Multivariate assessment of lipophilicity scales-computational and reversed phase thin-layer chromatographic indices

Filip Andrić, Dávid Bajusz, Anita Rácz, Sandra Šegan, K. Heberger

Research output: Contribution to journalArticle

26 Citations (Scopus)

Abstract

Needs for fast, yet reliable means of assessing the lipophilicities of diverse compounds resulted in the development of various in silico and chromatographic approaches that are faster, cheaper, and greener compared to the traditional shake-flask method. However, at present no accepted "standard" approach exists for their comparison and selection of the most appropriate one(s). This is of utmost importance when it comes to the development of new lipophilicity indices, or the assessment of the lipophilicity of newly synthesized compounds. In this study, 50 well-known, diverse compounds of significant pharmaceutical and environmental importance have been selected and examined. Octanol-water partition coefficients have been measured with the shake-flask method for most of them. Their retentions have been studied in typical reversed thin-layer chromatographic systems, involving the most frequently employed stationary phases (octadecyl- and cyano-modified silica), and acetonitrile and methanol as mobile phase constituents. Twelve computationally estimated logP-s and twenty chromatographic indices together with the shake-flask octanol-water partition coefficient have been investigated with classical chemometric approaches-such as principal component analysis (PCA), hierarchical cluster analysis (HCA), Pearson's and Spearman's correlation matrices, as well as novel non-parametric methods: sum of ranking differences (SRD) and generalized pairwise correlation method (GPCM). Novel SRD and GPCM methods have been introduced based on the Comparisons with One VAriable (lipophilicity metric) at a Time (COVAT). For the visualization of COVAT results, a heatmap format was introduced. Analysis of variance (ANOVA) was applied to reveal the dominant factors between computational logPs and various chromatographic measures. In consensus-based comparisons, the shake-flask method performed the best, closely followed by computational estimates, while the chromatographic estimates often overlap with in silico assessments, mostly with methods involving octadecyl-modified silica stationary phases. The ones that employ cyano-modified silica perform generally worse. The introduction of alternative coloring schemes for the covariance matrices and SRD/GPCM heatmaps enables the discovery of intrinsic relationships among lipophilicity scales and the selection of best/worst measures. Closest to the recommended logK OW values are ClogP and the first principal component scores obtained on octadecyl-silica stationary phase in combination with methanol-water mobile phase, while the usage of slopes derived from Soczewinski-Matyisik equation should be avoided.

Original languageEnglish
JournalJournal of Pharmaceutical and Biomedical Analysis
DOIs
Publication statusAccepted/In press - Oct 29 2015

Fingerprint

Silicon Dioxide
Correlation methods
Octanols
Methanol
Water
Cluster analysis
Coloring
Analysis of variance (ANOVA)
Covariance matrix
Principal component analysis
Computer Simulation
Visualization
Pharmaceutical Preparations
Principal Component Analysis
Cluster Analysis
Analysis of Variance

Keywords

  • Benzodiazepines
  • Comparison with one variable at a time-COVAT
  • Generalized pairwise correlation method-GPCM
  • Heatmap
  • Lipophilicity
  • Phenols
  • Polyaromatic hydrocarbons
  • Reversed-phase thin-layer chromatography
  • Sum of ranking differences-SRD

ASJC Scopus subject areas

  • Analytical Chemistry
  • Drug Discovery
  • Pharmaceutical Science
  • Spectroscopy
  • Clinical Biochemistry

Cite this

Multivariate assessment of lipophilicity scales-computational and reversed phase thin-layer chromatographic indices. / Andrić, Filip; Bajusz, Dávid; Rácz, Anita; Šegan, Sandra; Heberger, K.

In: Journal of Pharmaceutical and Biomedical Analysis, 29.10.2015.

Research output: Contribution to journalArticle

@article{9eadbb29b8d64e7f9629d2a39ee4c056,
title = "Multivariate assessment of lipophilicity scales-computational and reversed phase thin-layer chromatographic indices",
abstract = "Needs for fast, yet reliable means of assessing the lipophilicities of diverse compounds resulted in the development of various in silico and chromatographic approaches that are faster, cheaper, and greener compared to the traditional shake-flask method. However, at present no accepted {"}standard{"} approach exists for their comparison and selection of the most appropriate one(s). This is of utmost importance when it comes to the development of new lipophilicity indices, or the assessment of the lipophilicity of newly synthesized compounds. In this study, 50 well-known, diverse compounds of significant pharmaceutical and environmental importance have been selected and examined. Octanol-water partition coefficients have been measured with the shake-flask method for most of them. Their retentions have been studied in typical reversed thin-layer chromatographic systems, involving the most frequently employed stationary phases (octadecyl- and cyano-modified silica), and acetonitrile and methanol as mobile phase constituents. Twelve computationally estimated logP-s and twenty chromatographic indices together with the shake-flask octanol-water partition coefficient have been investigated with classical chemometric approaches-such as principal component analysis (PCA), hierarchical cluster analysis (HCA), Pearson's and Spearman's correlation matrices, as well as novel non-parametric methods: sum of ranking differences (SRD) and generalized pairwise correlation method (GPCM). Novel SRD and GPCM methods have been introduced based on the Comparisons with One VAriable (lipophilicity metric) at a Time (COVAT). For the visualization of COVAT results, a heatmap format was introduced. Analysis of variance (ANOVA) was applied to reveal the dominant factors between computational logPs and various chromatographic measures. In consensus-based comparisons, the shake-flask method performed the best, closely followed by computational estimates, while the chromatographic estimates often overlap with in silico assessments, mostly with methods involving octadecyl-modified silica stationary phases. The ones that employ cyano-modified silica perform generally worse. The introduction of alternative coloring schemes for the covariance matrices and SRD/GPCM heatmaps enables the discovery of intrinsic relationships among lipophilicity scales and the selection of best/worst measures. Closest to the recommended logK OW values are ClogP and the first principal component scores obtained on octadecyl-silica stationary phase in combination with methanol-water mobile phase, while the usage of slopes derived from Soczewinski-Matyisik equation should be avoided.",
keywords = "Benzodiazepines, Comparison with one variable at a time-COVAT, Generalized pairwise correlation method-GPCM, Heatmap, Lipophilicity, Phenols, Polyaromatic hydrocarbons, Reversed-phase thin-layer chromatography, Sum of ranking differences-SRD",
author = "Filip Andrić and D{\'a}vid Bajusz and Anita R{\'a}cz and Sandra Šegan and K. Heberger",
year = "2015",
month = "10",
day = "29",
doi = "10.1016/j.jpba.2016.04.001",
language = "English",
journal = "Journal of Pharmaceutical and Biomedical Analysis",
issn = "0731-7085",
publisher = "Elsevier",

}

TY - JOUR

T1 - Multivariate assessment of lipophilicity scales-computational and reversed phase thin-layer chromatographic indices

AU - Andrić, Filip

AU - Bajusz, Dávid

AU - Rácz, Anita

AU - Šegan, Sandra

AU - Heberger, K.

PY - 2015/10/29

Y1 - 2015/10/29

N2 - Needs for fast, yet reliable means of assessing the lipophilicities of diverse compounds resulted in the development of various in silico and chromatographic approaches that are faster, cheaper, and greener compared to the traditional shake-flask method. However, at present no accepted "standard" approach exists for their comparison and selection of the most appropriate one(s). This is of utmost importance when it comes to the development of new lipophilicity indices, or the assessment of the lipophilicity of newly synthesized compounds. In this study, 50 well-known, diverse compounds of significant pharmaceutical and environmental importance have been selected and examined. Octanol-water partition coefficients have been measured with the shake-flask method for most of them. Their retentions have been studied in typical reversed thin-layer chromatographic systems, involving the most frequently employed stationary phases (octadecyl- and cyano-modified silica), and acetonitrile and methanol as mobile phase constituents. Twelve computationally estimated logP-s and twenty chromatographic indices together with the shake-flask octanol-water partition coefficient have been investigated with classical chemometric approaches-such as principal component analysis (PCA), hierarchical cluster analysis (HCA), Pearson's and Spearman's correlation matrices, as well as novel non-parametric methods: sum of ranking differences (SRD) and generalized pairwise correlation method (GPCM). Novel SRD and GPCM methods have been introduced based on the Comparisons with One VAriable (lipophilicity metric) at a Time (COVAT). For the visualization of COVAT results, a heatmap format was introduced. Analysis of variance (ANOVA) was applied to reveal the dominant factors between computational logPs and various chromatographic measures. In consensus-based comparisons, the shake-flask method performed the best, closely followed by computational estimates, while the chromatographic estimates often overlap with in silico assessments, mostly with methods involving octadecyl-modified silica stationary phases. The ones that employ cyano-modified silica perform generally worse. The introduction of alternative coloring schemes for the covariance matrices and SRD/GPCM heatmaps enables the discovery of intrinsic relationships among lipophilicity scales and the selection of best/worst measures. Closest to the recommended logK OW values are ClogP and the first principal component scores obtained on octadecyl-silica stationary phase in combination with methanol-water mobile phase, while the usage of slopes derived from Soczewinski-Matyisik equation should be avoided.

AB - Needs for fast, yet reliable means of assessing the lipophilicities of diverse compounds resulted in the development of various in silico and chromatographic approaches that are faster, cheaper, and greener compared to the traditional shake-flask method. However, at present no accepted "standard" approach exists for their comparison and selection of the most appropriate one(s). This is of utmost importance when it comes to the development of new lipophilicity indices, or the assessment of the lipophilicity of newly synthesized compounds. In this study, 50 well-known, diverse compounds of significant pharmaceutical and environmental importance have been selected and examined. Octanol-water partition coefficients have been measured with the shake-flask method for most of them. Their retentions have been studied in typical reversed thin-layer chromatographic systems, involving the most frequently employed stationary phases (octadecyl- and cyano-modified silica), and acetonitrile and methanol as mobile phase constituents. Twelve computationally estimated logP-s and twenty chromatographic indices together with the shake-flask octanol-water partition coefficient have been investigated with classical chemometric approaches-such as principal component analysis (PCA), hierarchical cluster analysis (HCA), Pearson's and Spearman's correlation matrices, as well as novel non-parametric methods: sum of ranking differences (SRD) and generalized pairwise correlation method (GPCM). Novel SRD and GPCM methods have been introduced based on the Comparisons with One VAriable (lipophilicity metric) at a Time (COVAT). For the visualization of COVAT results, a heatmap format was introduced. Analysis of variance (ANOVA) was applied to reveal the dominant factors between computational logPs and various chromatographic measures. In consensus-based comparisons, the shake-flask method performed the best, closely followed by computational estimates, while the chromatographic estimates often overlap with in silico assessments, mostly with methods involving octadecyl-modified silica stationary phases. The ones that employ cyano-modified silica perform generally worse. The introduction of alternative coloring schemes for the covariance matrices and SRD/GPCM heatmaps enables the discovery of intrinsic relationships among lipophilicity scales and the selection of best/worst measures. Closest to the recommended logK OW values are ClogP and the first principal component scores obtained on octadecyl-silica stationary phase in combination with methanol-water mobile phase, while the usage of slopes derived from Soczewinski-Matyisik equation should be avoided.

KW - Benzodiazepines

KW - Comparison with one variable at a time-COVAT

KW - Generalized pairwise correlation method-GPCM

KW - Heatmap

KW - Lipophilicity

KW - Phenols

KW - Polyaromatic hydrocarbons

KW - Reversed-phase thin-layer chromatography

KW - Sum of ranking differences-SRD

UR - http://www.scopus.com/inward/record.url?scp=84964765652&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84964765652&partnerID=8YFLogxK

U2 - 10.1016/j.jpba.2016.04.001

DO - 10.1016/j.jpba.2016.04.001

M3 - Article

C2 - 27155738

AN - SCOPUS:84964765652

JO - Journal of Pharmaceutical and Biomedical Analysis

JF - Journal of Pharmaceutical and Biomedical Analysis

SN - 0731-7085

ER -