Design and analysis of small-molecule antineoplastic agents targeting brain tumors by utilizing pattern recognition methods and in silico optimized pharmaceutical properties

作者：Bartzatt Ronald^*

来源：Medicinal Chemistry Research, 2007, 16(7-9): 436-448.

DOI：10.1007/s00044-007-9084-x

• anticancer
• blood-brain barrier
• multivariate
• cluster analysis

The purpose of this work is the in silico design and characterization of small-molecule anticancer agents that target the central nervous system. All agents are cytotoxic nitrogen mustard agents, a class of drugs widely used in the clinical treatment of cancer. Molecular properties such as polar surface area, log P, and formula weight have previously been shown to be highly useful properties to utilize in optimizing drug design for penetration of the blood-brain barrier and subsequently the central nervous system. These properties are rigorously applied in this work for compiling the actual molecular structures of the anticancer agents. In silico molecular modeling and pattern recognition methods such as cluster analysis, non-metric multidimensional scaling, and K-means cluster analysis are applied to determine the importance of various pharmaceutical descriptors and interrelationships. Discriminant analysis of the molecular properties of 17 nitrogen mustard agents designed in this work determined the extent of dissimilarity among members of this group. Utilizing analysis of similarity (ANOSIM) on the ten pharmaceutical descriptors provides a discrete numerical result that indicates overall dissimilarity among these agents. K-means cluster analysis distinguished these agents into clusters of small, moderate, and high molecular weight. Discriminant analysis and cluster analysis generally discerned association of these 17 compounds on a basis of higher and lower formula weight criteria. Another pattern recognition method known as non-metric multidimensional scaling discerned finer associations that fell within categorizations by low, moderate, and higher formula weight. ANOSIM resulted in R = 0.5551, which indicates that there is a moderate level of dissimilarity within these brain-penetrating small anticancer agents. The use of in silico assisted drug design and multivariate analysis of these compounds discerned discrete interrelationships and information for evaluating potential clinical application.

• 出版日期2007-12