Inter- and Intraobserver Agreement of ¹⁸F-FDG PET/CT Image Interpretation in Patients Referred for Assessment of Cardiac Sarcoidosis

Abstract

Recent studies have reported the usefulness of ¹⁸F-FDG PET in aiding with the diagnosis and management of patients with cardiac sarcoidosis (CS). However, image interpretation of ¹⁸F-FDG PET for CS is sometimes challenging. We sought to investigate the inter- and intraobserver agreement and explore factors that led to important discrepancies between readers. <b>Methods:</b> We studied consecutive patients with no significant coronary artery disease who were referred for assessment of CS. Two experienced readers masked to clinical information, imaging reports, independently reviewed ¹⁸F-FDG PET/CT images. ¹⁸F-FDG PET/CT images were interpreted according to a predefined standard operating procedure, with cardiac ¹⁸F-FDG uptake patterns categorized into 5 patterns: none, focal, focal on diffuse, diffuse, and isolated lateral wall or basal uptake. Overall image assessment was classified as either consistent with active CS or not. <b>Results:</b> One hundred scans were included from 71 patients. Of these, 46 underwent ¹⁸F-FDG PET/CT with a no-restriction diet (no-restriction group), and 54 underwent ¹⁸F-FDG PET/CT with a low-carbohydrate, high-fat and protein-permitted diet (low-carb group). There was agreement of the interpretation category in 74 of 100 scans. The κ-value of agreement among all 5 categories was 0.64, indicating moderate agreement. For overall clinical interpretation, there was agreement in 93 of 100 scans (κ = 0.85). When scans were divided into the preparation groups, there was a trend toward higher agreement in the low-carb group versus the no-restriction group (80% vs. 67%, <i>P</i> = 0.08). Regarding the overall clinical interpretation, there was also a trend toward greater agreement in the low-carb group versus the no-restriction group (96% vs. 89%, <i>P</i> = 0.08). <b>Conclusion</b>: The interobserver agreement of cardiac ¹⁸F-FDG uptake image patterns was moderate. However, agreement was better regarding overall interpretation of CS. Detailed prescan dietary preparation seemed to improve interobserver agreement.