Presenter: Thomas Sutton MD | Oregon Health and Science University Background: Intrahepatic cholangiocarcinoma (ICC) is increasing in incidence and is potentially associated with non-alcoholic fatty liver disease (NAFLD) and steatohepatitis (NASH). The presence of both NAFLD and NASH has been shown to affect overall survival in patients with resected ICC. Steatosis, the primary criterion for NAFLD/NASH, may be under-reported on pathology and imaging, potentially impacting epidemiologic conclusions used to guide high-risk screening and treatment recommendations for ICC. We sought to evaluate the performance of standard-of-care (SOC) pathology, computed tomography (CT), and magnetic resonance imaging (MRI) for identifying steatosis. Methods: We identified n=52 patients with ICC treated with curative-intent resection from 2004-17. Hepatobiliary pathologists and radiologists specifically re-reviewed areas of non-malignant liver on tissue sections and pre-operative imaging, respectively, for evidence of steatosis. The pathologic criterion for steatosis was >5% of surface area involved by steatosis as evaluated by low to medium power examination. For MRI, chemical shift artifact seen on standard in and out-of-phase T1-weighted imaging was utilized to identify steatosis. For CT, density measurements were utilized, either absolute measurements on non-contrast CT or relative densities between liver and spleen when only post-contrast portal venous phase images were available. Using the dedicated pathology re-review as the reference, the accuracy of SOC pathology, contrasted CT, and MRI for diagnosing steatosis was evaluated by receiver operating characteristics (ROC) analysis. Results: The median patient age was 64-years with an equal gender distribution. The median body mass index was 30.3 kg/m2, and 11 (21%) patients had diabetes mellitus. Dedicated pathology review identified steatosis in 20 patients (38%) and was discordant from original reports in 8 cases (15%), with 4 instances each of over-diagnosis and under-diagnosis. On ROC analysis, the original pathology report correctly identified steatosis with an area under the curve (AUC) of 0.87 (P < 0.001 compared to chance), while MRI had an AUC of 0.83 (P < 0.001). Dedicated CT review was less accurate for steatosis (AUC 0.65, P=0.01), and SOC CT was not significantly better than chance (AUC 0.58, P=0.5)(Figure). Conclusion: SOC imaging, SOC pathologic examination, and dedicated review of imaging are less accurate in the diagnosis of steatosis compared to dedicated pathologic review of non-malignant liver tissue. The authors suspect this could be due to the lack of emphasis on documenting steatosis by pathologists and radiologists in cases of malignancy, where documenting cancer-associated features is understandably the primary focus. Epidemiologic studies relying on SOC imaging reports, and to a lesser extent those relying on SOC pathology or dedicated imaging review, are limited in their validity when associating NAFLD with primary liver cancers such as ICC and should be interpreted with caution.
Read full abstract
May 7, 2021
Melissa B Carroll + 3
Open Access