Diagnostic performance and inter-observer variability to differentiate between T1- and T2-stage gallbladder cancers using multi-detector row CT

Purpose: To evaluate the diagnostic performance and inter-observer variability of differentiating T1 and T2 gallbladder (GB) cancers using multi-detector row CT (MDCT). Methods: This retrospective study included 151 patients with surgically confirmed T1 (n = 49)- or T2 (n = 102)-stage GB cancer who underwent contrast-enhanced MDCT from 2016 to 2020. Five radiologists (two experienced and three less experienced) evaluated the T-stage with a confidence level calculated using a six-point scale. GB cancers were morphologically classified into three types: polypoid, polypoid with wall thickening, and wall thickening. The diagnostic performance of T-staging was assessed using receiver operating characteristic (ROC) curve analysis. Sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were calculated based on a binary scale (T1 = positive). Inter-observer agreement was assessed using Fleiss κ statistics. Results: The area under the receiver operating characteristic (ROC) curve of each reviewer for T-staging ranged from 0.69 to 0.80 (median 0.77). The overall accuracy of the five radiologists was 78% (95% confidence interval [CI] 71–84%). Sensitivity was higher and specificity was lower in experienced radiologists than in less experienced radiologists (P < 0.001). The overall inter-observer agreement was fair (κ = 0.36; 95% CI 0.31, 0.41). The overall accuracy for T-stage was 63% (95% CI 48–76), 78% (95% CI 63–88), and 87% (95% CI 77–93) for polypoid, polypoid with wall thickening, and wall thickening type, respectively. Conclusion: The accuracy of MDCT for differentiating T1 and T2 GB cancer is limited, and there is considerable inter-observer variability.