Reliable quantification of myocardial sympathetic innervation and regional denervation using [ 11 C]meta-hydroxyephedrine PET

Purpose Cardiac sympathetic nervous system (SNS) dysfunction is associated with poor prognosis in chronic heart failure patients. This study characterized the reproducibility and repeatability of [ 11 C]meta-hydroxyephedrine (HED) positron emission tomography (PET) quantification of cardiac SNS innervation, regional denervation, and myocardial blood flow (MBF). Methods Dynamic HED PET-CT scans were performed 47 ± 22 days apart in 20 patients with stable heart failure and reduced ejection fraction. Three observers, blinded to clinical data, used FlowQuant® to evaluate the test-retest repeatability and inter- and intra-observer reproducibility of HED tracer uptake and clearance rates to measure global (LV-mean) retention index (RI), volume of distribution ( V T ), and MBF. Values were also compared with and without regional partial-volume correction. Regional denervation was quantified as %LV defect size of values < 75% of the LV-maximum. Test-retest repeatability and observer reproducibility were evaluated using intra-class-correlation (ICC) and Bland-Altman coefficient of repeatability (NPC). Results Intra- and inter-observer correlations of both V T and RI were excellent (ICC = 0.93–0.99). Observer reproducibility (NPC = 3–13%) was lower than test-retest repeatability (NPC = 12–61%). Both regional (%LV defect size) and global (LV-mean) measures of sympathetic innervation were more repeatable using the simple RI model compared to V T (NPC = 12% vs. 19% and 30% vs. 54%). Using either model, quantification of regional denervation (defect size) was consistently more reliable than the global LV-mean values of RI or V T . Regional partial-volume correction degraded repeatability of both the global and regional V T measures by 2–12%. Test-retest repeatability of MBF estimation was relatively poor (NPC = 30–61%) compared with the RI. Conclusions Quantitative measures of global and regional SNS innervation were most repeatable using the simple RI method of analysis compared with the more complex V T . Observer variability was significantly lower than the test-retest repeatability using a highly automated analysis program. These results support the use of the simple RI method for reliable analysis of HED PET images in clinical research studies for future evaluation of new therapies and for risk stratification in patients with heart failure.