BACKGROUND: The aims of this study were to determine the optimal tracer kinetic model for [(11)C]-meta-hydroxyephedrine ([(11)C]HED) and to evaluate the performance of several simplified methods.
METHODS: Thirty patients underwent dynamic 60-min [(11)C]HED scans with online arterial blood sampling. Single-tissue and both reversible and irreversible two-tissue models were fitted to the data using the metabolite-corrected arterial input function. For each model, reliable fits were defined as those yielding outcome parameters with a coefficient of variation (CoV) <25%. The optimal model was determined using Akaike and Schwarz criteria and the F-test, together with the number of reliable fits. Simulations were performed to study accuracy and precision of each model. Finally, quantitative results obtained using a population-averaged metabolite correction were evaluated, and simplified retention index (RI) and standardized uptake value (SUV) results were compared with quantitative volume of distribution (V T) data.
RESULTS: The reversible two-tissue model was preferred in 75.8% of all segments, based on the Akaike information criterion. However, V T derived using the single-tissue model correlated highly with that of the two-tissue model (r (2) = 0.94, intraclass correlation coefficient (ICC) = 0.96) and showed higher precision (CoV of 24.6% and 89.2% for single- and two-tissue models, respectively, at 20% noise). In addition, the single-tissue model yielded reliable fits in 94.6% of all segments as compared with 77.1% for the reversible two-tissue model. A population-averaged metabolite correction could not be used in approximately 20% of the patients because of large biases in V T. RI and SUV can provide misleading results because of non-linear relationships with V T.
CONCLUSIONS: Although the reversible two-tissue model provided the best fits, the single-tissue model was more robust and results obtained were similar. Therefore, the single-tissue model was preferred. RI showed a non-linear correlation with V T, and therefore, care has to be taken when using RI as a quantitative measure.
2014. Vol. 4, 52