Quantifying dynamic contrast-enhanced MRI of the knee in children with juvenile rheumatoid arthritis using an arterial input function (AIF) extracted from popliteal artery enhancement, and the effect of the choice of the AIF on the kinetic parameters.

Quantification of dynamic contrast-enhanced (DCE) MRI based on pharmacokinetic modeling requires specification of the arterial input function (AIF). A full representation of the plasma concentration data, including the initial rise and decay parts, considering the delay and dispersion of the bolus contrast is important. This work deals with modeling of DCE-MRI data from the knees of children with a history of juvenile rheumatoid arthritis (JRA) by using an AIF extracted from the signal enhancement data from the nearby popliteal artery. Three models for the AIFs were considered: a triexponential (AIF1), a gamma-variate plus a biexponential (AIF2), and a biexponential (AIF3). The pharmacokinetic parameters obtained from the model were Ktrans', kep, and V'p. The results from AIF1 and AIF2 showed no statistically significant difference. However, some statistically significant differences were seen with AIF3, particularly for parameters Ktrans' and V'p in the synovium (SNVM). These results suggest the importance of obtaining an appropriate AIF representation in pharmacokinetic modeling of JRA. Specifically, the initial rising part of the AIF should be incorporated for optimal pharmacokinetic modeling results. The pharmacokinetic parameters (mean+/-SD) derived from AIF1, using the average plasma concentration data, were as follows: SNVM Ktrans'(min-1)=0.52+/-0.34, kep(min-1)=0.71+/-0.39, and V'p=0.33+/-0.16, and for the distal femoral physis (DFP) Ktrans'(min-1)=1.83+/-1.78, kep(min-1)=2.65+/-1.80, and V'p=0.46+/-0.31. The pharmacokinetic parameters in the SNVM may be useful for investigating activity and therapeutic efficacy in studies of JRA. Longitudinal studies are necessary to find or demonstrate the parameter that is more sensitive to disease activity.