Introduction and Epidemiology

Liver tumors in children account for approximately 1.1% of all malignancies in children younger than 20 years of age according to the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) reports. The age-adjusted incidence rate of liver tumors is approximately 11 per million in infants younger than the age of 1 year and approximately 6 per million in children between 1 and 4 years of age. This translates to approximately 100 to 150 new cases of liver cancer annually in children in the United States.

Liver malignancies in children can be divided into three broad classifications: hepatoblastoma, which accounts for two thirds of malignant liver cancers and the predominant malignant tumor type in young children; hepatocellular carcinoma (HCC), the predominant tumor type in adolescents; and other rare hepatic malignancies of childhood. A spectrum of benign tumor histologies broadens the differential diagnosis of the infant or child with a liver mass.

In an early meta-analysis of eleven separate series totaling 1256 primary liver tumors in children reported by Weinberg and Finegold, 43% were hepatoblastoma, 23% HCC, 13% benign vascular tumors, 6% mesenchymal hamartomas, 6% sarcomas, 2% adenomas, 2% focal nodular hyperplasia, and 5% other tumors. Rare tumors of the liver also include rhabdoid tumors and hepatic germ cell tumors. An international working group recently provided guidelines regarding the histologic classification of pediatric liver tumors.

Most liver tumors in childhood are more common in boys than girls. The male-to-female ratio observed for hepatoblastoma in cooperative group trials has ranged from 1.5 to 1 to 2 to 1; a combined analysis of multiple studies suggests that the male-to-female ratio in hepatoblastoma is 1.65, consistent with SEER data.

Hepatoblastomas have a unique age distribution ( Fig. 58-1 ). Two peak ages occur, one at birth or within the first month of life and a second broad peak between 16 and 20 months of age. Ninety percent of hepatoblastomas occur in children younger than 4 years of age, and 95% of hepatoblastomas occur by age 5. Although rare, hepatoblastoma has been seen in adults, as described in multiple sporadic case reports. A systematic review of hepatoblastoma in adults suggests that these tumors are associated with a particularly poor prognosis, although the time frame over which these cases were gathered spanned multiple decades. In contrast hepatoblastomas diagnosed very early in life (i.e., in the neonatal period) are associated with a favorable outcome.

Age distribution of hepatoblastoma derived from 256 sequential cases representing all stages and histologies of hepatoblastoma. An initial peak is seen at birth, reflecting tumors developing during gestation. The median age of all children with hepatoblastoma is 18 months, and the oldest age is 78 months. In 99% of cases hepatoblastoma was apparent by 5 years of age.

Hepatoblastoma in children older than 5 years of age is often more aggressive than the typical hepatoblastoma, has characteristics of HCC, and has been termed transitional liver cell tumor. More recently pathologists prefer to designate this poorly understood entity as hepatocellular tumor, not otherwise specified, pending a much-needed and more thorough understanding of the biology of these tumors with features of both hepatoblastoma and HCC. HCCs make up the bulk of liver tumors seen after age 10 years and throughout adolescence and adulthood. As with hepatoblastoma, HCC is more common in male subjects than female subjects. The fibrolamellar variant of HCC is a notable exception, with equal occurrence in male and female subjects.

The incidence of liver tumors has increased over recent decades, perhaps more than any other type of childhood tumor. During a single decade the incidence rates of hepatic tumors in infants younger than 1 year of age increased from 4 to 8 per million. The increase in rates was especially pronounced in female infants, in whom rates of hepatic cancer increased from 2 to 12 per million in a single decade. Ross and Gurney estimated that the average annual percentage change in incidence of hepatoblastoma between 1973 and 1992 was 5.2% for males and 8.2% for females. The rise in incidence of hepatoblastoma has been estimated to be 4% per year between 1992 and 2004 and continues to rise in recent years compared with even recent SEER data. The largest contributing factor to the increased incidence is thought to be the increased survival rates of premature infants who are at markedly increased risk of hepatoblastoma, as discussed in subsequent sections of this chapter.

Origins of Hepatoblastoma

Hepatoblastoma has long been thought to have prenatal origins, and increasing evidence points to prematurity, prenatal exposures, and overgrowth in early infancy as factors contributing to hepatoblastoma. It has long been presumed that embryonal tumors derive from primitive cells and that hepatoblastoma, as an embryonal tumor of the liver in which cells morphologically resemble cells in the developing embryonal and fetal liver, derives from a primitive hepatic cell. In rodents immature oval cells that proliferate after exposure to hepatocarcinogens express both biliary and hepatocytic markers. These cells are characterized by staining with OV-1 and OV-6 markers. Ruck and colleagues demonstrated the existence of a similar cell type in hepatoblastoma on the basis of marking with OV-1 and OV-6 in a small population of cells within the tumor. These OV-1 and OV-6 staining cells account for fewer than 1% of the cells within the tumor and provided evidence to support the existence of a population of stem cells in hepatoblastoma. Fiegel and coworkers analyzed hepatoblastomas for stem cell markers including CD34, Thy1, and c-kit and demonstrated the co-occurrence of these stem cell markers with hepatobiliary markers in ductal cells phenotypically resembling stem cells, which suggests that stemlike cells play a role in the histogenesis of hepatoblastoma. Other markers of hepatic progenitor cells including epithelial cell adhesion molecule (EpCAM), cytokeratin 19 (CK19), octamer-binding transcription factor 3/4 (Oct-3/4), and delta-like 1 homologue (DLK1) have been shown to be expressed in hepatoblastoma tumor cells. NOTCH2, which peaks in the second half of fetal development and then declines toward birth, is expressed in most hepatoblastomas. CITED1, expressed in embryonic mouse liver and a marker of hepatic progenitor cells, is shown to be present in hepatoblastoma, providing another similarity of hepatoblastoma to hepatic precursor stem cells.

Genetic Predisposition and Liver Tumorigenesis

Most cases of childhood liver malignancy are not associated with an apparent major genetic syndrome. However numerous syndromes have been observed in association with hepatoblastoma or HCC, as well as some benign hepatic tumors. A complete summary of genetic syndromes and liver tumors in children is provided in Table 58-1 . The major genetic syndromes associated with childhood liver tumors are discussed in the subsequent sections.

Familial Adenomatous Polyposis

Familial adenomatous polyposis (FAP) is caused by germline mutation of the APC gene, FAP is best characterized by the presence of colonic polyps, which without intervention are associated with the virtual certainty that adenocarcinoma of the colon will develop. Germline mutation of the APC gene is, however, also associated with a marked increased risk of hepatoblastoma. The risk of hepatoblastoma in children in FAP kindred is estimated to be 800-fold. The risk that a parent with an APC mutation will have a child with hepatoblastoma is 0.4%. Because FAP is an autosomal dominant disorder, genetic testing should demonstrate a predisposing mutation in 50% of offspring, such that the risk of hepatoblastoma in an infant who is a documented mutation carrier would be slightly less than 1%. Within a series of consecutive hepatoblastoma patients for whom family histories were obtained, approximately 8% of sequential cases of hepatoblastoma have a family history suggestive of FAP, and abnormalities of the APC gene were detectable in all cases.

Also reported in children from FAP kindreds, albeit less commonly, are other liver tumors, including HCCs. A benign precursor lesion, hepatocellular adenoma has been reported in a patient with a germline APC mutation and colonic polyps. Within resected liver tissue from children with germline APC mutations and hepatoblastoma, multiple hepatic nodules were consistent with precursor lesions with potential to progress to either hepatoblastoma or HCC. Presumably the timing of additional mutations or key developmental processes could determine the resulting histologic tumor type in the predisposed liver.

Significantly of nine reported cases of hepatoblastoma in siblings, at least four were in families with FAP; Of the remaining reported sibling pairs, one was affected with type 1a glycogen storage disease, and two sibling pairs had no apparent familial syndrome, and one recent report of a sibling pair with hepatoblastoma demonstrated an APC mutation in only one of the siblings. These reports suggest that whereas APC mutation accounts for a significant percentage of familial cases of hepatoblastoma, additional predisposition genetic changes may exist either as a distinct underlying cause of hepatoblastoma or as a modifying factor in infants with an existing APC mutation.

The percentage of sporadic hepatoblastoma patients (those without an obvious family history) who carry a germline mutation of the APC gene is not well defined. However it has been suggested both that children with hepatoblastoma be screened for germline APC mutations and that asymptomatic children in families with known polyposis be screened for hepatoblastoma. Within the APC gene genotype-phenotype correlations suggest regions of the gene that predict both the severity of polyposis as well as the presence of extracolonic manifestations. The literature suggests that germline mutations in FAP families with children who are affected by hepatoblastoma can occur throughout the gene ( Fig. 58-2 ) and can also include whole gene deletions or chromosome rearrangements such that the site or type of the individual APC mutation cannot be used to predict the occurrence of hepatoblastoma. This finding implies a risk to all APC mutation carriers.

Frequency histogram of mutations in the adenomatous polyposis coli (APC) gene. Left axis and black lines represent all known APC germline mutations. Right axis and orange lines represent APC germline mutations in children with hepatoblastoma. Germline mutations associated with hepatoblastoma are observed throughout the gene.

(From Hirschman B, Pollock B, Tomlinson G: The spectrum of APC mutations in children with hepatoblastoma from familial adenomatous polyposis kindreds. J Pediatr 147:263–266, 2005.)

Prematurity as a Risk Factor

One of the most intriguing aspects of hepatoblastoma is its association with premature or low-birth-weight infants. This association was initially reported by Japanese investigators who analyzed patient diagnoses and birth weights in the Japan Children’s Cancer Registry over the 9-year period from 1985 to 1993. It was found that 3.9% of patients with hepatoblastoma were of very low birth weight. Moreover it was observed that when data were subdivided over 4-year periods, the percentage of hepatoblastoma cases of low birth weight rose from 0.7% in 1985 to 1989 to 8.6% in 1990 to 1993 and demonstrated a linear increasing trend. The risk of hepatoblastoma has also been shown to be inversely proportional to birth weight. The relative risk for an infant between 2000 and 2999 grams is 1.21, whereas the relative risk for an infant of less than 1000 grams is 15.64.

Data from the Children’s Cancer Group survey confirmed the preponderance of former premature infants in a series of 76 patients with hepatoblastoma. A second confirmation in the United States comes from the California population-based cancer registry, which also reported an elevated risk of hepatoblastoma in children of very low birth weight. This study also pointed out that the age of diagnosis of hepatoblastoma in children of low birth weight was actually older than that of children of normal birth weight, which could partly be explained by gestational age.

It has been speculated that the increase in the percentage of low-birth-weight infants among children with hepatoblastoma may be partially caused by the increased survival rates of low-birth-weight infants. It is also believed that the increase in survival rates of premature infants at risk for hepatoblastoma may contribute to the overall increase in childhood liver tumors.

Indeed the rise in the rate of hepatoblastoma has been found to be consistent with the risk in births of low-birth-weight infants. However whether the pathways by which low-birth-weight infants develop hepatoblastoma involve key exposures in the newborn period, increased sensitivity of the liver in premature infants to potential carcinogens, disruption of the normal process of liver development by premature birth, or a combination of these factors has yet to be defined. One small single-institution study, in which treatment records of five infants who developed hepatoblastoma were compared with those of infants of similar birth weights who did not develop hepatoblastoma, suggests that perinatal intensive long-term treatments contribute to the development of hepatoblastoma. A somewhat larger study comparing 12 hepatoblastoma cases and 75 birth weight–matched controls suggests that the duration of oxygen therapy, use of furosemide, and length of time taken to regain body weight at birth were factors in predicting the likelihood that hepatoblastoma would develop in premature infants. A study from the United Kingdom reported that polyhydramnios and either eclampsia or preeclampsia are more common in children with hepatoblastoma compared with those without hepatoblastoma, and that the eclampsia or preeclampsia observed was also associated with low birth weight.

A recent case-control study of 600 COG patients with hepatoblastoma (AEPI04C1) has examined exposures in low-birth-weight and other infants with hepatoblastoma and has not found an association with maternal illness or medication use during pregnancy or infertility treatment and hepatoblastoma. Previously use of assisted reproductive technology was postulated to be a contributing factor, although it is thought that it may contribute to hepatoblastoma development only through the associated risk of BWS and assisted reproductive technology (ART), but not as an independent risk factor.

Environmental Exposures and Hepatoblastoma

An early epidemiologic case-control study of risk factors for hepatoblastoma by Buckley and colleagues, based on parental interviews, revealed an association of hepatoblastoma with maternal occupational exposure to metals, such as those used in welding or soldering (odds ratio [OR] = 8); petroleum products such as lubricating oils or greases (OR = 3.7); and paints or pigments (OR = 3.7). There was also a significant association with paternal occupational exposure to metals (OR = 3, P = .01) and a marginally significant association with paternal occupational exposure to petroleum products (OR = 1.9). Although the aforementioned study did not find an association of parental smoking with hepatoblastoma, this has been reported in several subsequent studies. The risk of hepatoblastoma in children of parents who smoked was approximately double in all three studies. Indeed in 2009 the International Agency of Research on Cancer declared that parental smoking is a carcinogen in the developing liver. Subsequent findings have been variable, however. A recent report from COG, the largest etiologic study to date, reported no association between smoking and hepatoblastoma. Variability among studies, however, could be based on small numbers and differences in reporting and overall smoking prevalence in the different populations. Although alcohol clearly plays a role in cirrhosis of the liver and HCC in adults, the COG study reported no association with parental alcohol use and hepatoblastoma in offspring.

Viral Hepatitis and Liver Tumors in Children

Viral hepatitis infection has historically accounted for a large percentage of HCC cases. Before the introduction of the hepatitis B vaccine in Taiwan, HCC accounted for 80% of cases of liver tumors in children. With the introduction of the hepatitis B vaccination in 1984, rates of liver tumors in Taiwanese children older than 6 years dropped significantly, from 0.70 per 100,000 from 1981 to 1986 to 0.36 per 100,000 between 1990 and 1994. The decrease in rates of HCC after introduction of the vaccine was more pronounced in boys than in girls. In fact the incidence of HCC decreased in boys but not in girls. The male-to-female incidence of HCC in Taiwan steadily decreased from 4.5 per 100,000 in 1981 to 1986 to 1.9 per 100,000 in 1990 to 1996, 6 to 12 years after the introduction of the vaccine. The reasons for these sex-related differences in responses to the vaccine are not known. It has been demonstrated that among children with hepatitis B the presence of a deletion of the pre-S region of the hepatitis B virus was an additional risk factor for the development of HCC (OR = 36.69).

Presentation and Evaluation of the Child with a Liver Mass

Aside from patients with known genetic syndromes (discussed in the preceding sections), most cases of hepatoblastoma occur in otherwise well children who present with an abdominal mass, often with some degree of abdominal discomfort.

Isosexual precocious puberty has been observed in some boys as a presenting finding in hepatoblastoma and occurs secondary to tumor production of chorionic gonadotropin. In one study of 48 children, the incidence of precocious puberty in children with liver tumors was 6%. To the authors’ knowledge, there are no known cases of precocious puberty associated with hepatoblastoma in girls.

Laboratory evaluation of the child with a liver mass should include a complete blood count and liver function tests, as well as AFP and β-human chorionic gonadotrophin tumor marker levels. The platelet count is often high in hepatoblastoma but is not diagnostic of hepatoblastoma among children with liver tumors. This thrombophilia is associated with high serum levels of thrombopoietin (TPO), also known as c-Mpl ligand. Normally synthesized in the liver, TPO has also been shown to be expressed in hepatoblastoma tumor tissues and is present at higher than normal levels in the serum of patients with hepatoblastoma. In a large series of patients of various ages (13 to 84 years) who had hepatic tumors, thrombocytosis was noted in 2.7%. The high platelet count was correlated with higher serum TPO levels. Other components of the complete blood count are usually normal insofar as bone marrow involvement with liver tumors has not been reported.

AFP is the primary serum marker that is used diagnostically and prognostically, as well as in surveillance. A sample should be obtained in the initial evaluation of any child with a liver mass. AFP levels are markedly elevated in more than 90% of hepatoblastomas and more than 50% of HCCs. Care should be taken in interpreting AFP levels in the young infant because levels are normally very high at birth and decline to less than 10 ng/dL over the first year of life. Of note is that some liver malignancies—notably the small cell undifferentiated subtype of hepatoblastoma as well as the fibrolamellar variant of HCC—are not associated with elevation of the AFP. Likewise, sarcomas and rhabdoid tumors are not associated with elevated AFP. The International Society of Oncology Epithelial Liver Group (SIOPEL) has demonstrated that hepatoblastomas with low AFP levels at diagnosis (less than 100 ng/mL) tend to present at a higher stage and are associated with a poor outcome. A low AFP level has also been noted to be a marker of poor outcome across multiple studies compiled and analyzed by the Children’s Hepatic Tumor International Consortium (CHIC) group. The β-human chorionic gonadotrophin level is only occasionally elevated in hepatoblastoma, usually in cases that present with precocious puberty. The β-human chorionic gonadotrophin level is also elevated in the rare choriocarcinoma of the liver, as well as teratomas.

Evaluation of a child with a liver tumor should include a thorough family history to document specific cancers in relatives, with notation made of any early-onset colon cancers or colectomies, in addition to the presence of thyroid cancers or medulloblastoma, which can be part of FAP syndrome. Testing for germline mutation of the APC gene may be indicated; if performed, it should include appropriate genetic counseling. These patients with hepatoblastoma and germline APC mutations will require long-term surveillance for colonic polyps and tumor development.

In addition the medical history should note the birth weight, history of prematurity, and presence or absence of stigmata of neonatal overgrowth, including omphalocele and hemihypertrophy, which would suggest BWS and a risk for other embryonal tumors.

Staging

Although North American trials have traditionally used a postsurgical diagnostic staging group, applied after the initial surgical procedure (biopsy or resection), the European system, based on the appearance by imaging before chemotherapy and surgical resection (PRETEXT), is being adopted worldwide. PRETEXT stages I through IV reflect the number of sectors that are involved with tumor. Other components of the staging system reflect extension into the vena cava , extension into the portal vein, extrahepatic abdominal disease, or distant metastases. The SIOPEL staging is shown schematically in Figure 58-3 . A comparison of the prior North American staging and the current SIOPEL staging is shown in Table 58-2 .

Pretreatment imaging-defined staging pro­tocol presurgical staging system used by the Inter­national Society of Pediatric Oncology to predict outcome. The liver image is divided into four sectors, and tumor stage is defined as the number of free sectors. In this system, stage is further defined by extension of tumor beyond the liver as V indicating extension into the vena cava or all three hepatic veins (or both), P indicating extension into the main or both left and right branches of the portal vein, E indicating extrahepatic disease proved by biopsy, and M indicating the presence of distant metastases.

Figure 58-4 demonstrates a small hepatoblastoma tumor image that was initially diagnosed by surveillance ultrasound in a patient with BWS. This tumor involved two sectors and as such is classified as PRETEXT II, but because it was completely surgically excised after diagnosis, it would be stage 1 using the North American staging system. Figure 58-5 demonstrates a multifocal hepatoblastoma that involves all four sectors and had metastatic lesions in the lung; therefore it is PRETEXT IV by the North American staging system. If the patient had not had metastatic lesions, the tumor would be stage 3 according to the North American system but still PRETEXT IV.

Computerized tomography scan of a low-stage hepatoblastoma detected by periodic surveillance with ultrasound in a patient at high risk of hepatoblastoma because of Beckwith-Wiedemann syndrome. The tumor was fully resected and classified as stage 1.

(Courtesy of Children’s Medical Center, Dallas, Tex.)

Computerized tomography of a high-stage hepatoblastoma. The tumor occupies most of the left lobe and a portion of the right lobe. Preoperative chemotherapy rendered this tumor resectable. This tumor was also accompanied by pulmonary metastases (not shown) and therefore was classified as stage 4 by North American staging system.

(Courtesy of Children’s Hospital, Boston, Mass.)

Both staging systems have been shown to have prognostic significance, but it is difficult to compare studies using different staging systems. In the future the PRETEXT system will be used by most groups internationally.

Pathology of Hepatoblastoma and Hepatocellular Carcinoma

Grossly hepatoblastomas occur as single expansile masses, usually in the right lobe of the liver. A series of international discussions originating at the Los Angeles COG International Pathology Liver Tumors Symposium in 2011 resulted in a consensus classification of liver tumors in children, summarized in Table 58-3 . A description of immunohistochemical properties currently recommended to aid in the classification of pediatric liver tumors is provided in Table 58-4 .

A representative resected hepatoblastoma tumor is shown in Figure 58-6 . Hepatoblastomas are composed primarily of epithelial cells that resemble different stages of the developing liver. Most hepatoblastomas are a mixture of cells resembling fetal cells and cells resembling embryonal cells. A typical hepatoblastoma of fetal and embryonal histology is shown in Figure 58-7, A . Because hepatoblastomas derive from immature cells that may retain the potential to differentiate, tumors often have foci of other types of more differentiated cells, including hematopoietic cells, as demonstrated in Figure 58-7, B . A subset of hepatoblastomas are characterized by cholangiocyte-like cells, which suggest the differentiation of the neoplastic cells along the cholangiocyte lineage as opposed to the hepatocyte lineage. When the epithelial cells are aligned in thick cordlike structures, the tumor is described as having macrotrabecular features. In approximately 5% of hepatoblastomas, only fetal-type cells are seen histologically, as shown in Figure 58-7, C . These fetal-only tumors are now further classified according to the degree of mitoses.

Gross pathology of resected hepatoblastoma tumor specimen. Tumor appears as a large expansile mass. The tumor has a whitish appearance compared with the darker adjacent liver in background. The cut surface is shown at the bottom of the image and has a variegated appearance.

(Courtesy of Victor Saldivar, MD, Children’s Hospital of San Antonio, Tex.)

A, Mixed fetal and embryonal epithelial hepatoblastoma. This hepatoblastoma shows two different morphologic patterns. The fetal pattern, seen in the lower part of the image, is composed of uniform, small cuboidal cells with distinct cell membranes, whereas the embryonal pattern, seen in upper part of the image, consists of smaller angulated cells with hyperchromatic nuclei, higher nuclear-to-cytoplasmic ratios, and indistinct cell membranes. This is the most common histologic pattern of hepatoblastoma. B, Hepatoblastoma with extramedullary hematopoesis. As is often observed in hepatoblastoma, islands of hematopoietic cells are observed as clusters of small dark cells with prominent nuclei within surrounding tumor tissue. C, Pure fetal epithelial hepatoblastoma. This is composed of relatively uniform, small cuboidal cells. The cells have a small round nucleus with inconspicuous nucleolus, abundant cytoplasm that may be clear or finely granular, and distinct cell membranes. The pure fetal variant accounts for approximately 5% of all hepatoblastomas and is associated with a favorable outcome. D, Small cell undifferentiated hepatoblastoma. The central portion shows cells resembling fetal hepatoblastoma. On the left, cells appear undifferentiated with round nuclei and minimal cytoplasm and fail to form tubules. At right, cells are more typical of rhabdoid-like histology.

( A , Courtesy of Dinesh Rakheja, MD, Children’s Medical Center, Dallas, Tex.; B , Courtesy of Victor Saldivar, MD, Children’s Hospital of San Antonio, Tex.; C , Courtesy of Dinesh Rakheja, MD, Children’s Medical Center, Dallas, Tex.; D , Courtesy of Milton Finegold, MD, Texas Children’s Hospital, Houston, Tex.)

Two histologic subtypes of hepatoblastoma have significant implications for predicting clinical outcome. Approximately 7% of hepatoblastomas are composed of pure, well-differentiated fetal cells. The well-differentiated histology, previously referred to as pure fetal histology, has been associated with an exceptionally favorable outcome. A recent COG study by Malogolowkin and coworkers demonstrated that if the hepatoblastoma is completely resected surgically at the time of diagnosis, further therapy may be unnecessary. The presence of this subtype and the desire to avoid the toxicity of chemotherapy remain compelling reasons to pursue upfront resection whenever possible.

The small cell undifferentiated variant accounts for approximately 5% of hepatoblastoma and also carries clinical significance. Histologically the cells are small and pleomorphic and show high mitotic rates, as shown in Figure 58-7, D . There are several reports in small cell undifferentiated hepatoblastoma of unique chromosome translocations involving chromosome 22q12, the region of SMARCB/INI1, the gene most often associated with rhabdoid tumors and atypical teratoid tumors. Emerging evidence suggests that from histologic, cytogenetic, molecular genetic, and clinical perspectives these tumors resemble rhabdoid tumors to varying degrees. Immunohistochemistry demonstrated loss of the INI1 protein in the small cell undifferentiated variant, and it is now recommended that INI1 staining be performed as part of the histologic characterization (see Table 58-4 ). Mounting evidence suggests that the small cell variant is associated with a poor outcome.

HCCs consist of cells that are well-differentiated and resemble normal hepatocytes. A typical HCC is shown in Figure 58-8 . The fibrolamellar variant of HCC is characterized by larger, polygonal cells mixed with fibrous stroma and often eosinophilic cytoplasm, occasionally with cystic degeneration ( Fig. 58-9 ). In contrast to hepatoblastoma, the adjacent liver tissue in patients with HCC is often marked by cirrhosis. The fibrolamellar variant is an exception in which surrounding liver tissue is usually noncirrhotic.

Hepatocellular carcinoma. This well-differentiated hepatocellular carcinoma is composed of two- to four-layer thick cords or trabeculae of cells resembling normal hepatocytes. Sinusoidal-like spaces lined by endothelial cells are present between the trabeculae, further mimicking the architecture of normal liver. No portal tracts are present.

(Courtesy of Dinesh Rakeja, MD, Children’s Medical Center, Dallas, Tex.)

Fibrolamellar hepatocellular carcinoma. Fibrolamellar hepatocellular carcinoma is composed of sheets of polygonal tumor cells that are larger than normal hepatocytes and have a deeply eosinophilic cytoplasm, a large nucleus with prominent eosinophilic nucleolus. As shown prominently in A, the cells are embedded in collagenous stroma, and variably thick fibrous septa course through the tumor. Often there are focal infiltrates of mononuclear inflammatory cells. B, Foci of cystic degeneration (pseudogland formation) are apparent.

(Courtesy of Dinesh Rakeja, MD, Children’s Medical Center, Dallas, Tex.)

Acquired Genetic Changes in Hepatoblastoma

Karyotypic changes in hepatoblastomas can be classified into two broad categories: numeric changes and structural changes of individual chromosomes, the most common of which involve translocations of chromosome 1. The initial recurring translocation described was the translocation involving chromosomes 1 and 4, t(1;4)(q12;q34). It has since been reported in a large series that hepatoblastoma is characterized by a family of chromosome translocations with similar breakpoints on either chromosome 1q12 or 1q21. Each such translocation observed is unbalanced, resulting in a gain of the long arm of chromosome 1, and is also frequently associated with numerous whole chromosomal gains.

Most of the numeric aberrations result in addition of whole chromosomes, but occasionally they result in loss of chromosomes. The specific numeric changes are nonrandom. Trisomies of chromosome 2, 8, and 20 are the most common recurring numeric aberrations. The most commonly observed chromosomal losses are chromosomes 4 and 18. Although these whole chromosome changes have been described previously by classical karyotype analysis, they can perhaps be better visualized by whole genome comparative genomic hybridization, as shown in Figure 58-10 . The clinical significance of these chromosomal abnormalities is not yet known..

Genome-wide copy number comparative genomic hybridization (CGH) scan of hepatoblastoma tumor compared with standard normal male control DNA. Hepatoblastoma is characterized by whole chromosome gains and losses as well as unbalanced translocations of chromosome 1 resulting in duplication of the long arm of chromosome 1q. This CGH plot demonstrates increased copy number of chromosome 8, 17, 19, and 20 as well as whole chromosome losses of chromosomes 4 7, 9, 10, 14, and 18. In addition a duplication of chromosome 1q and loss of chromosome 1p is observed. Numeric aberrations of X and Y reflect the female sex of the child rather than tumor-specific changes.

HCC is also characterized also by chromosomal gains and losses, with loss of the Y chromosome a notable finding. Duplication of chromosome 1q is also seen in HCC with increased expression of numerous chromosome 1 genes. Cytogenetic data on the fibrolamellar variant of HCC is sparse, but one childhood fibrolamellar carcinoma has been characterized by a hypertriploid karyotype with clonal evolution and multiple additional chromosomal gains as well as loss of the Y chromosome.