Quantitative Assessment Of Inflammatory (18)Fdg Pet Scans For Diagnosis Of Cardiac Sarcoidosis

650 Objectives: Identification of myocardial inflammation with FDG PET is increasingly used for diagnosis and management of cardiac involvement in sarcoidosis. Because consistent quantitative thresholds are lacking, interpretation of images can vary markedly from site to site. Incomplete metabolic preparation leading to incomplete suppression of physiologic uptake of FDG by cardiomyocytes is also common and difficult to recognize. The University of Michigan protocol [1] consistently suppresses myocardial uptake to levels that are equal to or below blood pool, which is not quantified by existing cardiac metabolic volume (CMV) and activity (CMA) metrics [2]. In this work, we retrospectively analyzed inflammatory FDG PET scans to establish quantitative thresholds for identifying positive studies and performed outcome-based validation. Methods: 316 inflammatory FDG scans performed at the University of Michigan between June 2015 and June 2018 were processed and reviewed in Corridor4DM (INVIA Medical Imaging Solutions, Ann Arbor, MI, USA). 15 studies were excluded for non-diagnostic preparation (high standardized uptake values (SUV) and low coefficient of variation) confirmed via further review of images, laboratory results, and patient history. Contours were transferred from perfusion studies to inflammatory datasets and polar maps were automatically sampled from the contours. Two myocardial sampling methods were investigated: 1) sampling the max SUV of the corresponding myocardial segment (MAX), and 2) sampling SUV at the midwall of the corresponding segment (MIDWALL). Presence of abnormal FDG uptake in the left ventricle (LV) assessed by the reading physicians was used as reference to validate the quantification calculated by the software. LV metabolic volume (LVMV) and activity (LVMA) were computed analogously to CMV and CMA, using a threshold of 1.5 × mean blood pool SUV. However, they were focused only on the LV, unlike the CMV and CMA which quantify both left and right ventricles. The ratio of maximum tissue SUV to mean blood pool SUV was used to create receiver operating characteristic (ROC) curves for both sampling methods. Cox survival analysis was performed on patients with ejection fraction (EF) ≥ 50% (N=172) for the composite of major adverse cardiac events (ventricular tachycardia, ventricular fibrillation, heart failure hospitalization, ventricular assist device, heart transplant, and death), adjusting for summed rest score (SRS). Results: MIDWALL sampling resulted in greater area under the ROC curve than MAX sampling (0.89 vs. 0.84, p=0.002). Importantly, LVMA > 0 (marked with X on Figure 1) corresponds to a threshold of 1.5 × mean blood pool SUV, illustrating poor sensitivity. Using a threshold of 1 × mean blood pool SUV with MIDWALL sampling, positive FDG quantification was a significant predictor of MACE (HR=2.65 [95% CI 1.07-6.56], p=0.03). Using the LVMA/CMA published threshold of 1.5 × mean blood pool SUV was not significant (HR=0.61 [95% CI 0.07-5.27], p=0.65). Conclusions: Quantification of LV myocardial FDG activity above blood pool SUV is a reliable quantitative method for diagnosis of myocardial inflammation and is associated with worse prognosis. References: [1] Larson, S.R., Pieper, J.A., Hulten, E.A. et al. Characterization of a highly effective preparation for suppression of myocardial glucose utilization. J. Nucl. Cardiol. (2019) doi:10.1007/s12350-019-01786-w [2] Ahmadian, A., Brogan, A., Berman, J. et al. Quantitative interpretation of FDG PET/CT with myocardial perfusion imaging increases diagnostic information in the evaluation of cardiac sarcoidosis. J. Nucl. Cardiol.21, 925-939 (2014) doi:10.1007/s12350-014-9901-9