Response-Derived Input Function Estimation for Dynamic Contrast-Enhanced MRI Demonstrated by Anti-DLL4 Treatment in a Murine U87 Xenograft Model.

Abstract

Dynamic contrast-enhanced magnetic resonance imaging (DCE MRI) is an accepted method to evaluate tumor perfusion and permeability and anti-vascular cancer therapies. However, there is no consensus on the vascular input function estimation method, which is critical to kinetic modeling and K trans estimation. This work proposes a response-derived input function (RDIF) estimated from the response of the tumor, modeled as a linear, time-invariant (LTI) system. In an LTI system, an unknown input can be estimated from the system response. If applied to DCE MRI, this method would eliminate need of distal image-derived inputs, model inputs, or reference regions. The RDIF method first determines each tumor pixel's best-fit input function, and then combines the individual fits into a single input function for the entire tumor. The method was tested with simulations and a xenograft study with anti-vascular drug treatment. Simulations showed successful estimation of input function expected values and good performance in the presence of noise. In vivo,significant reductions in K trans and AUC occurred 2days following anti-delta-like ligand 4 treatment. The in vivo study results yielded K trans consistent with published data in xenograft models. The RDIF method for DCE analysis offers an alternative, easy-to-implement method for estimating the input function in tumors. The method assumes that during the DCE experiment, the changes observed by MRI result solely from vascular perfusion and permeability kinetics, and that information can be used to model the input function. Importantly, the method is demonstrated in a murine xenograft study to yield K trans results consistent with literature values and suitable for compound studies.