(PC080) The Impact of WIfI Clinical Stage on Amputation Prediction Modeling for Patients with CLTI in the VQI Database

Objectives: Chronic limb-threatening ischemia (CLTI) is the most severe form of peripheral arterial disease (PAD), often requiring lower extremity revascularization to prevent major amputation. The Wound, Ischemia, and foot Infection (WIfI) classification has been validated as a measure of CLTI severity but is underutilized in prediction models. This study evaluates the utility of WIfI in improving major amputation risk prediction in patients with CLTI undergoing PVI.

Methods: Patients who underwent PVI for CLTI from 2013-2024 in the Vascular Quality Initiative were included. Patients who did not have adequate data from which to derive WIfI stages were excluded. The primary end point was major amputation at long-term follow up. The sample was divided randomly into 2/3 for model development and 1/3 for validation. Multivariable logistic regression models for major amputation were developed with and without the WIfI stages. Additional covariates include those whose marginal association meeting p< 0.1. Model discrimination was evaluated using AUC (C-statistic), with statistical comparisons made via DeLong's test.

Results: A total of 280,706 individuals underwent PVI procedures, 47% (N=130,575) of whom presented with CLTI. WIfI stage derivation was only possible for 33% (N=42,858) of patients, 28,727 of whom had long-term follow up. Among demographic factors, male sex (OR = 1.22[1.10-1.35]) and Hispanic ethnicity (OR = 1.32 [1.09-1.6]) were associated with a higher risk of major amputation. Patients with preoperative diabetes (OR = 1.68[1.5-1.9]) and end-stage renal disease (OR = 1.9[1.65-2.18]) were at significantly higher risk. A dose-response relationship was observed between WIfI stage and major amputation risk, with an incremental increase in odds ratios: (OR=1.27[1.11–1.47]) for Stage 2, (OR=2.06[1.78–2.38]) for Stage 3, and (OR=3.41[2.95–3.95]) for Stage 4 compared to Stage 1. (Table) The model demonstrated good fit and discrimination, with the AUC improving from 0.69 before adding the WIfI stage variable to 0.74 after its inclusion. (Figure) DeLong's test showed a statistically significant difference between the models indicating superior performance by the model that included the WIfI stage (p <.001).

Conclusions: The integration of WIfI clinical staging into predictive models significantly improves the ability to stratify the risk of major amputation in patients with CLTI undergoing PVI. Incorporation of WIfI into other modules of revascularization could improve the prediction of major amputation using the VQI database accounting for the revascularization strategy.