TCAR versus tfCAS: Machine Learning Suggests That One is Safer, Faster, and Easier to Master

Document Type

Conference Proceeding

Publication Date

6-1-2025

Publication Title

J Vasc Surg

Abstract

Objectives: This study aims to evaluate and compare the learning curves for transcarotid artery revascularization (TCAR) and transfemoral carotid artery stenting (tfCAS) based on procedural metrics and postoperative outcomes. Utilizing machine learning, we sought to identify procedural milestones, assess trends in operator performance, and explore the relationship between procedural proficiency and patient outcomes. Methods: Patients undergoing TCAR or tfCAS were identified using the Vascular Quality Initiative between 2005 and 2024. Physicians performing these procedures were assigned cumulative procedure counts by chronologically ordering their cases, enabling the analysis of operator experience over time. A machine learning model using neural network was implemented to predict how the risk of complications change with increasing procedure count. The learning curves for both TCAR and tfCAS were defined based on changes in the predicted probabilities of primary outcomes, which included stroke and major adverse cardiovascular events (MACE), comprising stroke, death, and myocardial infarction. Secondary outcomes included stroke/death and all-cause mortality. Procedural metrics, including contrast volume, fluoroscopy time, and operative time, were assessed to evaluate improvements in procedural efficiency over time. Results: The study included 48,536 TCAR and 37,561 tfCAS procedures, with an average of 20 TCAR procedures per operator and 16 tfCAS procedures per operator. MACE occurred at a rate 1.9% (2.4% symptomatic, 1.3% asymptomatic) for TCAR patients and 3.5% (4.5% symptomatic, 1.8% asymptomatic) for tfCAS patients. Among symptomatic patients, the rate of MACE was initially higher for tfCAS (5.01%) compared to TCAR (2.58%). While tfCAS showed a rapid decline early on, TCAR experienced a more gradual reduction. After the 50th procedure, the rates for both began to decline in parallel (Fig 1). The outcomes of stroke, death, and stroke/death followed similar downward trends. Asymptomatic patients displayed comparable patterns, but with overall lower rates of adverse events. Physician experience had a pronounced impact on procedural efficiency; for TCAR, the average procedure time (Fig 2) decreased from 79 minutes to 39 minutes, while for tfCAS, the reduction was less substantial (76 minutes to 56 minutes.) This improvement was mirrored by similar reductions in fluoroscopy time and contrast volume. Conclusions: TCAR consistently outperformed tfCAS, in both procedural and postoperative metrics. Notably, even the most experienced tfCAS operators achieved outcomes inferior to those of beginner TCAR operators. These findings highlight that TCAR is not only safer but easier to master. [Formula presented] [Formula presented]

Volume

81

Issue

6

First Page

e9

Last Page

e10

Find It @ Sladen

Share

COinS