Objective: Assess the sensitivity and specificity of computed tomography angiography (CTA) for carotid near-occlusion diagnosis interpreted in clinical practice against expert assessment.
Methods: CTAs were graded by two expert interpreters for near-occlusion. Findings were compared with clinical reports in 383 consecutive cases with symptomatic ≥ 50% carotid stenosis. In addition, 14 selected CTA exams (8 near-occlusions and 6 controls) were analyzed in a national effort by 13 radiologists experienced with carotid CTA.
Results: In clinical practice, imaging reports were 20% (95% CI 12-28%) sensitive for near-occlusion, ranging 0-58% between different radiologists; specificity was 99%. Among the 13 radiologists reviewing the same 8 near-occlusions, the average sensitivity was 8%, ranging 0-75%; specificity was 100%.
Conclusions: Carotid near-occlusion is systematically under-reported in clinical routine practice, caused by limited application of grading criteria when assessing CTA.
Key points: • Carotid near-occlusion is severe stenosis with distal artery collapse; this collapse is often subtle. • A fifth of near-occlusions were detected in routine practice. Many readers mistake near-occlusion for stenosis without distal artery collapse, either by not actively searching for subtle collapses or by not interpreting the collapse correctly when noticed. • On the other hand, the novice diagnostician should be cautioned to not over-diagnose near-occlusion; other causes of extracranial ICA asymmetry also exist such as distal disease and Circle of Willis anatomical variants.
Keywords: Carotid stenosis; Computed tomography angiography; Stroke.