| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, Gifu University School of Medicine, 40 Tsukasamachi,
Gifu 500-8705, Japan.
2 Department of Radiology, Osaka University School of Medicine, Yamadaoka 2-2,
Osaka 565-0871, Japan.
3 Department of Diagnostic Radiology, National Cancer Center Hospital, 5-1-1
Tsukiji, Chuo-Ku, Tokyo 104-0045, Japan.
Received January 15, 2002;
accepted after revision July 26, 2002.
Supported in part by the Grant for Scientific Research Expenses for Health,
Labor and Welfare Programs; the Foundation for the Promotion of Cancer
Research; and the Second-Term Comprehensive 10-Year Strategy for Cancer
Control.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Three independent off-site radiologists retrospectively reviewed MR images of the liver obtained in 72 patients with chronic liver damage, including 49 patients with untreated esophageal varices. All patients had undergone gastrointestinal endoscopy within 2 weeks of MR imaging. Both MR and endoscopic images were reviewed to determine whether esophageal varices were present and, if so, to determine the grade of the varices. Observer performances were tested with receiver operating characteristic curve analysis using the jackknife dispersion test. Correlations between the grades of the varices determined using MR images and those determined using endoscopic images were tested.
RESULTS. Sensitivity for detection of esophageal varices was significantly (p < 0.01) higher for the combination of unenhanced and gadolinium-enhanced MR images (81%) than for the unenhanced MR images alone (51%). The receiver operating characteristic curve analysis (area under the curve, [Az]) showed that performance using the combination of the unenhanced and gadolinium-enhanced MR images (Az = 0.641) was superior to that using unenhanced MR images alone (Az = 0.586). A statistically significant positive correlation (p < 0.05) was found between the grades determined using MR imaging and the grades determined using endoscopy.
CONCLUSION. Our results suggest the potential value of diagnosing the presence and grade of esophageal varices on MR imaging of the liver for patients with chronic liver damage. Gadolinium-enhanced MR imaging may increase the potential value.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Currently, the diagnosis of esophageal varices is established chiefly by gastrointestinal endoscopy, which has the advantage of allowing subsequent sclerotherapy or ligation therapy for esophageal varices. However, gastrointestinal endoscopy is an invasive and expensive procedure to include as part of the routine screening program in patients with chronic liver disease. Upper gastrointestinal radiography [14], Doppler sonography [12], CT [15], and MR angiography [16, 17] have been described as comparable to or less invasive than endoscopy for the detection of esophageal varices, although their clinical usefulness is still controversial. Researchers previously reported that sensitivities of transabdominal Doppler sonography [12] and phase-contrast MR angiography [16] were 89% and 85%, respectively.
Hepatocellular carcinoma is known to be the most common primary malignant neoplasm arising in patients with chronic liver damage related to current viral hepatitis or alcohol abuse, and a periodic screening program is believed to be effective for early tumor detection and improvement of prognosis [18]. Previous researchers have reported the usefulness of gadolinium-enhanced multiphasic MR imaging of the liver in screening programs or preoperative workup for patients with chronic liver damage or suspected hepatocellular carcinoma [19, 20].
We know of some anecdotal cases in which gadolinium-enhanced MR imaging being performed in daily clinical practice to detect and stage cirrhosis, hepatocellular carcinoma, or portosystemic shunt revealed and allowed accurate grading of esophageal varices in patients with cirrhosis, although the clinical usefulness of the procedure is as yet uncertain. The purpose of our study was to evaluate the value of MR imaging of the liver in the detection and grading of esophageal varices in patients with chronic liver damage.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The remaining 72 patients formed the study population—56 men and 16 women whose ages ranged from 50 to 84 years (mean age, 65.9 years). Chronic liver damage in these patients was caused by hepatitis B (n = 6), hepatitis C (n = 60), alcohol abuse (n = 2), or unknown agents (n = 4). Seventy patients had cirrhosis of the liver. For 13 patients, the diagnosis of cirrhosis was established by surgery and for 24 patients, by biopsy. The diagnosis for 33 patients was based on results on radiologic examinations (capsular nodularity and morphologic changes of the liver, splenomegaly, or extrahepatic portosystemic shunting) or on clinical and laboratory tests (Child-Pugh classification, the variables of ascites, encephalopathy, serum bilirubin and albumin values, and prothrombin time). Forty-nine patients received diagnosis of untreated esophageal varices detected on endoscopy. The grade of esophageal varices was graded according to the classification advocated by Beppu et al. [21] (Table 1).
|
MR Imaging of the Liver
MR imaging was performed using a scanner (Signa Horizon; General Electric
Medical Systems, Milwaukee, WI) with a superconducting magnet operating at 1.5
T. The unit provides a maximal gradient strength of 23 mT/m with a peak slew
rate of 120 mT/m per millisecond. All MR images were obtained in the axial
plane with a phase array multicoil for the body. The section thickness was
8-10 mm, with a 2- to 3-mm intersection gap for all pulse sequences.
The unenhanced MR imaging protocol comprised a T1-weighted spin-echo axial imaging sequence (19 patients in the early study period; TR/TE range, 500/8-9; matrix, 512 x 224; field of view, 29 x 29-32 x 32 cm; received bandwidth, ±16 kHz; signals acquired, 2; respiratory-ordered phase encoding; and acquisition time, 6 min 30 sec) or a T1-weighted gradient-recalled echo axial imaging with a fast multiplanar spoiled gradient-recalled echo acquisition under a steady-state free precession sequence (53 patients in the later study period; 150/1.6 [n = 39 patients] or 150/4.2 [n = 14 patients]; matrix, 512 x 224; flip angle, 90°; field of view, 29 x 22-32 x 24 cm; signal acquired, 1; 26-sec breath-hold acquisition once [n = 42] or twice [n = 11] to cover the entire liver).
Chemical shift selective fat-suppressed (n = 59) or non—fat-suppressed (n = 13) respiratory-triggered T2-weighted fast spin-echo axial imaging (effective TR range/effective TE range, 4000-8571/77-80; matrix, 512 x 256; echo-train length, 10-18 [median, 12]; field of view, 29 x 22-32 x 24 cm; received bandwidth, ± 62.5 kHz; signals acquired, 3 or 4; gradient moment nulling in the frequency-encoding direction; acquisition time, 3 min 12 sec—6 min [mean, 4 min 30 sec]); and gadolinium-enhanced triphasic gradient-recalled echo axial imaging with a fast multiplanar spoiled gradient-recalled echo acquisition under a steady-state free precession sequence (TR/TE range, 150/1.6-1.8; matrix, 512 x 224; flip angle, 90°; field of view, 29 x 22-32 x 24 cm; signal acquired, 1; and 26-sec breath-hold acquisition time in each phase) were also performed. The gadolinium-enhanced MR images were obtained after an antecubital IV bolus injection of 0.1 mmol/kg of gadopentetate dimeglumine (Magnevist; Schering. Berlin, Germany) delivered at 2.5-3 mL/sec flushed by 15 mL of sterile saline solution. The injection of contrast material and saline solution flush was performed manually. The scanning delays for triphasic MR imaging were 14 sec, 60 sec, and 3 min after initiation of the contrast injection, representing the hepatic arterial, portal venous, and equilibrium phases, respectively. Because the scanning delay for the hepatic arterial phase was set at 14 sec after the contrast injection, the image data obtained at approximately 27 sec were used to fill the central k-space lines to obtain entire image contrast of the hepatic arterial phase.
Image Analysis
We invited three radiologists from another institution with both clinical
and research experience as gastrointestinal radiologists for 6-10 years to
independently review the MR images. These three radiologists reviewed the
images off-site. They knew that the patients had chronic liver damage but knew
nothing else about the patients' histories.
In the image review, unenhanced T1-weighted spin-echo or gradient-recalled echo images and respiratory-triggered T2-weighted fast spin-echo images were combined and were reviewed first. Then unenhanced and gadolinium-enhanced triphasic gradient-recalled echo images were added for further review. The MR images were reviewed for the presence and grade of esophageal varices. The off-site radiologists were asked to assign one of five confidence level ratings for the presence of esophageal varices: 1 = definitely absent, 2 = probably absent, 3 = undetermined, 4 = probably present, or 5 = definitely present. The radiologists were instructed to assign a score of 1 if no dilated vessels were seen in the esophageal walls; a score of 3 if subtle, ill-defined vessels were seen in the esophageal mucosa or if evidently dilated vessels were seen adjacent to the esophageal walls; and a score of 5 if dilated vessels were definitely seen in the esophageal mucosa. Scores of 2 and 4 were assigned on the basis of each radiologist's subjective judgment.
The radiologists further assigned one of the four grades to the esophageal varices (0 = absent; 1 = small and straight; 2 = moderately sized, tortuous, and occupying less than one third of the esophageal lumen; or 3 = large, coiled, and occupying one third or more of the lumen). Finally, each radiologist counted the number of sections that depicted the esophagus on the MR images and determined one phase of gadolinium-enhanced triphasic gradient-recalled echo images in which the esophageal varices, if any, were most clearly revealed.
Statistical Analysis
The sensitivity of each imaging sequence for detection of esophageal
varices was determined using the number of patients whose images were assigned
a score of 4 or 5 (i.e., probably present or definitely present) of the total
number of 49 patients who had esophageal varices. Likewise, specificity was
determined using the number of patients whose images were assigned a score of
1 or 2 (i.e., definitely absent or probably absent) of the total number of 23
patients who had no esophageal varices. Sensitivities and specificities with
the two imaging sequences were compared using the McNemar test.
For each imaging sequence, a receiver operating characteristic curve was fitted to each radiologist's confidence rating using a maximum-likelihood estimation determined by LABMRMC1.0B software (Metz CE, University of Chicago, Chicago, IL). Observer performance with each imaging sequence for each radiologist was estimated by calculating the area under the receiver operating characteristic curve (Az). The difference between the averaged Az values was estimated for the two imaging sequences using jackknife dispersion and analysis of variance methods. Correlation between the grades of the varices based on MR imaging and those based on endoscopy was tested using the Spearman's rank correlation test.
To assess interobserver variability in assigning a confidence level for findings in each patient, we used multiple-observer kappa statistics to measure the degree of agreement. We used the nonweighted kappa statistics with binary data defined in terms of the presence (i.e., definitely present, probably present, or undetermined) or absence (i.e., probably absent or definitely absent) of esophageal varices. A kappa value of up to 0.20 showed slight agreement, a value of 0.21-0.40 showed fair agreement, a value of 0.41-0.60 showed moderate agreement, a value of 0.61-0.80 showed substantial agreement, and a value of 0.81 or greater showed nearly perfect agreement.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Twenty-four patients had grade 1 esophageal varices, 23 patients had grade 2, and two patients had grade 3, according to the endoscopic grades for esophageal varices described by Beppu et al. [21]. Sensitivities and specificities of the two imaging sequences for detection of esophageal varices are shown in Table 3. Sensitivity was significantly higher (p < 0.001) with unenhanced and gadolinium-enhanced MR images combined than with unenhanced MR images alone for all three radiologists and for the overall performance of the radiologists (Figs. 1A,1B,1C,1D and 2A,2B,2C,2D). Overall specificity was marginally (p < 0.08) higher with unenhanced MR images alone than with unenhanced and gadolinium-enhanced MR images combined (Fig. 3A,3B,3C,3D).
|
|
|
|
|
|
|
|
|
|
|
|
|
The Az values of the two unenhanced and gadolinium-enhanced sequences are shown in Table 4. The Az value was greater with unenhanced and gadolinium-enhanced MR images combined than with unenhanced MR images alone for all three radiologists and overall, although no statistically significant difference was confirmed. We found a significant positive correlation (p < 0.05) between the grades determined by MR imaging and those by endoscopy with unenhanced MR images for two radiologists and with unenhanced and gadolinium-enhanced MR images combined for all three radiologists (Table 5).
|
|
The multiple-observer kappa values were 0.476 for unenhanced MR images alone and 0.454 for unenhanced and gadolinium-enhanced MR images combined. Moderate agreement was obtained among the radiologists.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
The sensitivity for detecting esophageal varices was significantly higher with unenhanced and gadolinium-enhanced MR images combined than with unenhanced MR images alone. Although markedly dilated esophageal varices were readily detected as areas of flow void in the esophageal walls, varices of slight to moderate degree were not readily detected using only the unenhanced images. Contrast enhancement allowed the radiologists to confidently identify esophageal varices that had not been well depicted on the unenhanced images.
However, our results showed a trend toward the higher specificity with unenhanced MR images alone than with unenhanced and gadolinium-enhanced MR images combined. The lower specificity, despite the use of a contrast material, resulted from the radiologists' having more false-positive findings with the combination of unenhanced and gadolinium-enhanced images. In some patients with marked paraesophageal varices but without esophageal varices, radiologists often increased their confidence levels when using the gadolinium-enhanced images. Such false-positive interpretation with gadolinium-enhanced images took place because of the complexity of the anatomic relationship between the thin esophageal walls and the networks of paraesophageal and esophageal varices.
The observer performances for detecting esophageal varices determined by receiver operating characteristic curve analysis was not high enough on either the unenhanced images alone or the unenhanced and gadolinium-enhanced images combined. The three off-site radiologists were not familiar with interpreting liver MR images for the diagnosis of esophageal varices in their daily clinical practice, which might be the reason that their observer performances were not necessarily satisfactory. Radiologists' ability to diagnose esophageal varices with MR images of the liver may improve after recognizing the anatomic complexity of and gaining experience interpreting the structures in this region.
A significant positive correlation between the grades determined with MR imaging and those determined with endoscopy was confirmed. In our hospital, patients with mild esophageal varices are often followed up with periodic endoscopic examinations, whereas patients with moderate to severe esophageal varices undergo endoscopic ligation or sclerosing therapy. The study results suggest the potential usefulness of MR imaging of the liver in staging esophageal varices and determining endoscopic treatment.
There are some limitations to our study. We do not believe that MR imaging of the liver eliminates the need for endoscopic examination, but MR imaging may offer a good opportunity for radiologists to recommend endoscopy to the referring physicians. In our study, the MR images of the liver included the lower esophagus, in which involvement by esophageal varices and variceal rupture commonly occur. However, axial MR imaging performed mainly for imaging the liver may not reveal all the connections of esophageal varices. Although the off-site radiologists reviewed the images in conformity with our diagnostic criteria (based on Beppu's grading system), the radiologists might have been forced to subjectively judge whether borderline imaging findings showed esophageal or paraesophageal lesions. We may need to further assess the difference in imaging characteristics of paraesophageal and esophageal varices.
In conclusion, we found that our radiologists were able to provide valuable clinical information on the diagnosis of esophageal varices in patients with chronic liver damage by interpreting imaging findings of the esophagus identifiable on axial MR images of the liver. The radiologists were also able to assess the grade of esophageal varices with MR imaging of the liver. Although gadolinium-enhanced MR imaging increased sensitivity while somewhat decreasing specificity in our study, further refinement of diagnostic criteria may improve radiologists' performance. At present, however, radiologists should remember that MR imaging of the liver is a potentially valuable method of diagnosing and grading esophageal varices in patients with chronic liver damage, that gadolinium-enhanced imaging may increase the usefulness of MR images, and that endoscopy is mandatory for patients in whom MR imaging of the liver reveals esophageal varices.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  V. A. Sahni and K. J. Mortele The Bloody Pancreas: MDCT and MRI Features of Hypervascular and Hemorrhagic Pancreatic Conditions Am. J. Roentgenol., April 1, 2009; 192(4): 923 - 935. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. J. Kim, S. S. Raman, N. C. Yu, K. J. To'o, R. Jutabha, and D. S. K. Lu Esophageal Varices in Cirrhotic Patients: Evaluation with Liver CT Am. J. Roentgenol., January 1, 2007; 188(1): 139 - 144. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Annet, F. Peeters, Y. Horsmans, L. Hermoye, P. Starkel, and B. E. Van Beers Esophageal Varices: Evaluation with Transesophageal MR Imaging--Initial Experience Radiology, December 1, 2005; 238(1): 167 - 175. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
