| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 All authors: Department of Radiology, Yonsei University College of Medicine, YongDong Severance Hospital, 146-92 Dogok-Dong, Gangnam-Gu, Seoul 135-720, South Korea.
Received June 7, 2006;
accepted after revision October 11, 2006.
Address correspondence to J.-S. Yu
(yjsrad97{at}yumc.yonsei.ac.kr).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. In 38 patients with cirrhotic livers, focal
lesions 
5 mm containing fatty components were identified on chemical
shift gradient-echo MRI. Positive predictive values (PPVs) for benignity and
malignancy were calculated on the basis of lesion size, T1-weighted
hypointensity, T2-weighted hyperintensity, and arterial hypervascularity on
the initial MR images. The number of the fatty nodules (group A, up to 4;
group B, numerous) in individual patients was also correlated with the
malignant potential of the lesions that were verified pathologically or by
follow-up imaging studies.
RESULTS. In 31 group A patients, 21 (47%) of the 45 lesions showed a
malignant course, and their mean diameter (18.8 mm) was larger (p =
0.007) than that (10.5 mm) of benign lesions. In seven group B patients, all
35 lesions (the five largest lesions in each patient; mean diameter, 7.8 mm)
proved to be benign. The PPV of larger ( 15 mm) fat-containing nodules for
malignancy was 85% (11/13 lesions). Six (55%) of 11 immediately diagnosed
hepatocellular carcinomas were entirely hypointense on unenhanced in-phase
T1-weighted images. The PPV of T2-weighted hyperintensity and arterial
hypervascularity for the diagnosis of malignancy was 100% in group A
patients.
CONCLUSION. In the cirrhotic liver, large size ( 15 mm) and
T1-weighted hypointensity on in-phase images strongly suggest malignancy of
the fat-containing nodules. The presence of numerous nodules < 1 cm
suggests that the lesions are benign.
Keywords: cancer • cirrhosis • hepatocellular carcinoma • liver • liver disease • MRI
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
On the basis of our experience during the past decade, many fat-containing nodules are detected on chemical shift phase-selective gradient-echo imaging, with no malignant behavior extant on follow-up. The purpose of this study was to investigate the fate of fat-containing nodules via double-echo in-phase and opposed-phase gradient-echo imaging of the advanced cirrhotic liver. We discuss features that determine the frequency of malignancy and its clinical implications. To assess features that are suggestive of malignancy, we evaluated the size, signal intensity changes in and around fat-containing nodules on unenhanced and contrast-enhanced dynamic MR images, and the number of lesions in individual patients.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MRI was performed with a 1.5-T unit (Vision, Siemens Medical Solutions) and a phased-array multicoil. T2-weighted images were obtained by breath-hold multishot turbo spin-echo sequences (TR range/effective TE, 3,540-4,000/138; echotrain length, 29), with a chemical selective prepulse for fat suppression. In all cases, a double-echo chemical shift gradient-echo technique (TR/first-echo TE, second-echo TE, 140/2.7 [opposed-phase], 5.3 [in-phase]; flip angle, 90°) was performed before and after the injection of 15 mL of gadopentetate dimeglumine (Magnevist, Schering), with a 3 mL/s injection speed, into an antecubital vein, followed by a saline flush. An 8- to 10-mm slice thickness and a 1.6- to 2-mm intersection gap were used for 15 axial sections with a 19- to 21-second acquisition time, encompassing the entire liver during a single breath-hold.
Three phase (arterial, portal, and 5-minute delayed) contrast-enhanced dynamic imaging was performed. The delay time for arterial and portal phase imaging was determined after the test bolus examination and before dynamic imaging according to a previously described method [7]. Unenhanced opposed-phase images were automatically subtracted from the corresponding in-phase images and were routinely attached to the scan data. All scans were sent for interpretation on PACS workstations.
An attending radiologist with 10 years of experience in hepatic MRI reviewed the images of 73 patients on the PACS monitor, in addition to the patients' electronic medical records, for the initial selection of patients. To be enrolled in this study, each patient was required to have at least one nodular lesion showing a 5-mm or larger area of intralesional hyperintensity on the subtracted images of opposed-phase imaging from the corresponding in-phase images.
Malignancy of the lesions was determined by the following criteria:
histologically confirmed lesions; hepatocellular carcinomas ( 1 cm)
verified by coincidental hypervascularity and subsequent washout of contrast
enhancement on at least two vascular imaging techniques, including dynamic CT,
MRI, or conventional hepatic arteriography, and reinforced by sustaining
iodized oil accumulation after transarterial chemoembolization; and enlarged
lesions ( 2-mm increase in the longest dimension) on follow-up MRI within
1 year. For verification of the benignity of fat-containing nodules, the
nodule must have dis-appeared, decreased in size, or not changed after at
least 1 year or longer of follow-up with MRI or 2 years or longer of follow-up
with dynamic CT.
No 5-mm or larger fat-containing lesions were seen in 10 patients, and the image quality of the subjective MR images was not adequate for evaluation of focal lesions in another five patients, including one patient with technical failure of arterial phase imaging for the evaluation of lesion vascularity. Another 20 patients were excluded because of the absence of histologic or imaging verification or inadequate availability of follow-up studies. After exclusion of these 35 patients, the study population consisted of 38 patients (29 men and nine women) who ranged in age from 38 to 76 years (mean, 57 years).
Thirty-one patients had four nodules or fewer (n = 45 nodules), and seven patients had more than 10 nodules, in whom only the five largest lesions (n = 35 nodules) were counted. All of the 80 lesions on the initial subtraction images of opposed-phase from in-phase images were marked with electronic arrows, and these data were saved for subsequent analysis. The most common cause of cirrhosis was hepatitis B (n = 29, including one patient with concomitant hepatitis C and two patients with ethanol abuse). A limited number of patients had hepatitis C (n = 3) or ethanol abuse (n = 2). The origin of cirrhosis in four patients was unknown.
Data Analysis
Without knowing the final diagnoses or followup imaging findings of each
lesion, two attending radiologists with 6 and 12 years of experience,
respectively, in hepatic MRI examined the images. They determined the initial
MRI features of the fat-containing lesions on subtraction images (of
opposed-phase from in-phase) in terms of the signal intensity (hyper-, iso-,
hypointense) compared with the background parenchyma. They did this for the
unenhanced opposed-phase and in-phase and fat-suppressed T2-weighted images on
the PACS monitors. Arterial phase hypervascularity was determined by comparing
the unenhanced opposed-phase and in-phase with the corresponding arterial
phase images via dynamic imaging. This was done to decipher the degree of
increased signal intensity compared with the surrounding hepatic parenchyma.
The longest dimension of the lesion was measured on magnified views of
unenhanced opposed-phase or in-phase images using electronic calipers and was
recorded by the radiologists.
To reduce the measurement error, opposed-phase or in-phase was subjectively chosen for measurement, depending on the higher lesion-to-liver contrast, and the mean value of the two separate measurements was used for data description and statistical analyses.
Depending on the pathology reports or follow-up imaging findings, the 80
lesions were classified into six categories by the radiologist involved in the
initial case selection and another radiologist (who had 25 years of experience
in abdominal imaging) as follows: hepatocellular carcinoma, histologically
confirmed or verified by modified, noninvasive European Association for the
Study of the Liver (EASL) criteria for hypervascular focal lesions
[8,
9] immediately after initial
MRI reinforced by iodized oil CT within 1 month after chemoembolization
therapy (category I); hepatocellular carcinoma, verified during the follow-up
period, with the same criteria as for the diagnosis of a category I lesion
(category II); simply enlarged lesion ( 2-mm increase in the longest
dimension) on follow-up MRI, with no change in signal intensity and
enhancement pattern or histologically verified dysplastic nodules (category
III); lesion with no interval change in size (change of < ± 1 mm of
the longest dimension), contour, or signal intensity for 1 year or longer on
follow-up MRI (category IV); regressed or no longer visible lesion for 1 year
or longer on follow-up MRI (category V); and no hypervascular focal lesion
seen at the anatomically corresponding area for 2 years or longer of follow-up
dynamic CT (category VI). Categories I, II, and III were regarded as malignant
or premalignant lesions, and categories IV, V, and VI, as benign.
Statistical Analysis
The Student's t test was used to compare the mean size of the
malignant or premalignant lesions (categories I, II, and III) with that of the
benign lesions (categories IV, V, and VI) on the initial MR images. A
p value of less than 0.05 was considered statistically significant.
Positive predictive values (PPVs) for benignity and malignancy were calculated
on the basis of lesion size (10 and 15 mm, respectively), hypo- or
hyperintensity on in-phase images in addition to arterial hypervascularity,
and T2-weighted hyperintensity on the initial MR images. Patients with
numerous 5-mm or larger fat-containing nodules were categorized into two
groups (group A [31 patients], up to four lesions; group B [seven patients],
> 10 lesions), and the malignancy of the lesions in each group was
compared.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
The mean size of the 21 malignant lesions (18.8 mm) was larger (p
< 0.001) than that of all 59 benign lesions (8.9 mm) and even larger
(p = 0.007) than the 24 benign lesions (10.5 mm) of group A patients
(Table 1). No 15-mm or larger
lesions were found in group B patients. According to the size of the lesions,
the PPV for determination of benignity of 
10-mm lesions was 77% in group
A patients. The PPV for determining the malignancy of a fat-containing lesion
15 mm was 85%.
The signal intensity of each fat-containing nodule on the initial MR image is summarized in Table 2. Despite their fatty components, only four (36%) of 11 category I lesions showed hyperintensity on unenhanced in-phase images. Six (55%) of the category I lesions were exclusively hypointense on in-phase and opposed-phase images compared with the surrounding parenchyma (Fig. 1A, 1B, 1C, 1D, 1E). The PPV for the diagnosis of malignancy for hypointensity on in-phase images was 100% in any condition, and the PPV for the diagnosis of benignity for in-phase hyperintensity was only 57% in group A patients (Table 3). Despite chemical selective fat suppression, 10 malignant lesions (category I, 9/11; category II, 1/5) showed T2-weighted hyperintensity, and the PPV for the diagnosis of malignancy was 100% in group A patients (Fig. 1A, 1B, 1C, 1D, 1E). All 14 hypervascular lesions were malignant (category I, 10/11; category II, 3/5; category III, 1/5) in group A (PPV for malignancy, 100%). For 35 benign lesions in group B patients, T2-weighted hyperintensity, in-phase T1-weighted hypointensity, and arterial hypervascularity were not depicted at all.
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Chemical shift MRI is a proven technique for confirming the presence of fat in tissues and provides definitive differentiation of lesions that contain water from those that contain both water and fat [5, 6, 13, 14]. In unenhanced imaging, the double-echo chemical shift, phase-selective, gradient-echo technique, used in our study, is more helpful for detecting small amounts of lipids than for the visual inspection of in-phase and opposed-phase images, depending on the automatic in-phase-opposed-phase subtraction without slice misregistration [20-22].
Previously, during an MRI examination of the cirrhotic liver, the characterization of focal nodular lesions containing a fatty component was a diagnostic challenge. In the present study, well-known diagnostic criteria for classic hepatocellular carcinoma, including T2-weighted hyperintensity and arterial hypervascularity [23], were adjusted in the differential diagnosis of the fat-containing nodules. For T1-weighted images, six (55%) of the 11 category I hepatocellular carcinomas exclusively showed hypointensity on in-phase images. In addition to the proliferative cellular density and an intracellular copper component, fatty change is one of the factors that increases the T1-weighted signal intensity of hepatocellular carcinoma [24-26]. Meanwhile, T1-weighted hypointensity is a nonspecific finding of various neoplastic lesions containing excessive water content and loose cellular density with or without internal fibrosis, and we thought that somewhat excessive hypointensity masked the influence of hyperintensity from a fatty component in such cases of hypointense hepatocellular carcinomas on in-phase images. In other words, T1-weighted, hypointense, fat-containing lesions on in-phase and opposed-phase images could not be recognized as fat-containing lesions without the subtraction images that provide higher sensitivity for a small amount of intralesional fat. According to our study, hypointensity on in-phase images, in conjunction with a darker signal intensity on opposed-phase images, is a highly specific finding that suggests hepatocellular carcinoma with mild fatty change.
Except for the category I lesions that were diagnosed within 1 month of the initial MRI, most of the category II-VI lesions could not be readily characterized by initial imaging findings. Eguchi et al. [1] described high-grade dysplastic nodules that tend to be accompanied by marked fatty change, frequently seen in early-stage hepatocellular carcinomas and in dysplastic nodules containing cancerous foci. Such evidence strongly suggests that high-grade dysplastic nodules might be a transitional stage to hepatocellular carcinoma, or these nodules might already contain cancerous foci. Those authors found that the mean diameter of dysplastic nodules containing cancerous foci (1.6 cm) was significantly larger than the 1.0 cm diameter of dysplastic nodules not containing cancerous foci [1].
Regardless of the lesion categories or a patient's grouping, PPV of
malignancy in 15-mm fat-containing nodules was 85% in our study. Taking
heed of previous findings [1],
we know that fat-containing nodules larger than 15 mm might contain cancerous
foci. Kutami et al. [11]
reported that the number of intratumoral arteries in small ( 1.5-cm),
well-differentiated hepatocellular carcinomas with diffuse fatty change was
smaller than the number among other larger, moderately differentiated lesions.
Their results could suggest the presence of several smaller hypovascular
category II or III malignant lesions in the present study.
A close relationship between hepatocarcinogenesis and fatty change has been indicated in several reports [1, 4, 11, 27]. Animal experiments have shown that the hepatocellular metabolism was modified in chemically induced preneoplastic liver foci, resulting in glycogen or fat accumulation. Damage to the sinusoidal endothelium primarily affects the passage of lipoproteins and is thought to have a profound influence on the metabolism of lipoproteins [28]. In hepatocellular carcinomas, the endothelial cells of sinusoid-like vessels were shown to be thicker and to have fewer fenestrations than normal cells [29]. These endothelial changes, through a decrease in permeability, may contribute to fatty change in the early stages of hepatocellular carcinomas.
All 35 lesions in group B patients (all patients had B-viral cirrhosis, and one patient had diffuse steatosis of background parenchyma), among those with 10 or more fatty nodules, showed a benign course in our study. The mechanism of fatty changes in group B patients may be different from that in group A patients having only one or a few fatty nodules. Previous imaging descriptions of benign fatty nodules in the cirrhotic liver are limited [17, 29]. Local ischemia under steatogenic conditions, such as hyper- and hypoalimentation, alcoholism, drug reactions, obesity, diabetes, and Reye's syndrome, is considered an important pathogenetic factor of focal fatty lesions [23, 30, 31]. In the cirrhotic liver, fatty infiltration can also occur by the impaired extraction of fatty acids from damaged hepatocytes [23].
Seven (88%) of eight group B patients showed grossly macronodular cirrhotic changes, intermingled with numerous cirrhotic nodules and areas of fibrosis, on imaging studies. If poor tissue perfusion is indeed one of the causative factors of fatty nodules, then accentuation of local variations in tissue perfusion of advanced cirrhotic parenchyma, as postulated by Zimmon [32], could conceivably produce tissue hypoxia, leading to focal fatty change. That certain group A patients (categories IV, V, and VI) showed a benign course might be explained by this theory of localized anoxia in the cirrhotic liver. This may hold true among those with or without congenital or acquired vascular anomalies such as small portal venous thrombosis or an arterioportal shunt [30].
In a study of isolated benign, fatty nodules in otherwise normal livers, Shojania and Hogg [30] postulated that fatty nodules can represent a late stage of regeneration after anoxic insult. They also theorized that severe fatty change from prolonged ischemia can induce focal septal fibrosis and focal nodular regenerative activity [30]. The fate of nodules with focal fatty change is unknown. One case, described by Brawer et al. [31], showed extensive fibrosis, suggesting that, in some instances, focal fatty change may progress to a more fibrotic or regenerative pattern. Alternatively, nodules may simply regress, leaving normal parenchyma.
Our study has several limitations. Because of its retrospective nature, histologic specimens were not available for most lesions. Gun-needle biopsy of hypervascular lesions is not practically accepted by clinicians and is reserved for gradually enlarged lesions with no definite evidence of hypervascular features on imaging follow-up studies, such as the three histologically confirmed high-grade dysplastic nodules (category III) enrolled in our study. Moreover, this study was designed to describe the initial MRI features of the fat-containing lesions, depending on time-related changes for the benign lesions, rather than a correlation between imaging and pathology, for the final diagnoses. According to the recently modified criteria for noninvasive diagnosis of hepatocellular carcinomas of the European Association for Study of the Liver (EASL) [8, 9], reinforced by the iodized oil CT findings in our study, we believe that the number of false-positive lesions for the final diagnosis of hepatocellular carcinomas (categories I and II) would be limited.
The requirement for long follow-up periods to determine the malignancy of the lesions occasionally could not be met. However, we suggest that unchanged or regressed lesions that are small (mean initial diameter, 8.9 mm) on long-term (mean, 21.2 months) follow-up MRI (categories IV and V) have little chance of becoming malignant. For the category VI lesions, we thought that the follow-up period for CT (mean, 34.3 months) was long enough to exclude the malignant potential of the small lesions (mean, 8.4 mm). It is somewhat arbitrary to apply different follow-up periods for MRI and CT. For MRI, lesion-to-lesion comparison is feasible for small fat-containing lesions; however, it is not easy to directly compare the lesions on initial MRI with follow-up CT. We considered that 1-year or longer follow-up would be enough for observation of a subtle change on MRI, but 2-year or longer follow-up would be essential for CT to rule out gross change of small lesions.
Theoretically, chemical selective prepulse used during the fast spin-echo sequence in our study can effectively suppress the signals from the fatty component. We found, however, that fat suppression was not complete in the uppermost sectional images. Some T2-weighted hyperintensity of the fatty nodules could reflect the incomplete suppression of the inherently high signal intensity of the fatty component, regardless of the nature of the lesions. For many years, double-echo opposed-phase and in-phase gradient-echo imaging has been used as the base sequence for dynamic imaging. Combined interpretation of opposed-phase and in-phase IV dynamic contrast-enhanced images provided useful information for determining the temporal enhancement characteristics of focal lesions without incurring unpredictable variations of intra- or extralesional fatty content, including the phenomenon of paradoxical suppression of contrast enhancement or the misinterpretation of inherently hyperintense nodules as hypervascular lesions on in-phase contrast-enhanced imaging [22]. Discussion of the usefulness of this technique was not our focus in this study.
In conclusion, the presence of an intralesional fatty component itself is
not diagnostic of malignancy in the cirrhotic liver
[33]. Larger size ( 15
mm), T1-weighted hypointensity on in-phase images, and T2-weighted
hyperintensity or arterial hypervascularity are findings on initial MR images
that are highly suggestive of malignancy. Numerous smaller (< 15 mm) fatty
nodules strongly suggest benignity; however, for a single or a few small fatty
nodules in the cirrhotic liver, long-term serial follow-up imaging is required
to fully exclude malignancy.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
