Routine preprocedurale data voorspelt linkerventrikelherstel na TAVR en identificeert cardiorenal fenotype

BACKGROUND: Ejection fraction (LVEF) underpins risk stratification before transcatheter aortic valve replacement (TAVR), yet it does not indicate whether an impaired ventricle will recover once the stenosis is relieved, or what drives death in high-risk patients. We asked whether routine preprocedural data could address both. METHODS AND RESULTS: Among 1212 consecutive patients (1213 procedures) who underwent TAVR for severe aortic stenosis (2012-2026), unsupervised clustering of 18 routine variables (principal components, Ward linkage) defined four phenotypes: low-risk, elderly preserved-EF, reduced-EF, and cardiorenal. Five-year mortality ranged from 13.5% to 44.2% (P < 0.001). Among 180 patients with a baseline LVEF below 50%, a parsimonious model combining LV end-diastolic diameter, mean gradient, and coronary disease predicted a ≥ 10-point LVEF gain (optimism-corrected AUC, 0.75; 0.65 for LVEF alone, a difference that was not statistically significant in this sample), and an integer TAVR-RECOVER score stratified observed recovery from 69% to 98%. Reduced-EF patients recovered from a mean LVEF of 38% to 59% by 1 year. The cardiorenal phenotype carried a 3.4-fold age-adjusted mortality (95% CI, 2.0-5.6) but died chiefly of noncardiovascular causes (5-year incidence, 32% versus 13% cardiovascular); its excess risk became non-significant after adjustment for renal function and other routine laboratories (hazard ratio, 1.60; 95% CI, 0.61-4.17). Phenotype re-stratified mortality within each LVEF category, with an almost 3-fold gradient among preserved-EF patients. CONCLUSIONS: Routine preprocedural data can estimate which impaired ventricles recover after TAVR and identify a cardiorenal phenotype whose excess, largely noncardiovascular mortality is statistically explained by measurable renal dysfunction rather than captured by ejection fraction. Prospective external validation is warranted.