Hepatic tumors in childhood

Incidence

Hepatoblastomas are the most common primary hepatic tumors of childhood, constituting 43% to 64% of all hepatic neoplasms in one large series. Hepatoblastoma accounts for 91% to 96% of primary hepatic tumors in children younger than 5 years ^, but comprises less than 1% of hepatic malignancies when adult age groups are included. The Liver Cancer Study Group of Japan identified 30 hepatoblastomas (0.6%) in a cohort of 4658 patients of all ages diagnosed during a 2-year period.

Each year, hepatoblastoma affects 1 to 2.4 of every 1 million children younger than 15 years, ^, and approximately 50 to 70 new cases per year are reported in the United States, with a male/female ratio of 1.7:1. Although hepatoblastoma has been reported sporadically in adults, the median age at diagnosis is approximately 18 months, and most cases occur before the age of 3 years. Hepatoblastoma is the most prevalent malignant neoplasm of the fetus and neonate and results in death within 2 years if not treated. The incidence of hepatoblastoma from 2007 to 2011 was 4.2 per 1 million children younger than 20 years, and it may be increasing. ^, The incidence in the same age group from 1993 to 1997 was 1.2 per 1 million, and this was increased from 0.6 per 1 million from 1973 to 1977. However, this increase may simply be due to improving diagnostic modalities.

Hepatoblastoma may occur in siblings. It is most strongly associated with familial polyposis, ^, Gardner syndrome, and Beckwith-Wiedemann syndrome. ^, In familial polyposis, the incidence of hepatoblastoma seems to be increased in first-degree relatives of patients with polyposis. Beckwith-Wiedemann syndrome is associated with Wilms tumor, rhabdomyosarcoma, adrenocortical carcinoma, and hepatoblastoma with a possible association between hepatoblastoma and trisomies 2, 8, 18, and 20. ^,

Hepatoblastoma is also associated with low birth weight. ^, It is unknown whether the causative agent is developmental abnormalities associated with prematurity or interventions, such as early total parenteral nutrition. These tumors also are reported in patients with congenital anomalies, such as cleft palate, and cardiovascular and renal anomalies, including polycystic kidney and absence of the right adrenal gland. There are also at least two reports of hepatoblastoma occurring in patients with biliary atresia. There are sporadic findings of hepatoblastoma in patients with hepatitis B, but no correlation has been found between the two diseases. ^, To date, no evidence associates hepatoblastoma with hepatitis B or C infection or any other chronic viral hepatitis. These patients usually do not have cirrhosis or inborn errors of metabolism. ^,

A great advance in developing data-driven models of risk for hepatoblastoma was the institution of the Children’s Hepatic Tumors International Collaboration (CHIC). ^,

Pathology

Hepatoblastomas are large tumors that can contain fibrous bands, producing a spoked-wheel appearance (see Chapter 87 ). The five histologic subtypes observed in hepatoblastoma are (1) fetal, (2) embryonal, (3) mixed epithelial, (4) mesenchymal/macrotubular, and (5) anaplastic or small-cell undifferentiated. These subtypes are differentiated based on the findings with light microscopy, but all tumor cells appear smaller than nonneoplastic hepatocytes. Extramedullary hematopoiesis is notably present and may be related to constitutive cytokine production by the tumor cells. The fetal type grows in trabeculae and resembles fetal hepatic cells, whereas embryonic hepatoblastoma cells grow in noncohesive sheets and resemble embryonic cells. Some hepatoblastomas contain mesenchymal tissue along with the epithelial component. Calcification also may appear in these tumors, and one patient was reported with osteosarcomatous elements in the hepatoblastoma and associated pulmonary metastases. The anaplastic or small-cell undifferentiated type consists of small, round blue cells reminiscent of neuroblastoma. This subtype is rare but particularly aggressive, with a strong metastatic potential. The importance of subtyping in hepatoblastoma is due to the association between prognostic risk and subtype, illustrated in Fig. 93.2 . ^, Some studies have indicated that the fetal histologic subtype has a better prognosis; in contrast, patients with the rare small-cell undifferentiated variant usually do poorly. ^,

Basic biology (see Chapter 9B )

Few cellular models of hepatoblastoma exist, and immortalized cell lines have been difficult to establish. One cell line, isolated from a human hepatoblastoma in 1995, clearly expresses the c-MYC and HRAS1 oncogenes and epidermal growth factor receptor (EGFR). Antibodies that blocked the EGFR inhibited cell growth. Although there has not been a connection established between HRAS and hepatoblastoma, knockdown of c-MYC has led to growth inhibition in hepatoblastoma-derived cell lines HepG2 and Huh6. Another new cell line was established in 2009 from a 5-year-old hepatoblastoma patient and contains an identical genotype to tumor cells, with morphologic, molecular, and immunohistochemical confirmation. Three-dimensional cellular organoid models have also shown promise for maintaining hepatoblastoma lines in culture.

Perhaps the most exciting insight is the association between hepatoblastoma and familial adenomatous polyposis syndrome. ^, In one study of 13 hepatoblastomas obtained from nonfamilial adenomatous polyposis patients, 69% had mutations in the adenomatous polyposis coli (APC) gene. In one case of siblings with hepatoblastoma, a shared APC gene mutation was identified. To date, over 100 patients have been described with both hepatoblastoma and familial adenomatous polyposis, with compelling implications regarding screening for both diseases. In addition, the association between hepatoblastoma and β-catenin, an APC -regulated protein and transcription cofactor for many proliferation genes, is well studied. In an analysis of 52 hepatoblastoma samples, 48% showed mutations in a region known to regulate activation. Subsequent studies also noted mutations, as well as increased expression of β-catenin in as many as 88% of samples, where it frequently localized to the nucleus. When hepatoblastoma samples that contained β-catenin in the nucleus were compared with those that did not, nuclear β-catenin was associated with more aggressive histologic subtypes. ^, Moreover, when compounds known to inhibit β-catenin activity were added to hepatoblastoma cell lines in vitro, a reduction in nuclear localization and a dose-dependent inhibition of cell growth were observed. The proliferation-inducing transcriptional coactivator Yes-associated protein (YAP) has also been observed to localize to the nucleus in hepatoblastoma. One study involving 94 tumor samples demonstrated YAP nuclear localization in 85% of cases. Coimmunoprecipitation was then performed between YAP and β-catenin in HepG2 cells, which exhibited an association between the two. This association has been investigated in murine models, where constitutively active YAP and β-catenin led to the development of hepatoblastoma. Similar results were demonstrated when β-catenin was constitutively active with the YAP paralog, transcriptional cofactor with PDZ-binding motif (TAZ). There is evidence that mTOR is a downstream effector of YAP/TAZ/β-catenin activation in hepatoblastoma cells and that mTOR inhibitors have therapeutic potential. ^,

Elevated hepatocyte growth factor levels have been demonstrated in the serum of 10 (43%) of 23 patients with hepatoblastoma. Addition of hepatocyte growth factor to hepatoblastoma-derived cell lines has been shown to demonstrate both antiapoptotic and antiproliferative properties, highlighting the need for further study. ^,

Small epithelial cells, characteristic of hepatic stem cells, have been observed in human hepatoblastomas of various subtypes. Additionally, various genetic abnormalities have been reported in hepatoblastoma. Chromosome 8q amplification is associated with a worsened prognosis and has been correlated with overexpression of the transcription factor pleomorphic adenoma gene 1 (PLAG1) . Telomerase and its regulatory protein expression levels have been correlated with poor outcome in human hepatoblastoma, and tamoxifen may inhibit hepatoblastoma cells by reducing telomerase levels.

Loss of heterozygosity on chromosome 11p15.5, the region associated with Beckwith-Wiedemann syndrome, and on chromosome 1p36 has been described in hepatoblastoma. ^, Investigations into both regions suggest that each may contain a tumor suppressor gene, but this has not been proven. Trisomy 20 and trisomy of all or part of chromosome 2 also have been reported. In addition, an abnormality of chromosome 2q may provide a common genetic link between hepatoblastoma and rhabdomyosarcoma. Finally, frequent genetic losses found using comparative genomic hybridization included regions 13q21-q22 (28%) and 9p22-pter (22%), and the most frequent genetic gains were on chromosomes 2q23-q23 (33%) and 1q24-q25 (28%). Differentially expressed microRNA has been shown to be deregulated in hepatoblastoma.

In addition, the well-known thrombocytosis associated with untreated hepatoblastoma is fascinating, as is the presence of extramedullary hematopoiesis. It has been shown that hepatoblastoma cells secrete interleukin (IL)-1β, which causes secretion of IL-6 from surrounding fibroblasts and endothelial cells. Other factors, such as erythropoietin and stem cell factor, have been localized to the cytoplasm of hepatoblastoma cells. Thrombopoietin has been identified in hepatoblastoma tissue and serum from a patient, but its correlation with the thrombocytosis associated with this neoplasm is unclear.

Clinical findings

The most common presenting sign of hepatoblastoma is an asymptomatic abdominal mass. The child is often in good health, and the tumor usually is discovered incidentally, when an attentive parent, grandparent, or clinician discovers the mass on a routine examination or while bathing the child. Patients with the small-cell undifferentiated variant of hepatoblastoma, who often have distant metastases at diagnosis, are more frequently symptomatic. Accompanying symptoms such as pain, irritability, minor gastrointestinal disturbances, fever, and pallor occur in smaller numbers of patients. Significant weight loss is unusual, although patients may fail to thrive. In most series of hepatoblastomas and HCCs, a few patients present acutely with tumor rupture and intraperitoneal hemorrhage. Rarely, hepatoblastoma manifests with sexual precocity secondary to a β-human chorionic gonadotropin (β-hCG)–secreting tumor, and one patient with a hepatoblastoma was reported presenting with a biliary fistula. Finally, hepatoblastoma may present as a cardiac tumor.

A mild anemia associated with a markedly elevated platelet count is observed in most patients at diagnosis, and the platelet count can range into the millions. As noted previously, the cause is probably secondary to abnormal cytokine release.

Measurement of serum α-fetoprotein (AFP) is well established as an initial tumor marker in the diagnosis of hepatoblastoma and a means of monitoring the therapeutic response. The normal level in most laboratories is less than 20 ng/mL, whereas the AFP level at diagnosis in hepatoblastoma patients can range from normal to significantly elevated (7.7 × 10 ⁶ ng/mL); it is estimated that the AFP is elevated in 84% to 91% of patients with hepatoblastoma. One study reported a mean AFP level in hepatoblastoma of 3 million ng/mL, whereas the mean in pediatric patients with HCC was approximately 200,000 ng/mL. In infants younger than 1 year, the AFP is normally elevated and is highest at birth ( Fig. 93.3 ).

Some authors suggest that subfractionation more reliably indicates whether the increased AFP is secondary to a hepatoblastoma or HCC (see Chapter 89 ), an endodermal sinus tumor, or benign liver disease. The half-life of AFP is approximately 6 days, and in one study, 24 (77%) of 31 patients had levels decline postresection by at least 1 log before second-look surgery. Of these patients, 16 (50%) of 32 eventually had AFP levels decline to normal at the end of adjuvant therapy and had no clinical or radiographic evidence of hepatoblastoma at this point. Finally, 15 (94%) of 16 patients who attained a complete response also showed a decline in AFP levels of 2 logs or more before second-look surgery. A large, early decline in AFP levels was an independent predictor of survival in multivariate analysis. It has been suggested that a low initial AFP level, although exceedingly rare, is associated with worse survival. A retrospective analysis of International Society of Pediatric Oncology Liver Group (SIOPEL) groups 1 through 3, which focused on hepatoblastoma patients presenting with AFP levels less than 100 ng/mL ( N = 21 patients during a 14-year period), concluded that low initial AFP level was associated with extensive disease at diagnosis, poor response to chemotherapy, and poor outcome. Small-cell undifferentiated hepatoblastomas may also be associated with lower AFP levels. However, the small sample size limits multivariate analysis.

When interpreting the AFP level, it is important to realize that normal levels are very high at birth and decrease during the first 6 months of life. Premature newborns may have AFP levels in the range of 100,000 ng/mL. Term newborns also can have relatively high levels (10 ⁴ to 10 ⁵ ng/mL). By 2 months of age, most infants have levels ranging from 100 to 1000 ng/mL, and by 6 months, levels should be less than 100 ng/mL. Usually, levels decrease to normal (<20 ng/mL) after 6 to 7 months, but levels may remain elevated for 1 year after birth. AFP may also be elevated in the setting of liver damage, liver regeneration, or the presence of another tumor.

Imaging

The first imaging study is usually an abdominal ultrasound (US) (see Chapter 14 ). If duplex technique is used, tumor vascularity can be gauged, and the hepatic veins can be assessed. The ultrasonographer also should search for anomalies of the genitourinary system and rule out tumor thrombus in either the vena cava or the hepatic veins. Computed tomography (CT) (see Chapter 14 ) is useful to identify pulmonary metastases, identify diffuse hepatic involvement, and determine resectability. Oral and intravenous contrast material is used ( Fig. 93.4 A). CT scans can be performed quickly and can be completed in less than 2 minutes in helical scanners; this greatly shortens the required period of sedation for infants or small children, and it has the added advantage of being a quick and reliable screening method for pulmonary metastases. Hepatoblastoma will appear as a well-demarcated tumor without a capsule. CT angiography (CT portography) uses CT with fine cuts and an increased amount of intravenous contrast material to image hepatic tumors and the hepatic venous anatomy. CT portography may provide as much information as magnetic resonance imaging (MRI) (see Chapter 14 ), which is useful for evaluating hepatic lesions and their relationship to vascular structures. MRI can show the hepatic veins, vena cava, and bile ducts. MRI of a hepatoblastoma patient after neoadjuvant chemotherapy is shown in Fig. 93.4 B. Positron-emission tomography has been used to identify hepatoblastoma recurrence and to search for sites of metastatic disease, but it may not be reliable for lesions smaller than 6 to 10 mm ¹²² (see Chapter 18 ).

Staging

In the United States, the commonly used staging system is that from the Children’s Oncology Group (COG), based on operative findings ( Table 93.1 ). A tumor-node-metastasis (TNM) classification has also been used ( Table 93.2 ). The PRE- Treatment EXT ent (PRETEXT) system of disease staging has been used extensively by SIOPEL ( Fig. 93.5 ). It relies on radiographic findings before any therapy, including surgery, and does not account for the independent surgeon’s judgment at the time of surgery regarding resectability. This staging system is based on Couinaud’s system of segmentation of the liver and is thought to predict the degree of tumor infiltration, the extent of surgical resection, and the complexities involved in the resection. ^, This system classifies the tumor into one of four categories, depending on which sections of the liver do not include tumor ( Table 93.3 ). Additional criteria added in 2005 ( Table 93.4 ) further classify these tumors based on local extension, multifocality, rupture, and metastasis. ^,

PRETEXT was compared with pathologic findings in 110 patients and was correct in 51%, overstaged in 37%, and understaged in 12%. The authors compared this system with the Children’s Cancer Group/Pediatric Oncology Group (CCG/POG) and TNM schemes and claimed a better correlation with risk status. In this study, data from patients who had neoadjuvant chemotherapy were analyzed, whereas a study from the COG analyzed data from patients with a hepatoblastoma at diagnosis and reported that both the COG stage and PRETEXT were useful prognostic indicators. The PRETEXT system has been described as showing improved predictive value for survival compared with other staging classifications. Moreover, this system can be valuable for recognizing patients who are candidates for resection (PRETEXT stages I and II) and those who may benefit from lower dose chemotherapy. It is recommended that all liver tumor patients in future COG studies undergo PRETEXT staging. The PRETEXT system was recently updated and clarified to streamline its use in clinical trials.

Treatment

Multiple studies support the effectiveness of systemic chemotherapy combined with complete surgical resection of the primary hepatic tumor. ^, Survival depends on removal of the primary liver tumor, when imaging suggests that complete resection is feasible.

The first clinical decision is whether to initiate neoadjuvant chemotherapy or proceed with resection. Often, resection is not feasible if tumors are large and involve both hepatic lobes. Preoperative (neoadjuvant) chemotherapy results in tumor shrinkage and makes subsequent resection easier. In one study, the rate of shrinkage was high after initiation of chemotherapy, but it declined after two cycles had been administered ( Fig. 93.6 ). Another study focused exclusively on neoadjuvant therapy in PRETEXT stage III and IV patients and found that the majority of tumors that became resectable required only two cycles of chemotherapy, whereas several more required four cycles. Exquisite clinical judgment and good communication between members of the multidisciplinary team are crucial because approximately 60% of hepatoblastomas are resectable at diagnosis.

To confirm the diagnosis, an initial biopsy is required. For unresectable tumors, the initial surgical procedure should include a diagnostic biopsy and placement of a vascular access device for chemotherapy. A second laparotomy is performed after four cycles of chemotherapy if imaging studies show a good response and the tumor appears resectable.

Complete resection of the primary tumor is necessary for survival and may require extended hepatectomy and/or complex biliary reconstruction (see Chapters 101 , 118 , and 119 ). For hepatoblastoma, reports have suggested that gross total resection of the primary lesion may be adequate for cure in chemoresponsive tumors. ^, A SIOPEL report demonstrated that, for patients being treated with adjuvant chemotherapy, a microscopically positive resection margin did not affect event-free survival or overall survival.

For resected tumors (stage I) with fetal histology, further therapy is not recommended. All other stage I tumors without pure fetal histology, in addition to stage II patients, should receive four cycles of cisplatin, 5-fluorouracil (5-FU), and vincristine (C5V). Patients with stage III and IV disease should receive four cycles of chemotherapy, followed by resection or transplantation, followed by two more cycles of chemotherapy. C5V is administered, followed by doxorubicin if there is minimal response to C5V. Reports have suggested the use of doxorubicin from the start in this subgroup. ^, The combination of cisplatin plus doxorubicin was compared with cisplatin plus 5-FU plus C5V in a combined CCG/POG (intergroup) study. The efficacy was thought to be similar, but more complications resulted with the regimen containing doxorubicin, accounting for equivalent event-free survivals; however, a more detailed review of the analysis suggested that the doxorubicin-containing arm had an improved disease-specific survival. This finding implied that with better management of toxicity, patient outcome might be better with a doxorubicin-containing regimen. In 2009 a trial was published by SIOPEL 3 that randomized 255 “standard-risk” patients (defined as patients with PRETEXT stage I, II, or III and no evidence of extrahepatic disease) into two groups, the first treated with cisplatin alone and the second with cisplatin plus doxorubicin. They reported no difference in achieving resection, or in 3-year event-free or overall survival between these groups, indicating that standard-risk patients may be successfully treated with cisplatin alone. Clinical trials by COG and SIOPEL are planned to evaluate the use of doxorubicin, irinotecan, and other agents, especially in high-risk patients.

In patients with unresectable primary tumors, the use of liver transplantation is increasing (see Chapters 105 , 108 , and 110 ). An analysis published in 2005 showed an approximately 80% long-term disease-free survival in those receiving transplantation in large, solitary, or multifocal tumors invading all four sectors of the liver. The United Network for Organ Sharing (UNOS) database consists of more than 200 patients, with a median age of 2.9 years, who underwent orthotopic liver transplantation (OLT) for hepatoblastoma between 1987 and 2006. Approximately half of the patients had a recurrence. Overall survival was 80%, 69%, and 66% at 1, 5, and 10 years, respectively. A 2013 study that queried the UNOS and SEER databases from 1988 to 2010 and 1975 to 2007, respectively, determined that as many as 20% of hepatoblastoma cases are now being referred for OLT, and 5-year survival after transplant is estimated at 73%. In a multicenter review, data on 147 patients with hepatoblastoma were analyzed after liver transplantation. In almost three-quarters of these patients, the original surgery was OLT; the remaining patients either had residual disease after prior resection or had recurrent tumor. The first group of patients had an improved outcome, with 82% overall disease-free survival, compared with 30% in the second group. Smaller, single-center reports have reinforced findings that liver transplantation for hepatoblastoma has the best outcome when done as the primary procedure rather than as a salvage procedure.

Transplantation, however, does require the use of immunosuppressive treatment, which comes with its own set of side effects. Moreover, there is in increased chance of thrombosis of the hepatic artery after transplantation in children. The main causes of mortality after transplantation, accounting for 54% of this population, are metastases and recurrence. COG is continuing to investigate the role of liver transplantation in hepatoblastoma. A global database has been instituted to aid in this endeavor (see Chapters 108 , 110 , and 111 ).

In one study, the 1-year survival for patients presenting with metastases was no different from that in patients with localized tumors. In another study by SIOPEL, the 5-year overall and event-free survival for children with hepatoblastoma who presented with pulmonary metastases were 57% and 28%, respectively. This study suggested that 25% to 30% of patients with synchronous pulmonary metastases are curable. It is still necessary to resect the primary liver tumor, and pulmonary metastasectomy should be done only if the primary site is controlled. Many pulmonary metastases resolve with chemotherapy, but thoracotomy and resection are sometimes required for larger or persistent metastatic lesions.

No prospective studies are underway for pulmonary metastasectomy, but one study describes the advantage of pulmonary metastasectomy for diagnosed lesions that remain after neoadjuvant therapy. Some radiation oncologists have treated pulmonary metastases with external-beam radiotherapy in an approach similar to that used for Wilms tumors, but with 18 to 20 Gy administered ; however, this may be associated with significant pulmonary toxicity and has not resulted in cure. One case report notes long-term survival in a hepatoblastoma metastatic to the brain, but in general, there is almost no reported cure in patients with spread to sites outside of the lung or local lymph nodes.

Outcome

Following a gross total resection, the 5-year event-free survival is 83%, but this drops to 41% in patients with tumor remaining after surgery. Some patients with microscopic residual tumor are curable with continued chemotherapy and may benefit from external-beam radiotherapy to the primary hepatic site. Resection of many hepatoblastomas may be easier after chemotherapy, and complete resection of the primary hepatic tumor is necessary for survival. In multivariate analysis, factors that have been independent predictors of worse prognosis include a high TNM stage, unresectable tumor, bilobar involvement and multifocality, AFP less than 100 ng/mL or greater than 10 ⁵ ng/mL, distant metastases, embryonal versus fetal histology, and vascular invasion. COG has reported 3-year event-free survival rates of 90%, 50%, and 20% for stages I to II, III, and IV, respectively.

Prognostic factors

These data have been generated for a large number of patients by the CHIC collaboration. ^, Importantly, these prognostic factors are determined at enrollment on study. Negative prognostic factors are PRETEXT IV tumors, distant metastases at diagnosis, unresectable vessel involvement (portal or hepatic vein), extrahepatic tumor extension, multifocal tumor, tumor rupture at presentation, and age at diagnosis of greater than 8 years. Parameters associated with an improved prognosis include PRETEXT I and II tumors, and patients who are from 0 to 2 years of age at diagnosis. An initial AFP of 1000 to 1,000,000 is associated with the best prognosis, whereas an AFP less than 100 has the worst outcome. AFPs that are greater than 1,000,000, or from 100 to 999 have intermediate survival ( Figs. 93.7 , 93.8 , and 93.9 ).

Future directions

Advances in the surgical exploitation of the hepatic segmental anatomy (see Chapter 2 ) and improvements in surgical technique (see Chapters 101 , 102 , and 118 ) and perioperative management (see Chapter 26 ) have allowed more extensive resections so that even very large and bilobar tumors can be removed successfully. Segmental resection is also feasible and may allow resection of multifocal lesions (see Chapters 102B and 118B ).

Various novel treatments are currently under investigation. First, transcatheter selective arterial chemoembolization, which involves the direct injection of chemotherapeutic agents to the tumor, hypothetically decreases systemic toxicity (see Chapter 94 ). The average decrease in tumor size was 84% in one study, and unresectable hepatoblastomas may become resectable with this intervention. Doxorubicin, cisplatin, and fluorodeoxyuridine have been promising, because these agents have a high hepatic extraction; case reports exist with striking results. For example, Yokomori et al. describe the total regression of a tumor in a 4-month-old infant with fetal hepatoblastoma treated for 1.5 years with 5-FU, vincristine, doxorubicin, and cisplatin. No recurrence was seen after 6 years of follow-up. Risks involved with this technique include infection, thrombosis, or shifting of the catheter. Furthermore, it is challenging to carry out in children, and prospective studies are needed. Other new approaches include treatment with anti-AFP antibodies, IL-2, and viral transfection vectors to attack malignant hepatic cells.

Epidemiology

Hepatocellular carcinoma (HCC) (see Chapter 89 ) accounts for 22% to 23% of pediatric liver tumors but is rare in infancy. ^, Approximately 1.2 cases per 1 million children exist in the United States. The Lung Cancer Study Group of Japan reported no cases in children age 4 years or younger in a series of 2286 patients with histologically reviewed tumors. Historic series without pathology review may report a higher rate of infantile HCC owing to misdiagnosis of some early hepatoblastomas. A 2014 query of the SEER database, with data spanning from 1973 to 2009, reported 218 HCC patients diagnosed before 20 years of age. Of these, 9% were younger than 5 years at diagnosis, and nearly half of that group was in the first year of life. Sixteen percent of tumors were diagnosed in children from 5 to 9 years of age, 27% in children from 10 to 14 years of age, and 48% from 15 to 19 years of age. We have had personal experience of infants with well-documented HCC, and it affects approximately 0.5 children younger than 15 years per 1 million annually. SEER data indicate that HCC accounts for 93% of primary hepatic malignancies in the 15- to 19-year-old age group. The incidence is bimodal, with an early peak that is lower than that of hepatoblastoma. Most of these early cases occur before 5 years of age. A second peak occurs between 13 and 15 years of age. HCC has a male predominance (1.3 to 3.2:1), and in areas endemic for hepatitis B, the male/female ratio may be reversed, at 0.2:1. Approximately 35 to 40 new HCCs are diagnosed per year in the pediatric age group in the United States. Incidence reported for the years 1973 to 1977 versus 1993 to 1997 showed a decrease from 0.45 to 0.29 cases per 1 million.

In contradistinction to HCC in adults, conditions associated with cirrhosis occur in only 30% to 40% of children with HCC; the remaining 60% to 70% of tumors are present without any cirrhosis. However, there are certain known risk factors. Hepatitis B and C infection correlates with the incidence of HCC. In Asia, 85% of HCC patients, both adult and pediatric, are hepatitis B surface antigen positive, whereas this is found in only 10% to 25% of patients in the United States (see Chapter 68 ). The relative risk for the development of HCC is 250:1 for patients with chronic active hepatitis compared with patients without hepatitis surface antigen positivity. Hepatitis C antibodies are found in 20% of patients with HCC. In one report, an infant with a history of neonatal hepatitis developed HCC. A universal vaccination program against hepatitis B has reduced the incidence of HCC in Taiwan. The average annual incidence of HCC in children 6 to 14 years of age declined from 0.70 per 100,000 children in the years 1981 to 1986 to 0.36 in the years 1990 to 1994 ( P < .01) coincident with widespread administration of the hepatitis B vaccine. The mortality rate also decreased during this period. Antiviral therapy with lamivudine reduced the risk of development of HCC in patients with chronic hepatitis B infection and cirrhosis or fibrosis.

Other conditions associated with the development of HCC include cirrhosis, α ₁ -antitrypsin deficiency, tyrosinemia, aflatoxin ingestion, hemochromatosis, hepatic venous obstruction, androgen and estrogen exposure, Alagille syndrome (arteriohepatic dysplasia), Thorotrast administration, progressive familial intrahepatic cholestasis, glycogen storage disorders, and congenital portosystemic shunts ^, (see Chapter 89 ). One case of childhood HCC developing in a patient with neurofibromatosis has also been reported.

In one comparative study of pediatric HCC and hepatoblastoma, numerous discriminating features were reported. The mean age at presentation was 18 months for hepatoblastoma versus 10 years for HCC. The initial resectability of HCC was 45% and did not improve with chemotherapy, whereas 91% of hepatoblastomas could be completely resected before or after chemotherapy. Tumor rupture occurred in 36% of hepatoblastomas versus 9% of HCCs. Most important, the survival of patients with HCC was much worse, with an overall 5-year survival of only 24%.

Pathology (see Chapter 87 )

HCCs are highly invasive and often multicentric at diagnosis, with frequent hemorrhage and necrosis. Nuclear pleomorphism, nucleolar prominence, and the absence of extramedullary hematopoiesis are observed, and the cells are larger than normal hepatocytes. Low-grade HCCs may look similar to normal hepatocytes, especially if a limited amount of tissue is sampled. Invasiveness, and vascular invasion in particular, is a hallmark of these tumors. Extrahepatic dissemination to portal lymph nodes, lungs, and bones is frequent at diagnosis and strongly affects survival. HCCs naturally progress from capsular invasion to extracapsular extension, then vascular invasion, and finally to intrahepatic metastases. A strong correlation has been found between intrahepatic metastases and portal vein thrombosis; this suggests that efferent tumor vessels anastomose to the portal, rather than hepatic veins, allowing intrahepatic spread and explaining the multicentricity that is a hallmark of HCC.

Biology and molecular biology

Most investigations into the basic biology of HCC involve the study of hepatitis B and its relationship to carcinogenesis. In one in vivo model in which rats developed HCC after prolonged feeding with glyceryl trinitrate, KRAS point mutations were identified in 8 of 18 animals that developed tumors, and no TP53 mutations were seen. Another model in mice engineered an inducible mutation in the YAP gene, which led to overexpression when the mice were fed doxycycline. These mice began to develop HCC nodules after 8 weeks of YAP overexpression. Cytogenetic data indicate that chromosomal abnormalities are complex, and consistent patterns have historically been difficult to establish (see Chapters 9B and 9C ).

Clinical findings

Children and adolescents with HCC are often seen initially with palpable abdominal masses (40%), but many are asymptomatic at diagnosis. Pain is frequent (38%) and may occur in the absence of an obvious mass. Constitutional disturbances such as anorexia, malaise, nausea and vomiting, and significant weight loss occur with greater frequency. Jaundice is an uncommon feature of either disease, but AFP is elevated in approximately 85% of patients, with most levels greater than 1000 ng/mL. Although elevated, these levels are usually less than those measured in hepatoblastoma patients.

Staging

The PRETEXT staging system listed for hepatoblastoma is also used for HCC in childhood.

Treatment

For more than a quarter century, no significant progress has been made in treatment of the pediatric population with HCC. This tumor remains extremely resistant to current chemotherapy agents, and long-term survival is impossible without complete resection. Because of a high incidence of multifocality within the liver, extrahepatic extension to regional lymph nodes, vascular invasion, and distant metastases, complete resection is often impossible. Infiltration with thrombosis of portal and hepatic venous branches is common, and even the vena cava may be involved. Furthermore, the cirrhosis found in a number of HCC livers may preclude an extensive resection. The first prospective SIOPEL study reported a 36% resectability rate among children with HCC. Even with complete resection, however, the prognosis remains poor secondary to the high rate of recurrence; 5-year survival postresection is reported to be 40%.

Historically, the same chemotherapy protocols used for hepatoblastoma were also applied to HCCs in childhood; however, HCC is rather unresponsive to chemotherapy overall, although cisplatin in particular has shown activity against it (see Chapter 99 ). The addition of doxorubicin, as well as the VEGF and RAF kinase inhibitor sorafenib, has also shown a modest benefit. The utility of external-beam radiation therapy is unclear; it can aid with temporary control of gross disease, but it has not been shown to reduce the risk of relapse in patients with residual disease after resection. Because of the poor survival of patients with HCC, present thinking is to apply new and innovative approaches to these cancers.

Unresectable HCCs can be palliated with embolization with or without added chemotherapeutic agents or radioisotopes (see Chapter 94 ). Percutaneous intralesional injection of ethanol also has been of palliative benefit when lesions are small (see Chapter 96D ). Radiofrequency ablation (RFA) of these tumors, percutaneously or at laparoscopy/laparotomy, has been associated with tumor resolution and prolonged survival (see Chapter 96B ). Lin et al. reported that RFA was superior to ethanol injection in HCCs 4 cm or less in diameter. Moreover, preliminary research using metronomic chemotherapy and adjuvant antiangiogenic treatments is currently under way.

Because standard therapies have proven unsuccessful, liver transplantation has been used more widely (see Chapters 105 , 108 , and 110 ). The Milan criteria used for adults have been extrapolated to children. They stipulate a single tumor no larger than 5 cm in diameter or as many as three tumors each 3 cm or less in diameter, an absence of macroscopic portal vein invasions, and absence of extrahepatic disease; however, no current data support the efficacy of using the Milan criteria in this population. The UNOS database included information on 41 patients with HCC who received OLT from 1987 to 2006. The overall survival was 86%, 63%, and 58% for 1, 5, and 10 years, respectively. Analogous to hepatoblastoma, mortality was mainly secondary to recurrence, which occurred even more often than it did in hepatoblastoma (86% vs. 54%). A more recent SEER study demonstrated a 5-year survival of 85% after transplant.

Outcome

The overall survival from metastatic or unresectable HCC in childhood approaches zero, and it remains a therapeutic problem ( Table 93.5 ). Occasionally, resection of localized lesions results in long-term survival. The most recent query of the SEER database reports a 5-year overall survival of 45% after successful resection; 42% of patients who underwent surgery were alive after 10 years, and 28% after 20 years. The trend is to separate HCC from hepatoblastoma in clinical studies because of its greatly differing biologic behavior.

Fibrolamellar variant

Fibrolamellar HCC (FLH) is a histologic variant of the disease first described in 1956. Due to a specific genetic finding, it is becoming clear that FLH is a distinct tumor from HCC that is virally determined or otherwise idiopathic. It is characterized microscopically by bands of collagen that are arranged in a layered, or lamellar, pattern (see Chapters 87 and 89 ). It is relatively rare, with an age-adjusted incidence of 0.2 per 1 million, and for this reason, large series have been limited. However, a 2012 systematic review published in the Journal of the American College of Surgeons synthesized most of the published series, ranging back to 1980 and covering 575 patients. FLH is notable for multiple clinical characteristics that distinguish it from nonfibrolamellar HCC. Typically seen in older children and young adults, the age at diagnosis has been reported as anywhere from 1 to 62 years of age, with an overall median of 21 years. It is exceedingly rare in patients younger than 5 years. Unlike standard HCC, it is equally prevalent in males and females, and only 3% of patients have cirrhotic livers. Furthermore, on multivariate analysis, male gender has been shown to fare better, an observation not seen in standard HCC. At 5 and 10 years, overall survival in males was significantly better than in females, with a hazard ratio of 0.33. FLH has been described as having a lower mitotic index and following a more indolent course compared with standard HCC, and it was once thought to be associated with a better long-term prognosis ; however, studies indicate that when stage is controlled for, the survival is similar between standard HCC and its fibrolamellar variant. The tendency of FLH to grow slowly, as a single mass, compared with the early multifocality of standard HCC, may account for it being diagnosed more often at an early stage. Despite this, 10% of patients may come to medical attention with tumor rupture and a hemoperitoneum.

Biologically, FLH has been noted to coincide with an elevated serum unsaturated vitamin B ₁₂ binding capacity, as well as elevated neurotensin levels. In contrast to standard HCC, which can have an AFP greater than 100 ng/mL in as many as 85% of patients, AFP is elevated in FLH only 10% of the time. ^, One study performed on 15 patients with FLH found a translocation on chromosome 19, leading to a chimeric transcript that merged the activating region of heat-shock protein DNAJB1 with the functional region of protein kinase A, causing it to become hyperactive. Protein kinase A affects many cellular pathways and has been implicated in several other cancers. The translocation was present in 100% of tumor samples, compared with 0% in matched healthy liver controls derived from the same patients, suggesting it plays a role specific to the disease ( Fig. 93.10 ). More recently, four individuals without a DNAJB1-PRKACA fusion but with FLH were identified; three had Carney complex. All had evidence of inhibition of a PRKACA regulatory protein.

Similar to standard HCC, chemotherapy has little efficacy in FLH, and resection offers the only chance of cure (see Chapter 99 ). FLH is staged using the same PRETEXT criteria as hepatoblastoma and HCC, and like both of these diseases, patients may be offered partial hepatic resection or transplant, depending on the extent of disease. To date, the Milan criteria have typically been applied; however, as with standard HCC, the generalizability of these criteria to the pediatric population is questionable. In contrast to standard HCC, patients with FLH generally have a higher resection rate of 60%, as well as a higher 5-year survival rate of 59% after resection, compared with 40% in standard HCC. ^, Even without resection, 20% of FLH patients are reported to be alive after 5 years, compared with 3% of standard HCC patients.

Future directions

Gene therapy with viral vectors that attack dividing cells is being investigated. Hepatocytes rarely divide unless stimulated by liver resection. Viruses such as herpes attack dividing cells and can be molecularly manipulated to contain cytotoxic genes, and modified herpesvirus can be transfected efficiently into hepatoma cells. ^, One group has used an adenovirus vector to deliver murine endostatin to tumors in nude mice injected with HCC cells with a resultant reduction in tumor growth. A number of workers are investigating selective inhibition of the DNAJB1-PRKACA fusion.

Incidence

Although embryonal rhabdomyosarcoma of the extrahepatic bile ducts is extremely rare, it is the most frequently seen malignancy in the biliary tree of children. Ten cases were reported in intergroup rhabdomyosarcoma studies I and II, constituting 0.8% of confirmed tumors in those studies. Fewer than 100 cases have been reported since 1975. ^,

Pathology

Biliary rhabdomyosarcoma is categorized into five histopathologic subtypes: embryonal, alveolar, botryoid embryonal, spindle cell embryonal, and anaplastic (see Chapter 87 ). Most tumors are of the embryonal histopathologic subtype, which accounts for 60%. They often show botryoid characteristics similar to other rhabdomyosarcomas that arise in a hollow viscus. Immunohistochemistry and nuclear staining are informative for diagnosis of embryonal rhabdomyosarcoma, as well as desmin and muscle-specific actin. ^{, ,} Distant metastases develop in approximately 40% of patients, but mortality is most often due to the effects of local invasion, including biliary sepsis. Long-term survival is considered to be 60% to 70% and is not dependent on resectability. Rhabdomyosarcoma of the liver and the ampulla of Vater, but not involving the bile ducts, also has been reported but is very rare.