Interrater concordance of collateral arterial status scoring on CT angiography of patients with large vessel occlusion stroke

Saturday, May 6, 2023

First Author(s)

EA

Emily Avery, BA

Medical Student
Yale School of Medicine, Department of Radiology & Biomedical Imaging
La Jolla, California, United States

Co-Author(s)

AA

Anthony Abou-Karam, MD

Assistant Professor of Clinical Radiology and Biomedical Imaging
Yale School of Medicine, Department of Radiology & Biomedical Imaging, United States
SA

Sandra Abi-Fadel, MD

Assistant Professor of Clinical Radiology and Biomedical Imaging
Yale School of Medicine, Department of Radiology & Biomedical Imaging, United States
JB

Jonas Behland, n/a

Medical Student
Charité Universitätsmedizin Berlin, United States
AM

Adrian Mak, n/a

Medical Student
Charité Universitätsmedizin Berlin, United States
KS

Kevin Sheth, MD

Professor of Neurology and Neurosurgery
Yale School of Medicine, Department of Neurology, United States
CM

Charles Matouk, MD

Associate Professor of Neurosurgery
Yale School of Medicine, Department of Neurosurgery, United States
PS

Pina C. Sanelli, MD, MPH, FACR

Professor of Radiology
Northwell Health, Department of Radiology, United States
AM

Ajay Malhotra, MBBS, MD, MMM

Professor of Radiology and Biomedical Imaging
Yale School of Medicine, Department of Radiology & Biomedical Imaging, United States
SP

Seyedmehdi Payabvash, MD

Assistant Professor of Radiology and Biomedical Imaging
Yale School of Medicine, Department of Radiology & Biomedical Imaging, Connecticut, United States

Purpose: Large vessel occlusion (LVO) stroke is the leading cause of severe disability in the United States. Collateral arterial circulation beyond the site of occlusion is thought to modify individual extent of neurological damage among these patients. Collateral status has become increasingly relevant in acute LVO management and has recently been proposed to identify eligible patients for endovascular thrombectomy (ET). However, quantification of collateral status remains poorly standardized. Thus, we aimed to assess concordance of collateral status ratings among neuroradiologists.

Methods/Materials: For 598 acute LVO admission CTAs, 3 neuroradiologists quantified collateral status using the Miteff scale (good, moderate, or poor), with 3-rater consensus as ground truth. All neuroradiologists were board certified, each with at least 7 years of experience in reviewing brain CTAs. We calculated interrater correlation using Cohen’s Kappa and compared collateral status correlations with patient outcome (modified Rankin Score at discharge) using Pearson correlation.

Results: Our 598 patients had an average age of 70.215.2, average admission NIHSS of 15(10-19), and 46% were female. Interrater concordance for collateral status ranged from kappa=0.02 to 0.38, with an overall kappa=0.12 (p < 0.005) between the three raters. Individual neuroradiologist collateral status correlation with clinical outcome ranged from r=0.09 (p=0.37) to r=0.34 (p < 0.005).

Conclusions: Interrater correlation of collateral status among neuroradiologists was poor, with only modest correlation with clinical outcome. Our work depicts limited reliability of subjective collateral status, especially when used as a marker to determine ET eligibility or for other treatment decision-making in acute LVO stroke. Further development of automated machine-learning models may help provide objective, consistent quantification of collateral status.