Fritz-Hansen T. Quantification of the effect of water exchange in dynamic contrast MRI perfusion measurements

Measurement of myocardial and brain perfusion when using exogenous contrast agents (CAs) such as gadolinium-DTPA (Gd-DTPA) and MRI is affected by the diffusion of water between compartments. This water exchange may have an impact on signal enhancement, or, equivalently, on the longitudinal relaxation rate, and could therefore cause a systematic error in the calculation of perfusion (F) or the perfusionrelated parameter, the unidirectional influx constant over the capillary membranes (K i ). The aim of this study was to quantify the effect of water exchange on estimated perfusion (F or K i ) by using a realistic simulation. These results were verified by in vivo studies of the heart and brain in humans. The conclusion is that water exchange between the vascular and extravascular extracellular space has no effect on K i estimation in the myocardium when a normal dose of Gd-DTPA is used. Water exchange can have a significant effect on perfusion estimation (F) in the brain when using Gd-DTPA, where it acts as an intravascular contrast agent. Measurement of perfusion when using exogenous contrast agents (CAs) and MRI is affected by the diffusion of water between compartments. Several studies have indicated that this could affect the results of such perfusion measurements (1-6). However, contrast-enhanced MRI using T 1 -weighted imaging has been developed in order to measure perfusion or perfusion-related parameters of the heart in vivo without explicitly considering the effect of water exchange. Models of perfusion in animals and humans when using extravascular CAs such as gadolinium-DTPA (Gd-DTPA) or intravascular CA have been presented (7-12). These measurements are often based on inversion recovery turboFLASH (IR-turboFLASH) or saturation recovery turboFLASH (SAT-turboFLASH) imaging. Most of these studies measure the tissue enhancement curve and the arterial input function and rely on basic tracer kinetic models as formulated by Kety (13). In the ideal case, this approach gives the unidirectional influx constant K i (ml/100 g/min) for CA diffusion over the capillary membrane when extravascular CA is used and the perfusion F (ml/100 g/min) when intravascular CA is used. For extravascular CA, the perfusion and the unidirectional influx constant are related by K i ϭ EF There is now growing evidence that K i in particular provides valuable and clinically useful information related to perfusion (7-12,15). For example, it has been shown previously that infusion of dipyridamole in humans resulted in an increase of K i by a factor of 2-3 in the heart of healthy subjects. In patients with ischemic heart disease, K i was either unaffected or decreased after infusion of dipyridamole in areas supplied by an insufficient coronary artery Many assumptions need to be investigated in order to clearly associate directly the MRI-determined K i with the physiological parameter, the unidirectional influx constant of the CA diffusion over the capillary membrane in the heart. One of the main concerns is related to the fact that CA is measured indirectly through water relaxation. Water diffuses between compartments and this exchange of water between compartments can modify the MR signal enhancement of the tissue (1) and, therefore, influence the measured K i . Hence, water exchange can potentially affect the results obtained with both extravascular and intravascular CA. This study investigates quantitatively the effect of water exchange between intravascular and extravascular compartments and focuses on the effect on K i when using an extravascular CA and the effect on perfusion (F) when using an intravascular CA. The effect of water exchange was quantified by computer simulations. First, a simple approach was used by simulating two extreme situations: the fast water exchange regime, defined as 1/ ӷ ͉1/T 1b Ϫ 1/T 1myo ͉ and the no (or very slow) water exchange regime, defined as 1/ Ӷ ͉1/T 1b Ϫ 1/T 1myo ͉, where 1/ ϭ 1/ b ϩ 1/ myo with b and myo being the average residence times of water in the vascular and extravascular compartments, respectively, T 1b denotes blood longitudinal relaxation time, and T 1myo denotes myocardial longitudinal relaxation time (2). Second, using a specified MR sequence a two-site exchange model was incorporated into the perfusion model, allowing us to estimate how dependent K i or F was on the water exchange rate between compartments. In order to validate the predicted results of this model, we performed perfusion measurements using Gd-DTPA in the heart, where Gd-DTPA acts as an extravascular CA, and in the brain, where Gd-DTPA acts as an intravascular CA in normal volunteers (we did not have access to intravascular CA for the heart). The nomenclature in MR tracer kinetic models is somewhat confusing and in the following we will define transfer constants and concentrations carefully and adhere a