Abstract
Here we assessed the diagnostic value of a quantitative multiparametric magnetic resonance imaging (mpMRI) protocol for evaluation of renal allograft dysfunction with fibrosis. Twenty-seven renal transplant patients, including 15 with stable functional allografts (eGFR mean 71.5 ml/min/1.73m(2)), and 12 with chronic dysfunction/established fibrosis (eGFR mean 30.1 ml/min/1.73m(2)), were enrolled in this prospective single-center study. Sixteen of the patients had renal biopsy (mean 150 days) before the MRI. All patients underwent mpMRI at 1.5T including intravoxel-incoherent motion diffusion-weighted imaging, diffusion tensor imaging, blood oxygen level dependent (BOLD R(2)*) and T(1) quantification. True diffusion D, pseudodiffusion D*, perfusion fraction PF, apparent diffusion coefficient (ADC), fractional anisotropy (FA), R(2)* and T(1) were calculated for cortex and medulla. ΔT(1) was calculated as (100x(T(1) Cortex-T(1) Medulla)/T(1) Cortex). Test-retest repeatability and inter-observer reproducibility were assessed in four and ten patients, respectively. mpMRI parameters had substantial test-retest and interobserver repeatability (coefficient of variation under 15%), except for medullary PF and D* (coefficient of variation over 25%). Cortical ADC, D, medullary ADC and ΔT(1) were all significantly decreased, while cortical T(1) was significantly elevated in fibrotic allografts. Cortical T(1) showed positive correlation to the Banff fibrosis and tubular atrophy scores. The combination of ΔT(1) and cortical ADC had excellent cross-validated diagnostic performance for detection of chronic dysfunction with fibrosis. Cortical ADC and T(1) had good performance for predicting eGFR decline at 18 months (4 or more ml/min/1.73m(2)/year). Thus, the combination of cortical ADC and T(1) measurements shows promising results for the non-invasive assessment of renal allograft histology and outcomes.