Comparison of Angiographic Dissection Classification Systems in the Femoropopliteal Arteries Using IVUS Validation and Reliability Testing

Purpose: Objective assessment of dissection severity is difficult. Recognition of this has led to the creation of classification systems. This study investigated the performance of the National Heart Lung and Blood Institute (NHLBI) and Kobayashi systems at differentiating severity of femoropopliteal dissection using intravascular ultrasound (IVUS) as the reference standard. Comparison between the 2 systems and the inter- and intra-observer reliability were also investigated. Materials and Methods: Angiographic and IVUS imaging was assessed in 51 cases sourced from a RCT investigating the use of IVUS in femoropopliteal disease. A total of 2 readers independently scored the angiography images according to NHLBI and Kobayashi dissection classification systems and a consensus score was obtained for each system in each case. The NHLBI classification was condensed into 3 grades of dissection to allow comparison between systems. Dissection length, dissection arc, minimum lumen area, and lumen area stenosis were obtained from the IVUS imaging. IVUS parameters were compared between grades of severity for both systems. Agreement in grading between the systems was assessed and IVUS parameters for each level of dissection severity were compared between systems. Inter and intra-observer agreement tested for each system. Results: Dissection was present on IVUS in 92.2% (47/51) of cases and angiography identified 78.7% (37/47) of these. No difference was present in any IVUS parameters between mild and severe dissections with either classification system. No difference in IVUS findings was present for the same grades of dissection between systems. The 2 systems agreed on severity grade in 47 of 51 cases. The inter-observer agreement was for NHLBI was k=0.549 and k=0.627 for Kobayashi. Intra-observer agreement for NHLBI was k=0.633 and k=0.633 and for Kobayashi was k=0.657 and k=0.297. Conclusion: The lack of difference in IVUS parameters between mild and severe dissection for the NHLBI and Kobayashi systems raises doubts about their ability to effectively differentiate dissection severity. Weak to moderate reliability suggests that variability in interpreting dissection may be higher than acceptable. IVUS imaging is more sensitive for detecting dissection than angiography and research is required to establish the value of adding IVUS to dissection classification systems.