Heterogeneity in the efficacy of bosentan for hypertensive nephropathy: a study on individualized benefit prediction...

Many patients with hypertensive nephropathy progress to chronic kidney disease (CKD). Targeting the core pathophysiological pathway of endothelin-1 activation, this study aimed to elucidate response patterns to the endothelin receptor type A/B (ETA/ETB) antagonist bosentan and construct a personalized prediction model.

This single-centre retrospective cohort study enrolled 166 patients with hypertension-related nephropathy confirmed by renal biopsy. Risk stratification was performed based on a pathological injury scoring. Three efficacy trajectories were identified using latent variable growth mixture models. Efficacy drivers were screened via random forest algorithms. Finally, early response thresholds were established using receiver operating characteristic (ROC) curves.

Patients in the high-damage group exhibited significantly lower creatinine reduction than those in the low-damage group. Trajectory analysis revealed three patterns: sustained improvement (65.1%), with a 12-month creatinine reduction of 110.8 μmol/L, and deterioration (12.0%), with a deterioration inflection point at 5.3 ± 1.7 months and a subsequent creatinine increase of 40.2 μmol/L. Pathological injury score was the primary determinant of efficacy and was significantly negatively correlated with creatinine reduction (r = -0.61). Early response threshold analysis indicated that ΔCr_3M ≥ 42.3 μmol/L predicted 12-month efficacy, with an area under the curve (AUC) of 0.79. The early target-achievement group demonstrated significantly greater creatinine reduction than the nonachievement group. Compared with the non-sodium-glucose cotransporter-2 inhibitor (SGLT2i) group, the SGLT2i group demonstrated a 59% reduction in creatinine, while the diabetic subgroup showed a 47.1% smaller reduction than the nondiabetic subgroup.

Bosentan may improve glomerular haemodynamics in patients with hypertensive nephropathy. We propose pathological and early-response thresholds that could guide dynamic, precision interventions before critical inflection points, potentially paving the way for an upgrade from static to proactive management. These findings require validation in larger cohorts.