 107 BIO OK 89 94 5 2.70 2.63 0.68 1.04 2.07 -OP 108 EU OK 89 87 -2 1.88 1.68 0.26 1.73 2.25 -OP 109 RA Good 88 101 13 1.75 2.17 0.32 1.37 2.71 -OP 110 EU OK 88 97 9 1.72 2.47 0.32 1.53 2.64 -OP 111 RAP OK 88 79 -9 1.59 2.14 0.68 0.95 1.15 -OP 112 BIO OK 88 57 -31 1.62 2.06 0.61 2.34 1.67 -OP 113 EU OK 87 98 11 1.14 1.31 0.09 1.07 1.94 -OP 114 EU OK 87 81 -6 1.68 1.48 0.47 1.58 2.01 -OP 115 BIO OK 87 69 -18 2.34 2.36 0.51 1.92 1.53 -OP 116 EU OK 85 78 -7 2.10 1.68 0.55 1.76 2.06 -OP 117 BIO OK 85 85 0 2.02 3.15 0.42 1.84 2.37 -OP 118 BIO OK 85 93 8 1.85 1.94 0.19 1.74 2.54 -OP 119 RA+EF Good 84 80 -4 1.15 1.68 0.50 1.31 1.62 -OP 120 BIO OK 84 88 4 2.48 2.46 0.28 1.54 1.94 -OP 121 EU OK 84 84 0 0.93 1.35 0.59 0.80 1.29 -OP 122 RA Good 82 57 -25 2.55 2.71 0.93 2.39 2.00 -OP 123 BIO OK 82 84 2 2.18 3.23 0.72 2.02 2.90 -OP 124 EU OK 82 80 -2 0.76 2.38 0.00 2.28 2.36 -OP 125 BIO OK 81 91 10 0.82 1.80 0.54 1.41 2.60 -OP 126 BIO OK 81 89 8 2.30 1.90 0.26 1.88 2.55 -OP 127 BIO OK 81 78 -3 1.77 2.08 0.61 1.37 1.60 -OP 128 REV Good 80 83 3 1.60 2.40 0.58 1.97 2.76 -OP 129 REV Good 80 76 -4 0.78 0.79 0.22 1.81 1.83 -OP 130 BIO OK 80 73 -7 1.27 1.81 1.02 0.85 1.17 -OP 131 BIO OK 80 83 3 2.09 3.22 0.77 2.01 2.91 -OP 132 EU OK 80 84 4 1.85 1.60 O il 1.51 1.56 -OP 133 EU OK 80 66 -14 1.21 1.89 0.55 1.51 1.03 -OP 134 EU OK 80 82 2 2.35 3.41 0.42 2.16 2.52 -OP 135 BIO OK 80 109 29 1.61 1.59 0.00 1.25 2.71 -OP 136 BIO+RA? OK 77 88 11 1.98 2.71 0.25 1.83 2.34 -OP 137 RA Good 64 86 22 2.84 3.09 0.81 1.59 2.50 -OP 138 RA Good 60 50 -10 2.21 2.16 1.82 2.05 2.36 -OP 139 RA DP? 57 77 20 1.61 1.63 0.70 1.88 2.49 -OP 140 BIO+RAP OK 50 79 29 1.45 1.89 0.55 1.75 2.18 -OP 141 RAP OK 44 72 28 1.84 2.93 0.75 2.21 2.53 -OP 142 RA Good 44 56 12 1.41 1.44 1.28 1.71 1.70 -OP 143 BIO+RAP OK 38 57 19 2.61 2.53 1.07 2.47 2.26 -OP 144 RA Good 17 25 8 0.79 1.76 1.78 1.93 0.70 -OP 145 RAP OK 3 20 17 2.15 3.08 2.02 2.96 1.85 -OP 146 EF 92 96 4 1.67 2.46 0.32 1.62 2.71 -OP 147 62 58 -4 1.41 1.40 1.28 1.74 1.84 -OP 148 REV 60 50 -10 3.23 3.15 1.28 3.06 2.81 -OP 149 REV 28 56 28 1.88 2.36 0.84 3.17 3.10 -OP 150 REV 1 -4 -5 3.50 3.75 0.81 6.78 4.02 -OP 151 RA IVL 100 84 -16 2.39 2.45 1.28 1.60 3.13 -OP 152 BIO OK? 78 80 2 2.98 2.99 0.20 2.81 3.02 -OP 153 BIO OK? 76 78 2 1.34 1.40 0.55 1.12 1.28 -OP 154 EU OK? 71 88 17 1.26 1.58 0.59 1.33 2.30 -OP 155 EU OK? 73 66 -7 0.88 111 0.19 1.55 0.70 -OP 156 EU OK? 69(65-72) 55 -14 2.19 3.13 1.49 1.78 1.73 -OP 157 EU OK? 60 53 -7 1.39 1.59 1.08 2.22 1.94 -OP 158 EU OK? 57 73 16 1.87 2.33 0.82 1.88 2.27 -OP 280 Method for Quality of No obtaining the Observed Predicted Pred. - Rz Tlz” Zotz" Vx Error %Abs. %Abs. %Abs. %Abs. obs. Code Data 159 EUOK? 50 54 4 2.54 3.25 0.95 2.23 1.51 -OP 160 EUOK? 16 37 21 1.00 1.83 1.63 1.97 1.31 -OP 161 EUOK? 3 55 52 2.52 2.89 1.08 2.18 1.72 -OP 162 EU OK? 1 -6 -7 3.33 3.13 0.87 6.94 4.28 -OP 163 EUOK? 0.3 -8 -8 2.74 3.51 2.20 5.05 3.26 -OP 164 Check 100 90 -10 1.61 1.89 0.32 1.59 2.39 -OP 165 Check 100 110 10 1.29 1.38 0.00 1.20 2.79 -OP 166 Check 95 82 -13 1.64 1.96 0.35 1.84 2.23 -OP 167 Check 60 100 40 3.47 3.04 0.28 1.81 2.92 -OP 168 Check 50 77 27 1.62 1.99 0.11 1.83 1.55 -OP 169 Check 28 54 26 2.20 2.44 0.91 2.44 1.92 -OP 170 Check 100 90 171 Check 91 90 172 Check Poor 45 2.16 1.84 0.04 4.80 3.65 -OP Zwitterionic drugs 173 BIO Good 100 53 -47 2.72 3.07 1.12 2.80 2.49 -lOP 174 BIO Good 100 80 -20 2.27 2.64 0.00 2.54 2.50 -lOP 175 BIO Good 100 85 -15 2.05 2.45 0.00 2.22 2.41 -lOP 176 EU Good 100 69 -31 2.54 2.77 0.58 2.50 2.43 -lOP 177 EU Good 100 72 -28 2.34 2.74 0.58 2.32 2.39 -lOP 178 Check 100 70 -30 0.47 0.59 0.52 0.89 0.56 -lOP 179 BIO+RAP Good 93 52 -41 2.53 2.90 1.12 2.85 2.54 -lOP 180 BIO Good 90 100 10 1.93 1.56 0.59 1.23 2.89 -lOP 181 Check 90 113 23 3.90 3.15 0.59 1.71 3.98 -lOP 182 BIO OK 88 88 0 2.84 3.20 0.17 2.20 2.63 -lOP 183 EU OK 88 94 6 2.10 2.07 0.59 1.50 2.81 -lOP 184 EU OK 88 77 -11 0.86 1.11 0.59 0.78 0.89 -lOP 185 BIO OK 86 55 -31 1.36 1.30 1.36 1.50 1.43 -lOP 186 BIO+RAP OK 75 37 -38 3.65 4.11 1.73 3.48 3.04 -lOP 187 RAP OK 62 67 5 2.36 2.60 0.58 2.56 2.48 -lOP 188 EUOK? 58 72 14 0.56 0.76 0.77 0.93 1.09 -lOP 189 RA Good 55 76 21 0.62 0.83 0.77 0.91 1.30 -lOP 190 RAP OK 55 65 10 0.46 0.74 0.79 1.34 1.26 -lOP 191 Good 41 56 15 1.35 1.26 1.36 1.54 1.57 -lOP 192 Check I 43 42 4.47 5.36 1.28 4.08 3.48 -lOP Drugs with missing fragments from ABSOLV program 193 RA Good 8 96 88 1.75 2.23 0.00 2.30 3.32 -20P 194 RA Good 100 70 -30 2.10 1.99 0.49 2.01 1.82 -60P 195 BIO Good 98 89 -9 1.31 1.34 0.94 0.79 2.07 -60P 196 BIO Good 95 80 -15 1.60 1.49 0.64 1.16 1.53 -60P 197 EUOK 85 -70 -155 4.31 6.77 4.02 7.65 4.91 -60P 198 RA Good 72 60 -12 2.62 3.23 0.97 2.58 2.43 -60P 199 EUOK? 50 -45 -95 3.96 6.12 3.18 7.20 4.84 -60P 200 RA Good 32 65 33 1.45 1.89 0.43 2.01 1.47 -60P 201 BIO+RAP OK 23 116 93 0.87 0.62 0.00 0.13 1.72 -60P Dose-limited, dose-dependent and formulation-dependent drugs 202 EUBDL 73 108 35 2.25 3.74 0.00 1.82 3.17 -OP 203 BIO DL 50(25-75) 68 18 3.03 3.81 0.38 3.81 3.90 -OP 204 RAP DL 47 78 31 2.97 4.16 0.31 3.22 3.49 -OP 205 RA DL 44(36-52) 70 26 2.58 3.89 0.41 3.31 3.36 -OP 206 RAP+BIO DL 37 58 21 1.97 3.26 • 0.93 5.04 6.00 -OP 207 RA DL 36 90 54 1.61 3.25 0.50 2.92 4.47 -OP 281 Method for Quality of No obtaining the Observed Predicted Pred. - Error R2 7I2" %" Vx %Abs. %Abs. %Abs. %Abs. obs. Code Data 208 RA DL? 34 73 39 1.37 2.08 1.63 1.81 3.37 -OP 209 BIO DL 28(10-65) 36 8 3.97 6.84 1.54 8.65 10.02 -OP 210 REV DL 28 66 38 3.94 4.38 0.84 4.03 4.48 -OP 211 BIO+EU DL 12(0.7-23) 76 74 3.51 2.91 0.81 2.93 3.73 -OP 212 RAPDL? 1 58 57 2.66 3.38 0.91 2.93 2.73 -OP 213 RAP DL 1.9 78 76 3.44 3.39 1.57 1.33 2.50 -60P 214 EUDL? 59(56-61) 83 24 3.18 3.10 1.27 1.49 2.70 -lOP 215 EU DL? >37 101 64 2.34 2.43 0.28 1.84 3.27 -lOP 216 BIO+EF DL? 28(25-50) 69 41 1.79 2.42 0.96 2.56 3.19 -lOP 217 EU DL? 25(10-40) 74 49 1.60 2.18 0.78 2.08 2.66 -lOP 218 EUDL 3(2-5) 17 14 3.70 5.37 3.37 5.70 7.12 -lOP 219 RA DL? 1 50 49 2.77 4.36 1.15 3.18 2.76 -lOP 220 BIO DP 69(37-100) 106 37 2.28 2.57 0.43 1.91 3.89 -OP 221 BIO DP 64(39-88) 78 14 1.60 2.29 0.20 2.28 2.40 -OP 222 RA DP 49(36-61) 56 7 2.18 3.12 1.21 1.97 1.69 -OP 223 BIO DP 23(15-30) 63 40 2.34 2.67 0.83 1.87 1.52 -OP 224 EF DP 71 79 8 2.08 2.46 0.31 2.10 2.27 -lOP 225 RA DP 70(57-83) 56 -14 3.91 4.73 1.80 2.77 3.22 -lOP 226 BIO+EU DP 59(43-74) 78 19 0.63 0.83 0.77 0.93 1.44 -lOP 227 EUB FD 86(77-95) 87 1 3.40 3.81 0.25 2.38 2.74 -OP 228 EU DL?+M 24(17-31) 80 56 2.21 1.61 1.43 0.81 2.03 -OP Drugs expected to have higher absorption 228 EU DL7+M 24(17-31) 80 56 2.21 1.61 1.43 0.81 2.03 -OP 229 RAP >69 80 11 2.27 2.57 0.31 2.10 2.30 -lOP 230 BIO+RAP >33 44 II 1.71 2.66 1.13 2.44 1.58 -OP 231 RAP >29 80 51 1.41 1.75 0.67 2.01 2.84 -OP 232 RAP >10 73 63 2.84 3.17 1.46 1.50 2.34 -lOP 233 RAP >5 62 57 1.80 2.37 0.74 2.66 2.57 -OP 234 EU+EF 66(61-71) 89 23 1.50 2.29 0.28 2.09 2.94 -lOP 235 EU-EF 59(49-68) 76 17 2.20 2.58 0.40 2.28 2.44 -OP 236 EU >65 78 13 1.85 2.52 0.77 1.76 2.36 -OP 237 EU DL? >30 77 47 2.18 2.49 0.56 2.06 2.38 -OP 238 EU >20 64 44 2.05 2.25 0.98 1.82 1.77 -OP 239 EU DL >10 109 99 1.29 2.22 0.35 1.32 3.29 -OP 240 EU >0.4 67 66 3.10 3.64 1.35 2.41 3.04 -OP 241 BIO >35 61 26 1.97 3.55 1.02 4.71 5.77 -OP Error Code: 1. -OP: The descriptors are adequately calculated using the ABSOLV program. 2. -IGF: Zwitterionic molecule. The program calculates descriptors for the neutral form, but in many environments the charged form will dominate. 3. -20P: Charged molecule. There are no fragments for these molecules, such as CO;' or 4. -60P: Indicates missing fragment(s). The molecule contains an atom or atoms for which no fragment values have been defined. The descriptors calculated from ABSOLV will therefore be incorrect. 5. The drugs assigned an error code of-lOP, -20P or -60P were not used in the regression analyses. 6. Dose-limited drugs are defined as those for which : 100x[Dosex(l- fraction absorbed) - 0.25xSw]/Dose > 20%. 282 7.3 Rat Intestinal Absorption 7.3.1 Introduction Modem drug design not only focuses on the pharmacological activity of a compound but also considers its ability to be absorbed and to reach its site of action. One of the most common methods used to screen the ability of a compound to be absorbed is intestinal permeability in rats {in situ perfusion experiments). Whilst these methods provide the advantage of experimental control of conditions such as pH and provide insight into the mechanism of drug absorption, information obtained focuses on the rate of diffusion across the rat small intestine. Many other factors including solubility, formulation, food composition, gastric emptying time, and blood flow (detailed in Chapter 7.1.2) influence absorption and are not fully considered in perfusion experiments. Therefore, in vivo pharmacokinetic experiments in animals are often carried out in order to obtain the percentage of drug absorbed and the rat is one of the most common animals employed in preclinical oral absorption studies. Chiou et al recently reported an excellent overall linear correlation (r^=0.97) for the percentage of oral dose absorbed in humans and rats for 64 drug compounds with wide physicochemical and pharmacological properties. This indicates that the evaluation of in vivo rat absorption following oral administration of drug in a solution or rapidly released dosage form may be used as an alternative method to satisfactorily predict the extent of GI absorption in humans following oral administration of drugs in a solution or rapidly released dosage form. In Chapter 7.2, the human intestinal absorption of 241 drug and drug-like compounds with wide ranging physicochemical properties was evaluated. Among them, 169 drugs with % absorption considered to be reliable or relatively reliable were used for QSAR analysis. The results showed that the Abraham General Solvation Equation could successfully model human absorption. Furthermore, there is agreement with previous studies, in that hydrogen-bond acidity and basicity are found to be very important descriptors in human absorption drug m odelling^'A lthough many research groups have established quantitative structure relationships between human absorption and molecular descriptors^^'^^'^ 11-313^ only the study of Chiou and Barve^^ has compared human and rat absorption for a large data set (64 compounds). There is thus further scope to evaluate and assess rat oral absorption data from literature and to obtain a 283 robust multiple linear regression model. This will allow both the investigation of the relationship between drug absorption in humans and rats, and also the comparison of the General Solvation Equations derived for each of the models. To achieve these aims, rat oral absorption for 111 drug and drug-like compounds were evaluated from literature data. 7.3.2 Evaluation of Rat intestinal absorption data The rat intestinal absorption dosed orally by gavage was collected and evaluated from 111 sources of literature. An additional 36 absorption values were taken directly from the Chiou and Barve^^ paper. Methods of obtaining qualified oral absorption have previously been reported from literature^% and in Chapter 7.2.1. One of the best ways of obtaining oral absorption data is from the ratio of cumulative urinary excretion of drug and drug-related materials following oral and intravenous administration; this method being especially applicable in the case of low absorption compounds. Table 7.08 lists the absorption data evaluated mostly by this method, but also some absorption data obtained from the percentage of urinary and biliary excretion. The following key is used as an indication of the source and quality of the data. RA: The absorption is evaluated from the ratio of cumulative urinary excretion of compounds following oral and intravenous administration. EU: The absorption is obtained from the urinary excretion of compounds following oral administration and recovery in urine is greater than 80%. EUB: The absorption is obtained from the sum of urinary and biliary excretion of compounds following oral administration and total recovery in urine and faeces is greater than 80%. EBF: The absorption is evaluated from the excretion in faeces and bile to bile duct- cannulated rats and total recovery in urine and faeces is greater than 80%. OK: The absorption is evaluated from the original papers; or from the ratio of urinary excretion of compounds following oral and intravenous administration and urinary excretion following oral and intravenous administration is greater than 20%; or from 284 biliary excretion to bile duct-cannulated rats and total recovery in urine and faeces is greater than 80%. OK? (Uncertain): The absorption is evaluated from the RA but the total recovery is not given. If the total recovery in urine and faeces following ether oral or intravenous administration is lower than 80%, a large estimation error will result. IVL: The excretion in urine is so low following intravenous administration (<20%) that the absorption data may not be reliable based on the method of the ratio of urinary excretion following oral and intravenous administration. TL: The total recovery in urine and faeces is lower than 80%. The absorption may not be reliable. V: Absorption is variable because of variable excretion in urine or bile. DP: Absorption is dose-dependent based on original papers. The single oral dose level to intact or bile duct-cannulated rats (mg/kg) by gavage in solution is also listed in Table 7.08. All of the values in urine, faeces and bile were obtained from studies using radiolabelled compounds except drug 43 and 111. The above methods may mis-estimate drug absorption if metabolism occurs in the gastrointestinal tract. 285 Table 7.08 Rat absorption, percentage of excretion in urine, bile and faeces of drugs and drug-like compounds Excretion Excretion in Excretion Total O ral dose Method for Quality of the No Names %Abs.* % A bs/ Ref. in urine'’ bUe* in faeces'* recovery* (mg/kg) obtaining %Abs.' %abs.'' data Training set I Alprenolol 38/37 /56 44/55 81/92 10 100 RA OK 35 2 Cisapride -100/100' 40-160 100 OK 315 3 Clofibrate 100/100' 87/83 3.6/7 91/90 25 100 RAOK 316 4 1,3-diphenyl-1 -triazene 76/80 20 95 RAOK 317 5 Carfecillin 95' 95 5 100 740 95 EU OK 133 6 Torasemide 70/78 /30 13/16 83/94 10 90 RA OK 318 7 1,3-diphenylguanidine 32/36 45/46 77/82 321 89 RA OK 319 8 Amlodipine 33/38 58/60 91/98 10 87 RA OK 320 9 Trimethoprim 85 10 95 20 85 EU OK 321 (0 Ramatroban 83 (79) 4 83 EUB OK 322 11 Tetrapeptide >75 75/90 14/14 89/104 1 83 RA OK 323 12 Casodex -80 30/37 71/58 101/95 25 81 RA OK 324 13 Saccharin 100' 74-83 16 98-104 16-22 79 EU OK 325 14 Acifran 61/84 29/4.8 90/89 10 73 RA OK 326 15 Valaciclovir 65/95 98/99 25 68 RA OK 327 16 Colterol 43/83 43/23 86/107 1.41 52 RA OK 328 17 Fosfomycin 46/95 0.07/0.1 100 48 RA OK 329 18 YM17E >40 0.4(l)/0.6 39 97/95 97/96 10 40 EUB OK 330 19 Fosmidomycin 34/90 /0.2 61/3 95/93 10 38 RA OK 331 20 Doxycycline 9-11/30 87/71 99/101 10 33 RA OK 332 21 Bromocriptine 32-40/32-40' 23 (68) (91) 32 EBF OK 333 22 Pamaqueside 0.1(5.8)/(34) 11/40.6 98(76)/(1.3) 98(93) 100 24 EBF OK 334 23 Xamoterol 19 8.0/43 89/52 97/95 5 19 RA OK 335 24 Cyclosporin 39.5' 6.6/8.3 9.1/59 85/77 92/86 10-30 16 EUB OK 336 25 Acarbose 1-2 3.2-14/94 0/0.2 (80-103)/(0.91) 2-200 2 RA OK 105 26 Pamidronate 0.5 0.12 /0.07 20 0.5 OK Test set 27 Etintidine 70/63 28/23 98/86 20 100 RA OK 338 28 Felbamate 100 66(47)/(58) 52/40 24(0)/(0) 90 100 100 EBF OK 339 286 QuaUty of the Excretion Excretion in Excretion Total O ral dose % A bs/ Method for Ref. %abs.'' data No Names % A bs/ bile* (mg/kg) obtaining %Abs/ in urine'’ in faeces'* recovery* 29 Lorcainide 36/29 70 61/67 97/96 5 100 RAOK 340 30 Lormetazepam 100 20/19 0.25 100 RAOK 341 31 Nilvadipine -100 24/21 /74 69/80 93/101 10 100 RAOK 342 32 Propranolol 99' 68/52 27/24 95/76 10 100 RA OK 343 33 Propylthiouracil -100 90/75 1.5 20 100 RA OK 344 34 Rimantadine 81-87/81-87 60 100 RA OK 345 35 Salicylicacid 100' -100 1 1 100 5-50 100 EUB OK 346 36 Sultopride 100' 71-82 29 24-30 95-100 10-100 100 EUB OK 89 37 Timolol 58/51 26/28 84/79 10 100 RA OK 158 38 Tinidazole 100 64/65 25/25 89/90 100 99 RA OK 347 39 Tiopinac 99 60.7/61.3 31/33 92/94 2 99 RA OK 348 40 Acetylsalicylicacid -100 86/88 2.0/3 88/91 10 98 RAOK 349 41 Exaprolol 98 48/49 1 98 RAOK 350 42 Camazepam 97/97' 29/30 58/65 87/95 20 97 RA OK 111 43 Venlafaxine 91' 97 97 22.1 97 EU OK 351 44 Viloxazine 100' 91-101 1-8 95-101 4-250 96 EU OK 352 45 Nizatidine 100' 57(55) 22 27(6) 85 10 86(77-94) EUB-EBF OK 353 46 Acetaminophen 98' 92 100 92 EU OK 354 47 Fenclofenac 100' 2-4 79-84 86(1) 87-91 10 92(84-99) EUB-EBF OK 355 48 Granisetron 100/100' 35/38(32) 42/53 61/59 96/97 0.25-5 92 RA OK 120 49 Enciprazine 32/36 /79 59/57 91/93 20 89 RA OK 356 50 Dofetillde -100 54/61 41/36 95/97 7 88 RA OK 357 51 Omeprazole 23/26 73/72 96/98 34.5 88 RA OK 358 52 Ketorolac 87/87' 69/79 24/18 94/98 1 87 RA OK 359 53 Bitolterol 65/79 24/24 89/103 1.14 82 RA OK 328 54 Felodipine 100' 26/32(15) /74 55 81 1.92 81 RAOK 360,361 55 Pentacaine 79 32/40 2 79 RA OK 362 56 Dapsone (45) 24 (22) 5 74(69-79) EUB-EBF OK 363 57 Terbutaline 60' 12(41V48 44-29 61/35 73/83 1 78(70-85) EUBOK 364 58 Carvedilol (3.48)/(12) 68/81 (26)/(1.01) (98)/(94) 30 71 EUBOK 365 59 Loprazolam -68 6.7(13) 46 90(15) 97 5 68 OK 366 60 Pravastatin 65 4/4.8 58/90 87/83 91/88 20 62 EUB OK 300 61 Ayitriptan 9.8/22 62/54 85/70 95/92 20 59(45-72) RA-EUB OK 367 62 Atenolol 48-50' 42-49/88(71-89) /2.5 51-56/13 99/101(73-91) 9-80 52(48-56) RA OK 368,369 287 Quality of the No Names %Abs.' Excretion Excretion in Excretion Total O ral dose %Abs/ Method for Ref. %abs.*' bile' (mg/kg) obtaining %Abs.‘ in urine*’ in faeces'* recovery* data 63 MK-499 10/21.0 40 78/68 88/89 0.25 48 RA OK 370 64 PCE22716 29/61 53/33 82/94 5 43 RA OK 371 65 Fenoterol 42 42 RA OK 200 66 Recainam 25/59 72/39 97/98 8.5 42 RA OK 214 67 Ziprasidone (22) 20 (55) (97) 10 42 EUB OK 372 68 Bumetanide 15/37 79/61 94/98 5 41 RA OK 373 69 Nadolol 14-18/18' 8.6-11/62 88-84/31 96/93 20 18 310,374 70 Reproterol 18 58 18 RA OK 375 71 Alendronate -0.9 /35 /0.36 /35 10 0.9 RA OK 376 72 (-)-6-aminocarbovir 32/47 60 68 RA OK? 377 73 Pidotimod 31/76 100 41 RA OK? 378 74 Benidipine HCl 9-20(6-17) 42-51 73-91 1-0.5 58(48-68) EUB V 379 75 Sumatriptan -50/50* 12-30/45-71 63-71/17-23 1-5 -50 RA V 218,380 76 Sulpiride 15-20 20-30 75 90-95 43(35-50) EUB V 381 77 Carbovir 30-18/77-42 41-60/4-31 71-78/81-73 10-60 -40 RA V 382 78 Pafenolol 16-31 10-20/58-69 80/30 95/95 0.34-8.4 24(16-31) RA V 383 79 Mespirenone 15.5/10 63/31 79/41 2-200 100 RA TL 384 80 Nufenoxole 8 60 61 69 1 79(68-99) EUB-EBF TL 385 81 Carbidopa 16/66 8/1.0 52/10 68/76 20 54 EBF TL 386 82 Nileprost 12/27.0 /83 67/59 79/86 0.2 44 RA TL 387 83 ANSA 7-8/60 58-80/16 77/76 45 13 RA TL 388 84 Toremifene 2.8/3.5 70/62 73/65 3-48 80 RA IVL 389 85 Amodiaquine 7.0/10 77/91 84/101 8.6 70 RA IVL 390 86 Prazosin 8.3/12.5 40(24h) 81/75 89/88 1 66 RA IVL 391 87 Crisnatol 6.8/10.7 90/83 97/94 5 64 RA IVL 392 88 Spironolactone 4.69/7.46 74.19/90.18 79/97 5 63 RA rvL 393 89 Bromerguride 4.5/7.5 87/89 92/97 0.25 60 RA rvL 394 90 Lacidipine 3-8/5-14 88-95/76-96 97/96 2.5 -50 RA IVL 395 91 Lovastatin 29/29* 0.6/2.1 78/96 79/98 29 RA IVL+TL 300 92 Idarubicin -50 5.3/16 76/81 81/97 1 -50 RA IVL 396 93 BMS-18374 2.9 2.2(0.5)/18(14) 0.6/41 103/95 104/109 0.015 12 RA IVL 397 94 Tripeptoid 3-8 0.44/5.52 94/91 94/97 1 6 RA IVL 323 288 Quality of the No Names %Abs.* Excretion Excretion in Excretion Total O ral dose %Abs.' Method for Ref. %abs.'' bile* (mg/kg) obtaining %Abs.‘ in urine'’ in faeces'* recovery* data 95 Acyclovir 21' 8.0/90 88/9.5 100-900 9 RA DP 398 96 Diflubenzuron -50 23(21) 32-41 50-69 73-92 4 68(58-77) EUB+EBF DP 399 97 Ranitidine 63‘ 63 33 96 >63 EU 400 zwitterionic drugs or compounds for which ABSOLV cannot calculate descrip ors due to missi ng fragmen S 98 MK-711 29.5/38 55/44 85/82 1 78 RA OK 401 99 L-canavanine 61/83 2.0/0 63/83 2000 73 RA TL 402 100 Perindopril 33/44-54 59/41-36 92/86 0.5 67 RA OK 403 101 Ramipril 56 26 71 97 56 RA OK 404 102 Prulifloxacin 44 14(15.5) 35 83(44) 97 20 50 EUB OK 405 103 Apovincaminc acid 50 18/44 62/35 80/79 10 50 RA OK 406 104 Inogatran 30-35 16-21/52-60 67-72/35-45 95/98 -1 33(30-35) RA OK 407 105 CCS 16617 6.3/93 0.1-2.4 89/5.3 95/98 10 7 RA OK 408 106 NAD394 3.4/63 93/33 96/96 20 5.4 RA OK 405 107 Azidocillin 23/56 /52 60/39 83/95 15 41 RA OK 409 108 Pranolium chloride 8-13 7.2/40 <5/25 76/40 83/80 1 18 RA OK 410 109 Oxitropium bromide 14 /-50 12.0/14 /-50 14 RA OK 411 110 Ipratropium bromide 17-35 5.5/58 3.2/18 17-35 RA OK? 412 111 lothalamatesodium 2.1-7/1.9‘ 3.8-4.9/98 20-800 4.2 RA OK 413 ® Absorption data was given from the original literature/Chiou^'. ** Percentage of cumulative drug and its metabolites in urine following oral/intravenous administration to intact (bile duct-cannulated) rats. Percentage of cumulative drug and its metabolites in bile by oral/intravenous administration to bile duct-cannulated rats. ** Percentage of cumulative excretion in faeces by oral/intravenous administration to intact (bile duct-cannulated) rats. ® Total recovery of cumulative excretion in urine and faeces by oral/intravenous administration to intact (bile duct-cannulated) rats. ^ Absorption data (or averaged values) chosen here based on the analysis of excretion in urine, bile and faeces. ^ Method for obtaining absorption data (%Abs.^). ** Quality of the absorption data based on the analysis of literature values. ' Absorption is from Chiou data set^\ Notes for some of the drugs: 1. Drugs 17 and 111: absorption is obtained from the ratio of urinary excretion of parent drug following oral and intravenous administration. 2. Drug 26: amount absorbed within 72h was means used as radioactivity in urine and organs. 3. Drug 59: absorption is evaluated from the fraction of cumulative excretion in urine, faeces and amount in intestine gut (16%). 4. Drugs 65, 70, 101 and 109: absorption is obtained from the ratio of urinary excretion following oral and intravenous administration. 5. Drug 69: absorption is evaluated from the ratio of urinary excretion of the drug following oral and intraperitoneal administration. 6. Drug 71: absorption is based on the ratio of amount in bones following oral and intravenous administration. 289 7.3.3 Relationship between rat intestinal absorption and Abraham descriptors The General Solvation Equation (eqn 7) has been developed by the Abraham group and recently applied to human oral absorption in Chapter 7.2.2. SP = c + r.R2 + s.7i‘^ + a.^CL^ + 6.2^2^ + v. Vx (7) In this sub-chapter, SP will refer to % of drug absorbed (%Abs) in the rat intestine. Abraham descriptors and rat absorption data from the Chiou and Barve^^ data set (excluding PEG 900 and PEG 4000 for which compound structures are unavailable) are listed in Table 7.09. Abraham descriptors were calculated for each of these compounds using the ABSOLV program (details in Chapter 1.4.6, page 31). Regression analysis was then performed using only the drugs which ABSOLV had assigned a code of -OP (indicating that the descriptors were reliable). The resulting correlation (Fig. 7.09) between rat absorption and the Abraham descriptors for the Chiou dataset was very poor (r^=0.46, model 1) in comparison to the human absorption model (r^=0.83), see Table 7.10. Removal of the three significant outliers (acyclovir, lovastatin and nadolol), indicated as A in Fig. 7.09, did not substantially improve the regression coefficient (r^=0.66, model 2). This was inevitable because few low absorption data were available for analysis and also because some of the absorption data does not seem to be reliable enough (i.e. lovastatin in Table 7.08). Fig 7.09 Plot of observed vs predicted % absorption (model 1) for 51 drugs from Chiou and Barve data set 120 100 I 1Q. 0 20 40 60 80 100 Observed 290 Table 7.09 Observed rat absorption and absorption calculated from Abraham descriptors %Abs.“ %Abs. Pred.- No Names Zaz" Vx Code' (Pred.) obs. R2 Training set 1 Alprenolol 100 85 -15 1.25 1.03 0.10 1.25 2.16 -OP 2 Cisapride 100/100" 88 -12 2.30 3.40 0.46 2.04 3.40 -OP 3 Clofibrate 100/100" 100 0 0.93 1.23 0.00 0.69 1.82 -OP 4 1,3-diphenyl-1 -triazene 95 85 -10 1.82 1.30 0.20 0.85 1.58 -OP 5 Carfecillin 95/95" 73 -22 2.83 3.31 0.56 2.47 3.20 -OP 6 Torasemide 90 69 -21 2.14 2.95 1.12 1.90 2.58 -OP 7 1,3-diphenylguanidine 89 80 -9 2.07 1.84 0.39 1.22 1.72 -OP 8 Amlodipine 87 81 -6 1.82 2.37 0.35 2.07 3.02 -OP 9 Trimethoprim 85 79 -6 2.52 2.81 0.50 1.76 2.18 -OP 10 Ramatroban 83 72 -11 2.58 3.02 1.03 1.61 2.90 -OP 11 Tetrapeptide 83 64 -19 2.92 5.26 1.44 3.58 4.45 -OP 12 Casodex 81 104 23 2.02 3.85 0.50 1.34 2.71 -OP 13 Saccharin 79/100" 85 6 1.59 2.14 0.68 0.95 1.15 -OP 14 Acifran 73 74 1 1.39 0.88 0.50 1.09 1.56 -OP 15 Valaciclovir 68 71 3 2.35 2.92 0.59 2.41 2.34 -OP 16 Colterol 52 57 5 1.34 1.15 1.03 1.73 1.84 -OP 17 Fosfomycin 48 54 6 0.67 1.42 1.52 1.78 0.86 -OP 18 YM17E 40 64 24 3.41 3.49 1.03 2.21 4.57 -OP 19 Fosmidomycin 38 50 12 0.83 2.44 1.88 2.37 1.23 -OP 20 Doxycycline 33 43 10 3.22 3.92 1.62 3.16 3.10 -OP 21 Bromocriptine 32/32-40" 49 17 3.94 4.38 0.84 4.03 4.48 -OP 22 Pamaqueside 24 36 12 3.29 4.60 1.44 4.92 5.80 -OP 23 Xamoterol 19 63 44 1.80 2.37 0.74 2.66 2.57 -OP 24 Cyclosporin 16/39.5" 12 -4 3.97 6.84 1.54 8.65 10.02 -OP 25 Acarbose 2 -11 -13 3.31 4.47 2.53 6.19 4.38 -OP 26 Pamidronate 0.5 -5 -5 1.15 2.68 3.55 3.59 1.45 -OP Test set 27 Etintidine 100 70 -30 1.71 2.25 0.67 2.18 2.15 -OP 28 Felbamate 100 76 -24 1.44 1.48 0.70 1.12 1.77 -OP 29 Lorcainide 100 96 -4 2.20 2.31 0.00 1.19 2.96 -OP 30 Lormetazepam 100 85 -15 2.69 2.37 0.10 1.39 2.26 -OP 31 Nilvadipine 100 93 -7 1.67 2.80 0.32 1.69 2.83 -OP 32 Propranolol 100/99" 82 -18 1.85 1.36 0.10 1.29 2.15 -OP 33 Propylthiouracil 100 79 -21 1.34 1.40 0.55 1.12 1.28 -OP 34 Rimantadine 100 89 -11 0.88 0.60 0.18 0.65 1.57 -OP 35 Salicylicacid 100/100" 81 -19 1.05 0.89 0.72 0.38 0.99 -OP 36 Sultopride 100/100" 93 -7 1.77 3.25 0.22 2.18 2.71 -OP 37 Timolol 100 82 -18 1.47 1.81 0.10 2.03 2.38 -OP 38 Tinidazole 99 105 6 1.40 2.77 0.00 1.36 1.70 -OP 39 Tiopinac 99 81 -18 2.21 2.11 0.59 0.99 2.03 -OP 40 Acetylsalicylicacid 98 86 -12 0.93 1.35 0.59 0.80 1.29 -OP 41 Exaprolol 98 86 -12 1.31 1.10 0.10 1.23 2.52 -OP 42 Camazepam 97/97" 85 -12 2.63 2.56 0.00 1.84 2.67 -OP 43 Venlafaxine 97/97" 82 -15 1.24 1.32 0.35 1.36 2.37 -OP 44 Viloxazine 96/100" 82 -14 1.15 1.42 0.32 1.47 1.87 -OP 45 Nizatidine 86(77-94)/100" 80 -15 1.87 2.55 0.20 2.41 2.46 -OP 46 Acetaminophen 92/98" 77 -15 1.27 1.81 1.02 0.85 1.17 -OP 47 Fenclofenac 92(84-99)/100" 86 -6 1.80 1.76 0.59 0.62 1.98 -OP 48 Granisetron 92/100" 78 -14 2.18 2.53 0.37 2.03 2.45 -OP 49 Enciprazine 89 75 -14 2.27 2.34 0.17 2.47 3.32 -OP 50 Dofetilide 88 72 -16 2.47 3.62 1.10 2.21 3.23 -OP 51 Omeprazole 88 85 -3 2.35 3.41 0.42 2.16 2.52 -OP 291 %Abs.' %Abs. Pred.- No Names Zaz" Zpz" Vx Code" (Pred.) obs. 712 52 Ketorolac 87/87" 82 -5 1.69 2.02 0.59 1.23 1.87 -OP 53 Bitolterol 82 82 0 2.42 2.42 0.19 1.97 3.65 -OP 54 Felodipine 81/100" 89 8 1.75 2.17 0.32 1.37 2.71 -OP Pentacaine 79 83 4 1.64 1.70 0.37 1.40 3.09 -OP Dapsone 74(69-79) 92 14 2.32 3.24 0.50 1.25 1.80 -OP Terbutaline 78(70-85)/60" 54 -24 1.41 1.40 1.28 1.74 1.84 -OP Carvedilol 71 70 -1 3.09 2.49 0.32 2.12 3.10 -OP Loprazolam 68 78 10 3.60 3.99 0.00 3.05 3.23 -OP Pravastatin 62 57 -5 1.37 2.08 1.63 1.81 3.37 -OP Avitriptan 59(45-72) 61 2 3.11 3.64 0.82 2.96 3.27 -OP 52(48-56)/48-50 62 Atenolol 77 25 1.45 1.89 0.55 1.75 2.18 -OP 63 MK-499 48 70 22 2.90 3.54 0.90 2.32 3.44 -OP 64 PCE22716 43 61 18 2.65 2.18 0.58 2.26 2.50 -OP 65 Fenoterol 42 40 -2 2.21 2.16 1.82 2.05 2.36 -OP 66 Recainam 42 80 38 1.26 1.58 0.59 1.33 2.30 -OP 67 Ziprasidone 42 77 35 3.47 3.04 0.28 1.81 2.92 -OP 68 Bumetanide 41 61 20 2.20 2.73 1.41 1.76 2.64 -OP 69 Nadolol 18/18" 67 49 1.61 1.63 0.70 1.88 2.49 -OP 70 Reproterol 18 41 23 3.23 3.15 1.28 3.06 2.81 -OP 71 Alendronate 0.9 -4 -5 1.15 2.68 3.55 3.59 1.59 -OP Unreliable absorption 72 (-)-6-aminocarbovir 68 48 -20 2.16 2.45 1.02 3.04 1.82 -OP 73 Pidotimod 41 72 31 1.59 2.39 0.82 1.95 1.63 -OP 74 Benidipine HCl 58(48-68) 83 25 2.66 3.25 0.32 2.24 3.80 -OP 75 Sumatriptan -50/50" 70 20 1.87 2.33 0.82 1.88 2.27 -OP 76 Sulpiride 43(35-50) 76 26 1.84 2.93 0.75 2.21 2.53 -OP 77 Carbovir -40 67 27 2.61 2.73 0.92 1.70 1.73 -OP 78 Pafenolol 24(16-31) 70 39 1.41 1.75 0.67 2.01 2.84 -OP 79 Mespirenone 100 104 4 2.57 3.92 0.00 1.86 3.16 -OP 80 Nufenoxole 79(68-99) 94 15 2.39 2.23 0.00 1.12 3.10 -OP 81 Carbidopa 54 46 -8 1.49 1.52 1.45 2.08 1.67 -OP 82 Nileprost 44 67 23 1.36 2.06 1.28 1.59 3.17 -OP 83 ANSA 13 70 57 2.43 2.56 0.98 1.36 1.58 -OP 84 Toremifene 80 91 11 2.43 2.03 0.02 1.11 3.30 -OP 85 Amodiaquine 70 66 -4 2.73 2.47 0.74 1.87 2.74 -OP 86 Prazosin 66 80 14 3.40 3.81 0.25 2.38 2.74 -OP 87 Crisnatol 64 69 5 3.48 2.07 0.29 1.51 2.75 -OP 88 Spironolactone 63 106 43 2.25 3.74 0.00 1.82 3.17 -OP 89 Bromerguride 60 71 11 2.58 2.44 0.47 2.04 2.87 -OP 90 Lacidipine -50 88 30 1.65 2.34 0.32 1.80 3.62 -OP 91 Lovastatin 29/29" 95 67 1.29 2.22 0.35 1.32 3.29 -OP 92 Idarubicin -50 56 6 3.32 2.52 0.63 2.49 3.47 -OP 93 BMS-18374 12 49 37 1.39 2.69 2.04 2.23 3.24 -OP 94 Tripeptoid 6 80 74 1.78 4.23 0.89 3.15 4.11 -OP 95 Acyclovir 9/21" 68 59 2.34 2.67 0.83 1.87 1.52 -OP 96 Diflubenzuron 68(58-77) 86 18 1.79 2.67 1.04 0.73 1.99 -OP 97 Ranitidine >63/63" 81 18 1.60 2.29 0.20 2.28 2.40 -OP Zwitterionic drugs for which ABSOLV cannot calculate descriptors due to missing fragments 98 MK-711 78 77 -1 2.21 1.90 0.59 1.15 2.65 -lOP 99 L-canavanine 73 63 -10 1.04 1.18 0.57 2.28 1.30 -lOP 100 Perindopril 67 85 18 1.06 1.94 0.28 2.07 2.93 -lOP 101 Ramipril 56 84 28 1.68 2.37 0.28 2.12 3.26 -lOP 102 Prulifloxacin 50 84 34 2.89 3.48 0.00 2.64 3.01 -lOP 103 Apovincaminc acid 50 71 21 2.31 2.00 0.59 1.61 2.44 -lOP 104 Inogatran 33(30-35) 61 28 1.95 3.13 0.76 3.50 3.51 -lOP 292 %Abs.“ %Abs. No Names Res. Zoti" Vx Code" (Pred.) Rl 105 CGS 16617 7 59 52 2.00 2.14 0.96 2.26 2.62 -lOP 106 NAD394 5.4 74 69 2.61 2.74 0.31 2.26 2.33 -lOP 107 Azidocillin 41 69 28 2.50 2.43 0.40 2.33 2.59 -60P 108 pranolium chloride 18 87 69 1.47 1.11 0.26 0.76 2.45 -60P 109 Oxitropium bromide 14 87 73 1.31 1.57 0.35 1.19 2.54 -60P 110 Ipratropium bromide 17-35 90 78 1.10 1.27 0.35 0.86 2.73 -60P 111 lothalamatesodium 4.2/1.9" 59 54 3.44 3.39 1.57 1.33 2.50 -60P Chiou data set** 112 Antipyrine 100 93 -7 1.53 1.58 0.00 1.05 1.48 -OP 113 Caffeine 100 86 -14 1.94 1.81 0.00 1.47 1.36 -OP 114 Cimetidine 100 72 -28 1.53 2.11 0.59 2.14 1.96 -OP 115 Codeine 100 76 -24 2.02 1.78 0.26 1.75 2.21 -OP 116 Diclofenac 100 78 -22 1.97 1.88 0.78 0.87 2.03 -OP 117 Ethinylestradiol 100 77 -23 2.12 2.50 0.97 1.16 2.39 -OP 118 Flumazenil 100 92 -8 1.80 2.38 0.00 1.76 2.09 -OP 119 Imipramine 100 88 -12 1.97 1.56 0.00 1.15 2.40 -OP 120 Isradipine 100 90 -10 1.67 2.46 0.32 1.62 2.71 -OP 121 Ketoprofen 100 84 -16 1.63 1.78 0.59 0.86 1.98 -OP 122 Morphine 100 67 -33 2.10 1.68 0.55 1.76 2.06 -OP 123 Oxatomide 100 69 -31 3.43 2.83 0.33 2.25 3.40 -OP 124 Phenglutarimide 100 83 -17 1.61 1.89 0.32 1.59 2.39 -OP 125 Progesterone 100 104 4 1.58 2.47 0.00 1.16 2.62 -OP 126 Bomaprine 100 93 -7 1.29 1.38 0.00 1.20 2.79 -OP 127 Tolmesoxide 100 101 1 1.19 2.21 0.00 1.28 1.62 -OP 128 Verapamil 100 93 -7 1.70 2.48 0.00 2.07 3.79 -OP 129 Theophylline 97 77 -20 1.93 1.84 0.42 1.38 1.22 -OP 130 Hydrocortisone 95 81 -14 2.06 3.16 0.72 1.98 2.80 -OP 131 Naproxen 92 81 -11 1.62 1.40 0.59 0.75 1.78 -OP 132 Captopril >71 83 12 1.15 1.68 0.50 1.31 1.62 -OP 133 Hydrochlorothiazide 65 62 -3 2.19 3.13 1.49 1.78 1.73 -OP 134 Chlorothiazide 60 66 6 2.18 3.12 1.21 1.97 1.69 -OP 135 Furosemide 60 64 4 2.05 2.55 1.36 1.47 2.10 -OP 136 Azithromycin 45 43 -2 1.97 3.26 0.93 5.04 6.00 -OP 137 Adefovir 7.8 43 35 2.08 3.00 1.76 2.72 1.79 -OP Zwitterionic drugs for which ABSOLV cannot calculate descriptors due to missing fragments 138 Cefadroxil 95 52 -43 2.72 3.07 1.12 2.80 2.49 -lOP 139 Levodopa 87 54 -33 1.36 1.30 1.36 1.50 1.43 -lOP 140 Gabapentin 79 77 -2 0.63 0.83 0.77 0.93 1.44 -lOP 141 Benazepril 50 82 32 2.34 2.43 0.28 1.84 3.27 -lOP 142 Enalapril 34 84 50 1.50 2.29 0.28 2.09 2.94 -lOP 143 Enalaprilat 11 70 59 1.60 2.18 0.78 2.08 2.66 -lOP 144 Amphotericin B 5 -11 -16 3.70 5.37 3.37 5.70 7.12 -lOP 145 Clonidine 100 75 -25 1.60 1.49 0.64 1.16 1.53 -60P 146 Ximoprofen 100 75 -25 1.31 1.34 0.94 0.79 2.07 -60P 147 Bretyliumtosylate 20 100 80 0.87 0.62 0.00 0.13 1.72 -60P “Absorption values obtained from Table 7.08. **Absorption values obtained from the Chiou data set"^\ Error Codes: -OP: The descriptors are adequately calculated using the ABSOLV program. -IGF: Zwitterionic molecule. The program calculates descriptors for the neutral form, but in many environments the charged form will dominate. -60P: Missing fragment(s). The molecule contains an atom or atoms for which no fragment values have been defined. The descriptors will therefore be incorrect. None of the drugs assigned an error code of -lOP or -60P were used in the regression analyses. 293 Although a relatively large data set of 64 compounds was collated by Chiou and Barve^\ there was insufficient low absorption data available. However, there was potentially much more data available in the open literature, and absorption values for 111 drugs were therefore evaluated from original references. Among these, the absorption of 71 drugs (Drugs 1-71, which contained 21 drugs used by Chiou and Barve^^) considered to be relatively reliable were used for QSAR analysis in this work (Table 7.08). The absorption of 26 dmgs (Drugs 72-97, which contained 4 drugs used by Chiou and Barve^^) were considered to be un-reliable and as such were removed from the total data set. The training set was chosen from drugs 1-71 by use of the alternative space filling design technique developed by Kennard and Stone^®^. As such, the descriptors of the training set were chosen to cover the whole descriptor space of the total set. The remaining compounds were used as a test set, and 26 drugs from Chiou and Barve^^ were used as a cross-validation set (Table 7.09), The equations derived from the regression analysis between the Abraham descriptors and the training set are shown in Table 7.10. For comparative purposes, the human absorption models reported in Chapter 7.2.2 are also included in this table. The details of drugs used for the rat absorption analysis are listed in Table 7.11. 294 Table 7.10 Regression results of training, test and cross-validation sets using Abraham descriptors No Data Set Models from training sets Training Set •— Test Set — — (Cross-validation Set — r' (adj) n SD RMSE F n RMSE AAE AE n RMSE AAE AE 1 Rat (Chiou) %Abs. = 96 + 4.I6R2+ 4.37% "-29.3% " 0.46 0.40 51 21 19 8 - 14.9% " + 4.88VX 2 Rat (Chiou) %Abs. = 100 + 1.17 R 2 + 4.20%" - 25.2% " 0.66 0.62 48 14 13 16 - I8 .6ZP2” + 8.70VX 3 Rat %Abs. = 100 - 10.3 R2 + 12.7%" - 22.7% " 0.77 0.71 26 17 15 30 45 18 15 2 26 17 14 9 -13.7SP2“ + 1.94VX 4 Rat %Abs. = 1 0 7 -1 1 .7 R 2+ 12.7%" - 0.84 0.80 25 14 12 20 45 17 14 0 26 16 13 7 -9.28Zp2” -0.897Vx 5 Rat %Abs. = 105 - 24.6202" - 6.54ZP2” 0.79 0.77 25 16 14 41 45 19 14 -1 26 14 12 7 6 Human %Abs. = 90 + 2.11 R2+ 1.70%" - 20.7202 " 0.83 0.80 38 16 14 31 131 14 11 -1 - 22.32P2" + 15.0VX 7 Human %Abs. = 92 -2 0 .0 % " - 21.92^2“ + 17.2VX 0.82 0.81 38 15 14 53 131 14 11 -1 8 Rat %Abs. = 112.4 - 16.0 R2+ 12.6%" - 25.52o2" 0.69 0.67 71 17 16 29 (Total data) -II.72 P2" + 1.60VX 9 Human %Abs. = 92 + 2.94 R2+ 4.IO712" - 2 I.7202 " 0.74 0.73 169 14 14 93 (Total data) - 2 I.I2 P2" + 10.6VX (adj) : Adjusted r-square, (adj) = l-((l-r^)(n-l/n-p)), where p is the number of descriptors used in the model. RMSE: Root mean square error, RMSE=[Z(calc-obs)%]®'^, AAE: average absolute error, AAE=E|calc-obs|/n., AE: average error, AE=Z(calc-obs)/n. 295 Table 7.11 Details of drugs used in analysis Drug no. in Tables Data set Number of drugs^ 7.08 and 7.09 Training set 26 (6) 1-26 Test set 45 (15) 27-71 Cross-validation set 26 (26) 112-137 Un-reliable data 27(4) 72-97 Zwitterionic drugs 16(7) 98-106,138-144 Drugs for which descriptors 8(4) 107-111, 145-147 could not be assigned^ Numbers in parentheses indicate drug compounds present in the Chiou dataset ^ Descriptors could not be calculated by ABSOLV because of missing fragments Considering the training sets, if the rat absorption model 3 is compared with the human absorption model 6, the regression results indicate that the R 2 and 712^ descriptors have a stronger influence on rat absorption than on human absorption. However, step-wise regression shows that the significant descriptors are S a 2^ and Sp 2^ (Model 5). This is in agreement with the result of human oral absorption in that polar terms dominate the absorption process (Model 7). Comparison of the total data sets for rat (Model 8) and humans (Model 9) reveals the same trend. The differing effects of R 2 and 712^ and Vx descriptors is discussed in the following text. Fig. 7.10 shows the predicted absorption for the test, cross-validation and training set. In most cases, the predicted absorption correlates well with the observed absorption; the notable exceptions being xamoterol in the training set and nadolol in the test set (highlighted as o and A respectively in Fig. 7.10). If human absorption is considered to be similar to rat absorption, nadolol may be a dose-dependent drug because in humans, absorption decreases with increasing dose^^^’^^"^. The method of obtaining the absorption data may be a contributing factor to the difference since the rat absorption of nadolol is obtained from the ratio of urinary excretion between oral and intraperitoneal administration It is not clear why the predicted value is much higher than the observed absorption for xamoterol. Removal of xamoterol from Model 3 improves the regression coefficient, and the resulting training set (Model 4) Model 4 is able to predict absorption for the test and cross-validation sets slightly more accurately than Model 3. 296 Fig. 7.10 Plot of predicted against observed rat % absorption 120 100 80 60 % 40 20 Training set Test set 0 Cross-validation set -20 20 40 60 80 100 Observed Table 7.09 also lists the predicted absorption values for the cross-validation set taken from the Chiou and Barve^' data set; in general, the predicted absorption compares well with the observed absorption. Only adefovir (0 in Fig. 7,10) has a large prediction error. Examining the reference'^''* from which this absorption was obtained, it was found that the absorption value was evaluated from the ratio of urinary excretion of parent drug following oral and intravenous administration. However, the details were not provided from the reference‘s A large estimation error is possible if the drug undergoes significant first-pass effect or if total recovery in urine and faeces is lower than 80%. The oral bioavailability of adefovir was reported in a previous reference's S'S as being 11%; the absorption should therefore be equal to or higher than this bioavailability value. For compounds with unreliable absorption data (Drugs 72-97), the absorption prediction for some of these drugs is in good agreement with the observed absorption, whilst for others there is little agreement with the observed absorption (Table 7.09). For 69% of these unreliable drug values, the difference between predicted and observed absorption is less than 30%. However, if drugs 1-71 are compared (for which the data is considered to be reliable) the percentage of drugs for which the difference between predicted and observed absorption is less than 30% is 94. 297 It is useful to examine the similarity between human and rat absorption models and to assess the possible application of the rat model to predict human absorption. Thus the rat absorption model was used to predict the human absorption values of 169 drugs detailed in Chapter 7.2. The result (Fig. 7.11) shows that in general the observed human absorption corresponds well to the absorption predicted from the rat absorption Model 2, especially for high absorption drugs. In comparison with the human absorption model, the rat absorption model was found to predict absorption 8% higher than the human model for 25 low human absorption drugs (1-50%) [where, 8% = Z(Predieted"'®‘^^' ^ - Predicted"""^^' ^) /25]. However, for 143 high human absorption drugs (51-100%), the prediction from the human and rat absorption models is similar [Z(Predieted absorption"’”*^^' ^-Predicted absorption"'®^^^' ^)/144 = -5.6%]. This general trend can be seen in Fig. 7.11. Fig. 7.11 Prediction of human intestinal % absorption from the rat absorption m odel 120 % A Rat model # Human model 20 80 100 Observed If the dosage forms are considered, it is not surprising that there are differences between the human and rat absorption models. The vast majority of drug administered in humans was in the form of a tablet or capsule with 100-300 ml of water. In contrast to humans, the rats were all orally administered with drugs dosed in solution or suspension, and in some cases with addition of organic solvents such as ethanol. 298 methylcellulose or com oil (see Table 7.08). Examining the human absorption administered orally both in solution and tablet form, it is evident that absorption for drugs dosed in solution is usually equal to or higher (sometimes significantly higher) than in solid forms^^’^^^’"^^^’"^^^. For solid drug forms, absorption may not only be controlled by passive diffusion, but also by the in vivo dissolution rate in the small intestinal fluid. Furthermore, the organic solvents used in the administration may also promote intestinal absorption of the This offers one reason as to why R%, 712^ and Vx descriptors have differing magnitudes of coefficients and hence influence in the human and rat absorption models. Since oral administration to rats is in solution or suspension, it is reasonable to assume that in vivo dissolution rate does not significantly affect the extent of rat intestinal absorption"^'^. This situation does not apply to human absorption in which drugs with poor solubility may exhibit incomplete dissolution and hence incomplete absorption^^^^^. Dissolution rate is the rate-determining step of absorption for these drugs"^'^. 7.4 Comparison Between Human and Rat Intestinal Absorption In Chapter 7.2, 241 human oral absorption data were evaluated from over 250 original papers and reviews. Among them, 169 absorption data were considered to be sufficiently reliable. Of these, 18 drugs were separated from the total of 241 drugs as dose-limited drugs because dissolution may be the rate-determining step of absorption for these compounds. These dose-limited drugs cannot be rapidly released into G1 gut fluid if a drug is dosed as a tablet or capsule and as such the extent of absorption will be poor/variable. To further compare the similarity or difference between the human and rat absorption data, 63 absorption values which are considered to be reliable for both humans and rats is selected from Table 7.08, the Chiou and Barve data set^^ (Table 7.09) and Chapter 7.2. It is believed that these drugs can be rapidly released into intestinal fluid from solid forms following human oral administration The 63 drugs in Table 7.09 are 1-3, 5-6, 13, 16-17, 19, 25, 28, 30, 32-33, 35-37, 39-40, 42-48, 51-52, 54, 62, 65, 67-68, 70, 112-133, 135, 137-139 and 145-147. The relationship between human and rat absorption for these compounds is shown in Fig. 7.12. 299 Fig. 7.12 Relationship between human and rat absorption 100 c n E mC3 w .Q < 0 20 40 60 80 100 %Abs,(rat) %Abs.(human) = 0.88%Abs.(rat) + 9.83 n = 60, = 0.88, SD = 8.2, F = 408 The data shows that for 92% of the drugs, the absorption difference between humans and rats is less than 20% and for 94% of drugs that the difference is less than 30%. There are only three drugs, indicated as o in Fig. 7.12, for which human absorption is significantly different from rat absorption. These anomalies are bumetanide (human/rat=94/41) reproterol (human/rat=60/18) and cimetidine (human/rat=64/100). The difference for cimetidine may be due to an estimation error of absorption in humans. Percentages of 62, 85 and 98 of human absorption for cimetidine were reported in references 58, 63 and 68, respectively, so that the true absorption figure is not conclusive. The human absorption value for reproterol was given in a review paper and therefore the method by which the absorption data was obtained needs to be checked. A large estimation error may result for low absorption drugs if intravenous administration is very low or total recovery in urine and faeces is lower than 80% following oral and intravenous administration, for example drugs 80-94. In the case of bumetanide, although both rat and human absorption data was evaluated from the ratio of urinary excretion following oral and intravenous administration, there is 55% absorption difference between the two species. The reason for this difference is unknown and experimental error may certainly be involved. Removal of these 3 significant outlier drugs resulted in a regression coefficient of 0.88 (Fig. 7.12). Inclusion of the three drugs gave a regression coefficient of 0.75. 300 Overall, rat (oral) intestinal absorption is quite similar to human (oral) intestinal absorption based on the above analysis of 63 drugs and this is in agreement with the results reported by Chiou et al^\ However, there are notably a few drugs with large absorption differences between species and the regression coefficient (r^=0.76) between rat and human data presented in this chapter is not as significant as that in the Chiou paper (r^=0.97). Furthermore, there are relatively few drugs available with low and qualified absorption data to be compared in humans and rats, and more in-depth studies using an even larger number of drugs seems warranted"^^^. 301 7.5 The Mechanism of Human Intestinal Absorption The mechanism of intestinal absorption has been discussed in numerous books and reviews, 1-2,6,418-420 and a general scheme may be formulated as shown in Fig. 7.13. If a drug is administered as a solid, then dissolution, kdiss, is the first step. The drug then crosses the lumen-gut interface, diffuses across the membrane, exits the membrane, and is then removed by perfusion into the blood stream. The rate-determining step may sometimes be dissolution of a solid and this has been discussed in depth by Dressman et al. However, if a drug crosses the intestinal membrane very rapidly, then the perfusion step may be rate-controlling. Even if neither the dissolution nor perfusion steps are rate determining, there are still various possible rate-determining steps. Fig. 7.13 A general mechanism for intestinal absorption; —> and <— represent kinetic paths with a given rate constant. up Solid Lumen Membrane kd^s perf k back Waterbeemd et al. have expanded the scheme shown in Fig. 7.13. In their mechanism, the rate constants at each of the membrane interfaces are themselves composite constants, as shown in Fig. 7.14 for the lumen/membrane interface. A drug has to diffuse through a stagnant layer, SL, on the aqueous side, cross the actual interface, and then diffuse through a SL organic layer on the membrane side. Each one of these stages offers some resistance to drug transport. Waterbeemd et al. constructed a set of equations that related to Fig. 7.14, mostly in terms of 302 water/membrane partition coefficients, and their detailed analysis helps to provide an explanation for the models shown in Fig. 7.13 and Fig. 7.14. Fig. 7.14 A mechanism for the path across the lumen/membrane interface; and <— represent kinetic paths with a given rate constant. Lumen Interface Membrane aq. SL org. SL Waterbeemd et al. measured the water/octanol and water/dibutylether partition, P, of a series of sulfonamides, and also the transport rate constants for the transfer of these sulfonamides from the aqueous to the organic phase, kup. From P and k^p they were also able to calculate the rate of the reverse process which is off-loading of the solutes from the organic phase back to water through the equation, P — kup / ky^ck (9) where kup and ky^ck are the overall rate constants, see Fig. 7.13. It was found "^^^422 that a tripartite curve was observed when log kup or log kyack was plotted against log P, as shown in Fig. 7.15 for log kup. The first section shows log kup increasing linearly with log P, the second section is an intermediate which is parabolic in nature. This leads into the third section where log kup is essentially constant, and hence independent of log P because diffusion through the aqueous diffusion layer becomes rate determining for the interfacial transport. Waterbeemd et al. interpreted the plot of log kup against log P as follows. In the first section, the factors that influence log kup are similar to those that influence log P, but as kup becomes larger and larger it reaches a diffusion controlled rate limit in section three. This rate limiting step is due 303 to diffusion through the stagnant layer, SL, between the two phases. A similar rate limiting step takes place as kback becomes larger and larger corresponding to small values of log P. Leahy et al. have confirmed these findings through a study of a diverse range of solutes in the water/octanol, chloroform, isooctane and propylene glycol dipelargonate systems, and were able to construct plots that corresponded to sections A and B, Fig.7.15. In the systems of Waterbeemd et and Leahy et al.,"^^^ no membrane is involved at all, so that there is no question of any ‘membrane control’, cf ref.57, and it is reasonable to ascribe the diffusion limit to k„p, to the effect of the aqueous SL. Fig. 7.15 The scheme of Waterbeemd et al. for partition and uptake from water to solvents, showing how log kup varies with log P. log kup logP More recently, Lennemas has suggested that the aqueous SL contributes little to overall intestinal permeability, and that intestinal absorption is ‘membrane controlled’. Lennemas also showed that in vitro measurements of permeability across segments of human intestine were well related to the in vivo estimated time to maximal absorption (though rather poorly to the % absorption) for five drugs. In contrast, Larhed et al. suggest that the mucus layer that covers the surface of the gastrointestinal tract can act as a barrier to drug absorption. Interestingly, Larhed et al."^^"^ showed that the effect of charge on diffusion through the mucus was not very large. 304 A rather different type of system in which the organic phase is a Cig Empore™ extraction disk consisting of 90 % w/w octadecyl-silica sorbent embedded in 10 % w/w PTFE microfibrils has been examined by Abraham and Green. Values of kup from water were obtained for a series of diverse solutes. There is a very small dependence on solute structure, see eqn 10, log kup = -5.34 + 0.08 Rz + 0.20 - 0.08 Zctz" - 0.28 SPz" + 0.33 Vx (10) n = 21, r^ = 0.95, SD =0.08, F = 30 This small dependence might be because passage through the SL is not quite diffusion controlled, or that there is a contribution from the actual transfer across the interfacial boundary. Either of these effects also explain the positive v-coefficient in eqn 10. However, even for strict diffusion, there may still be a small dependence on solute properties. Chen and Chan have conducted a study on limiting diffusion of 27 aromatic compounds in ethanol, with limiting diffusion coefficients, D, in 10^ m^ s '\ eqn 11 is derived, 1/D = 0.197 + 0 .0 2 2 R2 - + 0.853 Zaz" + 0.105 zPz" + 0.468 Vx (11) n = 27, / = 0.989, SD = 0.027, F = 385 However, in order to compare diffusion coefficients with the absorption function, log k, log D values must be correlated. log D = 0.412 + O.OO 8 R2 - 0.037%" - 0.388 Saz" - 0.051 sPz" - 0.226 Vx ( 12) n = 27, ? = 0.911, SD = 0.018, F =186 Eqns 11 and 12 show the well known effect that for diffusion in homogeneous solution, D is inversely proportional to solute size; this leads to the negative v- coefficient in eqn 12, although even for limiting diffusion there are effects due to hydrogen-bonding. 305 However, there has been no study in which human in vivo data, especially the % absorption, have been generally related (ie for a large number of drugs of varied structure) to some model of intestinal absorption. In particular, there is a marked lack of comment on the role of strong acids and bases in in vivo % absorption. Various workers 63,65,66,241 have related % absorption to drug physicochemical properties do not refer to the problem at all. Yet Rowland and Tozer ^ use gastrointestinal absorption as an example to show that strong acids and strong bases that are predominantly ionized in the intestinal tract should be transported very slowly, and Sugawa et al. correlate perfusion rates in rat jejunum with various physicochemical descriptors, taking neutral, cationic and anionic compounds separately. Therefore, the aim of this chapter is to construct a model for passive intestinal absorption, and to relate this to the large data base of compounds presented in Chapters 7.2 and 7.3 of in vivo % absorption. In Chapter 7.2.2, the Abraham General Solvation equation was obtained for human % drug absorption leading to eqn 13. %Abs = 92 + 2.94 R 2 + 4.10 712"-21.7 I a 2” - 21.1 ZP2” + 10.6 Vx (13) n = 169, = 0.74, SD = 14%, F = 93 However, in order to deal with questions about mechanism, the % absorption has to be transferred into a free-energy related quantity, of which log k, where k is the overall rate constant for absorption, is the most appropriate. In intestinal absorption, the drug concentration on the receiving site (portal vein) is usually negligible in relation to that on the driving site (intestinal tract), where a quantity of the substance is dissolved in the small intestinal fluid. In addition, only a mechanism that does not involve rate- determining dissolution of a solid is considered, see Fig. 7.13. The absorption rate- determining step is therefore passive diffusion across the membrane system. With these reasonable simplifications, the rate of diffusion follows first-order kinetics, with an overall rate constant, k, given by eqn 14 where Cw is the drug concentration in the intestinal tract at a time t. Integration of eqn 14 leads to eqn 15, where Cw° is the drug concentration at t = 0. If Cw° is taken as 100% (that is the initial concentration) and Cw as the observed % absorption at some given time, t, then eqn 16 and eqn 17 follow. 306 dCw/dt = -k Cw (14) ln(Cw°-C„)/Cw“ = -kt (15) In [l-(%/100)] = -kt (16) log k = log{ln [100/(100-%)]} - log t (17) Eqn 16 and eqn 17 collapse when the % absorption is 0 or 100, and so can only be applied to drugs that are absorbed neither too rapidly nor too slowly. Under these conditions, it may be assumed that the time to maximum absorption, and hence the log t term in eqn 17, is constant from one drug to another. Hence, the % absorption in eqn 17 is the absorption at a given time, t. This does not conflict with results summarized by Lennemas,^^ who showed that for drugs with 100 % absorption, the time to maximum absorption was shorter than for drugs with smaller % absorption. With the above assumption, eqn 18 follows; the constant log t term is now subsumed into the equation c-constant. log{ln[100/(100-%)]} = c 4- rR ; -t- ^ 2" + a Zaz" + K k " + vVx (18) From the data set of human intestinal absorption, 128 compounds have a % absorption other than 0 or 100, and application of eqn 18 leads to eqn 19, log{ln[100/(100-%Abs.)]} = 0.568 - 0.036 R 2 + 0.141 - 0.414 - 0.507 ZPz" + 0.232 Vx (19) n = 128, = 0.766, SD =0.314, F = 80 The same process may be applied to the rat intestinal absorption. Model 8 (Chapter 7.3.3) to obtain eqn 20 log{ln[100/(100-%Abs.)]} = 0.652 - 0.224 Rz + 0.230 Ttz" - 0.524 2% " - 0.313 + 0.128 Vx (20) n = 6 6 , r^ = 0.79, SD = 0.28, F = 45 It is noteworthy that the constants in eqn 19 and 20 bear no relation at all to those in numerous equations that have previously been constructed for water/solvent partition 307 equilibria, as log P, or for rates of transfer from water to another phase as log k. This can be seen from Table 7.12, that includes some representative water/solvent partitions 427,428 as well as partition from water to plant matrix,to a Cig Empore disk, and to blood; the latter is obtained from gas/water and gas/blood partitions. In addition, there are rate processes in Table 7.12 for skin permeation and plant cell permeation 432 both from water. Table 7.12 System coefficients for water/phase transfers Phase SP r s a b V Human intestinal abs, eqn 19 logk -0.04 0.14 -0.41 -0.51 0.23 Rat intestinal absorption, eqn 20 logk -0.22 0.23 -0.52 -0.31 0.13 Water-octanol log K(P) 0.56 -1.05 0.03 -3.46 3.81 Water-cyclohexane log K(P) 0.78 -1.68 -3.74 -4.93 4.58 Water-chloroform log K(P) 0.16 -0.39 -3.19 -3.44 4.19 Water-plant matrix log K 0.60 -0.41 -0.51 -4.10 3.91 Water-Ci8 disk log K(P) 0.35 -0.07 -0.63 -1.96 2.85 Water-brain log K 0.67 -1.81 0.00 -2.15 3.20 Water-human skin log kup 0.44 -0.49 -1.48 -3.44 1.94 Water-plant cell log kup 0.00 -0.87 -3.14 -1.66 0.73 Although Waterbeemd et gave numerical values of kup for all three sections of Fig. 7.15, it is not possible to carry out an analysis through the Abraham General Solvation equation because there are not enough solutes in the separate sections A and C. However, the coefficients in eqn 19 (and also 20) can be compared with those for which diffusion seems to be the major process, as shown in Table 7.13. Table 7.13 System coefficients for diffusion processes Phase SP r s a b V Human intestinal abs, eqn 19 logk -0.04 0.14 -0.41 -0.51 0.23 Rat intestinal absorption, eqn 20 logk -0.22 0.23 -0.52 -0.31 0.13 Ci8 disk, eqn 10 log kup 0.08 0.20 -0.08 -0.28 0.33 Ethanol diff. eqn 12 log D ^ 0.00 0.00 -0.39 -0.05 -0.02 D is the limiting diffusion coefficient in bulk ethanol solvent 308 There is a certain degree of similarity between the intestinal absorption equations and eqns 10 and 12. Since neither eqn 10 nor eqn 12 refer to diffusion through a membrane, it is not necessary to postulate that such a process is part of the rate- limiting step. In this context, the work of Johnson et al. and of Mitragotri et al. is extremely important. The former workers examined permeation through the skin stratum comeum and concluded that lateral diffusion in the membrane bilayer was the primary transport step. On the other hand, Mitragotri et al. have argued that in general it is the actual interfacial transport at the membrane boundary that is the rate determining step. The two comments are not incompatible. In skin permeation a solute has to diffuse across about 100 bilayers; the rate determining step then becomes lateral diffusion along a bilayer, as the solute seeks to find a hole through which to pass to the next bilayer. The results presented in this work appear to be compatible with this scenario in that the equations for intestinal absorption are close to those for passive diffusion in systems that include no membrane at all, see Table 7.13, and is quite different to that for skin permeation. As such, the available evidence points to diffusion in the stagnant mucose layer together with the actual interfacial (mucus/membrane) transfer as the rate determining step in human intestinal absorption. 7.5.1 Bronsted acids and bases Eqn 13 includes Bronsted acids and bases, with no correction for ionization at all. However, unlike many previous workers, who have related %Abs to drug physico chemical properties, the influence of strong Bronsted acids and bases was considered in Chapter 7.2.2. An indicator variable for acids with pKa < 4.5 and bases with pKa >8.0 was incorporated, but the resulting equation (Chapter 7.2.2 eqn 8), is very close to eqn 13, Since the indicator variable term contributes to the % absorption only very slightly, this original analysis suggested that the effect of these ionizable compounds on the observed % absorption is very small - in the sense that strong Bronsted acids and bases are absorbed only 3% less than calculated from properties of their neutral forms. % Abs = 94 + 2.90 Rj + 2.71 712” -20.7 -20.9SPa” +I1.2Vx-3.I4I n = 169, = 0.74, SD = 14%, F = 78 (Chapter 7.2.2, eqn 8) 309 This analysis will now be extended in order to explain the observation of the small effect of ionized species on absorption. Now since the pH of the upper small intestine is around 6 -7, both strong Bronsted acids and bases will be mainly in an ionized form. But how are such ionized species absorbed ? One possibility is that the ionic <=> neutral equilibrium, eqn 21, provides a pathway for the ionic species to be absorbed in an indirect way, through the mechanism shown in eqn 22, where refers to the second (rate-determining) step. I <=> N > Absorption (22) The rate of absorption is given by r = k'^xN (23) and if eqn 21 and eqn 23 are combined, eqn 24 is obtained, where H denotes [H^], and T denotes the formal solute concentration, [I] + [N], r= k'^xT/(l+Ka/H) (24) The observed rate constant is then given by, kObs _ ^ Ka/H) (25) If it is supposed that H = 10 then the expression for can be calculated for a series of acids with various values of Ka, as shown in Table 7.14. Acids and bases that are ionized will appear to permeate by orders of magnitude less than the neutral species. This would greatly impact on the % absorption, and it can concluded that the mechanism shown in eqn 22 does not operate in any major way, as far as can be deduced from the data on % absorption in humans. 310 Table 7.14 Analysis of absorption of Bronsted acids on eqn 25, with H = 10^^ Ka Expression for 10'^ 0.99 k^ 10'^ 0.76 k^ 10'^ 0.03 k^ 10'^ 0.0003 k^ However, if it is supposed that both the ionic species, I, and the neutral species, N, are directly absorbed, then the observed rate constant will be given by, xT = k'xl + k'^xN (26) Eqn 27 is derived from eqn 21 and 26, k°'’* xT = N[k'xKa/H + k^] (27) Now the total substrate, T, is given by, T = N + I = N + KaxN/H (28) and if eqn 27 and eqn 28 are combined, is found to depend on H and Ka k°'’'‘ = [(k'xKa + k'^xH]/(H + Ka) (29) If a constant pH is again considered so that H = 10^^, the expression for substrates of various Ka values can then be obtained, as shown in Table 7.15. 311 Table 7.15 Analysis of the absorption of Bronsted acids on the mechanism of eqn 29, with H = 10*® Ka Expression for k°*^ ------Expression for k°** when ------k'/k'^ = 0.5 k’/k ^ = 0.01 k‘ / k'^ = 0.001 10'® (10'®'®xk‘ + k'^) 1.00 k'^ 1.00 k'^ 1.00 k"^ 10'*® (0.5 k' + 0.5 0.75 k'^ 0.51 k^ 0.50 k'^ 10'® (0.97 k' + 0.031 k'^) 0.52 k'^ 0.04 k"^ 0.03 k'^ 10'® (k' + 10'®-®xk'^) 0.50 k*^ 0.01 k'^ 0.001 k'^ It is apparent that if the ratio k' / becomes very small (ie the ionic species is very poorly absorbed), then the observed rate constant will decrease enormously for species that are strong Bronsted acids or bases; this is not the case for human intestinal absorption. On the other hand, as k* / approaches unity so the effect of the ionic species on the observed rate constant becomes smaller and smaller. It can be calculated from eqn 17 that if k^ / k^ is 0.87, then strong Bronsted acids and bases will reduce the observed % absorption by 5%. The analysis using an indicator variable. Chapter 7.2.2 eqn 8, suggests a reduction of only 3%, so with an allowance for some experimental error, it can be concluded that in general k^ / k^ cannot be less than about 0.87 in human intestinal absorption. Morishita et al. have studied the intestinal absorption of a number of acidic sulfonamides in the rat through an in situ perfusion method that allowed the pH to be adjusted. Under these conditions, it was possible to determine rate constants for absorption of the neutral and the anionic forms. Values of k^ / k^ ranged from 1.0 to 0.46 so that in this, admittedly different system, ionic species are absorbed almost as readily as the neutral species. Interestingly, the effect of compound structure on k^ is quite small. Values of log k^ cover only 1.06 units, as compared to 3.15 log units for the variation in log Poet, so that this absorption may also be diffusion controlled. The above calculations on ionic absorption are thus compatible with the findings of Morishita et al.,"^^^ and also with the observation of Larhed et al."^^"^ that the effect of 312 charge on diffusion through the mucus is not very large. However, it is worth noting that although the analysis in this work, and the findings of Morishita et al., have suggested that ionic species can be absorbed directly, the actual mechanism of ionic absorption remains to be elucidated. If an anionic species is absorbed from the aqueous phase, then either a cationic counter-ion must also be absorbed, or some other anionic species must be transferred in the reverse direction. Furthermore, because m vivo membranes are not homogeneous, it is possible that ionized species can cross them via pores or water-filled channels. Quite recently, Yoshida and Topliss have constructed a QSAR for classification of human oral bioavailability. In contrast to the findings of this work, and those of Morishita et al. on intestinal absorption, Yoshida and Topliss find that Bronsted acids have a better bioavailability than neutral species, and that Bronsted bases have lower bioavailability than neutral species. This may be attributable to the mechanism of first-pass metabolism and suggests that metabolic enzymes may have differing affinity for ionized species; certainly there are many other factors involved in bioavailability than just intestinal absorption. 7.5.2 Characterization of the Absorption System It has been suggested that the LFER equation for logk for intestinal absorption, eqn 19, is rather similar to those for rat intestinal absorption, the rate of uptake from water onto a CIS disk, and for diffusion in ethanol. Eqn 10 and eqn 12. However, it is beneficial to use a more rigorous and quantitative method for the comparison of these LFERs. Ishihama and Asakawa have recently put forward a very elegant method for the comparison of LFERs, say SPI and SP2, as regards correlation (detailed in Chapter 5.5.2a, page 151). They define an angle, 0, between the vectors of the coefficients in SPI and SP2 such that if the dependent variable in SPI is well correlated with that in SP2 then 0 is near to zero. However, if the two dependent variables are not well correlated then 0 deviates from zero. Since 0 can be calculated from the coefficients in SPI and SP2, this method is extremely convenient. Table 7.16 contains values of 0 obtained by the method of Ishihama and Asakawa taking logk for human intestinal absorption as the standard system. 313 The Ishihama and Asakawa 0-method of analysis indicates that none of the processes 3-12 listed in Table 7.16 has 0 anywhere near to zero, and so as regards correlation, none of them will be very good models for intestinal absorption. This has considerable implications as regards predictive algorithms based on water/octanol partition, as log Poet- If 0 is far from zero for log k and log Poet, then the latter will be a very poor linear correlative predictor of log k. This suggestion is verified for the 128 human intestinal absorption data set by a poor linear correlation of log k with ClogP (calculated logPoct from ClogP for Windows software. Biobyte version 2.0.0b, Claremont, CA), as shown in eqn 30. An even poorer relationship exists between log k and the 63 experimentally measured log Pœt values (obtained from Medchem 2000, Biobyte Corp., in co-operation with Daylight) as shown in eqn 31. logk = -0.060 + 0.197 ClogP (30) n = 128, = 0.703, SD = 0.348, F = 301 logk = -0.055 + 0.177 log Poet (31) n = 63, ? = 0.407, SD = 0.346, F = 44 It has also been found that 0 is far from zero for % Abs and log Poet (0=50), and so the use of log Poet in any linear correlation of % Abs is not likely to be of any generality. Palm et al. showed that there was almost no correlation whatsoever between % Abs and ClogP for 20 drugs. This has further been exemplified by the poor (sigmoidal) relationship obtained between % Abs and ClogP for the 169 drugs from which eqn 19 was derived, and also % Abs and measured log Poet values for 84 drugs. However, although the 0-parameter is useful as a linear correlative comparison, it is not very useful as regards any chemical comparison of LFERs. For example, if the coefficients in SP2 were all 1/10 of those in SPI, the two systems would be regarded chemically as quite different, and yet 0 would still be zero. A very simple way of comparing coefficients in two LFERs is to sum theabsolute differences between the coefficients, C|, in the two LFERs as shown in eqn32. The factor 1/5 is introduced so as to give the average difference in the five coefficients r to v. Of course, such a simple method will only work if the dependent variables in SPI and SP2 refer to the 314 same type of quantity (in this work either the logarithm of a rate constant or the logarithm of an equilibrium constant), and if the various descriptors cover about the same range of values. Ô = (EI Cl - C2I )/5 c = r to V (32) The obtained values of Ô are also given in Table 7.16 Table 7.16 Characterization of Systems Phase No SP 0 Ô Human Intestinal abs, eqn 19 1 logk 0 0 Rat intestinal absorption, eqn 20 2 logk 29 0.14 Water-octanol 3 log K(P) 50 1.75 Water-cyclohexane 4 log K(P) 31 2.94 Water-chloroform 5 log K(P) 29 2.07 Water-plant matrix 6 logK 41 1.71 Water-Gig disk 7 log K(P) 42 0.97 Water-brain 8 logK 60 1.53 Water- human skin 9 log kup 26 1.36 Water-plant cell 10 log kup 34 1.09 Water-Gig disk, eqn 10 11 log kup 38 0.16 Diffusion in ethanol, eqn 12 12 logD 65 0.21 If Ô is considered, the three examples of water/solvent partition. Nos 3-5, are all chemically far away from intestinal absorption, but the rate of uptake onto the Cig disk (No 11) and diffusion in ethanol (No 12) are both chemically close to intestinal absorption, in that the chemical factors that influence the three processes will be quantitatively very similar. The use of principle components analysis (PGA) is another method that can be used to compare LFERs. PGA has been carried out on the coefficients of the systems shown in Table 7.16, and it is found that the first two PGs account for 89% of the information. 315 The 2-dimensional space defined by the two largest eigenvectors of the correlation matrix indicates the presence of a cluster of points, see Fig. 7.16. This scores plot of PC2 vs PCI shows that processes Nos 11 and 12 are very close to intestinal absorption, whilst the three water/solvent partitions. Nos 3-5 are far away. The relationship between the twelve systems in Table 7.16 can also be shown graphically through the technique of non-linear mapping, NLM."^^^ In Fig. 7.17 is a two-dimensional non-linear map in which the similarity between the LFERs is given by the distance between the corresponding points. It is again apparent that processes Nos 1,2,11 and 12 are clustered together whilst the three water/solvent partitions. Nos 3-5, are far away. Fig. 7.16. A plot of PC2 vs PCI for the coefficients in Tables 7.12 and 7.13. Numbers refer to the systems in Table 7.16 2 • 10 1.5 # 5 1 • 4 0.5 • 2 Ü 0 e 9 Q. -0.5 # 7 • 6 1 • 3 1.5 e 8 2 2 1 0 1 2 33 4 PCI 316 Fig. 7.17 A plot of results of non-linear mapping for the coefficients in Tables 7.12 and 7.13. Numbers refer to the systems in Table 7.16 120 - • 10 300 7.5.3 Conclusion The simple eqn 32, and the more complicated analyses through PCA, and non-linear mapping, all lead to the same conclusion that the factors that influence human intestinal absorption are quantitatively not the same as those that influence water/solvent partitions and a number of other processes in which solutes are transferred from water to other phases. However, the factors are similar to those that influence two particular processes, viz. (1) the rate of diffusion of solutes in ethanol solution, and (2) the rate of uptake of solutes onto a Cig disk. Analysis using these three methods produces almost the same rank order of difference from intestinal absorption, see Table 7.17. The similarity between results from PCA and NLM is due in part to the fact that the first two PCs account for 89% of the information. If the information content of the remaining PCs becomes significant, then it cannot be expected that PCA and NLM will yield similar results. The three above methods appear to refer to ‘chemical’ differences, i.e. quantitative differences in the chemical factors that influence transport systems. Conversely, the rank order of difference from log k for intestinal absorption as obtained from the 0-parameter is not at all similar to those from the other three methods, see Table 7.17. As mentioned previously, this is 317 because the 0-method refers to linear correlative differences, whilst the other rriethods used refer to chemical differences. Table 7.17 Rank order of difference from log k for human intestinal absorption Rank order Ô PCANL 0 1 1 1 1 1 2 2 12 2 3 3 11 11 11 10 4 12 2 12 5 5 7 7 10 3 6 10 10 7 8 7 9 9 9 9 8 8 5 8 11 9 6 6 3 2 10 3 3 5 6 11 5 8 6 7 12 4 4 4 8 318 7,6 Partitioning of Compounds onto a C18 Disk In Chapter 7.5, it was shown that the log k equation obtained for prediction of human intestinal absorption was chemically most similar to the equation for the uptake rate constant (kup) of a solute as it partitions from the aqueous phase onto the bulk organic phase (Ci8 disk). The aim of this subsequent work is to experimentally measure the rate of uptake for a few drug compounds onto a Cig disk and to analyse the results as an extension of work that has previously been carried out. In addition, there is also the opportunity to improve the equations for the uptake rate constant, equilibrium constant (Keq) and also offloading rate constant (koff) by increasing the number and diversity of compounds on which they are based. 7.6.1 Experimental Acetaminophen, acetylsalicylic acid, caffeine, furosemide, and theophylline were purchased from Aldrich; all compounds have >98% purity. Nadolol and phénobarbital were purchased from Sigma and purity was not specified. Stock solutions of each of the compounds were prepared in methanol, except for caffeine and theophylline which were sufficiently soluble in water. The pesticide diuron was used as a test solute to ensure that the results were consistent with the previous rate and equilibrium constants obtained. The for each compound was measured using a Shimadzu UV 2041 spectrophotometer and hyper UV software and the respective molar absorption extinction coefficients were determined from absorbance versus calibration curves. Sample solutions were prepared by spiking the relevant volume of distilled water with stock solution to obtain a concentration of -IxlO ’"^ mol L '\ Uptake experiments were carried out in an aqueous environment where the drug compound is in the un-ionized form. Extraction apparatus consisted of a Uvikon 810 dual-beam UV spectrometer, Gilson Model 303 HPLC PUMP used at a flow rate of 2ml m in'\ water-bath, stainless steel cage (internal diameter 48mm, breadth Smm)"^"^* and Cig 3M Empore^^ disks (47mm diameter, 90% Cig, 10% PTFE fibres)"^"^' which were purchased form Phenomenex. For measurements in neutral conditions, the collection of data was carried out continuously on-line. The whole system was flushed through with acetone (I5mins) 319 and then pure distilled water for at least one hour to remove any contaminants from previous experiments and to zero the spectrophotometer. For diuron, caffeine, theophylline, acetaminophen and phénobarbital, sample solutions of -IxlO'"^ mol L'^ were then prepared by spiking 1.95L of distilled water with the stock solution in 2L glass bottles. Each solution was continuously stirred with a PTFE magnetic stirrer at 200 rpm (revolutions per minute) and allowed to equilibrate at 23°C for at least 16 hours to obtain a steady UV reading and to ensure that the analytes were not being absorbed by the apparatus. The Cig disk was conditioned by being soaked in methanol for 1 hour prior to use. The initial absorbance at the Àmax of the analyte under investigation was recorded, after which the preconditioned disk was placed in the steel cage which was then immediately submerged in the sample solution. The absorbance of the sample solution was recorded at regular intervals of ~30mins until there was no change in absorbance and hence equilibrium had been reached. In cases where extreme acidic (pH2) or basic (pH 12) conditions were used, measurements were taken using an offline procedure in which the sample solution was not circulated through the spectrophotometer; thus avoiding damage to the pump and stainless steel piping. These measurements were carried out using 0.4L water due to practical limitations associated with using buffer. To determine when equilibrium had been reached for these compounds, aliquots of the aqueous solutions were analysed by UV at regular time intervals until the change in absorbance was zero. The aliquots were returned to the solution after each reading to ensure that the volume of liquid in the system remained constant. Since the surface area of the disk remains unchanged, these uptake rate constants may be compared to those obtained by the on-line method with the proviso that the difference in the magnitude of the phase ratio is accounted for in the following calculations. 7.6.2 Calculations The concentration of the analyte in solution is monitored from its absorbance which declines as the analyte is extracted from water onto the Cig disk; the Beer-Lambert law is used to calculate the degree of absorption onto the disk. A = G Ic (33) 320 where A is absorbance (no units), e is the molar absorption coefficient (L mol'^ cm'*), 1 is the path length of the sample which in this case is 1 cm, and c is the concentration of the compound in solution (mol L'^). Both the equilibrium constant, Keq and the rate of uptake, kup of analyte onto the Cig are calculated from a single dataset. The equilibrium constant, Keq is simply calculated from'^'^^ Keq = CD/Cw=/(C“w-Cw)/Cw (34) where Cd is the concentration of solute in the disk at equilibrium, C w is the initial concentration of solute in the water, and Cw is the final concentration of solute in the water at equilibrium; the concentrations are all expressed in mol L '\ / is the phase ratio (^ V w/Vd , where Vw and Vd are the volume of water and Cig in the disk in litres) The rate of uptake, kup is then obtained by plotting absorbance versus time for the experimental data and then curve-fitting the data to a kinetic equation through the graphical package Table Curve 2D, Jandel Scientific, CA, USA, The equation takes the form of A, = (6 M/ZkeqVo) + [ Ao - (e M/ZeqVD)]exp‘^“'’^‘ (35) where Aq is the initial absorbance of solute in water, At is the absorbance after time t, G is the molar absorption coefficient, M is the total number of moles in the system, kup is the uptake rate constant ( s '\ t is the exposure time of the experiment and Z = 1+Vw/Keq.Vo, where Vd and Vw are the volumes of the Cig in the disk and water respectively; Vd has previously been calculated to be and Keq is the equilibrium constant determined from eqn 34. 7.6.3 Results An example of the curve obtained from these uptake experiments is shown in Fig. 7.18 which shows the change in absorbance for an aqueous solution of phénobarbital at 23°C, [A]w° = 2x10 ^^ mol L '\ The curve clearly shows the decrease in absorption 321 as the concentration of phénobarbital in the water decreases due to partitioning onto the Ci8 disk. The continuous monitoring of this change in absorbance allows the identification of the equilibrium position where absorbance will remain constant. Subsequently the rate of uptake onto the disk may be calculated from c = -kupZ. The results of this analysis for each of the drug compounds studied is shown in Table 7.18. Fig. 7.18 Plot showing change in absorbance for phénobarbital Eqn of form y = (a-b)exp(-cx)+b where in this case, a = 1.408, b = 1.229, c = 0.0001 (r^ = 0.999, Fitting error = 0.00135, F = 14916) 1.45 1.40 - cs I S38 1.30 - < 1.25 - ooo o 1.20 0 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000 Time (secs) Within the available time-scale it was only possible to obtain experimental values for seven extra compounds and so the results presented in this chapter are certainly only preliminary in nature. However, they will briefly be discussed within the context of previous work conducted by C.E.Green'^'^\ The initial selection of the drug compounds was driven by interest in the relationship between log kup and log {In [100/(100-% Abs.)]} in the human gastrointestinal tract, i.e. do compounds with poor absorption partition more slowly from an aqueous phase to an organic phase? Furthermore, it was of interest to study more hydrophilic compounds than previously considered; logPoct<1.57. The experimental requirements of sufficient water solubility and UV activity restricts the use of some compounds which otherwise would have been of interest. 322 Table 7.18 Descriptors and experimental data for diuron and drug compounds Compound R2 Zaz" z k " Vx pH V\v Àmax loge log kup log keq log %Abs (L) Poet Diuron^ 1.28 1.60 0.57 0.70 1.5992 7 1.95 249.0 4.34 -4.73 4.05 2.68 - Amphetamine*’ 0.94 0.77 0.18 0.63 1.2389 12 0.40 -4.92 2.55 1.76 90 Caffeine‘S 1.50 1.63 0.00 1.29 1.3632 7 1.95 272.0 3.96 -4.71 3.09 -0.07 100 Theophylline'S 1.50 1.60 0.54 1.34 1.2223 7 1.95 271.0 4.02 -4.88 2.63 -0.02 100 Phenobarbital's 1.63 1.80 0.73 1.15 1.6999 7 1.95 213.0 4.09 -4.85 3.29 1.47 - Acetaminophen's 1.27 1.81 1.02 0.85 1.1724 7 1.95 242.8 3.97 -4.73 2.69 0.51 80 Acetylsalicylic acid's 0.93 1.35 0.59 0.80 1.2879 2 0.40 230.9 3.68 -4.77 2.93 1.19 84 Furosemide's 2.05 2.55 1.36 1.47 2.1032 2 0.40 342.0 3.69 -4.64 3.59 2.03 61 Nadolol'S 1.61 1.63 0.70 1.88 2.4923 12 0.40 221.0 182 -4.71 2.91 0.71 57 “ Single experiment. Used for comparative purposes to ensure that result was consistent with previous experimental results for diuron. Work by C. Green'*'*®; log Keq = 4.04, logkup = -4.69 ** ** Experimental values measured by C.Green, ref 441 Experimental values obtained in this work, log Keq and log k„p are taken as an average of at least duplicate measurements. 323 The results in Table 7.18 indicate that log kup for these compounds is generally independent of logKgq and therefore diffusion-controlled through the static boundary layer at the water-disk interface. Since the slope of the log kup/log Poet relationship (Fig. 7.19) is negligible, it can be concluded that log kup is not influenced by log Poet, and these values correspond to section C of the Waterbeemd plot (shown in Fig. 7.15). However, the log kup values appear to become constant at a water-octanol partition coefficient (log Poet) below 1.5. This coincides with the point at which log Keq also becomes more constant and may indicate that kup for these compounds is becoming slightly more dependent on log Keq, corresponding to the end of section B on the Waterbeemd plot. Nevertheless, the relationship may still be regarded as linear. Fig. 7.19 Plot of log kup vs log Poet for 28 compounds (compounds studied by Green'*'*^ represented as open circles) -3.0 3.5 y = 0.0985X - 4.9579 = 0.6278 3 4.0 O) Odrugs 4.5 5.0 -5.5 2 0 24 6 8 log Poet Fig. 7.20 Plot showing relationship between log K^q and logPoct for 41 compounds 8.0 y = 0.6485x + 2.0728 6.0 - = 0.856 O’ o ❖ drugs 0.0 1 1 3 5 7 9 log Poet 324 Since there is at present data available for only eight drug compounds, it is not possible to make a reliable comment on the relationship between the experimental log kup and log{ln[100/(100-%Abs.)]}. Further experimental work is required and selection of compounds with yet greater hydrophilicity is required so that section B of the Waterbeemd plot may be investigated. However, the additional values for log kup and log Keq can be used to update the respective equations (Table 7.19). Indirect calculation of the offloading rates, log koff through Keq = kup / koff will also yield an improved log koff equation so that the minimal criteria of at least five data points per regressor variable is surpassed. Table 7.19 Equations for log Keq, log kup and log koff C r 5 a h V n rZ SD F Updated equations log Keq 1.34 0.45 0.08 -0.88 -2.05 2.41 41 0.91 0.37 70 log kup -5.26 0.07 0.22 -0.15 -0.29 0.28 28 0.87 0.09 28 log koff -6.47 -0.36 -0.27 0.87 1.68 -1.82 28 0.93 0.32 56 Old equations log Keq 1.14 0.26 -0.63 -0.67 -2.25 3.36 34 0.95 0.28 110 log kup -5.34 0.09 0.20 -0.08 -0.28 0.33 21 0.91 0.08 30 log koff -6.23 -0.27 0.27 0.55 1.68 -2.52 21 0.97 0.22 100 The results are generally consistent with the old equations. The only noticeable difference by first inspection is the difference in magnitude of the 5-coefficient in the log Keq equation. However, the low t-statistic for the 5-coefficient (t = 0.25, p=0.81) clearly indicates that it is not significant, and this is in agreement with the result previously obtained by C.E.Green'^V Overall the statistics are marginally worse for the extended data set, although this must be considered within the context of the enhanced range of physicochemical parameters which has been provided by the addition of these more polar drug compounds. 325 The equations may be interpreted as follows: log Keq : The dispersion and dipolarity / polarizability interactions between solute and solvent do not significantly influence retention. The hydrogen-bond basicity of the disk is slightly lower than that of water, and the larger and negative ^-coefficient indicates that hydrogen-bond acidity of water is greater than that of the Cis disk. Thus, retention of hydrogen-bond basic compounds is favoured in water. Conversely, the large and positive v coefficient indicates that cavity formation is favoured in the disk and will aid retention of solutes from water. log kup : The values are all comparatively close to zero so that there is very little competition between the disk and water for the retention of hydrogen-bonding compounds or indeed large molecules. The most significant contribution to the uptake rate is not from the five physicochemical properties, but from the equation intercept. This translates to an additional effect which is most prominent in governing the rate of uptake, and this is attributed to diffusion. This provides an explanation for the consistency in log kup values irrespective of the structural and physicochemical diversity of the drug compounds. log koff : Again, the main contribution to the rate of off-loading, log koff, is the c coefficient. However, the solute physicochemical properties play a larger role. Most notably, the hydrogen-bond acidity of the water favours offloading of hydrogen-bond bases from the disk, whilst the largely negative v-coefficient disfavours off-loading because cavities are formed more easily in the disk than the water. Consequently, the plot of log Keq vs log kup/log koff (Fig. 7.21) verifies the opposite influence of physicochemical parameters on log kup and log koff. Theoretically, the coefficients for the log kup equation minus those for the log koff equation should equal the coefficients in the log Keq equation. If this subtraction is carried out, the ‘calculated log Keq equation’ from this procedure is given by: log Keqfca/cj = 1,21 + O.43R2 + 0.49712” -1.02X02” - I.97XP2” +2 .10VX These calculated coefficients agree reasonably well with the coefficients obtained for log Keq in Table 7.19. There are minor differences, but these may be ascribed to 326 differing numbers of compounds used in the log Keq and log kup/log koff regressions, and also errors in the equation coefficients which are largest for the r- and 5 - coefficients. Fig. 7.21 Plot of log kup/log koff vs log K^q for 28 compounds (closed symbols represent drug compounds) 4 5 6 1 7 Q. O log kup 3 8 JC O log koff D) O 9 CO 10 11 12 0 1 2 3 4 5 6 7 8 log Keq 327 7.7 References 1. Smith, H.J. (Ed.) Introduction to the principles of drug design and action. Third edition, Harwood academic publishers 1988, Amsterdam BV. 2. Physiological pharmaceutics, biological barriers to drug absorption. Wilson, C.G.(ed); Washington, N. Ellis Harwood Limited 1989, England. 3. Lennernais, H. Does fluid-flow across the intestinal mucosa affect quantitative oral drug absorption - is it time for a re-evaluation. Pharm. Res. 1995,12,1573-1582. 4. Kramer, S.D. Absorption prediction from physicochemical parameters. Fharm. Sci. Tech. Today 1999,2, 36-42. 5. Martin, Y.C.; Kutter, E.; Austel, V. (Eds.) Modern Drug Research : Paths to better and safer drugs. Dekker, New York , 1989. 6. Rowland, M.; Tozer, T.N.(Eds) Clinical pharmacokinetics : concepts and applications. Second edition. Lea and Febiger, Philidelphia, 1989. 7. Artursson, P.; Ungell, A.L.; Lofroth, I.E. selective paracellular permeability in 2 models of intestinal absorption - cultured monolayers of human intestinal eptithelial cells and rat intestinal segments. Pharm. Res. 1993, 10, 1123-1129. 8. Artursson, P.; Palm, K.; Luthman, K. Caco-2 monolayers in experimental and theoretical predictions of drug transport. Adv. Drug Deliv. Rev. 1996, 22, 67-84 9. Hunter, J.; Hirst, B.H.; Simmons, N.L. Drug absorption limited by p-glycoprotein mediated secretory drug transport in human intestinal epithelial caco-2 cell layers. Pharm Res. 1993, 10, 743-749 10. Schinkel, A.H.; Wagenaar, E.; Vandeemter, L.; Mol, C.A.M.M.; Borst, P. Absence of the MDRIA p-glycoprotein in mice affects tissue distribution and pharmacokinetics of dexamethasone, digoxin and cyclosporin-A. J.Clin. Invest. 1995, 96, 1698-1705. 11. Charman, W.N.; Porter, C.J.H.; Mithani, S.; Dressman, J.B. Physicochemical and physiological mechanisms for the effects of food on drug absorption: The role of lipids and pH. J.Pharm.Sci. 1997, 86,269-282 12. Fleisher, D.; Li, C.; Zhou, Y.; Pao, L.H.; Karim, A. Drug, meal and formulation interactions influencing drug absorption after oral administration - clinical implications. Clin. Pharmacokinet. 1999, 36,233-254 13. Gubbins, P.O.; Bertch, K.E. Drug absorption in gastrointestinal disease and surgery - clinical pharmacokinetic and therapeutic implications. Clin. Pharmacokinet. 1991,21,431-447. 14. Shefter, E.; Higuchi, T. Dissolution behaviour of crystalline solvated and non-solvated forms of some pharmaceuticals. J.Pharm. Sci. 1963, 52, 781-791 15. Kristi, A.; Srcic, S.; Vrecer, F.; Sustar, B.; Vojnovic, D. Polymorphism and pseudopolymorphism: influencing the dissolution properties of the guanine derivative acyclovir. Int. J. Pharm. 1996, 139,231-235. 16. Grant, D.G.W.; Higuchi, T. Solubility behaviour of organic compounds; John Wiley & Sons, New York, 1990; p386 328 17. Robinson, D.H.; Narducci, W.A.; Veda, C.T. Drug delivery and administration. In Dipiro et al. (Eds). Pharmacotherapy ; A pathophysiologic approach. Elsevier Science Publishing. 18. Schultz, N.J. Principles of drug therapy. J.Enterostomal Ther. 1987, 14, 212-215. 19. Shargel, L.; Yu, A (Eds). Applied biopharmaceutics and pharmacokinetics; Fourth edition, Appleton & Lange, Stamford, 1999. 20. Aguiar, A.J.; Krc, J.; Kinkel, A.W.; Samyn, J.C. Effect of polymorphism on the absorption of chloramphenicol from chloramphenicol palmitate. J.Pharm.Sci. 1967, 56, 847-853 21. Shore, P.A.; Brodie, B.B.; Hogben, C.A.M. The Gastric secretion of drugs - A pH partition hypothesis. J. Pharmacol. Exp. Ther. 1957, 119, 361-369. 22. Schanker, L.S.; Shore, P.A.; Brodie, B.B.; Hogben, C.A.M. Absorption of drugs from the stomach. I. The rat. J.Pharmacol. Exp. Ther. 1957, 20, 528-539. 23. Nimmo, W.S. Gastric emptying and drug absorption. In Prescott & Nimmo (Eds) Drug absorption: Proceedings of the Edinburgh International Conference, ADIS Press; Auckland, 1979. 24. Welling, P.G.; Huang, H.; Krock, P.A.; Craig, W.A.; Madsen, P.O. Bioavailability of ampicillin and amoxicillin in fasted and non-fasted subjects. J. Pharm. Sci. 1977, 66, 549-552. 25. Willis, J.V.; Jack, D.B.; Kendall, M.J.; John, V.A. The influence of food on the absorption of diclofenac as determined by the urinary excretion of the unchanged drug and its major metabolites during chronic administration. Clin. Pharmacol. 1981, 19, 39-44. 26. Willis, J.V.; Kendall, M.J.; Jack, D.B. The influence of food on the absorption of diclofenac after single and multiple oral doses. Eur. J. Clin. Pharm. 1981, 19, 33-37. 27. Charman W.N.; Porter C.J.H.; Mithani S.; Dressman, J.B. Physicochemical and physiological mechanisms for the effects of food on drug absorption. J. Pharm. Sci. 1997, 86, 269-282. 28. Doluisio, J.T.; Tan, G.H.; Billups, N.F.; Diamond, L. Drug absorption ILEffect of fasting on intestinal drug absorption. J.Pharm.Sci. 1969, 58, 1200-1202. 29. Melander, A. Influence of food and different nutrients on drug bioavailability. Wld, Rev. Nutr. Diet 1984, 43, 34-44. 30. Welling, P.G. Interactions affecting drug absorption. Clin. Pharmacokinet. 1984, 9, 404-434. 31. Welling, P.G. Effects of food on drug absorption. Pharmacol. Ther. 1989,43, 425-441. 32. Reppas, C.; Eleftheriou, G.; Macheras, P.; Symillides, M.; Dressman, J.B. Effect of elevated viscosity in the upper gastrointestinal tract on drug absorption in dogs. Eur. J. Pharm. Sci. 1998, 6, I3I-I39 33. Yamaguchi, T.; Ikeda, C.; Sekine, Y. Intestinal absorption of a ^-adrenergic blocking agent - nadolol.2. Mechanism of the inhibitory effect on the intestinal absorption of nadolol by sodium cholate in rats. Chem. Pharm. Bull. 1986, 34, 3836-3843. 34. Bates, T.R.; Gibaldi, M.; Kanig, J.L. Rate of dissolution of griseofulvin and hexoestrol in bile salt solution. Nature 1966, 210, 1331-1332. 35. Bakatselou, V.; Oppenheim, R.C.; Dressman, J.B. Solubilization and wetting effects of bile-salts on the dissolution of steroids. Pharm. Res. 1991, 8, 1461-1469. 36. Rosoff, M.; Serajuddin, A.T.M. Solubilization of diazepam in bile-salts and in sodium cholate- lethicin water phases. Int. J. Pharm. 1980, 6, 137-146. 329 37. Kassem, M.A.; Mattha, A.G.; El-Nimir, A.E.M.; Omar, S.M. Study of the influence of sodium taurocholate (STC) and sodium glycocholate (SGC) on the mass-transfer of certain drugs - digoxin. Int. J. Pharm. 1982, 12, 1-9. 38. Toothaker, R.D.; Welling, P.G. The effect of food on drug bioavailability. Ann.Rev.Pharmacol. Toxicol 1980, 20, 173-199. 39. Dawson, J.R.; Bridges, J.W. Intestinal microsomal drug metabolism - A comparison of rat and guinea-pig enzymes, and of rat crypt and villous tip cell enzymes. Biochem. Pharmacol 1981, 30, 2415-2420. 40. Benet, L.Z., Wu, C-Y., Hebert, M.F., Wacher, V.J. Intestinal drug metabolism and anti transport processes: A potential paradigm shift in oral drug delivery. J.Controlled Release 1996, 39, 139- 143. 41. Wacher, V.J., Silverman, J.A., Zhang, Y., Benet, L.Z. Role of P-glycoprotein and cytochrome P450 3A in limiting oral absorption of peptides and peptidomimetics. J. Pharm. ScL 1998, 11, 1322-1330 42. Johnson, D.A.; Amidon, G.L. the effect of enzymatic reaction on dissolution rate-theoretical analysis and experimental test. J.Pharm.Scl 1986, 75, 195-203. 43. Lauritsen, K.; Laursen, L.S.; Raskmadsen, J. Clinical pharmacokinetics of drugs used in the treatment of gastrointestinal diseases Part(I). Clin Pharmacokinet. 1990, 19, 11-31. 44. Goodman, L.S. The Pharmacological Basis of Therapeutics, 9th edition. McGraw Hill, 1996, 641, 1712-1792. 45. Walker, M. C.; Patsalos, P. N. Clinical pharmacokinetics of new antiepileptic drugs. Pharmacol Ther. 1995,67, 351-384. 46. Physicians’ Desk Reference 53rd ed. Montvale, N.J. Medical Economics Data Production Company. 1999, pl323, 3257, 1252, 2414-2415, 2072. 47. Sinko, P.J. Drug selection in early drug development: Screening for acceptable pharmacokinetic properties using combined in vitro and computational approaches. Curr. Opin. Drug Discov. Develop. 1999, 2, 42-48. 48. Waterbeemd, H.V.D, Camenisch, G.; Folkers, G.; Raevsky, O.A. Estimation of caco-2 cell permeability using calculated molecular descriptors. Quant. Struct-Act Rel 1996, 15, 480-490. 49. Rubas, W.; Jezyk, N.; Grass, G.M. Comparison of the permeability characteristics of a human colonic epithelial (Caco-2) cell line to colon of rabbit, monkey, and dog intestine and human drug absorption. Pharm. Res. 1993, 10, 113-118 50. Stenberg, P.; Luthman, K.; Artursson, P. Virtual screening of intestinal drug permeability. J. Cont.Rel 2000, 65, 231-243 51. Stewart, B.H.; Chan, H.O.; Lu, R.H.; Reyner, E.L.; Schmid, H.L.; Hamilton, H.W.; Steinbaugh, A.; Taylor, M.D. Comparison of intestinal permeabilities determined in multiple in vitro and in situ models: Relationship to absorption in humans. 52. Schanker, L.S.; Tocco, D.J.; Brodie, D.B.; Hogben, C.A.M. Absorption of drugs from the rat small intestine. J. Pharmacol Exp. Ther. 1958, 123, 81-88. 53. Dolusio, J.T.; Billups, N.F.; Dittert, L.W.; Sugita, E.T.; Swintosky, J.V. Drug absorption. I. An in situ rat gut technique yielding realistic absorption rates. J. Pharm. S ci 1969, 58, 1196-1200. 330 54. Taylor, D.C.; Grundy, R.; Loveday, B. Chronic dog intestinal loop model for studying drug absorption as exemplified by P-adrenoceptor blocking agents, atenolol and propranolol. J.Pharm. Sci. 1981,70,516-521. 55. Stewart, B.H.; Chan, O.H.; Jezyk, N.; Fleisher, D. Discrimination between drug candidates using models for evaluation of intestinal absorption. Adv. Drug Deliv. Rev. 1997, 23, 27-45. 56. Lennernas, H.; Ahrenstadt, O.; Hallgren, R.; Knutson, L.; Ryde, M.; Paalzow, L.K. Regional jejunal perfusion, a new in vitro approach to study oral drug absorption in man. Pharm. Res. 1992,9, 1243-1251. 57. Lennernas, H. Human Intestinal Permeability. J. Pharm. Sci. 1998, 87, 403-410. 58. Yee, S. In vitro permeability across caco-2 cells (colonic) can predict in vivo (small intestinal) absorption in man— fact or myth. Pharm. Res. 1997, 14,763-766. 59. Biganzoli, E.; Cavenaghi, L. A.; Rossi, R.; Brunati, M. C.; Nolli, M. L. Use of a caco-2 cell culture model for the characterization of intestinal absorption of abtibiotics. IL Farmaco 1999, 54, 594-599. 60. Irvine, J. D.; Takahashi, L.; Lockhart, K.; Cheong, J.; Tolan, J. W.; Selick, H. E.; Russell Grove, J. DCK (madin-darby canine kidney) cells: a tool for membrane permeability screening. J. Pharm. Sci. 1999, 88, 28-33. 61. Chiou, W. L.; Barve, A. Linear correlation of the fraction of oral dose absorbedof 64 drugs between humans and rats. Pharm. Res. 1998, 15, 1792-1795. 62. Lipinski, C. A.; Lombardo, P.; Dominy, B. W.; Feeney, P. J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 1997, 23, 3-25. 63. Clark, D. E. Rapid calculation of polar molecular surface area and its application to the prediction of transport phenomena. 1. prediction of intestinal absorption. J. Pharm. Sci. 1999, 88, 807-814. 64. Palm, K.; Luthman, K.; Ungell, A. L.; Strandlund, G.; Artursson, P. Correlation of drug absorption with molecular surface properties. J. Pharm. Sci. 1996, 85, 32-39. 65. Palm, K.; Stenberg, P.; Luthman, K.; Artursson, P. Polar molecular surface properties predict and the intestinal absorption of drugs in humans. Pharm Res. 1997, 14, 568-571. 66. Wessel, M. D.; Jurs, P. C.; Tolan, J. W.; Muskal, S. M. Prediction of human intestinalabsorption of drugs from molecular structure. J. Chem. Inf. Comput. Sci. 1998, 38, 726-735. 67. Sietsema, W. K. The absorption oral bioavailability of selected drugs. Int. J. Clin. Pharmacol. Ther. Toxicol. 1989, 27, 179-211. 68. Yazdanian, M.; Glynn, S. L.; Wright, J. L.; Hawi, A. Correlating partitioning and caco-2 cell permeability of structurally diverse small molecular weight compounds. Pharm. Res. 1998, 15, 1490-1494. 69. Vanpeer, A.; Woestenborghs, R.; Verlinden, M.; Meuldermans, W.; Heykants, J. Pharmacokinetics of cisapride in healthy volunteers. Digestion 1986, 34, 138. 70. Meuldermans, W.; Vanpeer, A.; Hendrickx, J.; Lauwers, W.; Swysen, E.; Bockx, M.; Woestenborghs, R.; Heykants, J. Excretion and biotransformation of cisapride in dogs and humans after oral administration. Drug Metab. Dispos. 1988, 16i 403-409. 331 71. Wiseman, L.R.; Faulds, D. Cisapride. An updated review of its pharmacology and therapeutic efficacy as a prokinetic agent in gastrointestinal motility. Drugs 1994, 47, 116-152. 72. Klotz, U.; Antonin, K. H. Pharmacokinetics and bioavailability of sodium valproate. Clin. Pharmacol Ther. 1977, 21, 736-743. 73. Latini, R.; Cerletti, C.; de Gaetano, G.; Dejana, E.; Galletti, P.; Urso, R.; Marzot M. Comparative bioavailability of aspirin from buffered, enteric-coated and plain preparations. Int. J. Clin. Pharmacol. Ther. Toxicol. 1986, 24, 313-318. 74. Rowland, M.; Riegelman, S. Pharmacokinetics of acetylsalicylic acid and salicylic acid after intravenous administration in man. J. Pharm. Sci. 1968, 57, 1313-1319. 75. Mandelli, M.; Tognoni, G.; Garattini, S. Clinical pharmacokinetics of diazepam. Clin. Pharmacokinet. 1978, 3, 72-91. 76. Olkkola, K.T.; Brunetto, A.V.; Mattila, M.J. Pharmacokinetics of oxicam nonsteroidal anti inflammatory agents. Clin. Pharmacokinet. 1994, 26, 107-120 77. Rodighiero, V. Effects of liver disease on pharmacokinetics - An update. Clin. Pharmacokinet. 1999, 37,399-431. 78. Labbe, L.; Turgeon, J. Clinical pharmacokinetics of mexiletine. Clin. Pharmacokinet. 1999, 37, 361-384. 79. Greene, D.S.; Barbhaiya, R.H. Clinical pharmacokinetics of nefazodone. Clin. Pharmacokinet. 1997, 33, 260-275. 80. Calvo, M.V.; Dominguez-Gil, A.; Minalles, J.M.; De Pablo, F. Pharmacokinetics of naproxen in healthy volunteers and in patients with diabetic microangiopathy. Int. J. Clin. Pharm. Biopharm. 1979, 17,486-491. 81. Elwes, R. D.; Binnie, C. D. Clinical pharmacokinetics of newer antiepileptic drugs - Lamotrigine, vigabatrin, gabapentin and oxcarbazepine. Clin. Pharmacokinet. 1996, 31, 403-415. 82. Lloyd-Jones, J.G.; Henson, R.; Nichols, J.D.; Greenslade, D.; Clifford, J.M. Pharmacokinetics of intravenous and oral tolmesoxide. Eur. J. Clin. Pharmacol. 1981, 19, 119-125. 83. Brien, J.; Loomis, C. Disposition and pharmacokinetics of disulfiram and calcium carbamide (calcium cyanamide). Drug. Metab. Rev. 1983, 14, 113-126. 84. Frishman, W. H.; Tepper, D.; Lazar, E. J.; Behrman, D. Betaxolol - a new long-acting betal - selective adrenergic blocker. J. Clin. Pharmacol. 1990, 30, 686-692. 85. Meier, J. Pharmacokinetic comparison of pindolol with other beta-adrenoceptor-blocking agents. J. Am. Heart J. 1982, 104, 364-373. 86. Fishman, J.; Roffwarg, H.; Heilman, L. Disposition of naloxone-7,8-3H in normal and narcotic- dependent man. J. Pharmacol. Exper. Ther. 1973, 187, 575-580. 87. Sonders, R. C. Pharmacokinetics of terazosin. Am. J. Med. 1986, 80, 20-24. 88. Davies, N.M.; Watson, M.S. Clinical pharmacokinetics of sulindac - A dynamic old drug. Clin. Pharmacokinet. 1997, 32, 437-459. 89. Kobari, T.; Iguro, Y.; Ito, T.; Namekawa, H.; Kato, Y.; Yamada, S. Absorption, distribution and excretion of sultopride in man and several animal species. Xenobiotica 1985, 15, 605-613. 90. Sachdeo, R.C. Topiramate - Clinical profile in epilepsy. Clin. Pharmacokinet. 1998, 34, 335-346. 332 91. narrower, A.D.B. Pharmacokinetics of oral antihyperglycaemic agents in patients with renal insufficiency. Clin. Pharmacokinet. 1996,31, 111-119. 92. Haustein, K.O.; Huiler, G. The pharmacokinetics and metabolism of propiverine in man. Eur. J. Drug Metab. Pharmacokinet. 1988, 13, 81-90. 93. Hinderling, P.H.; Hartmann, D. Pharmacokinetics of digoxin and main metabolite derivatives in healthy humans. Therap Drug Monit. 1991, 13, 381-401. 94. James, C.A.; Mant, T.G.K.; Rogers, H.J. Pharmacokinetics of intravenous and oral sodium 2- mercaptoethane sulfonate (mesna) in normal subjects. Brit. J. Clin. Pharmacol. 1987, 23, 561- 568. 95. Benet, L. Z. J. Pharmacokinetics/pharmacodynamics of furosemide in man : A review. Pharmacokinet. Biopharm. 1979, 7, 1-27. 96. Tucker, G.T.; Casey, C.; Phillips, P.J.; Connor, H.; Ward, J.D.; Woods, H.F. Metformin kinetics in healthy-subjects and in patients with diabetes-mellitus. Brit. J. Clin. Pharmacol. 1981, 12, 235- 246. 97. Scheen, A.J. Clinical pharmacokinetics of metformin. Clin. Pharmacokinet. 1996, 30, 359-371. 98. Euans, M.E.; Shenfield, G.M.; Thomas, N.; Walker, S.R.; Paterson, J.W. The pharmacokinetics of rimiterol in man. Xenobiotica 1974, 4, 681-692. 99. Strobach, H.; Wirth, K.E.; Rojsathaporn, K. Absorption, metabolism and elimination of strophanthus glycosides in man. Naunyn Schmiedebergs Arch. Pharmacol. 1986, 334, 496-500. 100. Yung, S.; Mayersohn, M.; Robinson, J.B. Ascorbic-acid absorption in humans - a comparison among several dosage forms. J. Pharm. Sci. 1982, 71, 282-285. 101. Goto, M.; Sugiyama, M.; Nakajima, S.; Yamashina, H. Fosfomycin kinetics after intravenous and oral-administration to human volunteers. Antimicrob. Agents Chemother. 1981, 20, 393-397. 102. Shimizu K, Fosfomycin: absorption and excretion. Chemotherapy 1977, 23(Suppl 1),153-158. 103. Kuemmerle, H P.; Murakawa, T.; Desantis, F. Pharmacokinetic evaluation of fosfidomycin, a new phosphonic acid antibiotic. Chemioterapia 1987, 6 , 113-119. 104. Cundy, K.C. Clinical pharmacokinetics of the antiviral nucleotide analogues cidofovir and adefovir. Clin. Pharmacokinet. 1999,36, 127-143. 105. Ahr, H.J.; Boberg, M.; Krause, HP.; Maul, W.; Muller, F.O.; Ploschke, H.J.; Weber, H.; Wunsche, C. Pharmacokinetics of acarbose .1. Absorption, concentration in plasma, metabolism and excretion after single administration of [c-14]acarbose to rats, dogs and man. Arzneimittelforschung 1989, 39-2, 1254-1260. 106. Holdiness, M R. Clinical pharmacokinetics of the antituberculosis drugs. Clin. Pharmacokinet. 1984, 9,511-544. 107. Kunin, C.; Chalmers, T. C.; Leevy, C. M.; Sebastyen, S. C.; Lieber, C. S.; Finland, M. Absorption of orally administered neomycin and kanamycin. New England J. Med. 1960, 26, 9380. 108. Ukabam, S.O.; Clamp, J R.; Cooper, B.T. Abnormal small intestinal permeability to sugars in patients with Crohn’s disease of the terminal ileum and colon. Digestion 1983, 27, 70-74. 109. Ukabam, S. O.; Cooper, B. T. Small intestinal permeability to mannitol, lactulose, and polyethylene glycol 400 in celiac disease. Dig. Dis. Sci. 1984, 29, 809-816. 333 110. Maxton, D.G.; Bjarnason, I.; Reynolds, A.P.; Catt, S.D.; Peters, T.J.; Menzies, I.S. Lactulose CR- 51 labelled ethylenediamine tetra-acetate, L-Rhamnose and polyethyleneglycol-500 as probe markers for assessment in vivo of human intestinal permeability. Clin. ScL 1986, 71,71-80. 111. Morino, A.; Nakamura, A.; Nakanashi, K.; Tatewaki, N.; Sugiyama, M. Species differences in the disposition and metabolism of camazepam. Xenobiotica 1985, 15, 1033-1043. 112. Oberbauer, R.; Krivanek, P.; Turnheim, K. Pharmacokinetics of indomethacin in the elderly. Clin. Pharmacokinet. 1993, 24, 428-434. 113. Taburet, A. M.; Schmidt, B. Pharmacokinetic optimization of asthma treatment Clin. Pharmacokinet. 1994, 26, 396-418. 114. Paton, D.M.; Webster, D.R. Clinical pharmacokinetics of HI-receptor antagonists (the antihistamines). Clin. Pharmacokinet. 1985, 10, 477-497. 115. Sallee, F. R.; Pollock, B. G. Clinical pharmacokinetics of imipramine and desipramine. Clin. Pharmacokinet. 1990, 18, 346-364. 116. Henson, R.; Lloyd-Jones, J.G.; Nichols, J.D.; Jordan, B.J. Pharmacokinetics of fenclofenac following single and multiple doses. Eur. J. Drug Metab. Pharmacokinet. 1980, 5, 217-223. 117. Welling, P.G. Influence of food and diet on gastrointestinal drug absorption: A review. Pharmacokinet. Biopharm. 1977, 5, 291-335. 118. Humpel, M.; Nieuweboer, B.; Milius, W.; Hanke, H.; Wendt, H. Kinetics and biotransformation of lormetazepam. J. Clin. Pharmacol Ther. 1980, 28, 673-679. 119. Davies, N.M.; Anderson, K.E. Clinical pharmacokinetics of diclofenac : Therapeutic insights and pitfalls. Clin. Pharmacokinet. 1997, 33, 184-213. 120. Clarke, S.E.; Austin, N.E.; Bloomer, J.C.; Haddock, R.E.; Highams, F.C.; Hollis, F.J.; Nash, M.; Shardlow, P.C.; Tasker, T.C.; Woods, F.R.; Allen, G.D. Metabolism and disposition of 14C- granisetron in rat, dog and man after intravenous and oral dosing. Xenobiotica 1994, 24, 1119- 1131. 121. Shinohara, Y.; Baba, S.; Kasuya, Y. Absorption, metabolism, and excretion of oral testosterone in humans by mass fragmentography. J. Clin. Endocrinol Metab. 1980, 51, 1459-1462. 122. Busto, U.; Bendayan, R.; Sellers, E.M. Clinical pharmacokinetics of non-opiate abused drugs. Clin. Pharmacokinet. 1989, 16, 1-26. 123. Harvey, S. C.; Winthrow, C. D. In Remington’s Pharmaceutical Sciences, Vol. 17; Gennaro, A. R. Ed.; Mack Publishing Company, Easton, PA, 1985, pp 951-1001. 124. Orme, M.; Back, D.J.; Ward, S.; Green, S. The pharmacokinetics of ethinylestradiol in the presence and absence of gestodene and desogestrel. Contraception 1991,43, 305-316. 125. Roila, F.; Del Favero, A. Ondansetron clinical pharmacokinetics. Clin. Pharmacokinet. 1995, 29, 95-109. 126. McTavish, D.; Sorkin, E.M. Verapamil: An updated review of its pharmacodynamic and pharmacokinetic properties, and therapeutic use in hypertension. Drugs 1989, 38, 19-76. 127. Schedl, H. P. Absorption of steroid hormones from the human small intestine. J. Clin. Endocrinol Metab. 1965,25, 1309-1316. 128. Mahler, C.; Verhelst, J.; Denis, L. Clinical pharmacokinetics of the antiandrogens and their efficacy in prostate cancer. Clin. Pharmacokinet. 1998, 34, 405-417. 334 129. Sotelo, J.; Jung, H. Pharmacokinetic optimisation of the treatment of neurocysticercosis. Clin. Pharmacokinet. 1998, 34, 503-515. 130. Hildebrand, M.; Staks, T.; Schutt, A.; Matthes, H. Pharmacokinetics of cicaprost in healthy- volunteers. Prostaglandins 1989, 37, 259-273. 131. Agundez, J.A.G.; Martinez, C.; Benitez, J. Metabolism of aminopyrine and derivatives in man - in-vivo study of monomorphic and polymorphic metabolic pathways. Xenobiotica 1995, 25, 417- 427. 132. Greenblatt, D.J.; Divoll, M.K.; Soong, M.H.; Boxenbaum, H.G.; Harmatz, J.S.; Shader, R.I. Desmethyldiazepam pharmacokinetics : Studies following intravenous and oral desmethyldiazepam, oral clorazepate, and intravenous diazepam. J Clin. Pharmacol. 1988, 28, 853-859. 133. Filer, C.W.; Humphrey, M.J.; Jeffrey, D.J.; Jones, K.H.; Langley, P.P. The metabolism of carfecillin in rat, dog and man. Xenobiotica 1980, 10, 761-769. 134. Vermeulen, A. The metabolism of 4-14 C prednisolone. J. Endocrin. 1959, 18, 278-291. 135. Frey, B.M.; Brandenberger, A.W.; Frey, F.J.; Widmer, H.R. The systemic availability of prednisolone after orally administered prednisone. Pharmaceutica Acta. Helvetiae 1984, 59, 301- 309. 136. Dressman, J. B.; Amidon, G. L.; Fleisher, D. Absorption potential-estimating the fraction absorbed for orally-administered compounds. J. Pharm. Sci. 1985, 74, 588-589. 137. Borgstrom, L.; Johansson C. G.; Larsson H.; Lenander R. Pharmacokinetics of propanolol. J. Pharmacokinet. Biopharm. 1981, 9,419-429. 138. Bayliss, P.F.C.; Case, D.E. Blood level studies with viloxazine hydrochloride in man. Br. J. Clin. Pharmacol. 1975, 2, 209-214. 139. Pisani, F. Pharmacokinetics of the antidepressant drug viloxazine in normal subjects and in epileptic patients receiving chronic anticonvulsant treatment. Psychopharmacology 1986, 90, 295- 298. 140. Hoi ford, N. H. Clinical pharmacokinetics and pharmacodynamics of warfarin - understanding the dose-effect relationship. Clin. Pharmacokinet. 1986, 11, 483-504. 141. Beermann, B.; Hellstrom, K.; Rosen, A. The gastrointestinal absorption of atropine in man. Clin. Sci. 1971,40, 95-106. 142. Kirsten, R.; Nelson, K.; Kirsten, D.; Heintz, B. Clinical pharmacokinetics of vasodilators - Part I. Clin. Pharmacokinet. 1998, 34, 457-482. 143. Sedaghat, A.; Ahrens, Jr, E.H. Lack of effect of cholestyramine on the pharmacokinetics of clofibrate in Man. Eur. J. Clin. Invest. 1975, 5, 177-185. 144. Cayen, M.N. Disposition, metabolism and pharmacokinetics of antihyperlipidemic agents in laboratory animals and man. Pharmac. Ther. 1985, 29, 157-204. 145. Schwartz, D. E.; Ziegler, W. H. Postgrad. Assay and pharmacokinetics of trimethoprim in man and animals. Postgrad. Med. J. 1969, 45, Suppl:32-37. 146. Troy, S.M.; Parker, V.P.; Hicks, D R.; Pollack, G.M.; Chiang, S.T. Pharmacokinetics and effect of food on the bioavailability of orally administered venlafaxine. J. Clin. Pharmacol. 1997, 37, 954-961. 335 147. Eichelbaum, M.; Ochs, H. R.; Roberts, G.; Somogyi, A. Pharmacokinetics and metabolism of antipyrine (phenazone) after intravenous and oral-administration. Drug Res. 1982, 32, 575-578. 148. Pentikainen, P. J.; Penttila, A.; Neuvonen, P. J.; Gothoni, G. Fate of [14 C]-bumetanide in man. Br. J. Clin. Pharmacol. 1977, 4, 3944. 149. Kirsten, R.; Nelson, K.; Kirsten, D.; Heintz, B. Clinical pharmacokinetics of vasodilators - Part II. Clin. Pharmacokinet. 1998, 35, 9-36. 150. Debruyne, D.; Ryckelynck, J-P. Clinical pharmacokinetics of fluconazole. Clin. Pharmacokinet. 1993,24, 10-27. 151. Anttila, M.; Arstila, M.; Pfeffer, M.; Tikkanen, R.; Vallinkoski, V.; Sundquist, H. Human pharmacokinetics of sotalol. Acm. Pharmacol. Toxicol. (Copenh.) 1976, 39, 118-128. 152. Hanyok, J.J. Clinical pharmacokinetics of sotalol. Am. J. Cardiol. 1993, 72, 19A-26A. 153. Quiding, H.; Anderson, P.; Bondesson, U.; Boreus, L.O.; Hynning, P.-A. Plasma concentrations of codeine and its metabolite, morphine, after single and repeated oral administration. Eur. J. Pharmacol. 1986, 30, 673-677. 154. Bodd, E.; Beylich, K.M.; Christopherson, A.S.; Morland, Oral administration of codeine in the presence of ethanol - a pharmacokinetic study in man. J. Pharmacol. Toxicol. 1987, 61, 297-300. 155. Whitwan, J.G.; Amein, R. Pharmacology of flumazenil. Acta Anaesthesiologica Scandinavica 1995, 39(suppl. 108), 3-14. 156. Lee, E. J. D.; Williams, K.; Day, R.; Graham, G.; Champion, D. Stereoselective disposition of ibuprofen enantiomers in man. Br. J. Clin. Pharmacol. 1985, 19, 669-674. 157. Bodem, G.; Chidsey, C. A. Pharmacokinetic studies of practolol, a beta adrenergic antagonist in man. Clin. Pharmacol Ther. 1973, 14, 26-29. 158. Tocco, D.J.; Dunacan, E.W.; Deluna, F.A.; Hucker, H.B.; Gruber, V.F.; Vandenheuvel, W.J.A. Physiological disposition and metabolism of timolol in man and laboratory animals. Drug metab. dispos. 1975, 3, 361-370. 159. Johansson, R.; Regardh, C. G.; Sjogren, J. Absorption of alprenolol in man from tablets with different rates of release. Acta Pharm. Suec. 1971, 8, 59-70. 160. Artursson, P.; Karlsson J. Correlation between oral drug absorption in humans and apparent drug permeability coefficients in human intestinal epithelial (caco-2) cells. Biochem. Biophys. Res. Commun. 1991, 175, 880-885. 161. Jamali, F.; Brocks, D. R. Clinical pharmacokinetics of ketoprofen and its enantiomers. Clin. Pharmacokinet. 1990, 19, 197-217. 162. Mlynaryk, P.; Kirsner, J.B. Absorption and excretion of 1,2,H^-hydrocortisone in regional enteritis and ulcerative colitis, with a note on hydrocortisone production rates. Gastroenterology 1963,44,257-260. 163. Peterson, R. E.; Wyngaarden, J. B.; Guerra, L. S.; Brodie, B. B.; Bunim, J. J. The physiological disposition and metabolic fate of hydrocortisone in man. J. Clin. Invest. 1955, 34, 1779-1794. 164. Brocks, D.R.; Jamali, F. Clinical pharmacokinetics of Keterolac Tromethamine. Clin. Pharmacokinet. 1992, 23, 415-427. 336 165. Davies, N.M.; Skjodt, N.M. Clinical pharmacokinetics of meloxicam. Clin. Pharmacokinet. 1999, 36, 115-126. 166. Jung, D.; Powell, J. R.; Walson, P.; Perrier, D. Effect of dose on phenytoin absorption. Clin. Pharmacol Ther. 1980, 28, 479-485. 167. Glazko, A. J.; Dill, W. A.; Kazenko, A.; Wolf, L. M.; Carnes, H. E. Antibiot. Biochemical studies on chloramphenicol (chloromycetin). II. Tissue distribution and excretion studies. Chemother. 1958,8,516-526. 168. Ambrose, P. Clinical pharmacokinetics of chloramphenicol and chloramphenicol succinate. J. Clin. Pharmacokinet. 1984, 9, 222-238. 169. Shumaker, R. C.; Fantel, C.; Kelton, E.; Wong, K.; Weliky, I. Evaluation of the elimination of 14C Felbamate in healthy men. Epilepsia 1990, 31, 642. 170. Tegeder, I.; Lotsch, J.; Geisslinger, G. Pharmacokinetics of opioids in liver disease. Clin. Pharmacokinet. 1999, 37, 17-40. 171. Heinig, R. Clinical pharmacokinetics of nisoldipine coat-core. Clin. Pharmacokinet. 1998, 35, 191-208. 172. Sonne, J.; Loft, S.; Dossing, M.; Vollmer-Larsen, A.; Olesen, K.L.; Victor, M.; Andreasen, P.; Andreasen, P.B. Bioavailability and pharmacokinetics of oxazepam. Eur. J. Clin. Pharmacol. 1988, 35, 385-389. 173. Gardner, M. J.; Wilner, K. D.; Hansen, R. A.; Fouda, H. G.; McMahon, G. F. Single and multiple dose pharmacokinetics of tenidap sodium in healthy subjects. Br. J. Clin. Pharmacol. 1995, 39(sl), sll-15. 174. Edgar, B.; Regardh, C. G.; Johnsson, G.; Johansson, L.; Lundborg, P.; Lofberg, I.; Ronn, O. Felodipine kinetics in healthy men. Clin. Pharmacol Ther. 1985, 38, 205-211. 175. Soons, P.A.; Breimer, D.D. Stereoselective pharmacokinetics of oral and intravenous nitrendipine in healthy male subjects. Br. J. Clin. Pharmacol. 1991, 32, 11-16. 176. Mikus, G.; Fischer, C. Application of stable isotope methodology to study the pharmacokinetics, bioavailability and metabolism of nitrendipine after iv and po administration. Br. J. Clin. Pharmacol. 1987, 24, 561-569. 177. Sweatman, T.W.; Renwick, A.G.; Burgess, C D. The pharmacokinetics of saccharin in man. Xenobiotica 1981, 11, 531-540. 178. Findlay, J.W.A.; Van Wyck Fleet, J.; Smith, P.G.; Butz, R.F.; Hinton, M.L.; Blum, M.R.; Schroeder, D.H. Pharmacokinetics of bupropion, a novel antidepressant agent, following oral administration to healthy patients. Eur. J. Clin. Pharmacol. 1981, 21, 127-135. 179. Johnson, M.A.; Moore, K.H.P.; Yuen, G.J.; Bye, A.; Pakes, G.E. Clinical pharmacokinetics of lamivudine. Clin. Pharmacokinet. 1999, 36, 41-66. 180. Glare, P.A.; Walsh, T.D. Clinical pharmacokinetics of morphine. Ther. Drug Monit. 1991, 13, 1- 23. 181. Kripalani, K. J.; Mckinstry, D. N.; Singhvi, S. M.; Willard, D. A.; Vukovich, R. A.; Migdalof, B. H. Disposition of captopril in normal subjects. Clin. Pharmacol Ther. 1980, 27, 636-641. 337 182. Duchin, K.L.; McKinstfy, D.N.; Cohen, A.I.; Migdalof, B.H. Pharmacokinetics of captopril in healthy subjects and in patients with cardiovascular diseases. Clin. Pharmacokinet. 1988, 14, 241-259. 183. Raaflaub, J.; Speiser-Courvoisier, J. Zur pharmakokinetik von bromazepam beim menschen. Arzneimittelforschung 1974, 24, 1841-1844. 184. Macheras, P.; Reppas, C.; Dressman, J. B. Estimate of volume flow ratio of gastrointestinal (GI) fluids in humans using pharmacokinetic data. Pharm. Res. 1990,7, 518-522. 185. Roberts, M.S.; Rumble, R.H.; Wanwimolruk, S.; Thomas, D.; Brooks, P.M. Pharmacokinetics of aspirin and salicylate in elderly subjects and in patients with alcoholic liver disease. Eur. J. Clin. Pharmacol. 1983, 25, 253-261. 186. Everett, D.W.; Olsen, S.J.; Sugerman, A.A.; Ita, C.; Stewart, M.B.; Sherman, J.W.; Kripalani, K.J. Disposition of SQ 32,756 (BV-araU) in healthy male subjects. Pharm. Res. 1992, 9, S316. 187. Lew, K. H.; Ludwig, E. A.; Milad, M. A.; Donovan, K.; Middleton, E. Jr.; Ferry, J. J.; Jusko, W. J. Gender-based effects on methylprednisolone pharmacokinetics and pharmacodynamics. Clin. Pharmacol Ther. 1993, 54,402-414. 188. Lie-A-Huen, L.; Proost, J. H.; Kingma, J. H.; Meijer, D. K. F. Absorption kinetics of oral and rectal flecainide in healthy subjects. Eur. J. Clin. Pharmacol. 1990, 38, 595-598. 189. Guentert, T. W.; Upton, R. A.; Holford, N. H.; Bostrom, A.; Rigelman, S., Gastrointestinal absorption of quinidine from solutions and commercial tablets. J.Pharmacokinet. Biopharm. 1990, 8, 243-255. 190. Greenblatt, D.G.; Pfeifer, H.J.; Ochs, H.R.; Franke, K.; McLaughlin, D.S.; Smith, T.W.; Koch- Wesser, J. Pharmacokinetics of quinidine in humans after intravenous, intramuscular and oral administration. J. Pharmacol. Exp. Ther. 1977, 202, 365-378. 191. Foster, R. T.; Carr, R. A. In Analytical Profiles of Drug Substances, Vol. 19; Florey, K., Ed.; Academic: New York,1990, pp 1-26. 192. Lee, C. S.; Gambertoglio, J. G.; Brater, D. G.; Benet, L. Z. Kinetics of oral ethambutol in the normal subject. Clin. Pharmacol Ther. 1977,22, 615-621. 193. Forrest, J.A.H.; Clements, J.A.; Prescott, L.F. Clinical pharmacokinetics of paracetamol. Clin. Pharmacokinet. 1982, 7, 93-107. 194. Stewart, B. H.; Kugler, A. R.; Thompson, P. R.; Bockbrader, H. N. A saturable transport mechanism in the intestinal absorption of gabapentin is the underlying cause of the lack of proportionality between increasing dose and drug levels in plasma. Pharm. Res. 1993, 10, 276- 281. 195. Duggan, D.E.; Yeh, K.C.; Matalia, N.; Ditzler, C.A.; McMahon, F.G. Bioavailability of oral dexamethasone. Clin. Pharmacol. Ther. 1975, 18, 205-209. 196. Holmes, B.; Brogden, R.N.; Heel, R.C.; Speight, T.M.; Avergy, G.S. Guanabenz - a review of its pharmacodynamic properties and therapeutic efficacy in hypertension. Drugs 1983, 26, 212-229. 197. Gill, K.S.; Wood, M.J. The clinical pharmacokinetics of famciclovir. Clin. Pharmacokinet. 1996, 31, 1-8. 338 198. Tilstone, W.J.; Dargie, H.; Dargie, E.N.; Morgan, H.G.; Kennedy, A.C. Pharmacokinetics of metolazone in normal subjects and in patients with cardiac or renal failure. Clin. Pharmacol. Ther. 1974, 16, 322-329. 199. Heel, R.C.; Brogden, R.N.; Speight, T.M.; Avery, G.S. Fenoterol; Review of its pharmacological properties and therapeutic efficacy in asthma. Drugs 1978, 15, 3-32. 200. Rominger, V.K.L.; Polimann, W. Comparative pharmacokinetic studies on fenoterol hydrobromide in rat, dog and man. Arzneimmittelforschung 1972, 22, 1190-1196. 201. Cruickshank, J.M. The clinical importance of cardioselectivity and lipophilicity in beta blockers. Am. Heart J. 1980, 100, 160-178. 202. Morrison, R. A.; Singhvi, S. M.; Greasey, W. A. and Willard D. A. Dose proporptionality of nadolol pharmacokinetics after intravenous administration to healthy subjects. Eur. J. Clin. Pharmacol. 1988, 33, 625-628. 203. Bressolle, F.; Bres, J.; Faure-Jeantis, A. Absolute bioavailability, rate of absorption, and dose proportionality of sulpiride in humans. J. Pharm. Sci. 1992, 81, 26-32. 204. Dengler, H.J.; Hengstmann, J.H. Metabolism and pharmacokinetics of orciprenaline in various animal species and man. Arch. Int. Pharmacodyn. Ther. 1976, 223, 71-87. 205. Sjovall, J.; Karlsson, A.; Ogenstad, S.; Sandstrom, E.; Saarimaki, M. Pharmacokinetics and absorption of foscarnet after intravenous and oral administration to patients with human immunodeficiency virus. Clin. Pharmacol. Ther. 1988,44, 65-73. 206. Tse, F.L.S.; Jaffe, J.M. Pharmacokinetics of PN200-110 (isradipine), a new calcium-antagonist, after oral administration in man. Eur. J. Clin. Pharmaol. 1987, 32, 361-365. 207. Borgstrom, L.; Nyberg, L.; Jonsson, S.; Lindberg, C.; Paulson, J. Pharmacokinetic evaluation in man of terbutaline given as separate enantiomers and as the racemate. Br. J. Clin. Pharmac. 1989, 27, 49-56. 208. Morgan, D.J. Clinical pharmacokinetics of beta-agonists. Clin. Pharmacokinet. 1990, 18, 270- 294. 209. Jack, D.B. Handbook of Clinical Pharmacokinetic Data. Basingstoke : Macmillan Publishers Ltd 1992. 210. Lennernas, H.; Eager, G. Pharmacodynamics and pharmacokinetics of the HMG-CoA reductase inhibitors. Clin. Pharmacokinet. 1997, 32, 403-425. 211. Tse, F. L.; Jaffe, J. M.; Troendle, A. Pharmacokinetics of fluvaststatin after single and multiple doses in normal volunteers. J. Clin. Pharmacol. 1992, 32, 630-638. 212. Kampmann, J.; Skovsted, L. The kinetics of propylthiouracil in Hyperthyroidism. Acta Pharmac. Tox. 1975, 37,201-210. 213. Kampmann, J.; Skovsted, L. The pharmacokinetics of propylthiouracil. Acta Pharmac. Tox. 1974, 35, 361-369. 214. Scatina, J.A.; Wells, D.S.; Kimmel, H.B.; Kemper, C.J.; Sisenwine, S.F. Excretion and metabolism of recainam, a new antiarrhythmic drug, in laboratory-animals and humans. Drug Metab. Dispos. 1990, 18, 746-752. 215. Beermann, B.; Groschinsky-Grind, M. Pharmacokinetics of hydrochlorothiazide in man. Eur. J. Clin. Pharmacol. 1977, 12, 297-303. 339 216. Mcllhenny, H.M.; Ghaly, M.S.D. Biotransformation of pirbuterol by the rat, dog, and human. Federation Proceedings 1979, 38, 442. 217. Fowler, P. A.; Lacey, L. P.; Thomas, M.; Keene, O. N.; Tanner, R.J.; Baber, N.S. The clinical pharmacology, pharmacokinetics and metabolism of sumatriptan. Eur. Neurol. 1991, 31, 291-294. 218. Dixon, C. M.; Saynor, D. A.; Andrew, P. D.; Oxford, J.; Bradbury, A.;. Tarbit, M. H. Disposition of sumatriptan in laboratory animals and humans. Drug Metab. Dispos. 1993, 21, 761-769. 219. Scott, A.K. Sumatriptan clinical pharmacokinetics. Clin. Pharmacokinet. 1994, 27, 337-344. 220. Grayson, M.F.; Smith, A.J.; Smith, R.N. Absorption, distribution and elimination of 14C- amiloride in normal human subjects. Br. J. Pharmacol. 1971, 43, 473-474. 221. Jacobson, M. A.; de Miranda, P.; Cederberg, D. M.; Burnette, T.; Cobb, E.; Brodie, H. R.;Mills, J. Human pharmacokinetics and tolerance of oral ganciclovir. Antimicrob. Agents Chemother. 1987, 31, 1251-1254. 222. Lobley, R.W.; Burrows, P.C.; Warwick, R.; Dawson, D.J.; Holmes, R. Simultaneous assessment of intestinal permeability and lactose tolerance with orally administered raffinose, lactose and L- arabinose. Clin. Sci. 1990, 79, 175-183. 223. Data obtained from Roche Products Ltd. 224. Welling, P.G.; Selen, A.; Pearson, J.G.; Kwok, P.; Rogge, M.C.; Ifan, A.; Marrero, D.; Craig, W.A.; Johnson, C.A. A pharmacokinetic comparison of cephalexin and cefadroxil using HPLC assay procedures. Biopharm. Drug Dispos. 1985, 6 , 147-157. 225. Douglas, J.G.; McLeod, M.J. Pharmacokinetic factors in the modern drug treatment of tuberculosis. Clin. Pharmacokinet. 1999, 37(2), 127-146. 226. DeSante, K. A.; Zeckel, M. L. Pharmacokinetic profile of loracarbef. Am. J. Med. 1992, 93(S6A), S16-S19. 227. Sjovall, J.; Alvan, G.; Huitfeldt, B. Intra- and inter-individual variation in pharmacokinetics of intravenously infused amoxycillin and ampicillin to elderly volunteers. Br. J. Clin. Pharmac. 1986,21, 171-181. 228. Sinko, P. J.; Leesman, G. D.; Amidon, G. L. Predicting fraction dose absorbed in humans using a macroscopic mass balance approach. Pharm. Res. 1991, 8, 979-988. 229. Spyker, D. A.; Rugloski, R. J.; Vann, R. L.; Obrien, W. M. Pharmacokinetics of amoxicillin : Dose dependence after intravenous, oral, and intramuscular administration. Antimicrob. Agents Chemother. 1977, 11, 132-141. 230. Sjovall, J.; Alvan, G.; Westerlund, D. Dose-dependent absorption of amoxycillin and bacampicillin. Clin. Pharmacol. Ther. 1985, 38, 241-250. 231. Arancibia, A. Dose-dependent bioavailability of amoxycillin. Int. J. Clin. Pharmacol. Ther. Tox. 1998,26, 300-303. 232. Haria, M.; Lamb, H.M. Trovaflaxacin. Drugs 1997, 54, 435-445. 233. Simons, F.E.R.; Simons, K.G. J. Clinical pharmacology of new histamine H-1 receptor antagonists. Clin. Pharmacokinet. 1999, 36, 329-352. 234. Robertson, D.R.C.; Wood, N.D.; Everest, H.; Monks, K.; Waller, D.G.; Renwick, A.G.; George, C.F. The effect of age on the pharmacokinetics of levodopa administered alone and in the presence of carbidopa. Br. J. Clin. Pharmacol. 1989, 28, 61-69. 340 235. Pfeffer, M.; Gaver, R. C.; Ximenez, J. Human intravenous pharmacokinetics and absolute oral bioavailability of cefatrizine. Antimicrob. Agents Chemother. 1983, 24, 915-920. 236. Adithan, C.; Sriram, G.; Swaminathan, R.P.; Shashindran, C.H.; Bapna, J.S.; Krishnan, M.; Chandrasekar, S. Differential effect of type-i and type-ii diabetes-mellitus on serum ampicillin levels. Int. J. Clin. Pharmacol. Ther. 1989, 27,493-498. 237. Pilbrant, A.; Schannong, M.; Vessman, J. Pharmacokinetics and bioavailability of tranexamic acid. Eur. J. Clin. Pharmacol. 1981, 20, 65-72. 238. Haegele, K.D.; Aiken, R.G.; Grove, J.; Schechter, P.J.; Kochweser, J. Kinetics of alpha- difluoromethylornithine - an irreversible inhibitor of ornithine decarboxylase. Clin. Pharmacol. Ther. 1981,30,210-217. 239. Campbell, N.R.C.; Skerjanec, A.; Tam, Y.; Robertson, S.; Burgess, E. Methyldopa kinetics before and after ingestion of methyldopa for 8 weeks. Eur. J. Clin. Pharmacol. 1995, 48, 397-400. 240. Amidon, G.L.; Lee, H.J. Absorption of peptide and peptidomimetic drugs. Annu Re.v Pharmacol. Toxicol. 1994, 34, 321-341. 241. Kansy, M.; Senner, F.; Gubernator, K. Physicochemical higher throughput screening: parallel artificial membrane permeation assay in the description of passive absorption precess. J. Med. Chem. 1998,41, 1007-1010. 242. Vree, T.B.; Waitzinger, J.; Hammermaier, A.; Radhofer-Welte, S. Absolute bioavailability, pharmacokinetics, renal and biliary clearance of distigmine after a single oral dose in comparison to i.v. administration of C-14-distigmine-bromide in healthy volunteers. Int. J. Clin. Pharmacol. Ther. 1999, 37, 393-403. 243. Klecker, R.W.; Collins, J.M.; Yarchoan, R.; Thomas, R.; Jenkins, J.F.; Broder, S.; Myers, C.E. Plasma and cerebrospinal fluid pharmacokinetics of 3'-azido-3'-deoxythymidine - a novel pyrimidine analog with potential application for the treatment of patients with aids and related diseases. Clin. Pharmacol. Ther. 1987, 41, 407-412. 244. Acosta, E. P.; Page, L. M.; Fletcher, C. V. Clinical pharmacokinetics of zidovudine - an update. Clin. Pharmacokinet. 1996, 30, 251-262. 245. Taylor, I.W.; Taylor, T.; James, I.; Doyle, G.; Dorf, G.; Darragh, A.; Chasseaud, L.F. Pharmacokinetics of the anti-inflammatory drug ximoprofen in healthy subjects and in disease states. Eur. J. Clin. Pharmacol. 1991, 40, 101-106. 246. Davies, D. S.; Wing, A. M.; Reid, J. L.; Neill, D. M.; Tippett, P.; Dollery, C. T. Pharmacokinetics and concentration-effect relationships of intravenous and oral clonidine. Clin. Pharmacol Ther. 1977,21,593-601. 247. Lowenthal, D. T.; Matzek, K. M.; MacGregor, T. R. Clinical pharmacokinetics of clonidine. Clin. Pharmacokinet. 1988, 14, 287-310. 248. Lowenthal, D. T. Pharmacokinetics of clonidine. J. Cardiovasc. Pharmacol. 1980, 2(suppl.l), S29-S37. 249. Adam, D.; Timmler, R. Pharmacokinetics of ceftizoxime with and without administration of probenecid. Arzneimittelforschung 1982, 32, 416-419. 250. Kreis, W.; Gordon, C.S.; DeJager, R.; Krakoff, I.H. Physiological disposition of 2,2'-anhydro-l- beta-D-arabinofuranosyl-5-fluorocytosine in humans. Cancer Res. 1975, 35, 2453-2460. 341 251. Rapeport, W.G. Clinical pharmacokinetics of Bretylium. Clin. Pharmacokinet. 1985, 10, 248-256. 252. Overdiek, H.; Merkus, W. The metabolism and biopharmaceutics of spironolactone in man. Rev. Drug Metabol. Drug Interact. 1987, 5,273-302. 253. Clark, P. I.; Slevin, M. L. The clinical pharmacology of etoposide and teniposide. Clin. Pharmacokinet. 1987, 12, 223-252. 254. Nissen, N. I.; Hansen, H. H.; Pedersen, H.; Stroyer, I.; Dombernowsky, P.; Hessellund, M. Clinical trial of the oral form of a new podophyllotoxin derivative, VP-16-213 (NSC-141540), in patients with advanced neoplastic disease. Cancer Chemother. Rep. 1975, 59, 1027-1029. 255. Hayton, W.L.; Walstad, R.A.; Thurmannnielsen, E.; Kufaas, T.; Kneer, J.; Ambros, R.J.; Rugstad, H E.; Monn, E.; Bodd, E.; Stoeckel, K. Pharmacokinetics of intravenous cefetamet and oral cefetamet pivoxil in children. Antimicrob. Agents Chemother. 1991, 35, 720-725. 256. Finn, A.; Straughn, A.; Meyer, M.; Chubb, J. Effect of dose and food on the bioavailability of cefuroxime axetil. Biopharm. Drug Dispos. 1987, 8, 519-526. 257. Lalak, N.J.; Morris, D.L. Azithromycin clinical pharmacokinetics. Clin. Pharmacokinet. 1993, 25, 370-374. 258. Murdoch, D.; McTavish, D. Fosinopril - A review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential in essential hypertension. Drugs 1992, 43, 123-140. 259. Hui, K.; Duchin, K. L.; Kripalani, K.J.; Chan, D.D.; Kramer, P.K.; Yanagawa, N. Pharmacokinetics of fosinopril in patients with various degrees of renal function. Clin. Pharmacol Ther. 1991, 49, 457-467. 260. Singhvi, S. M.; Pan, H. Y.; Morrison, R. A.; Willard, D. A. Disposition of pravastatin sodium, a tissue selective HMG-CoA reductase inhibitor, in healthy subjects. Br. J. Clin. Pharmacol. 1990, 29, 239-243. 261. Fahr, A. Cyclosporin clinical pharmacokinetics. Clin. Pharmacokinet. 1993, 24, 472-495. 262. Gomez, D.Y.; Wacher, V.J.; Tomlanovich, S.J.; Hebert, M.F.; Benet, L.Z. The effects of ketoconazole on the intestinal metabolism and bioavailability of cyclosporine. Clin. Pharmacol.Ther. 1995,58, 15-19. 263. Kopitar, Z.; Vrhovac, B.; Povsic, L.; Plavsic, F.; Francetic, I.; Urbancic, J. The effect of food and metoclopramide on the pharmacokinetics and side-effects of bromocriptine. Eur. J. Drug Metab. Pharmacokinet. 1991, 16, 177-181. 264. Jauhiainen, K.; Eksborg, S.; Kangas, L.; Peril, M.; Alfthan, O. The absorption of doxorubicin and mitomycin C in perioperative installation - an experimental and clinical study. Eur. Urol. 1985, 11,269-272. 265. Speth, P.A.J.; Vanhoesel, Q.G.C.M.; Haanen, C. Clinical pharmacokinetics of doxorubicin. Clin. Pharmacokinet. 1988, 15, 15-31. 266. O’Callaghan, C H.; Sykes, R.B.; Ryan, D.M.; Foord, F.D.; Muggleton, P.W. Cefuroxime - a new cyclosporin antibiotic. J.Antibiot. (Tokyo) 1976, 29, 29-37. 267. Prueksaritanont, T.; Chiou, W.L. Absorption of lothalamate after oral administration - A preliminary study in humans and interspecies differences. Biopharm. Drug Dispos. 1987, 8, 99- 101. 342 268. Pieniaszek, H. J. Jr.; Resetarits, D. E.; Wilferth, W. W.; Blumenthal, H. P.; Bates, T. R. Relative systemic availability of sulfapyridine from commercial enteric-coated and uncoated sulfasalazine tablets. J. Clin. Pharmacol. 1979, 19, 39-45. 269. Balfour, J.A.; Goa, K.L. Benazepril - A review of its pharmacodynamic and pharmacokinetic properties and therapeutic efficacy in hypertension and congestive heart failure. Drugs 1991, 42, 511-539. 270. Waldmeier, F.; Kaiser, G.; Ackermann, R.; Faigle, J.W.; Wagner, J.; Barner, A.; Lasseter, K.C. The disposition of [14C]-labelled benazepril HCl in normal adult volunteers after single and repeated oral dose. Xenobiotica 1991,21,251-261. 271. Lancaster, S.G.; Todd, P.A. Lisinopril - a preliminary review of its pharmacodynamic and pharmacokinetic properties, and therapeutic use in hypertension and congestive heart failure. Drugs 1988, 35, 646-669. 272. Kelly, J. G.; O’Malley, K. Clinical pharmacokinetics of the newer ACF inhibitors - a review. Clin. Pharmacokinet. 1990, 19, 177-196. 273. Lennernas, H.; Ahrenstedt, O.; Ungell, A. L. Intestinal drug absorption during induced net water absorption in man - a mechanistic study using antipyrine, atenolol and enalaprilat. Br. J. Clin. Pharmacol. 1994, 37, 589-596. 274. Kubo, S.H.; Cody, R.J. Clinical pharmacokinetics of the angiotensin converting enzyme inhibitors - a review. Clin. Pharmacokinet. 1985, 10, 377-391. 275. Ulm, F.H.; Hichens, M.; Gomez, H.J.; Till, A.F.; Hand, F.; Vassil, T.C.; Biollaz, J.; Fnalapril maleate and a lysine analog (MK-521) - disposition in man. Bruner, H R.; Schelling, J.L. Br. J. Clin. Pharmacol. 1982, 14, 357-362. 276. Janknegt, R.; Demarie, S.; Bakkerwoudenberg, I.A.J.M.; Crommelin, D.J.A. Liposomal and lipid formulations of amphotericin-B - clinical pharmacokinetics. Clin. Pharmacokinet. 1992, 23, 279- 291. 277. Swabb, F.A.; Sugerman, A.A.; Stern, M. Oral bioavailability of the monobactam aztreonam (sq 26,776) in healthy-subjects. Antimicrob. Agents Chemother. 1983, 23, 548-550. 278. Mattie, H. Clinical pharmacokinetics of aztreonam. Clin. Pharmacokinet. 1988, 14, 148-155. 279. Welker, H.A.; Wiltshire, H.; Bullingham, R. Clinical pharmacokinetics of mibefradil. Clin. Pharmacokinet. 1998, 35, 405-423. 280. Osman, M. A.; Patel, R. B.; Irwin, D. S.; Craig, W. A.; Welling, P. G. Bioavailability of chlorothiazide from 50, 100, and 250 mg solution doses. Biopharm. Drug Dispos. 1982, 3, 89-94. 281. Laskin, O.L. Clinical pharmacokinetics of acyclovir. Clin. Pharmacokinet.l9S3, 8, 187-201. 282. Holmes, B.; Brogden, R.N.; Richards, D.M. Norfloxacin - A review of its antibacterial activity, pharmacokinetic properties and therapeutic use. Drugs 1985, 30,482-513. 283. Stein, G. F. Review of the bioavailability and pharmacokinetics of oral norfloxacin. Am. J. Med. 1987, 82(6B), 18-21. 284. Chabner, B. A.; Donehower, R. C.; Schilsky, R. L. Clinical pharmacology of methotrexate. Cancer Treat. Rep. 1981, 65(S1), 51-54. 285. Bleyer, W. A. The clinical pharmacology of methotrexate. Cancer 1978, 41, 36-51. 343 286. Henderson, E. S.; Adamson, R. H.; Oliverio, V. T. The metabolic fate of titrated methotrexate II. Absorption and excretion in man. Cancer Res. 1965, 25, 1018-1024. 287. Vincent, J.; Meredith, P.A.; Reid, J.L.; Elliot, H.L.; Rubin, P.C. Clinical pharmacokinetics of prazosin - 1985. Clin. Pharmacokinet. 1985, 10, 144-154. 288. Sorgel, F.; Jaehde, U.; Naber, K.; Stephen, U. Pharmacokinetic disposition of quinolines in human body fluids and tissues. Clin. Pharmacokinet. 1989, 16(1S), 5-24. 289. Ryde, E. M.; Ahnfelt, N. O. The pharmacokinetics of olsalazine sodium in healthy volunteers after a single iv dose and after oral doses with and without food. Eur. J. Clin. Pharmacol. 1988, 34, 481-488. 290. Drusano, G.L.; Standiford, H.C.; Plaisance, K.; Forrest, A.; Leslie, J.; Caldwell, J. Absolute oral bioavailability of ciprofloxacin. Antimicrob. Agents Chemother. 1986, 30, 444-446. 291. Morse, G.D.; Shelton, M.J.; Odonnell, A.M. Comparative pharmacokinetics of antiviral nucleoside analogs. Clin. Pharmacokinet. 1993, 24, 101-123. 292. Regardh, C.G.; Lundborg, P.; Gabrielsson, M.; Heggelund, A.; Kylberghanssen, K. Pharmacokinetics of a single intravenous and oral dose of pafenolol - a beta 1-adrenoceptor antagonist with atypical absorption and disposition properties - in man. Pharm. Res. 1990, 7, 1222-1227. 293. Brater, D C.; Day, B.; Anderson, S.; Seiwell, R. Azosemide kinetics and dynamics. Clin. Pharmacol. Ther. 1983, 34, 454-458. 294. Bastain, W.; Boyce, M.J.; Stafford, L.E.; Morton, P.B.; Clarke, D.A.; Marlow, H.F. Pharmacokinetics of xamoterol after intravenous and oral administration to volunteers. Eur. J. Clin. Pharmacol. 1988, 34, 469-473. 295. Ulm, E.H. Fnalapril maleate (MK-421), a potent, nonsulfhydryl angiotensin-converting enzyme inhibitor: absorption, disposition, and metabolism in man. Drug Metab. Rev. 1983, 14, 99-110. 296. Ranadive, S. A.; Chen, A. X.; Serajuddin, A. T. M. Relative lipophilicities and structural pharmacological considerations of various angiotensin-converting enzyme (ACE) inhibitors. Pharm. Res. 1992, 9, 1480-1483. 297. Macfayden, R.J.; Meredith, P.A.; Elliot, H.L. Fnalapril clinical pharmacokinetics and pharmacokinetic-pharmacodynamic relationships: An overview. Clin. Pharmacokinet. 1993, 25, 274-282. 298. Hellstrom, K.; Rosen, A.; Swahn, A. Absorption and decomposition of potassium - 35 S- phenoxymethyl penicillin. Clin. Pharmacol. Ther. 1974, 16(5, parti), 826-833. 299. Barza, M.; Weinstein, L. Pharmacokinetics of the penicillins in man. Clin. Pharmacokinet. 1976, 1,297-308. 300. Duggan, D.E.; Chen, I.W.; Bayne, W.F.; Halpin, R.A.; Duncan, C.A.; Schwartz, M.S.; Stubbs, R.J.; Vickers, S. The physiological disposition of lovastatin. Drug Metab. Dispos. 1989,17,166- 173. 301. Walker, S. R.; Richards, A. J.; Paterson, J. W. The absorption, excretion and metabolism of disodium [14C] cromoglycate in man. Biochem. J. 1971, 125, 27P. 344 302. Periti, P.; Mazzei, T.; Mini, E.; Novelli, A. Clinical pharmacokinetic properties of the macrolide antibiotics - effects of age and various pathophysiological states. 1. Clin. Pharmacokinet. 1989, 16, 193-214. 303. Sue, Y. J.; Shannon, M. Pharmacokinetics of drugs in overdose. Clin. Pharmacokinet. 1992, 23, 93-105. 304. Curatolo, W. Physical chemical properties of oral drug candidates in the discovery and exploratory development settings. PS7T1998, 1, 387-393. 305. Meylan, W. M.; Howard, P. H.; Boethling, R. S. Improved method for estimating water solubility from octanol water partition coefficient. Environ. Toxicol. Chem. 1996, 15, 100-106. 306. Benet, L. Z.; Wu, C-Y.; Hebert, M. F.; Wacher, V. J. Intestinal drug metabolism and antitransport processes: A potential paradigm shift in oral drug delivery. J. Controlled Release 1996, 2-3, 139- 143. 307. Wacher, V.J.; Silverman, J.A.; Zhang, Y.; Benet, L.Z. Role of P-glycoprotein and cytochrome P450 3A in limiting oral absorption of peptides and peptidomimetics. J. Pharm. Sci. 1998, 87, 1322-1330. 308. Loebstein,R.; Lalkin, A.; Koren, G. Pharmacokinetic changes during pregnancy and their clinical relevance. Clin. Pharmacokinet. 1997, 33, 328-343. 309. Kennard, R. W.; Stone, L. A. Computer aided design of experiments. Technometrics 1969, 11, 137-148. 310. Dreyfuss, J.; Shaw, J. M.; Ross, J. J. Jr. Absorption of the beta-adrenergic-blocking agent, nadolol, by mice, rats, hamsters, rabbits, dogs, monkeys, and man: an unusual species difference. Xenobiotica 1978, 8,503-508. 311. Yoshida, P.; Topliss, J.G. QSAR model for drug human oral bioavailability. J. Med. Chem. 2000,43,2575-2585. 312. Breitkreutz, J. Prediction of intestinal drug absorption properties by three-dimensional solubility parameters. Pharm. Res. 1998, 15, 1370-1375. 313. Sugawara, M.; Takekuma, Y.; Yamada, H.; Kobayashi, M.; Iseki, K.; Miyazaki, K. A general approach for the prediction of the intestinal absorption of drugs: regression analysis using the physicochemical properties and drug-membrane electrostatic interaction. J. Pharm. Sci. 1998, 87, 960-966. 314. Bodin, N.O.; Borg, K.O.; Johansson, R.; Obianwu, H.; Svensson, R. Absorption, distribution and excretion of alprenolol in man, dog and rat. Acta Pharmacol. Toxicol. 1974, 35, 261-269. 315. Michiels, M.; Monbaliu, J.; Hendriks, R.; Geerts, R.; Woestenborghs, R.; Heykants, J. Pharmacokinetics and tissue distribution of the new gastrokinetic agent cisapride in rat, rabbit and dog. Arzneimittelforschung 1987, 37,1159-1167. 316. Cayen, H.N.; Ferdinandi, E.S.; Greselin, E.; Robinson, W.T.; Dvrnik, D. Clofibrate and clofibric acid: comparison of the metabolic disposition in rats and dogs. J. Pharmacol. Exp. Ther. 1977, 200, 33-43. 317. Mathews, J.M; De Costa, K.S. Absorption, metabolism, and disposition of 1,3-diphenyl-1- triazene in rats and mice after oral, I.V., and dermal administration. Drug Metab. Dispos. 1999, 27,1499-1504 345 318. Ghys, A.; Denef, J.; De Suray, J.M.; Gerin, M.; Georges, A.; Delarge, J.; Willems, J. Pharmacological properties of the new potent diuretic torasemide in rats and dogs. Arzneimittelforschung 1985, 35,1520-1526. 319. loannou, Y.M.; Matthews, H.B. Absorption, distribution, metabolism, and excretion of 1,3- diphenylguanidine in the male F344 rat. Fundam. Appl. Toxicol. 1984, 4, 22-29. 320. Beresford, A.P.; MacRae, P.V.; Stopher, D.A. Metabolism of amlodipine in the rat and the dog: A species difference. Xenobiotica 1988, 18,169-182. 321. Meshi, T.; Sato, Y. Studies on sulfamethoxazole-trimethoprim. Absorption, distribution, excretion and metabolism of trimethoprim in rat. Chem. Pharm. Bull. 1972, 20, 2079-2090. 322. Boberg, M.; Ahr, H.J.; Beckermann, B.; Buhner, K.; Siefert, H.M.; Steinke, W.; Wunsche, C.; Hirayama, M. Pharmacokinetics and metabolism of the new thromboxane A2 receptor antagonist ramatroban in animals. 1st communication: absorption, concentrations in plasma, metabolism, and excretion after single administration to rats and dogs. Arzneimittelforschung 1997,47, 928-938. 323. Wang, Y.; Lin, H.; Tullman, R.; Jewell, C.F. Jr; Weetall, M.L.; Tse, F.L.S. Absorption and disposition of a tripeptoid and a tetrapeptide in the rat. Biopharm. Drug Dispos. 1999, 20, 69-75. 324. Boyle, G.W.; McKillop, D.; Phillips, P.J.; Harding, J.R.; Pickford, R.; McCormick, A.D. Metabolism of Casodex in laboratory animals. Xenobiotica 1993, 23, 781-798. 325. Ball, L.M.; Renwick, A.G.; Williams, R.T. The fate of [14C]saccharin in man, rat and rabbit and of 2-sulphamoyl[14C]benzoic acid in the rat. Xenobiotica 1977, 7, 189-203. 326. Cayen, M.N.; Gonzalez, R.; Ferdinandi, E.S.; Greselin, E.; Hicks, D R.; Kraml, M.; Dvornik, D. The metabolic disposition of acifran, a new antihyperlipidemic agent, in rats and dogs. Xenobiotica 1986, 16, 251-263. 327. Burnette, T.C.; de-Miranda, P. Metabolic disposition of the acyclovir prodrug valaciclovir in the rat. Drug Metab. Dispos. 1994, 22, 60-64. 328. Shargel, L.; Dorrbecker, S.A.; Levitt, M. Physiological disposition and metabolism of N-t- butylarterenol and its di-p-toluate ester (bitolterol) in the rat. Drug Metab. Dispos. 1976,4, 65-71. 329. Murakawa, T.; Sakamoto, H.; Fukada, S.; Konishi, T.; Nishida, M. Pharmacokinetics of fosmidomycin, a new phosphonic acid antibiotic, Antimicrob. Agents Chemother. 1982, 21, 224- 230. 330. Uchida, T.; Nakamura, E.; Usui, T.; Imasaki, H.; Kawakami, R.; Watanabe, T.; Higuchi, S. Disposition and metabolism of a novel diurea inhibitor of acyl CoA: cholesterol acyltransferase (YM17E) in the rat and dog. Xenobiotica 1994,24, 1223-1236. 331. Tsuchiya, T.; Ishibashi, K.; Terakawa, M.; Nishiyama, M.; Itoh, N.; Noguchi, H. Pharmacokinetics and metabolism of fosmidomycin, a new phosphonic acid, in rats and dogs. Eur. J. Drug Metab. Pharmacokinet. 1982, 7, 59-64. 332. Schach, V.W.; M.; Schachaavonwittenau, M.; Twomey, T.M.; Swindell, A.C. The disposition of doxycycline by the rat. Chemotherapy 1972,17, 26-39. 333. Schran, H.F.; Tse, F.L.; Bhuta, S.I. Pharmacokinetics and pharmacodynamics of bromocriptine in the rat. Biopharm. Drug Dispos. 1985, 6, 301-311. 334. Dalvie, D.; ODonnell, J. Metabolism of pamaqueside, a cholesterol absorption inhibitor, in Long- Evans rat: Effect of bile duct cannulation on absorption. Xenobiotica 1999, 29, 1043-1056. 346 335. Marten, T.R.; Bourne, G.R.; Miles, G.S.; Shaker, B.; Rankine, H.D.; Dutka, V.N. The metabolism of ICI 118,587, a partial agonist of beta 1-adrenoceptors, in mice, rats, rabbits, dogs, and humans. Drug Metab. Dispos. 1984,12, 652-660. 336. Wagner, O.; Schreier, E.; Heitz, F.; Maurer, G. Tissue distribution, disposition, and metabolism of cyclosporine in rats. Drug Metab. Dispos. 1987, 15, 377-383. 337. Wingen, F.; Schmahl, D. Pharmacokinetics of the osteotropic diphosphonate 3-amino-1- hydroxypropane-1,1 -diphosphonic acid in mammals. Arzneimittelforschung 1987, 37, 1037-1042. 338. Wong, F.A.; Lloyd, J.R.; Graden, D.W. The metabolism of etintidine in rat, dog, and human. Drug Metab. Dispos. 1990, 18, 949-953. 339. Adusumalli. V.E.; Yang, J.T.; Wong, K.K.; Kucharczyk, N.; Sofia, R.D. Felbamate pharmacokinetics in the rat, rabbit, and dog. Drug Metab. Dispos. 1991,19,1116-1125. 340. Meuldermans, W.; Hurkmans, R.; Swysen, E.; Hendrickx, J.; Thijssen, J.; Lauwers, W.; Heykants, J. Excretion and metabolism of lorcainide in rats, dogs and man. Eur. J. Drug Metab. Pharmacokinet. 1983, 8, 335-349. 341. Girkin, R.; Baldock, G.A.; Chasseaud, L.F.; Humpel, M.; Hawkins, D.R.; Mayo, B.C. The absorption, distribution and excretion of [(14)C] lormetazepam in dogs, rabbit, rats and rhesus monkeys. Xenobiotica 1980, 10, 401-411. 342. Tokuma, Y.; Fujiwara, T.; Noguchi, H. Absorption, distribution and excretion of nilvadipine, a new dihydropyridine calcium antagonist, in rats and dogs. Xenobiotica 1987, 17,1341-1349. 343. Hayes, A.; Cooper, R.G. Studies on the absorption, distribution and excretion of propranolol in rat, dog and monkey J. Pharmacol. Exp. Ther. 1971,176, 302-311. 344. Si tar, D.S.; Thornhill, D P. Propylthiouracil; absorption, metabolism and excretion in the albino rat. J. Pharmacol. Exp. Ther. 1972, 183,440-448. 345. Loh, A.C.; Szuna, A.J.; Williams, T.H.; Sasso, G.J.; Leinweber, F.J. The metabolism of [(14)C] rimantadine hydrochloride in rats and dogs. Drug Metab. Dispos. 1991, 19, 381-387. 346. McMahon, T.F.; Diliberto, J.J.; Birnbaum, L.S. Effects of age and dose on disposition and metabolism of salicylic acid in male Fischer 344 rats. Drug Metab. Dispos. 1990, 18,494-503. 347. Wood, B.A.; Rycroft, D.; Monro, A.M. The metabolism of tinidazole in the rat and dog. Xenobiotica 1973, 3, 801-812. 348. Mroszczak, E.J.; Lee, F.W. Tiopinac absorption, distribution, excretion, and pharmacokinetics in man and animals. Drug Metab. Dispos. 1980, 8, 415-421. 349. Iwamoto, K.; Takei, M.; Watanabe, J. Gastrointestinal and hepatic first-pass metabolism of aspirin in rats. J. Pharm. Pharmacol. 1982, 34,176-180. 350. Trnovec, T.; Zemanek, M.; Faberova, V.; Bezek, S.; Durisova, M.; Ujhazy, E.; Tomcikova, O. Disposition of exaprolol, a new beta-blocker, in rats. Drug Metab. Dispos. 1982, 10, 547-550. 351. Howell, S.R.; Husbands, G.E.; Scatina, J.A.; Sisenwine, S.F. Metabolic disposition of 14C- venlafaxine in mouse, rat, dog, rhesus monkey and man. Xenobiotica 1993, 23, 349-359. 352. Case, D.E.; Illston, H.; Reeves, P.R.; Shuker, B.; Simons, P. The disposition and metabolism of I.e.I. 58,834 (viloxazine) in animals. Xenobiotica 1975, 5, 83-111. 353. Morton, D.M. Pharmacology and toxicology of nizatidine. Scand. J. Gastroenterol. 1987, 136(Suppl), 1-8. 347 354. Jang, S.H.; Lee, M.H.; Lee, M.G. Pharmacokinetics of acetaminophen after intravenous and oral administration to spontaneously hypertensive rats and normotensive Wistar rats. J. Pharm. Sci. 1994, 83,810-814. 355. Greenslade, D.; Havler, M.E.; Humphrey, M.J.; Jordan, B.J.; Ranee, M.J. Species differences in the metabolism and excretion of fenclofenac. Xenobiotica 1980, 10, 753-760. 356. Scatina, J.A.; Lockhead, S.R.; Cayen, M.N.; Sisenwine, S.F. Metabolic disposition of enciprazine, a non-benzodiazepine anxiolytic drug, in rat, dog and man. Xenobiotica 1991,21,1591-1604. 357. Smith, D.A.; Rasmussen, H.S.; Stopher, D.A.; Walker, D.K. Pharmacokinetics and metabolism of dofetilide in mouse, rat, dog and man. Xenobiotica 1992,22, 709-719. 358. Regardh, C.G.; Gabrielsson, M.; Hoffman, K.J.; Lofberg, I.; Skanberg, I. Pharmacokinetics and metabolism of omeprazole in animals and man - an overview. Scand. J. Gastroenterol. 1985, 20(Suppl. 108), 79-94. 359. Mroszczak, E.J.; Lee, F.W.; Combs, D.; Sarnquist, F.H.; Huang, B.L.; Wu, A.T.; Tokes, L.G.; Maddox, M.L.; Cho, D.K. Ketorolac tromethamine absorption, distribution, metabolism, excretion, and pharmacokinetics in animals and humans. Drug Metab. Dispos. 1987, 15, 618-626. 360. Weidolf, L. Bimodal column switching liquid chromatographic assay of six metabolites of [14C] felodipine in rat urine. J. Chromatogr. 1985, 343, 85-97. 361. Sutfm-TA; Gabrielsson-M; Regardh-CG. Effects of biliary excretion on the disposition of felodipine and metabolites in the rat. Xenobiotica 1987, 17, 1203-1214. 362. Bezek, S.; Scasnar, V.; Trnovec, T.; Durisova, M.; Faberova, V.; Benes, L. The disposition of the local anaesthetic, pentacaine, in rats and mice. Biopharm. Drug Dispos. 1986, 7,137-150. 363. Andoh, B.Y.; Renwick, A.G.; Williams, R.T. The excretion of (35S)dapsone and its metabolites in the urine, faeces and bile of the rat. Xenobiotica 1974,4, 571-583. 364. Nilsson, H.T.; Persson. C.G.; Persson, K.; Tegner, K.; Ryrfeldt, A. The metabolism of terbutaline in dog and rat. Xenobiotica 1973, 3, 615-623. 365. Schaefer, W.H.; Politowski, J.; Hwang, B.; Dixon, F. Jr; Goal win. A.; Gutzait, L.; Anderson, K.; DeBrosse, C.; Bean, M.; Rhodes, G.R. Metabolism of carvedilol in dogs, rats, and mice. Drug Metab. Dispos. 1998,26, 958-969. 366. Illing, H.P.; Ings, R.M.; Johnson, K.I.; Fromson, J.M. Disposition of 14C-loprazolam in animals and man. Xenobiotica 1983, 13, 439-449. 367. Marathe, P H.; Greene, D.S.; Barbhaiya, R.H. Disposition of [14C] avitriptan in rats and humans. Drug Metab. Dispos. 1997, 25, 881-888. 368. Reeves, P.R.; Barnfield, D.J.; Longshaw, S.; McIntosh, D.A.D.; Winrow, M.J. Disposition and metabolism of aterolol in animals. Xenobiotica 1878, 8, 305-311. 369. Belpaire, F.M.; de Smet, F.; Vynckier, L.J.; Vermeulen, A.M.; Rosseel, M.T.; Bogaert, M.G.; Chauvelot Moachon, L. Effect of aging on the pharmcokinetics of atenolol, metoprolol and propranolol in the rat. J. Pharmacol. Exp. Ther. 1990, 254, 116-122. 370. Vickers, S.; Duncan, C.A.; Slaughter, D.E.; Arison, B.H.; Greber, T.; Olah, T.V.; Vyas, K.P. Metabolism of MK-499, a class III antiarrhythmic agent, in rats and dogs. Drug Metab. Dispos. 1998,26, 388-395. 348 371. Cocchiara, G.; Battaglia, R.; Fontana, E.; Benedetti, M.S. Metabolic fate of FCE 22716, a new antihypertensive agent, and of its N-oxide (FCE 24220) in the rat. Drug Metab. Drug Interact. 1992,10, 185-197. 372. Prakash, C.; Kamel, A.; Anderson, W.; Howard, H., Metabolism and excretion of the novel antipsychotic drug ziprasidone in rats after oral administration of a mixture of 14C- and 3H- labeled ziprasidone. Drug Metab. Dispos. 1997, 25, 206-218. 373. Kolis, S.J.; Williams, T.H.; Schwartz, M.A. Identification of the urinary metabolites of 14C- bumetanide in the rat and their excretion by rats and dogs. Drug Metab. Dispos. 1976,4, 169-76, 374. Yamaguchi, T.; Ikeda, C.; Sekine, Y. Intestinal absorption of a beta-adrenergic blocking agent nadolol. Enhancement of in situ and in vivo absorption of nadolol in rats. Int. J. Pharm. 1987, 37, 127-134. 375. Niebch, G.; Obermeier, K.; Vergin, H.; Thiemer, K. Studies on pharmacokinetics and biotransformation of reproterol in animal and man (author’s transi). Arznemittelforschung 1977, 27, 37-45. 376. Lin, J.H.; Duggan, D.E.; Chen, I.W.; Ellsworth, R.L. Physiological disposition of alendronate, a potent anti-osteolytic bisphosphonate, in laboratory animals. Drug Metab. Dispos. 1991, 19, 926- 932. 377. Yeager, R.L.; Brouwer, K.R.; Miwa, G.T, (-)-6-Aminocarbovir pharmacokinetics and relative carbovir exposure in rats. Drug Metab. Dispos. 1991, 19, 462-466. 378. Coppi, G.; Silingardi, S. Pharmacokinetics of pidotimod in rats and dogs. Arzneimittelforschung 1994,44(12A),1460-1464. 379. Kobayashi, H.; Kobayashi, S. Absorption, distribution, metabolism and excretion of 14C-labelled enantiomers of the calcium channel blocker benidipine after oral administration to rat. Xenobiotica 1998, 28, 179-197. 380. Dallas, F.A.; Dixon, C.M.; McCulloch, R.J.; Saynor, D.A. The kinetics of 14C-GR43175 in rat and dog. Cephalagia 1989,9(Suppl. 9),53-56. 381. Segura, J.; Borja, L.; Bakke, O.M. Pharmacokinetics of sulpiride after oral and intravenous administration in the rat and the dog. Arch. Int. Pharmacodyn. Ther. 1976, 223-224, 88-95. 382. Walsh, J.S.; Patanella, I.E.; Unger, S.E.; Brouwer, K.R.; Miwa, G.T. The metabolism and excretion of carbovir, a carbocyclic nucleoside, in the rat. Drug Metab. Dispos. 1990, 18, 1084- 1091. 383. Lennernas, H.; Regardh, C.G. Dose-dependent intestinal absorption and significant intestinal excretion (exsorption) of the beta-blocker pafenolol in the rat. Pharm. Res. 1993,10, 727-31. 384. Hildebrand, M.; Krause, W.; Kuhne, G.; Hoyer, G.A. Pharmacokinetics and metabolism of mespirenone, a new aldosterone antagonist, in rat and cynomolgus monkey. Xenobiotica 1987, 17, 623-634. 385. Cook, C.S.; Campion, J.G.; Hribar, J.D;. Karim, A. Metabolism and pharmacokinetics of nufenoxole in animals and humans: An example of stereospecific hydroxylation of an isoquinuclidine ring. Xenobiotica. 1990, 20,1065-1080. 349 386. Vickers, S.; Stuart, E.K.; Bianchine, J.R.; Hucker, H.B.; Jaffa, M.E.; Rhodes, R.E.; Vandenheuvel, W.J. Metabolism of carbidopa (l-(-)-aIpha-hydrazino-3,4-dihydroxy-a!pha- methylhydrocinnamic acid monohydrate), an aromatic amino acid decarboxylase inhibitor, in the rat, rhesus monkey, and man. Drug Metab. Dispos. 1974, 2, 9-22. 387. Krause, W.; Schulze, P.E.; Totzek, M. Pharmacokinetics of the stable prostacyclin analogue, nileprost, in the rat. Prostaglandins Leukotrienes Med. 1983, 11, 241-257. 388. Larsen, J.C.; Tarding, F. Absorption and excretion of l-amino-2-naphthol-6-sulphonic acid in rats and rabbits. ArcA. Toxicol. 1978,(Suppl. 1),251-254. 389. Sipila, H.; Kangas, L.; Vuorilehto, L.; Kalapudas, A.; Eloranta, M.; Sodervall, M.; Toi vola, R.; Anttila, M. Metabolism of toremifene in the rat. J. Steroid. Biochem. 1990, 36,211-215. 390. Winstanley, P.A.; Edwards, G.; Curtis, C.G.; Orme, M.L.; Powell, G.M.; Breckenridge, A.M. Tissue distribution and excretion of amodiaquine in the rat. J. Pharm. Pharmacol. 1988, 40, 343- 349. 391. Taylor, J.A.; Twomey, T.M.; von Wittenau, M.S. The metabolic fate of prazosin. Xenobiotica 1977,7, 357-364. 392. Patel, D.K.; Woolley Jr, J.L.; Shockcor, J.P.; Johnson, R.L.; Taylor, L.C.; Sigel, C.W. Disposition, metabolism, and excretion of the anticancer agent crisnatol in the rat. Drug Metab. Dispos. 1991, 19,491-497. 393. Karim, A.; Kook, C.; Zitzewitz, D.J.; Zagarella, J.; Doherty, M.; Campion, J. Species differences in the metabolism and disposition of spironolactone. Drug Metab. Dispos. 1976, 4(6), 547-555. 394. Hilderbrand, M.; Humpel, M.; Krause, W.; Tauber, U. Pharmacokinetics of bromerguride, a new dopamine-antagonistic ergot derivative in rat and dog. Eur. J. Drug Metab. Pharmacokinet. 1987, 12, 31-40. 395. Pellegatti, M.; Grossi, P.; Ayrton, J.; Evans, G.L.; Harker, A.J. Absorption, distribution and excretion of lacidipine, a dihydropyridine calcium antagonist, in rat and dog. Xenobiotica 1990, 20,765-777. 396. Zini, G.; Vicario, G.P.; Lazzati, M.; Arcamone, P. Disposition and metabolism of [14-14C] 4- demethoxydaunorubicin HCl (idarubicin) and [14-14C] daunorubicin HCl in the rat. A comparative study. Cancer Chemother. Pharmacol. 1986, 16,107-115. 397. Lan, S.J.; Hsieh, D C.; Hillyer, J.W.; Fancher, R.M.; Rinehart, K.J.; Warrack, B.M.; White, R.E. Metabolism of alpha-phosphonosulfonate squalene synthase inhibitors; I. Disposition of a farnesylethyl alpha-phosphonosulfonate and ester prodrugs in rats. Drug Metab. Dispos. 1998, 26, 993-1000. 398. de Miranda, P.; Krasny, H.C.; Page, D.A.; Elion, G.B. The disposition of acyclovir in different species. J. Pharmacol. Exp. Ther. 1981, 319, 309-315. 399. Willems, A.G.; Overmars, H.; Scherpenisse, P.; De Lange, N.; Post, L.C. Diflubenzuron: intestinal absorption and metabolism in the rat. Xeuobionca 1980, 10, 103-112. 400. Eddershaw, P.J.; Chadwick, A.P.; Higton, D.M.; Fenwick, S.H.; Linacre, P.; Jenner,, W.N.; Bell, J.A.; Manchee, G.R. Absorption and disposition of ranitidine hydrochloride in rat and dog. Xenobiotica 1996, 26, 947-956. 350 401. Hucker, H.B.; Ballettto, A.J.; White, S.D.; Zacchei, A.G. The disposition and metabolism of an orexigenic agent, 10,1 l-dihydro-5-(l-methyl-4-piperidinylidene)-5H-dibenzo[a,d] cyclohepten-3- c arboxylic acid, in the rat, dog, and cat. Drug Metab. Dispos. 1981,9,428-433. 402. Thomas, D.A.; Rosenthal, G.A. Metabolism of L-[guanidinooxy-(14)C]canavanine in the rat. Toxicol. Appl. Pharmacol. 1987, 91, 406-414. 403. Grislain, L.; Mocquard, M.T.; Dabe, J.F.; Bertrand, M.; Luijten Marchand, B.; Resplandy, G.; Devissaguet, M. Interspecies comparison of the metabolic pathways of perindopril, a new angiotensin-converting enzyme (ACE) inhibitor. Xenobiotica 1990, 20, 787-800. 404. Eckert, H.G.; Badian, M.J.; Gantz, D.; Kellner, H.M.; Volz, M. Pharmacokinetics and biotransformation of 2-[N-[(S)-l-ethoxycarbonyl-3-phenylpropyl]-L-alanyl]-(lS, 3S, 5S)-2- azabicyclo [3.3.0]octane-3-carboxylic acid (Hoe 498) in rat, dog and man. Arzneimittelforschung 1984, 34,1435-1447. 405. Okuyama, Y.; Momota, K.; Morino, A. Pharmacokinetics of prulifloxacin. 1st communication; absorption, distribution and excretion in rats, dogs and monkeys after a single administration. Arzneimithel-forschung 1997, 47, 276-284. 406. Pudleiner, P.; Vereczkey, L. Study on the absorption of vinpocetine and apovincaminic acid. Eur.J. Drug Metab. Pharmacokinet. 1993, 18, 317-321, 407. Eriksson, U.G.; Renberg, L.; Bredberg, U.; Teger Nilsson, A.C.; Regardh, C.G. Animal pharmacokinetics of inogatran, a low-molecular-weight thrombin inhibitor with potential use as an antithrombotic drug. Biopharm. Drug Dispos. 1998, 19, 55-64. 408. Egger, H.; Kochak, G.; Robertson, P.; lannucci, R.; Rufino, F.A.; Stancato, F. Physiological disposition of CGS 16617 in rat, dog, and man. Drug Metab. Dispos. 1989, 17, 669-672. 409. Ramsay, C.H.; Bodin, N.O.; Hansson, E. Absorption, distribution, biotransformation, and excretion of azidocillin, a new semi-synthetic penicillin, in mice, rats and dogs Arzneimittelforschung 1972, 22, 1962-1969. 410. Barrow, A.; Brownsill, R.D.; Spalton, P.M.; Walls, C.M.; Gunn, Y.; Haskins, N.J.; Rose, D.A.; Palmer, R.F. Disposition of pranolium chloride in small mammals. Xenobiotica 1980, 10, 219- 228. 411. Wahl, D.; Forster, H.J.; Pook, K.H.; Richter, I. Biochemical studies with oxitropium bromide. 1. Pharmacokinetics and metabolism in the rat and dog. Arzneimittelforschung 1985, 35, 255-265. 412. Forster, H.J.; Kramer, I.; Pook, K.H.; Wahl, D. Studies on the pharmacokinetics and biotransformation of ipratropium bromide in the rat and dog. Arzneimittelforschung 1976, 26, 992-1005. 413. Prueksaritanont, T.; Lee, M.G.; Hsu, F.H.; Chiou, W.L. Absorption of iothalamate after oral administration and absorption enhancement by amino acids in dogs and rats. Biopharm. Drug Dispos. 1986, 7, 463-478. 414. Starrett, J.E. Jr; Tortolani, DR.; Russell, J.; Hitchcock, M. J.; Whiterock, V.; Martin, J.C.; Mansuri, M.M. Synthesis, oral bioavailability determination, and in vitro evaluation of prodrugs of the antiviral agent 9-[2-(phosphonomethoxy)ethyl]adenine (PMEA). J. Med. Chem. 1994, 37, 1857-1864. 351 415. Faulkner, R.D.; Fernandez, P.; Lawrence, G.; Sia, L.L.; Falkowski, A.J.; Weiss, A.I.; Yacobi, A.; Silber, B.M. Absolute bioavailability of cefixime in man. J. Clin. Pharmacol. 1988, 28, 700-706. 416. Flesch, G.; Muller, P.; Lloyd, P. Absolute bioavailability and pharmacokinetics of valsartan, an angiotensin II receptor antagonist, in man. Eur. J. Clin Pharmacol. 1997, 52, 115-120. 417. Chiou, W.L.; Jeong, H.Y.; Chung, S.M.; Wu, T.C. Evaluation of using dog as an animal model to study the fraction of oral dose absorbed of 43 drugs in humans. Pharm. Res. 2000,17, 135-140. 418. Dressman J. B.; Amidon, G. L.; Reppas, C.; Shah, V.P. Dissolution testing as a prognotic tool for oral drug absorption; immediate release dosage forms. Pharm. Res. 1998, 15, 11-22. 419. Brodie, B.B. in Absorption and distribution of drugs, Binns,T.B., Ed.; E.& S. Livingstone, Edinburgh, 1964. 420. Gumbleton, M.; Forbes, B.J. Principles in the absorption, distribution and elimination of pharmaceuticals. Pestic.Sci. 1994, 42, 223-240. 421. Waterbeemd, J.Th.M. Van de; Van Boekel, C.A.A.; De Sevaux, R.L.F.M.; Jansen, A.C.A.; Gerritsma,K.W. Transport in QSAR. IV. The interfacial drug transfer model. Relationships between partition coefficients and rate constants of drug partitioning. Pharm. Weekblad. Sci.Ed. 1981, 3, 224- 236. 422. Waterbeemd, J.Th.M. van de; Jansen,A.C.A. Transport in QSAR. V. Applications of the interfacial drug transport model. Pharm. Weekblad. Sci.Ed. 1981, 3, 587-594. 423. Leahy, D.E.; De Mere, A.L.J.; Wait, A.R.; Taylor, P.J.; Tomenson, J.A.; Tomlinson, E. A general description of water-oil partitioning rates using the rotating diffusion cell. Int. J.Pharm. 1989, 50, 117-132. 424. Larhed,W.A.; Artursson,P.; Grasjo,J.; Bjork,E. Diffusion of drugs in native and purified gastrointestinal mucus. J.Pharm.Sci. 1997, 86, 660-665. 425. Abraham,M.H.; Green, C.E. unpublished work. 426. Chen,N.; Chan,T.C. Experimental study of hydrogen bonding by mutual diffusion. Chem.Commun. 1997, 7, 719-720. 427. Abraham,M.H.; Chadha,H.S.; Whiting,G.S.; Mitchell,R.C. Hydrogen bonding. 32. An analysis of water-octanol and water-alkane partitioning, and the AlogP parameter of Seiler. J.Pharm.Sci. 1994, 83, 1085-1100. 428. Abraham, M.H.; Platts,J.A.; Hersey,A.; Leo,A.J.; Taft,R.W. Correlation and estimation of gas- chloroform and water-chloroform partition coefficients by a linear free energy relationship. J.Pharm.Sci. 1999,88, 670-679. 429. Platts,J.A.; Abraham,M.H. The partition of volatile organic compounds from air and from water into plant cuticular matrix: an LEER analysis. Env.Sci.Technol. 2000, 34, 318-323. 430. Abraham,M.H.; Weathersby,P.K. Hydrogen bonding. 30. Solubility of gases and vapors in biological liquids and tissues. J.Pharm.Sci. 1994, 83, 1450-1456. 431. Abraham,M.H.; Martins,F.; Mitchell,R.C. Algorithms for skin permeability using hydrogen bond descriptors : the problem of steroids. J.Pharm.Pharmacol. 1997, 49, 858-865. 432. Platts,J.A.; Abraham,M.H.; Hersey,A.; Butina,D. Estimation of molecular free energy relationship descriptors. 4. Correlation and prediction of cell permeation. Pharm.Res. in the press. 352 433. Johnson,M.E.; Blankschtein,D.; Langer,R. Evaluation of solute permeation through the stratum corneum:Lateral bilayer diffusion as the primary transport mechanism. J.Pharm.Sci. 1997, 86, 1162- 1172. 434. Mitragotri,S.; Johnson,M.E.; Blankschtein,D.; Langer,R. an analysis of the size selectivity of solute partitioning, diffusion, and permeation across lipid bilayers. Biophys.J. 1999, 77, 1268- 1283. 435. Personal communication from Professor Samir Mitragotri. 436. Morishita,T.; Yamazaki,M.; Yata,N.; Kamada,A. Studies on the absorption of drugs. VIII. Physicochemical factors affecting the absprption of sulfonamides from the rat small intestine. Chem.Pharm.Bull. 1973, 27, 2309-2322. 437. Ishihama,Y.; Asakawa,N. Characterization of lipophilicity scales using vectors from solvation energy descriptors. J.Pharm.Sci. 2000, 8, 1305-1312. 438. Sammon, J.W.Jr. A nonlinear mapping for data structure analysis. IEEE Trans on computers, 1969, C-18,4401-4407. 439. Du, C.M.; Valko,K.; Bevan,C.; Reynolds,D.; Abraham,M.H. Characterizing the selectivity of stationary phases and organic modifiers in reversed phase - high perforrmance liquid chromatography by the general solvation equation using gradient elution. J.Chromatogr.Sci. in the press. 440. Green, C.E.; Abraham, M.H. Investigation into the effects of temperature and stirring rate on the solid-phase extraction of diuron from water using a Cig extraction disk. J. Chromatogr. A. 2000, 885,41-49. 441. Green, C.E. An experimental and modelling investigation into the Solid-Phase Extraction of pollutants from water. University College London, PhD thesis, 2000. 353 Chapter 8 Visualisation of the Abraham General Solvation Equation The Abraham General Solvation Equation has utility in drug design. A simple example is given in Table 8.1 where the method has been applied to predict aqueous solubility and human intestinal drug absorption for a series of analogues using equations derived in Chapters 6.3.2 and 7.2.2 respectively; log Sw (mol L-') = 0.394 - 0.954 Rz +0.318 Jtj" + 1.157 Saj" + 3.255 Epj” -0.786 Sai" X Zg:" - 3.329 V, (33) %Abs. = 92 + 2.94R2+4.107t2” - 21.7ZÜ2"- 2 I.IZP2” + 10.6Vx (34) Table 8.1 Calculated change in aqueous solubility and human intestinal absorption with respect to change in substituent group Parent Substituent, R log Sw %Abs OH -H -3.22 65 H i -OH -3.05 50 -NH2 -3.16 58 OH -CONH2 -3.23 51 NH -NO2 -3.53 67 -C2H5 -4.17 68 -Cyclohexyl -5.36 72 -Phenyl -5.60 74 The method is made conceptually more user-friendly by visualising the Abraham General Solvation Equation in two dimensions by separating the five variable model into dominantly polar and non-polar components : Polar terms SP = c + r.R2 + S.7C2” + a.Sa2” +6 .2 ^2" + v.Vx Non-polar terms 354 The sum of the polar eomponents [E | ( 5712” + I ] vs the sum of the non polar components [EI (rRi + vVx) | ] for each compound in a given data set may be plotted. The SP value for each data point is then used to colour code the respective point to represent a specific level of the dependent variable, ie. low, medium, or high value. An example is given for aqueous solubility which is shown in Fig.8.1. For aqueous solubility, there is an additional cross-term such that: y - co-ordinates given by:[ I (0.318712^ + 1.157Ea2^ + 3.255EP2*^-0.786(Ea2”x E P 2 " ) I ] X - co-ordinates given by:[ I (-0.954R2 - 3.329Vx) | ] Fig. 8.1 Plot of polar vs non-polar components for 1071 compounds which are colour coded according to whether log Sw is low, medium, or high. N o n -p o la r • low( log Sw<-3.5) a medium (log S w -3.5 to -1) ■ high log (Sw>-1) Fig. 8.1 shows that there is a general separation of low, medium and high values according to the relative contribution of the polar and non-polar components to solubility. Block colouring of the separation of these categories gives a clear indication of the fundamental physicochemical parameters that either promote or hinder the given process. This visualisation procedure has been conducted for both the intestinal absorption and water solubility equations. The result. Fig 8.2 shows explicitly that an inverse relationship exists, viz. 355 • Polar components promote solubility, non-polar eomponents hinder solubility • Polar components hinder intestinal absorption, non-polar components promote absorption. The substituent changes with respeet to the parent compound shown in Table 8.1 could therefore be translated onto Fig. 8.2 by means of arrows; the direction and magnitude of whieh would reflect the change in polar and non-polar components, and hence the change in SP. This visualisation step can easily be applied to any Abraham Solvation Equation. If the overall process could be computerised, it is envisaged that the procedure could be useful as an aid to guide synthetic chemists in choosing new chemical entities, not with just optimal biological activity, but also with preferable physicochemieal properties. Fig. 8.2 Visualisation diagrams for human intestinal absorption and aqueous solubility Aqueous Solubility Intestinal Absorption 200 0-33% non-polar 34-66% 67-100% 8 10 12 14 16 10 30 50 70 N on-polar This concept is chemically intuitive and Fig. 8.2 clearly illustrates that foeusing on improving only one avenue of lead development may have a detrimental effect on another. Thus, whilst the aim is often to improve lipophilicity and aid compound entry into cells, consideration must also be given to dissolution which is often the rate- limiting step in absorption and inherently dependent on solubility. 356 Chapter 9 Conclusions and Suggestions for Future Work This work has successfully yielded interpretable algorithms for a number of processes which will briefly be presented in turn. The solubility of vapours in a homologous series of n-alkanols has provided gas- solvent partition equations which have subsequently yielded indirect water-solvent partition equations. The equation coefficients indicate that characteristically, the alcohols have considerable hydrogen-bond basicity and moderate hydrogen-bond acidity. Contrary to other measures of solvent hydrogen-bond basicity, there is an inference that water and alcohols have almost identical hydrogen-bond basicity. All- in-all, the solvation of solutes in the alcohols more resembles solvation in nonaqueous polar solvents than solvation in water. This work has largely been completed and the equations are generally satisfactory. Nevertheless, if new data is made available, especially in the case of hexan-l-ol, heptan-l-ol and decan-l-ol, the respective equations may be updated, although it is unlikely that the equation coefficients will alter substantially. The main focus of future work would be to obtain an LFER equation for nonan-l-ol as solvent; an objective which cannot be met at present due to lack of data. Once this aim is achieved, the series of n-alkanol equations will be complete, and in addition to predictive properties, may be utilised in compound descriptor determination. Gas-solvent and water-solvent partition equations have been found to be effective in assigning solute descriptors, given that partition measurements are available in a sufficient number of solvent systems. Unfortunately, the implementation of more conveniently obtained reversed phase HPLC capacity factors as an alternative was inadequate to obtain all the required descriptors. The problem lay in the similarity between the equation coefficients obtained, irrespective of the column stationary phase employed. The r-, s- and ^-coefficients encoding the difference in n-n interactions, dipolarity/polarizability, and hydrogen-bond basicity between the mobile phase and the stationary phase were all small. The most significant coefficients were b and V corresponding to the difference in hydrogen-bond acidity and cohesiveness and dispersion of the stationary and mobile phases respectively. However, even b and v were of smaller magnitude than the corresponding coefficients in the water-solvent 357 partition measurements. The result is that descriptor prediction using this method is impaired by the high relative error inherently associated with the HPLC equation coefficients. The conclusion, is that the use of HPLC in conjunction with a water- solvent partition system would likely provide the best compromise in terms of accuracy and efficiency of descriptor determination. The Abraham General Solvation Equation had to be amended so that an additional X term accounting for compound crystallinity could be incorporated to deal with prediction of water solubility. The LFER technique was then successfully applied to a training set of 803 diverse compounds, yielding an equation which enabled prediction of aqueous solubility 0.67 log units for a test set of 268 compounds. From comparison with other available prediction methods, the results were certainly promising and the Abraham Method was found to be the most accurate in prediction for the given test set. However, the training set may certainly be improved by future addition of new solubility data for larger compounds i.e. Vx>3.5, which are at present lacking. Thus, the limitation of this method at present is that predictions may only be made with some degree of confidence for compounds for which Vx<4. Equations containing both the cross-term and also a melting point term were also derived and are in general slightly more accurate in prediction. The obvious disadvantage is that a reliable melting point value must be assigned before the prediction can be made. Using an extensively evaluated set of literature absorption data for both humans and rat, it was possible to derive equations for prediction of intestinal absorption. From the analysis of data, it may generally be assumed that the extent of drug absorption in vivo is the same irrespective of whether it is administered orally to rats or humans. However, there were three notable discrepancies and the existence of such anomalies must be borne in mind when using rat data to predict oral absorption in humans. The equations indicated that hydrogen-bonding terms were detrimental to absorption, whilst solute volume promoted the process. Ionization was shown to have minimal influence on the extent of absorption and may be attributed to the ability of ionized compounds to cross the intestinal membrane. Transformation of the % absorption data to overall rate constants allowed comparison of the equations to available water-solvent and diffusion system equations. Statistical 358 analysis revealed that the intestinal absorption equations are chemically most similar to the mainly diffusion processes of solute uptake onto a Cig disk and limiting diffusion of solutes in ethanol. Experimental measurement of the rate of uptake, kup, onto a Ci8 disk for eight drug compounds has been performed. However, far more drug data will need to be determined if the relationship between the rate of absorption and logkup is revealed with any confidence. It would be of interest to investigate whether the Waterbeemd tripartitite curve of logkup vs log P for this system exists and this could certainly form the basis of future work. Finally, by separating the five variable model into dominantly polar (& 712", ar. Eaz", 6 .ZP2") and non-polar (r.R 2, v.Vx) components, it is possible to visualise the Abraham General Solvation Equation in two dimensions. This visualisation concept could be developed as a tool to aid synthetic chemists in isolating compounds with optimal properties. A computer program could be introduced which would enable a chemist to observe the effect of introducing a particular substituent onto a parent compound (P). After selection of a substituent, descriptors would be calculated for the new chemical entity (NCE) and the resulting polar vs non-polar co-ordinate plotted on the property of interest. The direction of movement of the new entity with regard to the parent compound would indicate the influence of the substituent on the given property. This is illustrated taking intestinal absorption as the property of interest, where L, M and H represent low, medium, and high absorption respectively. A bsorption 200 0-33% (L) NCE? 34-66%(M) 67-lGG%(H) 30 50 Non-polar 359