Cerebral metabolic rate of glucose quantification with the aortic image-derived input function and Patlak method: numerical and patient data evaluation.

Abstract

Quantitative maps of cerebral metabolic rate of glucose (CMRGlu) from fluorine-18 fluorodeoxyglucose-PET are useful in brain studies, but are challenging to acquire because of technical constraints, which hinder their use in clinical routine. Aortic image-derived input functions (IDIFs) combined with Sokoloff's method were proposed as a suitable solution. However, Sokoloff's method requires the use of standard kinetic constants, which may produce biased estimates. Patlak's method would be more appropriate, but concern can arise when used with an aortic IDIF from unavailability of a complete brain curve acquired starting from injection. The aim of this study was to develop a CMRGlu quantification technique that combines Patlak's method with aortic IDIFs in a clinical setting. A simple acquisition protocol for aortic IDIF measurement was developed and applied on a sample of patients with different degrees of hypometabolism (one healthy control, four patients with a neurodegenerative condition, and one coma patient). CMRGlu estimates in vivo were obtained with both the Sokoloff method and the Patlak method. Computer simulations were performed to assess the causes of bias affecting Sokoloff and Patlak estimates and interpret the results obtained in patients. Simulations showed that Sokoloff's method is less stable than Patlak's method as the extent of bias changed across different physiological states, potentially leading to misinterpretation of clinical data. In clinical patients, Sokoloff and Patlak estimates were correlated on the whole, but deviations emerged for critical physiological states. CMRGlu quantification with the Patlak method and aortic IDIF is feasible, easy to implement in clinical practice, and superior to Sokoloff's method from a personalized medicine perspective.