Structure-antitumor activity relationships of 9-anilinoacridines using pattern recognition

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacology

Background:

• 9-anilinoacridine derivatives are investigated for antitumor properties.
• Traditional quantitative structure-activity relationship (QSAR) methods like Hansch analysis face challenges in modeling certain biological responses.
• Computer-aided drug design requires robust methods for predicting compound activity.

Purpose of the Study:

• To evaluate the utility of computer-generated molecular descriptors for classifying 9-anilinoacridine antitumor agents.
• To determine if pattern-recognition techniques can differentiate active from inactive compounds within this chemical class.

Main Methods:

• Pattern-recognition analysis was conducted using the ADAPT system.
• A training set of 213 compounds was randomly selected from 776 structures.
• Biological activity was measured by maximal increase in lifespan at the LD10 dosage.
• 18 molecular descriptors (fragment, substructure, physicochemical) were identified.

Main Results:

• The developed model correctly classified 94% of compounds in the training set (97% active, 85% inactive).
• Misclassified inactive compounds often contained amino substituents, suggesting ionization effects.
• Application to prediction sets yielded 73-86% correct classification, demonstrating generalizability.

Conclusions:

• Pattern-recognition analysis with computer-generated descriptors is a valuable tool for screening antitumor agents.
• This approach can aid in identifying potentially active compounds and refining drug design strategies.
• The method shows promise for both novel and existing chemical entities.