Assessment of intra- and interobserver reproducibility of rest and cold pressor test-stimulated myocardial blood flow with 13N-ammonia and PET

We investigated the intraobserver reproducibility of myocardial blood flow (MBF) measurements with PET at rest and during cold pressor test (CPT), and the interobserver agreement.Twenty normal volunteers were studied. Using (13)N-ammonia, MBF was measured at rest and during CPT and measurement was repeated in a 1-day session (short-term reproducibility; SR). After a follow-up of 2 weeks, MBF was measured again at rest and during CPT and compared with the initial baseline measurement (long-term reproducibility; LR). In addition, adenosine-induced hyperemic MBF increases were assessed.Assessment of the SR did not show a significant absolute difference in MBF at rest, MBF during CPT or the endothelium-related change in MBF from rest to CPT (DeltaMBF) (0.09 +/- 0.10, 0.11 +/- 0.09, and 0.08 +/- 0.05 ml/g/min; p = NS), and they were linearly correlated (r = 0.72, r = 0.76 and r = 0.84; p0.0001). Corresponding values for standard error of the estimate (SEE), as indicative for the range of MBF measurement error, were 0.14, 0.14, and 0.09 ml/g/min. The LR yielded relatively higher but non-significant absolute differences in the MBF at rest, MBF during CPT and DeltaMBF (0.10 +/- 0.10, 0.14 +/- 0.10, and 0.19 +/- 0.10 ml/g/min; p = NS), and paired MBFs significantly correlated (r = 0.75, r = 0.71, and r = 0.60; p0.001). Corresponding SEEs were 0.13, 0.15, and 0.16 ml/g/min. The interobserver analysis yielded a high correlation for MBF at rest, MBF during CPT, and hyperemic MBF (r = 0.96, SEE=0.04; r = 0.78, SEE=0.11; and r = 0.87, SEE=0.28; p0.0001, respectively), and also a good interobserver correlation for DeltaMBF (r = 0.62, SEE=0.09; p0.003).Short- and long-term MBF responses to CPT, as an index for endothelium-related coronary vasomotion, can be measured reproducibly with (13)N-ammonia PET. In addition, the high interobserver reproducibility for repeat analysis of MBF values suggests the measurements to be largely operator independent.