Drug repositioning strategies for antihypertensive drug discovery via machine learning

Abstract Introduction The employment of drug repositioning strategies has emerged as a critical approach for the optimisation of the development of antihypertensive drugs. This strategy can be regarded as an innovative drug development model. Methods A combination of machine learning models was utilised for the purpose of screening 3,285 non-antihypertensive drugs in the FDA database for the presence of any potential antihypertensive properties. The integration of retrospective clinical data analysis, animal experiments and cell-based studies was employed for the systematic assessment of the concordance between the outcomes of the predictive models and the observed data. Results In vivo, it was demonstrated that rivaroxaban and tioconazole exhibited significant antihypertensive activity, with rivaroxaban demonstrating an effect similar to that of clinical antihypertensive drugs like candesartan and amlodipine in Angiotensin II (Ang II)-induced hypertensive mouse models. In vitro, rivaroxaban, ibuprofen, risperidone, and tioconazole were found to significantly inhibit Ang II-induced Vimentin and Col 1a mRNA levels in endothelial cells and to reduce smooth muscle cell migration after Ang II stimulation. Meanwhile, molecular docking revealed that rivaroxaban primarily exhibited strong binding affinity for renin and the glucose-dependent insulinotropic receptor. Conclusions This study provides scientific evidence for the development of novel antihypertensive strategies. Rivaroxaban consistently exhibited superior antihypertensive effects and endothelial remodelling benefits across multiple validations, providing a potential novel therapeutic option for patients with non-valvular atrial fibrillation and concomitant hypertension.