Survival of children after liver transplantation for hepatocellular carcinoma

Hepatocellular carcinoma (HCC) in children is rare and accounts for 0.5%-2.0% of all pediatric tumor cases.1-4 Only 0.5%-1.0% of HCC cases occur in patients under the age of 20, with the majority presenting between 10 and 14 years of age. The clinical and scientific perception of HCC focuses therefore on adult patients who account for more than 99% of all HCC cases.5-7 The majority of childhood HCCs worldwide occurs secondary to chronic hepatitis B infection. In highly endemic areas for hepatitis B, the incidence of HCC in children and adolescents ranges from 0.09 to 0.45 cases per million per year and decreases with successful implementation of vaccination programs.8 In countries where hepatitis B prevalence is low, most childhood HCC cases appear to be associated with inherited liver disease.9, 10 Metabolic inherited conditions such as tyrosinemia type I (TTI),11 bile salt export pump (BSEP) deficiency,12 α-1-antitrypsin deficiency (A1ATD),13, 14 or glycogen storage disease (GSD)15 are known risk factors for malignant transformation with or without prior cirrhosis. It has been suggested that HCCs arising in patients with inherited disease form a distinct clinical entity given that the pathophysiology behind these tumors differs considerably from other HCC types.16-18 Evidence to support this hypothesis is, however, scarce. It is still unknown whether HCC in inherited metabolic liver disease is induced via a common specific mechanism (ie, by inflammatory or toxic injury, nuclear factor kappa B activation [TTI], impaired autophagy, and gain-of-function mechanisms [A1ATD]), or via a combination of these mechanisms.18-21

Curative therapy of childhood HCC requires complete surgical resection. In nonresectable tumors, liver transplantation (LT) may be considered.4 Although several case series indicated a favorable outcome of LT in childhood HCC,22-25 there are no studies available comparing survival rates of pediatric and adult HCC patients taking into account the underlying cause of HCC. A systematic assessment of the outcome of children who underwent LT for HCC was reported in 2006 for the United Network for Organ Sharing (UNOS) database26 and updated to an extended cohort27 in 2010 showing a 5-year patient survival rate of 53.5%. In this study, however, no information on underlying disease was reported so that HCCs based on inherited liver disease and acquired noninherited liver disease could not be differentiated.

The aim of our study was to use the advantages of the large European Liver Transplant Registry (ELTR) to analyze if there are differences in survival after LT for HCC between children and adults on the one hand and between patients with underlying inherited and noninherited liver disease on the other hand. We hypothesized that childhood HCC arising from underlying inherited disease may have superior graft and patient survival after LT when compared with HCC in children with noninherited liver disease or to HCC in adults. We further hypothesized that LT for HCC arising from underlying inherited metabolic disease yields a better outcome than in sporadic HCC regardless of age.

Patients and Methods

Results

In total, 570 patients undergoing LT for HCC with a number of associated liver diseases were included (Table 1). Type of associated liver disease differed considerably by age in those HCC patients with an associated inherited liver disease. Patients suffering from BSEP received liver transplants in childhood at a median age of 2.4 years, in contrast to patients with alpha-1-antitrypsin deficiency who were adults aged 59.4 (median) when undergoing first liver transplantation (Fig. 1).

COMPARISON PINHCC VERSUS PHCC

Pediatric patients with underlying inherited liver disease (PinHCC, n = 38) were younger (P < 0.001), more likely to be urgently transplanted (P = 0.007), and more likely to receive a partial graft (P = 0.01) than children without an underlying inherited cause (pHCC, n = 137; Table 2). Overall 5-year patient (57.6%; 95% confidence interval [CI], 47.9%-65.1%) and graft survival rates (56.3%; 95% CI, 39.4%-56.3%) of pediatric HCC patients were very similar to those reported by the US-based UNOS registry on pediatric HCCs (n = 58; Fig. 2; Table 3). In the multivariate survival analyses, we found evidence that PinHCC showed a better longterm patient survival than pHCC (hazard ratio [HR], 0.29; 95% CI, 0.10-0.90; P = 0.03, adjusted for sex, year of transplantation, and using a shared frailty term for transplantation center). Age at LT showed almost distinct distributions between both groups (median ages 7 versus 11 years, with the inherited HCC patients being clearly younger). It was therefore not possible to fully differentiate the effect of age and underlying disease. Survival rates increased with increasing age in pHCC patients (HR per year, 1.11; 95% CI, 1.04-1.18; P = 0.001), whereas there was no age effect in PinHCC patients (HR per year, 0.97; 95% CI, 0.81-1.16; P = 0.71). When adjusting the multivariate Cox regression model for age, the protective effect of underlying inheritable disease became slightly smaller (HR, 0.45; 95% CI, 0.16-1.31; P = 0.15). Effect estimates did not change considerably in the sensitivity analysis restricting pHCC patients to those with a defined secondary diagnosis (HR, 0.54; 95% CI, 0.16-1.83; P = 0.32). Results for graft survival were generally comparable with the ones for patient survival (Supporting Information).

Table 2. Baseline Characteristics of Patients Undergoing LT Because of HCC Stratified by Age at Transplantation and Underlying Disease

	Children (n = 175)		Adults (n = 395)		P Values
	Noninherited (n = 137)	Inherited (n = 38)	Inherited (n = 79)	Noninherited (n = 316)	Noninherited Versus Inherited Children	Inherited Children Versus Inherited Adults	Noninherited Versus Inherited Adults
Age in yearsa	11 (8-15)	7 (3-11)	52 (41-61)	53 (43-62)	<0.001	<0.001	0.36
Sexb					0.79	0.09	0.99
Male	76 (55.5)	22 (57.9)	58 (73.4)	232 (73.4)
Female	61 (44.5)	16 (42.1)	21 (26.6)	84 (26.6)
Type of donorb, c					0.34	0.11	0.67
Deceased	117 (85.4)	30 (79.0)	70 (89.7)	289 (91.5)
Living	20 (14.6)	8 (21.0)	8 (10.3)	27 (8.5)
Urgency (yes)b, c	5 (4.1)	6 (17.7)	4 (5.9)	7 (2.2)	0.007	0.08	0.09
Days on waiting lista, c	17 (5-51)	33 (11-78)	123 (32-230)	158 (48-288)	0.22	<0.001	0.27
Number of retransplantationsb					0.86	0.45	0.50
0	122 (89.0)	34 (89.5)	74 (93.7)	288 (91.1)
1	13 (9.5)	4 (10.5)	5 (6.3)	28 (8.9)
2	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
3	2 (1.5)	0 (0.0)	0 (0.0)	0 (0.0)
Encephalopathyb, c					0.54	0.28	0.06
Grade III-IV	0 (0.0)	1 (5.0)	3 (6.7)	2 (1.2)
Grade I-II	1 (2.3)	0 (0.0)	5 (11.1)	31 (19.1)
No	42 (97.7)	19 (95.0)	37 (82.2)	129 (79.6)
Ascitesb, c					0.38	0.048	0.02
Refractory	1 (2.9)	1 (5.0)	5 (12.5)	8 (4.9)
Controlled	4 (11.8)	0 (0.0)	8 (20.0)	69 (41.8)
No	29 (85.3)	19 (95.0)	27 (67.5)	88 (53.3)
Ischemia timea, c	465 (330-613)	450 (213-570)	510 (348-625)	466 (342-600)	0.28	0.08	0.30
Preservation liquidb, c					0.06	>0.99	0.68
UW	91 (75.8)	19 (59.4)	41 (59.4)	203 (64.2)
Other	29 (24.2)	13 (40.6)	28 (40.6)	109 (35.8)
Bypassb, c					0.06	<0.001	<0.001
Extracorporal	10 (10.2)	0 (0.0)	16 (32.7)	102 (35.3)
Vena cava	16 (16.3)	2 (6.7)	10 (20.4)	125 (43.9)
None	72 (73.5)	28 (93.3)	23 (46.9)	58 (20.3)
Type of graftb, c					0.01	<0.001	0.53
Full size	72 (54.9)	12 (31.6)	73 (92.4)	284 (89.9)
Partial	59 (45.1)	26 (68.4)	6 (7.6)	32 (10.1)

• NOTE: Data are given as n (%) and median (IQR).
• ^a Wilcoxon rank sum test.
• ^b Fisher's exact test or chi-square test as appropriate.
• ^c Number of missing values for type of donor (n = 1), urgency (n = 30), days on waiting list (n = 438), encephalopathy (n = 292), ascites (n = 311), ischemia time (n = 35), bypass (n = 108), and type of graft (n = 6).

Table 3. Comparison of Baseline Characteristics and Survival Rates of Pediatric HCC Patients From ELTR With UNOS Data

	ELTR (n = 175)	UNOS (n = 58)
Sex, male, n (%)	98 (56.0)	30 (51.7)
Age at transplantation, yearsa	10.2 (6.0-14.3)	10.5
Patient survival, %b
1-year survival	81.2 (74.2-86.5)	85.2
2-year survival	68.3 (60.5-75.8)	68.8
5-year survival	57.6 (47.9-65.1)	53.5
Graft survival, %a
1-year survival	73.6 (66.0-79.7)	83.7
2-year survival	58.5 (50.0-66.0)	60.6
5-year survival	56.3 (39.4-56.3)	42.8

• NOTE: See Guiteau et al.27 (2010).
• ^a For ELTR median and IQR are reported; for UNOS only data for the median were available.
• ^b For ELTR survival rates are presented together with 95% CIs; this was not possible for UNOS data as no uncertainty measures were reported in the original publication and no information on lost to follow-up was given.

COMPARISON PINHCC VERSUS ADINHCC

In the second step of our analysis, we compared children (PinHCCs, n = 38) and adults (AdinHCCs, n = 79) with HCC and associated inherited liver disease. Baseline characteristics revealed that adult patients (AdinHCC) were less likely to receive a partial graft (P < 0.001; Table 2) and appeared to have more advanced liver disease. They were more often diagnosed with encephalopathy and ascites (P = 0.048), were more likely to be treated with a bypass (P < 0.001), and were longer on the waiting list (P < 0.001; Table 2). Although there was almost no difference in patient survival in the first 2 years after LT between both groups (HR, 1.22; 95% CI, 0.55-2.70; P = 0.63), patients with PinHCCs showed a better longterm survival (HR, 0.27; 95% CI, 0.06-1.25; P = 0.09, piecewise Cox regression model; Fig. 3). Graft survival analyses results were similar to those for patient survival (Supporting Information).

COMPARISON ADINHCC VERSUS AHCC

In the last step of our analysis, we compared AdinHCC patients (n = 79) with their matched aHCC patients (n = 316) with defined other underlying diseases (Table 1). AdinHCC patients were less likely to suffer from ascites (P = 0.02) and less likely to be treated with a bypass (P < 0.001). In contrast, there was almost no difference in patient survival (HR, 1.05; 95% CI, 0.66-1.66; P = 0.84, aHCC as reference group; Fig. 4) and graft survival (HR, 1.04; 95% CI, 0.68-1.15; P = 0.85; Supporting Information) in the multivariate survival analyses.

Recurrence of HCC did not account for graft loss leading to retransplantation in any of the 4 groups (Supporting Information). The incidence of death due to HCC recurrence was higher in children (38.0 deaths [95% CI, 26.4-54.7] per 1000 patient years) than in adults (24.2 deaths [95% CI, 17.9-32.7] per 1000 patient years). No death due to HCC recurrence occurred in patients with underlying inherited liver disease.

Discussion

In this study, we investigated the hypothesis that children transplanted for HCC with associated inherited liver disease show better LT outcomes than other HCC patients. For this purpose, we compared these children with 2 other HCC groups, children with sporadic HCC, not based on inherited liver disease, and adults with HCC and inherited liver disease. Contrary to previous expectations, most childhood HCCs registered in ELTR were not associated with inherited liver disease. We showed that children with HCC based on inherited disease had a better longterm survival than children with a noninherited HCC or adult patients with an inherited HCC. Although not all comparisons were significant on a 5% level due to the small sample size in the groups of pediatric HCCs with associated inherited liver disease, the hazard of death was 3 to 4 times smaller in pediatric HCC patients with inherited liver disease than in their comparison groups.

One explanation for the better outcome of pediatric HCC patients with inherited liver disease might be that they are generally followed up in regular intervals, which may lead to timely HCC detection and earlier LT compared with sporadic HCC patients. This explanation is supported by the finding that adult patients (and especially those with no underlying inherited disease) were generally sicker, had more ascites at transplantation, and had longer waiting times before LT. However, these findings could also be explained by longer waiting times for adult LT due to a higher scarcity of organs for adults. Our results suggest that younger age is a prognostic factor for better survival rates after LT for HCC. However, some of the effect of inherited liver disease in children remained apparent (HR, 0.45) after adjusting for age so that inherited liver disease can be seen as an additional prognostic factor for better survival rates in pediatric HCC. A survival advantage based on associated inherited disease was not seen in adults, where HCC patients with and without inherited liver disease showed similar survival rates. This might be attributable to the different types of inherited liver diseases, which are associated with HCC in children and in adults (Fig. 1).

This is the first study which assessed comprehensively the competing effect of pediatric HCC and underlying inherited liver disease on patient and graft survival after LT. Crude overall survival rates were thereby in line with the largest previously published patient cohort from the UNOS database.27 Data on pediatric HCC patients undergoing LT are otherwise sparse.

Because of small numbers of patients with childhood HCC, there is little evidence regarding optimal treatment for this group of patients. Children with HCC are therefore evaluated for LT extrapolating experience from adult patients by applying the conventional Milan criteria. The criteria have been modified repeatedly, and these and other more liberal criteria like the University of California, San Francisco criteria have been a matter of ongoing debate.30, 31 The ELTR data presented here provide the most comprehensive cohort of children who underwent liver transplantation for HCC. Our results suggest that prognosis of a child with underlying inherited liver disease is generally better than anticipated.

However, our analyses and the underlying data set have several limitations. One major limitation is the lack of characterization of the individual HCC with respect to the histological and radiological stages and tumor volume. In the majority of patients (more than 85%), there were no data on tumor stage or grade for the various classifications such as conventional Milan criteria, Barcelona score, or the pediatric pretreatment extension score Pretreatment Extent of Disease. Moreover, no data on individual cirrhosis status or chemotherapy for HCC were available, limiting the interpretability of study results. Another drawback of our analysis originates from the nature of the source data. Several selection biases may apply to registry data, ie, the methods of diagnosing (or labeling) a patient in the source data, the methods of transplantation eligibility, selection on who was entered in the registry, and how complete follow-up was in the participating centers of ELTR. We tried to account for this by including the transplantation center as a shared frailty term in our multivariate models, so that our models accounted for heterogeneities between LT centers. However, given the high proportion of patients lost to follow-up after the first 5-10 years (especially in pediatric patients who were transferred to adult care), longterm survival estimates need to be interpreted with caution. Moreover, the group of patients with pediatric HCC but no underlying inherited disease was poorly characterized, and it cannot be excluded that some of these patients suffered from undocumented inherited liver disease. We accounted for this by performing a sensitivity analysis including only those patients with a confirmed underlying noninherited cause for HCC and sufficient follow-up information. Results remained virtually unchanged reassuring us that the original analysis showed valid results. The potential longterm benefits observed for pediatric HCC patients with underlying inherited disease in comparison to their adult peers might at least partially be due to the relative survival advantages of children when compared with adults. We did not account for this in the analysis of this study (eg, by using relative survival) because we were explicitly interested in the true survival rates of the respective patient groups.

In summary, we were able to analyze the largest ever reported cohort of pediatric HCC patients treated with LT. We have shown that pediatric HCC patients with underlying inherited liver disease have a survival advantage when compared with pediatric HCC patients with no underlying inherited disease and to adult HCC patients. We suggest that our results are taken into consideration for future transplant organ allocation. However, studies with more detailed information on tumor characteristics, waiting times, and disease severity are necessary to fully interpret our findings.