Liver tumors are rare in children. Primary liver tumors account for approximately 1% of the tumors in children. About 50-60% of the primary liver tumors in children are malignant, and more than 65% of those malignant liver tumors are HBL [1–4]. The constant annual incidence of HBL in children is 0.5-1.5 per 1 million in Western countries . The currently accepted hypothesis is that the HBL cells are derived from the pluripotent hepatic stem cells. Those stem cells retain the ability to differentiate into both the hepatocytes and the billiary epithelial cells, and retain the ability to express markers for both the cell types as a feature in HBL . HCC is the second most common pediatric primary malignant liver tumors, accounting for about 20% of the primary liver tumors in children [1, 4]. The two diseases can be treated by chemotherapy and surgical resection, or by liver transplantation as a last resort [8, 9]. It seems that children with HBL have a better survival rate when compared with children with HCC . Recently, approximately 75% of children with HBL can be cured completely .
The results of this study have indicated that children with HCC have a higher incidence of the positive HBV than those with HBL. Many studies have revealed that HBV is related to HCC in children. HBV has been considered a major etiological factor of HCC in children beyond the age of 4 years . However, little is known about the etiology of HBL. The most well-established risk factors of HBL are Beckwith-Wiedemann syndrome, family history of familial adenomatous polyposis, low birth weight, and smoking by either or both of the parents . This may explain the different incidence of the HBV infection related to HBL and HCC in children in this study.
HBL was mainly a tumor occurring in early childhood [6, 7], and 30-50% of the HBL cases occurred in the first year of childhood and 50-90% before the age of 5 years [2, 3, 6, 7]. The onset age of HCC was older than that of HBL in children. Lee, et al. reported that there was an HCC onset peak at 12 years old . In our series, the average HBL onset-age was 2.9 years, but the average HCC onset-age was 9.3 years. Frequently, HCC was one of the long-term results of chronic viral infection . And 80% of the HCC cases developed in the cirrhotic livers . In developing countries, young patients were apt to suffer from chronic hepatitis B virus infection [13, 14]. The time from liver cirrhosis to HCC might be the main reason for the older onset-age of HCC. In our study, positive HBV was found in 66.7% of the HCC cases but in 13.3% of the HBL cases. These findings have supported the idea that HBV is a major etiological factor of HCC in children beyond the age of 4 years.
There are more boys than girls who suffered from HBL or HCC, which was one of the common characteristics of those two types of tumors occurring in children . The results of our study have also supported this finding. The imaging study is important in evaluation liver neoplasms. CT, MRI and ultrasound are the most-commonly-used modalities for pediatric doctors in their medical researches as well as their clinical practice. Ultrasound is accepted as a first-line imaging method because of its less irradiation, greater convenience and better real-time . Ultrasound is extremely valuable in detecting much smaller lesions, especially in detecting fluid and blood-flow in a lesion, and it also can evaluate the hepatic vascular anatomy . As a rule, the initial diagnosis of live tumor is usually made by the abdominal ultrasound examination. Complete surgical resection is a key to the permanent cure of the disease, and so ultrasound examination can be used to exactly localize the tumor and assess the extent of the tumor development. Scintigraphy with 99mTc-labelled monoclonal anti-AFP is mainly used in the tumor staging. However, the clinical usefulness of this technique for HBL is not completely clear .
The lesion sizes were significantly different between the two groups. The long-axe diameter of HBL was greater than that of HCC. The younger onset-age of the tumor was a possible reason. The children with HBL could not tell exactly about their early symptoms, so the diagnosis was made relatively late. Although the average long-axe diameter of HBL was greater than that of HCC, there were some overlaps between the two groups. Therefore, the lesion size could not be used as an important indicator for differentiating HBL from HCC.
The two types of tumors in this study had some similarities in the ultrasound features. Firstly, most of the lesions in both HBL and HCC groups were located in the right lobe of the liver in more than 40% of the cases. Our study showed that the location of HBL was similar to that reported in the previous studies, in which the tumor location was in the right lobe of the liver in 60-70% of the HBL cases [7, 12]. Secondly, most of the patients with HBL or HCC had solitary lesions (66.7% in the HCC group, 73.3% in the HBL group), which coincided with what the medical literature had reported . However, these two diseases had many different imaging features. Dachman, et al. reported that liquefaction might be caused by intralesional necrosis or hemorrhage . Typically, ultrasound could identify liquefaction as a hypoechoic area at the center of the lesion. In our series, liquefaction appeared in 25% of the HCC cases but appeared in 56.7% of the HBL cases; therefore, a significant difference could be observed between the two groups. So, liquefaction could be considered significant for the HBL diagnosis.
Calcification occurred in such liver diseases as HBL, HCC, teratoma and involuting haemangioma. So, calcification was not a specific indicator in the differential diagnosis . Calcification might cause acoustic shadowing, which was better depicted by CT or ultrasound rather than MRI . Microscopically, the calcification presence on radiographs was often associated with the osteoid presence . Histologically, calcification was often related to the mixed type of HBL, and the reason was probably that there was a formation of the osteoid foci . Calcification rarely occurres in the HCC patients, which is related to the radiation therapy . Jha, et al.  reported that calcification was observed in about 40% of the HCC cases and observed in 50% of the HBL cases [2, 8]. Our study showed that calcification was occasionally found in the two kinds of tumors that had not been treated before. But our study still failed to give any evidence that calcification is a specific indicator in the imaging studies for a differentiation of HBL from HCC. The reason for the great difference in calcification between the two groups might be that there were quite a few cases of the mixed epithelial/mesenchymal type of tumors.
Septa were formed by the fibrous tissue band around or inside the lesions. Histologically, HBL is usually present as a solitary large solid mass, which could contain fibrous bands, leading to a "spoked-wheel appearance" . The typical fibrous septa could be observed in the grey-scale ultrasound images in our study. A previous case-study suggested that in a multifocal setting, the presence of a dominant lobulated mass was a clue to the diagnosis of HBL . In our series, the septa were more commonly observed in untreated HBL than in HCC. Thus, this ultrasound feature can be used as an indicator for differentiating HBL from HCC. Among the ultrasonic and clinical features of the HBL lesions, the negative HBS-Ag was the most sensitive parameter, and the septa were the most specific parameter. When a combination of liquefaction, septa, onset age (< 5 years) and negative HBS-Ag was used, the sensitivity was raised to 90%, the accuracy was raised to 88%, and the negative predictive value was raised to 73%.
Our study still had some limitations including a long duration (1993-2009) of the materials collection. Besides, at the beginning of the study, the Color Doppler Flow Imaging (CDFI) and elastography were not introduced into the practice. Whereas contrast enhanced ultrasound (CEUS) has already been mature in the diagnosis of the liver occupying lesions in recent years, childhood is still a contraindication to the contrast agent SonoVue. Thus, we were unable to give enough information about the CDFI, CEUS and elastography manifestation of HBL and HCC in children. But we think that our finding is still valuable for the pediatric physicians.