| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Département de Pathologie, Hôpital Henri Mondor, Assistance
Publique-Hôpitaux de Paris et Université Paris 12-Val de Marne,
Créteil 94010, France.
2 Service d'Anatomie Pathologique, Centre Hospitalier Universitaire de
Bicêtre, 78 rue du Général Leclerc, Le
Kremlin-Bicêtre, Cedex 94275, France.
3 Service d'Imagerie Médicale, Hôpital Henri Mondor, Créteil
94010, France.
4 Service de Chirurgie Digestive, Hôpital Henri Mondor, Créteil
94010, France.
5 Service d'Hépatologie et de Gastroentérologie, Hôpital
Henri Mondor, Créteil 94010, France.
Received June 24, 2002;
accepted after revision November 6, 2003.
Address correspondence to S. Ferlicot.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. We retrospectively reviewed MRI and pathology findings in 27 focal nodular hyperplasia lesions with atypical MRI features. Six criteria for typical focal nodular hyperplasia were required: iso- or hypointensity on T1-weighted sequences and iso- or slight hyperintensity on T2-weighted sequences; homogeneous signal intensity; central hyperintense area on T2-weighted sequences; marked lesion contrast enhancement; accumulation of gadolinium chelates within the central area on delayed contrast-enhanced T1-weighted sequences; and absence of capsule.
RESULTS. The most common atypical radiology features included absence of, or an atypical, stellate area; heterogeneity on both T1- and T2-weighted images; and high-intensity signal on T1-weighted sequences. MRI–pathology correlation showed that T1 hyperintensity with no other atypical MRI feature (n = 3) could be explained by steatosis, sinusoidal dilatation, or hemorrhage. In addition, in two patients with lesions smaller than 3 cm in diameter, the only atypical MRI feature was absence of a stellate area.
CONCLUSION. These findings suggest a lesion that is hyperintense on T1-weighted sequences or that lacks a stellate area but is smaller than 3 cm in diameter can be diagnosed as focal nodular hyperplasia provided all other MRI criteria for this diagnosis are present. In such cases, close monitoring on MRI without invasive diagnostic procedures may be warranted. However, in large lesions (> 3 cm) without a stellate area and in lesions with heterogeneity, histopathology examination is mandatory to rule out other diagnoses.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-glutamyl transpeptidase levels
may be slightly increased [4].
Because no risk of bleeding or malignant transformation exists, resection is
not required [5,
6]. This implies a need for a
noninvasive investigation capable of establishing the diagnosis. Among imaging
procedures, it has been suggested that MRI has 70% sensitivity and 98%
specificity for the diagnosis of focal nodular hyperplasia
[4]. MRI criteria for typical
focal nodular hyperplasia have been described by Mattison et al.
[7] and extended by Mathieu et
al. [8]. In cases in which MRI
features are atypical and might raise diagnostic difficulties with other liver
tumors that must be resected (e.g., hepatocellular adenoma > 4 cm or
hepatocellular carcinoma), definite histologic proof must be obtained
[4]. The aim of the present
work was therefore to study and correlate the atypical MRI features of focal
nodular hyperplasia and pathology findings to increase the number of patients
who avoid additional invasive diagnostic procedures. |
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Imaging Techniques
All 116 patients underwent MRI at our institution using a high-field
(1.5-T) magnet (Magnetom SP3, Siemens). The protocol, which remained unchanged
during the 5-year study period, included axial transverse spin-echo
T1-weighted images (TR/TE, < 500/15) and T2-weighted images (2,200/90)
acquired on contiguous, 8-mm-thick slices of the entire liver. Dynamic
gadolinium chelate–enhanced subsecond gradient echo T1-weighted images
(Turbo FLASH, Siemens; 7/3; inversion time, 400 msec; flip angle, 10°)
were then obtained after a manual bolus injection of gadolinium chelates
(Dotarem [gadoterate dimeglumine], Laboratoire Guerbet) through an antecubital
vein. This turbo fast low-angle shot (FLASH) sequence was started in the area
of the lesion immediately after bolus injection (scanning delay ranging from 2
to 40 sec). A total of 20 images were acquired every 2 sec. Finally, 4 min
after the bolus injection, delayed gadolinium chelate–enhanced spin-echo
T1-weighted images were obtained using the same parameters as in the
unenhanced study. Total duration of the diagnostic MRI study was 30 min.
For several lesions with hyperintensity on T1-weighted images, a fat-suppressed technique was performed by transmitting a chemically selective radiofrequency pulse centered on the frequency of lipid resonance to drop out the fat signal.
Image Analysis
All MRI documents were reviewed retrospectively by two radiologists
specialized in liver imaging. In each patient, the radiologists sought to
establish a consensus about MRI features that may have a pathology
correlative. The two radiologists were blinded to the histology findings. The
six imaging criteria for typical focal nodular hyperplasia on MRI were iso- or
hypointensity on T1-weighted sequences and isointensity or slight
hyperintensity on T2-weighted sequences; homogeneous signal intensity;
presence of a central hyperintense area on T2-weighted sequences; marked
lesion enhancement at the arterial phase after bolus injection of a
paramagnetic contrast agent; accumulation of gadolinium chelates within the
central area on delayed contrast-enhanced T1-weighted sequences; and absence
of capsule surrounding the lesion on these sequences
[7–9].
The 90 patients with typical focal nodular hyperplasia on MRI satisfied all
six criteria.
Pathology Study
Gross findings were abstracted from the medical charts, and the histology
slides were reviewed on a double-headed microscope by two pathologists. One of
the two pathologists was a liver specialist with 25 years' experience. The
following gross features were recorded: number, size, shape, color,
lobulation, and presence or absence of a stellate area; hemorrhage; and
necrosis. Histologic sections of paraffin-embedded material were stained with
H and E, picrosirius red, Perls' stain for iron, and modified orcein and
rhodanine for copper-associated protein and copper. The following
histopathology features were recorded: presence or absence of fibrosis (i.e.,
stellate fibrous area on resected specimens and fibrous bands on resected
specimens or needle biopsies), vascular lesions (i.e., arterial or venous
abnormalities, sinusoidal dilatation, and hemorrhagic foci), inflammatory
infiltration, cholangiolar proliferation, hepatocellular alterations (i.e.,
cytologic atypia, ballooning degeneration, cytoplasmic inclusions, and
steatosis), and pigment accumulation. The diagnosis of typical focal nodular
hyperplasia was made when the following pathology features were observed:
presence of hyperplastic nodules made of normal hepatocytes and separated by
fibrous septa containing abnormal vascular structures, cholangiolar
proliferation, and inflammatory infiltrates. Focal nodular hyperplasia was
considered atypical when these three pathology criteria were not observed
altogether. Normal interlobular bile ducts were never seen within fibrous
areas. Other pathology features that are unusual in focal nodular hyperplasia
(i.e., steatosis, hemorrhagic foci, sinusoidal dilatation, necrosis, and
pigment accumulation) were recorded and semiquantified (mild, moderate, and
marked scale).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
No stellate area could be identified on T2-weighted and on both unenhanced and contrast-enhanced T1-weighted sequences in eight lesions (Fig. 1). In four other lesions, the stellate area was atypical, with hypointensity on T2-weighted sequences (n = 1), persistent hypointensity on contrast-enhanced sequences (n = 2) (Fig. 2A, 2B), or visibility before but not after contrast injection (n = 1).
|
|
|
Eight lesions had heterogeneous signal intensity on both T1- and T2-weighted sequences (Figs. 1, 2A, 2B, 3). Seven lesions were hyperintense by comparison with the healthy liver on T1-weighted sequences.
|
Lesion diameter of the 16 resected focal nodular hyperplasia lesions ranged from 1.5 to 14 cm (mean, 7 cm). Two of these lesions were less than 3 cm in diameter, and their only atypical MRI feature was the absence of a stellate area.
Diameter on MR images of the 11 nonresected focal nodular hyperplasia lesions ranged from 2 to 4 cm (mean, 3.3 cm). Two of these lesions were less than 3 cm in diameter.
Other atypical findings were observed, including hypointensity on T2-weighted sequences, absence of or heterogeneous enhancement after contrast material injection, presence of a peripheral rim simulating a capsule or of hypointense foci on T1- and T2-weighted sequences, increase in size on two successive MRI examinations, and intrahepatic biliary duct dilatation due to large bile duct compression by a lesion located near the hilum.
Pathology Findings
Macroscopically, among the 16 resected focal nodular hyperplasia lesions,
three showed red hemorrhagic areas.
Fourteen (52%) of the 27 focal nodular hyperplasias showed typical histology patterns. The remaining 13 lesions (48%) had one or more of the following atypical histology features: absence of a stellate area or of thick fibrous bands (n = 5), absence of abnormal vessels (n = 7), and absence of cholangiolar proliferation (n = 3). Among the 27 focal nodular hyperplasia lesions, 18 (67%) exhibited sinusoidal dilatation—five of mild degree, seven of moderate degree, and six of marked degree. Ten lesions (37%) contained hemorrhagic areas, including three mild, four moderate, and three marked. Steatosis was present in 23 lesions (85%)—of mild degree in 13, moderate in nine, and marked in one.
Other findings included bile (n = 6), copper-associated protein and copper accumulation (n = 12), iron accumulation (n = 10), and presence of foci of atypical hepatocytes (n = 4).
Radiology–Pathology Correlation
In two of the eight lesions with no visible stellate area on both T1- and
T2-weighted sequences, no thick fibrous bands were seen at pathology
examination. Among the five patients with a radiologically atypical stellate
area, two had no thick fibrous bands, two had fibrosis with abnormal vessels,
and one had fibrosis with normal vessels. In one patient, the stellate area
seen on MRI was not found in the needle biopsy specimen.
Of the eight lesions that were heterogeneous on T1- and T2-weighted sequences, seven showed sinusoidal dilatation, accompanied in six cases by hemorrhagic foci. In three cases, sinusoidal dilatation was marked. Steatosis was observed in six of these eight cases.
Sinusoidal dilatation was seen in six of the seven lesions that were hyperintense on T1-weighted sequences. Hemorrhagic foci were visible in three of these seven lesions, always in association with sinusoidal dilatation. Steatosis was noted in six of these seven lesions and was accompanied by sinusoidal dilatation or hemorrhage in five lesions. On fat-suppressed T1-weighted images, four lesions became hypointense, whereas one remained hyperintense (Fig. 4A, 4B). Pathology examination showed accumulation of fat within the hepatocytes in the four lesions that became hypointense. In the lesion that remained hyperintense on fat-suppressed images, moderate steatosis was associated with sinusoidal dilatation and marked hemorrhagic areas.
|
|
In one case (Fig. 5), MRI showed a peripheral rim that was hyperintense on T2-weighted sequences and hypointense on T1-weighted sequences and thus suggested the presence of a pseudocapsule produced by edema or compressed liver tissue. In this case, histopathology showed only compression of adjacent liver tissue.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The prevalence of typical features of focal nodular hyperplasia on MRI has ranged from 8% to 63% [7, 11–14]. This wide variation may be ascribed to variations in the stringency of the criteria sets used to diagnose focal nodular hyperplasia, some of which included a dynamic contrast enhancement pattern. Another possible explanation is recruitment bias: patients recruited in surgical departments are more likely to have atypical MRI features requiring an invasive procedure to establish the definitive diagnosis.
Absence of a stellate area on MR images is not uncommon [7, 12–14]. In this study, eight (30%) of the 27 atypical focal nodular hyperplasia lesions, which represent 4% of the 186 focal nodular hyperplasia lesions seen among 116 patients at our institution during the same period, had no stellate area on MRI, and two of these eight had no fibrous area at pathologic examination. These findings are similar to those reported previously [7, 12–14]. In addition, the imaging diagnosis of focal nodular hyperplasia measuring less than 3 cm in diameter remains difficult [14], mainly because of the absence of a stellate area; nonetheless, in patients without any clinical suspicion of malignancy, a reasonable strategy may be to monitor small lesions without a stellate area rather than to perform a biopsy, provided all other typical MRI criteria for focal nodular hyperplasia are present.
Tissue heterogeneity revealed on MRI has been reported in 2–66% of focal nodular hyperplasia lesions [13–15]. We found that eight (30%) of 27 lesions were heterogeneous on T2-weighted sequences. It must be stressed that lesion heterogeneity on MR images occurs in malignant hepatic tumors such as hepatocellular carcinoma, cholangiocarcinoma, or hepatoblastoma and can be observed in hepatocellular adenoma, a benign tumor that must be resected [7, 16]. It therefore remains mandatory to obtain a histopathology diagnosis when a lesion is heterogeneous on MRI.
Hyperintensity of a liver lesion on T1-weighted sequences is usually ascribed to steatosis, sinusoidal dilatation, hemorrhagic foci, or copper accumulation [17]. Our findings confirm previous reports that steatosis is common in focal nodular hyperplasia [10, 18, 19]. Sinusoidal dilatation has also been reported in focal nodular hyperplasia lesions occurring in both oral contraceptive users and nonusers [10, 20, 21]. Marked sinusoidal ectasia with hemorrhagic areas is occasionally seen, especially in telangiectatic focal nodular hyperplasia. Of the 27 lesions in this study, seven (26%) were hyperintense on T1-weighted sequences as compared with the adjacent normal liver. All seven lesions showed at least one of the following histopathology features: sinusoidal dilatation, hemorrhagic foci, steatosis, and copper accumulation. These features may explain the hyperintensity that was the only atypical MRI feature in three of these seven lesions. Another lesion showed not only hyperintensity but also hypointense foci on T2-weighted sequences that probably corresponded to hemorrhage. In the remaining three cases, both hyperintensity on T1-weighted sequences and lesion heterogeneity were found. Although hyperintensity on T1-weighted sequences has been reported in more than half of hepatocellular adenomas, attributed to steatosis, necrosis, hemorrhage, or vascular lesions [17, 22], the other MRI features of focal nodular hyperplasia are not found.
The radiology–pathology correlation reported here should be interpreted in the light of several limitations of our study. First, the MR images were obtained between 1995 and 1999, and since then protocols for T1- and T2-weighted image acquisition have been improved; however, diagnostic criteria for focal nodular hyperplasia have not changed, to our knowledge. A recent study showed that helical MRI of Hyperplasia of the Liver CT also could be helpful for diagnosing focal nodular hyperplasia [23]. The CT features have been compared with pathology findings but not with MRI. Second, in 11 of the 27 focal nodular hyperplasia lesions in our patients, the histologic diagnosis was made on a core biopsy obtained using a 14-gauge needle. In theory, another lesion could have been missed by this procedure. This is highly unlikely, however, given that the biopsy was performed under sonographic guidance and that the diagnosis of focal nodular hyperplasia was based on the strict criteria we have described. Because atypical features were more often observed in patients who underwent a needle biopsy, it is possible that a sampling error may have been the major contributor to pathology atypia. On the other hand, pathologically typical lesions may look atypical on MRI; in such cases, dynamic gadolinium-enhanced MRI and liver-specific contrast agents could be helpful for diagnosis [24, 25].
Despite these limitations, our findings suggest that conservative treatment may be reasonable in patients with liver lesions that are hyperintense on T1-weighted images but fulfill all other criteria for focal nodular hyperplasia (three lesions in this study). We also suggest that lesions that are smaller than 3 cm in diameter and lack a stellate area on MRI but fulfill all other MRI criteria for focal nodular hyperplasia (two lesions in the present study) could be monitored by serial MRI studies (e.g., at yearly intervals for a period of 3 years, terminating if no change occurs over time). However, in large lesions without a stellate area and in lesions with heterogeneity, histopathology examination is mandatory to rule out other diagnoses.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Kottke, M. Horger, H. Schimmel, and M. Wehrmann Atypical Focal Nodular Hyperplasia with Cluster-Like Internal Cysts Due to Fibrinoid Necrosis Am. J. Roentgenol., November 1, 2007; 189(5): W247 - W250. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
