| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Radiology, Université Catholique
de Louvain, St-Luc University Hospital, Ave. Hippocrate 10, B-1200 Brussels,
Belgium.
2
Center of Biostatistics and Medical Documentation,
Université Catholique de Louvain, Mont-Godinne
University Hospital, Ave. Thérasse 1, B-5530
Yvoir, Belgium.
3
Laboratory of Gastroenterology, Université
Catholique de Louvain, St-Luc University Hospital, B-1200 Brussels,
Belgium.
Received June 30, 2000;
accepted after revision August 16, 2000.
Supported in part by research grants from Fonds National de la Recherche
Scientifique (3.4578.00), Belgium, and Association pour l'Etude et la
Recherche en Radiologie, France.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Dynamic contrast-enhanced single-section CT scans of the liver were obtained in 40 individuals who included six control subjects, 16 patients with noncirrhotic chronic liver disease, and 18 patients with cirrhosis. Hepatic, aortic, and portal venous time—density curves were fitted to a dual-input one-compartment model to calculate the liver perfusion, arterial fraction, distribution volume, and mean transit time.
RESULTS. Liver perfusion decreased in patients with cirrhosis (67 ± 23 mL · min-1 · 100 mL-1 versus 108 ± 34 mL · min-1 · 100 mL-1 in control subjects [p = 0.009] and 98 ± 36 mL · min-1 · 100 mL-1 in patients with noncirrhotic chronic liver disease [p = 0.003]), and the arterial fraction and the mean transit time increased (41 ± 27% and 51 ± 79 sec versus 17 ± 16% and 16 ± 5 sec in control subjects, and 19 ± 6% and 17 ± 8 sec in patients with noncirrhotic chronic liver disease [p < 0.05]). A significant correlation was seen between these three perfusion parameters and the severity of chronic liver disease based on clinical and biologic data (p < 0.001). No significant change in distribution volume was observed.
CONCLUSION. Hepatic perfusion parameters measured with CT were significantly altered in cirrhosis and correlated with the severity of chronic liver disease.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In chronic liver disease, important changes occur in the liver circulation. The increase of intrahepatic vascular resistance decreases the portal fraction of liver perfusion [2, 3]. This decrease of portal perfusion is partially compensated by an increase of arterial inflow [2, 4, 5]. In addition, capillarization of the sinusoids is observed in the form of endothelial defenestration, deposition of collagen in the extravascular Disse's spaces, and formation of basal laminas [6, 7]. The sinusoidal capillarization affects the transit time of small and large molecules [8, 9]. These quantitative and qualitative hemodynamic changes in cirrhosis have a profound effect on hepatic function and on the clearance of endo- and xenobiotics [10,11,12].
Various methods exist for the determination of hepatic perfusion [13,14,15,16,17]. Most of them, however, are invasive or controversial [13, 18]. In this study, we have quantified liver perfusion parameters noninvasively by measuring with CT the extravascular distribution of commercially available small-molecular-weight iodinated contrast agents, as described previously [19]. This method is based on the use of a dual-input single-compartment model of liver circulation [16, 19]. The hepatic inflow rate constant, which reflects liver perfusion, can be calculated; and the arterial fraction of liver perfusion, the distribution volume, and the mean transit time can be measured. A comprehensive view of liver perfusion parameters is thus obtained. The purpose of our study was to assess the changes in the perfusion parameters measured with CT in patients with chronic liver disease and to determine if these changes correlate with disease severity.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Imaging
After an overnight fast, the patients were asked to lie down for 1 hr
before the CT examination to minimize physiologic variations in portal flow.
The CT examinations were performed on a Twin RTS scanner (Elscint, Haifa,
Israel). On transverse scout images of the upper abdomen, a level was
determined at which the liver, aorta, and portal vein were clearly visualized.
The CT perfusion protocol comprised 40 scans that were obtained at this single
level with the following parameters: 120 kVp, 100 mA, 512 x 512 matrix,
10-mm slice thickness, 1-sec scan time, and 3-sec cycle time.
A nonionic iodinated contrast agent containing 350 mg I/mL (Omnipaque [iohexol]; Nycomed Imaging, Oslo, Norway) was administered IV through a 16- or 18-gauge catheter at the start of the CT perfusion study. Forty milliliters of the contrast agent, pushed by 30 mL of saline solution, was injected with a power injector (CT 9000 ADV; Liebel-Flarsheim, Cincinnati, OH) at a rate of 7 mL/sec via an antecubital vein. The patients kept breathing quietly during the perfusion study. To avoid respiratory motion artifacts, patients were clearly informed of a possible flushing sensation during contrast agent injection.
Data Analysis
The images were transferred to a workstation (Silicon Graphics, Mountain
View, CA) for data analysis using programs written in IDL and C (Research
Systems, Boulder, CO). Three large regions of interest (ROIs) were drawn in
the aorta, the portal vein, and the right liver lobe by two independent
observers. Because of respiratory movements, the portal venous ROI often had
to be moved on each slice. The density measurements on the unenhanced first
scans were averaged and subtracted from the measurements on the ensuing
contrast-enhanced scans. After this normalization, the contrast-enhanced
density measurements or signal increases are directly proportional to the
concentration of the contrast agent in the tissue
[22].
A dual-input single-compartment model was used to fit the data, as
previously described [19].
Briefly, the liver, including sinusoids, interstitium, and cells, was
considered to be a single compartment
[23]; and two inflow rate
constants, k1a and k1p (aorta and
portal vein, respectively) were used because the liver receives its blood
supply from both vessels. One outflow rate constant, k2,
was also included in the model, resulting in the following equation:
 | (1) |
a and p, which represent the
transit time from the aorta and the portal vein to the liver ROI, we obtain:
 | (2) |
 | (3) |
Statistical Analysis
Concordance between the measurements of the two observers was assessed
using intraclass correlation coefficients. The results of the four perfusion
parameters (liver perfusion, arterial fraction, volume of distribution, and
mean transit time) in the three groups (normal control subjects, patients with
noncirrhotic chronic liver disease, and patients with cirrhosis) were compared
using the Kruskal-Wallis analysis of variance. This was followed, in case of
significant heterogeneity, by two-by-two comparisons using the Wilcoxon's rank
sum test. The sensitivity and specificity of the perfusion parameters for
diagnosing cirrhosis were analyzed with receiver operating characteristic
(ROC) curves constructed by maximum likelihood estimation with the LABROC 1
program (Metz CE, Chicago, IL). Areas under the curves were compared using the
z score test with the CLABROC program (Metz CE, Chicago, IL). The
severity of liver disease was categorized in five classes (normal;
noncirrhotic chronic liver disease; and Child A, Child B, and Child C
cirrhosis). The correlation between this severity score and each of the four
perfusion parameters was assessed using Spearman's rank correlation
coefficients. Data are given as mean ± standard deviation. All
statistical tests are two-tailed. Statistical significance was defined as a
p value of less than 0.05.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Figures 1A,1B,1C and 2A,2B show typical time—density curves and fits for normal and cirrhotic patients. The results of the liver perfusion, arterial fraction, distribution volume, and mean transit time in the control subjects, the patients with noncirrhotic chronic liver disease, and the patients with cirrhosis are given in Figure 3A,3B,3C,3D. Liver perfusion was significantly decreased in the patients with cirrhosis (67 ± 23 mL · min-1 · 100 mL-1 versus 108 ± 34 mL · min-1 · 100 mL-1 in control subjects [p = 0.009] and 98 ± 36 mL · min-1 · 100 mL-1 in patients with non-cirrhotic chronic liver disease [p = 0.003]). The arterial fraction was significantly increased in patients with cirrhosis (41 ± 27%, versus 17 ± 16% in control subjects [p = 0.022] and 19 ± 6% in patients with noncirrhotic chronic liver disease [p = 0.004]). The mean transit time was also significantly increased in patients with cirrhosis (51 ± 79 sec, versus 16 ± 5 sec in control subjects [p < 0.001] and 17 ± 8 sec in patients with noncirrhotic chronic liver disease [p < 0.001]). No significant difference of distribution volume was observed between the groups (25.5 ± 4.4% in control subjects, 24.1 ± 4.3% in patients with noncirrhotic chronic liver disease, and 28.9 ± 8.6% in patients with cirrhosis [p = 0.22]). No significant differences were found between control subjects and patients with noncirrhotic liver disease for any parameter.
|
|
|
|
|
|
|
|
|
In the diagnosis of cirrhosis, the areas under the ROC curves were 0.81 ± 0.07 for liver perfusion, 0.78 ± 0.08 for the arterial fraction, and 0.89 ± 0.05 for the mean transit time. The areas under the ROC curves did not differ significantly (liver perfusion versus arterial fraction, p = 0.69; liver perfusion versus mean transit time, p = 0.14; arterial fraction versus mean transit time, p = 0.13) (Fig. 4). The best cutoff point to differentiate patients with cirrhosis from patients without cirrhosis was considered to be a mean transit time of 22.6 sec, giving a sensitivity and a specificity of 81%.
|
The severity of liver disease categorized in five classes correlated significantly with the liver perfusion (r = -0.55, p < 0.001), the arterial fraction (r = 0.59, p < 0.001), and the mean transit time (r = 0.70, p < 0.001). No significant correlation was observed between the severity of liver disease and the distribution volume (r = 0.29, p = 0.07) (Table 1).
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Other striking changes during cirrhosis are the transformation of the fenestrated sinusoids into continuous capillaries and the deposition of collagen in the extravascular Disse's spaces located between the sinusoidal endothelium and the hepatocytes [6, 7]. These alterations in the sinusoids and extravascular Disse's spaces modify the transit time and the distribution volume of small and large molecules, as previously shown in multiple indicator dilution studies [8, 9]. In these ex vivo studies, it was observed that large molecules such as albumin had a decreased distribution volume in cirrhosis because the molecules remained intravascular because of capillarization of the sinusoids. In contrast, small molecules such as sucrose had a constant distribution volume in cirrhosis because they were still able to pass into the Disse's spaces. The transit time of small molecules, however, was increased because their diffusion in the extravascular space itself was restricted by collagenization. Our results agree with these findings of the indicator dilution studies. Indeed, we observed that the small-molecular-weight contrast agent used in this study had an increased transit time and a constant distribution volume in cirrhosis. We propose that the use of contrast agents with different molecular weights in future CT studies may give the opportunity to further probe the sinusoidal permeability in a noninvasive way.
We observed that the perfusion parameters measured with CT tended to change in patients with noncirrhotic chronic liver disease (Table 1). Some hemodynamic changes may occur in the liver before cirrhosis develops [25]. However, the differences between the control subjects and the patients with noncirrhotic chronic liver disease did not reach a statistically significant level in our group of patients. In contrast, the perfusion parameters were significantly altered in cirrhosis.
Our ROC analysis showed that patients with cirrhosis can be differentiated from patients without cirrhosis with reasonable accuracy on the basis of their hepatic perfusion parameters. However, because of intersubject variability in liver perfusion, an absolute threshold to separate patients with cirrhosis from patients with noncirrhotic chronic liver disease may be difficult to apply in clinical practice. Despite its limitations, liver biopsy will remain the gold standard for the diagnosis of cirrhosis [26]. A more interesting finding of our study was that the perfusion changes in chronic liver disease significantly correlated with disease severity. Therefore, the value of CT perfusion measurements in the follow-up of patients with chronic liver disease and in the determination of their prognosis should be further assessed [3].
Several noninvasive methods have been proposed to quantitate liver flow in clinical practice. These proposals include imaging methods based on measurements performed with Doppler sonography, nuclear medicine, or CT; and methods based on the clearance of xenobiotics such as sorbitol. Liver flow at the prehepatic level can be measured with Doppler sonography [17]. However, the reproducibility of portal venous flow measurements with Doppler sonography remains controversial, and arterial flow measurements are even more difficult to obtain with this method because of the small diameter of the hepatic artery [27, 28]. In addition, liver flow can be measured with Doppler sonography, but liver perfusion per gram or milliliter of tissue is unknown unless the liver volume is also calculated. For example, in a cirrhotic patient with hepatomegaly, liver flow measured with Doppler sonography at the prehepatic level may be increased, whereas the liver perfusion per gram of tissue is actually decreased.
Nuclear medicine techniques have been used to study hepatic perfusion [16, 29]. These techniques are hampered by their limited spatial resolution. In particular, noninvasive direct measurement of the activity in the portal vein cannot be obtained even with positron emission tomography, which has the best spatial resolution of the nuclear medicine techniques.
Reports have indicated that functional liver flow can be estimated by the hepatic clearance of sorbitol [12, 14]. Because of its high extraction fraction in normal subjects, the clearance of sorbitol is flow-limited and reflects the perfusion through functional sinusoids. However, the use of sorbitol as a marker of functional perfusion in chronic liver disease remains controversial because the hepatocyte extraction of sorbitol may decrease in cirrhosis [30]. Moreover, the sorbitol clearance method is global because separate measurements of arterial and portal flows cannot be performed. In addition, urine has to be collected for several hours to determine the renal clearance of sorbitol, which can create logistic problems in a busy clinical department [30].
The high spatial and temporal resolution of modern CT scanners is particularly suited for perfusion measurements. Miles et al. [15] and Blomley et al. [31] have shown that CT can be used for the noninvasive quantification of liver perfusion. These authors used a slope measurement technique. We used a well-established compartmental modeling technique [16, 19]. With the slope method, only the peak time points of the aortic and portal time—density curves are used to calculate perfusion, whereas with a compartmental model, all points of the time—density curves are used. In addition, a compartmental model can measure not only the liver perfusion and arterial fraction but also the mean transit time and distribution volume of the tracers. The measurement of these parameters is relevant because our study and previous ex vivo studies with multiple indicators show that these perfusion parameters are modified in cirrhosis. CT of the whole liver can be performed after the perfusion study by injecting a second bolus of contrast agent to detect hepatocellular carcinoma in the cirrhotic liver. It is thus feasible to obtain morphologic and functional information in one CT examination.
The radiation dose may be a limitation to the frequent use of perfusion CT. However, as CT technology evolves, the radiation exposure is reduced because of decreased scanning time and improved detector sensitivity. To limit the radiation dose, our CT protocol was performed with 100 mAs and 3-sec time resolution. One may argue that this time resolution is too coarse to assess the peak aortic enhancement. However, in contrast to the slope method, the compartmental model uses all the points of the aortic time—density curve. In addition, we performed a computer simulation in which we compared 1-sec versus 3-sec sampling rates and determined that the perfusion value was underestimated by less than 10% when 3-sec sampling was used as compared with 1-sec sampling. Scanning was performed during 120 sec because the outflow rate constant (k2) had to be estimated to calculate the distribution volume and the mean transit time. In future studies, the CT protocol can be improved by decreasing the scanning interval in the 10- to 30-sec period and increasing it after 30 sec. Furthermore, the use of MR imaging as an alternative to CT for the quantification of liver perfusion should be explored [32].
No extravasation of contrast material occurred in any patient in our study whose injection was performed with a high flow rate of 7 mL/sec. IV injections of contrast material have been performed with flow rates up to 10 mL/sec in previous perfusion studies in which the slope method was used [31, 33]. Indeed, the slope method is critically dependent on the measurement of the peak of the vascular input function [3, 31, 33]. The need for a sharp vascular bolus is less acute with a compartmental model as used in our study because no assumption on bolus shape is made with compartmental models. In practice, however, it is always better to obtain a sharp input bolus because the tissue response is easier to measure. The optimal injection rate for perfusion studies remains to be determined.
We used ROIs in the aorta as a surrogate for the hepatic arterial inflow. The hepatic inflow can be affected by atherosclerosis or other diseases in the celiac trunk or the hepatic artery itself. In theory, it should thus be preferable to draw ROIs in more peripheral arteries when liver perfusion measurements are performed. However, to do so is often impracticable because the hepatic artery has a small diameter and the anatomy of the arterial supply to the liver varies.
Because of the long scanning period in our perfusion studies, the patients were allowed to breathe quietly during scanning. A drawback of this method was that the portal venous ROIs had to be changed often from slice to slice because of respiratory motion. To avoid this problem, the perfusion study could be performed during several breath-holding intervals, at least in cooperative patients. In addition, several slices can be scanned simultaneously with multirow helical CT scanners.
A further limitation of our study was the use of a 10-mm slice thickness. This slice thickness was chosen to obtain a high signal-to-noise ratio. Thick slices were also used in previous liver perfusion studies [3, 15, 33]. However, the density measurements in the vessels may be affected by partial volume effects, especially in the portal vein, which crosses the imaging plane at an acute angle. These effects may influence the results of the perfusion measurements. Therefore, the use of thinner slices is recommended in future perfusion studies.
The clinical relevance of the measurements of perfusion changes with CT in cirrhosis should be assessed further. In addition to the possible role of CT in the determination of disease prognosis, the quantification with CT of liver perfusion parameters may be an important factor in the determination of the systemic availability of drugs given orally. Indeed, depending on liver perfusion and permeability, the hepatic removal of drugs may be reduced and the systemic availability increased [11]. Finally, because the portal pressure is the product of liver flow and resistance, the quantification of liver perfusion may improve our understanding of the hepatic hemodynamic effects of vasoactive drugs and interventional procedures for the treatment of portal hypertension [34,35,36].
In conclusion, our CT results show that the liver perfusion, the arterial fraction of liver perfusion, and the mean transit time of an iodinated contrast agent are significantly altered in cirrhosis, and that these parameters correlate with the degree of hepatic dysfunction on the basis of clinical and biologic data in chronic liver disease. These findings underscore the importance of perfusion as a marker of liver function. CT and compartmental modeling can be used as noninvasive tools to quantify hepatic perfusion parameters in chronic liver disease.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Luciani, A. Vignaud, M. Cavet, J. Tran Van Nhieu, A. Mallat, L. Ruel, A. Laurent, J.-F. Deux, P. Brugieres, and A. Rahmouni Liver Cirrhosis: Intravoxel Incoherent Motion MR Imaging--Pilot Study Radiology, December 1, 2008; 249(3): 891 - 899. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S MIYAZAKI, K MURASE, T YOSHIKAWA, S MORIMOTO, Y OHNO, and K SUGIMURA A quantitative method for estimating hepatic blood flow using a dual-input single-compartment model Br. J. Radiol., October 1, 2008; 81(970): 790 - 800. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. S. Koh, C. H. Thng, P. S. Lee, S. Hartono, H. Rumpel, B. C. Goh, and S. Bisdas Hepatic Metastases: In Vivo Assessment of Perfusion Parameters at Dynamic Contrast-enhanced MR Imaging with Dual-Input Two-Compartment Tracer Kinetics Model Radiology, October 1, 2008; 249(1): 307 - 320. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Hagiwara, H. Rusinek, V. S. Lee, M. Losada, M. A. Bannan, G. A. Krinsky, and B. Taouli Advanced Liver Fibrosis: Diagnosis with 3D Whole-Liver Perfusion MR Imaging--Initial Experience Radiology, March 1, 2008; 246(3): 926 - 934. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Zhang, Y.-Y. Duan, J.-M. Li, and J.-K. Yin Hemodynamic Features of Doppler Ultrasonography in Patients With Portal Hypertension: Intraoperative Direct Measurement of Portal Pressure in the Portal Venous System J. Ultrasound Med., December 1, 2007; 26(12): 1689 - 1696. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K Murase, S Miyazaki, and X Yang An efficient method for calculating kinetic parameters in a dual-input single-compartment model Br. J. Radiol., May 1, 2007; 80(953): 371 - 375. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. E. Stewart, X. Chen, J. Hadway, and T.-Y. Lee Correlation between Hepatic Tumor Blood Flow and Glucose Utilization in a Rabbit Liver Tumor Model Radiology, June 1, 2006; 239(3): 740 - 750. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Sitek and R. G. Sheiman Small-Bowel Perfusion Measurement: Feasibility with Single-Compartment Kinetic Model Applied to Dynamic Contrast-enhanced CT Radiology, November 1, 2005; 237(2): 670 - 674. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. V. Pandharipande, G. A. Krinsky, H. Rusinek, and V. S. Lee Perfusion Imaging of the Liver: Current Challenges and Future Goals Radiology, March 1, 2005; 234(3): 661 - 673. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C Harvey and M Blomley Imaging investigation of liver haemodynamics in patients at risk for hepatic metastatic disease Br. J. Radiol., February 1, 2005; 78(926): 103 - 104. [Full Text] [PDF]
|
|
|
|
|
|
A Nakashige, J Horiguchi, A Tamura, T Asahara, F Shimamoto, and K Ito Quantitative measurement of hepatic portal perfusion by multidetector row CT with compensation for respiratory misregistration Br. J. Radiol., September 1, 2004; 77(921): 728 - 734. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K A Miles Perfusion CT for the assessment of tumour vascularity: which protocol? Br. J. Radiol., December 1, 2003; 76(suppl_1): S36 - S42. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Annet, R. Materne, E. Danse, J. Jamart, Y. Horsmans, and B. E. Van Beers Hepatic Flow Parameters Measured with MR Imaging and Doppler US: Correlations with Degree of Cirrhosis and Portal Hypertension Radiology, November 1, 2003; 229(2): 409 - 414. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K A Miles and M R Griffiths Perfusion CT: a worthwhile enhancement? Br. J. Radiol., April 1, 2003; 76(904): 220 - 231. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
