Striving for consistency in the measurement of LVEF and left ventricular volume

Abstract

Myocardial perfusion imaging (MPI) with positron emission tomography (PET) is a robust imaging modality for evaluation of patients at risk for developing or with known coronary artery disease and it has several advantages over other nuclear imaging modalities. The advantages of PET include high spatial and temporal resolution, routine use of either 68-Germanium source or computed tomography (CT) for attenuation correction, use of short-lived radiotracers such as Rb or NH3 and low radiation exposure. 1 ECG-gated PET permits assessment of left ventricular ejection fraction (LVEF) and volumes, and also has the additional advantage of the ability to measure myocardial blood flow (mL/min/g) and flow reserve. All of those cardiac PET imaging parameters have been well studied and play an important role in diagnosis and prognosis of coronary disease. Given that many critical prognostic and therapeutic decisions are based on left ventricular size and function, LVEF, LV end-diastolic volume (LVEDV), and LV end-systolic volume (LVESV) need to be accurately measured. In addition to accuracy, reproducibility is also important as are intra-observer and inter-observer variability. Software vendors do their best to demonstrate those qualities before they market their product. However, there is no independent, accepted, gold standard to which a vendor can compare its product. Therefore, demonstration of the reproducibility and the accuracy of the measurement of those variables is of great importance in clinical practice. Using ECG gating, functional and volumetric measurements are typically performed in conjunction with myocardial perfusion imaging. LVEF is defined quantitatively by analysis of a reconstructed threedimensional data set using software with automated edge detection where few geometric assumptions need to be made about the LV cavity shape. Several different proprietary, quantitative software products such as QPET (Cedars-Sinai), Syngo MBF (Siemens Healthcare), FlowQuant (University of Ottawa Heart Institute), and Corridor4DM (INVIA, LLC, Ann Arbor, USA) and others are currently available for PET MPI processing. They differ in subtle and sometimes, not so subtle ways that may result in differing quantitative results as has been demonstrated for SPECT imaging. We know from previous studies of SPECT imaging, that there is variability among the different software products especially in regard to the quantitative assessment of ventricular size and function. Bravo et al evaluated reference values for LVEF, LVESV, and LVEDV in gated Rb Cardiac PET/CT using four different commercial software packages (CardIQ Physio (a dedicated PET software)) and three SPECT software packages (Emory Cardiac Toolbox, Quantitative Gated SPECT, and 4DM-SPECT) applied to the PET data and found that LVEF and LV volumes from gated Rb PET/ CT varied significantly among available software programs and therefore could not be used interchangeably. Additionally, Oliveira et al also demonstrated that when using different PET software products, the mean values of myocardial blood flow and myocardial perfusion reserve were statistically significantly different. The authors of that study advised caution when using different software packages as the differences may introduce quantitative variation which could be clinically significant. In this issue of the Journal, Byrne et al also discovered poor concordance in healthy volunteers between the measurements of LVEF and LV volume when comparing two different PET software packages. Forty Reprint requests: Steven Port, MD, Aurora Cardiovascular Services, Advocate Aurora HealthCare,Milwaukee,WI, USA; [email protected] J Nucl Cardiol 2020;27:1246–8. 1071-3581/$34.00 Copyright 2019 American Society of Nuclear Cardiology.