44. Assessment of Pediatric Liver Tumors - Part 1 . By Prof. Dr. Alexander Towbin
With Dr. Alexander J. Towbin
Chapter 3 of 12
Hemangioma imaging
Case 1: Hepatic Hemangioma - Imaging Features and Classification
Cued at 8:17 · press play
Expert statements on this page
No expert statements were drawn from this page.
Host summaries · secondary, not cited in answers
Educational content from recorded physician discussions — not medical advice. Talk to your (or your child's) care team about your situation.
More from this expert
Dr. Alexander J. TowbinPodcast
Customer Service in Radiology: Satisfying Your Patients and Referrers
Alexander J. Towbin · Published Nov 2026
Video
Measuring Radiology's Impact: Core Concepts for Tracking Patient-Oriented Outcomes & High-Value Care
Alexander J. Towbin · 10 min · Published Sep 2026
Video
Five Minutes in Healthcare Featuring Dr. Alexander Towbin
Alexander J. Towbin · 11 min · Published Jul 2020
Video
Get to Know Your Radiologist Dr. Towbin
Alexander J. Towbin · 6 min · Published Mar 2017
Video
ED-Radiology: Turn Around Time
Alexander J. Towbin · 3 min · Published Jan 2015
What the experts said
Hepatic hemangiomas are the most common benign liver tumor in children and are more common in girls.
About 85% of patients diagnosed with hepatic hemangiomas are diagnosed in the first two months of life.
Focal hepatic hemangiomas are the only type that are congenital hemangiomas, present at birth and can be diagnosed in utero.
Focal hepatic hemangiomas are not GLUT1 positive, unlike other types of pediatric hepatic hemangiomas.
Rapidly involuting congenital hemangiomas regress within the first two years of life.
Patients with more than five cutaneous hemangiomas should have a liver ultrasound to look for hepatic lesions.
Infantile hemangiomas are not present at birth, grow rapidly in the first year of life, and slowly involute over the next five years, usually gone by school age.
Patients with multifocal and diffuse hepatic hemangiomas can present with congestive heart failure.
Patients with hepatic hemangiomas may present with hypothyroidism.
Hepatobiliary agents are used for almost all indications and all types of liver imaging in children.
Hepatoblastoma is the most common primary liver tumor of childhood and the third most common abdominal malignancy in childhood, after Wilms tumor and neuroblastoma.
Hepatoblastoma is more common in males with a median age of around 19 months.
A very small percent of hepatoblastoma patients present after age four.
The most important association with hepatoblastoma is prematurity, particularly extreme prematurity and extremely low birth weight, thought to be related to oxygen and free radicals.
PRETEXT stands for pre-treatment extent of tumor and is a radiologic staging system used to predict surgical complexity.
Surgery is curative for hepatoblastoma; chemotherapy is important but cures only happen because of surgery.
PRETEXT is used in all cooperative group trials and has been shown to be an independent predictor of overall survival.
The PRETEXT group determines which sections of the liver are involved with tumor and represents the number of contiguous sections that need to be resected to completely remove the tumor.
For PRETEXT 1, tumor must be in the right posterior section or left lateral section, the only locations where conventional resection can remove the tumor completely.
A caudate-only tumor is defined as PRETEXT 2 because either a left or right hepatectomy is required to remove the caudate; the caudate is not removed in isolation.
Central tumors involving the right anterior and left medial sections are PRETEXT 3 because surgeons would perform an extended left or extended right hepatectomy rather than a mesohepatectomy, which is considered the most heroic surgery with suboptimal outcomes.
PRETEXT 4 means every section of the liver is involved with tumor and almost always leads to transplant; multifocal tumors are by far the most common PRETEXT 4 presentation.
Vascular involvement in PRETEXT staging gives important information regarding surgical complexity and prognostic information, and increases risk stratification.
Hepatic venous involvement is defined as either tumor obliterating or encasing more than 50% of all three hepatic veins (right, middle, left), or tumor thrombus in any one hepatic vein, or the intrahepatic IVC encased by more than 50% or obliterated.
Obliteration of a vessel means it is completely not visible, not just squished or showing a small sliver; usually there will be enlargement of azygos vessels if IVC obliteration occurs.
Portal venous involvement requires two main portal vein branches (right and left) to be obliterated/encased >50%, or tumor thrombus in any portal vein branch.
Cavernous transformation of the portal vein is assumed to represent portal vein thrombosis and is considered tumor involvement.
Bland thrombus is not distinguished from tumor thrombus in PRETEXT staging; any thrombus is called tumor thrombus or tumor in vein.
Vascular involvement must be in the main branches of vessels: for hepatic veins before any bifurcation, and for portal veins before bifurcation (not including the umbilical segment).
PRETEXT annotation factors predict surgical complexity and worsened prognosis; there are six additional annotation factors beyond portal and hepatic vein involvement.
Extrahepatic disease is diagnosed by peritoneal nodules (one >1 cm or two or more >5 mm), tumor crossing tissue planes, or normal tissue surrounding tumor by more than 180 degrees.
Multifocal disease requires normal liver interspersed between lesions; if tumors connect in any way, it is not considered multifocal.
Hepatobiliary phase imaging often reveals satellite lesions not visible on other phases, which can change staging from PRETEXT 3 to PRETEXT 4 and alter surgical planning from resection to transplant.
Tumor rupture is diagnosed by one or more findings of hemorrhage: on CT, fluid density >25 Hounsfield units (chosen for higher specificity); on MRI, blood degradation products; on ultrasound, heterogeneous echogenic debris or internal complexity/septations in fluid.
The caudate is specifically defined anatomically with boundaries at the ligamentum venosum (left), porta hepatis and ligamentum teres (anterior), perpendicular to the IVC axis (right), liver dome (superior), and between main portal vein and IVC (inferior).
Lymph node metastases are defined mostly by size criteria and are extremely rare in hepatoblastoma but more common in hepatocellular carcinoma.
Pulmonary metastases are defined as either one nodule >5 mm in diameter, or two or more nodules >3 mm in diameter.
Hepatocellular carcinoma is the second most common liver malignancy in children but is quite rare.
Around 30% of pediatric HCC patients have a pre-existing liver condition, compared to earlier literature suggesting up to 50%.
Pediatric HCC tumors are often C-met mutated, have higher rates of loss of heterozygosity on 13Q, and have lower levels of cyclin D1 compared to adult HCC.
LI-RADS (Liver Imaging Reporting and Data System) is a comprehensive system to standardize terminology, technique, interpretation, reporting, and data collection of liver imaging.
LI-RADS studies in adults show substantial inter-reader agreement with kappa around 0.7 and accuracy in the 80% range, often high 80s.
The LI-RADS document explicitly states do not apply to people less than 18 years of age.
Very few pediatric patients have had hepatitis B or C long enough to develop HCC, and pediatric cirrhosis is often due to conditions excluded from LI-RADS application.
Only about 50 patients are diagnosed with HCC in the pediatric age group in the United States each year, compared to 25,000 in adults.
Children often have larger tumor size at diagnosis compared to adults, making it easier to identify malignancy without needing to distinguish subtle features.
In post-Fontan livers, benign hyperenhancing masses may demonstrate washout and be mistaken for HCC by imaging criteria, though most lesions in one study were not biopsied and HCC development in Fontan patients can be slow.
The only study evaluating LI-RADS in children showed positive predictive value of 33-75% for LI-RADS 5 lesions and low negative predictive value for LI-RADS 1-3 lesions, problematic for a screening test.
The pediatric LI-RADS work group statement has been accepted to Radiology and addresses imaging parameters, image interpretation, and reporting of pediatric liver tumors.
The work group evaluated evidence quality using GRADE classification (A=high quality multi-center trials, D=expert opinion) and strength of recommendation via survey (1-1.49=very strong, 3-4=weak).
The work group gave a grade A, very strong recommendation for application of PRETEXT staging in pediatric hepatoblastoma or HCC at all imaging time points, based on multiple cooperative group trials showing PRETEXT is prognostic.
The work group gave a grade C, very strong recommendation (stronger than PRETEXT) for MRI with hepatobiliary agents for initial assessment of known or suspected focal pediatric liver lesions, based on one study showing improved diagnostic confidence, lesion detection, and radiologist confidence.
The work group recommended MRI for imaging patients with resected lesions (grade C, very strong), as hepatobiliary phase helps define tumor from areas of poor perfusion in post-surgical liver remnants.
The work group made no recommendation on screening for liver cancer in children with diffuse liver diseases due to wide variation in predisposing syndromes and overall rarity of pediatric liver tumors.
The work group recommended CT only for contraindications to MRI, when anesthesia cannot be provided in one setting or at all, or when chest CT is needed and cannot be done under one anesthesia setting (grade based on anesthesia evidence, strong but not very strong recommendation).
If CT is performed for known or suspected hepatoblastoma or HCC, the work group recommends arterial phase and late portal phase (very little evidence, strong recommendation).
For screening patients (e.g., Fontan hepatopathy) who require CT, the work group recommends three phases to identify worrisome lesions using LI-RADS features (fairly weak recommendation overall).