CASE REPORTS

Undifferentiated Embryonal Sarcoma withUnusual Features Arising withinMesenchymal Hamartoma of the Liver:Report of a Case and Review of theLiterature

MAUREEN J. O’SULLIVAN,1 PAUL E. SWANSON,1 JOAN KNOLL,2

EUGENIO M. TABOADA,3 AND LOUIS P. DEHNER1*1L. V. Ackerman Laboratory of Surgical Pathology, Washington University Medical Center, Campus Box 8118,660 S. Euclid Avenue, St. Louis, MO 63110, USA2Department of Cytogenetics, Children’s Mercy Hospital, 2401 Gillham Road, Kansas City, MO 64108, USA3Department of Pathology, Children’s Mercy Hospital, 2401 Gillham Road, Kansas City, MO 64108, USA

Received May 19, 2000; accepted January 31, 2001.

ABSTRACTUndifferentiated embryonal sarcoma (UES) is a rare andhighly malignant hepatic neoplasm, affecting almost ex-clusively the pediatric population. It has replaced malig-nant mesenchymoma, under which diagnostic term thefirst three cases were described. A link between embry-onal sarcoma and mesenchymal hamartoma of the liver(MHL) has long been proposed, because of clinicopath-ologic overlaps of these entities; however, until recently,this association remained tenuous. Cases of UES arisingin a background of mesenchymal hamartoma of the liverhave previously been reported in two teenage girls. Dis-covery of a similar genetic abnormality in MHL and UEShas clinched the supposed link between them. Therehave also been two reports of UES with prominent cys-tification, one associated with peripheral eosinophilia,and thereby masquerading as hydatid cyst of the liver.We report a case of UES arising in a young boy withMHL, with unusual histologic features, including large

mesothelial-lined cysts and ectopic adrenal cortical tis-sue under Glisson’s capsule.

Key words: undifferentiated embryonal sarcoma, mes-enchymal hamartoma, mesothelium

INTRODUCTIONUndifferentiated embryonal sarcoma (UES) of the

liver is a unique tumor type, first described by

Willis as a rhabdomyoblastic mixed tumor, in his

discussion of hepatic neoplasms of childhood [1].

This was identified by Willis from two reports in

the literature of intrahepatic neoplasms as a “dis-

tinct group” of tumors, possibly related to embry-

onal rhabdomyosarcoma of the extrahepatic bile

ducts. In 1973, Stanley and associates reported

three cases of UES with the appelation of “malig-

nant mesenchymoma” [2]; this study was then fol-

lowed by a study of 31 cases by Stocker and Ishak,*Corresponding author

Pediatric and Developmental Pathology 4, 482–489, 2001

DOI: 10.1007/s10024-001-0047-9

Pediatric and Developmental Pathology

© 2001 Society for Pediatric Pathology

who proposed the designation “undifferentiated(embryonal) sarcoma” [3].

Two themes have been recurrent in the liter-ature since UES came to be accepted as an entity:the suggestion of a link between UES and mesen-chymal hamartoma of the liver (MHL) [2–6], andthe related question about the progenitorship orhistogenesis of this neoplasm. There is a consensusthat UES is not an intrahepatic rhabdomyosar-coma. Support for a relationship between UES andMHL is based on two reports in the literature [7,8],one of which demonstrated a shared genetic aber-ration in the UES and MHL areas of the tumor.This study documents our experience with a com-posite MHL and UES in the liver of a 3-year-oldchild.

CASE REPORTA 35-month-old, previously healthy Hispanic boypresented with anorexia, abdominal pain, andincreasing abdominal girth, which preventedhim from bending over. On presentation, he wasanicteric and lacked ecchymoses or evidence of ableeding diathesis. Radiologic studies showed aheterogeneous, predominantly cystic intrahe-patic mass with dense septations (Fig. 1A). Nointraabdominal lymphadenopathy was identifiedand the lung fields were clear. Laboratory inves-tigations showed a normal serum a-fetoproteinlevel (1 mg/dl), normal ferritin, albumin, plasmathromboplastin (PT) and activated partialthromboplastin time (APTT), elevated aspartateaminotransferase (AST) (56 U/liter), alanine ami-notransferase (ALT) (65 U/liter), and low fibrin-ogen (108 mg/dl). The clinical impression wasthat this represented a mesenchymal hamar-toma.

A hepatic trisegmentectomy was performed.The gross specimen weighed 2090 g (460 g fol-lowing drainage of the cyst fluid) and measured19.5 3 17.5 3 12.5 cm. The lesion was welldemarcated from the surrounding hepatic paren-chyma by a fibrous pseudocapsule (Fig. 1B). Am-ber, slightly mucoid fluid filled the multiple thin-walled cysts, which ranged from 0.3 to 3.0 cm indiameter.

Five cycles of chemotherapy comprising cis-platin, vincristine, cyclophosphamide, actinomy-cin-D, and doxorubicin were well tolerated. Mag-

netic resonance imaging (MRI) demonstrated noevidence of disease 1 month post-surgery. Fivemonths post-surgery, the child remains clinicallywell.

METHODSImmunohistochemistryAntibodies to vimentin, pan-cytokeratin (cocktailof AE1/AE3, CAM5.2, and MAK6), epithelial mem-brane antigen (EMA), desmin, muscle-specific ac-tin, a-1-antitrypsin, and a-fetoprotein were appliedto 3- to 5-mm-thick formalin-fixed, paraffin-embed-ded tissue sections on poly-L-lysine–coated slides(Table 1). All primary antibodies were incubatedwith tissue sections for 18 h at 4°C in moisturechambers. Slides were then developed using amodified avidin-biotin peroxidase complex methodas previously described [9], using Vector Elite ABCkits (Vector Labs, Burlingame, CA). Heat-inducedepitope retrieval (HIER) with citrate buffer (0.01M, pH 6.0) was carried out for each of the antibod-ies prior to primary incubation. All HIER was com-

Figure 1. Ultrasound scan demonstrates a compositehepatic lesion with dense septations (A); grossly, themass is well delineated and contains multiple thin-walled cysts separated by dense fibrous tissue (B).

UNDIFFERENTIATED EMBRYONAL SARCOMA 483

pleted in a 900 W microwave oven at 70% powerfor 7 min.

Electron microscopic analysisElectron microscopy was performed with freshstarting tissue fixed in 2% glutaraldehyde in phos-phate buffer, postfixed in 1% osmium tetroxide,and embedded in Spurr’s epoxy resin. Additionalmaterial was dissected from the formalin-fixed,paraffin-embedded tissue block, deparaffinized inxylene, rehydrated, fixed in glutaraldehyde, andpostfixed in osmium tetroxide prior to embeddingin epoxy resin. Thick sections, stained with tolu-idine blue, were used to select areas for furtherstudy. Ultrathin sections, stained with uranyl ace-tate and lead citrate, were examined using a Phil-ips CM10 electron microscope.

Cytogenetic evaluationGTG-banding was carried out on metaphase chro-mosomes according to protocol [10] prior to karyo-typic analysis.

RESULTSA population of primitive oval-to-spindled cellswith hyperchromatic nuclei and containingscant amounts of cytoplasm in a myxoid back-ground (Fig. 2A) was condensed, but withoutevidence of cambium layer formation, and sub-jacent particularly to the larger cystic spaces(Fig. 2A inset) within this mass. Occasional, iso-lated hyaline globules were identified within tu-

mor cells. The appearance of the mass was oth-erwise that of a mesenchymal hamartoma of theliver, with densely hyalinized, collagenous col-lars ensnaring serpiginous bile ducts and associ-ated islands of unremarkable hepatocytes (Fig.2B). Focal infiltration of these hepatocyte islandsby tumor cells was noted. The tumor cell popu-lation formed a minor component (,5% total;present in 9 of 20 sections) of the mass (Fig. 2C).Subjacent to Glisson’s capsule was ectopic adre-nal cortical tissue (Fig. 2D). The cystic spaceswithin the mass were lined by mesothelial cells(Fig. 2E inset), the nature of which was con-firmed by electron microscopic examination(Fig. 2E). These lining cells were co-reactive forvimentin and cytokeratin (Fig. 3A, B). The tumorcells showed diffuse, strong reactivity for vimen-tin and a-1-antitrypsin (Fig. 3A, C). Punctateparanuclear reactivity for cytokeratin was alsonoted (Fig. 3B) and isolated tumor cells ex-pressed desmin (Fig. 3D). Epithelial membraneantigen, muscle-specific actin, and a-fetoproteinstains were all negative. These findings are inkeeping with those previously reported for UES[11–13]. In all, electron microscopic examinationled to the identification of four different cell pop-ulations, including hepatocytes, bile duct epithe-lium, mesothelial-like cells lining the cysticspaces, and a population of cells without differ-entiating features, interpreted as representingthe tumor. These cells lacked intercellular junc-tions, had high nuclear/cytoplasmic ratios, and

Table 1. Immunoperoxidase panel

AntibodySpecies andisotype Dilution Pretreatment Manufacturer

Vimentin Mouse IgG 1:10,000 Citrate HIER Biogenex, San Ramon, CA

CK-MAK6 Mouse IgG 1:50 Citrate HIER Zymed, South San Francisco, CA

CK-AE1/AE3 Mouse IgG 1:150 Citrate HIER Boehringer-Mannheim, Indianapolis, IN

CK-CAM 5.2 Mouse IgG 1:50 Citrate HIER Becton-Dickinson, San Jose, CA

EMA Mouse IgG 1:3,000 Citrate HIER Dako, Carpinteria, CA

Desmin Mouse IgG 1:400 Citrate HIER Dako, Carpinteria, CA

MSA Mouse IgG 1:400 Citrate HIER Biogenex, San Ramon, CA

a-1-Antitrypsin Rabbit IgG 1:20,000 Citrate HIER Dako, Carpinteria, CA

a-Fetoprotein Rabbit IgG 1:10,000 Citrate HIER Dako, Carpinteria, CA

HIER, heat-induced epitope retrieval.

484 M.J. O’SULLIVAN ET AL.

prominent nucleoli. Organelles were not readilyappreciated in these cells. Haphazardly arrangedcollagen bundles invested these cells without anyintervening basal lamina.

Cytogenetic examination of 20 cells at the400- to 450-band resolution level, as determined bythe trypsin-Giemsa method, showed the presenceof a balanced t(11;19)(q11;q13.3/13.4) chromo-somal translocation in all cells examined (Fig. 4).

DISCUSSIONUndifferentiated embryonal sarcoma is a very un-common primary malignant neoplasm of the liver[4]. It has a predilection for children between theages of 6 and 10 years [3] and is the third mostcommon primary hepatic tumor between 5 and 20years of age [14], exceeded only by hepatocellularcarcinoma and focal nodular hyperplasia in thisage-group. Individual cases have been reported inadults. Until recently, the clinical outcome for UEShas been uniformly dismal [3], but aggressive mul-tidrug chemotherapy has resulted in some durableremissions [11,15].

Two previous cases [7,8] and the present onehave reported the coexistence of UES with MHL,supporting the earlier hypothesis of a histogeneticrelationship between these two tumefactive lesionsof the liver [2–6]. Stocker and Ishak noted thepresence of MHL-like features in the backgroundof UES [3]. Even assuming that some or all UESsoriginate in the setting of MHL, the lineage andphenotype of this neoplasm have not yet been es-tablished with any degree of certainty. Suggestionsthat UES is a sarcomatoid variant of hepatoblas-toma [12] have not found acceptance; rather, thereis broad agreement that the neoplastic cells havemesenchymal rather than epithelial characteristicsat the ultrastructural and immunophenotypiclevel. The tumor cells have primitive to poorly dif-ferentiated features, limiting the value of these an-cillary techniques in the attempt to identify theprogenitor of UES. At one time or another, myoid[16], myofibroblastic [17]; fibroblastic [6], and fi-brohistiocytic [18,19] lineages have all been pro-posed as the constituent of UES. Parham and as-sociates have suggested a pluripotent stem cellorigin [20]. These investigators found that approx-imately 50% of UESs expressed desmin and/ormuscle-specific actin. A mucoid anaplastic hepato-

blastoma reported by Joshi and associates [21] hadsome features reminiscent of UES with bizarregiant cells, but an elevated a-fetoprotein and junc-tional complexes by electron microscopy led to theformer rather than the latter interpretation. Seruma-fetoprotein is generally not elevated in cases ofUES. One might perhaps anticipate a greater de-gree of pleomorphism in UES than was appreci-ated in this particular case. However, there is ananalogy here with the cystic type of pleuropulmo-nary blastoma I and the complex multipatterningtype III. As these tumors enlarge and progress,there is a tendency for the morphology to becomemore complex, and this is usually what happenswith UES as we understand it as a clinical entity,rather than what we observed in this particularcase of an incidental finding of UES within anMHL.

Mesenchymal hamartoma has been regardedas a developmental anomaly of the liver, with nowell-established pathogenesis, other than that itappears to represent some aberration at the level ofthe portal tracts, with participation of the bileducts and supporting mesenchyme. Given the sim-ilarities between the bile duct abnormalities inMHL and those in von Meyenburg complexes, bileduct hamartomas, Caroli disease, and congenitalhepatic fibrosis [22]; a primary bile duct plate mal-formative etiology [23] has been proposed forMHL as well. Serial dissection studies have dem-onstrated a single portal tract as being the sourceof the lesion [24,25]. If there is a histogenetic rela-tionship between UES and MHL, it is not readilyappreciated from the perspective of a bile ductplate maldevelopment. Von Schweinitz and asso-ciates [26] examined four MHLs by immunohisto-chemistry and found that the stromal cells ex-pressed desmin and a-actin, which is interesting inlight of the myogenic immunophenotype of themalignant cells in some UESs, including the onereported here. The stromal cells of MHL also ex-pressed the proliferation antigen Ki67 [26].

The notion of malignant transformation oc-curring in a dysgenetic or hamartomatous lesion isone with any number of examples, such as adeno-carcinomas arising in bronchogenic and chole-dochal cysts, Wilms tumor from perilobar nephro-genic rests, and pleuropulmonary blastoma frompresumed congenital lung cysts. It has become

UNDIFFERENTIATED EMBRYONAL SARCOMA 485

Figure 2. The undifferentiated embryonal sarcoma iscomposed of primitive spindle cells with scant cytoplasmand hyperchromatic nuclei in a myxoid background (A);there is condensation of the tumor cells beneath the cystlining (A, inset). Directly adjacent, the typical features ofmesenchymal hamartoma of the liver (MHL) are noted, in-cluding densely hyalinized collagenous collars around ser-piginous bile ducts and interspersed islands of benign, en-trapped hepatocytes (B). A low-power overview indicates

relative proportions of tumor to mesenchymal hamartomain the liver mass (C). Ectopic adrenal cortical tissue is locatedunder Glisson’s capsule (D). The cysts are lined by hobnailedcells with bland-appearing nuclear features, consistentwith mesothelial cells (E, inset), the nature of which is con-firmed by electron microscopy demonstrating roundedcells with bland nuclear features and surface microvilli. In-tercellular junctions are noted and the cells contain looselyaggregated filament bundles (E).

clear that many of the genes critical to develop-ment, differentiation, and growth are the very onesthat are mutated in association with a particular

tumor type. For instance, the WT1 gene product, a

zinc finger transcriptional activator of several

growth factor genes and repressor of others, is

expressed in developing kidney, ovary, and testis

[27,28]. A point mutation in WT1 is associated

with Denys-Drash syndrome, which includes oc-

currence of Wilms tumor [29]. It should therefore

perhaps come as no major surprise that there is a

pathogenetic association between UES and MHL.

Although most MHLs have been diploid by

flow cytometry, a minority have been aneuploid

[30]. An identical breakpoint on chromosome 19

has been reported in three separate cases of MHL

[31–33]; two of which had t(11;19)(q13;q13.4)

and one with t(15;19)(q15;q13.4). One of these,

the t(11;19)(q13;q13.3) translocation, was de-

tected in our case. The presence of a consistent,

identical breakpoint on one of the involved chro-

mosomes is a well-recognized phenomenon that

Figure 3. Strong and diffuse reactivity of tumor cellsand mesothelial lining cells is noted for vimentin (A).Cytokeratin is strongly reactive in the cyst lining cells andthe entrapped hepatocytes and produces punctate

paranuclear staining in the tumor cells (B); a-1-antitryp-sin produces a granular cytoplasmic stain in the tumorcells (C). Desmin reactivity is noted in occasional tumorcells (D).

Figure 4. Karyotypic analysis demonstrates the pres-ence of a chromosomal translocation involving chromo-somes 19q13.3/13.4 and 11q11.

UNDIFFERENTIATED EMBRYONAL SARCOMA 487

has been observed in many other tumors of di-verse type [34].

Our case is unique and distinct from the pre-viously reported examples of MHL with the break-point at 19q13.4, since it contained a microscopi-cally detected UES. We are unable to state withcertainly whether the cells from this tumor withthe translocation were from the hamartomatous orsarcomatous areas. Although the dominant com-ponent had the histologic features of a MHL, thecultured cells did not resemble fibroblasts. Onemight surmise from the limited molecular geneticdata available that the breakpoint at 19q13.4 isinsufficient in and of itself for the development ofa sarcoma, since the overwhelming majority ofMHLs have a benign clinical course. Rather, ourfinding would support the hypothesis that MHL isa precursor of some UESs.

Yet other unusual aspects of our case of amicroscopic UES arising in a background of MHLwere the presence of ectopic adrenal tissue andmesothelial-lined cysts. To our knowledge suchcysts have yet to be described in either MHL orUES. Both MHL and UES arising within MHLhave been associated with other perturbations indevelopment, such as an annular pancreas andtracheoesophageal fistula [35]. The adrenal heter-otopia in our case is fascinating from the perspec-tive that one of the two previously reported cases ofUES arising in a MHL also had ectopic adrenaltissue beneath Glisson’s capsule [7].

The final interesting and provocative issueraised by our case is that of the best approach tothe clinical management of MHL. While there is noquestion about the therapeutic strategy in a patientwith potentially resectable UES, conservative man-agement of MHL has in the past been advocated,since there is a potential for spontaneous regres-sion [36]. A lesson should perhaps be learned fromthe parallel scenario in which cystic pleuropulmo-nary blastoma (PPB) may radiologically and mac-roscopically resemble cystic adenomatoid malfor-mation (CCAM). Just as the PPB within a CCAMmay be missed because of less than extensive sam-pling, so too, apparently, might a UES withinMHL. If MHL truly has the potential to progress toUES and we are currently unable to identify inwhich particular case it will do so, complete surgi-

cal resection followed by extensive histologic sam-pling would appear to be warranted in all cases.

R E F E R E N C E S1. Willis RA. The Borderland of Embryology and Pathology,

2nd ed. London: Butterworth, 1962;435–437.2. Stanley RJ, Dehner LP, Hesker AE. Primary malignant

mesenchymal tumors (mesenchymoma) of the liver inchildhood. An angiographic pathologic study of threecases. Cancer 1973;32:973–984.

3. Stocker JT, Ishak KG. Undifferentiated (embryonal) sar-coma of the liver. Report of 31 cases. Cancer 1978;42:336–348.

4. Goodman ZD. Histologic diagnosis of hepatic tumors. AnnClin Lab Sci 1984;14:169–178.

5. Lack EE. Mesenchymal hamartoma of the liver. A clinicaland pathologic study of nine cases. Am J Pediatr HematolOncol 1986;8:91–98.

6. Cozzutto C, De Bernardi B, Comelli A, Soave F. Malignantmesenchymoma of the liver in children: a clinicopatho-logic and ultrastructural study. Hum Pathol 1981;12:481–485.

7. De Chadarevian J-P, Pawel BR, Faerber EN, WeintraubWH. Undifferentiated (embryonal) sarcoma arising in con-junction with mesenchymal hamartoma of the liver. ModPathol 1994;7:490–493.

8. Lauwers GY, Grant LD, Donnelly WH, et al. Hepatic un-differentiated (embryonal) sarcoma arising in a mesenchy-mal hamartoma. Am J Surg Pathol 1997;21:1248–1254.

9. Hsu, S-M, Raine L, Fanger H. Use of avidin-biotin-perox-idase complex (ABC) in immunoperoxidase techniques: acomparison between ABC and unlabeled antibody (PAP)procedures. J Histochem Cytochem 1981;29: 577–580.

10. Seabright M. A rapid banding technique for human chro-mosomes. Lancet 1971;2:971–972.

11. Steiner M, Bostrum B, Leonard AS, Dehner LP. Undiffer-entiated (embryonal) sarcoma of the liver. A clinicopath-ologic study of a survivor treated with combined techniquetherapy. Cancer 1989;64:1318–1322.

12. Miettinen M, Kahlos T. Undifferentiated (embryonal) sar-coma of the liver. Epithelial features as shown by immu-nohistochemical analysis and electron microscopic exam-ination. Cancer 1989;64:2096–2103.

13. Chou P, Mangkornkanok M, Gonzalez-Crussi F. Undiffer-entiated (embryonal) sarcoma of the liver: ultrastructure,immunohistochemistry, and DNA ploidy analysis of twocases. Pediatr Pathol 1990;10:549–562.

14. Stocker JT. An approach to handling pediatric liver tu-mors. Am J Clin Pathol 1998;109:S67–S72.

15. Urban CE, Mache CJ, Schwinger W, et al. Undifferentiated(embryonal) sarcoma of the liver in childhood. Successfulcombined-modality therapy in four patients. Cancer 1993;72:2511–2516.

16. Harris MB, Shen S, Weiner MA, et al. Treatment of pri-mary undifferentiated sarcoma of the liver with surgeryand chemotherapy. Cancer 1984;54:2859–2862.

17. Pieterse AS, Smith M, Smith LA, Smith P. Embryonal(undifferentiated) sarcoma of the liver. Fine-needle aspi-ration cytology and ultrastructural findings. Arch PatholLab Med 1985;109:677–680.

18. Keating S, Taylor GP. Undifferentiated (embryonal) sar-coma of the liver: ultrastructural and immunohistochem-ical similarities with malignant fibrous histiocytoma. HumPathol 1985;16:693–699.

19. Iliszko M, Czauderna P, Babinska M, Stoba C, Roszk-iewicz A, Limon J. Cytogenetic findings in an embryonal

488 M.J. O’SULLIVAN ET AL.

sarcoma of the liver. Cancer Genet Cytogenet 1998;102:142–144.

20. Parham DM, Kelly DR, Donnelly WH, Douglass EC. Im-munohistochemical and ultrastructural spectrum of he-patic sarcomas of childhood: evidence for a common his-togenesis. Mod Pathol 1991;4:648–653.

21. Joshi VV, Kaur P, Ryan B, Saad A, Walters TR. Mucoidanaplastic hepatoblastoma: a case report. Cancer 1984;54:2035–2039.

22. MacSween RNM, Anthony PP, Scheuer PJ, Burt AD, Port-mann BC. Pathology of the Liver. New York: ChurchillLivingston, 1994.

23. Desmet VJ. Cholangiopathies: past, present and future.Semin Liv Dis 1987;7:67–75.

24. Lennington WJ, Gray GF, Page DL. Mesenchymal hamar-toma of liver. A regional ischemic lesion of a sequesteredlobe. Am J Dis Child 1993;147:193–196.

25. Okeda R. Mesenchymal hamartoma of the liver—an au-topsy case with serial sections and some comments on itspathogenesis. Acta Pathol Jpn 1976;26:229–236.

26. von Schweinitz D, Dammeier BG, Gluer S. Mesenchymalhamartoma of the liver—new insights into histogenesis.J Pediatr Surg 1999;34:1269–1271.

27. Rauscher F. The WT1 Wilms tumor gene product: a devel-opmentally regulated transcription factor in the kidneythat functions as a tumor suppressor. FASEB J 1993;7:896–903.

28. Hastie ND. The genetics of Wilms tumor—a case of dis-rupted development. Annu Rev Genet 1994;28:523–558.

29. Pelletier J, Bruening W, Kashtan C, et al. Germline muta-tions in the Wilms tumor suppressor gene are associatedwith abnormal urogenital development in Denys-Drashsyndrome. Cell 1991;67:437–447.

30. Otal TM, Hendricks JB, Pharis P, Donnelly WH. Mesen-chymal hamartoma of the liver. DNA flow cytometric anal-ysis of eight cases. Cancer 1994;74: 1237–1242.

31. Mascarello JT, Krous HF. Second report of a translocationinvolving 19q13.4 in a mesenchymal hamartoma of theliver. Cancer Genet Cytogenet 1992;58:141–142.

32. Speleman F, De Telder V, De Potter KR, et al. Cytogeneticanalysis of a mesenchymal hamartoma of the liver. CancerGenet Cytogenet 1989;40:29–32.

33. Bove KE, Blough RI, Soukup S. Third report oft(19q)(13.4) in mesenchymal hamartoma of liver withcomments on link to embryonal sarcoma. Pediatr DevPathol 1998;1:438–442.

34. Heim S, Mitelman F. Cancer Cytogenetics, New York; AlanR. Liss, 1987.

35. Keeling JW. Liver tumors in infancy and childhood.J Pathol 1971;103:69–85.

36. Barnhart DC, Hirschl RB, Garver KA, Geiger JD, HarmonCM, Coran AG. Conservative management of mesenchy-mal hamartoma of the liver. J Pediatr Surg 1997;32:1495–1498.

UNDIFFERENTIATED EMBRYONAL SARCOMA 489