This study is aimed at finding Quantitative Structure Activity Relationships (QSAR) for drugs reported to result in fatal consequence due to kidney failure and categorized as Adverse Drug Reactions (ADR). Study is based on the reports from open source Canada Vigilance Adverse Reaction Online database. Biological toxicity of small molecules has been predicted as a function of molecular structural features represented by their molecular descriptors. QSAR methods used have identified the structural features of the drugs/molecules and predicted their toxicity. Drugs suspected to cause kidney failure as ADR were analyzed. The molecular descriptors of these drugs were obtained using DRAGON web interface. The structural characteristics that distinguish drugs reported to cause death due to kidney failure as ADR against drugs not causing death but causing kidney failure as ADR were checked. Three QSAR methods used to find the relationships were Simple Kmeans clustering, decision tree and linear regression analysis. The greater value of the descriptor MAXDP is favorable for preventing death has been illustrated by all three models. The 9-membered ring of the benzimidazole substructure can be inferred from Pubchem database to contribute positively towards death. The descriptor, T(N..P), sum of topological distances between N..P 2D atom pairs, would prevent death if its value is lowered. A decrease in value of the descriptor, PCR - ratio of multiple path count over path count, will result in a decrease in probability of fatal consequences. The study predicts that renal toxicity could be decreased if the above mentioned molecular descriptors are modified.
