Warfarin derivatives as free radical scavengers: a coumarin scaffold-based linear regression model with in vitro validation

Antioxidant agents that efficiently scavenge reactive oxygen species (ROS) are of great interest in medicinal chemistry for their potential to mitigate oxidative stress-related pathologies. In this work, we developed an interpretable Multiple Linear Regression (MLR) QSAR model using seven molecular descriptors (D/Dr05, MATS2v, MATS8p, Mor24m, L2s, HATS3u, H8m) to predict the free radical scavenging activity of coumarin-based compounds as measured by the IC₅₀ in the DPPH assay. The MLR-QSAR model showed strong goodness-of-fit and robust internal and external validation parameters (<i>r</i>² = 81.04, <i>Q</i>²_LOO = 77.93, <i>Q</i>²_boot = 76.78, <i>r</i>²_ext = 75.38, <i>y</i>_{<i>s</i>cramble}<i>r</i>² = 0.25), supporting its predictive reliability. We applied the model to predict the antiradical potential of a novel set of Warfarin derivatives, a class of molecules historically known for anticoagulant properties but with unexplored antioxidant potential. Experimental in vitro DPPH assays on the seven Warfarin derivatives (WD) revealed a positive correlation (<i>r</i> = 0.63) with the predictions, validating the MLR-QSAR as a screening tool. Furthermore, all WD exhibited significant DPPH radical scavenging activity, demonstrating the chemical antioxidant potential of an anticoagulant-derived scaffold. This dual in silico-in vitro strategy highlights the value of interpretable QSAR models for guiding compound prioritization and structural optimization towards new coumarin-based antioxidants.