Ashish P Wasnik1, Mathew S Davenport2, Ravi K Kaza2, William J Weadock2, Aaron Udager3, Nahid Keshavarzi4, Bin Nan5, Katherine E Maturen2. 1. Department of Radiology, University of Michigan Health System, 1500 E. Medical Center Dr., Ann Arbor, MI 48109, United States. Electronic address: ashishw@med.umich.edu. 2. Department of Radiology, University of Michigan Health System, 1500 E. Medical Center Dr., Ann Arbor, MI 48109, United States. 3. Department of Pathology, University of Michigan Health System, 1500 E. Medical Center Dr., Ann Arbor, MI 48109, United States. 4. Michigan Institute of Clinical & Health Research (MICHR), University of Michigan Health System, 1500 E. Medical Center Dr., Ann Arbor, MI 48109, United States. 5. Department of Statistics, School of Public Health, University of Michigan, Ann Arbor, MI 48109, United States.
Abstract
OBJECTIVE: To determine the diagnostic accuracy of multi-detector CT (MDCT) for differentiating gallbladder cancer from acute and xanthogranulomatous cholecystitis using previously described imaging features. METHODS: In this IRB approved HIPAA-compliant retrospective cohort study, contrast-enhanced MDCT of histologically confirmed acute cholecystitis (n = 17), xanthogranulomatous cholecystitis (n = 25), and gallbladder cancer (n = 18) were reviewed independently by three abdominal radiologists blinded to outcome. The primary outcome was the diagnostic accuracy of MDCT for the differentiation of gallbladder cancer from cholecystitis (acute and xanthogranulomatous) using various imaging parameters. Kappa (κ) statistics and two-way mixed-model single-measure intra-class correlation statistics (ICC) were calculated for each imaging feature and the final radiologic diagnosis. RESULTS: Inter-rater agreement was moderate to substantial (κ = 0.43-0.70), sensitivity 0.67-0.78, specificity 0.22-0.33 and the positive likelihood ratio was 4.28-8.56 for the differentiation of gallbladder cancer from benign gallbladder pathology. Only three imaging findings: disrupted gallbladder mucosa (κ = 0.68), intraluminal gallstones (κ = 0.66), and gallbladder wall thickness (ICC = 0.63) had substantial inter-rater agreement. The following had slight or no agreement: intramural hypoattenuating nodules (κ = 0.17), transient hepatic attenuation differences (κ = 0.14), gallbladder wall calcification (κ = -0.01), gallbladder wall enhancement (κ = 0.18), and omental or mesenteric invasion (κ = 0.08). In the final multivariate model, the following were significant predictors useful in making or excluding diagnosis of gallbladder cancer: focal gallbladder wall thickening (p = 0.003, OR: 13.09 [95% CI: 2.40-71.48]), pericholecystic "fat stranding" (p = 0.018, OR: 0.10 [95% CI: 0.01-0.66]), and maximum short axis lymph node diameter (p = 0.043, OR: 1.18 [95% CI: 1.00-1.38]). CONCLUSION: MDCT has moderate sensitivity, poor specificity, and moderate-to-substantial inter-rater repeatability for the differentiation of gallbladder cancer from acute and xanthogranulomatous cholecystitis.
OBJECTIVE: To determine the diagnostic accuracy of multi-detector CT (MDCT) for differentiating gallbladder cancer from acute and xanthogranulomatous cholecystitis using previously described imaging features. METHODS: In this IRB approved HIPAA-compliant retrospective cohort study, contrast-enhanced MDCT of histologically confirmed acute cholecystitis (n = 17), xanthogranulomatous cholecystitis (n = 25), and gallbladder cancer (n = 18) were reviewed independently by three abdominal radiologists blinded to outcome. The primary outcome was the diagnostic accuracy of MDCT for the differentiation of gallbladder cancer from cholecystitis (acute and xanthogranulomatous) using various imaging parameters. Kappa (κ) statistics and two-way mixed-model single-measure intra-class correlation statistics (ICC) were calculated for each imaging feature and the final radiologic diagnosis. RESULTS: Inter-rater agreement was moderate to substantial (κ = 0.43-0.70), sensitivity 0.67-0.78, specificity 0.22-0.33 and the positive likelihood ratio was 4.28-8.56 for the differentiation of gallbladder cancer from benign gallbladder pathology. Only three imaging findings: disrupted gallbladder mucosa (κ = 0.68), intraluminal gallstones (κ = 0.66), and gallbladder wall thickness (ICC = 0.63) had substantial inter-rater agreement. The following had slight or no agreement: intramural hypoattenuating nodules (κ = 0.17), transient hepatic attenuation differences (κ = 0.14), gallbladder wall calcification (κ = -0.01), gallbladder wall enhancement (κ = 0.18), and omental or mesenteric invasion (κ = 0.08). In the final multivariate model, the following were significant predictors useful in making or excluding diagnosis of gallbladder cancer: focal gallbladder wall thickening (p = 0.003, OR: 13.09 [95% CI: 2.40-71.48]), pericholecystic "fat stranding" (p = 0.018, OR: 0.10 [95% CI: 0.01-0.66]), and maximum short axis lymph node diameter (p = 0.043, OR: 1.18 [95% CI: 1.00-1.38]). CONCLUSION: MDCT has moderate sensitivity, poor specificity, and moderate-to-substantial inter-rater repeatability for the differentiation of gallbladder cancer from acute and xanthogranulomatous cholecystitis.
Authors: Vikas Gupta; K S Vishnu; Thakur D Yadav; Yashwant R Sakaray; Santosh Irrinki; B R Mittal; N Kalra; K Vaiphei Journal: J Gastrointest Cancer Date: 2019-12