Predictors of survival and incidence of hepatoblastoma in the paediatric population

Abstract

Objectives

This study evaluates current trends in incidence, clinical outcomes and factors predictive of survival in children with hepatoblastoma (HB).

Methods

The Surveillance, Epidemiology and End Results (SEER) database was queried for the period 1973–2009 for all patients aged <20 years with HB.

Results

A total of 606 patients were identified. The age-adjusted incidence was 0.13 patients per 100 000 in 2009. An annual percentage change of 2.18% (95% confidence interval (CI) 1.10–3.27; P < 0.05) was seen over the study period. Overall survival rates at 5, 10 and 20 years were 63%, 61% and 59%, respectively. Ten-year survival rates significantly improved in patients with resectable disease who underwent operative treatment in comparison with those with non-resectable HB (86% versus 39%; P < 0.0001). Multivariate analysis showed surgical treatment (hazard ratio (HR) = 0.23, 95% CI 0.17–0.31; P < 0.0001), Hispanic ethnicity (HR = 0.61, 95% CI 0.43–0.89; P = 0.01), local disease at presentation (HR = 0.43, 95% CI 0.29–0.63; P < 0.0001) and age < 5 years (HR = 0.63, 95% CI 0.41–0.95; P < 0.03) to be independent prognostic factors of survival.

Conclusions

The incidence of paediatric HB has increased over time. Hepatoblastoma is almost exclusively seen in children aged < 5 years. When HB presents after the age of 5 years, the prognosis is most unfavourable. Tumour extirpation markedly improves survival in paediatric patients with local disease.

Introduction

Primary hepatic malignancies are rare during childhood, representing 1% of all paediatric neoplasms.^1–4 A total of 80% of all hepatic malignancies can be identified as hepatoblastoma (HB),^3,5,6 which is arguably the most common primary liver cancer in infants and children and occurs at an annual incidence of 0.05–0.15 patients per 100 000 population in subjects aged < 15 years.^1,7 In children aged < 5 years, 91% of liver malignancies are diagnosed as HB^3,5,6 and only 5% of HBs are seen in patients aged >4 years.^5,8

Hepatoblastoma most frequently affects children between the ages of 6 months and 3 years, at a median age of 18 months.^1,8 It is slightly more predominant in White males,¹ is seen more frequently in association with premature birth and very low birthweight^2,9 and in patients with genetic alterations such as familial adenomatous polyposis syndrome.¹

Complete tumour resection remains the cornerstone of treatment,¹⁰ but since the addition of chemotherapy in the 1980s, the overall survival rate has increased markedly from 30%.¹¹ Various groups have suggested different approaches to treatment; the International Childhood Liver Tumours Strategy Group (SIOPEL) suggests chemotherapy should be administered before the resection of the tumour, whereas the Children's Oncology Group (COG), along with Japanese and German groups, suggests the tumour should first be resected for staging and chemotherapy administered subsequently.¹⁰ Both approaches have achieved similar clinical outcomes with 5-year overall survival rates of >70%.^1,7,10–12

The purpose of the present study was to analyse current trends in tumour incidence and clinical outcomes and the factors predictive of survival in a large sample of paediatric patients with HB.

Materials and methods

The Surveillance, Epidemiology and End Results (SEER) database released on July 2012 was analysed to identify all incident patients with primary HB diagnosed between 1973 and 2009. Tumour location and histology were based upon topography and morphology codes according to the International Classification of Diseases for Oncology, third edition (ICD-O-3). The SEER database was first queried for all patients aged < 20 years with primary malignant hepatic tumours (topography code: C22.0). A total of 946 patients with primary malignant liver tumours were identified in the SEER database. Of these, 606 patients were identified as diagnosed with HB (ICD-O-3 code: 8970) and included in this analysis. There were no duplicate patients. Only percentages based on available data for each individual variable are given. Unknown values were excluded from all calculations. SEER*Stat software (Version 7.1.0; National Cancer Institute, Bethesda, MD, USA) was used to analyse incidence rates and trends from 1973 to 2009. All incidence data were age-adjusted and normalized to the 2000 US Standard Population. The annual percentage change (APC) was calculated using the weighted least squares method.

The tumour staging criteria used in this analysis were adopted from the SEER summary staging guidelines and differ from the tumour–node–metastasis (TNM) staging system. In this study, local staging represents disease that does not extend beyond the primary organ, whereas regional disease includes tumour extension to adjacent organs, regional lymph nodes or both. Documentation of distant metastases during the perioperative period led to the classification of affected patients as having distant disease.

Statistical analysis was performed using IBM spss Statistics 20.0 (IBM Corp., Armonk, NY, USA). Correlations between categorical variables were made using the chi-squared test. Data with a normal distribution were compared using Student's t-test and non-parametric data were compared using the Mann–Whitney U-test. Five, 10- and 20-year overall survival rates were calculated using life tables. The Kaplan–Meier method was used to calculate survival curves. Survival was calculated from the time of initial diagnosis to the date of last contact (or the date of death if the patient was deceased). The effects of demographic, clinical and treatment variables on survival were tested by utilizing the log-rank test for categorical values. Values are expressed as appropriate as the mean ± standard deviation, median (interquartile range) or number (percentage). Cox's logistic regression was used to identify predictors of survival using all variables found to be significant (P < 0.05) or near significant (P < 0.10) in univariate analysis or if clinically significant.

Results

Patient demographics and tumour characteristics

For the study period, a total of 606 patients were identified. The overall age-adjusted incidence of paediatric malignant liver tumours was 0.174 patients per 100 000 children with a statistically significant increase in incidence per year, leading to an APC of 1.79% [95% confidence interval (CI) 0.74–2.84; P < 0.05]. This increase in incidence may be largely attributed to an increase in incidence per year of HBs, resulting in an APC of 2.18% (95% CI 1.10–3.27; P < 0.05) (Fig. 1). Patient demographics and tumour characteristics are summarized in Table 1.

Figure 1.

Incidence of hepatoblastoma in children over a 37-year period. The incidence of hepatoblastoma has increased over time (annual percentage change = 2.18%; P < 0.05), in line with increases in the incidence of all primary tumours of the liver in the paediatric population

Table 1.

Demographic characteristics of 606 paediatric patients with hepatoblastomas registered in the Surveillance, Epidemiology and End Results (SEER) database

Age, years, median (range)	1 (0–19)
Gender, n (%)
Male	360 (59.4%)
Female	246 (40.6%)
Age, n (%)
<1 year	209 (34.5%)
1–4 years	346 (57.1%)
5–9 years	33 (5.4%)
10–14 years	13 (2.1%)
15–19 years	5 (0.8%)
Race, n (%)
White	489 (80.7%)
Black	46 (7.6%)
Asian	66 (10.9%)
Other	5 (0.8%)
Ethnicity, n (%)
Non-Hispanic	438 (72.3%)
Hispanic	168 (27.7%)
Tumour stage, n (%)
Local	253 (41.7%)
Regional	164 (27.1%)
Distant	163 (26.9%)
Unstaged	26 (4.3%)
Surgery, n (%)
No	142 (23.4%)
Yes	464 (76.6%)
Radiation, n (%)
No	587 (96.9%)
Yes	19 (3.1%)

Survival

Overall survival rates at 5, 10 and 20 years for the entire cohort were 63%, 61% and 59%, respectively. These results were consistent for up to 20 years after diagnosis. Overall survival rates are summarized in Table 2. Notably, age at diagnosis emerged as an important clinical indicator of disease prognosis. Adolescent patients had the lowest 5-year survival rate and none of these patients remained alive at 10 years after diagnosis. However, the occurrence of HB after the age of 10 years is exceedingly rare and only 18 such patients were reported in the SEER database for the study period. With reference to interventions, tumour resection improved survival in patients with resectable disease (Fig. 2).

Table 2.

Overall survival at 5, 10 and 20 years in 606 paediatric patients with hepatoblastoma

	Survival at 5 years	Survival at 10 years	Survival at 20 years	P-value
Gender				0.001
Male	62%	60%	60%
Female	63%	63%	58%
Age group				0.001
<1 year	64%	64%	64%
1–4 years	64%	63%	59%
5–9 years	44%	44%	44%
10–14 years	34%	34%	34%
15–19 years	20%	0%	NR
Tumour stage				0.001
Local	80%	80%	80%
Regional	59%	57%	52%
Distant	42%	39%	39%
Surgery				0.001
No	28%	25%	25%
Yes	74%	74%	74%

NR, not reached.

Figure 2.

Kaplan–Meier survival curves for paediatric patients with hepatoblastoma and resectable disease at presentation. Survival in patients with local disease is significantly better in those undergoing tumour resection compared with that in patients who do not undergo surgery (P < 0.0001)

Results of the univariate and multivariate analyses using a Cox regression model are summarized in Table 3. Age at diagnosis, patient ethnicity, tumour stage and surgical resection were all identified as independent predictors of survival.

Table 3.

Predictors of survival by univariate and multivariate analyses for children with hepatoblastomas according to the Surveillance, Epidemiology and End Results

	Univariate analysis		Multivariate analysis
	% surviving	P	HR	95% CI	P
Gender		0.507			0.890
Male	68		Reference
Female	69		–	–	–
Age (years)		0.001
<5	70		0.63	0.41–0.95	<0.030
>5	47		Reference
Race		0.010			0.192
White	69		Reference
Black	50		–	–	–
Asian	77		–	–	–
Other	100		–	–	–
Ethnicity		0.004
Non-Hispanic	65		Reference
Hispanic	77		0.61	0.43–0.89	0.010
Tumour stage		<0.0001
Local	83		0.43	0.29–0.63	<0.0001
Regional	65		0.77	0.54–1.08	0.130
Distal	50		Reference
Surgery		<0.0001
No	30		Reference
Yes	80		0.23	0.17–0.31	<0.0001
Radiation		0.001			0.260
No	70		Reference
Yes	32		–	–	–

HR, hazard ratio; 95% CI, 95% confidence interval.

Discussion

Although HB is the most frequent primary malignant hepatic tumour in children, it is quite rare and occurs in fewer than one subject per 100 000 births.¹³ Nonetheless, the prompt diagnosis and management of HB with chemotherapy and/or surgical resection remain paramount in treating these patients, who most often present with a mass in the upper right quadrant of the abdomen or with an evident epigastric abdominal mass. This study was conducted to analyse current trends in tumour incidence, and clinical outcomes and factors predictive of survival in a large sample of paediatric patients with HB.

Similarly to findings in recent reports on HB,^{1,3,5,6,13,14} analysis of data in the present cohort of patients showed that HB appeared most frequently in males (59.4%), Whites (80.7%) and children aged < 5 years (91.6%). Furthermore, an increasing number of patients are now diagnosed antenatally by routine sonography.^15–19 Specifically, within this analysis, the second most commonly affected age group comprised infants aged < 1 year (34.5%), followed by children aged 5–9 years (5.4%), children aged 10–14 years (2.1%) and, lastly, adolescents aged 15–19 years (0.8%). As expected, the age group most commonly affected by HB in this study included children aged 1–4 years (57.1%).

Several studies have reported the incidence of malignant liver tumours to be increasing in the USA.^8,20,21 For the period 1972–1992, the SEER database showed an annual increase of 5% in the rate of primary paediatric hepatic cancer.²⁰ Other data showed that liver cancer represented 2% of all malignancies in infants in the early 1980s, with the incidence doubling to 4% over the next 10 years.²¹ Similarly, the present study determined a statistically significant increase in the overall age-adjusted incidence of paediatric malignant liver tumours for the period 1973–2009, leading to an APC of 1.79%. This increase may potentially be attributed to an increase in the annual incidence of HB, leading to an APC of 2.18% over the 37-year period. Lastly, it has been suggested that improvements in technology, patient care and outcomes for premature infants may account for the increased incidence of paediatric HB. Other factors, such as better diagnosis and coding, may also have contributed to the observed increase in incidence.

Complete surgical resection remains the cornerstone of treatment in paediatric patients with HB; without this, patient prognosis remains grave. The COG recommends a post-surgical staging system based on operative findings. According to the COG, stage I disease refers to a tumour that is completely resected. Microscopic residual tumour after resection is designated as stage II. Stage III is used for tumours with no distant metastasis but positive lymph nodes or unresectable or gross residual tumour. Lastly, stage IV is applied to cases of distant metastasis irrespective of tumour extent.¹⁰ If left untreated, patients with HB may die within < 2 years.^22–25 Over three-quarters of the patient population within this study underwent surgical resection for the presence of an HB. Furthermore, 68.8% of patients presented with resectable disease situated either locally or regionally, and the remainder had either distant metastasis (26.9%) or unstaged disease (4.3%).

Whereas HB was formerly a disease with a dreadful prognosis, children with HB now have a 5-year survival rate of >70% as a result of the advent of treatment regimens comprising both systemic chemotherapy and surgery.¹¹ Because of the different therapeutic strategies used, which involve adjuvant versus neoadjuvant chemotherapy, there is no universal system for selecting patients for treatment. In general, patients with tumours that are completely resected at the time of surgery and favourable tumour histology can be cured with surgery alone.²⁶ Conversely, patients with unfavourable tumour biology, unresectable tumours prior to surgery, tumours that are incompletely resected at the time of surgery or locoregional disease benefit from chemotherapy.^27,28 Nevertheless, additional issues regarding risk group stratification, and the treatment of recurrent or refractory disease remain to be determined and treatment strategies remain under debate.¹⁰ In this study, survival rates at 5, 10 and 20 years post-HB resection were consistently 74%, and were significantly greater than those in patients who did not undergo surgical intervention. Furthermore, girls achieved better survival rates to 10 years than boys. However, there is no scientific evidence of gender-specific factors that might account for the observed differences in survival. Chemotherapy information is not reported in the SEER database.

In the 1980s, 5-year overall survival in children with HB was about 30%.¹¹ Within this surgical series, overall survival rates post-treatment were 63% at 5 years, 61% at 10 years and 59% at 20 years. The highest survival rates were found for males in comparison with females (P = 0.001), the presence of local disease as opposed to regional or distant metastasis (P = 0.001) and surgical extirpation versus no operative management (P = 0.001). Specifically, patients with local disease achieved a survival rate of 80% at 5, 10 and 20 years post-diagnosis, whereas patients with distant disease achieved survival rates close to 40% at all three time-points. Furthermore, survival rates were highest for children diagnosed at < 5 years of age and peaked to 64% in the neonatal group at 5, 10 and 20 years post-diagnosis. Therefore, age is an important clinical indicator of prognosis; adolescent patients had the lowest 5-year survival rate at 20% and no adolescent patients remained alive at 10 years after diagnosis. However, the incidence of HB in subjects aged >10 years is low and only 18 such patients were identified in the SEER database for the study period. In addition to age and surgical resection, multivariate analysis determined that ethnicity and tumour stage are also independent predictors of survival in patients with HB.

There are several limitations inherent within the SEER database. This was a retrospective review of data collected by the National Cancer Institute, the coverage of which is limited to about 26% of the total US population. The study sample size was relatively small and thus there is a potential decrease in power, leading to the possibility of a type II error. The influence of prematurity and the impact of associated anomalies on outcomes cannot be evaluated. Furthermore, assessing patient symptomatology on initial presentation, determining the resolution of such symptoms after treatment and evaluating the use of chemotherapy as a treatment regimen remain beyond the scope of this database. Lastly, given that minority racial and ethnic groups are of special interest to the SEER programme, they may be somewhat over-represented within the database. Nonetheless, the SEER database is considered the standard among cancer registries and its value has been validated in numerous publications pertaining to paediatric cancer outcomes.^29–35

In conclusion, HB is an uncommon liver malignancy seen mostly in children aged < 5years. The dramatic increase in survival can be attributed, in part, to surgical resection of the disease. However, given the increase in disease incidence over time, the evaluation of reliable prognostic factors remains paramount in supporting the prompt diagnosis and optimum management of paediatric patients with HB.

Conflicts of interest

None declared.