Holmes SP, Schmiedt CW, Brown SA.
IntroductionPurpose: Glomerular filtration rate (GFR) is considered the gold standard measure of renal function. Single kidney GFR (SK-GFR) is often more clinically relevant, but also more challenging to acquire in a reliable and easily applicable manner. Nuclear scintigraphy and computed tomography, respectively, have been used clinically and in research in veterinary medicine. Both have recognized limitations. Magnetic resonance (MR) renography offers superior soft tissue contrast, corticomedullary volumes of each kidney, individual renal blow flow measures, and dynamic contrast-enhancement (DCE) has been shown to accurately estimate global and SK-GFR in human patients. We hypothesize that global and SK-GFR could accurately be estimated by monitoring gadolinium MR contrast transit using a simplified multicompartment model.
Methods: Clinically normal research dogs (n=8) were anesthesized for imaging in a GE 3.0 Tesla Signa HDx twin gradient MRI. A coronal dynamic three-dimensional fast spoiled gradient echo (3D-FSPGR) sequence was used for MR renography, with the following parameters: TRITE/flip angle = 11.7 msec/1.4 msec/25°, matrix of 128 x 80, slice thickness 5 mm and 128 phases. After sequence initiation, a bolus injection of 0.1mmol/kg of Gd- DTPA at 2 mils (followed by a 20-ml saline flush) was administered. GFR was estimated based on signal intensity-time calculations and using a 3-compartment model described by Lee et al. (AJP Renal. 292: F1548-59. (2007)) and Zhang et al. (Magn Reson Med. 59: 278- 88 (2008)). Four-hour iohexol clearance assessment was started simultaneously for estimation of global GFR (l-GFR).
Results: MR-GFR scan time was approximately 7 minutes. The diagnostic quality was considered good in 7/8 dogs and non-diagnostic in 1/8 dogs. Image noise and patient motion made GFR estimation impossible in 1 dog. The average right and left kidney GFRs were 3.65 ± 3.59 and 2.79 ±1.95 ml/min/kg; these were not significantly different (p = 0.30). The mean global MR-GFR was 6.45 ± 5.41 ml/min/kg, whereas the mean globall-GFR was 3.12 ± 0.76 ml/min/kg. The global MR-GFRs were not significantly different from lohexol- based global GFR (p = 0.13), but the Pearson’s correlation poor (r = 0.51). The MR-GFR mildly overestimated the l-GFR in 3 dogs and mildly underestimated it in 2 dogs. There were 2 dogs where the MR-GFR (range: 10.67-16.34 ml/min/kg) markedly overestimated the l-GFR (range: 3.297 – 3.626 ml/min/kg).
Discussion/Conclusion: Accurate MR-GFR may not be possible on all MR units. Our MR acquisition technique did not completely match that described by Lee et al. (2007), because the MRI lacked parallel imaging capabilities. Therefore, the first-pass peak of the arterial input function (AIF) could not be captured and is likely our most significant contributor to the overestimation of GFR. Motion and small patient size may have been contributing factors.