Prognostic Nutritional Index Versus CALLY, CONUT, and CAR for Predicting Restenosis After Endovascular Treatment of Superficial Femoral Artery Chronic Total Occlusion.

OBJECTIVE: Inflammation- and nutrition-based indices have been associated with adverse outcomes after peripheral endovascular interventions, but their comparative performance remains unclear. We compared the associations and discriminatory performance of the prognostic nutritional index (PNI), C-reactive protein-albumin-lymphocyte (CALLY) index, Controlling Nutritional Status (CONUT) score, and C-reactive protein-to-albumin ratio (CAR) in relation to 6-month restenosis after endovascular treatment of superficial femoral artery chronic total occlusion (SFA CTO). METHODS: This retrospective cohort included 227 technically successful SFA CTO procedures performed between 2015 and 2021 with imaging-based 6-month restenosis assessment. The four indices were compared using ROC analysis and separate multivariable logistic regression models adjusted for clinical and lesion-related factors. Multiplicity correction and patient-cluster bootstrap internal validation were performed. RESULTS: Restenosis occurred in 79 procedures (34.8%). PNI showed the highest discrimination (AUC 0.728; 95% CI 0.662-0.794), compared with CONUT (0.625), CALLY (0.582), and CAR (0.568). After multivariable adjustment and multiplicity correction, PNI (adjusted OR 2.72 per 1-SD worsening; 95% CI 1.78-4.14) and CONUT (adjusted OR 1.61; 95% CI 1.17-2.22) remained independently associated with restenosis, whereas CALLY and CAR did not. Adding PNI to the clinical base model increased the AUC from 0.670 to 0.782 and reduced the Brier score from 0.209 to 0.178. The improvement persisted after bootstrap validation (optimism-corrected AUC 0.740). CONCLUSIONS: PNI showed the strongest association with and the greatest incremental discrimination for 6-month restenosis after SFA CTO intervention. These findings are limited to this retrospective cohort and don't establish PNI as a clinically validated predictive tool; external validation is required.