Peripheral endocannabinoid microdialysis: in vitro characterization and proof-of-concept in human subjects

In vivo endocannabinoid (EC) microdialysis has only seldom been performed, mostly in rodent brain tissue. Low solubility in aqueous media, adsorption to surfaces, and instability with co-present human serum albumin (HSA) are the major obstacles in EC microdialysis. The addition of hydroxypropyl-ß-cyclodextrine (HPCD) to the perfusion fluid has been previously described to facilitate lipid microdialysis, but the general biophysical properties of HPCD, especially with respect to peripheral EC microdialysis, have not been described before. We report on the characterization of EC microdialysis using an in vitro system using Ringer's solution with 10% HPCD as the perfusion fluid and with fatty acid-free HSA as the matrix fluid. The endocannabinoids anandamide (AEA) and 2-arachidonoyl glycerol (2AG) were measured using LC-MS/MS. AEA was stable in the perfusion and matrix fluids, whereas 2AG was only stable in the perfusion fluid. In the matrix fluid, 2AG underwent rapid isomerization to 1-arachidonoyl glycerol. A relative recovery of 3.5% for AEA was found with 10% HPCD in the perfusion fluid and a flow rate of 1 μL/min. For 2AG, a similar relative recovery of 3.5% was estimated. Since 2AG was found unstable in the matrix fluid, a reliable calculation of the relative recovery rates was not possible. Delivery and recovery experiments revealed unequal inward and outward EC transport across the microdialysis membrane. Contrary to usual microdialysis findings, we observed increasing recovery rates for AEA with increasing flow rates. Long equilibration times of several hours were necessary to obtain constant relative recovery rates. In a proof-of-concept study in humans, we collected AEA from subcutaneous abdominal adipose tissue employing the described methodology. Our study suggests that the microdialysis technique is not suitable for the exact quantification of tissue EC concentrations, but it allows for their rough estimation.