Pretreatment prognostic factors and treatment results in children with hepatoblastoma

A Report from the German Cooperative Pediatric Liver Tumor Study HB 94

Authors


Abstract

BACKGROUND

In the past 20 years, a dramatic improvement in the prognosis of patients with hepatoblastoma (HB) has been achieved by combining surgery with chemotherapy in several national and international trials. A worldwide, unsolved problem remains the treatment of patients with advanced or metastatic HB.

METHODS

The German Cooperative Pediatric Liver Tumor Study HB 94 was a prospective, multicenter, single-arm study. The study ran from January 1994 to December 1998. The protocol assessed the efficiency of chemotherapy consisting of cisplatin, ifosfamide, and doxorubicin (CDDP/IFO/DOXO) and/or etoposide and carboplatin (VP16/CARBO). The prognostic significance of the surgical strategy, pretreatment factors, and tumor characteristics for disease free survival (DFS) were analyzed.

RESULTS

Sixty-nine children with HB were treated in the HB 94 study. The median follow-up of survivors was 58 months (range, 32–93 months). Fifty-three of 69 patients (77%) remained alive, and 16 of 69 patients (23%) died. Long-term DFS was as follows: 26 of 27 patients had Stage I HB, 3 of 3 patients had Stage II HB, 19 of 25 patients had Stage III HB, and 5 of 14 patients had Stage IV. A complete resection of the primary tumor was achieved in 54 of 63 patients (86%). Six children (8%) had no surgical treatment. Twenty-two tumors were resected primarily, and 41 children underwent surgery after initial chemotherapy. Two children underwent liver transplantation. There was no perioperative death. Forty-eight children received primary chemotherapy with CDDP/IFO/DOXO. Forty-one of 48 children achieved partial remission after CDDP/IFO/DOXO. Eighteen children with advanced or recurrent HB underwent VP16/CARBO chemotherapy, with a response achieved by 12 children. The relevant pretreatment prognostic factors were growth pattern of the liver tumor (P = 0.0135), vascular tumor invasion (P = 0.0039), occurrence of distant metastases (P = 0.0001), initial α-fetoprotein level (P = 0.0034), and surgical radicality (P < 0.0001).

CONCLUSIONS

The current results underline the necessity of preoperative chemotherapy in all children with HB. Complete tumor resection is one of the main prognostic factors. Cancer 2002;95:172–82. © 2002 American Cancer Society.

DOI 10.1002/cncr.10632

Hepatoblastoma (HB) is the most frequent epithelial liver tumor in infants and toddlers.1, 2 At the time of diagnosis, a total extirpation of the tumor is possible in < 50–60% of pediatric patients with hepatic malignancies.3 In the past 20 years, a dramatic improvement in the prognosis of these patients has been achieved by combining surgery with chemotherapy in several national and international trials.4, 5 The treatment modalities of the prospective studies differed with respect to primary surgery versus delayed surgery after chemotherapy, the cytotoxic drugs used, and the number of preoperative and/or postoperative courses of chemotherapy. Although chemotherapy sometimes may shrink a nonresectable tumor to a resectable size, cytotoxic drugs alone cannot eradicate HB.6, 7 Clinical observations in the prior German Cooperative Pediatric Liver Tumor Study (HB 89) showed the development of drug resistance in patients with HB after they received four or five courses of ifosfamide, cisplatin, and doxorubicin (CDDP/IFO/DOXO). Preliminary results of basic research also have suggested different mechanisms of drug resistance.8, 9 Therefore, some new treatment modalities have been evaluated in the HB 94 study. First, the intensity of postoperative chemotherapy in patients with Stage I HB or Stage II HB was reduced to two or three courses of chemotherapy, respectively. Second, a maximum of four courses of conventional CDDP/IFO/DOXO chemotherapy was allowed, and a new treatment arm for patients with advanced or recurrent HB who received etoposide and carboplatin (VP16/CARBO) was evaluated. Third, children age 6 months to 3 years with significantly elevated serum α-fetoprotein levels and/or tumor in both liver lobes, with or without distant metastases, were treated without a prior histologic diagnosis.10 Furthermore, possible clinical and pathologic prognostic factors were analyzed and compared with the experience of other groups studying patients with HB. This report focuses on the treatment results and pretreatment prognostic factors of children with HB.

MATERIALS AND METHODS

Principles of Study Protocols and Patients

The German Cooperative Pediatric Liver Tumor Study HB 94 commenced as a prospective study in January 1994 and closed in December 1998. One hundred eight children with different liver tumors were registered (Table 1). The study protocols were approved by the Ethical Committee of the Medical School, Hannover and the Executive Board of the German Society for Pediatric Oncology and Hematology.

Table 1. Tumor Entities in the German Cooperative Pediatric Liver Tumor Study HB 94
EntityNo.
Hepatoblastoma69
Hepatocellular carcinoma26
Hemangioendothelioma7
Focal nodular hyperplasia4
Hamartoma2
Total108

Sixty-nine children had HB, and all were treated according to the study protocol. The protocol prescribed a primary tumor resection only in patients with tumors that clearly were confined to one liver lobe (Stage I), whereas patients with tumors that involved both liver lobes or patients with distant metastases (Stage IV) underwent biopsy only (Stage III). Only children age 6 months to 3 years with significantly elevated serum α-fetoprotein levels and/or tumor in both liver lobes, with or without distant metastases, were treated without a prior histologic diagnosis (Fig. 1). Depending on the tumor stage, patients with HB were treated with chemotherapy consisting of CDDP (20 mg/m2 × 5 on Days 4–8), IFO (500 mg/m2 bolus on Day 1 and 3000 mg/m2 over 72 hours on Days 1–3), and DOXO (60 mg/m2 over 48 hours on Days 9 and 10). The children received two cycles of CDDP/IFO/DOXO if they had Stage I HB and received three cycles if they had Stage II HB postoperatively. All of the children described above who were without a prior histologic diagnosis received two cycles CDDP/IFO/DOXO preoperatively and, after undergoing complete tumor resection (Stage I), they received one course postoperatively. In children with Stage III–IV HB, two to three cycles were administered preoperatively. If there was a tumor recurrence or advanced HB, then the children were given CARBO (800 mg/m2 over 96 hours on Days 1–4) and VP16 (400 mg/m2 over 96 hours on Days 1–4). A total of 18 children received addition treatment with CARBO/VP16.

Figure 1.

Study design of the German Pediatric Liver Tumor Study HB 94. IFO/CDDP/DOXO: ifosfamide, cisplatin, and doxorubicin; VP16/CARBO: etoposide and carboplatin; PR: partial response; SD/PD: stable disease/progressive disease.

Staging Procedure

The postsurgical staging system was used and the following criteria were applied, according to the results of the initial clinical evaluation and surgery: Stage I, completely resected tumors; Stage II, microscopic residual tumor; Stage III, macroscopic residual tumor after resection or tumor biopsy; and Stage IV, distant metastases.

Treatment Principles in Surgery

The treatment protocol allowed atypical (synonymous wedge) resection and anatomic tumor resection. Patients with advanced stage HB, with invasion of liver veins and/or tumor masses in the inferior cava vein, underwent resection under extracorporal bypass and hypothermia. Patients with multifocal HB without lung metastases were eligible to undergo transplantation after two cycles of CDDP/IFO/DOXO.

Evaluation of Clinical Data and Determination of Tumor Response

Abdominal ultrasound, computed tomography (CT) scan and/or magnetic resonance imaging, and chest X-ray/CT scan were used to evaluate the initial tumor size and extension and tumor response after chemotherapy. Tumor volume was calculated using the formula 0.5 × a (length) × b (width) × c (height) (cm3). Serum α-fetoprotein levels were measured before chemotherapy and on Day 21 or at the end of chemotherapy and also were measured regularly during follow-up.

Tumor remission was evaluated after patients received two courses of CDDP/IFO/DOXO and/or VP16/CARBO. Complete remission was defined as complete disappearance of the tumor and no elevated serum α-fetoprotein levels. A partial response was defined as tumor volume shrinkage and a reduction > 50% in serum α-fetoprotein levels. Stable disease was defined as tumor regression and a reduction < 50% in α-fetoprotein levels. Progressive disease was defined as enlargement of the tumor mass, evidence of new metastases, or an increase in serum α-fetoprotein levels.

Macroscopic or microscopic residual tumor after last resection (second-look or third-look surgery) was documented histologically. Clinical involution of lung metastases without histologic confirmation was registered as potential microscopic residue despite complete resection of the liver tumor. After the completion of treatment, all children were followed with physical examinations, ultrasound scans, and monitoring of the α-fetoprotein levels monthly during the first year, every 3 months during the second and third years, and every 6 months during the fourth and fifth years.

Histologic Investigation

All material obtained at biopsy and/or resection was reviewed centrally by one pathologist (D.H.). The tumors were examined for histologic type, as proposed by Ishak and Glunz, and were classified as purely epithelial or mixed HB and, according to the epithelial differentiation, as the fetal subtype or the embryonal subtype.3

Toxicity of Chemotherapy and Dose Modifications

Toxicity was assessed after each course of chemotherapy and was graded according to the World Health Organization toxicity criteria. All children had a venous access device (Port-A-Cath® or Hickman/Broviac catheter®) implanted for the administration of drugs. Cardiac function was checked by electrocardiogram and echocardiogram; liver function was determined by aspartate aminotransferase, alanine aminotransferase, choline esterase, alkaline phosphatase, and γ-glucoronosyltransferase tests; and renal function was determined by measuring creatinine, urea, serum electrolytes, and creatinine clearance. Also, all children had audiograms. The doses recommenced for children age < 1 year were calculated per kg based on a formula in which 1 m2 of surface area corresponded to a body weight of approximately of 30 kg. Other guidelines for treatment modification, dosage reductions up to 80%, or therapy break were leucopenia < 2000/μL, thrombopenia < 75,000/μL, and creatinine clearance < 75 mL per minute/1.73 m2. Treatment with granulocyte-colony stimulating factor (GCSF) was not recommended explicitly in the study protocol.

Data Collection and Statistical Analysis

The results of the study are expressed in terms of response to CDDP/IFO/DOXO or VP16/CARBO chemotherapy, the complete surgical resection rate, the disease free survival (DFS), event free survival (EFS), prognostic factors, and toxicity of chemotherapy. DFS was measured from the start of treatment to the date of death or last follow-up, taking into account that all surviving patients eventually were disease free. At the time of last follow-up, all patients had no residual tumor. EFS was defined as the time from the date of diagnosis to the date of VP16/CARBO treatment because of the failure of the CDDP/IFO/DOXO therapy, the date of recurrence, the date of death, or the date of last follow-up. The Kaplan-Meier method was used to estimate survival curves, and the log-rank test was used to compare them. The prognostic factors are listed in Table 4 and also were analyzed using the log-rank test. Significance was assumed at the P < 0.05 level.

Table 4. Disease Free Survival Analysis of Single Prognostic Factors in Patients with Hepatoblastoma (n = 69 patients)
ParameterP value
Tumor size (< 1000 m3 or >1000 m3)0.5013
One or two liver lobes0.5901
Histologic subtype0.672
Growth pattern in the liver0.0135
Vascular invasion0.0039
Distant metastasis0.0001
Surgical radicality0.0001
Initial α-fetoprotein0.0034
Response to chemotherapy0.003

RESULTS

Epidemiologic Data

One hundred eight children with liver tumors were registered between 1994 and 1998. Sixty-nine children had an HB (Table 1). The median age at diagnosis for children with HB was 16 months (range, 28 days to 16 years). Forty-nine patients were male, and 20 patients were female. In two of the children, HB was associated with Beckwith-Wiedemann syndrome. The median follow-up of survivors was 58 months (range, 32–93 months). In the group of patients who received primary chemotherapy without a histologic diagnosis, the diagnosis of HB was confirmed in all children.

Overall Therapy Results

DFS at the end of follow-up was 96% (26 of 27 patients) for children with Stage I HB, 100% (3 of 3 patients) for children with Stage II HB, 76% (19 of 25 patients) for children with Stage III HB, and 36% (5 of 14 patients) for children with Stage IV HB (Table 2, Fig. 2). EFS was 89% (24 of 27 patients) for children with Stage I HB, 100% (3 of 3 patients) for children with Stage II HB, 68% (17 of 25 patients) for children with Stage III HB, and 21% (3 of 14 patients) for children with Stage IV HB. These differences were significant in log-rank analyses (P < 0.0001).

Table 2. Therapy Results of Hepatoblastoma in the German Cooperative Pediatric Liver Tumor Study HB 94 (n = 69 patients)
StageDisease freeDeadTotal
I26127
II33
III19625
IV5914
Total531669
Figure 2.

Disease free survival (A) and event free survival (B) of children with hepatoblastoma (n = 69 patients.).

Fifty-three children (77%) are alive, and 16 children (23%) have died. Twelve patients had progressive disease with lung metastases or local recurrence. Two children died due to side effects of CDDP/IFO/DOXO chemotherapy. Two patients in this group died due to other causes, e.g., bronchopulmonary dysplasia and gastrointestinal bleeding/esophageal varicosis. In the study, seven children with HB were born prematurely, and one of them died due to bronchopulmonary dysplasia.

Surgery

The surgical procedures were performed by pediatric surgeons and transplantation surgeons in 31 different institutions. Four or more liver resections were performed in only three of these hospitals. Sixty-three (91%) from 69 HBs were resectable. In six children with metastatic disease, liver tumor resection was impossible. Complete resection of HB after second-look or third-look surgery was achieved in 54 of 63 patients (86%).

In 65 first-look operations, 24 tumor biopsies and 41 liver resections were performed. Twenty-seven of 41 liver resections as first-look operations were without macroscopic or microscopic residual tumor (Stage I). Fourteen of of these 27 children underwent tumor resection without primary chemotherapy, and 13 children underwent surgery after receiving two courses of chemotherapy without a prior histologic diagnosis due to age, elevated serum α-fetoprotein level, and tumor extension (Fig. 3). Thirty-five second-look operations and 4 third-look operations followed. There was no surgical mortality.

Figure 3.

Flow diagram indicating results of surgical therapy after the first operation for hepatoblastoma (n = 65 patients).

Twenty-one children with HB underwent liver resection without primary chemotherapy and received adjuvant chemotherapy with CDDP/IFO/DOXO. There were 7 patients (33%) with microscopic or macroscopic residual tumor. In contrast, 40 of 69 patients underwent resection after receiving primary chemotherapy and, in only 7 patients (18%), an incomplete tumor resection was achieved after first-look operation. Two of 63 patients with resected tumors (3%) underwent primary transplantation due to multifocal spreading of the HB. None of these patients had lung metastases. One child underwent orthotopic liver transplantation, and one patient underwent a split-liver transplantation that was donated by the mother. Both children who underwent transplantation survived.

Of 63 liver resections, 21 were atypical liver resections, 40 were anatomic liver resections (22 hemihepatectomies and 18 trisegmentectomies), and 2 were liver transplantations. Two right trisegmentectomies were performed under extracorporal blood circulation and hypothermia. In both of these patients, the remnant liver veins were reconstructed with pericard after resection of tumor infiltration, and the retrohepatic cava vein was reconstructed with Gore-Tex®. Both children recovered well from liver resection and showed no signs of postperfusion syndrome. One child survived for 3 years after surgery; however, unfortunately, the other child died 10 weeks later due to sepsis after receiving adjuvant chemotherapy.

The correlation between the radicality of tumor resection and the probability of survival was significant (P < 0.0001). The DFS rate in children with tumors that were resected completely was 95%. Children with microscopic residual tumor had a DFS rate of 87%, and children with macroscopic residual tumor had a DFS rate of 25%.

Chemotherapy and Toxicity

One hundred eighty-five cycles of CDDP/IFO/DOXO were administered to 68 children (Table 38, Fig. 4). Forty-eight children received primary chemotherapy, and 41 children (85%) responded after receiving two courses of chemotherapy. Seven children (14%) were nonresponders, and they all died (P = 0.0003). Acute Grade 3 and 4 toxicity was reported in 39 of 68 children. Two children died from sepsis due to severe aplasia of the bone marrow.

Table 3. Acute Grade 3–4 Toxicity of Chemotherapy with Cisplatin, Ifosfamide, and Doxorubicin (n = 185 cycles) or Etoposide and Carboplatin (n = 34 cycles)
VariableCourses (%)
CDDP/IFO/DOXOVP16/CARBO
  • CDDP/IFO/DOXO: cisplatis, ifosfamide, and doxorubicin; VPIB/CARBO. etoposide and carboplatin.

  • a

    Sepsis with necessity of intravenous application of antibiotics and partially hypotension.

Leucocyte < 1000/μL55 (∼30%)19 (∼56)
Platelet count < 30000/μL57 (∼31%)23 (∼68)
Clearance < 69 mL/minute 1.73m24 (∼2%)2 (∼6)
Severe infectionsa18 (∼10%)1 (∼3)
Summary of courses38 of 68 (∼56)11 of 18 (∼61)
Figure 4.

Disease free survival of children in relation to their response to chemotherapy with either (IFO/CDDP/DOXO) (A) or etoposide and carboplatin ifosfamide, cisplatin, and doxorubicin (VP16/CARBO) (B).

Thirty-four cycles of VP16/CARBO were given to 18 children with metastatic HB or advanced HB. Twelve children responded, and, in one child with Stage IV HB, a complete disappearance of lung metastases was registered. Six patients were nonresponders: Five of them had an unfavorable histology, such as undifferentiated or embryonal HB with anaplastic components. All children who did not have a tumor response died (P = 0.0013). There was a minor difference in response to CARBO/VP16 in patients who had metastatic HB compared with patients who had advanced HB. The children with metastatic HB (n = 7 patients) had a partial response rate of 50%, and the children with advanced HB (n = 11 patients) had a response rate of 66%. Acute Grade 3 and 4 toxicity was observed in 11 of 18 children.

In total, 22 patients from the CDDP/IFO/DOXO group and 6 patients from the VP16/CARBO group had a planned dose reduction of 80% during their chemotherapy. The reasons for these reductions were mainly severe leucopenia (seven patients), age < 1 year (three patients), febrile neutropenia (five patients), creatinine clearance < 50 mL per minute/1.73 m2 (three patients), and cardiotoxicity after three courses of CDDP/IFO/DOXO with ejection fraction < 50% and shortening fraction < 25% (one patient). In nine patients, the reasons were unknown. GCSF was used successfully in four children for the treatment of myelosuppressive complications at a dosage of 5 μg/kg per day subcutaneously.

Histology

Tumor samples were available for histologic examination from 69 resected or biopsied HBs. Twenty-eight patients (41%) had the pure epithelial type of HB, and 39 patients (56%) had a mixed epithelial/mesenchymal HB. Ten of these children (14%) had a pure fetal HB. Two children (3%) had teratoid/undifferentiated HB without elevated serum α-fetoprotein level, and no response to chemotherapy: All three children died. Kaplan-Meier analysis revealed no difference between patients with an epithelial HB or a mixed HB (P = 0.672). Compared with patients who had untreated HB, the histologic sections from HBs after patients received primary chemotherapy showed signs of tumor regression, e.g., tumor bleeding, necrosis, and cystic or fibrotic transformation. In patients with mixed HB, osteoid formation was generally more abundant after pretreatment compared with the formation found in patients who underwent primary surgery. There was also a tendency toward more predominant fetal tissue in patients with embryonal/fetal HB after they received chemotherapy.

Prognostic Factors

Tumor size was measured in 68 patients before treatment. Twenty-two children had a tumor size > 1000 cm3 and a DFS rate of 77%, compared with 46 children who had a tumor size < 1000 cm3 with a DFS rate of 85% (P = 0.5013). There also was no difference in the DFS rate between children who had HB that involved one liver lobe or two liver lobes (P = 0.5901). The growth pattern of the liver tumor significantly influenced the DFS of these children (P = 0.0135). An exact analysis was possible in 54 operated children: Twenty-three children had only 1 tumor node in the liver (DFS rate, 95%), 11 children had 2 tumor nodes in the liver (DFS rate, 100%), and 20 children had a multifocally disseminated tumor (DFS rate, 70%). The vascular tumor invasion of large vessels and/or the microscopic tumor infiltration were detected in 14 of 69 patients with a DFS rate of 55%, compared with a DFS rate of 86% for patients without tumor invasion (P = 0.0039). The occurrence of distant metastases had a highly significant prognostic impact, with a DFS rate of 36% in 14 children with metastasis and a DFS rate of 92% in 55 children without metastases (P < 0.0001).

Pretreatment α-fetoprotein serum levels were recorded in 68 patients. Seven children had an α-fetoprotein level < 100 ng/mL, and 42% survived disease free. In 12 children, the initial α-fetoprotein level ranged from 100 ng/mL to 10,000 ng/mL, with a DFS rate of 83%. Forty-six children had an α-fetoprotein level between 10,000 ng/mL and 1,000,0000 ng/mL, with an 89% DFS rate, and, in only 3 patients, the α-fetoprotein was > 1,000,000 ng/mL with a DFS rate of 66%. These differences were significant (P = 0.0034). The differences in the outcome of the children with initial α-fetoprotein levels < 100 ng/mL (n = 7 patients; DFS rate, 43%) and the other children (n = 61 patients; DFS rate, 87%) was significant (P = 0.0005).

Tumor Recurrence

Fourteen children had 26 recurrent tumors. Nine tumors recurred in the liver, 12 tumors recurred in the lungs, and 5 tumors recurred in other organs. Two tumors were located in the brain, one tumor was located in the spinal cord with ascending paralysis, one tumor was located in the spleen, and one tumor was located in the adrenal gland.

Eight of 14 children (57%) with recurrent tumors survived. All children with both local recurrence in the liver and lung metastases (3 of 14 patients) and children with metastases in other organs (3 of 14 patients) died.

Seven children with recurrent tumors received chemotherapy with VP16/CARBO. Three of seven children (42%) responded to chemotherapy. Four children were nonresponders and died.

However, all surviving patients underwent complete resection of their lung metastasis or local recurrence. Ten children underwent 15 surgeries for recurrent tumors, and 6 children benefited from the radical surgical approach. In all tumor specimens from patients with recurrent tumors, vital tumor cells were found on histologic examination.

DISCUSSION

The prognosis of children with HB has improved significantly during the last 20 years. Due to standardized chemotherapy in several multicenter, cooperative trials, the tumor size was reduced successfully, and, in many patients, complete tumor resection became possible.5, 11–13 However, all multicenter HB trials have shown clearly that these tumors cannot be eliminated by chemotherapy alone; complete surgical resectability remains the most important prognostic factor. These clinical findings were underlined by an analysis of experimental data, e.g., the efficiency of chemotherapy in several cell lines and animal models of HB was analyzed.14–16 However, a worldwide unsolved problem remains regarding the treatment of patients with advanced or metastatic HB.9, 17–19

Comparisons of the treatment results from the International Society of Pediatric Oncology (SIOP) study (SIOPEL), United States-Intergroup, and Japanese studies are difficult, because different staging systems were used. These ranged from the PRETEXT-system postsurgical staging system to the TNM classification. The overall survival rate for children was approximately 75% in all studies, which is equal to the overall survival of 77% in our study after a median follow-up of 56 months.9, 18, 20–22

There is a tendency toward better treatment results in the HB 94 study compared with the prior HB 89 study. In particular, the DFS rate increased in patients with Stage II HB (100%) and Stave IV HB (36%) compared with the DFS rates in the HB 89 study (50% and 29%, respectively). A possible explanation may be an increase in complete tumor resections from 80% to 86% and new chemotherapy regimes for patients advanced or recurrent HB with VP16/CARBO.23

Like the prior HB 89 study, complete tumor resection and the occurrence of lung metastases were the two most powerful prognostic factors. A complete tumor resection rate of 86% is acceptable and compares favorably with the tumor resection rate of 77% in the SIOPEL group that was achieved after intensive preoperative chemotherapy with PLADO (CDDP; DOXO).8, 24, 25 However, the resection rate should be improved further. New radical surgical approaches, such as tumor resection under extracorporal blood circulation with hypothermia or extended atypical left hepatectomy, as described by Superina et al., are possible answers for specific patients.26–28 Conversely, liver transplantation provides a realistic chance for survival in children with multifocal HB. Our two children who underwent transplantation, in addition to two such children in HB 89, who had multifocal HBs but did not have distant metastases, survived. Other groups have performed liver transplantation successfully after the complete disappearance of lung metastases during chemotherapy.22, 29–31

Black, Feusner et al., and Oliveira et al. proposed the aggressive excision of pulmonary metastases in children with hepatic tumors.17, 32, 33 Fourteen of our children with recurrent tumors or metastases also benefited from a radical surgical approach.

Additional factors that were of prognostic relevance for survival in our study were focality of the tumor in the liver, vascular invasion, the response to chemotherapy, and the initial level of α-fetoprotein.25, 32, 34, 35 In an analysis of the SIOPEL 1 study and in HB 89, the growth pattern of the tumor was significant for the outcome of children with HB. Our data regarding tumor extension and involvement of liver lobes are not comparable with the results of the SIOPEL 1 study due to the use of different staging systems. Brown et al. found that intrahepatic extension of the tumor and distant metastases were predictors for EFS and DFS.18 The SIOP group now divides their children with HB into a low-risk group and a high-risk group depending on these factors. The treatment strategies for patients with high-risk or advanced/metastatic HB are very different in international trials. The SIOPEL and United States-Intergroup studies intensified the chemotherapy with more cycles of chemotherapy and new drugs.4, 20, 36, 37 In contrast, a small group of patients with advanced and/or recurrent or metastatic HB in our HB 89 study were treated with VP16/CARBO as were all patients in the HB 94 study, and we observed a survival rate of 50% in 14 patients.23 In the current protocol, 18 children with HB were treated with VP16/CARBO. The treatment results underline the efficacy of this new chemotherapy arm: Sixty-six percent of the children with advanced HB or recurrent tumors responded, and surgical treatment was possible. In one patient, the complete disappearance of lung metastases was observed, and the child survived. The toxicity of VP16/CARBO was acceptable. Thrombopenia was observed more frequently compared with that seen with conventional chemotherapy. Therefore, high-dose VP16/CARBO chemotherapy after stem cell separation was recommended in the subsequent German study (HB 99) for all children who had Stage III HB with vascular invasion and/or multifocal, spreading tumor and Stage IV HB.

However, standard chemotherapy, consisting of CDDP/IFO/DOXO, was effective, with a response rate of at least 85%. There are great differences in the number of preoperative/postoperative chemotherapy courses among several international trials. On one hand, the toxicity should be reduced; and, on the other hand, an effective tumor treatment is necessary. The incidence of toxicity from CDDP/IFO/DOXO therapy was greater in the HB 94 study compared with the HB 89 study. One explanation may be the exact registration of side effects in the new study protocol after each course of chemotherapy. The main problems were leucopenia and thrombopenia with severe infection in 10% of all cycles. Although the dosage was reduced, two children age < 1 year died due to side effects. Those children underwent extended surgical procedures. Mild renal failure in four courses was caused by the administration of IFO. von Schweinitz et al. found a subclinical renal tubulopathy in 17% of children who were treated with CDDP/IFO/DOXO.9 Therefore, in the subsequent HB 99 study, the loading dose of 500 mg/m2 IFO will not be used in the chemotherapy regimes. However, it was found that IFO was an effective drug in the treatment of HB in animal studies.15, 38

In the current study, we observed three adverse events after patients underwent primary complete tumor resection (Stage I), and one of these children died. The cause of death was severe bleeding due to portal hypertension with esophageal varicosis, and not tumor recurrence. This is a deterioration compared with the prior HB 89 study. There are several possible reasons for this deterioration: First, all children with Stage I disease in the HB 94 study received only two courses of chemotherapy, compared with three postoperative courses in the HB 89 protocol. Second, from a surgical point of view, it is interesting to note that all three patients with adverse events underwent atypical tumor resection and not anatomic liver resection. Third, the role of preoperative chemotherapy has to be reconsidered. Some recent publications demonstrated the advantages of delayed surgery after chemotherapy. Preoperative chemotherapy, therefore, has been instituted for children with all stages of HB in the subsequent HB 99 study.4, 18, 39

The discussion about the phenomena of vascular invasion and the initial α-fetoprotein level is controversial in the literature. Both factors play a significant role in the HB 89 and HB 94 studies, in which 141 HBs were analyzed. However, Brown et al. found no correlation between the outcome of children with HB and vascular invasion.8, 18, 40

There are some reports correlating survival and α-fetoprotein decline during chemotherapy.24, 25 Van Tornout et al. found that the early changes in α-fetoprotein levels can be used for the identification of patients who are poor responders to treatment.41 This is concordant with our data about the poor outcome of patients who were nonresponders in both chemotherapy groups. In our study, there were two children who did not have elevated α-fetoprotein levels at the time of diagnosis. Both children died due to recurrent tumors. They had primarily hilar lymph node involvement and an unfavorable (teratoid/undifferentiated) histology.

Compared with the HB 89 study, there was no significant difference in DFS among the related histologic subtypes, both embryonal versus fetal and epithelial versus mixed differentiation. The same observation was made by Brown et al. and Conran et al.18, 42 Haas et al. found that the pure fetal histologic type was associated with improved survival compared with all other histologic patterns of HB (92% vs. 57%).4, 43 It is interesting to note that the amount of mixed HB was higher in the HB 94 study compared with the prior study. In addition, there is a trend toward higher numbers of mixed HB tumors after patients receive preoperative chemotherapy compared with the numbers in the group who underwent primary resection. These histopathologic features were described by Saxena et al., who found that the most notable feature was the extensive presence of osteoid in patients who were treated with chemotherapy.44

In conclusion, the certainty of the diagnosis of HB without biopsy in children age 6 months to 3 year with liver tumors and elevated α-fetoprotein is high. Chemotherapy using CDDP/IFO/DOXO and VP16/CARBO is effective. Complete tumor resection is one of the main prognostic factors, and atypical liver resection should be avoided. New surgical strategies, e.g., resection of advanced HB under extracorporal blood circulation with hypothermia or liver transplantation, may improve the outcome of these children.

Acknowledgements

The authors thank Mrs. Börner for excellent editorial assistance. They also acknowledge the kind collaboration of the patients' parents and the following principal clinical investigators: Dr. E. Pongratz (Josephinum, Augsburg); Dr. A. Gnekow (Städtisches Klinikum; Augsburg); Dr. G. F. Wündisch (Kinderklinik Städtische Krankenanstalten, Bayreuth); Dr. G. Henze (Universitätsklinikum Rudolf Virchow, Berlin); Dr. U. Bode (Universitäts-Kinderklinik, Bonn); Dr. H.-J. Spaar, Prof. (Hess-Kinderklinik, Bremen); Dr. J.-D. Thaben (Landeskrankenhaus, Coburg); Dr. I. Meyer (Vestische Kinderklinik, Datteln); Dr. H. Breu (Städtische Kliniken, Dortmund); Dr. U. Göbel (Universitätskinderklinik, Düsseldorf); Dr. A. Lemmer (Kinderklinik des Städtischen Klinikums, Erfurt); Dr. W. Havers (Kinderklinik der Universität, Essen); Dr. B. Kornhuber (Kinderklinik der Universität, Frankfurt/M); Dr. B. Brandis (Kinderklinik der Universität, Freiburg/br.); Dr. W. Schröter (Universitätskinderklinik Göttingen, Göttingen); Dr. C. Urban (Universitätskinderklinik, Graz); Dr. G. Günther (St. Barbarakrankenhaus, Halle/S); Dr. P. Reifferscheid (Altonaer Kinderkrankenhaus, Hamburg); Dr. R. Erttmann (Kinderklinik der Universität, Hamburg); Dr. U. Hofmann (Kinderkrankenhaus auf der Bult, Hannover); Dr. P. Weinel (Kinderklinik Medizinische Hochschule, Hannover); Dr. R. Ludwig (Universitätskinderklinik, Jena); Dr. M. Wright (Kinderkrankenhaus Park Schönfeld, Kassel); Dr. W. Sternschulte (Kinderklinik, Städtische Krankenanstalten, Köln); Dr. H.C. Dominik (Kinderklinik St. Marien und St. Annastift, Ludwigshafen); Dr. U. Mittler (Universitätskinderklinik, Magdeburg); Dr. P. Gutjahr (Universitätskinderklinik, Mainz); Dr. W. Tillmann (Kinderklinik Städtisches Klinikum, Minden); Dr. C. Bender-Götze (Dr. von Hauner'sches Kinderspital/Universität, München); Dr. S. Müller-Weihrich (Kinderklinik Schwabing, München); Dr. U. Jürgens (Universitätskinderklinik, Münster); Dr. A. Jobke (Cnopf′sche Kinderklinik, Nürnberg); Dr. Gröbe (Nürnberg-Klinikum Süd, Nürnberg); Dr. E. G. Huber (Kinderklinik, Landeskrankenanstalten, Salzburg); Dr. U. Leuthold (DRK-Kinderklinik, Siegen); Dr. R. Dickerhoff (Johanniter Kinderklinik, St. Augustin); Dr. J. Treuner (Olgahospital, Stuttgart); Dr. H. Gardner (St. Anna Kinderspital, Wien); Dr. R. Dopfner (Universitätskinderklinik, Tübingen); Dr. J. Kühl (Universitätskinderklinik, Würzburg); Dr. R. Geib (Winterbergkliniken GmbH, Saarbrücken); and Dr. v. Scharfe (Städtische Kinderklinik, Dresden).

Ancillary