Abstract
Measuring glomerular filtration rate (GFR) by dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) as part of standard of care clinicalMRIexams (e.g., in pediatric solid tumor patients) has the potential to reduce diagnostic burden. However, enthusiasm for this relatively newGFRtest may be curbed by the limited amount of cross-calibration studies with referenceGFRtechniques and the vast variety ofMRtracer model algorithms causing confusion on the choice of model. To advanceMRI-basedGFRquantification via improvedGFRmodeling and comparison with associated(99m)Tc-DTPA-GFR, 29 long-term Wilms' tumor survivors (19.0-43.3 years, [median 32.0 ± 6.0 years]) treated with nephrectomy, nonnephrotoxic chemotherapy ± radiotherapy underwentMRIwith Gd-DTPAadministration and a(99m)Tc-DTPA GFRtest. ForDCE-MRI-basedGFRestimation, a subject-specific two-compartment (SS-2C) model was developed that uses individual hematocrit values, automatically defines subject-specific uptake intervals, and fits tracer-uptake curves by incorporating these measures. The association between reference(99m)Tc-DTPA GFRandMR-GFRs obtained bySS-2C, three published 2C uptake, and inflow-outflow models was investigated via linear regression analysis. Uptake intervals varied from 64 sec to 141 sec [96 sec ± 21 sec] and hematocrit values ranged from 30% to 49% [41% ± 4%]; these parameters can therefore not be assumed as constants in 2C modeling. OurMR-GFRestimates using theSS-2C model showed accordingly the highest correlation with(99m)Tc-DTPA-GFRs (R(2) = 0.76,P < 0.001) compared with other models (R(2)-range: 0.36-0.66). In conclusion,SS-2C modeling ofDCE-MRIdata improved the association betweenGFRobtained by(99m)Tc-DTPAand Gd-DTPA DCE-MRIto such a degree that this approach could turn into a viable, diagnosticGFRassay without radiation exposure to the patient.