Skip Navigation

Determination of the Confidence Interval for Calculations of Glomerular Filtration Rate Following Infusion of Gadolinium Chelate and MR Imaging

Author: Nivedita Raghunath, MS

Management of patients with impaired renal function is significantly limited by the lack of a non-invasive test to follow renal function, diagnose dysfunction,or monitor treatment response. Techniques for determining glomerular filteration rate (GFR) and renal blood flow (RBF) using MRI combined with dynamic imaging of the kidney following administration of gadolinium-chelate (Gd) have been developing. Gd is a filtered agent and can be used for determining GFR. We have developed a compartment model to consider the temporal and spatial extent of Gd in the kidney. For the derived GFR and RBF to be effectively used in clinical practice, the reliability of the calculation must be known. This work was performed to determine the 68% and 95% confidence intervals (CI) for the GFR when using 30 and 90s constant infusions of Gd.

Highly accelerated renal perfusion imaging was developed using a 3D spoiled gradient echo technique with fat saturation and centric-radial k-space acquisition.

 

Renal Perfusion 4D imaging: Selected images from a series of 30 slices per volume at 0.9 sec per scan 3mm voxels

Download AVI file

Figure 1

The kidneys and descending aorta (not shown) were segmented to derive arterial blood and total kidney time activity curves.

Figure 3

Typical arterial, left and right kidney curves resulting from a constant infusion of Gd over 30s.

Gd was infused at a constant rate over either 30 or 90s.GFR was determined from the model and the CI for each patient GFR was calculated using the BCa bootstrap method. The method creates similar datasets by flipping data points across the model estimate by randomly changing the sign of 50% of the residuals.1000 such datasets were created for each kidney and aorta curve from each patient and 1000 GFR estimates were determined from curve fitting. The intervals that encompass 68% and 95% of the values were determined with the BCa method.

We performed analysis on 15 subjects (left and right kidneys) with 90s (n=8) and 30s infusions (n=7).The CI was much better for the 30s infusion. Most likely, GFR is better identified with a 30s infusion because the wash-in and wash-out phases are about equal in duration as opposed to the 90s infusion where Gd continues to accumulate over most of the study. A 30s infusion of Gd is preferred. When so done, the GFR per kidney can be calculated with a precision of approximately 6 ml/min, which is sufficient for a very useful clinical test.

Team members: John Votaw, PhD, Nivedita Raghunath, MS, Courtney Mitchell, BS, Khalil Salman, MD, Puneet Sharma, PhD, Diego R. Martin, MD, PhD, John Carew, PhD

Team Members

John Votaw, PhD, Nivedita Raghunath, MS, Courtney Mitchell, BS, Khalil Salman, MD, Puneet Sharma, PhD, Diego R. Martin, MD, PhD, John Carew, PhD