AJR:182, May 2004

1227

MRI of Atypical Focal Nodular Hyperplasia of the Liver:

Radiology–Pathology Correlation

OBJECTIVE.

The purpose of our study was to determine the MRI features of atypical fo-cal nodular hyperplasia of the liver and to compare them to pathology findings.

MATERIALS AND METHODS.

We retrospectively reviewed MRI and pathology find-ings in 27 focal nodular hyperplasia lesions with atypical MRI features. Six criteria for typicalfocal nodular hyperplasia were required: iso- or hypointensity on T1-weighted sequences andiso- or slight hyperintensity on T2-weighted sequences; homogeneous signal intensity; centralhyperintense area on T2-weighted sequences; marked lesion contrast enhancement; accumu-lation 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 atyp-ical, stellate area; heterogeneity on both T1- and T2-weighted images; and high-intensity sig-nal on T1-weighted sequences. MRI–pathology correlation showed that T1 hyperintensitywith no other atypical MRI feature (

n

= 3) could be explained by steatosis, sinusoidal dilata-tion, 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 se-quences or that lacks a stellate area but is smaller than 3 cm in diameter can be diagnosed asfocal nodular hyperplasia provided all other MRI criteria for this diagnosis are present. Insuch cases, close monitoring on MRI without invasive diagnostic procedures may be war-ranted. However, in large lesions (> 3 cm) without a stellate area and in lesions with heteroge-neity, histopathology examination is mandatory to rule out other diagnoses.

ocal nodular hyperplasia of theliver is a benign tumorlike lesionthat probably reflects a hyperplas-

tic response of the liver to a focal blood flowincrease related to a preexisting arterial mal-formation [1, 2]. The characteristic histologicpattern is a central or peripheral fibrous stellatearea with radiating septa containing mal-formed vascular structures and cholangiolarproliferation [1, 3]. Clinical and biochemicaldata are usually normal, although serum

γ

-glutamyl transpeptidase levels may be slightlyincreased [4]. Because no risk of bleeding ormalignant transformation exists, resection isnot required [5, 6]. This implies a need for anoninvasive investigation capable of establish-ing the diagnosis. Among imaging procedures,it has been suggested that MRI has 70% sensi-tivity and 98% specificity for the diagnosis offocal nodular hyperplasia [4]. MRI criteria fortypical focal nodular hyperplasia have been

described by Mattison et al. [7] and extendedby Mathieu et al. [8]. In cases in which MRIfeatures are atypical and might raise diagnosticdifficulties with other liver tumors that must beresected (e.g., hepatocellular adenoma > 4 cmor hepatocellular carcinoma), definite histo-logic proof must be obtained [4]. The aim ofthe present work was therefore to study andcorrelate the atypical MRI features of focalnodular hyperplasia and pathology findings toincrease the number of patients who avoid ad-ditional invasive diagnostic procedures.

Materials and Methods

Patients

From 1995 to 1999, at our institution, a diagno-sis of focal nodular hyperplasia was made in 116patients without underlying chronic liver disease.Ninety patients (77.6%) had the six typical MRIfeatures of focal nodular hyperplasia detailed in the

Sophie Ferlicot

1,2

Hicham Kobeiter

3

Jeanne Tran Van Nhieu

1

Daniel Cherqui

4

Daniel Dhumeaux

5

Didier Mathieu

3

Elie Serge Zafrani

1

Received June 24, 2002; accepted after revision November 6, 2003.

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. Address correspondence to S. Ferlicot.

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.

AJR

2004;182:1227–1231

0361–803X/04/1825–1227

© American Roentgen Ray Society

F

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by U

nive

rsity

of

Uta

h on

07/

04/1

4 fr

om I

P ad

dres

s 15

5.97

.178

.73.

Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved

1228

AJR:182, May 2004

Ferlicot et al.

following text, and none was biopsied. In the re-maining 26 patients (22.4%), including 22 womenand four men (mean age, 40 years; age range, 16–62 years), MRI findings were atypical, and focalnodular hyperplasia was diagnosed either on thesurgically resected lesion (

n

= 16) or on a liver corebiopsy obtained under laparoscopy using a 14-gauge needle (

n

= 11). Because one patient had twofocal nodular hyperplasia lesions, radiology–pa-thology correlations were studied in 27 focal nodu-lar hyperplasia lesions.

Imaging Techniques

All 116 patients underwent MRI at our institu-tion using a high-field (1.5-T) magnet (MagnetomSP3, Siemens). The protocol, which remained un-changed during the 5-year study period, includedaxial 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 en-tire liver. Dynamic gadolinium chelate–enhancedsubsecond gradient echo T1-weighted images(Turbo FLASH, Siemens; 7/3; inversion time, 400msec; flip angle, 10°) were then obtained after amanual bolus injection of gadolinium chelates(Dotarem [gadoterate dimeglumine], LaboratoireGuerbet) through an antecubital vein. This turbo fastlow-angle shot (FLASH) sequence was started in thearea of the lesion immediately after bolus injection(scanning delay ranging from 2 to 40 sec). A total of20 images were acquired every 2 sec. Finally, 4 minafter the bolus injection, delayed gadolinium che-late–enhanced spin-echo T1-weighted images wereobtained using the same parameters as in the unen-hanced study. Total duration of the diagnostic MRIstudy was 30 min.

For several lesions with hyperintensity on T1-weighted images, a fat-suppressed technique wasperformed by transmitting a chemically selective ra-diofrequency pulse centered on the frequency oflipid resonance to drop out the fat signal.

Image Analysis

All MRI documents were reviewed retrospec-tively by two radiologists specialized in liver imag-ing. In each patient, the radiologists sought toestablish a consensus about MRI features that mayhave a pathology correlative. The two radiologistswere blinded to the histology findings. The six im-aging criteria for typical focal nodular hyperplasiaon MRI were iso- or hypointensity on T1-weightedsequences and isointensity or slight hyperintensityon T2-weighted sequences; homogeneous signalintensity; presence of a central hyperintense areaon T2-weighted sequences; marked lesion en-hancement at the arterial phase after bolus injectionof a paramagnetic contrast agent; accumulation ofgadolinium chelates within the central area on de-layed contrast-enhanced T1-weighted sequences;and absence of capsule surrounding the lesion onthese sequences [7–9]. The 90 patients with typi-cal focal nodular hyperplasia on MRI satisfied allsix criteria.

Pathology Study

Gross findings were abstracted from the medicalcharts, and the histology slides were reviewed on adouble-headed microscope by two pathologists. Oneof the two pathologists was a liver specialist with 25years’ experience. The following gross features wererecorded: number, size, shape, color, lobulation, andpresence or absence of a stellate area; hemorrhage;and necrosis. Histologic sections of paraffin-embed-ded material were stained with H and E,

picrosiriusred, Perls’ stain for iron, and modified orcein andrhodanine for copper-associated protein and copper.The following histopathology features were re-corded: presence or absence of fibrosis (i.e., stellatefibrous area on resected specimens and fibrousbands on resected specimens or needle biopsies),vascular lesions (i.e., arterial or venous abnormali-ties, sinusoidal dilatation, and hemorrhagic foci), in-flammatory infiltration, cholangiolar proliferation,hepatocellular alterations (i.e., cytologic atypia,ballooning degeneration, cytoplasmic inclusions,and steatosis), and pigment accumulation. The di-agnosis of typical focal nodular hyperplasia wasmade when the following pathology features wereobserved: presence of hyperplastic nodules madeof normal hepatocytes and separated by fibroussepta containing abnormal vascular structures, cho-langiolar proliferation, and inflammatory infil-trates. Focal nodular hyperplasia was consideredatypical when these three pathology criteria werenot observed altogether. Normal interlobular bileducts were never seen within fibrous areas. Otherpathology features that are unusual in focal nodularhyperplasia (i.e., steatosis, hemorrhagic foci, sinu-soidal dilatation, necrosis, and pigment accumula-tion) were recorded and semiquantified (mild,moderate, and marked scale).

Results

MRI Findings

Among the 27 radiologically atypical focalnodular hyperplasia lesions, more than oneatypical MRI feature was noted in 10 cases.

No stellate area could be identified on T2-weighted and on both unenhanced and contrast-enhanced T1-weighted sequences in eightlesions (Fig. 1). In four other lesions, the stellatearea was atypical, with hypointensity on T2-weighted sequences (

n

= 1), persistent hy-pointensity on contrast-enhanced sequences(

n

= 2) (Fig. 2), or visibility before but not aftercontrast injection (

n

= 1).Eight lesions had heterogeneous signal inten-

sity on both T1- and T2-weighted sequences(Figs. 1–3). Seven lesions were hyperintense bycomparison with the healthy liver on T1-weighted sequences.

Lesion diameter of the 16 resected focalnodular hyperplasia lesions ranged from 1.5to 14 cm (mean, 7 cm). Two of these lesionswere less than 3 cm in diameter, and theironly atypical MRI feature was the absence ofa stellate area.

Diameter on MR images of the 11 nonre-sected focal nodular hyperplasia lesions rangedfrom 2 to 4 cm (mean, 3.3 cm). Two of these le-sions were less than 3 cm in diameter.

Other atypical findings were observed, includ-ing hypointensity on T2-weighted sequences, ab-sence of or heterogeneous enhancement aftercontrast material injection, presence of a periph-eral rim simulating a capsule or of hypointense

Fig. 1.—43-year-old woman with focal nodular hyperplasia. Spin-echo T2-weighted image shows 6-cm lesion(arrows) in segment VII of liver, displacing both right hepatic vein and inferior vena cava. Lesion is heteroge-neous in posterior part and slightly hyperintense with no central scar. Microscopically, lesion showed hyperplas-tic nodules with no stellate area or thick fibrous septa.

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by U

nive

rsity

of

Uta

h on

07/

04/1

4 fr

om I

P ad

dres

s 15

5.97

.178

.73.

Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved

MRI of Hyperplasia of the Liver

AJR:182, May 2004

1229

foci on T1- and T2-weighted sequences, in-crease in size on two successive MRI exami-nations, and intrahepatic biliary duct dilatationdue to large bile duct compression by a lesionlocated near the hilum.

Pathology Findings

Macroscopically, among the 16 resected fo-cal nodular hyperplasia lesions, three showedred hemorrhagic areas.

Fourteen (52%) of the 27 focal nodular hy-perplasias showed typical histology patterns.The remaining 13 lesions (48%) had one or

more of the following atypical histology fea-tures: absence of a stellate area or of thick fi-brous bands (

n

= 5), absence of abnormalvessels (

n

= 7), and absence of cholangiolar pro-liferation (

n

= 3). Among the 27 focal nodularhyperplasia lesions, 18 (67%) exhibited sinusoi-dal dilatation—five of mild degree, seven ofmoderate degree, and six of marked degree. Tenlesions (37%) contained hemorrhagic areas, in-cluding three mild, four moderate, and threemarked. 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 presenceof foci of atypical hepatocytes (

n

= 4).

Radiology–Pathology Correlation

In two of the eight lesions with no visiblestellate area on both T1- and T2-weighted se-quences, no thick fibrous bands were seen atpathology examination. Among the five pa-tients with a radiologically atypical stellatearea, two had no thick fibrous bands, two hadfibrosis with abnormal vessels, and one had fi-brosis with normal vessels. In one patient, thestellate area seen on MRI was not found in theneedle biopsy specimen.

Of the eight lesions that were heteroge-neous on T1- and T2-weighted sequences,seven showed sinusoidal dilatation, accom-panied in six cases by hemorrhagic foci. Inthree cases, sinusoidal dilatation wasmarked. Steatosis was observed in six ofthese eight cases.

Sinusoidal dilatation was seen in six ofthe seven lesions that were hyperintense onT1-weighted sequences. Hemorrhagic fociwere visible in three of these seven lesions,always in association with sinusoidal dilata-tion. Steatosis was noted in six of theseseven lesions and was accompanied by sinu-soidal dilatation or hemorrhage in five lesions.On fat-suppressed T1-weighted images, fourlesions became hypointense, whereas one re-mained hyperintense (Fig. 4). Pathology ex-amination showed accumulation of fat

A B

Fig. 2.—46-year-old woman with focal nodular hyperplasia.A, Spin-echo T1-weighted image shows slightly heterogeneous lesion (arrow) in left lobe. Lesion contains central hypointense scar (ar-rowhead).B, Contrast-enhanced spin-echo T1-weighted image obtained 4 min after bolus injection reveals persistent hypointensity of central scar(arrowhead).

Fig. 3.—37-year-old woman with focal nodular hyperplasia. Spin-echo T2-weighted image shows hyperintenseand heterogeneous round lesion (arrow) of left lobe, with round highly hyperintense area in its center. Micro-scopically, lesion showed large area of hemorrhagic necrosis within hepatic nodule.

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by U

nive

rsity

of

Uta

h on

07/

04/1

4 fr

om I

P ad

dres

s 15

5.97

.178

.73.

Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved

1230

AJR:182, May 2004

Ferlicot et al.

within the hepatocytes in the four lesionsthat became hypointense. In the lesion thatremained hyperintense on fat-suppressedimages, moderate steatosis was associatedwith sinusoidal dilatation and marked hem-orrhagic areas.

In one case (Fig. 5), MRI showed a pe-ripheral rim that was hyperintense on T2-weighted sequences and hypointense onT1-weighted sequences and thus suggestedthe presence of a pseudocapsule produced byedema or compressed liver tissue. In thiscase, histopathology showed only compres-sion of adjacent liver tissue.

Discussion

Several focal nodular hyperplasia morpho-logic variants, which account for approxi-mately 20% of cases, have been described: thetelangiectatic variant, the mixed hyperplasticand adenomatous variant, and lesions withlarge cell hepatocellular atypia [10]. These lastchanges resembling large cell hepatocellulardysplasia may represent an adaptive responseof the liver, for instance to intralesionalcholestasis [10]. In our series, five of the 13histologically atypical lesions had no stellatearea or thick fibrous septa on pathologic exam-ination and exhibited microscopic features

mimicking liver cell adenoma. Presence of ab-normal vessels and minimal cholangiolar pro-liferation were the two findings in favor offocal nodular hyperplasia in three cases. Thediagnosis was more difficult in one case with-out cholangiolar proliferation and in anothercase with no abnormal vessels and minimalcholangiolar proliferation. This last patternwas consistent with the mixed hyperplasticand adenomatous form of focal nodular hyper-plasia, part of the lesion having been missed atbiopsy [10].

The prevalence of typical features of focalnodular hyperplasia on MRI has ranged from8% to 63% [7, 11–14]. This wide variationmay be ascribed to variations in the strin-gency of the criteria sets used to diagnose fo-cal nodular hyperplasia, some of whichincluded a dynamic contrast enhancementpattern. Another possible explanation is re-cruitment bias: patients recruited in surgicaldepartments are more likely to have atypicalMRI features requiring an invasive proce-dure to establish the definitive diagnosis.

Absence of a stellate area on MR images isnot uncommon [7, 12–14]. In this study, eight(30%) of the 27 atypical focal nodular hyper-plasia lesions, which represent 4% of the 186focal nodular hyperplasia lesions seen among116 patients at our institution during the sameperiod, had no stellate area on MRI, and two ofthese eight had no fibrous area at pathologicexamination. These findings are similar tothose reported previously [7, 12–14]. In addi-tion, the imaging diagnosis of focal nodularhyperplasia measuring less than 3 cm in diam-eter remains difficult [14], mainly because of

A B

Fig. 4.—22-year-old woman with focal nodular hyperplasia of left lobe.A, Spin-echo T1-weighted image shows slightly hyperintense foci (arrows) circumscribing lesion.B, Fat-suppressed spin-echo T1-weighted image reveals persistent hyperintensity of lesion (arrows).

Fig. 5.—39-year-old woman with focal nodular hyperplasia of left lobe. Spin-echo T2-weighted image shows hy-perintense rim (arrow) simulating capsule around lesion. This peripheral rim was hypointense on T1-weightedsequences and corresponded to compressed adjacent liver tissue on histology examination.

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by U

nive

rsity

of

Uta

h on

07/

04/1

4 fr

om I

P ad

dres

s 15

5.97

.178

.73.

Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved

MRI of Hyperplasia of the Liver

AJR:182, May 2004

1231

the absence of a stellate area; nonetheless, inpatients without any clinical suspicion of ma-lignancy, a reasonable strategy may be to mon-itor small lesions without a stellate area ratherthan to perform a biopsy, provided all othertypical MRI criteria for focal nodular hyper-plasia are present.

Tissue heterogeneity revealed on MRI hasbeen reported in 2–66% of focal nodular hy-perplasia lesions [13–15]. We found that eight(30%) of 27 lesions were heterogeneous onT2-weighted sequences. It must be stressedthat lesion heterogeneity on MR images oc-curs in malignant hepatic tumors such as hepa-tocellular carcinoma, cholangiocarcinoma, orhepatoblastoma and can be observed in hepa-tocellular adenoma, a benign tumor that mustbe resected [7, 16]. It therefore remains man-datory to obtain a histopathology diagnosiswhen a lesion is heterogeneous on MRI.

Hyperintensity of a liver lesion on T1-weighted sequences is usually ascribed to ste-atosis, sinusoidal dilatation, hemorrhagic foci,or copper accumulation [17]. Our findingsconfirm previous reports that steatosis is com-mon in focal nodular hyperplasia [10, 18, 19].Sinusoidal dilatation has also been reported infocal nodular hyperplasia lesions occurring inboth oral contraceptive users and nonusers [10,20, 21]. Marked sinusoidal ectasia with hem-orrhagic areas is occasionally seen, especiallyin telangiectatic focal nodular hyperplasia. Ofthe 27 lesions in this study, seven (26%) werehyperintense on T1-weighted sequences ascompared with the adjacent normal liver. Allseven lesions showed at least one of the fol-lowing histopathology features: sinusoidal di-latation, hemorrhagic foci, steatosis, andcopper accumulation. These features may ex-plain the hyperintensity that was the only atyp-ical MRI feature in three of these sevenlesions. Another lesion showed not only hy-perintensity but also hypointense foci onT2-weighted sequences that probably corre-sponded to hemorrhage. In the remaining threecases, both hyperintensity on T1-weighted se-quences and lesion heterogeneity were found.Although hyperintensity on T1-weighted se-quences has been reported in more than half ofhepatocellular adenomas, attributed to steato-sis, necrosis, hemorrhage, or vascular lesions[17, 22], the other MRI features of focal nodu-lar hyperplasia are not found.

The radiology–pathology correlation re-ported here should be interpreted in the light ofseveral limitations of our study. First, the MRimages were obtained between 1995 and 1999,and since then protocols for T1- and T2-weighted image acquisition have been im-proved; however, diagnostic criteria for focalnodular hyperplasia have not changed, to our

knowledge. A recent study showed that helicalCT also could be helpful for diagnosing focalnodular hyperplasia [23]. The CT featureshave been compared with pathology findingsbut not with MRI. Second, in 11 of the 27 fo-cal nodular hyperplasia lesions in our patients,the histologic diagnosis was made on a corebiopsy obtained using a 14-gauge needle. Intheory, another lesion could have been missedby this procedure. This is highly unlikely,however, given that the biopsy was performedunder sonographic guidance and that the diag-nosis of focal nodular hyperplasia was basedon the strict criteria we have described. Be-cause atypical features were more often ob-served in patients who underwent a needlebiopsy, it is possible that a sampling error mayhave been the major contributor to pathologyatypia. On the other hand, pathologically typi-cal lesions may look atypical on MRI; in suchcases, dynamic gadolinium-enhanced MRIand liver-specific contrast agents could behelpful for diagnosis [24, 25].

Despite these limitations, our findings sug-gest that conservative treatment may be rea-sonable in patients with liver lesions that arehyperintense on T1-weighted images but fulfillall other criteria for focal nodular hyperplasia(three lesions in this study). We also suggestthat lesions that are smaller than 3 cm in di-ameter and lack a stellate area on MRI butfulfill all other MRI criteria for focal nodularhyperplasia (two lesions in the present study)could be monitored by serial MRI studies(e.g., at yearly intervals for a period of 3years, terminating if no change occurs overtime). However, in large lesions without astellate area and in lesions with heterogeneity,histopathology examination is mandatory torule out other diagnoses.

References

1. Ishak KG, Rabin L. Benign tumors of the liver.

MedClin North Am

1975;59:995–10132. Wanless IR, Mawdsley C, Adams R. On the patho-

genesis of focal nodular hyperplasia of the liver.

Hepatology

1985;5:1194–12003. Craig JR, Peters RL, Edmondson HA.

Tumors of theliver and intrahepatic bile ducts: atlas of tumor pa-thology,

2nd series, fasc. 26. Washington, DC:Armed Forces Institute of Pathology, 1989:8–62

4. Cherqui D, Rahmouni A, Charlotte F, et al. Man-agement of focal nodular hyperplasia and hepato-cellular adenoma in young women: a series of 41patients with clinical, radiological, and pathologicalcorrelations.

Hepatology

1995;22:1674–16815. Mathieu D, Vilgrain V, Mahfouz AE, et al. Benign

liver tumors.

Magn Reson Imaging Clin N Am

1997;5:255–2886. Belghiti J, Pateron D, Panis Y, et al. Resection of

presumed benign liver tumours.

Br J Surg

1993;80:380–383

7. Mattison GR, Glazer GM, Quint LE, et al. MR im-

aging of hepatic focal nodular hyperplasia: charac-terization and distinction from primary malignanthepatic tumors.

AJR

1987;148:711–7158. Mathieu D, Rahmouni A, Anglade MC, et al. Focal

nodular hyperplasia of the liver: assessment withcontrast- enhanced turboFLASH MR imaging.

Ra-diology

1991;180:25–309. Mortele KJ, Praet M, Van Vlierberghe H, de Hemp-

tinne B, Zou K, Ros PR. Focal nodular hyperplasiaof the liver: detection and characterization withplain and dynamic-enhanced MRI.

Abdom Imaging

2002;27:700–70710. Nguyen BN, Flejou JF, Terris B, et al. Focal nodular

hyperplasia of the liver: a comprehensive pathologicstudy of 305 lesions and recognition of new histologicforms.

Am J Surg Pathol

1999;23:1441–145411. Golli M, Mathieu D, Anglade MC, et al. Focal nod-

ular hyperplasia of the liver: value of color DopplerUS in association with MR imaging.

Radiology

1993;187:113–11712. Irie H, Honda H, Kaneko K, et al. MR imaging of

focal nodular hyperplasia of the liver: value of con-trast-enhanced dynamic study.

Radiat Med

1997;15:29–3513. Lee MJ, Saini S, Hamm B, et al. Focal nodular hy-

perplasia of the liver: MR findings in 35 provedcases.

AJR

1991;156:317–32014. Vilgrain V, Fléjou JF, Arrivé L, et al. Focal nodular

hyperplasia of the liver: MR imaging and patho-logic correlation in 37 patients.

Radiology

1992;184:699–703

15. Schiebler ML, Kressel HY, Saul SH, et al. MR im-aging of focal nodular hyperplasia of the liver.

JComput Assist Tomogr

1987;11:651–65416. Powers C, Ros PR, Stoupis C, et al. Primary liver

neoplasms: MR imaging with pathologic correla-tion.

RadioGraphics

1994;14:459–48217. Mathieu D, Paret M, Mahfouz AE, et al. Hyperin-

tense benign liver lesions on spin-echo T1-weightedMR images: pathologic correlations.

Abdom Imag-ing

1997;22:410–41718. Benz EJ, Baggenstoss AH. Focal cirrhosis of the

liver: its relation to the so-called hamartoma (ade-noma, benign hepatoma).

Cancer

1953;6:743–75519. Ishak KG. Hepatic neoplasms associated with con-

traceptive and anabolic steroids. In: Lingeman CH,ed.

Recent results in cancer research

. New York,NY: Springer-Verlag, 1979:73–128

20. Zafrani ES, Pinaudeau Y, Dhumeaux D. Drug-in-duced vascular lesions of the liver.

Arch Intern Med

1983;143:495–50221. Colle I, de Beeck BO, Hoorens A, et al. Multiple fo-

cal nodular hyperplasia.

J Gastroenterol

1998;33:904–908

22. Arrivé L, Fléjou JF, Vilgrain V, et al. Hepatic ade-noma: MR findings in 51 pathologically proved le-sions.

Radiology

1994;193:507–51223. Brancatelli G, Federle MP, Grazioli L, Blachar A,

Peterson MS, Thaete L. Focal nodular hyperplasia:CT findings with emphasis on multiphasic helicalCT in 78 patients.

Radiology

2001;219:61–6824. Mahfouz AE, Hamm B, Taupitz M, Wolf KJ. Hy-

pervascular liver lesions: differentiation of focalnodular hyperplasia from malignant tumors withdynamic gadolinium-enhanced MR imaging.

Radi-ology

1993;186:133–13825. Ba-Ssalamah A, Schima W, Schmook MT, et al.

Atypical focal nodular hyperplasia of the liver: im-aging features of nonspecific and liver-specific MRcontrast agents.

AJR

2002;179:1447–1456

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by U

nive

rsity

of

Uta

h on

07/

04/1

4 fr

om I

P ad

dres

s 15

5.97

.178

.73.

Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved