Unified multi-target approach for the rational in silico design of anti-bladder cancer agents

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Oncology

Background:

• Bladder cancer (BLC) is a prevalent and aggressive malignancy requiring effective chemotherapy.
• Current computer-aided drug design (CADD) lacks methods to predict anti-BLC activity across diverse cell lines.
• Development of potent anti-BLC agents necessitates rational drug discovery approaches.

Purpose of the Study:

• To establish the first unified quantitative structure-activity relationship (QSAR) approach for predicting anti-BLC activity.
• To develop multi-target (mt) QSAR models for classifying compounds against four distinct BLC cell lines.
• To identify structural features influencing compound activity and propose novel anti-BLC drug candidates.

Main Methods:

• Construction of two multi-target (mt) QSAR models using a large, heterogeneous compound database.
• Model 1: Linear discriminant analysis (mt-QSAR-LDA) with fragment-based descriptors.
• Model 2: Artificial neural networks (mt-QSAR-ANN) with global 2D descriptors.

Main Results:

• Both mt-QSAR-LDA and mt-QSAR-ANN models achieved over 90% accuracy in classifying active and inactive compounds.
• Models demonstrated high predictive performance on both training and independent prediction sets.
• Identified key substructural patterns correlated with anti-BLC activity or inactivity.

Conclusions:

• The developed unified QSAR models provide a robust computational tool for predicting anti-BLC activity.
• These models facilitate the rational design of more potent and versatile anti-bladder cancer chemotherapy agents.
• The study suggests novel molecular entities with potential as future anti-BLC therapeutics.