| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Genitourinary Imaging |
1 From the Department of Diagnostic Radiology, Justus-Liebig Universität Giessen, Klinikstr 36, 35385 Giessen, Germany. Received October 3, 2002; revision requested November 27; final revision received March 30, 2003; accepted May 19. Supported by a grant from Schering, Berlin, Germany. Address correspondence to A.C.L. (e-mail: alexander.langheinrich@radiol.med.uni-giessen.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Fifty patients without acute renal disorder were included. GFR was calculated by using a two-point Patlak plot technique. The computed tomography (CT) protocol consisted of a plain examination followed by two contrast material–enhanced examinations in the arterial and portovenous phase. Each examination included the entire kidneys and was performed after injection of 120 mL iopromide and 300 mg of iodine per milliliter given per 75 kg of body weight. All examinations were performed with a standard abdominal protocol. Section thickness was 4 x 2.5 mm, and table advance was 12.5 mm. Bolus triggering commenced 10 seconds after the start of contrast medium injection. Twelve dynamic scans were obtained with reduced tube current every 3 seconds to obtain sufficient arterial input function data. Correction for hematocrit level was made by using the unenhanced attenuation of the aorta. As a reference method, plasma clearance of the contrast medium injected for CT was calculated from three iodine plasma concentration measurements obtained 3, 4, and 5 hours after injection. Linear correlation was performed.
RESULTS: GFR was calculated from CT data in 48 patients. Two patients were excluded because of breathing errors. Mean GFR was 80 mL/min (range, 17–153 mL/min) as measured with iopromide plasma clearance and 82 mL/min (range, 28–148 mL/min) as measured with CT. Linear correlation between the two methods was r = 0.889; GFR calculated with the two-point Patlak plot was equal to 15 plus 0.83 times GFR (plasma clearance). The mean difference between GFRs as determined with the two methods was -1.2 mL/min (95% CI: -27.1, 24.6).
CONCLUSION: Total GFR can be measured accurately with minimally extended triphasic CT in patients without acute renal disorder by using a two-point Patlak plot technique.
© RSNA, 2003
Index terms: Computed tomography (CT), multi–detector row, 81.12111, 81.12115 • Kidney, CT, 81.12111, 81.12115 • Kidney, function
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
We introduced a new practical approach for measuring single-kidney GFR with the Patlak plot by performing multiphasic diagnostic CT (7,8). In the present study, we examined a modification of this technique. Only two points in the Patlak plot were calculated from the arterial and parenchymal scans. GFR was calculated by using this two-point Patlak plot. Thus, the purpose of our study was to determine the accuracy of the two-point Patlak plot in the calculation of GFR.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Fifty adults without acute renal disorder were included (21 women and 29 men). Patients underwent a CT examination between July 2001 and January 2002. The age range was 26–94 years (mean age, 64 years). Eight patients were younger than 50 years. Reasons for referral for CT were as follows: exclusion of abdominal tumor (n = 26), staging of colonic carcinoma (n = 2), and evaluation of lung cancer (n = 2) or carcinoma of the bladder (n = 1), breast, (n = 1) or ovary (n = 1). Seven patients had chronic pancreatitis, one had diverticulitis, and one had a retroperitoneal abscess. The remaining eight patients were referred for exclusion of abdominal abscess or inflammation.
Seven patients had diabetic nephropathy, and four had chronic renal failure that was due to arterial hypertension or an unknown cause. Many patients had diabetes mellitus (n = 12), arterial hypertension (n = 14), or cardiac insufficiency (n = 5). One patient previously underwent nephrectomy because of renal cell carcinoma. Two patients were undergoing chemotherapy with carboplatin and vepesid.
Contrast Medium Injection
All patients received one injection of 120 mL iopromide (Ultravist 300; Schering, Berlin, Germany) (300 mg of iodine per milliliter given per 75 kg of body weight). In all patients, the contrast medium injection lasted 40 seconds, with a flow rate of 3 mL/sec per 75 kg of body weight through a cannula that was placed in the forearm. A CT power injector (EnVision; Medrad, Indianola, Pa) was used for the injections.
GFR Determination with Plasma Clearance of Iopromide
Plasma clearance of iopromide was used as the reference value for GFR. The clearance was calculated by dividing the injected dose by the area under the plasma concentration curve (9). As shown by Bröchner-Mortensen (10) and Brown and O’Reilly (11), the integral of the plasma concentration curve can be accurately estimated by measuring at least two plasma concentrations approximately 3 and 4 hours after contrast medium injection and by applying a correction for the early part of the plasma concentration curve. The later phase is approximated by a single-exponential function.
In the present study, the gradient and intercept of the exponential function were calculated from three plasma samples that were obtained 3, 4, and 5 hours after injection of contrast material. For correction of the early component, the clearance was calculated in accordance with a calculation used by Bröchner-Mortensen (10).
A baseline blood sample was obtained just before CT to exclude baseline plasma iodine concentration. All blood samples were obtained by using a different cannula than the one used for contrast medium injection. The blood was centrifuged, and the iodine concentration in the plasma was measured by using an x-ray fluorescence analyzer (EG&G Ortec, Munich, Germany).
Determination of Hematocrit Level
The hematocrit level of all patients was determined with a blood sample obtained just before CT. By comparing this laboratory-determined hematocrit level with unenhanced aortic attenuation numbers, an equation for determination of hematocrit level from unenhanced attenuation of the aorta was calculated.
GFR Determination with Triphasic CT
CT was performed with a four–detector row CT scanner (Somatom Plus 4 Volume Zoom; Siemens, Erlangen, Germany). The CT protocol consisted of an unenhanced examination followed by two contrast medium–enhanced examinations in the arterial and portovenous phase, and each included the entire kidneys. The arterial examination was started when bolus-triggering scans showed an aortic attenuation of 300 HU; alternatively, the arterial examination was started 20 seconds after the initial rise of aortic attenuation. All examinations were performed by using our standard abdominal protocol with section thickness of 4 x 2.5 mm, table feed of 12.5 mm, tube voltage of 120 kV, and tube current of 205 mAs.
Bolus triggering commenced 10 seconds after the start of contrast medium injection and was used to obtain optimum contrast in the arterial phase for diagnostic purposes and to provide information about arterial input function.
In addition to the routine triphasic CT protocol, 12 dynamic scans were obtained between the arterial and parenchymal scans to gain sufficient data about arterial input function.
The bolus-triggering scans and the additional scans were obtained with a section thickness of 10 mm at the level of the renal hilum without table feed. Interscan delay was 3 seconds. Tube voltage was 120 kV, and tube current was 20 mAs for bolus-triggering and 40 mAs for dynamic examinations. Complete examamination of both kidneys lasted approximately 5–6 seconds for each of the three examinations (unenhanced, arterial phase, and parenchymal phase).
Measurement of Whole Kidney Attenuation
Mean attenuation and area of the left and the right kidney were measured by drawing a region of interest on every CT section. All regions of interest were drawn by the same investigator (C.W). The region of interest was drawn as close as possible to the kidney surface. Only the renal parenchyma was included; larger cysts and renal hilum with fat and vessels were excluded. Whole kidney attenuation K was calculated separately for both kidneys by multiplying mean attenuation (in Hounsfield units per cubic millimeter) by area (in square millimeters) and thickness (in millimeters) for each kidney section. K was calculated from the unenhanced scan and the arterial and parenchymal phase scans.
The net attenuation of the right and the left kidney was calculated by subtracting the kidney attenuation before contrast material injection from the kidney attenuation after contrast medium injection. K(t) is the net attenuation of one kidney at time and is measured in Hounsfield units.
Determination of the Aortic Attenuation Curve
The aortic attenuation curve b(t)—also known as the input function—has to be determined for calculation of GFR calculated with the two-point Patlak plot. Four parts of the aortic attenuation curve were measured on CT images, including images obtained with bolus triggering and with dynamic techniques and images obtained during the arterial phase and during the parenchymal phase. Aortic attenuation was measured on all these images by placing a circular region of interest inside the aortic lumen. The unenhanced aortic attenuation was measured on five CT scans at the level of the kidneys and subtracted from the enhanced aortic attenuation numbers. Figure 1 shows the aortic attenuation curve of one patient. The missing parts of the curve were interpolated linearly.
|
|
It is assumed that the amount of contrast medium in the vascular space B(t) is proportional to the concentration of the contrast medium in the aorta b(t) (measured in Hounsfield units):
|
|
|
Dividing Equation (4) by the aortic attenuation b(t) results in the Patlak plot function (Fig 2):
|
|
|
|
A correction for hematocrit level had to be made for the calculation of GFR with the Patlak plot method. hctCT represents hematocrit level calculated with Equation (8) by using the individual unenhanced aortic attenuation measured with CT. A correlation was calculated between individual unenhanced aortic attenuation bun measured by using CT and individual hematocrit level measured by using the blood specimen, resulting in
|
|
So far, all calculations have been performed separately for each kidney, resulting in a single kidney GFR. For comparison with plasma clearance, GFR of the left and the right kidneys as calculated with the two-point Patlak plot was added.
GFR calculated with the two-point Patlak plot and iopromide plasma clearance were calculated independently by using the methods described above.
Statistical Analysis
Total GFR, as measured with plasma clearance and the two-point Patlak plot, is given as mean ± SD. As the purpose of the present study was to determine the accuracy with which GFR can be measured with a CT-based Patlak plot technique compared with plasma clearance, a Bland-Altmann statistic was calculated. For each patient, we calculated the difference between GFR measured with plasma clearance and GFR calculated with the Patlak plot from CT data. From this data, mean difference d and SD were calculated; furthermore the "limits of agreement" are given as mean difference ± 2 SD. The Spearman rank correlation coefficient was calculated for correlation of the two methods used to measure total GFR. Because a correlation between plasma clearance and GFR as calculated with the two-point Patlak plot was already shown in previous studies, a P value for statistical significance was not defined.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The mean initial increase of the aortic attenuation curve occurred 21 seconds after injection (range, 10–37 seconds). The mean start of arterial imaging was 30 seconds after the initial increase of the aortic attenuation curve (range, 22–42 seconds). The mean start of dynamic imaging (12 scans times 3 seconds delay before each single dynamic scan = 36 seconds) was 49 seconds after the initial increase of the aortic attenuation curve (range, 39–63 seconds). The mean start of parenchymal imaging was 102 seconds after the initial increase of the aortic attenuation curve (range, 91–120 seconds).
Mean plasma clearance was 80 mL/min (range, 17–153 mL/min), and mean GFR as calculated with the two-point Patlak plot was 82 mL/min (range, 28–148 mL/min). Linear correlation between the two methods was r = 0.889; GFR calculated with the two-point Patlak plot was equal to 15 plus 0.83 times GFR (plasma clearance). Figure 3 shows the correlation between the two methods. Figure 4 illustrates the results from the Bland-Altmann test.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Dawson and Peters (3) showed that single-kidney GFR can be measured with contrast-enhanced CT by using the Patlak plot model. The technique used was single-location dynamic CT. After intravenous injection of 40 mL of contrast medium, a single CT scan was obtained every 5 seconds for approximately 2 minutes. Omitting the earliest scans, they derived Patlak plots with 20 single data points. GFR was then calculated in milliliters per minute, per milliliter of renal tissue, by determining the slope of a regression line through these multiple data points (3). Tsushima et al (14) measured whole-kidney GFR by using the same single-location dynamic CT technique and multiplying relative GFR per milliliter of renal tissue by whole kidney volume, which was obtained with planimetry from a separate whole-kidney scan. They investigated 24 patients with diabetes and compared GFR measured at CT with GFR measured by means of 24-hour creatinine clearance. They found a correlation of r = 0.87 and y = 29.2 + 0.64x.
In 2001, Tsushima et al (15) compared single-kidney GFR measured by using single-location dynamic CT and multiplied by renal volume with split renal function measured by using technetium 99m (99mTc) diethylenetetraaminepentaacetic acid (DTPA) scintigraphy (n = 24) and total GFR measured by using 24-hour creatinine clearance (n = 12). Seventeen patients had hydronephosis, and the other patients had arterial hypertension, diabetic nephropathy, or atrophic kidneys. They found nearly identical split renal function when they compared CT scans with scintigrams (r = 0.97, y = 2 + 0.98x, n = 24). The correlation between total GFR measured with the Patlak plot technique and 24-hour creatinine clearance was also close (r = 0.92) The equation of the linear correlation was not given in the article.
All these studies included single-location dynamic CT as an extra investigation that was performed in addition to routine abdominal CT. It seemed difficult to us to include single-location dynamic CT in a routine abdominal CT examination. We thought that several breathing maneuvers during dynamic CT scanning for about 2 minutes might produce a degree of error, as the exact breathing position might differ between two distinct inspirations. Measuring GFR with single-location dynamic CT necessitates the extrapolation of relative GFR per milliliter of renal tissue to the whole kidney volume, which might introduce an error in inhomogeneous kidney disease.
In two preceding studies, we introduced a practical approach to the measurement of single-kidney GFR with the Patlak plot model. By imaging the entire kidneys three times after contrast medium injection, only three data points for the Patlak plot were measured. The measurement was performed in patients who underwent CT for clinical reasons (7,8). Fifty patients without acute renal disease were investigated. GFR, as determined with CT, was compared intraindividually with plasma clearance of contrast medium administered for CT. We found a promising correlation of r = 0.84 and y = 7.5 + 0.94x (8); however, this technique involved an increase of radiation exposure compared with triphasic routine CT of the abdomen, as an additional early parenchmyal scan was obtained. Another drawback was inaccurate determination of the input function.
In the present study, we tried to overcome these earlier problems by replacing the second contrast-enhanced helical CT scan of our previous CT study protocol with 12 dynamic CT scans with reduced tube current. This resulted in more complete registration of the input function and, at the same time, reduced the additional radiation exposure. The additional radiation exposure caused by 12 dynamic scans with reduced tube current is estimated to be 1.86 mSv for women and 1.61 mSv for men (16). The radiation exposure is much lower compared with the single-location dynamic CT technique, which requires that 12 or more scans of 10-mm section thickness be obtained with usual tube current.
In the present study, hematocrit level was measured by using the unenhanced aortic attenuation values and Equation (8). By using this approach, GFR calculated with the two-point Patlak plot technique was determined only from CT data, without withdrawing a blood specimen. Alternatively, hematocrit levels measured in a blood specimen could be used instead.
With the use of a four–row detector CT scanner, examination times for the kidneys were reduced compared with examination times reported in previous studies, resulting in the reduction of errors caused by patient movement.
We found a good correlation between GFR measured with the two-point Patlak plot technique and GFR measured with iopromide plasma clearance. The correlation coefficient was better than that found in the previous study (8).
The two-point Patlak plot technique used in the present study showed a good correlation of total GFR with the reference method. The correlation was as good as that reported for the single-location dynamic CT examination, which measures multiple points for the Patlak plot (14,15). It seems likely that the improvement of the statistical noise by measuring the whole kidney volume instead of only one section outweighs the loss of redundancy by minimizing the number of points measured for the Patlak plot. As the reference method of the present study measured only total GFR, a definitve statement regarding the accuracy measuring single kidney function cannot be made from this datum alone. From a theoretical standpoint, however, it seems logical that the method can also be used to accurately measure relative kidney function. An accurate measurement of the relative kidney function was found in previous studies investigating single-location dynamic CT (14,15).
CT imaging of both kidneys in the present study lasted 5–6 seconds. The time needed for whole kidney imaging introduces an error, because the upper pole of the kidney is imaged before the lower pole, which means that at the time when imaged, the lower portions of the kidney have filtered more iodine into the nephron than the upper portion of the kidney.
Another error might result from different amounts of contrast medium in the vascular bed. The maximal difference from a kidney pole to the middle portion of the kidney, to which the arterial function fits, is 3 seconds. The error of the upper half might be corrected partially by the lower half because the differences in the densities of the upper and lower halves offset each other. As 3 seconds is a short time compared with the complete time of one examination, the expected error appears to be small.
Measuring GFR with the Patlak plot technique is a simplification of the renal physiology, as it neglects the renal interstitium as a third space. Bohle et al (17) measured interstitial volume of the kidney by using light microscopy in control subjects and patients with acute renal failure. The relative interstitial volume in the renal cortex was 8.4% for control subjects and 16.7% for patients with acute renal failure. The relative interstitial volume in the outer stripe of the outer medulla was 17.6% for control patients and 27.2% for patients with acute renal failure.
All approaches to measuring GFR with the Patlak plot inherit this "interstitium problem," which applies not only to the approach used in the present study but also to the single-location dynamic CT technique (3,14,15) and to the use of the Patlak plot technique with 99mTc-DTPA scintigraphy (19).
To our understanding, an enlarged interstitial space leads to an overestimation of GFR, as measured with the Patlak plot technique (8,18,19), which—to our knowledge—has not been proved in a study. Consequently, patients with inflammation, acute obstruction, or acute renal failure were excluded.
GFR can be measured accurately with minimally extended triphasic CT of the abdomen in patients without acute renal disorder by using a two-point Patlak plot technique. Thus, GFR measurement can be performed in patients who must undergo triphasic helical CT of the abdomen for clinical reasons. This is of clinical interest in patients with renal cell carcinoma or chronic pyelonephritis who might be eligible for nephrectomy if renal function of one kidney is too low. The technique presented here might also be useful for measuring GFR in the CT evaluation of living renal donors. As the two point Patlak plot technique presented in this study was tested only against total GFR, further study is required to evaluate its accuracy in the measurement of single-kidney GFR.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, N.H.; study concepts, all authors; study design, N.H.; literature research, all authors; clinical studies, N.H., C.W.; data acquisition and analysis/interpretation, all authors; statistical analysis, N.H., W.S.R.; manuscript preparation, definition of intellectual content, editing, revision/review, and final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. Boss, P. Martirosian, M. Gehrmann, F. Artunc, T. Risler, N. Oesingmann, C. D. Claussen, F. Schick, K. Kuper, and H.-P. Schlemmer Quantitative Assessment of Glomerular Filtration Rate with MR Gadolinium Slope Clearance Measurements: A Phase I Trial Radiology, March 1, 2007; 242(3): 783 - 790. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. G. Sommer Can Single-Kidney Glomerular Filtration Rate Be Determined with Contrast-enhanced CT? Radiology, February 1, 2007; 242(2): 325 - 326. [Full Text] [PDF]
|
|
|
|
|
|
E. Daghini, L. Juillard, J. A. Haas, J. D. Krier, J. C. Romero, and L. O. Lerman Comparison of Mathematic Models for Assessment of Glomerular Filtration Rate with Electron-Beam CT in Pigs Radiology, February 1, 2007; 242(2): 417 - 424. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
