There Is Poor Agreement between the Subjective and Quantitative Adjudication of Aneurysm Wall Enhancement.

Abstract

Background and purpose: The determination of aneurysm wall enhancement (AWE) by human readers on visual inspection alone is subjective and prone to error. A 3D method for quantifying the signal intensity (SI) of the aneurysm enables objective determination of AWE. Interreader agreement and agreement between subjective and objective determination of AWE were assessed in this study.

Materials and methods: Patients with saccular intracranial aneurysms (IAs) were imaged with high-resolution MRI. In the subjective assessment, 2 internal adjudicators visually determined AWE if the degree of enhancement was equal to or higher than that of the pituitary stalk. An experienced internal neuroradiologist resolved disagreements. This internal adjudication was compared with an external adjudication to assess interrater agreement among centers. In the objective assessment, the distribution of SI across the aneurysm wall after normalizing the SI to the corpus callosum was determined with an in-house code. The normalized mean SI on postcontrast T1 MRI was defined as 3D-circumferential AWE (3D-CAWE). If the 3D-CAWE value was higher than 1, an IA was defined as objectively "enhancing." Interrater agreement was analyzed with κ coefficients. Intertechnique agreement between the subjective and objective assessments was performed using κ statistics. Univariate regressions were used to identify which morphologic characteristics influenced subjective adjudication of enhancement.

Results: A total of 113 IAs were analyzed. The agreement of the internal assessment was moderate (κ = 0.63), 49.5% of IAs (56) were classified as "enhancing;" and 50.5% (57) as "nonenhancing" after consensus. Interrater agreement between internal and external adjudication was weak (κ = 0.52) for the presence of AWE. There was no agreement between the subjective assessment of AWE and objective 3D-CAWE (κ = 0.16, P = .02). Subjective assessment was less likely to reliably adjudicate enhancement when assessing multiple aneurysms (OR, 0.4; 95% CI, 0.16-0.97; P = .04) and IAs larger than >7 mm (OR, 0.22; 95% CI, 0.09-0.55; P = .002) despite being objectively nonenhancing.

Conclusions: Subjective adjudication of AWE has poor interrater agreement, and no agreement with an objective 3D method of determining AWE. It is also less likely than objective quantification to identify enhancement in aneurysms of >7 mm or when multiple aneurysms are present. Objective 3D quantification, such as the technique used in this study, should, therefore, be considered when assessing AWE, especially in patients with multiple aneurysms and aneurysms of >7 mm in size.