Noninvasive assessment of renal allograft function with multiparametric MRI: DWI, DTI, and arterial spin labeling compared with eGFR.

PURPOSE: To evaluate diffusion-weighted imaging (DWI), diffusion tensor imaging (DTI), and arterial spin labeling (ASL) for noninvasive assessment of renal allograft function using estimated glomerular filtration rate (eGFR) as the reference.

METHODS

Eighty-four kidney transplant recipients and 30 age-matched healthy volunteers underwent mono-exponential DWI, DTI, and pseudocontinuous ASL on a 1.5-T MRI scanner. Cortical and medullary apparent diffusion coefficient (ADC), fractional anisotropy (FA), and cortical renal blood flow (RBF) were measured. Recipients were grouped by eGFR, and intergroup differences, correlations with eGFR, and receiver operating characteristic performance were analyzed.

RESULTS

Cortical ADC, medullary ADC, and cortical RBF decreased with worsening allograft function and correlated positively with eGFR (r = 0.513, 0.489, and 0.615, respectively; all p < 0.001). Cortical and medullary FA were not significantly correlated with eGFR. Cortical RBF showed the highest single-parameter area under the curve for differentiating group 1 from group 2 (0.919; sensitivity, 95.7%; specificity, 71.4%) and group 1 from group 3 (0.969; sensitivity, 85.7%; specificity, 100%). When distinguishing group 1 from group 2, the AUCs obtained by combining cortical RBF with cortical ADC or medullary ADC, as well as the combination of all three parameters, were significantly higher than those of cortical ADC alone or medullary ADC alone (P < 0.05).

CONCLUSION

DWI-derived ADC and ASL-derived cortical RBF are useful noninvasive MRI markers of renal allograft function, with cortical RBF showing the strongest diagnostic performance. DTI-derived FA appeared exploratory in this cohort.