Development and initial clinical application of the HKL Thrombus Burden Classification for endovascular salvage of...

The absence of a standardised method for assessing thrombus burden limits endovascular management of thrombosed haemodialysis vascular access. Studies comparing thrombectomy techniques rarely account for lesion complexity, introducing treatment-selection bias. We evaluated the HKL Thrombus Burden Classification, a structured framework designed to stratify thrombus complexity and guide treatment.

This retrospective cohort study included consecutive endovascular thrombectomy procedures for thrombosed haemodialysis access at a tertiary centre (January-June 2025). Thrombus burden was classified prospectively using the HKL Classification (thrombus length, organisation, venous outflow, aneurysmal burden), and treatment followed a predefined classification-guided strategy. The primary outcome was 6-month primary patency. Kaplan-Meier analysis and Cox regression identified independent predictors of access failure; thrombectomy technique was retained as an adjustment variable, multicollinearity was assessed using variance inflation factors, and a post-hoc power calculation contextualised non-significant findings.

Of 87 procedures in 75 patients (12 contributing more than one procedure), 32 (36.8%) were Grade 1, 38 (43.7%) Grade 2 and 17 (19.5%) Grade 3. Six-month primary patency was 68.8% (95% CI 52.7-84.8), 57.9% (42.2-73.6), and 52.9% (29.2-76.7) respectively (log-rank p = 0.482). Increasing HKL grade independently predicted access failure (adjusted hazard ratio 2.42, 95% CI 1.12-5.25; p = 0.025), whereas thrombectomy technique was not independently associated with outcome after adjustment for thrombus burden. Variance inflation factors ranged from 1.12 to 3.47. With 34 events among 87 procedures, the study had 80% power to detect a hazard ratio of approximately 0.31 or 3.27.

The HKL Classification provides a pragmatic hierarchical framework for pre-procedural assessment of thrombosed haemodialysis access and independently predicts primary patency. These findings support structured thrombus burden assessment to guide treatment planning. Prospective multicentre validation, including interobserver reliability testing, is required before clinical implementation.