| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 University of Alabama at Birmingham School of Medicine, Birmingham, AL.
2 Department of Radiology, University of Alabama at Birmingham, 619 19th St.,
South, JTN363, Birmingham, AL 35249-6830.
3 Department of Surgery, Division of Urology, University of Alabama at
Birmingham, Birmingham, AL.
4 Department of Biostatistics, School of Public Health, University of Alabama at
Birmingham, Birmingham, AL.
Received March 24, 2008;
accepted after revision May 23, 2008.
Address correspondence to M. E. Lockhart
(mlockhart{at}uabmc.edu).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. The records of 173 renal donor candidates who had undergone CT and radionuclide renography between March 1, 2005, and February 28, 2006, were retrospectively evaluated; of those 173 patients, 152 met study inclusion criteria. A blinded investigator using 3D models that were created semiautomatically from the unenhanced, arterial, and excretory phase data made measurements of CT renal volumes and attenuations. The mean renal attenuation and volume were used to calculate the net accumulation of contrast material and split renal function for comparison with radionuclide renography. Split function from CT was calculated in the arterial and excretory phases as well as based on split renal volume and the Patlak method.
RESULTS. All four CT-based methods for the calculation of split renal function showed correlation with no significant difference from radionuclide renography (p > 0.05, Student's t test). Pearson's correlation coefficients varied from 0.36 to 0.63 (p < 0.001 for each). Difference scores revealed that the excretory and renal volume splits had the narrowest range and showed a linear, nonzero relationship to the renography splits. Bland-Altman analysis confirmed that the majority of difference scores between each CT method and the radionuclide renography were within the 95% CI of the differences.
CONCLUSION. Split renal function based on 3D CT models can provide a "one-stop" evaluation of both the anatomic and the functional characteristics of the kidneys of living potential kidney donors. The excretory phase data and the split renal volume data show the best correlation and the smallest difference scores compared with radionuclide renography data.
Keywords: 3D kidney modeling • kidney transplantation • renal artery • renography • split renal function
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Split renal function is then calculated by comparing the radioactive tracer accumulation from each side in the first 2 minutes divided by the total accumulation in both kidneys over the same period. Renography results have been shown to have a relatively wide variance in normal values because of anatomic variations in the location and depth of the kidneys, patient body habitus, and inconsistencies between examiners in the selection of the region of interest (ROI) and the time to peak enhancement, as well as a variety of algorithms for calculating renal clearance. The average variance between repeat scans by the same operator on consecutive days is ± 2%, but up to 8% variation has been reported in standardization trials [3, 4]. Other patient factors such as state of diuresis can compound this variation, further reducing the reproducibility of this method. However, radionuclide renography assessment remains the reference standard because of the lack of a suitable alternative.
Iohexol is a radiopaque IV-administered contrast agent that is freely filtered in the glomerulus, is neither secreted nor reabsorb ed in the renal tubules, and is physiologically similar to 99mTc MAG3 [5–7]. Consequently, iohexol is an ideal radiopaque contrast agent for the assessment of renal function. Because iodinated contrast agents are already used for CT evaluation of potential renal donors for surgical candidacy, the data acquired have the potential to be used, in addition, to evaluate split renal function [8]. In 1991, Miles [9] reported a method of evaluating renal tissue perfusion by CT. The method utilizes dynamic CT to quantify the accumulation of radiopaque contrast medium in renal tissue and determines functional information about the kidney using the attenuation–time curve [9]. Based on this method, the accumulation of contrast medium in each kidney can be estimated and the split function determined.
CT angiography (CTA) has been examined by several groups as a possible method for evaluating split renal function. These studies [10–14] were small, but each showed that various 2D CT methods for the calculation of split renal function produce results only slightly different from those obtained by the current radionuclide renography standard in populations with suspected or known renal disease, including renal artery stenosis, hydronephrosis, stones, renal artery embolization, renal tuberculosis, and hematuria. However, slow CT scanning speeds with older equipment as well as anatomic differences in the location of the renal artery origin, renal parenchymal size, and pathologic conditions of the kidneys make estimation of total renal volume and attenuation from individual 2D slices problematic.
With the advent of faster helical CT machines and 3D reconstruction techniques, faster and potentially more accurate measurements of the accumulation of contrast material in the kidneys are possible. However, the calculation of split function by 3D reconstruction has not been evaluated in a large-scale study of a healthy population, to our knowledge. The objective of this study was to examine the correlation between CT-based and radionuclide renography–based measures of split renal function in a healthy population of live potential kidney donors using 3D models generated from CTA.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
Iohexol 350 mg I/mL (Omnipaque, GE Healthcare) was the contrast agent injected IV according to a standardized protocol, with injection rates ranging between 3 and 5 mL/s according to examination features such as the gauge and location of the peripheral IV access in individual patients. Patients with partial studies due to unsuccessful administration of contrast material because of extravasation were not included. If the CT examination did not include the entire renal volumes or if there were any focal lesions in the renal parenchyma, the patient was excluded from the study. Patients with renal lesions were excluded primarily because the presence of indeterminate renal lesions is a criterion at our institution for temporary or permanent exclusion from donation regardless of the determination of split renal function by radionuclide renography.
Creating the 3D Kidney Model
A workstation (Advantage Workstation, version 4.1, GE Healthcare) was used
by one author to generate 3D CT models of the patients' kidneys. This
investigator was blinded to the radionuclide renography findings.
Three-dimensional tools in the commercial software package have an Add Object
function that was used to build a virtual representation of the kidneys, left
and right separately, for each scan series. This tool functions by
automatically filling a space that contains similar voxel values. Selecting a
piece of the renal parenchyma allows the software to identify the remaining
surrounding renal parenchyma but not adjacent fat or fluid in the renal
pelvis. It was sometimes necessary to use the opacity filters, color settings,
and cutting tool to free the margins of the 3D kidney from adjacent structures
of similar attenuation (e.g., liver, spleen, psoas muscle, intestine).
Structures in the 3D model with attenuation values different from that of the
renal parenchyma, such as the contrast-filled renal pelvis, could be
subtracted from the model with the Remove Object tool. The final 3D model of
each kidney was visually compared with the cross-sectional images to ensure
that only the renal parenchyma was included in the virtual representation
(Fig. 1A,
1B,
1C).
|
|
Calculating Split Function
Algebraic determination of split renal function followed the form put
forward by Frennby et al. [10]
in 1995, which we modified only to account for volume instead of
cross-sectional area. For each side, the difference between the mean
attenuation in the arterial series (AArt) and in the
unenhanced series (AUn) was multiplied by the mean of the
respective parenchymal volumes (VArt and
VUn) to measure the accumulation of contrast material in
each individual kidney during the arterial phase (AccArt)
as follows:
 |
Once this procedure was completed for both the left and right kidneys in
the arterial phase, the total arterial contrast accumulation on the right
(AccRArt) was divided by the sum of the total arterial
contrast accumulation in the right and left (AccLArt)
kidneys to determine the relative clearance of contrast medium from the right
kidney in the arterial phase as follows:
 |
These calculations were repeated using data from the excretory phase series to determine the right excretory split function. These data sets were generated from both the threshold-corrected data (to remove fat) and the uncorrected data. For the enhanced renal volume split function calculation, the volume (evaluated both with and without a threshold) of the right kidney in the excretory phase was divided by the total volume of the two kidneys (also in the excretory phase). The algebraic method of calculating split function yielded six data sets.
The two-point Patlak integral method has been studied in the determination of split function as well, and we also performed this calculation from our data [13]. Contrast clearance for each kidney is calculated using data from both the arterial and excretory phases. When calculating split function, because the clearance of the right kidney is divided by the sum of the left and right clearances, the integrals in the numerator and denominator cancel out. The Patlak analysis yields two more data sets, one with and one without thresholding for fat attenuation, for a total of eight distinct measures of split renal function generated from the same 3D models of the kidneys.
Data Analysis
The relationships between each method and radionuclide renography were
evaluated for systematic differences and also correlation. Paired two-tailed
Student's t tests were performed between each of the eight right
split function measures compared with the right split function reported by the
renography method. Pearson's correlation coefficients were also examined for
each data set relative to the renogram and for all data sets relative to the
creatinine clearance. For the Student's t test, a p value of
> 0.05 indicated no significant systematic difference between the
CT-measured and the radionuclide-determined split renal function. For
Pearson's correlation coeffici ents, a p value of < 0.05 was
considered statistically significant, indicating a linear relationship between
CT and radionuclide measures of split renal function or between the split
function measure and the creatinine clear ance. Linear regression model
equations were generated. Difference scores were computed for each patient to
compare each of the right split function obtained from CTA with the right
split function by renography. The difference score is equal to the
renography-measured split renal function minus the CT-measured split renal
function [13]. A Bland-Altman
analysis was gen erated for comparison of CT-based and radio nuclide
renography–based methods of assessing split renal function.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Paired-sample Student's t tests (Table 1) showed no significant systematic difference between any of our CT-based measures of split renal function and the standard radionuclide-generated split renal function (p > 0.05). The threshold-corrected arterial split renal function measurement, with a p value of 0.06, approached significance and tended to be higher than the other CT-based measurements. The mean difference score was approximately zero for all methods except the threshold-corrected arterial measurement, which had a mean difference score of –0.7.
|
All data sets correlated with the renography results as indicated by Pearson's correlation coefficients that were significantly different from zero (all, p < 0.001). However, the correlation coefficients for arterial split function were significantly less than those for excretory split renal volume and split renal function both with (p = 0.006) and without (p = 0.018) threshold correction.
More descriptive of the relationship between our CT-calculated measures of split renal function and the radionuclide renography measures of split renal function is a diagram showing the range of difference scores in relation to the radionuclide renogram split. Figure 3 shows a representative diagram of the difference scores plotted against the radionuclide standard split function—in this case, for the CT-calculated split function based on enhanced renal volume without threshold correction (n = 152) showing a linear, nonzero relationship. The CT-calculated split function was generally in the same direction, but lesser in degree, than the reference standard; that is, radionuclide renography values tended to be lower than the CT-calculated splits when the right kidney had less than 50% of the function and tended to be higher than the CT-calculated splits when the right kidney had more than 50% of the function, but the differences were small.
|
|
Correction equations based on regression models were generated for each data set to determine whether there was systematic vari ation that could be eliminated statistically, but the R2 values were not high enough to show that correction improved the data's inherent predictive power (Table 2).
|
There was no correlation between any of the split function data sets (including the renogram) and the creatinine clearance (p > 0.05) (Table 3).
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Improved software and faster CT scanning have made it possible to produce 3D representations of the kidneys from CT angiograms that measure both the attenuation and the volume semiautomatically. Faster scanners reduce the impact of perfusion on the continued accumulation of contrast material during scanning because the entire kidney can be scanned in less than 3 seconds. Another advantage of this method is less extrapolation than is needed using 2D approaches because the software allows the 3D model to be sampled simultaneously over the entire volume one time, instead of adding multiple independent ROIs. A 3D reconstruction method has been examined in one small study [14], and the results were within the standard error of the nuclear medicine split determination. Despite the limitations of that study [14], the authors still concluded that 3D determination of split renal function was as accurate as renography and was also much faster.
The results of our study show that 3D modeling of CTA data to determine split renal function yields results that are not different from renal functional assessment by radionuclide renography among a population of live potential renal donors. Furthermore, the renal volume split, calculated solely by measuring the volume of the renal parenchyma in renal donors without adjusting for contrast accumulation, is the most accurate measure of split renal function regardless of threshold correcting. The enhanced renal volumes used for this split were recorded from the excretory phase; data from this series were generally the most reproducible and easiest to create automatically in our experience. Because chronic glomerular damage that reduces renal filtration is typically associated with parenchymal scarring and volume loss, volume may prove to be the best and easiest measure of split renal function, but the approach lacks the direct physiologic measure that is the goal of this study. Using a mathematic method for calculating contrast accumulation from excretory phase images to determine the split function is slightly less accurate than renal volume split function, but it still is not statistically different from the renography and so allows a physiologically more satisfying method of assessment.
Using 3D modeling of the kidneys for the calculation of split renal function has multiple advantages over the current protocol for renal donor assessment. The use of one technique substantially reduces the cost of the evaluation. Conducting only one study also speeds up the evaluation process and reduces radiation exposure. The potential for automating the software to measure renal volumes and determine splits is very real; automation would greatly reduce the time and effort required to determine split renal function. CT-measured split renal function is also potentially more accurate than radionuclide renography–measured split function because the latter varies significantly with patient habitus, kidney position, and the operator. Modifying the current CT protocol for renal donor evaluation would produce all the information necessary for the selection of the best kidney for donation at reduced cost and reduced time and with equal or improved accuracy compared with the commonly used radionuclide renography method.
The Bland-Altman analysis attempts to compare the accuracy of two clinical measures—in this case, CT-based versus renography-based measures of split function—and shows a significant degree of concordance for patients with normal split renal function. However, it is difficult to directly compare the CT-based and renography-based measures of split function in the determination of split renal function in donors with a wide split function for a number of reasons. There is no universal cutoff value for split function in the evaluation of live renal donors. At our institution we generally use a 60/40 split as a relative contraindication to donation. Of six donor evaluations showing split function greater than or equal to 60/40 by renography, five were excluded. Three of the 152 candidates were excluded because of hypertension, malignancy, or multiple renal arteries and two were excluded solely on the basis of uneven split function. One patient with a 60/40 split was a successful donor. If we apply the same 60/40 limits from the renography to the CT data set for uncorrected contrast-enhanced volume split function, we identify only one of the six abnormal split functions identified by renography.
There are limitations of this study that deserve to be addressed in future work. The correlations in our study are not as high as those reported previously by other groups (range, 0.43–0.98) that were generated from 2D CT methods of data collection and analysis applied to a markedly abnormal patient population with known renal artery stenosis, hydronephrosis, hematuria, and other renal diseases [8, 10]. Our novel method for the determination of split renal function from enhanced renal volume compares slightly more favorably with renography than 2D CT measurements. The apparent low correlations with renography seen in our study may be due to our homogeneous population of live potential kidney donors and the inherent limitations of the retrospective nature of this study.
First, our population was found to have a mean and SD for the radionuclide split function more similar to those reported in radionuclide standardization studies than to other studies that focus on renal artery stenosis. The screening of renal donors is a multistep process and the determination of split renal function is the last step of that process; consequently, renal donors evaluated for split renal function are a relatively homogeneous group without significant medical history or metabolic or anatomic features that would make them poor donor candidates. In our sample, most donor candidates who were excluded because of lesions detected by CT were not further evaluated by nuclear renography, so CT data collected on those patients could not be used in the data analysis. The nine patients in this study with a renography-determined split function and abnormal findings on CT scans underwent nuclear renography before CT. Of note, recent guidelines published in the transplantation literature now make isolated simple renal cysts only a relative contraindication to donation [15]. It is possible that a broader population including those with known renal abnormalities could improve the correlation between CT-based and renography-based methods of determination of split function that we detected in the present study.
Second, we were restricted to standard data sets acquired for live renal donor evaluation at our institution, which include unenhanced, arterial, and excretory phases. Unenhanced images are considered essential by our surgeons for the detection of nephrolithiasis. Arterial images are used for the evaluation of the highly variable renal vasculature. Excretory phase data are acquired several minutes after contrast administration so that the contrast material has left the renal tubules and passed into the renal pelvis, ureters, and bladder for visualizing the urinary collecting system. In previous prospective studies, nephrographic phase scans allowed the precise determination of single-pass contrast accumulation in the renal parenchyma by acquiring images before excretion of contrast material from the renal tubules. Nephrographic phase images might have made our correlations compare more favorably with the correlations reported in other published studies. Adding a nephrographic phase scan to the CTA protocol for live renal donor evaluation would not extend the duration of the examination and might yield more accurate data, but at the cost of increased radiation exposure. Because each scan series is uniquely important in the evaluation of the renal donor, no series was considered expendable; however, based on our results, the determination of split function based on renal enhancement volume would not require an unenhanced scan and so could theoretically be used in other contexts for the determination of split renal function when anatomic considerations are not as critical as they are in live renal donor evaluations.
Furthermore, this 3D method may be lacking in its ability to identify subjects with wider split renal function, as shown by the linear trend of the difference scores, the flat slope of the correlation coefficient for all data sets, and the trend seen with the Bland-Altman analysis. The wider the split renal function was on radionuclide renography, the more the 3D CTA methods underestimated that magnitude. This may be due to autoregulation of GFR related to nephrotoxicity of iodinated IV contrast agents or again may be due to the use of excretory phase data instead of nephrographic phase data.
Despite the limitations of this study, determination of renal function split by semiautomated 3D volumetric analysis of CTA and nephrographic data sets remains a promising alternative to nuclear renography. In patients with normal renal function determined by serum chemistries and no apparent renal anomalies or anatomic defects on CT, 3D CTA is a suitable screening test for determining split renal function.
Acknowledgments
Special thanks to Trish Thurman for assistance in manuscript
preparation.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  P.-H. Vivier and J.-N. Dacher Pharmacokinetics of 99mTc-MAG3 Am. J. Roentgenol., October 1, 2009; 193(4): W361 - W361. [Full Text] [PDF]
|
|
|
|
|
|
A. L. Summerlin, M. E. Lockhart, and J. K. Smith Reply Am. J. Roentgenol., October 1, 2009; 193(4): W362 - W362. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
