Absolute Quantitation of Cardiac ^99mTc-Pyrophosphate Using Cadmium-Zinc-Telluride-Based SPECT/CT

The primary aims of this study were to determine the correlation between absolute quantitative ^99mTc-pyrophosphate metrics and traditional measures of cardiac amyloid burden and to measure the intraobserver repeatability of the quantitative metrics. Methods: We studied 72 patients who underwent ^99mTc-pyrophosphate SPECT/CT using a novel general-purpose cadmium-zinc-telluride-based SPECT/CT system. The clinical standard for these studies is visual grading (with grades of 0, 1, 2, and 3 indicating myocardial uptake absent, less than rib uptake, equal to rib uptake, or more than rib uptake, respectively). A visual grade of 2 or more was considered positive. For 72 patients, SUV_max, SUV_mean, cardiac amyloid activity (CAA; i.e., SUV_mean × left ventricular [LV] volume), and percentage injected dose (%ID) were calculated, and visual grading was performed. The correlation was determined between the 4 quantitative metrics or visual grades and the LV mass index (LVMI) (indexed to body surface area on echocardiography, 67 patients). For a subset of 11 patients, the correlation was determined between the visual or quantitative metrics and the extracellular volume (ECV) on cardiac MRI. Normal linear regression was used to compare the standardized association of each of the 4 quantitative metrics with LVMI, as a surrogate for amyloid burden. Receiver-operating-characteristic curve analysis was used to determine the diagnostic accuracy of quantitative metrics, using visual grading as the reference standard. The intraobserver repeatability of generating quantitative metrics was also determined. Results: All 4 quantitative metrics were highly accurate, with an area under the receiver-operating-characteristic curve of more than 0.96 for diagnosis of transthyretin cardiac amyloidosis. SUV_max, SUV_mean, CAA, %ID, and visual grade were moderately positively correlated with LVMI (r = 0.485 for %ID) and strongly positively correlated, albeit in a small cohort, with ECV (r = 0.873, SUV_max). Intraobserver repeatability was excellent, with less than a 2% coefficient of variation for SUV_max, %ID, and CAA and 3.8% for SUV_mean All 4 quantitative metrics had a standardized effect of more than 0.324 on LVMI; the largest standardized effect was 0.485, for %ID. Conclusion: In this first (to our knowledge) study of ^99mTc-pyrophosphate cardiac imaging using a novel cadmium-zinc-telluride SPECT/CT scanner, SUV_max, SUV_mean, CAA, and %ID measured by absolute quantitation of ^99mTc-pyrophosphate were moderately correlated with LVMI and strongly correlated, albeit in a small cohort, with ECV. The intraobserver repeatability of generating the quantitative metrics was excellent.